Evaluation of agonistic interactions and behavioural …...2 Furthermore, pigs with shorter latencies in the human approach test were more aggressive in mixing situations (Brown et

Aus dem Institut für Tierzucht und Tierhaltung

der Agrar- und Ernährungswissenschaftlichen Fakultät

der Christian-Albrechts-Universität zu Kiel

Evaluation of agonistic interactions and behavioural tests

concerning systematic influences and genetic aspects

of pigs at different age levels

Dissertation

zur Erlangung des Doktorgrades

der Agrar- und Ernährungswissenschaftlichen Fakultät


vorgelegt von

Dipl. Ing. agr. Katharina Scheffler

aus Sangerhausen

Dekan: Prof. Dr. Dr. h.c. Rainer Horn

Erster Berichterstatter: Prof. Dr. Joachim Krieter

Zweiter Berichterstatter: Prof. Dr. Georg Thaller

Tag der mündlichen Prüfung: 29. Januar 2014

Die Dissertation wurde mit dankenswerter finanzieller Unterstützung aus Mitteln des

Bundesministeriums für Bildung und Forschung im Rahmen des Kompetenznetzes der Agrar-

und Ernährungsforschung PHÄNOMICS angefertigt.

MEINEN ELTERN

TABLE OF CONTENTS

GENERAL INTRODUCTION .................................................................................................................. 1

CHAPTER ONE

Characterisation of pigs into different personalities using the behavioural

tests backtest and human approach test……….…………………………………………5

CHAPTER TWO

Estimation of genetic parameters for agonistic behaviour of pigs

at different ages ....................................................................................................................... 23

CHAPTER THREE

Genetic analysis of the individual pig behaviour in backtests and

human approach tests ............................................................................................................ 43

CHAPTER FOUR

Relationship between behavioural tests and agonistic interactions

at different age levels in pigs ............................................................................................. 61

GENERAL DISCUSSION ..................................................................................................................... 83

GENERAL SUMMARY ........................................................................................................................ 93

ZUSAMMENFASSUNG ........................................................................................................................ 95

1

GENERAL INTRODUCTION

Animal welfare aspects are of increasing interest in modern pig production systems (Piñeiro et

al., 2013). In such systems, the perhaps most challenging and stressful situation, usually

implemented as a standard procedure for pigs, is the mixing of unacquainted animals of different

ages (Ismayilova et al., 2013). In this stressful situation, fighting is important to establish a

stable group hierarchy and to prevent permanent stress within the group. It is in this context that,

fighting is considered a normal behavioural pattern (Frädrich, 1974). However, there are large

individual differences in this behaviour can be observed. Observations have been made of on the

one hand pigs with enhanced aggressive behaviour and on the other hand pigs which show

extremely submissive behaviour. To prevent excessive stress, increased injuries and large

weight loss, highly aggressive pigs should be removed from the group (Tan et al., 1991; Stookey

and Gonyou, 1994; Tuchscherer, 2000). Due to the fact that video observations of agonistic

behaviour in such mixing situations are time-consuming and lavish, the usage of standardised

indicators to predict the agonistic interactions of the individual animals is necessary. For

practical application, these indicators should be easy to measure or to record under standardised

conditions. Possible indicators might be behavioural tests. In literature, studies using social tests

e.g. the backtest (Hessing et al., 1993) and the open-field test (Spoolder et al., 1996) as well as

non-social behavioural tests e.g. the human approach test (Thodberg et al., 1999) and the open-

door test (van Erp-van der Kooij et al., 2002). Here, the backtest is the most standardised test

with easy measurable and recognisable traits. In this test, pigs of this study are laid on their

backs and escape attempts are recorded in this stressful situation (according to Hessing et al.,

1993). The second behavioural test used in this study was the human approach test. In this test

situation, a stockperson stands in the pen and records the latency of the pigs to approach and

touch the stockperson (according to Thodberg et al., 1999). The human approach test provides

the opportunity to record the behaviour in a stressful, social situation with all pigs in the pen

simultaneously. Hence, the social aspects of the individual pigs in the human approach test e.g.

the rank order in the group, might be comparable in the mixing situations. Thus, the backtest and

the human approach test might be suitable indicators of the prediction of agonistic interactions

and also comply with the criteria of feasibility under practical conditions.

Relations between these two behavioural tests were found by Ruis et al. (2000). Low-reactive

pigs in the backtest showed more hesitation to approach and touch the stockperson. Connections

between the behavioural test and the agonistic behaviour were investigated in studies by Melotti

et al. (2011). They stated that highly reactive pigs in the backtest initiated more fights.

2

Furthermore, pigs with shorter latencies in the human approach test were more aggressive in

mixing situations (Brown et al., 2009).

So far, the analysis of agonistic behaviour as well as of backtests and human approach tests have

already been carried out in several studies producing partially contradictory results. However,

until now, investigations into the ontogenetic development of agonistic behaviour in relation to

the reaction of these pigs in the backtest and the human approach test under standardised testing

conditions have not been carried out. Neither were well documented evaluations of the agonistic

interactions and behavioural tests on a large number of animals available as well as the genetic

aspects of the used behavioural tests in general. Taking this into account, the aim of the present

study was to assess the agonistic behaviour of pigs and investigate the ontogenetic development

of this behavioural pattern. Furthermore, the examination of the behaviour of these animals in

backtests and human approach tests as suckling pigs, weaned pigs and gilts was carried out, in

order to find consistencies in behaviour between the same and different behavioural tests. To

examine the possibilities of using these traits in selections strategies, heritabilities of all

behavioural traits were estimated. Finally, the phenotypic and genetic relations between these

traits were investigated to verify whether these behavioural tests could be used as simple

indicators for agonistic interactions for mixing situations.

The assessment of systematic influences on the backtest and the human approach test in three

age groups (suckling pigs, weaned pigs and gilts) was carried out in Chapter One. Furthermore,

phenotypic correlations within the traits of the same test and between the traits of the backtest

and human approach tests were investigated to find behavioural consistencies of the reactions of

the individual pig. With the help of these results the animals were categorised in groups with

high or low reactions in both behavioural tests.

Chapter Two deals with the genetic analyses of the agonistic behaviour of pigs in different

mixing situations especially emphasising the ontogenetic development of the agonistic

behaviour in order to provide information about possible implementations in breeding programs.

Furthermore, systematic and random effects which had an impact on the specific agonistic

behavioural traits were analysed to improve the welfare of the pigs e.g. in mixing procedures as

in the required group housing of sows.

3

In the Chapter Three the genetic aspects of the behaviour of the pigs in the backtest and human

approach tests were analysed and heritabilities of the behavioural test traits and also the genetic

correlations between the traits of the same test and across traits of different tests were examined

to investigate whether the backtest and the human approach test can be used for implementation

in selection programs. The analysis also examined, whether both tests had the same genetic

base.

The connection of the agonistic behaviour at different ages to the backtest and the human

approach tests of weaned pigs and gilts was evaluated in Chapter Four. For this purpose,

phenotypic and genetic correlations of specific agonistic interaction traits and behavioural test

traits were estimated to examine the use of the backtest or human approach test as functional

and easy obtainable indicators to predict the agonistic behaviour of pigs in common mixing

situations at different ages.

References

Brown, J.A., Dewey, C., Delange, C.F.M., Mandell, I.B., Purslow, P.P., Robinson, J.A., Squires,

E.J., Widowski, T.M., 2009. Reliability of temperament tests on finishing pigs in group-

housing and comparison to social tests. Appl. Anim. Behav. Sci. 118, 28-35.

Frädrich, H., 1974. 'A comparison of behaviour in the Suidae'. The Behaviour of Ungulates and

its Relation to Management Vol 1, 133-143.

Hessing, M.J.C., Hagelsø, A.M., van Beek, J.A.M., Wiepkema, R.P., Schouten, W.G.P.,

Krukow, R., 1993. Individual behavioural characteristics in pigs. Appl. Anim. Behav.

Sci. 37, 285-295.

Ismayilova, G., Oczak, M., Costa, A., Thays Sonoda, L., Viazzi, S., Fels, M., Vranken, E.,

Hartung, J., Bahr, C., Berckmans, D., 2013. How do pigs behave before starting an

aggressive interaction? Identification of typical body positions in the early stage of

aggression using video labelling techniques [engl]. Wie verhalten sich Schweine vor

Beginn einer aggressiven Interaktion? Identifizierung typischer Körperpositionen im

frühen Stadium aggressiver Auseinandersetzungen anhand von Video-Labelling-

Techniken. Berl. Münch. Tierärztl. Wschr. 8, 113-120.

Melotti, L., Oostindjer, M., Bolhuis, J.E., Held, S., Mendl, M., 2011. Coping personality type

and environmental enrichment affect aggression at weaning in pigs. Appl. Anim. Behav.

Sci. 133, 144-153.

4

Piñeiro, M., Morales, J., Vizcaino, E., Murillo, J.A., Klauke, T., Petersen, B., Piñeiro, C., 2013.

The use of acute phase proteins for monitoring animal health and welfare in the pig

production chain: The validation of an immunochromatographic method for the

detection of elevated levels of pig-MAP. Meat. Sci. 95, 712-718.

Ruis, M.A.W., Brake, J., Van de Burgwal, J.A., de Jong, I.C., Blokhuis, H.J., Koolhaas, J.M.,

2000. Personalities in female domesticated pigs: behavioural and physiological

indications. Appl. Anim. Behav. Sci. 66, 31-47.

Spoolder, H.A.M., Burbidge, J.A., Lawrence, A.B., Simmins, P.H., Edwards, S.A., 1996.

Individual behavioural differences in pigs: intra-and inter-test consistency. Appl. Anim.

Behav. Sci. 49, 185-198.

Stookey, J.M., Gonyou, H.W., 1994. The effects of regrouping on behavioral and production

parameters in finishing swine. J. Anim. Sci. 72, 2804-2811.

Tan, S.S.L., Shackleton, D.M., Beames, R.M., 1991. The effect of mixing unfamiliar individuals

on the growth and production of finishing pigs. Anim. Sci. 52, 201-206.

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.

Tuchscherer, M.P., B., 2000. Dominance status affects immune response after social disturbance

in pigs. Arch. Tierz. Dummerstorf 43, 227.

van Erp-van der Kooij, E.V., Kuijpers, A.H., Schrama, J.W., van Eerdenburg, F.J.C.M.,

Schouten, W.G.P., Tielen, M.J.M., 2002. Can we predict behaviour in pigs?: Searching

for consistency in behaviour over time and across situations. Appl. Anim. Behav. Sci.

75, 293-305.

5

CHAPTER ONE

Characterisation of pigs into different personalities using the

behavioural tests backtest and human approach test

K. Scheffler, I. Traulsen and J. Krieter

Institute of Animal Breeding and Husbandry,

Christian-Albrechts-University,

Kiel, Germany

Submitted for publication in Livestock Science

6

Abstract

The knowledge of the reaction of pigs in challenging situations is important for breeding and

husbandry. The aim of the study was to investigate whether the backtest and the human

approach test, as simple and practical behavioural tests, can be used to characterise pigs

according to different behaviour in stressful situations i.e. – so-called “coping styles”. The

conditions for coping are the consistency of behaviour over time and across situations. The

backtest was performed twice with 1,382 suckling piglets. The human approach test was

performed at different ages: twice with suckling pigs (n=1,318), four times with weaned pigs

(n=1,317) and once with gilts (n=230). Significant effects on the traits of the number of escape

attempts (NEA), duration of escape attempts (DEA) and latency to the first escape attempt

(LEA) in the backtest were batch, test number and birth weight. The results of the latency (LC)

trait in the human approach tests were significantly influenced by batch, test number, and gender

(additionally the distance of the pen with weaned pigs to the door; additionally the body weight

with gilts). The correlations between the backtest traits were high (NEA – DEA rp = 0.73 to

0.79, NEA – LEA rp = -0.43 to -0.53, DEA – LEA rp = -0.43 to -0.54). Therefore, it is sufficient

to record only the NEA traits in further studies. The correlations between the two backtests

(rp = 0.31 – 0.43) and the Kappa-Coefficients (ĸ = 0.14 – 0.21), as a measurement of agreement

of the classes of behaviour, showed that the behaviour of the piglets in the first backtest was

different to the second one. The first backtest might thus be more the convincing test. The

relations between the human approach tests at different ages showed that the smaller the time

difference between was the tests, the higher was the correlation (rp = 0.20 – 0.52). A good

distinction between pigs was observed in weaned pigs and gilts. The human approach test with

gilts might be the more satisfying test with regard to breeding issues. The relation between the

backtests and the human approach tests was poor. Therefore, both tests seem to measure

different behaviour and the variation in behaviour is only random.

Keywords: pig, behaviour, backtest, human approach test, correlation, Kappa

7

Introduction

Stressful situations play an important role in all pig production. Mixing pigs which is a fairly

standard procedure in pig production, is perhaps the most challenging situation in the life of a

pig. Therefore, it is important to know how pigs deal with these situations and to investigate

existing differences in the behaviour of the individual pigs. Stress affects health and welfare as

well as production parameters. Mastering these different challenging situations with behavioural

and physiological effort is called coping (Koolhaas et al., 1999). Coping can be found in lots of

animals and has manifested itself in experiments with rodents. E.g. mice and rats, were

categorised into two different types in a social context (territorial behaviour), i.e. active and

passive coping styles (Benus et al., 1991). The two types showed endocrine and neuroendocrine

differences (Korte et al., 1992; Sgoifo et al., 1996). For example, the active rodents had a higher

sympathetic-adrenal activity, more adrenalin and noradrenalin and higher heart rates than the

passive ones. Additionally, the active copers had an decreased hypothalamic-pituitary-adrenal

(HPA) activity and corticosteroids (Korte et al., 1996). In literature, coping styles show a

consistency of HPA and sympathetic reactions over time and are characteristic of a certain

group of animals without a clarified association to behavioural characteristics. However, a few

studies find these two characteristics of coping styles in the behaviour of pigs also with

physiological differences, as has been observed in rodents (Hessing et al., 1994; Korte et al.,

1996; Ruis et al., 2001). Nevertheless, other authors have found no behavioural characteristics in

pigs and furthermore have criticised the method of recording coping styles (Forkman et al.,

1995; Jensen, 1995; Jensen et al., 1995). In practice, the measurement of the physiological part

of behaviour is more expensive and time-consuming. Therefore, several tests have been

developed which concentrate on the behavioural response of pigs to particular challenges.

Examples are the backtest (Hessing et al., 1993), the open field test (Spoolder et al., 1996), the

human approach test (Thodberg et al., 1999) and the open door test (van Erp-van der Kooij et

al., 2002). For a practical application, it is necessary for the tests to be simple to perform under

standardised conditions. With these tests, it is possible to describe behaviour over time and

across situations. The backtest measures the reaction of piglets to the fixation in the supine

position. The human approach test is performed with animals of different age groups and under

various test conditions (Thodberg et al., 1999; van Erp-van der Kooij et al., 2002; Janczak et al.,

2003; de Sevilla et al., 2009). The observer stands motionless in the pen for a defined duration

and the latency of the animal to touch the person is recorded. A relation between the backtest

and the response of the animals to the human approach test are given in Ruis et al. (2000).

Important for the evaluation of behavioural tests are the intra-situation consistency of one

8

specific test (across the same situation at different ages) and the inter-situation consistency of

the tests (between different stressful situations) (Jensen, 1995). However, there are only few

studies with a great animal number which have analysed backtest results in combination with

results of the human approach test. Therefore, the aim of the present study was to assess the

reactions of pigs to these different behavioural tests with a high number of animals. On this

basis, the relation within and among the backtests and the human approach tests was observed to

assess consistencies in behaviour. Through repeated test situations, differences in the individual

behaviour of pigs in three age groups - suckling piglets, weaned pigs and gilts - were recorded in

order to test the stability of the test measurements. Hence, a methodical validation was

performed to assess the individual pig behaviour in the two behavioural tests.

Material and methods

Animals and housing

The data were collected on the “Hohenschulen” research farm of the Institute of Animal

Breeding and Husbandry of the University Kiel (Germany) from December 2010 till August

2012. The herd consisted of purebred and crossbred animals of the German Landrace (DL) and

Large White breeds. The piglets from 139 litters (16 sows per batch) were kept in farrowing

pens for 26 days post partum (suckling period). The conventional farrowing stable consisted of

four compartments each with eight pens. These pens measured 2.2 m x 1.7 m and had a tiled and

metal base floor with no substrate. A piglet feeder was present from the first week after

farrowing. The lactating sows received commercial lactating feed in accordance with the

German norm (GfE, 2006). Water was accessible through nipple drinkers. At the first day of

age, each live-born piglet was marked and weighed individually (average weight 1.54 kg). In the

first three days, the piglets were cross-fostered to standardise the litter size for each sow and all

male piglets were castrated.

At weaning, the pigs were weighed individually (average weight 8.8 kg) and then housed in a

flatdeck pen. There were four compartments with 10 pens each. The dimension of one pen was

2.05 x 1.36 m and had a concrete and metal base floor with no substrate. Each pen had two

nipple drinkers for non-stop use. The pigs were fed ad libitum with solid, pelleted feed in

conformity with the German norm (GfE, 2006). The room temperature was maintained at a

minimum of 24°C. The pigs were re-mixed and sorted by the smallest level of familiarity and by

nearly equal weight. 8 till 10 piglets were housed in each pen. The pigs stayed in the flatdeck

pen for six weeks (on average 44 days).

The growing pigs were re-mixed in groups of 20 to 25 animals and housed in the growing

stable. The pens had a size of 3.25 x 2.40 m with a half-slatted and half-solid floor. Pigs had ad

9

lib access to water, which was accessible through nipple drinkers. The growing pigs were fed a

commercial diet by automatic mash feeding machine (GfE, 2006). The temperature was 22°C.

The pigs were sorted by the smallest level of familiarity and by nearly equal body size. A

maximum of two pigs already acquainted with each other from the flatdeck pens were housed

together.

The mixing and housing of gilts in groups of 17 to 28 sows in the breeding area (arena pen) was

in the 22nd week of age. The pen had a dimension of 7.2 x 5.4 m and a half-slatted and half-solid

floor. The gilts were fed gilt feed by automatic mash feeding machine with according to GfE

(2006). Water was accessible through nipple drinkers. The gilts were sorted by the smallest level

of familiarity, which means a maximum of five out of all pen mates were already acquainted

from the growing pens.

Backtest

At the age of 12 and 19 days, all piglets (n=1,382) were subjected to a backtest. The test was

performed in the home compartment of the piglet. The piglets were put on their back in a special

y-shaped device. Each piglet was taken out of the pen and tested individually. After the test, the

piglet was replaced and the next pen mate was tested. The experimenter held the piglet loosely

with his left hand and restrained it in this supine position (after Hessing et al., 1993). The test

began when the piglet lay still, and ended after one minute. During the test time, the number of

escape attempts (NEA), the latency to the first escape attempt (LEA) and the duration of all

escape attempts (DEA) were recorded.

The 25th and 75th percentiles of number and duration of escape attempts and the latency to the

first escape attempt were used to categorise the piglets into HR (high-reactive), LR (low-

reactive) pigs. All other piglets were classified as D (doubtful) (Bolhuis et al., 2005). NEA,

DEA and LEA were single categories assigned for the first and second backtests and for each

trait. From this, it follows that in sum every piglet had six separated categories of HR, LR or D.

The cut-offs were determined after all piglets had performed both backtests and was calculated

from the whole test values of the NEA, DEA and LEA. The piglets which showed more than

three escape attempts, a duration of more than 16 seconds and a latency of less than 8 seconds

were classified as high-reactive. Piglets which struggled fewer than twice, not longer than five

seconds and with a latency of more than 37 seconds were low-reactive.

10

Human approach test

The human approach test was performed with pigs which had also performed the backtest. The

human approach test took place two times (2nd and 3rd week of age) in the farrowing pen

(n=1,318), four times (at age of 6, 7, 8 and 9 weeks) in the flatdeck (n=1,317) and one time (22

weeks of age) with gilts (n=272). In the farrowing pen the stockperson crouched motionless in

the front of the pen, in the flatdeck and the arena pen the person stood still in front of the pen for

one minute. The gilts were observed in the arena pen. During this time the experimenter noted

which pigs made physical contact with the stockperson. Additionally, the experimenter recorded

the latency to touch the stockperson.

Statistical analysis

Statistical analyses were performed using the SAS® statistical software package

(SAS, 2008). The SAS procedure GLIMMIX was used for generalised linear mixed models.

The fixed effects were added stepwise in the model. A pseudo-likelihood method was used to

test the different models. The fit statistics AICC “Akaike’s information criterion corrected”

(Hurvich and Tsai, 1989) and the BIC “Bayesian information criterion” (Schwarz, 1978) were

used to compare the different models. The models with the smallest AICC and BIC were chosen

for the analysis. Significant differences of the least square means were adjusted with the

Bonferrone-correction (p < 0.05) (Westfall et al., 2011).

The analysis of the NEA backtest trait was carried out with a poisson distribution since this trait

represents count data (Haight, 1967). The duration of escape attempts (DEA) was separated into

classes of 10 seconds (class1: 0 s; class 2: 0 – 10 s; class 3: 11 – 20 s; class 4: 21 – 30 s; class 5:

31 - 60 s) and also analysed underlying a poisson distribution. The latencies to the first escape

attempt in the backtest (LEA) were separated into binary data (0: struggle, 1: no struggle). All

backtest traits included the fixed effects: batch (group 1 - 10) and test number (1 - 2). The birth

weight of the piglets was linear and quadratically included in the models as covariates.The

effects of the parity of the sow, cross-fostering, number of pen mates, gender and pen were

removed based on the model fitting. Additionally, these effects showed no significant influences

on the traits. The piglet was included in the model as a random effect. The latencies during the

human approach tests (latency class = LC) were analysed as binary data (0: touched the person;

1: did not touch the person). The model for the suckling piglets included the fixed effects of

batch (group 1 - 10), gender (male, female) and test number (1 - 2). The model for the weaned

pigs used the fixed effects of batch (group 1 – 10), test number (1 - 4), gender (male, female)

and the distance of the pen to the door (pens in front of the compartment: front, pens in back of

the compartment: back). The weight at weaning was included as a linear covariate in the model.

11

The effects of parity, cross-fostering and pen were removed from the model since the fitting

statistics showed no improvement and the effects had no significant impact. The pig was

included in the models for weaned as well as for suckling pigs as a random effect. The fixed

effect in the model for the gilts was only batch (group 5 -10) and as the covariate weight at time

of testing. The effects of the parity of the dam, cross-fostering and number of pen mates were

excluded because of the AICC and BIC of the model. The pen as a random effect was not used

and caused no improvement in the model fitting statistics.

The relationship between the traits and the tests were analysed using Spearman rank correlations

(PROQ CORR;SAS, 2008). The correlation between the first and second backtests and between

the traits NEA, DEA and LEA were calculated based on the residuals of the models. The

correlation coefficients of the human approach test result within one age group were also

estimated by the residuals of the models. Due to the different number of tests, the correlation

between different ages in the human approach test and between the backtest and the human

approach test were calculated with the animal effects.

Cohen’s Kappa-Coefficients (ĸ) (Landis and Koch, 1977) were used to test the consistency of

the coping styles (HR, LR, D) of the three backtest traits (NEA, DEA and LEA). The calculation

was separated into Kappa-Coefficients between the traits and within one test and between the

tests within one trait. The procedure Proq Freq of SAS was used to calculate Kappa (PROQ

FREQ;SAS, 2008).

Results

Fixed effects and covariates

The backtest traits NEA, DEA and LEA showed higher reactions in the first backtest (Table 1).

In the second backtest, the number of escape attempts was approximately 0.18 smaller than in

the first one. The duration class of the struggling was significantly higher in the first test. 88 %

of the piglets struggled in the first backtest and only 82 % in the second one.

Table 1: Least Square Means (LSMeans) and Standard Error (Std Error) for the test number of

the backtest traits number of escape attempts (NEA), duration class of escape attempts (DEA)

and latency class to the first escape attempt (LEA).

NEA (n = 1,382 ) DEA (n = 1,382 ) LEA (n = 1,382 )

LSMeans SE LSMeans SE LSMeans SE

Test number 1 1.88a* 0.03 2.60a 0.03 0.11a 0.01 2 1.70b 0.03 2.43b 0.03 0.16b 0.01

*Within columns LSMeans with different letters are significantly different (p < 0.05)

12

The results of the fixed effects of the human approach tests are shown in Figure 1. The human

approach test with suckling piglets showed higher latencies in the first test (LC= 0.95) than in

the second test (LC = 0.92). Additionally, the gender of the piglets significantly influenced the

results of the human approach test. Female piglets had shorter latencies (LC = 0.91) until the

first contact with the person than the male piglets (LC = 0.96). The human approach test of the

weaned pigs showed that the more often the test was carried out, the more piglets touched the

person; this proportion increased from 21 % to 50 %. The gender effect showed that female pigs

had shorter latencies (LC = 0.60) than male pigs (LC = 0.74). Moreover, weaned pigs in pens at

the front of the compartment had a shorter latency (LC = 0.62) than pigs in pens at the back of

the compartment (LC = 0.72). The frequency of gilts approaching the contact person was 35 %.

Figure 1: Least Square Means (LSMeans) and Standard Error (Std Error) for the latency class

(LC) in the human approach tests in suckling piglets and weaned pigs for the effects test

numbers (1, 2, 3, 4), gender (male: m; female: f) and distance of the pen to the door (back,

front). Within different age levels and within effects (test number, gender, distance) LSMeans

with different letters were significantly different. p<0.05.

The results of the backtests were significantly influenced by the birth weight of the pigs. Piglets

with a birth weight of 1.0 kg had the highest number of NEA (NEA = 1.89) and the highest

DEA (DEA = 2.54). The higher the birth weight of the piglets was the faster was the decrease in

13

NEA and DEA (Figure 2). The same tendencies were found in LEA (not shown). Lighter piglets

had a lower LEA than heavier piglets. This implies that the heaviest piglets at birth showed the

smallest reaction in the test situation of the backtests. The results of the human approach test

with suckling piglets and with weaned pigs were independent of the birth weight or the weight

at weaning. The influence of the weight on the human approach test latencies with gilts showed

that the higher the weight of the gilts were, the shorter was the latency to touch the person (LC90

kg = 0.80, LC120 kg = 0.58).

Figure 2: Number of escape attempts (NEA) and duration class of escape attempts (DEA)

depending on birth weight of piglets in the backtest. (p<0.05)

Correlations

The correlations between the backtest traits showed high coefficients between the NEA and

DEA of rp = 0.73 in the first backtest and one of rp = 0.79 in the second test (Table 2). A

negative relation was calculated between an NEA and LEA of rp = -0.43, as well as between a

DEA and LEA of rp = -0.43 in the first test. The correlations between the traits were about 0.1

higher in the second backtest. When comparing the first and second backtests, a medium

correlation of rp = 0.31 – 0.43 in NEA, DEA and LEA was obtained. The correlations between

the human approach tests within one age group are not shown. A moderate correlation with a

value of rp = 0.22 was calculated between the two human approach tests with suckling piglets. In

general, with a decreasing time difference between the performances of the tests with weaned

14

pigs, the correlation coefficients of the tests increased (rp = 0.20 – 0.52). The highest relation

could be obtained between test three and test four of the weaned pigs (rp = 0.52).

Table 2: Spearman-Rank-Correlation coefficients within and between backtest traits: number of

escape attempts (NEA), duration class of escape attempts (DEA) and latency class (LEA) to first

escape attempt (n = 1,382).

NEA - DEA NEA - DEA DEA - LEA

1st backtest 0.73* -0.43* -0.43*

2nd backtest 0.79* -0.53* -0.54*

NEA 1 - NEA 2 DEA 1 - DEA 2 LEA 1 - LEA 2

0.31* 0.33* 0.43*

* Values were significantly different to zero. p < 0.01

The relationship between the backtest and the human approach tests are shown in Table 3. The

backtest and the human approach test with suckling piglets had low correlation coefficients (rp =

-0.07 to -0.17). No significant correlations were calculated between the backtests and the human

approach test with weaned pigs and gilts. Minor correlations could be obtained between the

human approach test results of suckling pigs and weaned pigs (rp = 0.11) and between weaned

pigs and gilts (rp = 0.16).

Table 3: Spearman-Rank-Correlation1) coefficients between backtest traits: number of escape

attempts (NEA), duration class of escape attempts (DEA), latency class to first escape attempt

(LEA) and human approach tests (HAT) latency class (LC) at various ages.

Human approach test

LC Suckling pigs

(n = 1,317)

LC Weaned pigs

(n = 1,318)

LC Gilts

(n = 230)

Backtest NEA -0.07* -0.04 -0.03

DEA -0.07* -0.04 -0.05

LEA 0.17* 0.03 0.04

HAT LC Suckling pigs 0.11* 0.06

LC Weaned pigs 0.16* 1) Correlation between animal effects.

*Values were significantly different to zero. p < 0.05

15

Kappa-Coefficients

The Kappa-Coefficients of the coping styles HR, LR and D for each backtest trait NEA, DEA

and LEA are shown in Table 4. The relations between the backtest traits were more consistent in

the second backtest (ĸ = 0.43 – 0.53) than in the first one (ĸ = 0.38 – 0.49). Kappa-Coefficients

of the traits between the two tests showed a slight strength of agreement (ĸ = 0.14 – 0.21).

Table 4: Kappa-Coefficients for number (NEA) and duration (DEA) of escape attempts and

latency (LEA) of the backtest categories HR (high-reactive), LR (low-reactive)

and D (doubtful) with n = 1,382.

NEA - DEA NEA - LEA DEA - LEA

1st backtest 0.49* 0.38* 0.39*

2nd backtest 0.53* 0.43* 0.45*

NEA 1 - NEA 2 DEA 1 - DEA 2 LEA 1 - LEA 2

0.17* 0.21* 0.14*

*Values are significant different to zero. p<0.01

Discussion

Fixed effects and covariates

The reaction of pigs in the backtests showed that the piglets had higher NEA, DEA and lower

LEA in the first backtest than in the second one. Van Erp-van der Kooij (2002) stated that the

need to master this challenging situation became less significant in a repeated test situation. This

could also be shown in the second backtest of the present study. An explanation for this finding

could be a higher stress level in the first test situation. Therefore, the first backtest could be

considered as a better indicator of the behaviour of the pigs in such stressful situations.

The backtest traits were significantly influenced by the birth weight of the piglets. The studies of

van Erp-van der Kooij (2002) and D’Eath (2002) showed similar results. Lighter piglets had to

assert themselves more strongly against their littermates for example when fighting for a teat or

to establish the teat order (van Erp-van der Kooij et al., 2002). Therefore, the reaction of lighter

pigs to challenging situations such as the backtest was more reactive than the behaviour of

heavier pigs. Contrarily, the results of Cassady (2007) showed no effect of birth weight on the

recorded backtest traits.

In suckling and in weaned pigs, the LC of the human approach test as the probabilities of

whether the pigs touched the human or not were significantly influenced by the test number.

16

Higher probabilities to contact the humans were obtained with higher test numbers. An

explanation for this finding could be the habituation of the animals to the test situation. This is

in accordance with Velie et al. (2009) and Marchant-Forde et al. (2003), who stated that the

approach to humans depends on the number of contacts to humans in general. These statements

could be emphasised by the significant influence of the distance of the pen to the door. In this

study, the behaviour of pigs in pens closer to the door was more courageous than the behaviour

of the animal in the pens at the back of the compartment.

Additionally, the results of the human approach tests showed that the probability to touch the

stockperson was smaller in male pigs. Except for the castration of male piglets in early age, all

pigs were treated in the same way regardless of their gender. However, this treatment could be

evaluated as a negative experience in handling. Furthermore, Hemsworth et al. (1986) showed

that good (human contact: gently touched ) and bad handling (human contact: shocked with a

battery-operated prodder) leads to different results in the human approach tests. Pigs handled

more pleasantly tend to approach a human faster.

Correlations - backtest

The correlations between NEA, DEA and LEA were at a medium to high level. D’Eath (2002),

Cassady (2007) and Velie et al. (2009) found similar correlation coefficients for NEA and DEA.

This indicates that the use of NEA or LEA could be sufficient in further studies, which also

implies a simplification and standardisation of the backtest. However to compare the present

results with literature, the NEA trait will be preferred as the record in further investigations. The

same proceeding has been carried out in several other studies (Hessing et al., 1993; Bolhuis et

al., 2005; Zebunke et al., 2013). Furthermore, the correlations between the traits showed higher

coefficients in the second test. In agreement with this, the comparison HR, LR or D behavioural

classes between the traits NEA, DEA and LEA was more convergent. Hence, the Kappa-

Coefficients were higher in the second backtest. Considering these results, it could be suggested

that the behaviour of the animals was more consistent in the second backtest. The study of van

Erp-van der Kooij (2002), which carried out the backtest with pigs at nearly the same age,

showed similar results. Considering these facts and the already mentioned influence of the test

number, which indicate that pigs in the second backtest showed less reaction to this challenging

situation, the behaviour in the first test was different from the behaviour in the second test.

Furthermore, regarding the results of the kappa coefficients it could be shown that are higher

values were obtained between the backtest traits in the first backtest than in the second. This

means that the behaviour was more consistent in the first backtest and therefore, the first

backtest was more convincing.

17

Correlations - human approach test

The correlations between the tests at different ages showed that the pigs behaved differently in

each test. The intra-test consistency of the human approach test between different age levels was

low. Janczak et al. (2003) calculated nearly the same correlations for the human approach test.

The present results showed that only 15 % of the suckling piglets approached the human. The

proportion of pigs willing to touch the stockperson increased with the age of the pigs. In weaned

pigs, the probabilities of approaching increased up to 0.48 in the fourth test and the highest

correlations were calculated for the third and fourth tests. These results can be explained by

habituation to humans on the one hand and the test situation on the other hand. This was in

accordance to the results of van Erp-van Kooij (2002). The approach to humans depends on the

quality of handling and animals’ fear of humans (Hemsworth et al., 1989; Andersen et al.,

2006). This could be emphasised by the results of the human approach test with gilts. In this

test, 35 % of the gilts approached the human. Regarding the correlation coefficients between

different tests and the frequencies of pigs touching the human, the human approach test with

weaned pigs and gilts provides a clear distinction between the individual pigs and hence reliable

reactions to the human approach test could be recorded. The human approach test is usually

performed with gilts (Hemsworth et al., 1990; Thodberg et al., 1999; Janczak et al., 2003;

Andersen et al., 2006) and the heritabilities are at a medium level (Hellbrügge et al., 2009).

Hence, the human approach test with gilts might be preferred especially regarding its

implementation in breeding programs.

Correlations – backtest and human approach test

Ruis et al. (2001) stated that the total time spent struggling (here: DEA) is associated with the

human approach test. Contrarily, the present correlations between the backtest and human

approach test showed no relationship. Therefore, no inter-test consistency between the different

test situations could be observed. The two tests seemed to measure different behaviour. A

consistency in individual pig behaviour could not be obtained, which is in accordance with

literature (Forkman et al., 1995; Jensen et al., 1995; van Erp-van der Kooij et al., 2002; Cassady,

2007; Velie, 2007; Velie et al., 2009; Spake et al., 2012). The same behavioural characteristics

in these challenging situations were not obtained.

18

Conclusion

The backtest results show that it is sufficient to record only NEA. Additionally, the first backtest

is the convincing test to describe the pig behaviour. Due to the low correlations between the first

and the second backtests, consistencies of behaviour were not found. The human approach test

with gilts provides satisfying results for a distinction between individual pigs. Although the

correlations between the human approach tests of the three different ages were low, the medium

phenotypic correlations between the tests with weaned pigs were estimated. The present study

found no relation between the backtest and the human approach test which indicates that the

tests measure different behaviour in pigs. Therefore, the consistency of behavioural

characteristics between different tests was not observed and hence the behaviour might be only

random individual variation in the pigs.

References

Andersen, I.L., Berg, S., Bøe, K.E., Edwards, S., 2006. Positive handling in late pregnancy and

the consequences for maternal behaviour and production in sows. Appl. Anim. Behav.

Sci. 99, 64-76.

Benus, R.F., Bohus, B., Koolhaas, J.M., Van Oortmerssen, G.A., 1991. Heritable variation for

aggression as a reflection of individual coping strategies. Experientia 47, 1008-1019.

Bolhuis, J.E., Schouten, W.G.P., Schrama, J.W., Wiegant, V.M., 2005. Individual coping

characteristics, aggressiveness and fighting strategies in pigs. Anim. Behav. 69, 1085-

1091.

Cassady, J.P., 2007. Evidence of phenotypic relationships among behavioral characteristics of

individual pigs and performance. J. Anim. Sci. 85, 218-224.

D'Eath, B.R., Burn, C.C., 2002. Individual differences in behaviour: a test of "coping style" does

not predict resident-intruder aggressiveness in pigs. Behaviour 139, 1175-1194.

de Sevilla, X.F., Casellas, J., Tibau, J., Fàbrega, E., 2009. Consistency and influence on

performance of behavioural differences in Large White and Landrace purebred pigs.

Appl. Anim. Behav. Sci. 117, 13-19.

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

piglets. Applied Animal Behaviour Science 45, 31-42.

GfE, 2006. Empfehlungen zur Energie und Nährstoffversorgung von Schweinen. DLG-Verlag

Frankfurt am Main, Germany.

Haight, F.A., 1967. Handbook of the Poisson distribution. 100 - 107.

Hellbrügge, B., Tölle, K.-H., Presuhn, U., Krieter, J., 2009. A note on genetic parameters of gilt

responses to humans. Appl. Anim. Behav. Sci. 121, 153-156.

19

Hemsworth, P.H., Barnett, J.L., Coleman, G.J., Hansen, C., 1989. A study of the relationships

between the attitudinal and behavioural profiles of stockpersons and the level of fear of

humans and reproductive performance of commercial pigs. Appl. Anim. Behav. Sci. 23,

301-314.

Hemsworth, P.H., Barnett, J.L., Hansen, C., 1986. The influence of handling by humans on the

behaviour, reproduction and corticosteroids of male and female pigs. Appl. Anim.

Behav. Sci. 15, 303-314.

Hemsworth, P.H., Barnett, J.L., Treacy, D., Madgwick, P., 1990. The heritability of the trait fear

of humans and the association between this trait and subsequent reproductive

performance of gilts. Appl. Anim. Behav. Sci. 25, 85-95.



Sci. 37, 285-295.

Hessing, M.J.C., Hagelsø¸, A.M., Schouten, W.G.P., Wiepkema, P.R., Van Beek, J.A.M., 1994.

Individual behavioral and physiological strategies in pigs. Physiol. Behav. 55, 39-46.

Hurvich, C.M., Tsai, C.-L., 1989. Regression and time series model selection in small samples.

Biometrika 76, 297-307.

Janczak, A.M., Pedersen, L.J., Bakken, M., 2003. Aggression, fearfulness and coping styles in

female pigs. Appl. Anim. Behav. Sci. 81, 13-28.

Jensen, P., 1995. Individual variation in the behaviour of pigs- noise or functional coping

strategies? Appl. Anim. Behav. Sci. 44, 245-255.

Jensen, P., Rushen, J., Forkman, B., 1995. Behavioural strategies or just individual variation in

behaviour? - A lack of evidence for active and passive piglets. Appl. Anim. Behav. Sci.

43, 135-139.

Koolhaas, J.M., Korte, S.M., De Boer, S.F., Van Der Vegt, B.J., Van Reenen, C.G., Hopster, H.,

De Jong, I.C., Ruis, M.A.W., Blokhuis, H.J., 1999. Coping styles in animals: current

status in behavior and stress-physiology. Neurosci. Biobehav. R. 23, 925-935.

Korte, S.M., Buwalda, B., Bouws, G.A.H., Koolhaas, J.M., Maes, F.W., Bohus, B., 1992.

Conditioned neuroendocrine and cardiovascular stress responsiveness accompanying

behavioral passivity and activity in aged and in young rats. Physiol. Behav. 51, 815-822.

Korte, S.M., Meijer, O.C., de Kloet, E.R., Buwalda, B., Keijser, J., Sluyter, F., van Oortmerssen,

G., Bohus, B., 1996. Enhanced 5-HT< sub> 1A</sub> receptor expression in forebrain

regions of aggressive house mice. Brain Res. 736, 338-343.

20

Landis, J.R., Koch, G.G., 1977. The Measurement of Observer Agreement for Categorical Data.

Biometrics 33, 159-174.

Marchant-Forde, J.N., Bradshaw, R.H., Marchant-Forde, R.M., Broom, D.M., 2003. A note on

the effect of gestation housing environment on approach test measures in gilts. Appl.

Anim. Behav. Sci. 80, 287-296.




Ruis, M.A.W., te Brake, J.H.A., Engel, B., Buist, W.G., Blokhuis, H.J., Koolhaas, J.M., 2001.

Adaptation to social isolation: Acute and long-term stress responses of growing gilts

with different coping characteristics. Physiol. Behav. 73, 541-551.

SAS, I.I., 2008. SAS Institute Inc. Cary, NC, USA.

Schwarz, G., 1978. Estimation the dimension of a model. Ann. Stat. 6, 461-464.

Sgoifo, A., De Boer, S.F., Haller, J., Koolhaas, J.M., 1996. Individual differences in plasma

catecholamine and corticosterone stress responses of wild-type rats: relationship with

aggression. Physiol. Behav. 60, 1403-1407.

Spake, J.R., Gray, K.A., Cassady, J.P., 2012. Relationship between backtest and coping styles in

pigs. Applied Animal Behaviour Science 140, 146-153.



Behav. Sci. 49, 185-198.






75, 293-305.

Velie, B.D., 2007. Repeatabilities and heritabilities of behavior traits and their relationship with

performance. Accessed July 28, 2013. http://www.lib.ncsu.edu/resolver/1840.16/1995.

Velie, B.D., Maltecca, C., Cassady, J.P., 2009. Genetic relationship among pig behavior,

growth, backfat, and loin muscle area. J. Anim. Sci. 87, 2767-2773.

Westfall, P.H., Tobias, R.R.D., Wolfinger, R.D., 2011. Multiple comparisons and multiple tests

using SAS. SAS Institute.

21

Zebunke, M., Nürnberg, G., D.Repsilber, Puppe, B., 2013. Der Backtest beim Schwein —

Individuelle Variation im Verhalten oder Verhaltensstrategie? Aktuelle Arbeiten zur

artgemäßen Tierhaltung 2013, KTBL-Schrift 503, 188–197.

22

23

CHAPTER TWO

Estimation of genetic parameters for agonistic behaviour of pigs

at different ages

K. Scheffler1, E. Stamer², I. Traulsen1 and J. Krieter1

1Institute of Animal Breeding and Husbandry,


Kiel, Germany

²TiDa Tier und Daten GmbH,

Westensee/Brux, Germany

Submitted for publication in Journal of Agricultural Science

24

Abstract

In commercial pig production, the mixing of unacquainted pigs is a standard procedure which

leads to agonistic interactions with a wide range of individual pig behaviour. Hence, the aims of

the present study were to assess the heritabilities of agonistic behaviour and to estimate

correlations between three different age groups (weaned pigs n = 1,111, growing pigs n = 446,

gilts n = 279). The behavioural observation analysis included a period of 17 h directly after

mixing in a flatdeck, growing stable and arena pen whereby the following agonistic traits were

obtained: number of fights (NF), duration of fights (DF), initiated fights (IF), received fights

(RF), fights won (FW) and fights lost (FL). The behaviour of the weaned and growing pigs was

significantly influenced by cross-fostering, weight at mixing and litter. Cross-fostered animals

showed fewer agonistic interactions caused by the higher degree of socialisation as weaned pigs

(LSMeans e.g. NFcross-fostered: 13.3 ± 0.05, NFnon cross-fostered: 15.0 ± 0.03; p<0.05) as well as as

growing pigs (LSMeans e.g. NFcross-fostered: 4.3 ± 0.09, NFnon cross-fostered: 6.0 ± 0.5; p<0.05). The

influence of weight revealed that heavier pigs had a higher NF score at weaning (slope of linear

regression b = 0.06 ± 0.01; p<0.05) and as growing pigs (b = 0.03 ± 0.01; p<0.05). The random

litter effect explained up to 8 % of the whole variance in weaned and 4 % in growing pigs

whereby this could partly be explained by litter size. Pigs from larger litters tended to have more

agonistic interactions. The heritabilities of the recorded traits were at a low to medium level

(h² = 0.01 to 0.37) but similar between age groups. There were high correlations between NF

and all other traits in weaned pigs. The relations of the trait IF showed that the more IF a pig

had, the fewer fights it lost (rg = -0.16 ± 0.54, rp = 0.20) and the more fights it won (rg = 0.87 ±

0.07, rp = 0.83). Comparable values were estimated for the phenotypic correlations of the

growing pigs but genetic correlations were different. The relations between the age groups

provided no uniform trend and consequently, it is supposed that, the agonistic behaviour is not

consistent between different age levels.

Keywords: pig, behaviour, aggression, heritability, correlation

25

Introduction

The importance of behaviour in the breeding and husbandry of pigs is becoming more important

especially regarding animal welfare aspects. Comparing the behaviour of pigs in commercial

production to that in the natural environment indicates the same behavioural pattern (Frädrich,

1974). The social groups in modern pig production are not stable, e.g. the mixing of

unacquainted pigs after weaning, at the beginning of the growing period or in breeding herds is a

common practice (Ismayilova et al., 2013). In order to establish a stable hierarchy and to prevent

permanent stress within the group, fights take place between the animals. The interaction of pigs

using aggressive and submissive behaviour is called agonistic behaviour. Less aggressive

animals do not influence the other pigs negatively, e.g. due to stress or injuries (Tuchscherer and

Manteuffel, 2000). Hence, the excessive aggression of pigs especially towards low-ranking pigs

can influence welfare, health and weight gain, which are the most important economic factors in

pig production (Tan et al., 1991; Stookey and Gonyou, 1994). Therefore, one possible way to

reduce aggression and increase animal welfare is the breeding of calm and less-aggressive pigs

(Erhard et al., 1997; D'Eath et al., 2009). Currently, breeding organisations include traits

regarding behaviour and aggression as subordinated goals in their breeding programs (Kanis et

al., 2005). For the use of traits concerning agonistic interactions it is important to know their

heritability. Only a few estimates have been published on this so far (Løvendahl et al., 2005;

Turner et al., 2008; Turner et al., 2009; Stukenborg et al., 2012). For growing pigs, heritabilities

with wide ranges have also been estimated (Turner et al., 2008; Turner et al., 2009). According

to Stukenborg et al. (2012), heritabilities for growing pigs and gilts are higher than those for

weaned pigs. Obviously, weaned pigs also seem to display playful manners (Chaloupková et al.,

2008; Silerova et al., 2010). Fighting in growing pigs and in gilts is mainly motivated by

establishing the rank order. Agonistic interactions have been recorded by behavioural

observations in several studies. In contrast, the ontogenetic analyses of the aggressive and

submissive behaviour have been less documented. Comparisons of agonistic interaction at

different stages of life show that higher relations exist between growing pigs and gilts than

between weaned and growing pigs as well as between weaned pigs and gilts (Stukenborg et al.,

2012). However, the authors worked with focus animals, meaning that not every agonistic

interaction was recorded.

The aim of the present study was to examine the systematic effects on different traits related to

agonistic interactions in weaned pigs, growing pigs and gilts. Furthermore, heritabilities of six

behavioural traits and correlations between these traits in the different three age groups were

estimated to describe the pig´s ontogenetic development of the aggression towards conspecifics.

26

The present study is part of a larger investigation analysing the behaviour of pigs. These results

will be compared with results of behavioural tests for the prediction of pig behaviour at early

ages using standardised test situations.


Animals and housing

Data were recorded from December 2010 till August 2012 on the research farm “Hohenschulen”

of the Institute of Animal Breeding and Husbandry of the University Kiel (Germany). The pigs

were pure-bred and cross-bred animals of the breeds German Landrace (DL) and German

Edelschwein (DE). The piglets from 139 litters (16 sows per batch) were kept in farrowing pens

for 26 days postpartum (suckling period). The stable consisted of four compartments each with

eight pens. These conventional farrowing pens measured 2.2 m x 1.7 m and had a tiled and

metal base floor with no substrate. In accordance with the German norm (GfE, 2006) the

lactating sow received a commercial lactating feed. Water was assessable through nipple

drinkers. A piglet feeder was open to the piglets from the first week after farrowing. Each live

born piglet was marked and weighed individually (average weight 1.54 kg) at the first day of

age. Within the first three days the piglets were cross-fostered to standardise the litter size. The

cross-fostered piglets were the heaviest piglets of the litter. All male piglets were castrated.

At weaning the pigs were weighed individually (average weight 8.8 kg) and then housed in

flatdecks. There were four compartments with 10 pens each. The dimension of one pen was 2.05

x 1.36 m and had a concrete and metal base floor with no substrate. Two nipple drinkers were

available in each pen for non-stop water use. The pigs were fed ad libitum with solid pelleted

feed in conformity with the German standards (GfE, 2006). The room temperature was

approximately 24°C. The pigs were re-mixed and sorted by the smallest level of familiarity and

by nearly equal weight. Eight to ten pigs were housed in one pen and no pig knew another pig

from the farrowing pens. The pigs stayed in the flatdecks for six weeks (on average 44 days).

After these weeks in flatdeck, the pigs were re-mixed and re-housed in the growing stable in

groups of 20 to 25 animals. The pens had a size of 3.25 x 2.40 m with a half-slatted and solid

floor. Nipple drinkers for non-stop water use were accessible. The growing pigs were fed by

mash automats with a commercial diet (GfE, 2006). The temperature was 22°C. The pigs were

sorted by the smallest level of familiarity and by nearly equal body size. In the pens, a maximum

of two pen mates already knew each other from the time spent in the flatdeck pens.

In the 22nd week of age, the gilts were re-mixed and housed in the pen in the breeding area

(arena pen) in groups of 17 to 28 sows. The pen had a dimension of 7.2 x 5.4 m and a half

27

slatted and solid floor. The gilts were fed according to standards of the GfE (2006). Water was

accessible through nipple drinkers. All gilts were sorted by the smallest level of familiarity;

hence a maximum of five gilts knew each other from the growing pens.

Behavioural observations

The video observations started at approximately at 12:00 hrs immediately after rehousing and

remixing in the flatdeck, growing stable or arena pen and recorded the pigs’ behaviour for four

days. Stukenborg et al. (2010) stated that there was a declined rate of agonistic behaviour during

the night and Meese and Ewbank (1973) showed that the fighting behaviour decreased

fundamentally after two days of observation. Therefore, the video recording was interrupted at

the night (from 18:00 h to 07:00 h). Due to the high number of animals in the study, the period

used for the analysis was limited to 17 hours (day of housing: approx. 12:00 – 18:00 h; 2nd day:

07:00 – 18:00 h). The HeitelPlayer software (Xtralis Headquarter D-A-CH, HeiTel Digital

Video GmbH, Kiel, Germany) was used for the video analysis of the agonistic interactions. All

pigs in a pen were marked individually on their backs and could be observed in the whole pen.

Data from 1,111 weaned pigs, 446 growing pigs and 279 gilts could be used in the statistical

calculation.

The start and end of the fight, the initiator or receiver and the winner or loser of an agonistic

interaction were recorded for all marked pigs in the flatdecks, growing stable or arena pen. If the

aggressor/receiver or the winner/looser was not clear, the fights were recorded with unclear

starter/finisher or as stand-off fights. From all fights, six traits were obtained: number of fights

(NF), duration of fights (DF), number of initiated fights (IF), number of received fights (RF),

number of fights won (FW) and number of fights lost (FL). A fight was defined as a physical

contact longer than one second with aggressive behaviour initiated by one pig towards another

and ended in the submissive behaviour of an involved pig, i.e. the loser of the fight (Tuchscherer

et al., 1998; Langbein and Puppe, 2004). ‘Head to head knocks’, ‘head to body knocks’,

‘parallel/inverse parallel pressings’, ‘bitings’ or ‘physical displacements’ were recorded as

agonistic behaviour (Puppe, 1998; Stukenborg et al., 2012; Ismayilova et al., 2013). Submissive

behaviour was defined as the stop in a fight, a turning away, displacement from a location and

fleeing (Tuchscherer et al., 1998; Langbein and Puppe, 2004; Stukenborg et al., 2012). The

video observations of the weaned pigs in the flatdecks were carried out by three different

observers who had been trained using video test sequences at the beginning of the video

analysis. The definition and identification of the agonistic behaviour was tested with an

unknown video sequence. The inter-observer agreement was more than 90 %. The videos of the

growing pigs and gilts were analysed by only one person.

28


All statistical analyses were performed using the SAS® statistical software package (SAS,

2008). All traits of the original data had a non-normal distribution. Therefore, the descriptive

statistics used the medians of the data. Further calculations were performed with log-

transformed data (Y = loge (1+observation value)) in order to reduce the skewness. After

transformation the skewness of weaned pigs ranged from -0.77 to -0.1, for the growing pigs

from 1.08 to 0.23 and for the gilts from -0.46 to 0.42 between the traits. The curtosis of the traits

had a range of -0.63 to 0.76 for weaned pigs, of -0.79 to 1.03 for growing pigs and of -0.05 to

-0.74 for the gilts. Therefore, after log-transformation, the agonistic behavioural traits were

approximate normally distributed also observed by the visual inspection of the residual plots.

The analysis of the relevant systematic effects was performed using the procedure MIXED in

SAS (SAS, 2008). The Maximum Likelihood Estimation was used to test different models. The

fixed effects were added stepwise in the models. Evaluation of goodness of fit to the different

models was carried out by using the AICC, Akaikes’ information criterion and BIC, Bayesian

information criterion (Schwarz, 1978; Hurvich and Tsai, 1989). The smaller the AICC and BIC

the better fit the model.

The models for the different traits were the same within one age group (Model I: weaned pigs;

Model II: growing pigs; Model III: gilts).

Model I Yijklmno = µ + Bi + OBj + PENk + CFl + b*Wijklmno + litm + anin+ eijklmno

Model II Yilmno

= µ + Bi + CFl + b*Wjlmno + litm + anin + eilmno

Model III Yimno

= µ + Bi + b*Wimno + litm + anin + eimno

Where Yijklmno = observations of traits NF, DF, IF, RF,FW, FL of weaned pigs;

Yilmno = observation of traits NF, DF, IF, RF, FW, FL of growing pigs and Yimno = observation

of traits NF, DF, IF, RF, FW, FL of gilts; µ = overall mean; Bi = fixed effect of batch (i = 1 to

10); OBj = fixed effect of observer (j = 1, 2, 3); PENk = fixed effect of pen (k = 1 to 40);

CFl = fixed effect of cross-fostering (l = 1, 2); b*Wijklmno = linear regression on weight at

rehousing, litm = random effect of litter (m = 1 to 139); anin = random additive genetic effect of

the nth animal (n = 1 to 1,273). Not included effects did not improve the model fitting and

showed no significance (e.g. gender, parity of the sow or number of pen mates).

The heritabilities of the traits and the genetic and phenotypic correlations between the traits in

the different age groups were estimated with an animal model using the program package

VCE-6 (Kovac and Groeneveld, 2007). Before using VCE the data were prepared with the

29

program package PEST (Groeneveld, 1990). The pedigree contained two generations

backwards.

Results

Behavioural performance

Descriptive statistics of the agonistic behaviour between three age groups are presented in Table

1. Weaned pigs had the highest number of fights (NF), whereas the fewest NF could be

observed for the gilts. The weaned pigs fought 387 s, the growing pigs fought 279 s and gilts

174 s. The maximum duration of fights (DF) was recorded in growing pigs with 7,672 s.

Considering all traits, it was shown that the older the pigs were, the fewer agonistic interactions

could be observed.

Table 1: Median, minimum and maximum of behavioural traits (number of fights: NF; duration

of fights: DF; initiated fights: IF; received fights: RF; fights won: FW; fights lost: FL)

for three different age groups.

Weaned pigs (n = 1,111) Growing pigs (n = 446) Gilts (n = 279)

Median Min Max Median Min Max Median Min Max

NF 15a 1 116 6b 0 39 4c 0 32

DF (s) 387a 2 4647 279b 0 7672 174c 0 3623

IF 5a 0 68 3b 0 24 1c 0 28

RF 6a 0 37 3b 0 18 1c 0 15

FW 5a 0 98 2b 0 24 1c 0 31

FL 7a 0 52 3b 0 19 2c 0 17

a,b,c Within row medians with different letters were significantly different (Wilcoxon rank-sum test; p< 0.01)

30

Fixed and random effects

Except for DF, the observer significantly influences all traits of the weaned pigs (p<0.01).

Testing linear contrasts of the three observers, it could be shown that one observer had

significant different results of the agonistic interactions from the other observers (p<0.05).

Cross-fostering influenced the agonistic interactions of the weaned pigs (back transformation of

LSMeans e.g. NFcross-fostered: 13.3 ± 0.05, NFnon cross-fostered: 15.0 ± 0.03; p<0.05) as well as all

traits of the growing pigs (back transformation of LSMeans e.g. NFcross-fostered: 4.3 ± 0.09,

NFnon cross-fostered: 6.0 ± 0.5; p<0.05). Pigs which had not been raised by their own dam showed

fewer agonistic interactions and were less aggressive than non cross-fostered animals. The

behaviour of the gilts was no longer influenced by cross-fostering.

The weight at rehousing at weaning or in the growing stable influenced the agonistic behaviour

of the pigs significantly. Heavier pigs fought more than lighter pigs. The slopes of the linear

regression were for the weaned pigs 0.06 ± 0.01 (p<0.05) and for the growing pigs 0.03 ± 0.01

(p<0.05). The fighting of gilts was not influenced by the weight of the animals.

The estimated values of the random animal effect (additive genetic effect), random litter effect

and the residual effect are shown in Table 2. In weaned pigs, the portion of litter ranged from 0

to 8 %. In growing pigs this portion was lower (0 – 4 %). The behaviour of gilts was not

influenced by the litter effect.

Heritabilities

For the weaned pigs, the heritabilities ranged from 0.06 for DF to 0.37 for the RF (Table 2).

Heritabilities for NF, FW and FL were at a common level (0.15 to 0.20). In growing pigs, the

heritabilities did not substantially differ between the recorded behavioural traits (0.11 to 0.18),

apart from DF, which was not heritable. The heritabilities for gilts ranged from 0.01 for NF to

0.12 for FW. High standard errors for the heritabilities were estimated in all age groups (0.03 –

0.19).

Table 2: Variances of the animal effect (δa² = additive genetic variance), the litter effect (δli² = permanent environmental effect) and residual effect

(δe²) for the behavioural traits (number of fights: NF; duration of fights: DF; initiated fights: IF; received fights: RF; fights won: FW; fights lost: FL)

in different ages (Weaned Pigs: NF, DF, FW, FL n = 1111 and IF, RF n = 778; Growing Pigs: n = 446; Gilts: n = 279).

Trait δa² δli² δe² h²

Weaned Growing Gilts Weaned Growing Gilts Weaned Growing Gilts Weaned Growing Gilts

pigs pigs pigs pigs pigs pigs pigs pigs

NF 0.05 0.12 0.07 0.02 0.00 0.00 0.30 0.52 0.66 0.15 (0.09)* 0.18 (0.08) 0.10 (0.11)

DF 0.06 0.01 0.49 0.07 0.13 0.08 0.74 3.51 5.33 0.06 (0.07) 0.01 (0.10) 0.08 (0.11)

IF 0.07 0.09 0.06 0.07 0.02 0.00 0.63 0.57 0.54 0.09 (0.16) 0.13 (0.19) 0.10 (0.11)

RF 0.14 0.05 0.20 0.00 0.00 0.00 0.24 0.41 0.49 0.37 (0.08) 0.11 (0.07) 0.04 (0.10)

FW 0.15 0.07 0.08 0.04 0.00 0.00 0.80 0.60 0.55 0.15 (0.11) 0.11 (0.08) 0.12 (0.11)

FL 0.11 0.07 0.00 0.00 0.00 0.00 0.38 0.37 0.00 0.22 (0.13) 0.16 (0.08) 0.01 (0.03)

*Standard errors in brackets

31

32

Correlations between behavioural traits within different age groups

The phenotypic and genetic correlations between the behavioural traits are shown in Table 3.

The correlation between all traits and the NF trait showed high phenotypic in all age groups as

well as genetic correlations. The IF is highly correlated with the trait FW whereas RF was

highly correlated with FL. There was a negative relation between FW and FL for the weaned

pigs. The traits IF and RF showed small correlations. These results were especially observed for

the phenotypic and genetic correlations in weaned pigs. The correlations for the growing pigs

and gilts between the recorded traits were high for almost all traits.

Correlations within behavioural traits between different age groups

Table 4 shows the phenotypic and genetic correlations between the three age groups. The

phenotypic correlations between the weaned and growing pigs and between growing pigs and

gilts were low. The coefficients ranged from -0.04 to 0.34 within the different traits. It is shown

that the shorter the time difference between the groups was, the higher were the correlations.

The phenotypic correlations between the age groups differed to the genetic correlations and also

the genetic relations varied extremely between the traits (rg = -0.06 to 0.99). Also, the standard

errors of the correlations were very high.

Table 3: Genetic (rp) and phenotypic (rp) correlation of different age stages between the behavioural traits (number of fights: NF; duration of fights:

DF; initiated fights: IF; received fights: RF; fights won: FW; fights lost: FL) within the three stages of life.

DF IF RF FW FL

rg rp rg rp rg rp rg rp rg rp

Weaned pigs NF 0.85 ± 0.17 0.82 0.79 ± 0.19 0.83 0.62 ± 0.55 0.68 0.67 ± 0.21 0.77 0.25 ± 0.75 0.45

DF 0.48 ± 0.30 0.68 0.75 ± 0.38 0.55 0.60 ± 0.26 0.67 0.08 ± 0.86 0.29

IF 0.02 ± 0.57 0.28 0.87 ± 0.07 0.83 -0.16 ± 0.54 0.20

RF -0.05 ± 0.49 0.33 0.67 ± 0.31 0.57

FW -0.54 ± 0.09 -0.03

Growing pigs NF 0.96 ± 0.07 0.88 0.98 ± 0.13 0.86 0.87 ± 0.03 0.81 0.99 ± 0.01 0.81 0.96 ± 0.06 0.78

DF 0.90 ± 0.15 0.73 0.97 ± 0.08 0.70 0.97 ± 0.10 0.69 0.85 ± 0.18 0.66

IF 0.77 ± 0.25 0.48 0.98 ± 0.03 0.84 0.99 ± 0.03 0.56

RF 0.88 ± 0.17 0.55 0.71 ± 0.21 0.75

FW 0.14 ± 0.07 0.37

Gilts NF 0.95 ± 0.04 0.93 0.97 ± 0.02 0.86 0.99 ± 0.00 0.88 0.98 ± 0.02 0.88 0.99 ± 0.01 0.89

DF 0.90 ± 0.08 0.77 0.96 ± 0.05 0.79 0.90 ± 0.06 0.79 0.95 ± 0.06 0.80

IF 0.97 ± 0.03 0.60 0.99 ± 0.01 0.86 0.99 ± 0.02 0.72

RF 0.98 ± 0.03 0.74 0.99 ± 0.01 0.83

FW 0.20 ± 0.06 0.64

33

Table 4: Genetic (rg) and phenotypic (rp) correlations for the behavioural traits (number of fights: NF; duration of fights: DF; initiated fights: IF;

received fights: RF; fights won: FW; fights lost: FL) between different age stages.

Weaned pigs – Growing pigs Weaned pigs – Gilts Growing pigs – Gilts

rg rp rg* rp rg

* rp

NF 0.38 ± 0.44 0.27 -0.06 ± 0.85 -0.03 0.39 ± 0.40 0.24

DF -0.99 ± 0.29 0.26 0.99 ± 0.11 -0.03 0.99 ± 0.01 0.25

IF 0.99 ± 0.01 0.34 0.99 ± 0.01 0.02 0.99 ± 0.01 0.29

RF -0.03 ± 0.54 0.14 0.40 ± 0.01 -0.07 0.30 ± 0.96 0.07

FW 0.99 ± 0.03 0.29 0.85 ± 0.97 0.06 0.45 ± 0.45 0.25

FL -0.40 ± 0.41 -0.04 -0.99 ± 0.01 -0.11 0.99 ± 0.01 0.15

34

35

Discussion

Fixed and random effects

Despite the intensive training on different video sequences, the observers had significant

influence on the recorded traits of the agonistic interactions in weaned pigs. This significant

effect of the observer implies the necessity for the definition of the agonistic interaction to be

absolutely clear for all observers and the inter-observer reliability must be continuously verified

during the complete analysis of the videos. Linear contrasts between the three observers showed

that one had significantly different results compared to the observations of the other observers.

The influence of the divergent observer on the present results especially on the heritabilities was

tested without these observations by excluding the animals observed by the observer (n = 348

animals). The estimated heritabilities without these data did not differ between the presented

results.

Pigs raised by their own dam had more agonistic interactions at weaning and rehousing in the

growing stable and were more aggressive than cross-fostered pigs. In literature, this effect is less

documented. D’Eath (2004) stated that the pigs which had been socialised before first mixing

with unacquainted pigs showed more consistently aggressive behaviour. These pigs established

the rank order faster due to the learning of social behavioural skills in earlier age (D'Eath, 2005).

The effect of cross-fostering could be explained in the same way. The cross-fostered pigs learnt

the social behaviour very early and how they had to react in mixing with unacquainted pigs.

Therefore, they did not fight very often to establish a rank order. The non cross-fostered pigs did

not have this experience in the first mixing and thus obtained fewer social skills. Experiences in

early age and success or failure in aggressive interactions had long-lasting effects on the animals

(D'Eath, 2004). In commercial pig production, the heaviest animals are used for cross-fostering.

The weight at weaning and rehousing in the growing stable of the cross-fostered and non cross-

fostered animals was compared in the present study to avoid an effect of the weight of the pigs

within the cross-fostering effect. Here, no significant differences were analysed. The cross-

fostering had no influence on the behaviour of the gilts.

In literature, different results due to the influence of the weight of the pigs on the agonistic

behaviour can be found. Litten et al. (2003) and Turner and Camerlink (2011) stated that there is

an influence of weight on the behaviour of the pigs. Heavier pigs were more active and more

dominant in their studies. In contrast, Fels and Hoy (2013) obtained no differences between

agonistic interactions in groups sorted by light and heavy pigs. The present results show an

influence of the weight, which means that the heavier the pigs were, the more aggressive they

36

were. This was observed in weaned pigs as well as in growing pigs. An explanation might be the

ability of the heavier pigs to protect the own rank position from the last group of conspecifics

(e.g. in flatdeck) against the new and unacquainted pigs (e.g. in growing stable) after remixing.

This could be emphasised by statements that the previous dominance rank had a prolonged

effect to the rank position in later groups (D’Eath (2004) and Otten et al. (1997)). The present

results also show that heavier pigs initiated more fights which also could be explained by the

consciousness of the pigs to their own rank position. The influence of the weight decreased with

the age of pigs and in gilts the weight had no impact on the agonistic interactions.

The random litter effect explained small parts of the whole variance in weaned and growing

pigs. The older the pigs were, the smaller the influence of the litter became and therefore the

highest portions of litter variance ware found for the weaned pigs. In contrast, the behaviour of

the gilts was not affected by the litter component. The litter effect describes the maternal

genetic and maternal environmental effect (Roehe et al., 2009). Stukenborg et al. (2012) stated

that the behaviour of the pigs was influenced by pre-weaning experiences, illustrated by the

litter effect. Events before farrowing could also substantially influence the development of the

brain and the behaviour of the animals (Champagne and Curley, 2005). Jarvis et al. (2006)

found that stress in pregnant sows had an effect on the aggressive behaviour of the offspring. In

order to obtain more information about the influence of behaviour of the dam on the agonistic

behaviour of its piglets, a separation test was carried out. All dams of the pigs used in this study

performed these separation tests at the first, 12th and 19th days after farrowing. The test recorded

the body position of the sow before and after separation from the piglets and validated the

reaction in five categories from no reaction to aggressive behaviour towards the stockperson

(after Hellbrügge et al., 2009). The results of these tests were compared with the agonistic

behaviour of the offspring in the present mixing situations. The aggressive behaviour of the

dam did not influence the agonistic behaviour of the offspring in the present study (Scheffler

and Krieter, 2013). D’Eath and Lawrence (2004) stated that pigs from larger litters were more

aggressive compared with smaller litters. It was explained by the competition of the piglets for a

teat or in general with the limited food resource (Fraser, 1975; Fraser and Jones, 1975). The

influence of the litter effect in the present study was also tested to determine whether there were

differences due to the litter size and the behaviour of the pigs. It could be shown that pigs of

larger litters tended to have more agonistic interactions (p≤0.1) and therefore it might be an

explanatory approach for the influence of the random litter effect.

37

Heritabilities

The heritabilities were at nearly the same level within traits and between age groups. These

findings could not emphasise the statements of Stukenborg et al. (2012), who explained the

differences between heritabilities of age groups with the enhanced playing behaviour of the

weaned pigs. Silerova et al. (2010) stated that the fighting and playing behaviour in weaned pigs

could not be separated from each other. In studies of Turner et al. (2008; 2009), heritabilities of

0.008 and 0.46 were estimated for agonistic interactions with weaned pigs. The values also had

a wide range in contrast to the present study. Løvendahl et al. (2005) estimated heritabilities for

the agonistic behaviour of sows with values of 0.17 to 0.24 for the number of aggressions and

with 0.04 to 0.06 for received aggressions. These results were in accordance with the

heritabilities of the present study. In literature it is stated that agonistic interactions are related to

reproductive traits. Tönepöhl et al. (2013) showed that sows which initiated more fights had

more piglets in total and more born alive. In contrast, sows with fewer agonistic interactions

cared better for their offspring (Løvendahl et al., 2005). Therefore, it seems that the traits NF

(h² = 0.10 to 0.18) and IF (h² = 0.09 to 0.13) should be the most important and will be discussed

in further investigations (Scheffler et al., 2013).

Correlations between behavioural traits within different age groups

The NF trait showed high genetic and phenotypic correlations to all traits and within all age

groups. The high genetic correlations of the traits show that the reactions depend on the same

genetic base. Turner et al. (2008) estimated genetic correlations between number of fights and

initiated fights of 0.79. These results were in accordance with the present findings. As expected,

there were high correlations between IF and FW and RF and FL and lower correlations between

IF and FL as well as between RF and FW. This means, that pigs which initiated fights were

more often the winner and pigs which received most of the fights lost the fights most often. The

traits IF and RF showed low correlations. Especially, the results of the weaned pigs confirmed

these findings. Due to the smaller number of growing pigs and gilts in the analysis, the genetic

correlations were overestimated especially if the heritabilities of the traits were very small.

Therefore, to complete the results of the correlation between the traits, the growing pigs and

gilts were separated into groups of 25 % quantiles according to the regarded traits. E.g. the

animals were separated into pigs with the 25 % highest and 25 % lowest NF and Wilcoxon rank-

sum tests were carried out to test the differences between these groups in IF. It could be shown,

that the expectations (see text above) were complied with. The separation of pigs into groups of

high and low IF showed that the group with high IF had significant more NF. Groups with high

and low IF differed significantly in the number of FW and groups with high and low RF differed

38

significantly in the number of FL. The low estimates for the correlations between FL and FW

could be explained by the statements of Rushen and Pajor (1987). They stated that there is a

balance between offensive and defensive interactions in pigs. These results agree with parameter

estimation of Stukenborg et al. (2012) and Turner et al. (2008). They also found low correlations

between the number of initiated and received fights.

Correlations within behavioural traits between different age groups

There were low phenotypic correlations between the age groups. The correlations were slightly

higher the smaller the time differences between the three groups were. This means there were

higher correlations between weaned and growing pigs than between weaned pigs and gilts. The

correlations were at nearly the same level for all traits. These results were in accordance with

Stukenborg et al. (2012). They estimated phenotypic correlations between 0 to 0.47. However,

the relations also increased with smaller time differences between the age groups and is not in

accordance with statements of Clark and D’Eath (2013). They stated that aggressive behaviour

is a stable and consistent temperament of the individual pig. The genetic correlations were again

much higher than the phenotypic and showed no general trend between the age groups. This can

also be explained by a too small number of animals for reliable estimation of genetic

correlations.

Conclusion

The agonistic interaction of pigs at different ages showed that cross-fostering is an important

effect especially on the behaviour of weaned and growing pigs. Thus, cross-fostered animals are

less aggressive as weaned and growing pigs. Hence, socialisation in early life of pigs leads to

less-aggressive behaviour in further mixing situations. The agonistic interactions in weaned and

growing pigs cannot be regarded without waiving the weight of the pigs. The heavier the pigs

are, the more aggressive they are. The most important traits to describe agonistic behaviour in

pigs are thus the number of fights and number of initiated fights with low to moderate but

consistent heritabilities for all age groups. Also, the correlations between and across traits and

age levels suggested that number of fights and number of initiated fights were the most suitable

traits for the assessment of the agonistic interactions and the aggressiveness of pigs and should

therefore be considered in further studies.

39

References

Chaloupková, H., Illmann, G., Pedersen, L.J., Malmkvist, J., Simecková, M., 2008. Sow

responsiveness to human contacts and piglet vocalization during 24 h after onset of

parturition. Appl. Anim. Behav. Sci. 112, 260-269.

Champagne, F.A., Curley, J.P., 2005. How social experiences influence the brain. Curr. Opin.

Neurobiol. 15, 704-709.

Clark, C.C.A., D'Eath, R.B., 2013. Age over experience: Consistency of aggression and

mounting behaviour in male and female pigs. Appl. Anim. Behav. Sci. 147, 81-93.

D'Eath, R.B., 2004. Consistency of aggressive temperament in domestic pigs: The effects of

social experience and social disruption. Aggressive Behav. 30, 435-448.

D'Eath, R.B., 2005. Socialising piglets before weaning improves social hierarchy formation

when pigs are mixed post-weaning. Appl. Anim. Behav. Sci. 93, 199-211.

D'Eath, R.B., Lawrence, A.B., 2004. Early life predictors of the development of aggressive

behaviour in the domestic pig. Anim. Behav. 67, 501-509.

D'Eath, R.B., Roehe, R., Turner, S.P., Ison, S.H., Farish, M., Jack, M.C., Lawrence, A.B., 2009.

Genetics of animal temperament: aggressive behaviour at mixing is genetically

associated with the response to handling in pigs. Animal 3, 1544-1554.

Erhard, H.W., Mendl, M., Ashley, D.D., 1997. Individual aggressiveness of pigs can be

measured and used to reduce aggression after mixing. Appl. Anim. Behav. Sci. 54, 137-

151.

Fels, M., Hoy, S., 2013. Einflüsse einer Gewichtssortierung von Aufzuchtferkeln auf

Lebendmasseentwicklung, aggressive Auseinandersetzungen und Verletzungsgrad nach

der Gruppierung. Effects of sorting weaner pigs by weight on growth performance,

aggressive interactions and skin lesion score after mixing. Berl. Münch. Tierärztl.

Wschr. 1, offen-offen.

Frädrich, H., 1974. 'A comparison of behaviour in the Suidae'. The Behaviour of Ungulates and

its Relation to Management Vol 1, 133-143.

Fraser, D., 1975. The teat order of suckling pigs: II. Fighting during suckling and the effects of

clipping the eye teeth. J. Agr. Sci. 84, 393-399.

Fraser, D., Jones, R.M., 1975. The teat order of suckling pigs: I. Relation to birth weight and

subsequent growth. J. Agr. Sci. 84, 387-391.



40

Groeneveld, E., 1990. PEST user's manual. Institute of Animal Husbandry and Animal

Behaviour Federal Research Centre, Neustadt, 1-80.












Jarvis, S., Moinard, C., Robson, S.K., Baxter, E., Ormandy, E., Douglas, A.J., Seckl, J.R.,

Russell, J.A., Lawrence, A.B., 2006. Programming the offspring of the pig by prenatal

social stress: Neuroendocrine activity and behaviour. Horm. Behav. 49, 68-80.

Kanis, E., De Greef, K.H., Hiemstra, A., van Arendonk, J.A.M., 2005. Breeding for societally

important traits in pigs. J. Anim. Sci. 83, 948-957.

Kovac, M., Groeneveld, E., 2007. VCE-6 users guide and Reference Manual Version 6.

University of Ljubljana, Slovenia, 75.

Langbein, J., Puppe, B., 2004. Analysing dominance relationships by sociometric methods - a

plea for a more standardised and precise approach in farm animals. Appl. Anim. Behav.

Sci. 87, 293-315.

Litten, J.C., Drury, P.C., Corson, A.M., Lean, I.J., Clarke, L., 2003. The Influence of Piglet

Birth Weight on Physical and Behavioural Development in Early Life. Neonatology 84,

311-318.

Løvendahl, P., Damgaard, L.H., Nielsen, B.L., Thodberg, K., Su, G., Rydhmer, L., 2005.

Aggressive behaviour of sows at mixing and maternal behaviour are heritable and

genetically correlated traits. Livest. Prod. Sci. 93, 73-85.

Meese, G.B., Ewbank, R., 1973. The establishment and nature of the dominance hierarchy in the

domesticated pig. Anim. Behav. 21, 326-334.

41

Otten, W., Puppe, B., Stabenow, B., Kanitz, E., Schön, P.C., Brüssow, K.P., Nürnberg, G.,

1997. Agonistic interactions and physiological reactions of top- and bottom-ranking pigs

confronted with a familiar and an unfamiliar group: Preliminary results. Appl. Anim.

Behav. Sci. 55, 79-90.

Puppe, B., 1998. Effects of familiarity and relatedness on agonistic pair relationships in newly

mixed domestic pigs. Applied Animal Behaviour Science 58, 233-239.

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. Livest. Sci. 121, 173-181.

Rushen, J., Pajor, E., 1987. Offence and defence in fights between young pigs (Sus scrofa).

Aggressive Behavior 13, 329-346.


Scheffler, K., Krieter, J., 2013. Analysing the relation between agonistic behaviour, behavioural

tests and litter size. Unpublished Results.

Scheffler, K., Stamer, E., Traulsen, I., Krieter, J., 2013. Relationship between behavioural tests

and agonistic interactions at different age levels in pigs. Unpublished Results.


Silerova, J., Spinka, M., Sarova, R., Algers, B., 2010. Playing and fighting by piglets around

weaning on farms, employing individual or group housing of lactating sows. Appl.




Stukenborg, A., Traulsen, I., Puppe, B., Presuhn, U., Krieter, J., 2010. Agonistic behaviour after

mixing in pigs under commercial farm conditions. Appl. Anim. Behav. Sci. 129, 28-35.

Stukenborg, A., Traulsen, I., Stamer, E., Puppe, B., Presuhn, U., Krieter, J., 2012. Heritabilities

of agonistic behavioural traits in pigs and their relationships within and between

different age groups. Livest. Sci. 149, 25-32.



Tönepöhl, B., Appel, A.K., Voß, B., König von Borstel, U., Gauly, M., 2013. Interaction

between sows' aggressiveness post mixing and skin lesions recorded several weeks later.

Appl. Anim. Behav. Sci. 144, 108-115.

42

Tuchscherer, M., Manteuffel, G., 2000. Die Wirkung von psychischem Stress auf das

Immunsystem. Ein weiterer Grund für tiergerechte Haltung (Übersichtsreferat). Arch.

Tierz., Dummerstorf 43, 547-560.

Tuchscherer, M., Puppe, B., Tuchscherer, A., Kanitz, E., 1998. Effects of social status after

mixing on immune, metabolic, and endocrine responses in pigs. Physiol. Behav. 64, 353-

360.

Turner, S.P., Farnworth, M.J., Mendl, M., Erhard, H.W., Lawrence, A.B., 2011. Evidence for

residence-induced enhancement of aggressiveness in the non-territorial pig. Appl. Anim.

Behav. Sci. 130, 10-19.

Turner, S.P., Roehe, R., D'Eath, R.B., Ison, S.H., Farish, M., Jack, M.C., Lundeheim, N.,

Rydhmer, L., Lawrence, A.B., 2009. Genetic validation of postmixing skin injuries in

pigs as an indicator of aggressiveness and the relationship with injuries under more

stable social conditions. J. Anim. Sci. 87, 3076-3082.

Turner, S.P., Roehe, R., Mekkawy, W., Farnworth, M.J., Knap, P.W., Lawrence, A.B., 2008.

Bayesian analysis of genetic associations of skin lesions and behavioural traits to

identify genetic components of individual aggressiveness in pigs. Behav. Genet. 38, 67-

75.

43

CHAPTER THREE

Genetic analysis of the individual pig behaviour in backtests and

human approach tests




Kiel, Germany



Submitted for Publication in Applied Animal Behaviour Science

44

Abstract

The most recent development in pig production has focused increasingly on the well-being of

the individual pig and animal-friendly housing conditions i.e. the launch of the group housing of

sows in the EU. Concerning this, however, in every commercial farm production, standard

procedures are undertaken (i.e. mixing, iron injections, vaccinations) which might be stressful to

the animals and thus have an impact on their health and welfare. Therefore, there is a need to

assess individual pig behaviour in such stressful situations and furthermore to take into

consideration differences between several age levels. Hence, pigs were evaluated for their

response in two standardised stress situations - the backtest and the human approach test. The

data were collected on one research farm using the races of German Landrace, Large White and

crossbred pigs. The backtest (n = 1,382) was performed on the 12th and 19th day of life and the

number of escape attempts (NEA), the duration of escape attempts (DEA) and the latency to the

first escape attempt (LEA) were recorded. Additionally, the human approach test was performed

four times with weaned pigs (n = 1,317) and once with gilts (n = 272) while recording the

latency trait (LC) of the pigs to touch the human. The heritabilities of the different traits were

estimated univariately and correlations between all observed traits were obtained from bivariate

analyses with the average information-restricted, maximum-likelihood procedure as

implemented in the DMU software package. The random litter effect had the largest impact on

the LEA backtest trait (15 %). Smaller values were obtained for NEA and DEA. The LEA

backtest trait and the LC trait of the human approach test of weaned pigs and gilts were not

influenced by litter effect. The highest heritability were estimated for LEA (h² = 0.29) and NEA

(h² = 0.19), followed by DEA (h² = 0.10). The heritability of the human LC approach test trait of

weaned pigs on the same level with h² = 0.20. However, the heritability of the LC of gilts was

low (h² = 0.03) but the estimation provide no reliable values due to the small number of gilts.

The genetic correlations between LEA and DEA were very high (rp = -0.88). Also, the first and

second backtests for all traits were highly genetically correlated (rp = 0.69 to 0.90). This means

that the traits and the first and second backtests shared the same genetic base. Therefore,

performing just one backtest is sufficient for practical breeding purposes. The genetic

correlations between four LC human approach test traits of the weaned pigs were very high

(rp = 0.65 to 0.87) especially between consecutive tests. Hence, under practical conditions, the

performance of one human approach test might be sufficient since the behaviour shown in all

the human approach tests with weaned pigs depended on the same genetic base. The genetic

correlations between backtest traits and human approach test traits of weaned pigs and gilts were

45

very low, which indicates that both tests partly measure different behavioural patterns and that

the reactions of the pigs in the tests were not related.

Keywords: pig, behaviour, heritability, backtest, human approach test

46

Introduction

Commercial pig production includes routinely stressful situations such as weaning, cross-

fostering or in general the mixing of unacquainted pigs after the suckling period, in growing

herds or in the breeding area for the pigs (Ismayilova et al., 2013). Hence, different stressors can

appear in the life of the pigs. These stressful situations, which effect the animals systematically,

influence physiological, neuroendocrinological and behavioural changes in pigs (van Erp-van

der Kooij et al., 2000). Murani et al. (2010) stated that these differences in behaviour are based

on the brain and transmitted by the neuroendocrine system of the individual animal and are

therefore characteristic for the individual pig. The reaction of the pigs to such stressful,

challenging situations and effects is called coping (Koolhaas et al., 1999). An increase in stress

is often related to a decrease in animal welfare and also in performance and reproduction traits

(Tan et al., 1991; Stookey and Gonyou, 1994). The reaction or the coping ability towards these

stressors could be measured using behavioural tests. The backtest based on Hessing et al. (1993)

measures the reaction of the pig to the stress situation excluded from social effect of the pen

mates. Due to the development to stables and groups with a large numbers of animals and the

automated monitoring of husbandry conditions, there is a decrease in animal-human interactions

(Rushen et al., 1999) and an increase in fear towards humans. Hence, the second behavioural

test used in this study was the human approach test, in order to reveal individual differences in

social group situations (Thodberg et al., 1999). These two tests are easy to perform with a large

number of animals under standardised conditions. Both tests were indicated as measurements of

the individual pig to cope with stressful situations (e.g. Hessing et al., 1993; Hessing et al.,

1994; Giroux et al., 2000; Ruis et al., 2000; Ruis et al., 2001; van Erp-van der Kooij et al., 2001;

van Erp-van der Kooij et al., 2002; Janczak et al., 2003; Bolhuis et al., 2005). With the

knowledge of these reactions of the individual pigs, improvements in housing conditions or the

selection of pigs are possible. Therefore, the genetic aspects of the backtest and the human

approach test are needed to understand the individual stress reactions of the pigs. Cassady

(2007) estimated on moderate repeatabilities within two backtests on the basis of 150 pigs.

These results were in accordance with Velie et al. (2009). Both sources performed the backtest

twice but Velie et al. (2009) used a larger sample size of piglets (n = 457). The author also

estimated moderate to high heritabilities of backtest traits of 0.31 to 0.53. Moreover, Rohrer et

al. (2013) estimated heritabilities of one backtest with 975 pigs at age of weaning. But these

values were on a lower level with h² = 0.15 to 0.19. Hence in literature, the values of the

heritabilities of the backtests are not consistent between studies. Regarding the human approach

test, Hemsworth et al. (1990) estimate a heritability of 0.38 in a five-minute human test for the

47

individual pig without conspecifics in the test area. In contrast to this, Velie et al (2009) found

that a human approach test in growing pigs was not heritable. The authors also did this test with

a duration of five minutes while all pigs in the pen were tested simultaneously. The relation

between the backtest and human approach test was investigated (Ruis et al., 2001; van Erp-van

der Kooij et al., 2002; Cassady, 2007). Genetic relations between the backtest and human

approach test were less documented. Therefore, the heritabilities of these two tests and the

genetic background of the relations between the tests were inconsistent but necessary to evaluate

the use of these tests in further studies.

Hence, the aim of the present study was to examine the heritabilities of the backtest and social

human approach test traits at different ages with standardised experimental conditions to

examine the less documented ontogenetic effect especially in a social human approach test. In

addition, phenotypic and genetic correlations were analysed between the same tests at different

times and across the different behavioural tests. Genetic relations especially between the

backtest and human approach test should indicate whether the behaviour of such social and non-

social stressful tests had the same genetic base.


Animals and housing

Data were recorded on the “Hohenschulen” research farm of the Institute of Animal Breeding

and Husbandry of the University Kiel (Germany) from December 2010 till August 2012.

Purebred and crossbred pigs of the German Landrace (DL) and Large White breeds were used in

the investigation. The piglets from 139 litters (16 sows per batch) were kept in farrowing pens

for the suckling period for 26 days. The conventional farrowing stable consisted of four

compartments each with eight pens (2.2 m x 1.7 m) with a tiled and metal base floor with no

substrate. A piglet feeder was open to the piglets from the first week after farrowing. The sow of

the piglets received commercial lactating feed in accordance with the German norm (GfE,

2006). Water was available through nipple drinkers for non-stop use. At the first days of age,

each live born piglet was individually marked and weighed (average weight 1.54 kg). Within the

first three days the piglets were cross-fostered to standardise the litter sizes of all sows in the

batch. Male piglets were castrated.

After the suckling period, the pigs were weighed individually again (average weight 8.8 kg),

weaned and re-housed in flatdeck pens. The four compartments consisted of 10 pens each. One

pen measured 2.05 x 1.36 m and had a concrete and metal base floor with no substrate. In each

pen, two nipple drinkers were accessible for non-stop use. The pigs in flatdeck were fed ad

48

libitum with solid pellet feed (GfE, 2006). The temperature in the compartment was pre-set to a

minimum temperature of 24°C. The pigs were re-mixed and sorted by the smallest level of

familiarity and by nearly equal weight after weaning. Hence, 8 to 10 piglets were housed in each

pen. The pigs stayed in the flatdeck pens for 44 days.

The growing pigs were re-mixed in groups of 20 to 25 animals. The pens had a half-slatted and

half-solid floor (3.25 x 2.40 m). Water for non-stop use was accessible through nipple drinkers.

The growing pigs were fed by automatic mash feeding machine a with a commercial diet in

conformity with the German norm (GfE, 2006). The temperature in the growing stable was

maintained at 22°C. Here again, the pigs were sorted by the smallest level of familiarity and by

nearly equal body size at this time. A maximum of two pigs already acquainted with each other

from the flatdeck pens were housed together.

The mixing and housing of gilts in groups of 17 to 28 sows in the arena pen was in the 22nd

week of age. The weight of the gilts at this time was also recorded. The arena pen had a

dimension of 7.2 x 5.4 m and a half-slatted and half-solid floor. The gilts were fed by automatic

mash feeding machine with gilt feed according to GfE (2006). Water was provided through

nipple drinkers. The gilts were sorted by the smallest level of familiarity which means a

maximum of five out of all pen mates were already acquainted from the pens in the growing

stable.

Backtest

At the age of 12 and 19 days, all piglets (n=1,382) were subjected to a backtest. This test was

performed in the home compartment of the piglets. The animals were put on their back in a

special y-shaped device while each piglet was individually taken out of the pen and tested. After

the test the piglet was replaced and the backtest was performed with the next pen mate. The

experimenter held the piglet loosely with his left hand and restrained it in this supine position

for one minute. The test began when the piglet lay still. During this time, the number of escape

attempts (NEA), the latency to the first escape attempt (LEA) and the duration of all escape

attempts (DEA) were recorded.

Human approach test

The human approach test was performed with pigs which had also performed the backtest. This

test took place four times (at age of 6, 7, 8 and 9 weeks) in the flatdeck (n = 1,317) and one time

(22 weeks of age) with gilts (n = 272). The person stood still in front of the pen for one minute.

During the test time, the experimenter recorded which pigs made physical contact with the

person and noted the latency of the pigs to touch the stockperson (LC).

49


The characteristics of the backtest traits number of escape attempts (NEA), duration of all

escape attempts (DEA) and latency to the first escape attempt (LEA) and the characteristics of

the human approach test latencies (LC) of weaned pigs and gilts are shown in Table 1. The

results indicate that all traits were not normally distributed and after examination of the fitting

distributions, a poisson distribution was chosen for the analysis of NEA. DEA was separated

into classes of 10 seconds (class1: 0 s; class 2: 0 – 10 s; class 3: 11 – 20 s; class 4: 21 – 30 s;

class 5: 31 - 60 s) and the analysis was also performed with an underlying poisson distribution.

The third recorded trait in backtest, LEA, was separated into binary data (0: struggle, 1: no

struggle) and analysed with threshold models. The latencies (LC), which were recorded during

the human approach tests of the two different ages (weaned pigs and gilts), were also separated

into two classes (0: touched the person; 1: did not touch the person).

The fitting of the models was evaluated using the AICC “Akaike’s information criterion

corrected” (Hurvich and Tsai, 1989) and the BIC “Bayesian information criterion” (Schwarz,

1978) of the SAS procedure Glimmix (SAS, 2008). All fixed effects were added stepwise in the

model. Finally, the models with the smallest AICC and BIC were chosen for the analysis.

Effects which had no impact on the model fitting (e.g. parity of the sow, number of pen mates,

cross fostering, pen) were not used in the final models.

The models of all backtest traits (NEA, DEA and LEA) included the systematic effects of batch

(group 1 to 10) and test number (test 1 and test 2). The birth weight of the piglets was included

as a linear covariate. The LC traits of the human approach test with weaned pigs used the fixed

effects of batch (group 1 to 10), test number (test 1 to test 4), gender (male and female) and the

distance of the pen to the door (pens in front of the compartment: front, pens in back of the

compartment: back) in the models. The weight of the pigs had no impact and was not used in

this analysis.

The model analysing the LC trait with gilts included one systematic effect (batch: group 5 -10).

Additionally, the weight at time of testing in the human approach test was used as a covariate.

On the basis of the selected models, genetic parameters were estimated by GLMM analyses was

performed using the average information (AI) restricted, maximum-likelihood procedure

implemented in the DMU program package (Madsen and Jensen, 2000). For the count variables

DEA and NEA a Poisson distribution was assumed. The link function between the linear

predictor and the observations was a log link. For the binomial distributed LEA and LC

threshold models were defined. The adopted probit link modelling the probability that the pig

50

does not struggled (LEA) or does not contact the stockperson (LC) is given by the inverse

normal cumulative density function. The residual variance was fixed to a value of 1.

The assessment of the correlations used the EM-REML procedure. The pedigree contained

information of two generations backwards with 104 sows (average 13 piglets) and 60 boars

(average 23 piglets). The estimation of heritabilities and correlations was performed with an

animal model. In addition to the above-described systematic effects and covariates, the genetic

estimations included as random effects the animal, the litter effect and the permanent

environmental effect. The separated analysis of the specific tests with only one observation per

animal was carried out without using the test number and the permanent environmental effect.

The correlations between the different traits were analysed with bivariate estimations and an

underlying normal distribution of all traits because the estimations otherwise did not reach

convergence. Mäntysaari et al. (1991) stated that linear correlations and correlations estimated

with an underlying distribution showed no difference.

Due to the fact that the estimations of genetic correlations as poisson estimations (NEA, DEA)

did not reach convergence for the separated analysis of backtests (first and second backtest),

these values were estimated with linear models.

Table 1: Median, minimum (min), maximum (max) and number of observations (n) for the

backtest traits and human approach test traits in weaned pigs and gilts.

Trait N Unit Median Min Max

Backtest

Number of escape attempts (NEA) 2,766 number 2 0 7

NEA 1st Backtest 1,382 number 2 0 7

NEA 2nd Backtest 1,382 number 2 0 7

Duration of all escape attempts (DEA) 2,766 s 10 0 60

DEA 1st Backtest 1,382 s 11 0 60

DEA 2nd Backtest 1,382 s 9 0 60

Latency to first escape attempt (LEA) 2,766 s 18 1 60

LEA 1st Backtest 1,382 s 14 1 60

LEA 2nd Backtest 1,382 s 22 1 60

Human approach test

Latency of weaned pigs (LC) 5,268 s 60 0 60

LC 1st Test 1,317 s 60 2 60

LC 2nd Test 1,317 s 60 0 60

LC 3rd Test 1,317 s 60 1 60

LC 4th Test 1,317 s 60 0 60

Latency of gilts (LC) 272 s 60 2 60

51

Results

Random effects

The analysis of the systematic effect of the backtest revealed higher reactions of the pigs in the

first backtest (p<0.05). Furthermore, lighter pigs struggled more during the backtests than

heavier pigs (p<0.05). In the human approach tests with weaned pigs, female pigs had lower

latencies than males. Additionally, the analysis of the systematic effects showed that the LC of

weaned pigs decreased with higher test numbers. Moreover, weaned pigs which were housed in

pens closer to the door of the compartment had smaller LC in the human approach tests. For

more information on the systematic effects of the backtest and human approach test traits, see

Scheffler et al. (2013).

The analysis of the backtest showed that the random litter effect explained 1 % of the NEA trait,

3 % of DEA and 15 % of LEA. The separated analysis of NEA showed a higher influence of the

litter component in the second backtest (19 %) than in the first one (2 %) in the linear

estimations. The portion of litter variance of DEA in the first and second tests was almost equal

(9 and 11 %). The influence of the litter effect, regarding the LEA trait of both backtests

separately was much higher than for the other backtest traits with, 18 % in the first test and 40 %

in the second backtest.

The portions of litter variance in the human approach tests were between 1 to 6 % for the

separated analysis of the human approach tests in both age groups. Only in the third test with

weaned pigs could influences of the litter be obtained (16 %).

Heritabilities

For the heritability of the backtest traits, the highest values were estimated for the LEA trait with

h² = 0.29, followed by NEA with h² = 0.19 and DEA with h² = 0.10 (Table 2). The heritabilities

of NEA in the first backtest and second backtest (linear estimation) were comparable with

h² = 0.24 and h² = 0.20, respectively. Moreover, the heritabilities of DEA in the separated

estimation (linear) were also on one level (h² = 0.11 and h² = 0.16). In the separated analysis of

the backtests, the LEA trait showed heritabilities with higher values in the first (h² = 0.15) than

in the second backtest (h² = 0.03).

The overall heritability of the human approach test of the weaned pigs was h² = 0.20. The values

decreased from the first human approach test (h² = 0.33) to the fourth test (h² = 0.10) with regard

to the separated estimations of each human approach test with weaned pigs. The heritability of

LC in gilts was almost zero with h² = 0.03 but with a small number of gilts in the estimation.

52

Table 2: Heritabilities and standard errors of backtest traits (number of escape attempts: NEA;

duration of escape attempts: DEA; latency to first escape attempt: LEA) and human approach

test traits (latency: LC) of suckling pigs, weaned pigs and gilts.

δani δper δlit δe h² SE

Backtest

NEA (total) 0.14 0.01 0.01 0.60 0.191 0.05

NEA 1st Backtest 0.38 - 0.03 1.10 0.24² 0.08

NEA 2nd Backtest 0.33 - 0.12 1.15 0.20² 0.11

DEA (total) 0.03 0.01 0.01 0.24 0.101 0.05

DEA 1st Backtest 0.09 - 0.10 0.68 0.11² 0.10

DEA 2nd Backtest 0.14 - 0.08 0.68 0.16² 0.11

LEA (total) 0.55 0.08 0.29 1.00 0.29³ 0.17

LEA 1st Backtest 0.22 - 0.27 1.00 0.15³ 0.22

LEA 2nd Backtest 0.05 - 0.71 1.00 0.03³ 0.22

Human approach test

LC Weaned pigs (total) 0.66 1.50 0.12 1.00 0.20³ 0.06

LC 1st Test 0.51 - 0.02 1.00 0.33³ 0.14

LC 2nd Test 0.42 - 0.04 1.00 0.29³ 0.12

LC 3rd Test 0.24 - 0.23 1.00 0.16³ 0.12

LC 4th Test 0.11 - 0.04 1.00 0.10³ 0.06

LC Gilts 0.06 - 0.74 1.00 0.03³ 0.19

1 Poisson model; ²Linear model; ³Threshold model δani: additive genetic variance, δper: permanent environmental variance,

δlit: litter variance, δe: residual variance, h²: heritability

53

Correlations within behavioural tests

Due to the small number of animals, the estimation of the relations between NEA and DEA and

NEA and LEA did not converge. The phenotypic and genetic correlations between the traits

DEA and LEA were very high at rp = -0.65 and rg = -0.88 (Figure 1). The phenotypic

correlations between the first and the second backtests were moderate for NEA (rp = 0.34) and

DEA (rp = 0.36) but lower for LEA (rp = 0.19). In contrast to the phenotypic correlations, the

genetic relations for NEA, DEA and LEA between both backtests were much higher with values

of rg = 0.90, rg = 0.89 and rg = 0.69, respectively.

The phenotypic correlations of the human approach tests between weaned pigs and gilts were

small at rp = 0.09 (not shown). However, the genetic correlation was moderate at rg = 0.52. The

phenotypic relations between each human approach test with weaned pigs were on a medium

level (rp = 0.23 – 0.53) (Figure 1). The genetic correlations between these tests ranged from

rg = 0.65 to 0.87. Furthermore, increasing correlations could be obtained with a smaller time

difference between two human approach tests.

Figure 1: Phenotypic (rp) and genetic (rg) correlations between the first (1) and second (2)

backtests for the number of escape attempts (NEA), duration of escape attempts (DEA) and

latency to the first escape attempts (LEA) and between the latency (LC) of the four ( 1 to 4)

human approach test with weaned pigs.

54

Correlations between behavioural tests

The relations between the backtest and the human approach tests at different age levels are

shown in Table 5. The phenotypic correlations between these two behavioural tests were small.

Furthermore, the genetic correlations between human approach tests with weaned pigs and the

backtests traits were small (rg = 0.04 – 0.21). Values between rg = -0.12 for NEA, rg = -0.19 for

DEA and rg = 0.21 for LEA could be obtained for the genetic correlations between the human

approach test with gilts and the backtest traits.

Table 3: Phenotypic and genetic correlations of the backtest traits number of escape attempts

(NEA), duration of escape attempts (DEA) and latency to the first escape attempt (LEA) and the

human approach test trait latency (LC) at different ages.

Discussion

Random effects

The variance of the litter effect was smaller than the additive genetic effect for all backtest traits.

This was in contrast to Rohrer at al. (2013), who illustrated that the litter variance was on the

same level as the additive genetic variance. Roehe et al. (2009) stated that the litter effect

represents the maternal genetic and maternal environmental effect. In order to assess the

maternal environmental impact in the present study, a separation test was performed with the

dams of the piglets (according to Hellbrügge et al., 2009; Scheffler and Krieter, 2013). The

reaction of changes of the sows’ body position was recorded at separation from their piglets. No

evidence was found that the reaction of the sow in the separation tests influenced the behaviour

of the piglets in the backtest.

Furthermore, the influence of the litter size on the reactions in the backtest was tested. It was

found that the litter size had no impact on the backtest behaviour of the pigs either. Hence,

further investigations are needed to qualify the influence of the dam within the litter effect. In

the present study, the human approach tests were not affected by the litter effect. This effect is

Backtest NEA DEA LEA

rp rg rp rg rp rg

Hu

man

ap

pro

ach

te

st

LC Weaned pigs -0.01 0.08 ± 0.31 -0.03 0.04 ± 0.37 0.03 0.21 ± 0.37

LC Gilts -0.02 -0.12 ± 0.38 -0.02 -0.19 ± 0.46 0.04 0.21 ± 0.43

55

also not discussed in the literature. Thus, experiences in early age had no effect on the behaviour

of such a test situation.

Heritabilities

The heritabilities of the backtest traits showed that the NEA and LEA traits were most heritable.

The DEA trait showed lower heritabilities. Velie et al. (2009) estimated values for the number of

struggles of h² = 0.54 and for the total time spent struggling of h² = 0.49. In contrast to the

present backtest, this test was not carried out on a defined day of age of the pigs. The test day

took place between the 7th and 14th day of life with a difference of one week between the two

backtests. The heritabilities of Velie et al. (2009) were higher than in the present study which

could be explained by the non-standardised test conditions i.e. in this case, the age of the pigs

during the backtest. Other investigations estimated heritabilities of h² = 0.14 to h² = 0.15 for the

backtest traits (Rohrer et al., 2013). These values were closer to the results of the current study,

especially closer to the separated analysis of the first and second backtests. Rohrer et al. (2013)

used only one backtest and performed the test after weaning at the 24th day of life. In this case,

direct behavioural influences on the reaction of the pigs in the backtests caused by the dams

could be eliminated. Hence, the results of the backtest traits in the study of Rohrer et al. (2013)

and the results of the present study showed that environmental effects such as the reaction of the

dams had less influence on the behaviour in the backtest. However, standard conditions such as

a defined age of the pigs might have an impact on the heritabilities.

Regarded separately, the values of the heritabilities in the first and second backtests for the NEA

trait in the present study were on the same level. Velie et al. (2009) also estimated similar

heritabilities for the separated analysis of the tests (h² = 0.38 and h² = 0.40, respectively).

However, in contrast to the present results these heritabilities were on a higher level. The

heritabilities of DEA and LEA showed smaller values than for the NEA trait. These findings

were in accordance to Rohrer et al. (2013).

The heritability of the LC human approach test trait with weaned pigs was on the same level

than that of the backtest traits NEA and LEA. Regarded separately, the heritabilities of LC of

the human approach test with weaned pigs, were less heritable the more often the test was

performed due to the better fitting of the threshold model to the data because the more often the

test was performed the more animals touched the stockperson. Threshold models especially

provide reliable estimations if the frequency manifestation of the traits is low. In contrast to

these results of the heritabilities, Velie et al. (2009) stated that a human approach test with

finishing pigs was not heritable. This could be a hint that the genetic determination of the

reaction in this test varied with the age of the animals.

56

The heritability of the human approach test with gilts was very low (h² = 0.03). Hellbrügge et al.

(2009) estimated a heritability of the human approach test with gilts of h² = 0.09 and therefore,

on a comparable level to the heritability of the human approach test with weaned pigs.

Furthermore, the results of Hemsworth et al. (1990) showed that the estimated heritability for

gilts in a human approach test (h² = 0.38). Therefore, the heritability of the gilts is small due to

the number of gilts in the present study which was too small for a reliable estimation of

heritability.

Correlations within behavioural tests

In literature, correlations between all backtests traits were in general high (Cassady, 2007; Velie

et al., 2009; Spake et al., 2012). In the present study, these estimations did not converge due to a

too small number of observations. The high correlations between DEA and LEA of the present

study were in accordance to Rohrer et al. (2013), who estimated correlations of rp = -0.70 and

rg = -0.91.

The correlations between the first and second backtests showed low to moderate phenotypic

correlations but very high genetic correlations. Thus, the behaviour in the first and second

backtests is highly related and therefore one backtest might be sufficient especially in practical

breeding programs concerning time effort and standardisation of the backtest under less

experimental conditions.

The smaller the time difference between the human approach tests was, the higher the

phenotypic and genetic correlations. The more often the test was carried out, the less fearful and

stressed the animals were. Also Janczak et al. (2003) supported this by their statement that fear

towards humans decreased with the age of the animals. Additionally, Marchant-Forde et al.

(2003) stated that the approach of pigs to humans generally depends on the number of contacts

in life. This could be emphasised by the correlations of the human approach tests with weaned

pigs and gilts in the present study. Here, the relations were small especially the phenotypic ones

(rp = 0.09; rg = 0.52). Due to the limited data set for the gilts, the genetic correlations gave only

an impression and further research is needed to confirm these results.

57

Correlations between behavioural tests

The phenotypic and genetic correlations between the NEA, DEA and LEA backtest traits and

the LC human approach tests traits of weaned pigs and gilts was very low or almost zero.

Comparable correlations were stated in Velie et al. (2009). They also estimated phenotypic and

genetic correlations between DEA and NEA and the human approach test traits with growing

pigs. Here, no relationship between these traits could be obtained. Hence, the backtest and the

human approach test measure different behaviour or different characteristics of the personality

of the individual pig (van Erp-van der Kooij et al., 2002). Also Janczak et al. (2003) found no

significant phenotypic correlations between the latency human approach test trait and the

duration of struggling in an immobility test, which is comparable with the backtest. Thus, no

consistency between the tests could be obtained.

Conclusion

The analyses of the behavioural tests showed that the behaviour of the pigs in the backtest was

more greatly influenced by the litter effect than the behaviour of the pigs in the human approach

test. To clarify the litter effect, described as the maternal genetic and maternal environmental

effect, further investigations are needed. For practical application it is sufficient to record only

one backtest. Here, particularly the number of escape attempts should be recorded due to the

highest heritabilities for this trait and furthermore due to the high genetic correlation of all traits

between first and second backtest for all traits. The heritability of the human approach test with

weaned pigs was higher than for gilts. The relation between the tests with both age groups was

low. Hence, partly different behavioural patterns were measured in weaned pigs and gilts.

However, due to the small number of gilts in the present study, the relations between the age

levels and also the heritabilities have to be investigated in further studies with a larger number

of animals. The low phenotypic and genetic correlations between the backtest and the human

approach tests indicate that both tests describe different behavioural patterns.

58

References



1091.





Giroux, S., Martineau, G.-P., Robert, S., 2000. Relationships between individual behavioural

traits and post-weaning growth in segregated early-weaned piglets. Applied Animal

Behaviour Science 70, 41-48.








Sci. 37, 285-295.

Hessing, M.J.C., Hagelsø¸, A.M., Schouten, W.G.P., Wiepkema, P.R., Van Beek, J.A.M., 1994.

Individual behavioral and physiological strategies in pigs. Physiol. Behav. 55, 39-46.












59




Madsen, P., Jensen, J., 2000. A user's guide to DMU. A package for analysing multivariate

mixed models. Version 6, 1-33.

Mäntysaari, E.A., Quaas, R.L., Gröhn, Y.T., 1991. Simulation study on covariance component

estimation for two binary traits in an underlying continuous scale. Journal of dairy

science 74, 580-591.




Murani, E., Ponsuksili, S., D'Eath, R., Turner, S., Kurt, E., Evans, G., Tholking, L., Klont, R.,

Foury, A., Mormede, P., Wimmers, K., 2010. Association of HPA axis-related genetic

variation with stress reactivity and aggressive behaviour in pigs. BMC Genetics 11, 74.

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. Livest. Sci. 121, 173-181.

Rohrer, G.A., Brown-Brandl, T., Rempel, L.A., Schneider, J.F., Holl, J., 2013. Genetic analysis

of behavior traits in swine production. Livest. Sci.

http://dx.doi.org/10.1016/j.livsci.2013.07.002.




Ruis, M.A.W., te Brake, J.H.A., Engel, B., Buist, W.G., Blokhuis, H.J., Koolhaas, J.M., 2001.

Adaptation to social isolation: Acute and long-term stress responses of growing gilts

with different coping characteristics. Physiol. Behav. 73, 541-551.

Rushen, J., Taylor, A.A., de PassillÃ©, A.M., 1999. Domestic animals' fear of humans and its

effect on their welfare. Applied Animal Behaviour Science 65, 285-303.


Scheffler, K., Krieter, J., 2013. Analysing the relation between agonistic behaviour, behavioural

tests and litter size. . Unpublished Results.

Scheffler, K., Traulsen, I., Krieter, J., 2013. Assessment of individual behaviour using backtest

and human approach test at different age levels. Submitted to Livest. Sci.

60



pigs. Appl. Anim. Behav. Sci. 140, 146-153.







van Erp-van der Kooij, E., Kuijpers, A.H., Schrama, J.W., Ekkel, E.D., Tielen, M.J.M., 2000.

Individual behavioural characteristics in pigs and their impact on production. Applied

Animal Behaviour Science 66, 171-185.

van Erp-van der Kooij, E., Kuijpers, A.H., van Eerdenburg, F.J.C.M., Tielen, M.J.M., 2001. A

note on the influence of starting position, time of testing and test order on the backtest in

pigs. Applied Animal Behaviour Science 73, 263-266.




75, 293-305.



61

CHAPTER FOUR

Relationship between behavioural tests and agonistic interactions

at different age levels in pigs




Kiel, Germany



62

Abstract

Fighting among pigs is a normal behavioural pattern to establish a stable rank order. Enhanced

aggressive behaviour in pigs in groups lead to increasing stress and injuries especially in mixing

situations used as a common procedure in modern pig production systems. In such systems, it is

usually not possible to avoid re-housing with unacquainted conspecifics. Hence, due to the

lavish analysis of direct or video observations of the agonistic interactions in such mixing

situations, there is a necessity to receive easy measurable and practical indicators for predicting

individual agonistic behaviour. Possible indicators are standardised behavioural tests such as the

backtest and the human approach test. The backtest was performed twice. In each test, the pigs

were laid on their backs and held loosely for one minute (n = 1,382). The number of escape

attempts (NEA) was recorded. In addition to this test, a human approach test was performed four

times with weaned pigs (n = 1,318) and once with gilts (n = 272). Here, the stockperson

recorded the latency of the pigs to approach and touch the person, i.e. the latency count (LC).

The agonistic interactions were recorded in a video observation period of 17 h while the traits

number of fights (NF) and number of initiated fights (IF) were recorded in mixtures of weaned

pigs (n = 1,111), growing pigs (n = 446) and gilts (n = 279). The estimations of heritabilities as

well as phenotypic and genetic correlations between these different traits were carried out with

animal models in bivariate analyses. The IF trait of weaned pigs and NEA were slightly

positively correlated (rg = 0.18). Pigs which initiated more fights after weaning, described in

literature as dominant pigs, had more escape attempts in the backtests. However, there were

negative genetic correlations between the agonistic interactions traits NF and IF traits and the

NEA backtest trait of growing pigs (rg = -0.14 and rg = -0.28). The genetic relation between the

agonistic NF and IF traits of weaned pigs and the human approach test LC trait of weaned pigs

were on a medium level (rg = -0.50 and rg = -0.45). The genetic correlations between IF and NF

of growing pigs and gilts and the human approach test LC trait in weaned pigs were lower but

also negatively correlated. Hence, pigs with more NF and IF in mixing had shorter latencies

during the human approach tests. Concluding these facts, the backtest and the human approach

test might be able to predict the agonistic behaviour of pigs in mixing situations. Nevertheless,

the reliability of the predictions of the behavioural tests depends on the age of the pigs at mixing

and the previous experiences of these animals.

Keywords: pig, behaviour, backtest, human approach test, aggression

63

Introduction

Focussing on animal welfare aspects, individual pig behaviour in standard situations of

commercial pig production is becoming more and more important. In the daily routine work of

pig farms, the mixing of unacquainted pigs is a common occurrence (Ismayilova et al., 2013).

To establish a stable rank order in the group, the pigs fight each other. The stable hierarchy is

usually achieved at the third day after re-housing and is needed to prevent permanent stress in

the groups (Meese and Ewbank, 1973). However, the specific fighting behaviour of pigs shows

a large variation between individuals. Here, animals with enhanced aggressive behaviour can

influence the health, welfare as well as the weight gain of especially low-ranking pigs (Tan et

al., 1991; Tuchscherer and Manteuffel, 2000). Former investigations had estimated moderate

heritabilities of aggressive and submissive traits (Løvendahl et al., 2005; Turner et al., 2008;

Turner et al., 2009b; Stukenborg et al., 2012). Therefore, knowledge regarding the behaviour of

pigs in agonistic interactions gives the opportunity for breeding of calm and less aggressive

animals (Kanis et al., 2005; D'Eath et al., 2009). However, due to the time-consuming and lavish

observation methods mostly by video techniques, agonistic interactions at different ages are not

easy to measure. Hence, selective breeding against increased aggressive animals might be

possible with easy measurable indicators. Turner et al. (2009a) described as a possible indicator

the lesion score, which counts the number of scratches on the body of a pig, distributed into

scores for each pig. The lesion score showed heritabilities of h² = 0.19 - 0.43 for the different

body regions of the pig. However, Stukenborg et al. (2010) stated that the lesion score (recorded

with the modified method of Turner et al.(2009a) concerning the counting of scratches in a

directly assigned score on a five-score scale from no wounds and scratches to many deep

wounds and scratches) could not be used to evaluate agonistic behaviour especially in growing

pigs. In literature, different behavioural tests are described which could also be used as easy

measurable indicators to predict agonistic interactions such as the non-social backtest and the

social human approach test (e.g. Hemsworth et al., 1990; Hessing et al., 1993; Thodberg et al.,

1999; Ruis et al., 2000b). Therefore, the present study focussed on these two behavioural tests.

In the backtest, the animals were put on their backs and the number of escape attempts was

recorded (after Hessing et al., 1993). The human approach test measures the latency of the pigs

to approach and touch the stockperson (Thodberg et al., 1999). According to Rohrer et al.

(2013), Cassady (2007) and Velie et al. (2009), both tests show also a large variation between

the individuals, similar to agonistic behaviour, with moderate heritabilities also in different age

groups . Moreover, Melotti et al. (2011) found that highly reactive pigs in the backtest spent

more time in self–initiated fights without regarding the submissive signals sent by their

64

opponents. Furthermore, the results of Brown et al. (2009) showed that pigs with shorter

latencies in the human approach tests were more aggressive at mixing. However, the results in

literature concerning aggressive behaviour in relation to the backtests or the human approach

test showed divergent results. On the one hand, investigations have shown that there is a relation

between these non-social and social tests and aggressive behaviour (e.g. Hessing et al., 1993;

Thodberg et al., 1999; Ruis et al., 2000b; O'Connell et al., 2004; Bolhuis et al., 2005). On the

other hand, some studies have obtained no relation and mentioned that these behavioural

patterns are completely different between the agonistic interactions and the behavioural tests

(Lawrence et al., 1991; Forkman et al., 1995; Jensen et al., 1995; Spoolder et al., 1996; D'Eath

and Burn, 2002; Janczak et al., 2003). Although the backtest has a highly standardised test

procedure, a direct comparison between the results in literature is not possible due to the

different methods used to evaluate the aggressive behaviour of an individual pig (e.g. recorded

during a resident intruder test, a food competition test or by video observations) and the

different performances of the human approach tests. Hence, so far, an ontogenetic approach with

standardised behavioural tests and standardised recording of aggression with a high number of

animals has not been investigated. Therefore, the aim of the present study was to clarify the

questions if there are behavioural tests performed at different ages could predict the agonistic

behaviour of pigs in different mixing situations. Hence, in this study, the heritabilities of the

behavioural test traits and agonistic interactions traits as well as phenotypic and genetic

correlations between behavioural test traits and agonistic interaction traits at different ages were

estimated for implementation in selection strategies.


Animals and housing

The data collection was from December 2010 till August 2012 on the research farm

“Hohenschulen” of the Institute of Animal Breeding and Husbandry of the University Kiel

(Germany). The pigs were pure-bred and cross-bred animals of the breeds German Landrace

(DL) and German Edelschwein (DE). The piglets from 139 litters (16 sows per batch) were kept

in farrowing pens for the suckling period of 26 days postpartum. In the stable were four

compartments each with eight pens. These conventional farrowing pens (2.2 m x 1.7 m) had a

tiled and metal based floor with no substrate. In accordance to the German norm (GfE, 2006) the

lactating sow received a commercial lactating feed. Water was available through nipple

drinkers. From the first week after farrowing a piglet feeder was provided. Each live born piglet

was marked and weighted individually (average weight 1.54 kg) at the first day of life. The

65

piglets were cross-fostered for a standardisation of the litter size for each sow until the third day.

Cross-fostered piglets were the heaviest of the litter. All male piglets were castrated.

At weaning the pigs were again weighted individually (average weight 8.8 kg) and then housed

in flatdecks. There were four compartments with 10 pens each. The pens (2.05 x 1.36 m) had a

concrete and metal based floor with no substrate. Two nipple drinkers were available in each

pen. The pigs were fed ad libitum with solid pelleted feed in conformity with the German

standards (GfE, 2006). The room temperature was approximately 24°C. The pigs were re-mixed

and sorted by the smallest level of familiarity and by nearly equal weight. Eight to ten pigs were

housed in one pen and no pig knew another pig from the farrowing pens. The pigs stayed in the

flatdecks on average for 44 days.

After these weeks in the flatdeck, the pigs were re-mixed and re-housed in the growing stable in

groups of 20 to 25 animals. The pens (3.25 x 2.40 m) had a half-slatted and solid floor. Nipple

drinkers for non-stop water use were present. The growing pigs recieved a commercial diet by

mash automats according to standard (GfE, 2006). The temperature was 22°C. The pigs were

sorted by the smallest level of familiarity and by nearly equal body size. In the pens, maximal

two pen mates already knew each other from the previous pens.

In the 22th week of age, the gilts were re-mixed and housed in the pen in the breeding area

(arena pen) in groups of 17 to 28 sows. The pen had a dimension of 7.2 x 5.4 m and a half-

slatted and solid floor. The gilts were fed according to standards of the GfE (2006). Water was

assessable through nipple drinkers. All gilts were sorted by the smallest level of familiarity.

Hence, maximal five gilts knew each other from the growing pens.

Backtest

At the age of 12 and 19 days, all piglets (n=1,382) were subjected to a backtest. The test was

performed in the home compartment of the animals. The piglets were put on their backs in a

special y-shaped device. Each animal was individually taken out of the pen and tested. After the

test, the piglet was replaced and the test was performed with the next pen mate. The stockperson

held the piglet loosely with his left hand and restrained it in this supine position. The test began

when the piglet lay still and ended after the test time of one minute. During this time, the

number of escape attempts (NEA), the latency to the first escape attempt (LEA) and the duration

of all escape attempts (DEA) were recorded.

66

Human approach test

The human approach test was performed with pigs which had also been used in the backtest.

The human approach test was carried out four times (at age of 6, 7, 8 and 9 weeks) in the

flatdeck (n=1,317) and once (22 weeks of age) with gilts (n=272). The gilts were analysed in the

arena pen. During the test time of one minute the stockperson noted which pigs made physical

contact with him or her. Additionally, the experimenter recorded the latency to touch the

stockperson (LC).

Behavioural observations

The video observations started circa at 12:00 h immediately after re-housing and re-mixing in

the flatdeck, growing stable or arena pen and recorded the behaviour of the pigs for four days.

Stukenborg et al. (2010) stated that there was a declined rate of agonistic behaviour in the night

and the results of Meese and Ewbank (1973) showed that the fighting behaviour decreased

fundamentally after two days of observation. Therefore, the recording was interrupted in the

night (from 18:00 h to 07:00 h) and due to the high number of animals in the study, the period

used for the analysis was limited to 17 hours (day of re-housing: ca. 12:00 – 18:00 h; 2nd day:

07:00 – 18:00 h). The HeitelPlayer software (Xtralis Headquarter D-A-CH, HeiTel Digital

Video GmbH, Kiel, Germany) was used for the observation of the agonistic interactions. All

pigs of a pen received a unique number on their backs and their behaviour could be observed in

the whole pen. Data from 1,111 weaned pigs, 446 growing pigs and 279 gilts were used in the

statistical analyses.

All marked pigs in the flatdeck, growing stable or arena pen were analysed with the help of

videotapes. Recorded parameters were the start and end of the fight, the initiator or receiver and

the winner or loser of an agonistic interaction. If the aggressor/receiver or the winner/looser was

not clear, the fights were recorded with unclear starter/finisher or as stand-off fights. Thus, six

traits were obtained: number of fights (NF), duration of fights (DF), number of initiated fights

(IF), number of received fights (RF), number of fights won (FW) and number of fights lost (FL).

A fight was defined as a physical contact longer than one second with aggressive behaviour

initiated from one pig to another and which ended in the submissive behaviour of an involved

pig, i.e. the loser of the fight (Tuchscherer et al., 1998; Langbein and Puppe, 2004). ‘Head to

head knocks’, ‘head to body knocks’, ‘parallel/inverse parallel pressings’, ‘bitings’ or ‘physical

displacements’ were identified as agonistic behaviour (Puppe, 1998; Stukenborg et al., 2012;

Ismayilova et al., 2013). The submissive behaviour was defined as the cessation of a fight,

turning away, displacement from a location and fleeing (Tuchscherer et al., 1998; Langbein and

Puppe, 2004; Stukenborg et al., 2012). The video observations of the pigs in the flatdecks were

67

carried out with three different observers who had been trained with a video test sequence at the

beginning of the video analysis. The definition and identification of the agonistic behaviour was

tested with an unknown video sequence. The inter-observer agreement was higher than 90 %.

The growing pigs and gilts were observed by only one person.


The NEA backtest trait (number of escape attempts), the LC human approach test trait of

weaned pigs and gilts and the NF and IF traits (number of total fights and number of initiated

fights) were statistically analysed. These traits had been obtained in previous studies (Scheffler

et al., 2013a; Scheffler et al., 2013c) as the most suitable ones for the description of the pig

behaviour in the behavioural tests or the agonistic behaviour. Descriptive statistics of these data

are shown in Table 1. As the results indicate, none of the traits were normally distributed.

Therefore, the data were analysed regarding the underlying distribution.

NEA is a count variable following a poisson distribution. The latencies of the human approach

tests of two different ages (weaned pigs and gilts) were defined as binary traits (LC = 0: touched

the person; LC = 1: did not touch the person). A threshold model was specified for this

binomially distributed trait. The number of fights (NF) and the number of initiated fights (IF) of

the agonistic interactions were log-transformed (Y = loge (1+observation value)) for reducing

skewness and curtosis. After this transformation, the agonistic behavioural traits NF and IF were

approximately normally distributed, which was also be observed by the visual inspection of the

residual plots.

Model fit was evaluated by Akaike’s information criterion corrected (AICC) (Hurvich and Tsai,

1989) and the Bayesian information criterion (BIC) (Schwarz, 1978) implemented in the SAS

procedure GLIMMIX (SAS, 2008). The model which minimised the AICC and BIC was

superior and was chosen for further analyses. Effects which had no impact on the model fitting

were not used in the final models. On the basis of the selected models genetic parameters were

estimated using the average information (AI) restricted maximum likelihood method

implemented in the DMU program package (Madsen and Jensen, 2000). For NEA, the link

function between the linear predictor and the observations was a log link. A threshold model

was defined for the binomial distributed LC trait. The adopted probit link modelling the

probability that the pig does not contact the stockperson is given by the inverse normal

cumulative density function. The residual variance was fixed to a value of 1. The pedigree

contained information of two generations backwards with 104 sows (average 13 piglets) and 60

boars (average 23 piglets). The estimation of correlations between different traits was performed

with an animal model as bivariate analyses. Due to the fact that the genetic estimations for the

68

poisson distributed NEA and for the binomial distributed LC in the separated analyses did not

reach convergence, the correlations for the NEA and LC traits were assumed as normally

distributed traits. Mäntysaari et al. (1991) stated that the genetic correlation from binomial traits

is equal to the correlation of normally distributed variables.

The fixed and random effects and covariates used in the linear models of the NEA backtest trait,

the LC human approach test trait of weaned pig and gilts and the NF and IF agonistic interaction

traits of weaned pigs, growing pigs and gilts are shown in Table 2.

Table 1: Median, minimum (Min) and maximum (Max) of the row data of the NEA backtest

trait (number of escape attempts), the LC human approach test trait of weaned pigs and gilts and

the NF (number of fights) and IF (number of initiated fights) agonistic interaction traits of

weaned and growing pigs and gilts.

Trait Unit N Median Min Max

Backtest

Number of escape attempts (NEA) number 2,764 2 0 7

NEA 1st Backtest number 1,382 2 0 7

NEA 2nd Backtest number 1,382 2 0 7

Human approach test

Latency of weaned pigs (LC) s 5,268 60 0 60

LC 1st test s 1,317 60 2 60

LC 2nd test s 1,317 60 0 60

LC 3rd test s 1,317 60 1 60

LC 4th test s 1,317 60 0 60

Latency of gilts (LC) s 272 60 2 60

Agonistic interactions

Weaned pigs

Number of total fights (NF) number 1,111 15 1 116

Number of initiated fights (IF) number 778 5 0 68

Growing pigs

Number of total fights (NF) number 446 6 0 39


Gilts

Number of total fights (NF) number 279 4 0 32


Table 2: Fixed and random effects and covariates of the different models of the NEA backtest traits (number of escape attempts), the LC human

approach test traits (latency) of weaned pigs and gilts and the NF (number of fights) and IF (number of initiated fights) agonistic interaction traits of

weaned pigs, growing pigs and gilts.

Fixed effects Random effects

Batch Test number

Gender Pen category

Observer Pen Cross- fostering

Weight

Ani Lit Perm Env

Bac

k-

test

Number of escape attempts (NEA) � � � � � � NEA 1st backtest � � � � NEA 2nd backtest � � � �

Hu

man

ap

pro

ach

tes

t Latency of weaned pigs (LC) � � � � � � � LC 1st test � � � � � LC 2nd test � � � � � LC 3rd test � � � � � LC 4th test � � � � � Latency of gilts (LC) � � � �

Ago

nis

tic

inte

ract

ion

s

Weaned pigs Number of fights (NF) � � � � � � � Number of initiated fights (IF) � � � � � � � Growing pigs Number of fights (NF) � � � � � Number of initiated fights (IF) � � � � � Gilts Number of fights (NF) � � � �

Number of initiated fights (IF) � � � �

Batch: Backtest: 1 to 10 Human approach test: Weaned pigs: 1 to 10, Gilts: 1 to 6 Agonistic interactions: Weaned pigs: 1 to 10, Growing pigs, Gilts: 1 to 6 Test number : Backtest: 1 to 2 Human approach test: 1 to 4 Gender: male, female Pen category: pen in front of compartment, pen in back of compartment

Observer: 1 to 3 Pen: 1 to 40 Cross-fostering: cross-fostered, non cross-fostered Weight: weight at time of mixing (Covariate) Ani: animal effect Lit: litter effect Perm env: permanent environmental effect

69

70

Results

Heritabilities

The heritability of the NEA backtest trait was h² = 0.19 (Table 3). Regarding the heritabilities of

the trait NEA in the first and second backtest separately, the values were on the same level with

h² = 0.24 and h² = 0.20, respectively. The heritability of the LC human approach test traits of

weaned pigs was small with h² = 0.20. The values in the separated estimation for the four human

approach tests with weaned pigs decreased from the first test with h² = 0.33 to the fourth test

with h² = 0.10. The heritability of the human approach test with gilts was very low (h² = 0.03).

Regarding the heritabilities of the agonistic interaction traits NF and IF, the values were on one

level within the traits between age groups. The NF traits showed values of h² = 0.15, h² = 0.18

and h² = 0.10 for the weaned pigs, growing pigs and gilts, respectively. The heritability of the IF

trait was h² = 0.09 for weaned pigs, h² = 0.13 for growing pigs and h² = 0.10 for gilts.

Table 3: Heritabilities (h²) and standard errors (SE) for the backtest trait number of escape

attempts (NEA), the human approach test trait latency (LC) and for the agonistic interaction

traits number of fights (NF) and number of initiated fights (IF) of weaned pigs, growing pigs

and gilts.

Trait h² SE Backtest

Number of escape attempts (NEA) 0.19 0.05 NEA 1st Backtest 0.241 0.08 NEA 2nd Backtest 0.201 0.11

Human approach test

Latency of weaned pigs (LC) 0.20 0.06 LC 1st test 0.33 0.14 LC 2nd test 0.29 0.12 LC 3rd test 0.16 0.12 LC 4th test 0.10 0.06 Latency of gilts (LC) 0.03 0.24


Weaned pigs

Number of total fights (NF) 0.15 0.09 Number of initiated fights (IF) 0.09 0.16 Growing pigs Number of total fights (NF) 0.18 0.08 Number of initiated fights (IF) 0.13 0.19 Gilts Number of total fights (NF) 0.10 0.11

Number of initiated fights (IF) 0.10 0.11 1 Linear model

71

Correlations between agonistic behavioural traits and the backtest

The phenotypic correlations between the NF and IF agonistic traits at different ages and the

NEA backtest trait were very low (Table 4). Estimations of genetic correlations were different

and generally higher than the phenotypic ones. There were slightly positive genetic correlations

between NEA and the IF of weaned pigs (rg = 0.18). The correlation between NEA and NF was

lower (rg = -0.08). This was similar to the separated analysis of both backtests. Moderate, but

negative genetic correlations were estimated for the relations NF and IF of growing pigs and the

NEA (rg = -0.14 to -0.37). The genetic correlations between IF and NEA were higher than

between NF and NEA. Negative genetic correlations of the NF and IF of gilts and NEA could be

obtained. However, the traits NF and NEA (rg = -0.17) showed higher relations than IF and

NEA (rg = -0.02). The separated analysis of NEA and the NF and IF of gilts showed ambiguous

results.

Correlations between agonistic behavioural traits and human approach tests

Phenotypic correlations were low between the traits describing agonistic behaviour and the

human approach tests (Table 2). The genetic correlations between the human approach test of

weaned pigs and the NF and IF showed medium values (rg = -0.50, rg = -0.45, respectively).

Moderate genetic correlations could also be obtained between the LC human approach test trait

of gilts and the NF and IF agonistic traits (rg = -0.18 and rg = -0.21). The estimations of the

genetic correlations ranged between rg = -0.23 and rg = -0.07 for the NF and IF of growing pigs

and the human approach tests of weaned pigs.

The separated analysis of the human approach tests in relation to agonistic traits showed also

small phenotypic correlations. The genetic correlations increased with higher test number of the

human approach test especially in NF and IF of weaned and growing pigs and partially also in

NF and IF of gilts. The highest correlations were estimated between NF and IF of weaned pigs

and the third and fourth human approach test (rg = -0.64 to 0.71). The genetic relation of NF and

IF and the human approach test with growing pigs was lower but showed the same trend such as

the genetic correlations in weaned pigs. The genetic correlations between the human approach

test with gilts and the agonistic behaviour traits of the different ages were high.

Table 4: Phenotypic and genetic correlations between the NF (number of fights) and IF (number of initiated fights) agonistic traits

of weaned pigs (n = 1,111), growing pigs (n = 446) and gilts (n = 279) and the NEA backtest trait (number of escape attempts) in the first and

second backtest (n = 1,382) and the human approach tests trait latency (LC) of weaned pigs (n = 1,317) and gilts (n = 272).

Weaned pigs Growing pigs Gilts NF IF NF IF NF IF rp rg rp rg rp rg rp rg rp rg rp rg

Bac

kte

st NEA -0.03 -0.08 ± 0.17 0.02 0.18 ± 0.22 -0.02 -0.14 ± 0.28 -0.05 -0.28 ± 0.30 -0.04 -0.17 ± 0.44 -0.02 -0.02 ± 0.44

1st test -0.03 -0.09 ± 0.21 0.02 0.19 ± 0.27 -0.04 -0.16 ± 0.28 -0.05 -0.22 ± 0.33 -0.08 -0.42 ± 0.52 -0.05 -0.18 ± 0.52

2nd test 0.00 0.01 ± 0.21 0.02 0.21 ± 0.25 -0.03 -0.27 ±0.38 -0.05 -0.37 ± 0.42 0.02 0.18 ± 0.53 0.00 -0.11 ± 0.53

Hu

man

ap

pro

ach

tes

t

LC Weaned pigs

-0.05 -0.50 ± 0.24 -0.06 -0.45 ± 0.26 -0.02 -0.23 ± 0.39 -0.01 -0.07 ± 0.37 0.00 -0.18 ± 0.54 -0.01 -0.21 ± 0.55

1st test 0.00 0.15 ± 0.36 -0.01 0.12 ± 0.44 -0.01 -0.01 ± 0.43 -0.01 0.01 ± 0.45 -0.06 -0.45 ± 0.66 -0.06 -0.44 ± 0.60

2nd test 0.01 -0.12 ± 0.29 -0.02 -0.10 ± 0.30 -0.02 -0.15 ± 0.38 0.00 -0.02 ± 0.38 0.02 -0.17 ± 0.58 0.01 0.05 ± 0.56

3rd test -0.08 -0.64 ± 0.25 -0.12 -0.67 ± 0.26 -0.02 -0.25 ± 0.50 0.00 -0.07 ± 0.45 0.01 -0.32 ± 0.96 0.01 -0.22 ± 0.81

4th test -0.08 -0.71 ± 0.31 -0.08 -0.65 ± 0.40 -0.03 -0.39 ± 0.55 -0.01 -0.19 ± 0.50 -0.01 -0.17 ± 0.71 -0.01 -0.24 ± 0.78

LC Gilts -0.17 -0.72 ± 0.45 -0.17 -0.65 ± 45 -0.02 -0.13 ± 0.51 -0.01 -0.10 ± 0.52 0.00 -0.15 ± 0.60 -0.07 -0.34 ± 0.57

72

73

Discussion

Heritabilities

The heritabilities of the backtest traits showed that the NEA and LEA traits were most heritable.

Velie et al. (2009) estimated values for the number of struggles of h² = 0.54 and for the total

time spent struggling of h² = 0.49. In contrast to the present backtest, this test was not carried

out on a defined day of age of the pigs and took place between the 7th and 14th day of life. The

heritabilities of Velie et al. (2009) were higher than in the present study which could be

explained by these non-standardised test conditions i.e. the age of the pigs during the backtest.

Hence, standard conditions such as a defined age of the pigs might have an impact on the

heritabilities. Other investigations estimated heritabilities of h² = 0.14 to h² = 0.15 for the

backtest traits which were closer to the results of the current study (Rohrer et al., 2013).

Regarded separately, the values of the heritabilities in the first and second backtests for the NEA

trait in the present study were on the same level. Velie et al. (2009) also estimated similar

heritabilities for the separated analysis of the tests (h² = 0.38 and h² = 0.40, respectively),

however, these values were higher than in the present study.

The heritability of the LC human approach test trait with weaned pigs was on the same level

than that of the backtest traits NEA and LEA. Regarded separately, the heritabilities of LC of

the human approach test with weaned pigs, were less heritable the more often the test was

performed due to the better fitting of the threshold models to the data. Threshold models in

general provide more reliable estimations if the frequency manifestation of the traits is low (the

often test was performed the more pigs touched the stockperson). In contrast to these results of

the heritabilities, Velie et al. (2009) stated that a human approach test with finishing pigs was

not heritable. This could be a hint that the genetic determination of the reaction in this test

varied with the age of the animals.

The heritability of the human approach test with gilts was very low (h² = 0.03). However,

Hellbrügge et al. (2009) estimated a value of the human approach test with gilts of h² = 0.09 and

therefore, on a comparable level to the heritability of the human approach test with weaned pigs.

Furthermore, the results of Hemsworth et al. (1990) showed a high heritability for gilts in a

human approach test (h² = 0.38). The low heritability of the gilts is caused by the small dataset

of gilts used in the present study.

The heritabilities of the agonistic interaction traits were at nearly the same level within the traits

NF and IF and between the age groups. Hence, the genetic determination of agonistic

interactions did not change with the age of the animals. Stukenborg et al. (2012) found different

74

values between the age groups and explained these differences with the enhanced playing

behaviour of the weaned pigs which could be emphasied by Silerova et al. (2010), who stated

that the fighting and playing behaviour in weaned pigs could not be separated from each other.

In studies of Turner et al. (2008; 2009a), heritabilities with a wide range were estimated for

agonistic interactions with weaned pigs. Løvendahl et al. (2005) estimated heritabilities for the

agonistic behaviour of sows which were more comparable to the results of the present study.

Correlations between agonistic behavioural traits and the backtest

In literature, divergent results can be found regarding the relation between backtest traits and

aggressive behaviour. Forkman et al.(1995) performed an owner/intruder test (comparable with

a resident intruder test) to record the aggression of pigs. They found no relation between the

backtest trait escape attempts and the attack latency of pigs in this test (rp = -0.15, not

significant). Also D'Eath (2002) found no phenotypic relation between the backtest traits and

aggressive behaviour in a resident intruder test. Contrarily, Hessing et al. (1993) stated that pigs

with more escape attempts in backtests were more aggressive in a social confrontation test in the

first two weeks of age (the confrontation test is comparable with the mixing of unacquainted

pigs). Furthermore, Ruis et al. (2000a) found relations between the aggressive behaviour in a

food competition test and the backtest results. More highly reactive pigs in the backtest were on

a higher rank in the food competition test. Additionally, also Hessing et al. (1994) observed

more fighting behaviour in groups which contained only pigs with high reactions in the backtest

than in groups of low-reactive or mixed groups (low- and high-reactive animals). To the best of

our knowledge, the measuring of aggression in pigs with the help of direct observations in

mixing situations and the comparison with backtest results is as yet sparsely documented,

especially the genetic aspect of this behaviour. Furthermore, the artificial creation of test

situations to record individual aggression might not be able to show the aggressive interaction in

a mixing situation of unacquainted pigs, which is a common feature of modern pig production.

In addition, in previous studies, a too-short observation period was used which ended before the

rank order of the pigs was established. Therefore, these studies did not contain all agonistic

information. The observation period in the present study was two days without recording the

night with the aim of reducing the effort with the high number of animals used in this study.

Meese and Ewbank (1973) stated that the fighting behaviour decreased fundamentally until the

second day. Stukenborg et al. (2010) stated that there was a significant decrease in agonistic

interactions during the night. Investigations by Bolhuis et al. (2005) used aggressive interactions

in mixing situations of 180 min after weaning. They found positive phenotypic relations

between pigs with high reactions in backtests and the number of initiated fights. These findings

75

were in accordance with the results of the present study. The genetic correlation between the

NEA and IF trait of weaned pigs was positive. The more escape attempts were recorded in the

backtest, the more fights were initiated by these pigs at mixing after weaning. Pigs which

showed active defence reactions in the stressful situation of the backtest thus initiated more

fights in the also as stressful experienced remixing with the unacquainted pigs. Enquist and

Leimar (1983) and Rushen et al. (1988) stated that the outcomes of previous fighting influenced

the fighting ability. Therefore, pigs with previous success in fighting were more likely to fight or

to initiate fights again. The correlations between the number of fights won and the number of

initiated fights in weaned pigs was high at rg = 0.87 ± 0.07 and rp = 0.83 (Scheffler et al.,

2013a). In literature, pigs which fought most and which had most initiated fights were described

as dominant pigs (Meese and Ewbank, 1973; Arey, 1999).

Contrary to these results, the correlations between NEA and the fighting behaviour in growing

pigs showed negative genetic correlations. Pigs with low reactions in the backtest had more

agonistic interactions. This could be explained by the confidence in the individual’s fighting

ability and the rank position. More dominant pigs as weaned pigs were less involved in agonistic

interactions in new mixing situations at re-housing in the growing stable. D’Eath et al. (2004)

emphasised this statement in their investigation that success in fighting had a long-lasting effect

on the dominance status of the individual pig in a new mixed group. Also Otten et al. (1997)

stated that the former rank position in a group influenced the agonistic interactions in the new

group. More dominant pigs fought less than pigs with lower rank positions.

The genetic relations between NEA and the NF and IF of gilts showed ambiguous results. This

could be explained by the small number of animals in this age group. Therefore, in further

investigations these analyses should be repeated with a larger sample size.

Correlations between agonistic behaviour traits and human approach tests

Similar to the backtest, for the relation of the human approach test traits and the aggressive

behaviour of pigs, different results could be found in literature. The tests were performed using

various test conditions. Forkman et al. (1995) found no significant correlations between the

approach to a novel object and the attack latencies in a resident intruder test. Contrary to this,

Thodberg et al. (1999) showed that the pigs with more explorative behaviour in the human

approach test attacked faster in the resident intruder tests. Also in studies of Brown et al. (2009),

a tendency for a negative correlation between the human approach test trait latency and the

lesion scores as an indicator of aggression in pigs could be observed. To our knowledge, the

relation of the LC human approach test trait of weaned pigs and gilts and the aggressive

behaviour recorded by video observations during the first two days after mixing at different ages

76

(weaned pigs, growing pigs and gilts) has not yet been investigated. The present results show

moderate negative genetic correlations between the NF and IF agonistic behavioural traits of

weaned pigs, growing pigs and gilts and the LC human approach tests trait in weaned pigs and

gilts. Van Erp-van der Kooij et al. (2002) stated that the human approach test measures the

dominance and the explorative behaviour of pigs. The reaction in social tests depends on the

pre-existing social experiences and the characteristic behaviour of other pigs in the group

(Manteca and Deag, 1993; Jensen, 1995; D'Eath and Burn, 2002). Animals which fought after

weaning were mostly the dominate ones (Meese and Ewbank, 1973; Arey, 1999). The pigs

which had shorter latencies in the human approach test had therefore a higher rank position and

were more dominant in the group. Thus, they were able to displace other pigs in approaching the

humans in the test (Thodberg et al., 1999; Ruis et al., 2000a). The human approach test was

generally performed after the establishment of a group hierarchy had been nearly completed.

Therefore, the dominance rank of pigs had been established in the group at time of testing.

Dominant pigs had a higher confidence, which they applied in the agonistic interactions after

mixing situations (Otten et al., 1997; Otten et al., 2002). The genetic correlations between NF

and IF and the LC human approach test trait of weaned pigs increased the more often the test

was performed.

The genetic relations between NF and IF of growing pigs and the LC human approach test trait

of weaned pigs and gilts also showed negative values. With the age of the pigs, the correlations

between the LC human approach tests trait and the NF or IF agonistic traits (influence of

dominance) decreased. Therefore, pigs which were the dominant ones in previous groups fought

not that muh in new groups than pigs which were low-ranking pigs in former mixing situations

(Otten et al., 1997; D'Eath, 2004, 2005). Similar results could be obtained for the gilts. But the

smaller number of animals used in the bivariate analysis with 446 animals in growing pigs and

272 in gilts should also be taken into consideration. However, the results were in accordance

with the genetic correlations between the LC trait with weaned pigs and the agonistic

interactions with weaned pigs.

The backtest and the human approach test seemed to be related to the dominance status of the

pigs. However, in previous studies, no phenotypic and genetic correlations were found between

both tests (Scheffler et al., 2013c). Thus, the dominance might not be the only behavioural part.

Consequently, there must be further underlying physiological, functional and motivational

systems influencing behaviour in the tests. Also Jensen et al. (1995) stated that the individual

variation is caused by different environmental and motivational systems.

77

Conclusion

Pigs with high reactions in the backtests indicate pigs which initiate more fights at mixing as

weaned pigs. However, these dominant pigs fought less in mixing as growing pigs, which was

caused by the confidence appropriated in previous fighting situations. Thus, pigs with a large

number of escape attempts in the backtest became less aggressive growing pigs. Consequently,

the relations between backtest and agonistic interactions depend on the age of the animals and

on the experiences of previous fights. The human approach test of weaned pigs and gilts is able

to predict the agonistic behaviour in mixing of weaned pigs, growing pigs and gilts. Pigs with

shorter latencies in the human approach test fight more and initiate more fights due to the

dominance status achieved in the mixing situations.

References

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.



1091.








D'Eath, R.B., 2004. Consistency of aggressive temperament in domestic pigs: The effects of

social experience and social disruption. Aggressive Behav. 30, 435-448.

D'Eath, R.B., 2005. Socialising piglets before weaning improves social hierarchy formation

when pigs are mixed post-weaning. Appl. Anim. Behav. Sci. 93, 199-211.

D'Eath, R.B., Roehe, R., Turner, S.P., Ison, S.H., Farish, M., Jack, M.C., Lawrence, A.B., 2009.

Genetics of animal temperament: aggressive behaviour at mixing is genetically

associated with the response to handling in pigs. Animal 3, 1544-1554.

Enquist, M., Leimar, O., 1983. Evolution of fighting behaviour: Decision rules and assessment

of relative strength. Journal of Theoretical Biology 102, 387-410.

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

piglets. Applied Animal Behaviour Science 45, 31-42.

78










Sci. 37, 285-295.

Hessing, M.J.C., Schouten, W.G.P., Wiepkema, P.R., Tielen, M.J.M., 1994. Implications of

individual behavioural characteristics on performance in pigs. Livestock Production

Science 40, 187-196.












Jensen, P., 1995. Individual variation in the behaviour of pigs- noise or functional coping

strategies? Appl. Anim. Behav. Sci. 44, 245-255.

Jensen, P., Forkman, B.r., Thodberg, K., KÃ¶ster, E., 1995. Individual variation and consistency

in piglet behaviour. Applied Animal Behaviour Science 45, 43-52.

Kanis, E., De Greef, K.H., Hiemstra, A., van Arendonk, J.A.M., 2005. Breeding for societally

important traits in pigs. J. Anim. Sci. 83, 948-957.

Langbein, J., Puppe, B., 2004. Analysing dominance relationships by sociometric methods - a

plea for a more standardised and precise approach in farm animals. Appl. Anim. Behav.

Sci. 87, 293-315.

79

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. Applied Animal

Behaviour Science 30, 73-86.

Løvendahl, P., Damgaard, L.H., Nielsen, B.L., Thodberg, K., Su, G., Rydhmer, L., 2005.

Aggressive behaviour of sows at mixing and maternal behaviour are heritable and

genetically correlated traits. Livest. Prod. Sci. 93, 73-85.

Madsen, P., Jensen, J., 2000. A user's guide to DMU. A package for analysing multivariate

mixed models. Version 6, 1-33.

Manteca, X., Deag, J.M., 1993. Individual differences in temperament of domestic animals: A

review of methodology. Animal Welfare 2, 247-268.

Mäntysaari, E.A., Quaas, R.L., Gröhn, Y.T., 1991. Simulation study on covariance component

estimation for two binary traits in an underlying continuous scale. J. Dairy. Sci. 74, 580-

591.



Melotti, L., Oostindjer, M., Bolhuis, J.E., Held, S., Mendl, M., 2011. Coping personality type

and environmental enrichment affect aggression at weaning in pigs. Appl. Anim. Behav.

Sci. 133, 144-153.

O'Connell, N.E., Beattie, V.E., Moss, B.W., 2004. Influence of social status on the welfare of

growing pigs housed in barren and enriched environments. Anim. Welfare 13, 425-431.

Otten, W., Puppe, B., Kanitz, E., Schön, P.C., Stabenow, B., 2002. Physiological and behavioral

effects of different success during social confrontation in pigs with prior dominance

experience. Physiology & Behavior 75, 127-133.

Otten, W., Puppe, B., Stabenow, B., Kanitz, E., Schön, P.C., Brüssow, K.P., Nürnberg, G.,

1997. Agonistic interactions and physiological reactions of top- and bottom-ranking pigs

confronted with a familiar and an unfamiliar group: Preliminary results. Appl. Anim.

Behav. Sci. 55, 79-90.

Puppe, B., 1998. Effects of familiarity and relatedness on agonistic pair relationships in newly

mixed domestic pigs. Applied Animal Behaviour Science 58, 233-239.




80


2000a. Personalities in female domesticated pigs: behavioural and physiological


Ruis, M.A.W., te Brake, J.H.A., van de Burgwal, J.A., de Jong, I.C., Blokhuis, H.J., Koolhaas,

J.M., 2000b. Personalities in female domesticated pigs: behavioural and physiological

indications. Applied Animal Behaviour Science 66, 31-47.

Rushen, J., 1988. Assessment of fighting ability or simple habituation: What causes young pigs

(Sus scrofa) to stop fighting? Aggressive Behavior 14, 155-167.


Scheffler, K., Stamer, E., Traulsen, I., Krieter, J., 2013a. Estimation of genetic parameters for

agonistic behaviour of pigs at different ages. Submitted to J. Agr. Sci. .

Scheffler, K., Stamer, E., Traulsen, I., Krieter, J., 2013b. Genetic analysis of the individual pig

behaviour in backtests and human approach tests. .

Scheffler, K., Traulsen, I., Krieter, J., 2013c. Assessment of individual behaviour using backtest

and human approach test at different age levels. Submitted to Livest. Sci.


Silerova, J., Spinka, M., Sarova, R., Algers, B., 2010. Playing and fighting by piglets around

weaning on farms, employing individual or group housing of lactating sows. Appl.




Behav. Sci. 49, 185-198.



Stukenborg, A., Traulsen, I., Stamer, E., Puppe, B., Presuhn, U., Krieter, J., 2012. Heritabilities

of agonistic behavioural traits in pigs and their relationships within and between

different age groups. Livest. Sci. 149, 25-32.





Tuchscherer, M., Manteuffel, G., 2000. Die Wirkung von psychischem Stress auf das

Immunsystem. Ein weiterer Grund für tiergerechte Haltung (Übersichtsreferat). Arch.

Tierz., Dummerstorf 43, 547-560.

81

Tuchscherer, M., Puppe, B., Tuchscherer, A., Kanitz, E., 1998. Effects of social status after

mixing on immune, metabolic, and endocrine responses in pigs. Physiol. Behav. 64, 353-

360.


Rydhmer, L., Lawrence, A.B., 2009a. Genetic validation of postmixing skin injuries in


stable social conditions. J. Anim. Sci. 87, 3076-3082.


Rydhmer, L., Lawrence, A.B., 2009b. Genetic validation of postmixing skin injuries in


stable social conditions. Journal of Animal Science 87, 3076-3082.

Turner, S.P., Roehe, R., Mekkawy, W., Farnworth, M.J., Knap, P.W., Lawrence, A.B., 2008.

Bayesian analysis of genetic associations of skin lesions and behavioural traits to

identify genetic components of individual aggressiveness in pigs. Behav. Genet. 38, 67-

75.




75, 293-305.



82

83

GENERAL DISCUSSION

The objective of the present study was the assessment of agonistic behaviour in relation to

behavioural tests, i.e. the backtest and the human approach test of pigs, at different age levels.

To evaluate influences on the behaviour of the animals in these stressful situations, different

systematic effects were examined for their impact. Furthermore, individual pig behaviour within

and between the behavioural tests was analysed to find consistent behavioural patterns. To

investigate the possibility of implementation in selection programs and to improve animal

welfare, the study especially emphasised the genetic aspects of the ontogenetic development of

the agonistic behaviour as well as of the two behavioural tests: the backtest and the human

approach test.

Backtest

Research on rodents has shown that there are individual differences in reactions to

environmental stressors. These animals could be categorised into two groups of behaviour,

namely active and passive, the so-called coping strategies, (Benus et al., 1991). Also in pigs, a

large variation in coping with stress has been found (Bolhuis et al., 2003). These facts lead to

the assumption that the two coping styles, i.e. active and passive, obtained in rats and mice, also

exist in pigs (Spake et al., 2012). The backtest is frequently used in literature and measures the

attempts of the pigs to escape from the stressful situation of being laid on their back (Hessing et

al., 1993; Koolhaas et al., 1999; D'Eath and Burn, 2002). According to Hessing et al. (1993),

pigs which struggled more in the backtest were assumed as active copers and those which

struggled less were classified as passive copers. Active copers have also been described as more

aggressive and quicker to investigate novel stimuli (Koolhaas et al., 1999).

In the backtest in this study, the pigs were restrained on their backs for one minute. The number

of escape attempts, the duration of all escape attempts and the latency to the first escape attempt

were recorded as traits during this test time. The performance of the backtest was simple and the

recorded traits were easy to recognise. Another advantage of this test was its high

standardisation, which enables a higher comparability to similar studies and an easy

transferability by different stockpersons in practical approaches. Other behavioural tests have

shown more variations in their test performances and have been therefore less reliable.

The number of escape attempts trait was sufficient due to the medium heritability. This, together

with the results of the genetic and phenotypic correlations between all backtest traits, indicated

84

that the recluse recording of this trait is sufficient to evaluate the behaviour of the pigs in the

backtest.

Furthermore, the results of the present thesis show that one backtest is considered to be enough

for practical application regarding time-saving aspects. This was indicated by the results of the

heritabilities of the backtest traits, especially for the number of escape attempts (h² = 0.19),

which did not differ between the first and second backtests. Moreover, high genetic correlations

between the first and second backtests were obtained (rg = 0.69 – 0.90).

However, in the research area it could also be of interest to perform repeated backtests to

achieve more reliable results particularly regarding the coping style theory in pigs. In the present

study, the pigs were separated into classes on the basis of their reactions in the backtests.

According to Ruis et al. (2000), the 25 % of the pigs which struggled the most and the 25 %

which struggled the least were categorised as high-reactive (HR) and low-reactive (LR) animals,

respectively. The calculated Kappa-Coefficients showed that there were low consistencies in the

categories between the first and the second backtests. Therefore, a repeated backtest could

evaluate the consistencies of the results concerning the behaviour of the animals and hence the

existence of coping styles. Nevertheless, the effect of habituation to the test situation should not

be neglected. In the present study, it was not possible to clarify this effect on the behaviour due

to the performance of only two backtests. It also has to be taken into consideration that the

assignment of the boundaries for the different categories is not exactly defined in literature and

therefore subjective and not standardised. In contrast to the present study, Hessing et al. (1993)

categorised pigs into HR if they had more than two escape attempts and as LR with fewer than

two escape attempts. Pigs which struggled exactly twice were classified as D. An agreement

regarding the definition of the category boundaries will improve further assessments concerning

the coping styles.

Besides the mentioned facts, the knowledge of the relation of the backtest to performance traits

is also of particular importance in breeding programs. In literature, it has been shown that pigs

with high reactions in the backtest have higher daily weight gains, and a higher lean meat

content and less backfat (Ruis et al., 2000; van Erp-van der Kooij et al., 2003). Furthermore,

Velie et al. (2009) stated that there were genetic correlations in the total struggling attempts and

an average daily weight gain of rg = 0.38. However, contrary results were shown by Rohrer et al.

(2013), who found no genetic correlations between backtest results and performance traits. A

selection of pigs according to their backtest responses would have no negative effects on the

performance traits.

85

Human approach test

Due to the simple and practical test conditions, the human approach test was the second

behavioural test chosen for analysis in the present study. In literature, several variations of the

test performance can be found, e.g. the animals are tested alone in a novel environment or they

are tested in their common pen with conspecifics (Ruis et al., 2001; Brown et al., 2009). In the

present study, after entering the pen the stockperson stood motionless in front of the pen and

recorded the latency of the individual pig to approach and touch him or her. This chosen

performance had the advantage that the individual differences between the animals could be

obtained without considerable effort because the pigs were tested in their home pens as their

common environment. Besides the different test performances, the test time is not consistent in

literature either. For example, Thodberg et al. (1999) and Janczak et al. (2003) performed the

test in for three and five minutes, respectively. In the present study, due to the high number of

animals and due to the experience gained from preliminary tests, the test time was maintained at

one minute regarding the time-saving aspects in practical application.

At the beginning of the data collection, it was also planned to perform the human approach test

with growing pigs. However, at this age level, it was not possible to differentiate between the

animals due to the fact that all pigs touched the stockperson immediately after entering the pen.

Therefore, in this case, the human approach test with growing pigs did not provide any reliable

data. Due to the repeated test performances in weaned pigs with short time differences between

these tests, the animals had habituated to the test situation or to humans in general (Hemsworth

et al., 1987; Pedersen, 1997; van Erp-van der Kooij et al., 2002; Marchant-Forde et al., 2003).

This habituation had already been observed at earlier ages, which was indicated by the results of

the test with suckling pigs and weaned pigs. Here, it was shown that with a higher age the

latencies to touch the stockperson decreased significantly. In contrast to growing pigs, it was

possible to differentiate between the gilts in the human approach test. Here, the time between the

last test performance as weaned pigs dated back nearly 12 weeks in the flatdeck stable. Hence,

the effect of habituation decreased fundamentally in this time. This might be due to the higher

number of contacts to humans throughout their lives, so that the gilts were less interested in

humans (Hemsworth et al., 1990). Otherwise, it is also possible that gilts had had more negative

experiences with humans and thereby the humans were less attractive for them (Hemsworth et

al., 1989; Andersen et al., 2006).

Concluding these facts, the human approach tests with weaned pigs and with gilts provide data

with a good distinction between the animals. The test was heritable only in weaned pigs

(weaned pigs: h² = 0.20; gilts: h² = 0.03) but the heritability of gilts will be expected to increase

86

using a higher number of gilts in the study. For application in breeding programs or in practice,

the human approach test with gilts might be preferred and could easily be applied to pig

production during the individual performance testing of the gilts.

Furthermore, due to the increasing effect of habituation to the test situation with higher test

numbers and due to the high genetic correlations between the tests in weaned pigs, it is

sufficient to perform just one human approach test.

In literature, no relation between the human approach test and reproduction and performance

traits has been described (Hemsworth et al., 1990). Thus, the selection of animals due to their

reactions in the human approach test should not decrease important reproduction and

performance traits.


Video observations provide the possibility to record agonistic interactions between animals in

detail excluding the influence of humans as direct observers. Another positive aspect of video

observations is that they can be analysed time-independently and therefore offer a more flexible

method. However, beside all these positive aspects, the examination of video recordings is very

time-consuming and lavish and in this case three observers were needed to analyse the agonistic

interactions of the weaned pigs (approximately 2,200 hours of video observations). All

observers were trained with one video sequence of the weaned pigs before starting with the

analysis of the agonistic behaviour. Furthermore, the results of the observers were examined and

compared with another test sequence where an inter-observer reliability of greater than 90 %

was obtained. Nevertheless, one observer had significantly different results in the number of

agonistic interactions. This shows that despite intensive training, a certain subjectivity in the

data collection method could not be eliminated. One possibility to avoid observer effects and

also to make video observations less time-consuming is the analysis of the video tapes by

specially developed automated software. This approach was used by Ismayilova et al. (2013)

and Oczak et al. (2013), who tried to find specific body positions of pigs initiating an agonistic

interaction to interrupt this behaviour before a fight took place. Kashiha et al. (2013) stated that

it is possible to perform automatic individual recognition by video techniques by painting

patterns on the backs of the pigs is possible. Additionally, with the help of this automatic

recording method, the activity status of pigs could be obtained with an average accuracy of

88.7 %. Nevertheless, the fighting behaviour of pigs is more complex behaviour than simple

activity. Therefore, this video technique has hitherto not been capable of providing a reliable

method of using this technique in practical applications.

87

In addition to the subjectivity of the video evaluation by different observers, some animals

showed enhanced mounting behaviour in the present study. According to Clark and D'Eath

(2013) and Hintze et al. (2013), this behavioural pattern indicates an interaction between a

dominant pig and a submissive one. Therefore, to improve further investigations, mounting

behaviour should also be recorded as agonistic interaction. However, it should be clearly

differentiated from the agonistic interactions recorded as fights.

The results of the present study indicate that the number of fights and number of initiated fights

are the most suitable agonistic interaction traits. Due to the constant heritabilities between all

age groups (h² = 0.09 - 0.18) and due to the estimated genetic correlations between these traits, a

practical application in selection programs for breeding more docile animals is possible

especially regarding the recent development of pig production towards large and therefore

instable groups.

To use the obtained results in breeding or selection programs, the relation of these agonistic

traits to performance or reproduction traits should also be taken into consideration. In literature,

Turner et al. (2006) stated that genetic selection to decrease aggression in pigs is possible

without reducing the growth rate or backfat depth. Furthermore, several authors found no

correlations between agonistic interactions and reproductive traits of sows (Jarvis et al., 2006;

Kranendonk et al., 2007; Stukenborg et al., 2010).

Relations

The results of the present study show that it is possible to predict agonistic interactions using the

backtest and the human approach test.

It is indicated that pigs with an increased number of escape attempts in the backtest were those

who fought more when they were weaned pigs. These relations were reversed in the mixing of

growing pigs, i.e. pigs with more escape attempts in the backtest were less aggressive when they

were growing pigs. Hence, between agonistic interactions and the backtest results, a clear age

effect was estimated. In literature, pigs which fight more are described as dominant, and

therefore, high-ranking pigs (Meese and Ewbank, 1973; Arey, 1999). Due to success in previous

fighting and the appropriate confidence gained from this, dominant weaned pigs fought less as

growing pigs. This age effect has to be taken into consideration when using the backtest as

indicator for the prediction of the agonistic behaviour of pigs in different ages.

Negative genetic correlations were estimated in the relations between agonistic interactions and

the human approach tests. This means that pigs with shorter latencies in the human approach

tests showed more agonistic behaviour. Due to the fact that the human approach test was carried

out after the establishment of the rank order in the group, enhanced aggressive pigs had already

88

gained confidence and were therefore able to displace other pigs in approaching the stockperson

in the test. Therefore, an age-independent prediction of agonistic behaviour of the individual

pigs is possible with the human approach test.

Even though both tests could be used as indicators for the agonistic behaviour of pigs, no

relations between the backtest and the human approach test were found. This can be explained

by the fact that beside dominance or rank order, both tests also measure different underlying

physiological, functional and behavioural patterns.

In contrast to the backtest, the human approach test could be simultaneously carried out with all

animals in the pen and it might partly reflect the social structure of the group. Furthermore, in

the present study, higher and consistent genetic correlations with the agonistic interaction traits

were obtained in the human approach test. Moreover, no age effect was found in the human

approach test. Concluding these facts, the results indicate that the human approach test might be

preferred to predict the agonistic interaction of pigs at different ages.

Recommendations

The improvement in animal health and welfare is a common challenge in commercial pig

production. The present study provides results which are feasible instruments for these

improvements. Important legal changes in pig production are the required group housing of

sows since 2013, the prohibition of castration of boars without anaesthesia after 2018 and the

increased tail-biting problem due to the law of 2008, which allows tail docking only in justified

cases. Especially in the group housing of sows and the fattening of boars, it is important to

reduce fighting by breeding calm and less aggressive animals. Although the behavioural tests

and the agonistic interactions were investigated in gilts in the present study, a larger sample in

this age group might further enhance the reliability of the results. Due to the higher aggression

level of boars (Giersing, 2006), further investigations should analyse whether the behavioural

tests are also feasible indicators of agonistic interactions in boars. In the case of tail biting, there

is still the problem of recognising the perpetrator of this behaviour in the group. Therefore, in

further investigations, it would be of particular interest to see whether there is a relation between

tail biting and behavioural tests or whether enhanced agonistic interactions between the pigs

were observed before the tail-biting outbreak. If the behavioural tests prove to be possible

indicators of the above-mentioned problems, the results of the tests can be applied for selection

or breeding programs to improve animal welfare as well as housing conditions, i.e. the

systematic re-mixing of pigs.

89

References

Andersen, I.L., Berg, S., Bøe, K.E., Edwards, S., 2006. Positive handling in late pregnancy and

the consequences for maternal behaviour and production in sows. Appl. Anim. Behav.

Sci. 99, 64-76.

Arey, D.S., 1999. Time course for the formation and disruption of social organisation in group-

housed sows. Appl. Anim. Behav. Sci. 62, 199-207.

Benus, R.F., Bohus, B., Koolhaas, J.M., Van Oortmerssen, G.A., 1991. Heritable variation for

aggression as a reflection of individual coping strategies. Experientia 47, 1008-1019.

Bolhuis, J.E., Parmentier, H.K., Schouten, W.G.P., Schrama, J.W., Wiegant, V.M., 2003.

Effects of housing and individual coping characteristics on immune responses of pigs.

Physiol. Behav. 79, 289-296.




Clark, C.C.A., D'Eath, R.B., 2013. Age over experience: Consistency of aggression and

mounting behaviour in male and female pigs. Appl. Anim. Behav. Sci. 147, 81-93.



Giersing, M.L., Jan; Forkman, Björn, 2006. Animal welfare aspects of preventing boar teint.

Acta Vet. Scand. 48 1-3.

Hemsworth, P.H., Barnett, J.L., Coleman, G.J., Hansen, C., 1989. A study of the relationships

between the attitudinal and behavioural profiles of stockpersons and the level of fear of

humans and reproductive performance of commercial pigs. Appl. Anim. Behav. Sci. 23,

301-314.

Hemsworth, P.H., Barnett, J.L., Hansen, C., 1987. The influence of inconsistent handling by

humans on the behaviour, growth and corticosteroids of young pigs. Appl. Anim. Behav.

Sci. 17, 245-252.






Sci. 37, 285-295.

90

Hintze, S., Scott, D., Turner, S., Meddle, S.L., D'Eath, R.B., 2013. Mounting behaviour in

finishing pigs: Stable individual differences are not due to dominance or stage of sexual

development. Appl. Anim. Behav. Sci. 147, 69-80.










Jarvis, S., Moinard, C., Robson, S.K., Baxter, E., Ormandy, E., Douglas, A.J., Seckl, J.R.,

Russell, J.A., Lawrence, A.B., 2006. Programming the offspring of the pig by prenatal

social stress: Neuroendocrine activity and behaviour. Horm. Behav. 49, 68-80.

Kashiha, M., Bahr, C., Ott, S., Moons, C.P.H., Niewold, T.A., Öberg, F.O., Berckmans, D.,

2013. Automatic identification of marked pigs in a pen using image pattern recognition.

Comput. Electron. Agric.93, 111-120.




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. J.

Anim. Sci. 85, 420-429.






Oczak, M., Ismayilova, G., Costa, A., Viazzi, S., Sonoda, L.T., Fels, M., Bahr, C., Hartung, J.,

Guarino, M., Berckmans, D., Vranken, E., 2013. Analysis of aggressive behaviours of

pigs by automatic video recordings. Comput. Electron. Agric.99, 209-217.

91

Pedersen, L.J., 1997. Socialisation to humans of pigs at different ages. In: Hemsworth, P.H.,

Spinka, M., Kostal, L. (Ed.), Proceedings of the 31st International Congress of the

International Society for Applied Ethology, Uhrineves, Czech Republic, p. p.208.







Ruis, M.A.W., de Groot, J., te Brake, J.H.A., Dinand Ekkel, E., van de Burgwal, J.A., Erkens,

J.H.F., Engel, B., Buist, W.G., Blokhuis, H.J., Koolhaas, J.M., 2001. Behavioural and

physiological consequences of acute social defeat in growing gilts: effects of the social

environment. Appl. Anim. Behav. Sci. 70, 201-225.


pigs. Appl. Anim. Behav. Sci. 140, 146-153.





Turner, S.P., White, I.M.S., Brotherstone, S., Farnworth, M.J., Knap, P.W., Penny, P., Mendl,

M., Lawrence, A.B., 2006. Heritability of post-mixing aggressiveness in grower-stage

pigs and its relationship with production traits. Anim. Sci. 82, 615-620.

van Erp-van der Kooij, E., Kuijpers, A.H., van Eerdenburg, F.J.C.M., Tielen, M.J.M., 2003.

Coping characteristics and performance in fattening pigs. Livest. Prod. Sci. 84, 31-38.




75, 293-305.



92

93

GENERAL SUMMARY

This thesis focussed on the ontogenetic analysis of agonistic interactions of pigs as well as on

two behavioural tests, i.e. the backtest and the human approach test. Heritabilities as well as

phenotypic and genetic correlations were estimated to examine the possibility of an

implementation in breeding programs and to assess whether the behavioural tests are feasible

indicators of the prediction of agonistic behaviour of pigs in common, stressful mixing

situations.

The data were recorded on the research farm of the University of Kiel from January 2011 to

February 2012. The pigs were pure-bred and cross-bred animals of the races Large White and

German Landrace. The backtest was performed on the 12th and 19th day post-partum. The pigs

(n =1,382) were laid on their backs and loosely held in this supine position for one minute.

During this test time, the number of escape attempts, the duration of all escape attempts and the

latency to the first escape attempt were recorded as traits. The second behavioural test used in

the present study was the human approach test. This test was carried out with suckling piglets

(n = 1,318), weaned pigs (n = 1,317) and gilts (n = 272). The latency of the animals to approach

the stockperson was recorded during the test time of one minute. In addition to these

behavioural tests, the agonistic interactions of weaned pigs (n = 1,111), growing pigs (n = 446)

and gilts (n = 279) was obtained in the first two days after common mixing situations. The

recorded agonistic interaction traits were the number of fights, the duration of all fights, the

number of initiated and received fights and the number of fights won and lost.

In Chapter One, different systematic influences were tested for their impact on the behaviour of

pigs in the backtest and the human approach test. Furthermore, to find consistencies in the

behaviour of the animals, phenotypic correlations were estimated between the traits of one test

and also between the traits of different tests. It was shown that lighter piglets struggled more in

the backtests than heavier ones. Moreover, female pigs in the human approach test had shorter

latencies than male pigs. The phenotypic correlations and the Kappa-Coefficients which indicate

consistencies of the high-reactive and low-reactive animals in the backtest, were low between

the first and second backtests. Hence, the behaviour in both backtests was different and due to

an effect of habituation the first backtest was the more convincing test. The relations between

the human approach tests showed that the shorter the time difference between the tests was the

higher was the phenotypic correlation. In weaned pigs as well as in gilts, the performance in the

human approach test provides reliable results for the distinction between animals. Regarding

breeding issues, the test with gilts might be the more feasible test. No phenotypic relations were

94

obtained between the backtest and the human approach test, indicating that, both tests measure

different behavioural patterns.

Chapter Two dealt with the agonistic interactions of weaned pigs, growing pigs and gilts. Also

in this chapter, systematic influences as well as the impact of random effects were analysed.

Furthermore, heritabilities and genetic correlations for the different agonistic interaction traits

were estimated especially to compare the ontogenetic development of the agonistic behaviour

between the age groups. It was shown that cross-fostered animals were less aggressive in mixing

situations due to their socialisation in early life. The weight of the pigs had an influence on the

agonistic interactions in weaned and growing pigs. Heavier pigs fought more than lighter ones.

Low to moderate heritabilities, which were consistent between the age groups, indicate the

number of fights and the number of initiated fights as the most suitable traits for further

investigations. The correlations of the traits between the age groups showed no uniform trend,

indicating that, the agonistic behaviour varied with the age of the pigs.

The aim of Chapter Three was the estimation of heritabilities and genetic correlations between

the backtest traits and the human approach tests traits at different ages. The analyses of the

behavioural tests showed that the behaviour of the pigs in the backtest was more greatly

influenced by the litter effect than the behaviour of the pigs in the human approach test. The

number of escape attempts is the most reliable trait in selection programs due to the highest

heritabilities compared with the other backtest traits. The high genetic correlation between the

first and the second backtests for all traits indicated the sufficiency of one backtest in further

practical applications. The heritability of the human approach test with weaned pigs was higher

than for gilts. Due to the small correlations between the human approach tests of these ages,

different behavioural patterns were measured in weaned pigs and gilts.

To receive feasible and easy obtainable traits to predict agonistic behaviour, the backtest and

human approach test of weaned pigs and gilts were analysed in relation to the agonistic

interactions of weaned pigs, growing pigs and gilts in the Chapter four. Especially focussing on

the genetic aspects of these relationships, both behavioural tests might be indicators of agonistic

behaviour. However, the relations between backtest traits and agonistic interaction traits

depended on the age of the pigs. Pigs with a high number of escape attempts in the backtests

were more aggressive as weaned pigs but less aggressive as growing pigs due to the confidence

appropriated in previous fights. The animals with shorter latencies in the human approach tests

fought more and initiated more fights in all age groups even if the genetic relation became

smaller with the age of the pigs caused by previous experiences.

95

ZUSAMMENFASSUNG

Das Ziel der vorliegenden Arbeit bestand darin, das Verhalten von Schweinen verschiedener

Altersstufen in zwei standardisierten Verhaltenstests, dem Backtest und dem Human-Approach-

Test, sowie das agonistische Verhalten der Tiere bei der Gruppenbildung zu beurteilen. Hierfür

wurden Erblichkeiten ebenso wie phänotypische und genetische Korrelationen zwischen den

erhobenen Merkmalen geschätzt, um Aussagen über die mögliche Einbindung in

Zuchtprogramme treffen zu können. Des Weiteren sollte überprüft werden, ob die genannten

Verhaltenstests als Methoden zur Vorhersage des agonistischen Verhaltens der Schweine

geeignet sind.

Die Daten wurden an Reinzucht- und Kreuzungstieren der Deutschen Landrasse und der Rasse

Large White von Januar 2011 bis Februar 2012 auf dem Versuchsbetrieb der Universität Kiel

erhoben. Der Backtest wurde am 12. und 19. Lebenstag der Ferkel (n = 1.382) durchgeführt.

Während des Tests wurden die Tiere auf dem Rücken liegend für eine Minute locker in dieser

Position fixiert. Ermittelt wurden die Merkmale Anzahl an Befreiungsversuchen, die Dauer aller

Befreiungsversuche und die Latenz bis zum ersten Befreiungsversuch. Der Human-Approach-

Test wurde zweimal bei Saugferkeln (n = 1.318), viermal bei abgesetzten Ferkeln (n = 1.317)

und einmal bei Jungsauen (n = 272) durchgeführt. Dabei stellte sich die Versuchsperson für die

Testdauer von einer Minute reglos in die Bucht und maß die Zeit bis zum ersten physischen

Kontakt des Einzeltieres mit der Versuchsperson. Neben den genannten Verhaltenstests wurde

das agonistische Verhalten der Schweine zu bestimmten Zeitpunkten analysiert. Die

Beobachtungen erfolgten über zwei Tage mittels Videoaufzeichnungen. Hierzu erfolgte eine

individuelle Kennzeichnung der Tiere, sodass die Auswertung auf Einzeltierbasis durchgeführt

werden konnte. Gewählt wurden die für die Schweineproduktion typischen Zeitpunkte für

Umgruppierungen: Absetzen (n = 1.111), Umstallung in die Mast (n = 446) respektive

Jungsauenaufzucht (n = 279). Die erhobenen Merkmale waren die Anzahl Kämpfe, die

Kampfdauer, die Zahl initiierter und empfangener Kämpfe sowie die Anzahl gewonnener und

verlorener Kämpfe pro Tier.

Im ersten Kapitel wurden verschiedene systematische Faktoren hinsichtlich ihres Einflusses auf

das Verhalten der Schweine im Backtest und im Human-Approach-Test analysiert. Weiterhin

wurden phänotypische Korrelationen zwischen den Merkmalen innerhalb eines Tests sowie

zwischen den Verhaltenstests geschätzt, um konstante Muster im Tierverhalten zu erkennen.

Leichtere Tiere zeigten dabei signifikant stärkere Reaktionen im Backtest als schwerere. Die

phänotypischen Korrelationen sowie die Kappa-Koeffizienten, die als Maß der

96

Übereinstimmung der Kategorieneinteilung in stark reagierende Tiere und in schwach

reagierende Tiere im Backtest herangezogen wurden, waren zwischen dem ersten und zweiten

Backtest gering. Zwischen den Merkmalen des Backtests Anzahl an Befreiungsversuchen,

Dauer aller Befreiungsversuche und Latenz bis zum ersten Befreiungsversuch ergaben sich gute

Übereinstimmungen. Somit scheint die Erfassung des Merkmals Anzahl an

Befreiungsversuchen und die Durchführung lediglich eines Backtests, für belastbare

Verhaltensbeurteilungen der Tiere auszureichen. Im Human-Approach-Test hatten weibliche

Saugferkel und weibliche abgesetzte Ferkel kürzere Latenzzeiten als die männlichen Tiere. Bei

geringerer zeitlicher Distanz zwischen zwei Human-Approach-Tests konnten höhere

Korrelationen geschätzt werden. Aussagekräftige Ergebnisse ergaben sich bei der Durchführung

des Human-Approach-Tests bei abgesetzten Ferkeln und Jungsauen, wobei der Test bei

Jungsauen, hinsichtlich der Nutzbarkeit in Zuchtprogrammen, geeigneter erscheint. Die

geringen Zusammenhänge zwischen den Backtest und Human-Approach-Test Merkmalen

weisen darauf hin, dass beide Tests ein unterschiedliches Verhalten beschreiben.

Das Ziel des zweiten Artikels war die Analyse des agonistischen Verhaltens der Schweine

unmittelbar nach dem Absetzen, beim Umstallen in die Mast und als Jungsauen. Dabei stellte

sich heraus, dass Tiere, die im Rahmen des Wurfausgleichs versetzt wurden, weniger aggressiv

waren, vermutlich aufgrund der frühen Sozialisierung mit unbekannten Artgenossen. Des

Weiteren verdeutlichen die Ergebnisse, dass das Gewicht eine entscheidende Rolle im

agonistischen Verhalten der Schweine spielt, wobei die schwereren Tiere die aggressiveren

waren. Die Anzahl Kämpfe und die Zahl initiierter Kämpfe konnten aufgrund konstanter,

geringer bis moderater Erblichkeiten (h² = 0,09 - 0,18) zwischen allen Altersgruppen und hoher

Korrelationen zu anderen Merkmalen, als geeignete Parameter zur Beschreibung des

agonistischen Verhalten identifiziert werden. Die Zusammenhänge zwischen den Altersgruppen

lassen kein einheitliches Bild erkennen, sodass davon ausgegangen wird, dass das Verhalten

zwischen den Altersgruppen unterschiedlich determiniert ist.

Der dritte Artikel befasst sich mit der Schätzung von Erblichkeiten und genetischen

Korrelationen zwischen den Merkmalen des Backtests und denen des Human-Approach-Tests.

Das Merkmal Anzahl an Befreiungsversuchen im Backtest wies die höchsten Erblichkeiten auf

(h² = 0,19), somit scheint es für Selektionsprogramme am geeignetsten. Aufgrund der hohen

genetischen Korrelationen zwischen dem ersten und zweiten Backtest kann (rg = 0,69 – 0,90),

vor allem unter praktischen Gesichtspunkten, auf die Durchführung eines zweiten Backtests

verzichtet werden. Die Erblichkeiten des Human-Approach-Tests waren bei abgesetzten Ferkeln

97

(h² = 0,20) höher als bei Jungsauen (h² = 0,03), wobei jedoch die geringe Anzahl an Jungsauen

in der Analyse beachtet werden muss. Die geringen genetischen Korrelationen beim Test

zwischen abgesetzten Ferkeln und Jungsauen weisen darauf hin, dass die genetische

Determination des Verhaltens im Human-Approach-Test altersabhängig ist.

Im vierten Artikel wurden phänotypische und genetische Korrelationen zwischen den

Verhaltenstests und dem agonistischen Verhalten der Tiere geschätzt, um eine Aussage über die

Eignung der Verhaltenstests zur Vorhersage des agonistischen Verhaltens von Schweinen

verschiedener Altersstufen treffen zu können. Die genetischen Korrelationen verdeutlichen, dass

beide Verhaltenstests in der Lage sind, das Verhalten im Voraus abzuschätzen. Allerdings

weisen insbesondere die genetischen Zusammenhänge des Backtest Merkmals Anzahl an

Befreiungsversuchen mit den agonistischen Verhaltensmerkmalen einen nicht zu

vernachlässigenden Alterseffekt auf. Schweine mit starken Reaktionen im Backtest zeigten mehr

agonistische Interaktionen nach dem Absetzen, waren in der Mast jedoch weniger aggressiv,

was durch die in den vorangegangen Kämpfen eingenommene Rangposition und die daraus

gewonnene Souveränität erklärt werden kann. Weiterhin wurden negative genetische

Korrelationen zwischen den Human-Approach-Tests und den agonistischen

Verhaltensmerkmalen geschätzt. Dabei waren Schweine, die sowohl als abgesetztes Ferkel als

auch als Jungsau geringe Latenzzeiten aufwiesen, aggressiver bei der Umgruppierung zu

praxisüblichen Zeitpunkten.

98

99

DANKSAGUNG

An dieser Stelle möchte ich mich bei all denen bedanken, die mit ihrer Unterstützung und Motivation zum Gelingen dieser Arbeit beigetragen haben.

Mein aufrichtiger Dank gilt meinem Doktorvater Herrn Prof. Dr. Krieter für die Überlassung des Themas, die wissenschaftliche Betreuung und die mir gewährten Freiräume bei der Anfertigung der Dissertation. Weiterhin möchte ich mich für die Möglichkeiten bedanken, meine Forschungsergebnisse im In- und Ausland präsentieren zu können.

Herrn Prof. Dr. Thaller danke ich für Übernahme des Koreferats.

Die finanzielle Förderung erfolgte dankenswerterweise durch das Bundesministerium für Bildung und Forschung im Rahmen des Kompetenznetzes der Agrar- und Ernährungsforschung PHÄNOMICS. In diesem Zusammenhang danke ich auch allen Projektpartnern insbesondere Herrn Prof. Dr. Puppe und Frau Dr. Manuela Zebunke vom FBN in Dummerstorf.

Ein herzliches Dankeschön geht an Herrn Dr. Eckard Stamer und Frau Dr. Imke Traulsen für die fachliche Unterstützung bei der statistischen Auswertung und der schriftlichen Ausarbeitung der Arbeit.

Ausdrücklich bedanken möchte ich mich bei den Mitarbeitern des Versuchsguts Hohenschulen, Herrn Jerzy Kampa und Herrn Juri Hahn, die mich während der Datenaufnahme tatkräftig unterstützten und mir meine „Sonderwünsche“ bei der Erfassung der Daten erfüllten.

Bei Eva Pohlmann, Anne Knifka, Cornelia Pielow und Kathrin Riepe bedanke ich mich herzlich für die Hilfe einerseits bei der Datenerhebung als auch andererseits bei der Auswertung der zahlreichen Videobeobachtungen.

Meinen lieben jetzigen und ehemaligen Kollegen/Innen und Bürokamaraden/Innen danke ich für die schöne und unvergessliche Zeit an diesem Institut. Mein besonderer Dank gilt Kathrin, Irena, Regina, Astrid, Anne und Karo, die für alle fachlichen und weniger fachlichen Probleme immer ein offenes Ohr hatten und zum Gelingen der Arbeit entscheidend beigetragen haben.

Mein größter Dank gilt meinen Eltern und meiner Familie für ihren uneingeschränkten Glauben an mich, die Unterstützung in allen Lebenslagen und den Rückhalt den sie mir gegeben haben. Weiterhin bedanke ich mich bei Sebastian für seine Hilfe und Motivation.

100

101

LEBENSLAUF

ALLGEMEINE INFORMATIONEN

Name

Katharina Scheffler

Geburtsdatum 21.11.1985

Geburtsort Sangerhausen

Staatsangehörigkeit Deutsch

BERUFLICHE TÄTIGKEIT

Seit 07.2013

Wissenschaftliche Mitarbeiterin

am Institut für Tierzucht und Tierhaltung


bei Prof. Dr. Joachim Krieter

STUDIUM

10.2005 – 06.2010

Agrarwissenschaften an der Martin-Luther-Universität

Halle/Wittenberg

Fachrichtung: Nutztierwissenschaften

Abschluss: Diplom-Agraringenieur

SCHULBILDUNG

08.2002 – 07.2005

Geschwister-Scholl Gymnasium Sangerhausen

Abschluss: Allgemeine Hochschulreife

Documents

Evaluation of agonistic interactions and behavioural …...2 Furthermore, pigs with shorter latencies in the human approach test were more aggressive in mixing situations (Brown et