INDIVIDUAL AND GENETIC DIFFERENCES IN FEARFULNESS: EFFECTS ON
FEATHER PECKING AND MEAT QUALITY OF TURKEYS
By
Marisa Anna Erasmus

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Animal Science – Doctor of Philosophy
2014

ABSTRACT
INDIVIDUAL AND GENETIC DIFFERENCES IN FEARFULNESS: EFFECTS ON
FEATHER PECKING AND MEAT QUALITY OF TURKEYS
By
Marisa Anna Erasmus
Research with several species has demonstrated that individual differences in behavior or
temperament influence behavior, well-being and economically important characteristics such as
meat quality. In particular, differences in fearfulness are related to feather pecking in laying hens
and meat quality in pigs and cattle. There has been scant research examining the behavior and
well-being of commercial turkeys and even less is known about fear responses of turkeys or the
possible relationships between fearfulness, feather pecking and meat quality. Fearfulness was
examined in male commercial turkeys using four tests of fear, including tonic immobility (TI),
open field, (OF), voluntary approach (VA), and novel object (NO) tests. Changes in fear
responses over time were described and the reliability of all four tests of fear was assessed.
Although behavior changed over time, most TI (latency to vocalize and number of
vocalizations), OF (latency to ambulate, numbers of steps taken, squares entered, and
defecations) and VA (latencies to move within two body lengths, one body length, approach and
peck) test responses were reliable as indicated by significant, moderate correlation coefficients.
The only reliable test measure for the NO test was the latency to peck the NO. The validity of the
four tests was assessed by comparing inter-test correspondence at two ages (4-6 weeks and 8-10
weeks) using two strategies: 1) turkeys showing extreme behavioral responses in each test were
selected and classified as high responders (HR) or low responders (LR) and their behavior was
compared across test situations; 2) test measures from each test were correlated with test
measures from all other tests. At 4-6 weeks, OF test behavior tended to be correlated with VA

test behavior. At 8-10 weeks, OF test behavior was correlated with TI test behavior and tended to
be correlated with NO test behavior. The OF test was subsequently used to 1) examine the
relationship between fearfulness and feather pecking in males and females of a commercial
(COMM) and randombred (RB) turkey strain, and 2) examine the relationship between
fearfulness and meat quality in COMM and RB males. No associations were found between
physiological (corticosterone levels) and behavioral OF responses and feather pecking or meat
quality. Some strain differences were observed in feather pecking behavior. Specifically,
plumage damage was worse in COMM compared to RB turkeys. Furthermore, a large percentage
of turkeys (> 39%) developed feather pecking. In conclusion, this research demonstrates that
behavior of turkeys in fear tests is reliable under the conditions of this research, but caution is
needed when interpreting responses to the NO test. New information was presented regarding
feather pecking behavior of turkeys. In contrast to laying hens, there does not appear to be an
association between feather pecking behavior and fear responses in an OF test, nor is there an
association with meat quality, contrary to what has been demonstrated with pigs and cattle.
Results may be useful in the development of welfare assessment programs for turkeys because
fear tests are often used to assess welfare. Moreover, results provide insights into feather pecking
behavior and identify important areas for future research regarding feather pecking and the
relationship between stress, meat quality and turkey temperament.

To my family and especially Ouma Anna for teaching me to be creative, follow my dreams and
“byt vas”.

iv

ACKNOWLEDGEMENTS
The road to the PhD is filled with bumps, forks, and adventures. Without the help and
support of my family, friends and colleagues, I would not have been able to complete this
journey.
First I would like to thank my adviser, Dr. Janice Swanson, for providing me with this
extraordinary opportunity and the freedom to explore my own research questions. Janice, thank
you so much for your patience, advice and for your willingness to listen no matter the issue.
Thank you for reminding and encouraging me to stay focused on the goal. Your guidance and
mentorship have shaped me into a more critical, confident scientist.
I am grateful to the members of my guidance committee, Dr. Dalen Agnew, Dr. Darrin
Karcher, Dr. Janice Siegford and Dr. Tina Widowski. Dalen – thank you for your willingness to
provide assistance and for helping me understand and incorporate immunohistochemistry into
the study of animal behavior and welfare so that I may see behavior from a different perspective.
Darrin – thank you for listening, providing advice and always having an open door. Thank you
for pushing me to “think outside the box”. I am also grateful to you for finding turkeys for my
experiments; they would not have been possible otherwise. Janice – thank you for making me a
part of your lab and for helping me develop my own personal mentorship and teaching style.
Through working and collaborating with you I have come to appreciate and enjoy teaching.
Tina – thank you for seeing me through another degree with turkeys. It is because of your
guidance and mentorship that I am where I am today. I know I can always count on you for
advice and support.

v

I would like to express my gratitude to Dr. Steve Bursian for his kindness and for always
being willing to listen and provide advice. I would like to thank Angelo Napolitano for his
patience, help and many lunch time conversations as I ran turkey project after turkey project.
This work would not have been possible without the help of the Animal Behavior and
Welfare Group and all of the student volunteers who helped with data collection at the farm and
watched hours and hours of turkey video. Many thanks to Dana Campbell, Courtney Daigle,
Melissa Elischer, Jackie Jacobs, Maja Makagon, David Orban, and Ali Witaifi for their support
and for keeping me (mostly) sane.
To my wonderful parents, Koos and Merlynn; sister, Annette; and brother, Louw – you
have supported me and provided encouragement no matter which path I chose and reminded me
to stay focused and stay positive. And lastly, to my husband, Jeffrey, this would not have been
possible without you and I am glad we are able to share our adventures and complete this
journey together.

vi

PREFACE
Chapter 2 was formatted for publication in Applied Animal Behaviour Science and corresponds
to the peer-reviewed version of Erasmus, M. and J. Swanson. 2014. Temperamental turkeys:
reliability of behavioural responses to four tests of fear. Applied Animal Behaviour Science. 157:
100-108.

Chapter 3 was formatted for publication in Animal Behaviour, Chapter 4 was formatted for
publication in Physiology & Behavior and Chapter 5 was formatted for publication in Poultry
Science.

vii

TABLE OF CONTENTS
LIST OF TABLES ......................................................................................................................... xi
LIST OF FIGURES ..................................................................................................................... xiii 
KEY TO ABBREVIATIONS ....................................................................................................... xiv
CHAPTER 1: GENERAL INTRODUCTION AND LITERATURE REVIEW ............................ 1
GENERAL INTRODUCTION....................................................................................................... 1
LITERATURE REVIEW ............................................................................................................... 3
Personality, temperament, coping styles and fearfulness ................................................... 3
Tests used to assess fear in poultry ..................................................................................... 9
Reliability and validity of fear tests .................................................................................. 13
Fear and stress physiology ................................................................................................ 16
Physiology of tonic immobility ....................................................................................... 19
Injurious pecking .............................................................................................................. 20
Feather pecking of laying hens ......................................................................................... 22
Feather pecking of turkeys ................................................................................................ 24
Stress, temperament and meat quality .............................................................................. 25
Pre-slaughter stress and meat quality................................................................................ 26
Temperament and meat quality ......................................................................................... 27
CHAPTER OVERVIEW .............................................................................................................. 28
REFERENCES ............................................................................................................................. 29
CHAPTER 2: TEMPERAMENTAL TURKEYS: RELIABILITY OF BEHAVIORAL
RESPONSES TO FOUR TESTS OF FEAR ................................................................................ 41
ABSTRACT.................................................................................................................................. 41
INTRODUCTION ........................................................................................................................ 42
MATERIALS AND METHODS.................................................................................................. 44
Animals and housing......................................................................................................... 45
Test procedures ................................................................................................................. 46
Test schedule ..................................................................................................................... 46
Test measures .................................................................................................................... 47
Statistical analysis ............................................................................................................. 49
RESULTS ..................................................................................................................................... 52
Changes in fear responses over time and with repeated testing........................................ 52
Test-retest reliability ......................................................................................................... 53
DISCUSSION ............................................................................................................................... 54
ACKNOWLEDGEMENTS .......................................................................................................... 60
APPENDIX ................................................................................................................................... 61
REFERENCES ............................................................................................................................. 66

viii

CHAPTER 3: USING INTER-TEST CORRESPONDENCE TO ASSESS THE VALIDITY OF
FOUR TESTS OF FEAR FOR TURKEYS .................................................................................. 71
ABSTRACT.................................................................................................................................. 71
INTRODUCTION ........................................................................................................................ 72
MATERIALS AND METHODS.................................................................................................. 75
General procedures ........................................................................................................... 75
Statistical analysis ............................................................................................................. 78
RESULTS ..................................................................................................................................... 79
Pair-wise correlations of test measures ............................................................................. 79
Differences between high (HR) and low (LR) responders ............................................... 79
DISCUSSION ............................................................................................................................... 80
ACKNOWLEDGEMENTS .......................................................................................................... 85
APPENDIX ................................................................................................................................... 86
REFERENCES ............................................................................................................................. 92
CHAPTER 4: FEATHER PECKING AND OPEN FIELD BEHAVIOR OF TURKEYS ........... 96
ABSTRACT.................................................................................................................................. 96
INTRODUCTION ........................................................................................................................ 97
MATERIALS AND METHODS.................................................................................................. 99
Animals and housing......................................................................................................... 99
Open field testing procedures ......................................................................................... 100
Feather pecking behavior ................................................................................................ 101
Plasma corticosterone levels ........................................................................................... 104
Statistical analysis ........................................................................................................... 105
RESULTS ................................................................................................................................... 107
Test-retest reliability of OF behavior .............................................................................. 107
Sex and strain differences in OF test responses .............................................................. 108
Sex and strain differences in feather pecking ................................................................. 110
Feather scores.................................................................................................................. 111
Feather pecking and open field responses ...................................................................... 112
DISCUSSION ............................................................................................................................. 112
ACKNOWLEDGEMENTS ........................................................................................................ 119
APPENDIX ................................................................................................................................. 120
REFERENCES ........................................................................................................................... 130
CHAPTER 5: INDIVIDUAL AND GENETIC DIFFERENCES IN OPEN FIELD BEHAVIOR
AND ITS RELATIONSHIP WITH MEAT QUALITY OF TURKEYS .................................... 134
ABSTRACT................................................................................................................................ 134
INTRODUCTION ...................................................................................................................... 135
MATERIALS AND METHODS................................................................................................ 137
Animals and housing....................................................................................................... 137
Open field (OF) testing ................................................................................................... 138
OF responses and meat quality ....................................................................................... 139
Blood sample collection and corticosterone assays ........................................................ 139
Processing ....................................................................................................................... 140
Muscle pH, R-value, and color measurements ............................................................... 140
ix

Statistical analysis ........................................................................................................... 141
RESULTS ................................................................................................................................... 142
Cluster analysis ............................................................................................................... 142
Strain and cluster differences in meat quality and carcass weight.................................. 143
DISCUSSION ............................................................................................................................. 143
ACKNOWLEDGEMENTS ........................................................................................................ 148
APPENDIX ................................................................................................................................. 149
REFERENCES ........................................................................................................................... 154
CHAPTER 6: GENERAL DISCUSSION AND CONCLUSIONS ............................................ 158
REFERENCES ........................................................................................................................... 165

x

LIST OF TABLES
Table 2.1. Testing schedule and test order for each group of four pens. Each test group was tested
three times in four tests of fear (VA = voluntary approach test, NO = novel object test, OF = open
field test, TI = tonic immobility)................................................................................................... 62 
Table 2.2. Differences in percentages of turkeys performing a particular behavior in the open
field, tonic immobility, voluntary approach and novel object tests between days (test Period 1 vs.
2) and between weeks (Period 1 vs. 3). ........................................................................................ 63 
Table 2.3. Differences in median (25-75% range) values of open field, tonic immobility,
voluntary approach and novel object test measures between Period 1 and Period 2 and Period 1
and Period 3. Testing occurred at 28 to 40 days (Period 1), 29 to 41 days (Period 2), and 56 to 68
days (Period 3). ............................................................................................................................. 64 
Table 2.4. Spearman rank correlation (rS) of test measures between days (Period 1 vs. 2) and
weeks (Period 1 vs. 3) for four tests of fear (open field, tonic immobility, voluntary approach,
and novel object tests)................................................................................................................... 65 
Table 3.1. Spearman rank correlation coefficients (rS) of open field (OF), tonic immobility (TI),
voluntary approach (VA), and novel object (NO) test measures. ................................................. 87 
Table 3.2. Wilcoxon rank sum test differences in median (25-75% quartiles) values of open
field (OF), tonic immobility (TI), voluntary approach (VA) and novel object (NO) test measures
for turkeys classified as high (HR) or low (LR) responders in each test at Age 1 (4 to 6 weeks).88 
Table 3.3. Wilcoxon rank sum test differences in median (25-75% quartiles) values of open
field (OF), tonic immobility (TI), voluntary approach (VA) and novel object (NO) test measures
for turkeys classified as high (HR) or low (LR) responders in each test at Age 2 (8 to 10 weeks).
....................................................................................................................................................... 90 
Table 4.1. Spearman rank correlation (rS) of open field (OF) behavior among test periods for
male and female commercial (COMM) and randombred (RB) turkeys. .................................... 121 
Table 4.2. Behavioral responses of commercial (COMM) and randombred (RB) male and female
turkeys to open field testing at 3 ages - Test 1: wk 1; Test 2: wk 4 (males) and wk 5 (females);
Test 3: wk 11 (males) and wk 12 (females). ............................................................................... 122 
Table 4.3. Sex and strain differences in feather pecking (FP) behavior (n = 4 pens of males and
females per strain). Behavior was recorded over 3 days. Raw data (mean ± SE) are presented. 124 
Table 4.4. Feather scores (lsmean ± SE) from the neck, tail, and right and left flight feathers of
male and female commercial (COMM) and randombred (RB) turkeys. .................................... 125 

xi

Table 4.5. Behavioral open field test responses of male and female turkeys that were observed to
perform feather pecking (PECK) and that did not perform feather pecking (NPECK) of a
commercial (COMM) and randombred (RB) strain. Open field testing was conducted at 3 ages (Test 1: wk 1; Test 2: wk 4 (males) and wk 5 (females); Test 3: wk 11 (males) and wk 12
(females). .................................................................................................................................... 126 
Table 5.1. Differences in live weight, carcass weight, pH, R-value, and L*, a*, and b* values
between male turkeys of commercial (COMM) and randombred (RB) strains.......................... 150 

xii

LIST OF FIGURES
a

a

a

a

Figure. 4.1. Plasma corticosterone levels (pre-test , post-test and change from pre- to post-test)
of male and female turkeys of a commercial (COMM) and randombred (RB) strain tested in an
open field test at 11 (males) and 12 (females) weeks of age. Raw data (mean ± SE) are presented.
Statistical analyses were conducted on log-transformed data..................................................... 128 
Figure. 4.2. Plasma corticosterone levels (pre-test , post-test and change from pre- to post-test)
of male and female turkeys of a commercial (COMM) and randombred (RB) strain that were
observed to perform feather pecking (PECK) and that did not perform feather pecking (NPECK).
Turkeys were tested in an open field test at 11 (males) and 12 (females) wk of age.................. 129 
Figure. 5.1. Dendogram (graphical representation of cluster analysis) of clusters of commercial
(COMM) male turkeys (n = 44) based on their latency to ambulate, number of steps and number
of defecations in an open field test at 11 weeks of age. Two clusters were identified (dotted line).
..................................................................................................................................................... 151 
Figure. 5.2. Dendogram (graphical representation of cluster analysis) of clusters of randombred
(RB) male turkeys (n = 47) based on their latency to ambulate, number of steps and number of
defecations in an open field test at 11 weeks of age. Two clusters were identified (dotted line).
..................................................................................................................................................... 152 
Figure. 5.3. Plasma corticosterone levels (pre-test, post-test and change from pre- to post-test) of
male turkeys of a commercial (COMM) and randombred (RB) strain tested in an open field test
at 11 wk of age. Turkeys were grouped into two clusters based on their responses in the open
field test. Raw data (mean ± SE) are presented. Statistical analyses were conducted on logtransformed data.......................................................................................................................... 153 

xiii

KEY TO ABBREVIATIONS
Ach = acetylcholine
ACTH = adrenocorticotropic hormone
AD = avoidance distance
ATP = adenosine triphosphate
COMM = commercial
CRH = corticotropin releasing hormone
d = day
DFD = dark firm dry
EEG = electroencephalogram
GABA = gamma-aminobutyric acid
h = hour
HFP = high feather pecking
HPA = hypothalamic-pituitary-adrenal
HR = high responder
HS = high stress
LAL = long attack latency
LFP = low feather pecking
LR = low responding
LS = low stress
LTI = long tonic immobility
min = minute
MP = moving person
xiv

NO = novel object
NPECK = non-feather pecking
OF = open field
PAG = periaqueductal gray matter
PECK = feather pecking
PSE = pale, soft exudative
PVN = paraventricular nucleus
QTL = quantitative trait locus
RB = randombred
s = second
SAL = short attack latency
SAM = sympathetic-adrenal-medullary
SP = stationary person
STI = short tonic immobility
TI = tonic immobility
TT = touch
VA = voluntary approach
wk = week

xv

CHAPTER 1: GENERAL INTRODUCTION AND LITERATURE REVIEW
GENERAL INTRODUCTION
The wild turkey is native to North America and was domesticated in Mexico over 2000
years ago (Buss, 1989; Crawford, 1992). Initially, turkeys were selected for plumage pattern and
th

color so that they could be used in shows, and it was only in the 20 century that selection for
meat production and conformation became important (Appleby et al., 2004). Along with
selection for production traits, there was also selection against dark feathering, resulting in most
domestic turkeys being white by the 1960s (Appleby et al., 2004). Selection for increased breast
muscle yield resulted in turkeys no longer being able to mate naturally and the modern
commercial turkey industry relies on artificial insemination to produce turkeys.
Although commercial turkey production comprises over 240 million turkeys annually in
North America (United States Department of Agriculture National Agricultural Statistics
Service, 2014), there has been very little research into the behavior and welfare of turkeys
compared to other poultry species. Feather pecking (the pecking at, plucking and sometimes
removal of feathers from conspecifics) is one of the main welfare problems of commercial turkey
production (Marchewka et al., 2013), but this abnormal behavior is poorly understood. The
majority of information regarding injurious pecking, particularly feather pecking, has come from
research with laying hens, and suggests that feather pecking is a complex, multifactorial problem
(Dalton et al., 2014). Research specifically examining feather pecking in turkeys has focused on
environmental factors (e.g. Crowe and Forbes, 1999; Duggan et al., 2014; Lewis et al., 1998;
Martenchar et al., 2001; Moinard et al., 2001; Sherwin et al., 1999), and the effects of genetics,
development and sex (Busayi et al., 2006; Hughes and Grigor, 1996). However, research with

1

laying hens has demonstrated that individual differences in behavior such as fearfulness also play
an important role in the development of feather pecking (reviewed in Rodenburg et al., 2013).
Individual differences in behavior, or animal temperament, are also associated with other
economically important characteristics such as meat quality. Meat quality is an important area of
research in turkey production because of the occurrence of pale, soft, exudative meat (PSE),
which is meat that is characterized by its pale color and inability to hold water (Sosnicki et al.,
1998). Research examining turkey meat quality has primarily focused on genetic (e.g. Updike et
al., 2005; Werner et al., 2008), environmental (e.g. Sarica et al., 2011) and pre-slaughter stress
(e.g. Owens and Sams, 2000) factors. As with feather pecking, there is no scientific literature
examining whether turkey temperament is related to meat quality.
Scientific literature regarding poultry temperament has focused on fearfulness, which can
be assessed using a number of different fear tests. Although tests used for hens and quail are
widely used for turkeys, research has not confirmed whether these tests are reliable or valid for
use in turkeys and very little is known about how turkeys respond to fear tests, or about their
fear-related behavior in general. Fear tests are also used to draw conclusions regarding animal
well-being. Using fear tests for turkeys without knowing whether these tests are reliable or valid
may lead to improper conclusions being drawn.
The objectives of this research were to provide more information regarding fear
responses of turkeys and how these responses differ between sexes and genetic strains.
Furthermore, this research examined whether fearfulness, as part of turkey temperament, is
related to feather pecking and meat quality.

2

LITERATURE REVIEW

Personality, temperament, coping styles and fearfulness
Many studies have documented the existence of correlated groups of individual
characteristics (Koolhaas et al., 2010) and personality traits (Stamps and Groothuis, 2010) in
different animal species. Interest in studying whether animals differ in behavior and whether
these differences are consistent across different situations and contexts has increased greatly in
recent years. The study of animal temperament has many applications ranging from human
medicine to animal breeding and production. Indeed, rodents are used as models for diseases and
disorders such as depression, anxiety and post-traumatic stress (reviewed in Bourin et al., 2007,
and Neumann et al., 2011). In production animal species, differences in temperament are
associated with differences in disease susceptibility (see Koolhaas et al., 1999) and other
economically important characteristics such as production, reproduction (see Jones and Boissy,
2011), and meat quality (e.g. Voisinet et al., 1997). There is evidence that individuals that differ
in feather pecking behavior (discussed below) differ in other behavioral characteristics as well.
Temperament is relevant to animal welfare because animals that differ in how they respond to
stressors in their environment may be at risk of reduced welfare if they are not able to
successfully cope with stressors. Animal well-being concerns “the state of the individual as
regards its attempts to cope with its environment” (Broom, 1986), and animals may develop
abnormal behaviors if they are unable to successfully cope with challenges (Wechsler et al.,
1995).
The study of animal personality is complicated, and made even more so by the various
terms that are used across disciplines. The terminology used in relation to individual variation in
animal behavior includes personality, temperament, coping styles, behavioral profiles and
3

behavioral syndromes (see Groothuis and Carere, 2005; Koolhaas et al., 2010; and Stamps and
Groothuis, 2010). In addition to the terms proactive and reactive, some other terms that are used
to differentiate between animals at the extremes of the behavioral spectrum include shy vs. bold,
active vs. passive, hawk vs. dove (Koolhaas et al., 2010) and fast vs. slow (see Groothuis and
Carere, 2005). Some of the specific behaviors or traits that are most often studied as part of, and
that are used to describe animal personality include activity, aggression, sociability, friendliness,
emotionality, confidence, anxiety, exploration, psychopathy, affinity, playfulness, dominance,
reactivity and fearfulness (reviewed in Gosling, 2001). Although there are some differences in
the literature regarding the definitions of terms relating to personality; temperament, personality
and coping styles are generally considered to have the same meaning (Cockrem, 2007; Stamps
and Groothuis, 2010).
Personality is defined as individual behavioral differences that are consistent across
contexts and time, and it concerns the behavior of individuals relative to one another rather than
the “absolute level” of an individual’s behavior (Stamps and Groothuis, 2010).
Behavioral syndromes refer to sets of behaviors that are correlated across different
situations (Sih et al., 2004; Jacobs, 2009), where a situation is defined as “a given set of
conditions at one point in time” (Sih et al., 2004). Stamps and Groothuis (2010) distinguish
between behavioral syndrome and personality, stating that behavioral syndrome refers to a suite
of behaviors that are correlated across time or across contexts; whereas personality refers to
behavior that is correlated across both time and context. Groothuis and Carere (2005) use the
term behavioral profiles as an adaptation of the definition of behavioral syndromes to include
both behavior and underlying physiology.

4

Temperament is often used instead of personality because the term personality is
perceived as being anthropomorphic (Gosling, 2001). In the animal behavior literature,
temperament is defined as the “characteristics of individuals that describe and account for
consistent patterns of feeling and behaving” and is usually described in terms of fearfulness or
coping styles (Jones and Boissy, 2011).
The term coping styles has been used widely in the biomedical and stress physiology
literature to refer to the strategies that individuals use to cope with environmental challenges (see
Groothuis and Carere, 2005), and a considerable amount of research has been dedicated to the
study of coping styles. Coping styles are defined as consistent behavioral and physiological
stress responses of individuals (Koolhaas et al., 1999) and “alternative response patterns in
reaction to a stressor” (Koolhaas et al., 2010). Therefore, coping styles are specifically concerned
with the responses to stressors. A stressor is defined as a stimulus that induces stress, which is
the state that results from activation of the hypothalamo-pituitary adrenocortical (HPA) axis and
glucocorticoid (stress hormone) secretion due to a stressor (Cockrem, 2007). Two coping styles
have been identified, namely proactive and reactive. Animals with proactive coping styles are
characterized as being bold, fast, aggressive, less fearful and having high sympathetic nervous
system activity and low corticosterone levels in response to stressors; whereas reactive animals
are slow, shy and cautious, more fearful and have lower sympathetic nervous system activity and
higher corticosterone levels in response to stressors (Cockrem, 2007; Carere et al., 2010).
Research with mice and pigs suggests that the two types of coping styles differ in behavioral
flexibility, such that animals with reactive coping styles behave based on environmental cues and
stimuli and fare better under “variable and unpredictable environmental conditions” (Koolhaas et

5

al., 2010). In contrast, proactive animals are more likely to develop routines and do better under
stable conditions (Koolhaas et al., 2010).
By definition, coping styles include both the behavioral as well as the physiological
response to a stressor. Although individual differences in physiological stress responses are
associated with individual differences in personality (reviewed in Cockrem, 2013), the link
between behavior and physiology is still debated. Carere et al. (2010) propose three mechanisms
to explain the relationship between behavior and glucocorticoid levels: behavior determines
particular physiological responses, such as an increase in glucocorticoid secretion, or vice versa,
or thirdly, behavior and stress physiology are both determined by other factors and are therefore
correlated, but do not determine one another (i.e. one does not cause the other). The authors state
that the third mechanism may provide a more comprehensive explanation of the link between
behavior and physiology (Carere et al., 2010). Along the same lines, Koolhaas et al. (2010)
distinguish between the quality (how animals respond (Coppens et al., 2010)) and the quantity
(how strongly animals respond (Coppens et al., 2010)) of the behavioral and physiological
responses to stress. Koolhaas et al. (2010) propose that qualitative and quantitative stress
responses should be considered as independent dimensions because individuals can vary along
both dimensions such that proactive and reactive copers may have either high or low stress
reactivity (rather than proactive animals having only low stress reactivity and reactive animals
having only high stress reactivity as stated in Cockrem (2007)).
Various genetic selection lines have been developed to study the relationship between
different behavioral and physiological traits and these genetic lines possess characteristics that
are consistent with proactive and reactive coping styles. Separate genetic lines of long (LAL) and
short (SAL) attack latency mice have been developed based on aggressive behavior which is

6

measured using the latency of the mice to attack an intruder that is introduced into their cage
(reviewed in Groothuis and Carere, 2005 and Cockrem, 2007). SAL mice display behavioral
characteristics and show physiological responses that are consistent with the proactive coping
style, whereas LAL mice possess characteristics consistent with the reactive coping style (see
Groothuis and Carere, 2005; and Carere et al., 2010). For bird species, the FAST (proactive
coping style) and SLOW (reactive coping style) genetic lines of the great tit (Parus major, a
passerine bird) were developed based on whether the birds took a long (SLOW) or short (FAST)
time to explore a novel environment or novel object (reviewed in Groothuis and Carere, 2005;
Cockrem, 2007; Carere et al., 2010). Two lines of quail were developed based on their tonic
immobility reactions (LTI – long tonic immobility and STI – short tonic immobility), as well as
two lines differing in corticosterone responses to restraint stress (LS – low stress and HS – high
stress) (see Cockrem, 2007). The quail lines are also consistent with proactive (LS and STI lines)
and reactive (HS and LTI lines) coping styles (Cockrem, 2007). In chickens, high (HFP) and low
(LFP) feather pecking lines of white leghorns were developed to examine differences in feather
pecking behavior and the relationship between feather pecking and other behavior such as
fearfulness (see Carere et al., 2010; Rodenburg et al., 2013). Although the physiological data
(corticosterone levels in response to stress) suggest that HFP chickens are proactive copers and
LFP chickens are reactive copers (Carere et al., 2010), behavioral data show the opposite: HFP
chickens (proactive) are less active in a novel environment and have longer durations of tonic
immobility compared to LFP chickens (reactive) (reviewed in Groothuis and Carere, 2005).
However, behavioral data are not consistent with what is seen in other species, and thus there
may be differences among animal species due to domestication or genetic selection that should
be taken into consideration when classifying animals as having a particular behavioral profile

7

(Groothuis and Carere, 2005). Furthermore, behavior may differ depending on which test is used
(e.g. TI vs. OF).
Although Cockrem’s (2007) description of proactive and reactive coping styles includes
differences in fearfulness, other researchers consider fearfulness to be an altogether separate
dimension of personality that is related to emotionality and not coping style (e.g. Koolhaas et al.,
1999; Jones and Boissy, 2011). Fearfulness refers to how susceptible an individual is to being
frightened (Boissy, 1995; Jones, 1996), whereas fear is the individual’s reaction to a perceived
threat or danger (Forkman et al., 2007).
In stark contrast to the behavioral ecology literature where personality of wild bird
species has been studied (e.g great tits - Carere et al., 2010, and geese - Kralj-Fišer et al., 2010),
research with domestic bird species, primarily laying hens and quail, has focused on fear and
fearfulness, and the terms temperament and personality have not traditionally been used. The
interest in fear in particular, rather than other aspects of poultry temperament, can be traced back
to the release of the Brambell Report in 1965 (Brambell, 1965) and the development of the Five
Freedoms which include freedom from hunger and thirst, freedom from discomfort, freedom
from pain, injury or disease, freedom to express normal behavior, and freedom from fear and
distress (Farm Animal Welfare Council, 1979). Consequently, fear and fearfulness have been
important areas of poultry behavior and welfare research and the ability to identify individual
differences in fearfulness and characteristics of fearfulness have received a great deal of
attention. Moreover, the development of tests for assessing fear and identifying fearful
individuals has been a major part of fear research.

8

Tests used to assess fear in poultry
Because an individual animal’s subjective state of fear can only be inferred indirectly, a
number of different tests have been developed for assessing fear in poultry (reviewed in Forkman
et al., 2007 and Jones, 1996). Stimuli that elicit fear have been classified into categories based on
the properties of the stimuli (Gray, 1979; reviewed in Boissy, 1995 and Forkman et al., 2007).
Categories of fear-eliciting stimuli include:
1) innate fears or stimuli associated with factors that are important to the species from an
evolutionary standpoint (e.g. a particular predator)
2) the specific characteristics associated with the stimulus, such as novelty, suddenness,
intensity and duration
3) stimuli that are perceived as threatening because previous experience and learning have
changed the animal’s perception of the stimulus (e.g. conditioned fear)
4) stimuli associated with social signals such as alarm calls or pheromones; for many farm
animal species, social isolation induces fear

Considering the characteristics of stimuli that induce fear, the most common types of fear
tests that are used for poultry expose poultry to predatory stimuli and/or restraint (e.g. tonic
immobility, and human approach or avoidance tests), or novelty (e.g. novel arena (open field)
and novel object tests), or sometimes elements of predatory stimuli and novelty are combined.
The most common poultry fear tests include the tonic immobility, open field, novel object,
human approach or avoidance and emergence tests (reviewed in Jones, 1996 and Forkman et al.,
2007).

9

Tonic immobility (TI) is an innate, unlearned response to capture and physical restraint,
characterized by temporary inhibition of the righting response; i.e. temporary paralysis (Gallup et
al., 1974). TI occurs in a wide range of animal species, including mammals, birds and reptiles
and has been referred to as death feigning and animal hypnosis (Gallup, 1974; Jones, 1986;
Leite-Panissi et al., 2006). According to Jones (1986), TI was “discovered” in 1636 by Daniel
Schwenter but TI occurring in a hen has been documented as early as 1562 (see Hoagland,
1928). Tonic immobility is believed to be the last reaction to an attack by a predator and
functions to protect the animal by improving its chances of escape (Suarez and Gallup, 1983).
Because TI is preceded by fear, TI is used as a test of fear for poultry. Specifically, the duration
of TI is used as an indicator of the level of fear of the bird. Longer TI duration is associated with
greater levels of fear prior to TI induction (Jones, 1996). General procedures for assessing TI are
described in Forkman et al. (2007). Briefly, hens and quail are placed on their side on a table or
on their back in a U-shaped cradle. One of the observer’s hands is placed on the bird’s sternum
and the other hand is placed on the head. Chicks are usually restrained on a cloth on a table top,
also with one hand on the sternum and one on the chick’s head. The bird is held in this manner
for 15 s and then released. If the bird rights itself within 10 s the procedure is repeated again,
usually up to a maximum of three times. Once TI has been achieved (if the bird remains on its
back for 10 s or longer), the amount of time taken before the bird moves its head for the first
time, and the duration of TI (amount of time taken before the bird rights itself) are recorded. The
number of inductions is also recorded. This methodology has been adapted for turkeys. Turkeys
are usually restrained on their side with one hand on the wing and the other hand holding the
shanks (Noble et al., 1996).

10

The open field (OF) test (also known as the novel arena test) was developed by Hall in
1938 to measure “emotionality” in rodents (Boissy, 1995). During the open field test, rodents are
exposed to several threatening stimuli, including novelty, bright light, and the absence of
landmarks (Archer, 1973). All of these stimuli are believed to induce fear (or anxiety). Rodents
that spend more time in the center of the open field are believed to be less anxious (Archer,
1973). The OF test was adapted for use in poultry in the 1980s because it was easy to use,
standardize and implement (Boissy, 1995) and is based on similar concepts to the rodent model.
However, species differences need to be taken into account when using the OF test. Specifically,
poultry are gregarious animals that have “evolved to hide in the undergrowth” and there is little
evidence that poultry find bright light aversive, therefore increasing light intensity in the open
field as is normally done for rodents is not likely to increase fear in poultry (Forkman et al.,
2007). Furthermore, poultry do not display thigmotaxis (staying close to the wall) as rodents do
and the time spent in the center of the OF may, therefore, not be a good indicator of a lack of fear
in poultry as it is for rodents. Nonetheless, the OF test for poultry is intended to measure fear of
novelty, and subjects animals to social isolation (because animals are removed from their flock
and tested individually) and is also used to assess social reinstatement motivation (Forkman et
al., 2007). It is believed that birds that are inactive and silent experience greater fear than birds
that move around the arena and vocalize (Forkman et al., 2007). Procedures for testing birds in
an OF test are described in Jones (1996): birds are removed from their home pen or cage and
individually tested in an unfamiliar arena that is usually larger and more brightly lit than their
home pen for a prescribed period of time. The open field is usually constructed as a square arena
that is divided into different areas by constructing a grid on the floor of the arena, but circular
arenas have also been used. Measures that are recorded include the latency to ambulate, the

11

number of steps taken and areas entered, the latency to vocalize and the number of vocalizations
emitted (e.g. Clarke and Jones, 2000).
The emergence test (also known as the hole-in-the-wall test) is another test that was
originally used to assess emotionality of rodents (Archer, 1973) and similar to the open field test,
has been applied for assessing fear in poultry. The emergence test is based on the premise that a
bird that is in a sheltered area, such as its home cage, will take longer to emerge into an adjacent
unfamiliar area (which could occur through a hole in the cage wall) if the bird is fearful (Jones,
1996; Forkman et al., 2007). The unfamiliar environment is usually more brightly lit than the
sheltered area. Forkman et al. (2007) concluded that the emergence test is a version of the OF
test, and the OF test is the better test to use.
Similar to the open field test, the novel object (NO) test also assesses the birds’ level of
fear of novelty (neophobia). An unfamiliar, inanimate object is presented to birds in their home
cage, and it is believed that the birds’ level of avoidance of the object is indicative of the level of
fear. Birds located farther away and taking longer to approach the novel object are interpreted as
experiencing a greater level of fear. The novel object test is fast and easy to implement (Forkman
et al., 2007), and has been incorporated into welfare assessment programs to evaluate the welfare
of poultry (Welfare Quality®, 2009a), pigs (Welfare Quality®, 2009b) and cattle (Welfare
Quality®, 2009c).
Up until relatively recently, laying hens were housed in conventional cages where
interactions with humans were limited to the interactions occurring at the front of the cage.
However, as laying hen housing systems are transitioning from caged systems to alternatives
such as aviaries, tests for assessing hens’ responses to humans in these larger systems have been
developed. Tests include the avoidance distance (approaching person) test, the moving person
12

test, the stationary person test and the touch test (Raubek et al., 2007). With the avoidance
distance (AD) test, the observer approaches a group of birds and determines the distance at
which birds withdraw (Raubek et al., 2007), or counts the number of birds within a prescribed
distance of the observer (Welfare Quality®, 2009a). With the stationary person (SP) test, the
person stands in the litter area with his or her back to a wall and counts the average number of
birds within a prescribed distance of the person (Raubek et al., 2007). The moving person (MP)
test combines elements of the AD and SP tests so that episodes of moving are interspersed with
stationary periods and the number of birds within a prescribed distance of the person when the
person is moving and stationary is counted (Raubek et al., 2007). The touch (TT) test is similar
to the SP test, except that the person attempts to touch three birds and also counts the number of
birds within reach (Raubek et al., 2007). In general, birds that take less time to approach the
person or show less avoidance when approached are perceived as being less fearful. The
avoidance distance test is used as part of the Welfare Quality® Assessment Protocol for broiler
chickens (Welfare Quality®, 2009a).
The aforementioned fear tests and versions thereof are have been used for wild birds (e.g.
Carrere et al., 2005; Kralj-Fišer et al., 2010) and other species, such as pigs (e.g. Brown et al.,
2009), fish (e.g. Martins et al., 2011; Castanheira et al., 2013) and cattle (e.g. Mazurek et al.,
2011) to classify animals as having a particular coping style or temperament.

Reliability and validity of fear tests
Before fear tests can be used to draw conclusions about animal welfare or to classify
animals as having a particular temperament, the tests need to be evaluated for reliability and

13

validity. Martin and Bateson (1993) define a reliable test as one that has few random errors
(precise) and is consistent over repeated applications; whereas a valid test is one that is accurate
(free from systematic errors), specific (the measurement measures what it is supposed to) and
scientifically valid (the measurement provides information relevant to the questions being
investigated). In addition to consistency and precision, reliability includes sensitivity (the test
measure changes when there are small changes in the true value of the measure) and resolution
(the test measure can detect small changes in the true value) (reviewed in Waiblinger et al.,
2006). Validity also includes convergent validity (independent test measures that are supposed to
measure the same thing are in fact correlated) and discriminant validity (independent test
measures that are not supposed to measure the same thing are not related), which form part of the
specificity of the test (see Waiblinger et al., 2006). In their review, Waiblinger et al. (2006)
further divide scientific validity into internal validity (whether the method used answers the
research question) and external validity (whether the results can be applied to situations outside
of the experimental environment).
The majority of information regarding the reliability and validity of fear tests for poultry
is presented in a review by Forkman et al. (2007) who concluded that TI is a reliable and valid
test of fear for hens and quail because previous research has shown TI to be repeatable when
birds are tested multiple times, and because there is a quantitative trait locus (QTL) for TI, which
suggests that the response is stable within individuals. Furthermore, TI is believed to be a valid
test of fear because research has shown that TI reactions are correlated with reactions in other
tests of fear: quail selected for long TI duration show increased fear responses to novel objects,
novel environments and humans compared to quail selected for short TI durations (reviewed in
Jones, 1996). Additionally, TI is influenced by other situations that affect fear responses.

14

Situations that are believed to cause fear, such as administering electric shocks to the animals or
housing animals in poor environments, are associated with increased duration of TI, whereas
situations believed to be associated with reduced levels of fear, such as housing animals in
enriched environments, are associated with reduced TI duration (reviewed in Forkman et al.,
2007).
Similar to the TI test, the OF test is reliable for testing fear in chickens and quail because
ambulation in the OF is heritable and because there is a QTL that is related to OF behavior,
indicating that the response is stable (reviewed in Forkman et al., 2007; Jensen et al., 2014). The
OF test is considered to be valid because behavior in the OF is affected by or associated with
other situations that affect fear. For example, electrical shocks induce fear in animals, and
animals that are shocked before OF testing vocalize and walk less (see Forkman et al., 2007).
Additionally, the OF test is believed to assess fear of novelty, and increasing the degree of
novelty present in the arena is associated with reduced activity levels and vocalizations and
increased latency to move (see Forkman et al., 2007). Conversely, situations believed to reduce
fear, such as housing birds in enriched environments, has been associated with greater activity
and shorter vocalization latency (see Forkman et al., 2007).
Although the emergence test has been validated and is correlated with the OF test and TI,
there is little information available regarding the reliability of the test, it is similar to the OF test,
and therefore the OF test is a better test to use (Forkman et al., 2007).
Forkman and colleagues (2007) concluded that the NO test is reliable for testing fear in
laying hens and quail because it is repeatable between days and weeks. Furthermore, birds reared
in enriched environments were more likely to approach the NO, indicating that it is a valid test of

15

fear, and birds housed in the top tier of battery cages (associated with high fear) that have high
TI durations also display longer latencies to approach the NO (reviewed in Forkman et al., 2007).
Raubek et al. (2007) assessed the reliability of the AD, SP, MP and TT tests for laying
hens. They concluded that the AD and TT tests were the “most promising” tests for assessing
hens’ reactions to humans because these tests had high repeatability within and between days
(Raubek et al., 2007). Graml et al. (2008) assessed the validity of the AD and TT tests and
concluded that the tests are valid because repeated handling of birds reduced the distance at
which hens moved away from the observer and increased the number of hens that approached the
person.
In conclusion, research with hens and quail has demonstrated the reliability and validity
of the OF, TI, NO, AD and TT tests for assessing fear in hens and quail. However, there has been
no research to date examining the reliability and validity of any of these fear tests for turkeys.

Fear and stress physiology
Emotions, especially fear, are important from an evolutionary perspective because they
promote survival. Therefore, fear responses are widely accepted as being “hard-wired” during
evolution (Davis, 1992). Fear is a “functional behavior system” that has been regarded as
defensive behavior (Misslin, 2003). Some of the most common defensive behaviors occurring in
response to predatory attacks and stimuli that induce stress include the “fight or flight” response,
freezing, tonic immobility, hypoalgesia (reduced pain sensitivity) and reactions mediated by the
autonomic nervous system (e.g. piloerection) (reviewed in Misslin, 2003).
Brain regions involved in fear responses in mammals include the amygdala,
hippocampus, thalamus, prefrontal cortex and brain stem structures (reviewed in Shin and

16

Liberzon, 2010). The amygdala is the most important brain region in mammals involved in the
response to fear-inducing stimuli. Through projections to the brainstem, the amygdala causes the
expression of fear responses, and projections to the cortex result in the “fear experience” and
cognitive processing of emotional stimuli (LeDoux, 2003). In birds, the fear response is mediated
by the nuclei teaniae amygdala, posterior pallial amygdala, and the subpallial amygdala, which
are homologous to the amygdala in mammals (Reiner et al., 2004).
A stimulus that elicits a fear response can be considered an emotional stressor because it
has both an affective component (the subjective experience of fear) and a physiological stress
component resulting in activation of the stress-related pathways (Armario et al., 2012). Cannon
(1914) is credited with developing the “fight or flight” concept which is concerned with the
actions of the sympathetic-adrenal medullary (SAM) axis. Hans Selye is credited with
developing the concept of stress, or the General Adaptation Syndrome as it was called, which is
concerned with activity of the hypothalamic-pituitary-adrenal (HPA) axis (Selye, 1932). The
SAM and HPA axes are the main pathways involved in the stress response. Upon detection of a
threat, the SAM and HPA axes are activated. The SAM axis is faster-acting and initiates the
immediate responses to a stressor, whereas the HPA axis is slower-acting and is involved in
responses that have metabolic and anti-inflammatory effects (reviewed in Mellor et al., 2000).
The activation and effects of the SAM and HPA axes have been reviewed in several books and
manuscripts (e.g. Matteri et al., 2000; Rodrigues et al., 2009; Amario et al., 2012).
The SAM axis is activated when centers in the brainstem receive signals from the
amygdala. Activation of the SAM axis results in the fight/flight response and the release of
epinephrine (adrenaline) and norepinephrine (noradrenaline) from sympathetic nerves and the
adrenal medulla. Epinephrine and norepinephrine circulate throughout the body and result in

17

various autonomic effects, including increased blood pressure and heart rate, and divert energy to
muscle and inhibit digestion.
In response to signals from the amygdala, the paraventricular nucleus (PVN) of the
hypothalamus releases corticotrophin-releasing hormone (CRH) and vasopressin. Corticotrophinreleasing hormone and vasopressin bind to receptors in the anterior pituitary gland, which
releases adrenocorticotropin-releasing hormone (ACTH). Adrenocorticotropin-releasing
hormone binds to receptors in the cortex of the adrenal gland, causing the secretion of
glucocorticoids. Corticosterone is the main glucocorticoid in birds, rodents, amphibians and
reptiles; cortisol is the main glucocorticoid in fish and most mammals. Glucocorticoids are then
released into the bloodstream and bind to receptors throughout the body and brain.
Glucocorticoids result in the release of glucose that is used for energy, and glucocorticoids also
have a negative feedback response that reduces the expression of CRH in the PVN of the
hypothalamus.
Further, fear processing and activation of the amygdala have monoaminergic, endocrine
and autonomic consequences (reviewed in Rodrigues et al., 2009). Monoaminergic consequences
include activation of the SAM axis and the release of the neurotransmitters norepinephrine,
acetylcholine, serotonin and dopamine throughout the brain, causing increased arousal and
vigilance and autonomic effects (e.g. increased heart rate). Endocrine consequences arise as a
result of stimulation of the HPA axis, causing the release of CRH and ACTH which lead to the
secretion of glucocorticoids (e.g. corticosterone). Physiological changes occurring during fear,
therefore, include increased heart rate, change in respiratory rate (breathing), decreased
salivation, and increased body temperature. Furthermore, animals may display behavior
associated with the fight/flight response, such as freezing or escape and piloerection (erection of

18

hair) or pteroerection (erection of feathers) and pupil dilation following the detection of a threat
and the initiation of fear. Other behavioral indicators of fear include urination, defecation,
vocalization, scanning and vigilance, and grooming (Davis, 1992).

Physiology of tonic immobility
Physiologically, the TI response is different from responses to other tests of fear because
the TI response has been regarded as a reflex (e.g. Klemm, 1976), and because fear does not
cause TI but the duration of TI is related to the degree of fear experienced before TI (reviewed in
Jones, 1986).
The TI response is mediated by areas of the brain that are involved in fear and motor
inhibition including the amygdala (Donatti and Leite-Panissi, 2009; Leite-Panissi et al., 2003),
hypothalamus (Oliveira et al., 1997) and periaqueductal gray matter (PAG) (Monassi et al.,
1999) as well as the reticular formation which is responsible for maintaining the waking state
(Moruzzi and Magoun, 1949; reviewed in Jones, 2008). Although TI is characterized by
temporary motor inhibition, the animal remains conscious (reviewed in Jones, 1986 and Jones
2008).
Gentle et al. (1989) examined electroencephalogram (EEG) activity during TI and
reported that the EEG of adult hens was characterized by high amplitude, slow wave activity
interspersed with low amplitude, fast wave activity during the period of immobility.
Furthermore, hens had reduced neck muscle tone during TI, an elevated heart rate during TI
induction which declined during the immobility period, an elevated shank temperature during the
immobility period (Gentle et al., 1989. In contrast, Nash et al. (1976) reported that heart rate was
lower during TI induction than before induction, and that cloacal temperature was also lower

19

from the onset of TI to TI termination compared to pre-TI levels. More recent results from a
study with rabbits found that heart rate is decreased during TI (Giannico et al., 2014).
Neurotransmitters involved in TI include acetylcholine (Ach) which is involved in
freezing and immobility (reviewed in Monassi et al., 1997 and Jones, 2008; Thompson et al.,
1974; da Silva and Menescal-de-Oliveira, 2006; Leite-Panissi et al., 2003), gamma-aminobutyric
acid (GABA) (reviewed in Jones, 2008; Donatti et al., 2009; Leite-Panissi and Menescal-deOliveira, 2002), and serotonin (Hennig, 1980; Wallnau and Gallup, 1977; Dennis et al., 2013;
reviewed in Clerici and Veneroni, 2011). Increased levels of serotonin are associated with longer
TI durations (e.g. Hill et al., 1994; Dennis et al., 2013), whereas GABA reduces TI duration as
indicated by the injection of GABA agonists in the PAG (Monassi et al., 1999; reviewed in
Clerici and Veneroni, 2011). Cholinergic neurons stimulate activation of the cerebral cortex
during the waking state and during TI (reviewed in Jones, 2008). Furthermore, the cholinergic
system mediates analgesia (inability to feel pain) which occurs during TI (da Silva et al., 2006;
Leite-Panissi et al., 2003). Hormones that are involved in TI include CRF and ACTH. Activation
of CRF in the amygdala of guinea pigs results in increased TI duration (Donatti and LeitePanissi, 2011) and CRF results in the release of ACTH during TI (Donatti and Leite-Panissi,
2011; Farabollini et al., 1990).

Injurious pecking
The majority of research regarding injurious or damaging pecking in poultry has been
conducted with laying hens. The various forms of injurious pecking in laying hens include
feather pecking, defined as the pecking, pulling and sometimes removal of feathers of
conspecifics; cannibalism, which is the pecking and consumption of skin and tissue; and vent

20

pecking, the pecking of the area at the top of the cloaca which usually occurs shortly after the
onset of lay (Savory, 1995). Injurious pecking in turkeys includes head pecking, feather pecking
and cannibalism. Head pecking is more prevalent in males than females (see Dalton et al., 2013),
whereas feather pecking occurs in both male and female turkey flocks. Unlike the other forms of
injurious pecking, head pecking in turkeys is believed to be related to aggressive pecking that is
used to establish or maintain the dominance hierarchy (reviewed in Dalton et al., 2013), whereas
feather pecking and cannibalism have similar manifestations as in laying hens, and are not
related to aggression.
In order to control and reduce injurious pecking, laying hens and turkeys are beaktrimmed within a few days after hatch. With beak trimming, the tip of the beak is removed using
a hot blade or through infra-red beak trimming, which are the two main methods of beak
trimming (reviewed in Nicol et al., 2013). In addition to beak trimming, injurious pecking is also
controlled by reducing the light intensity in barns. However, both methods of controlling
injurious pecking adversely affect well-being. Beak trimming is associated with pain (e.g. Freire
et al., 2008; Jongman et al., 2008; reviewed in Nicol et al., 2013) and reduced light intensity is
associated with altered eye development and reduced activity levels (e.g. Nickla et al., 2001;
reviewed in Nicol et al., 2013).
Feather pecking is the best-studied form of injurious pecking in poultry. Two forms of
feather pecking are recognized, including gentle and severe feather pecking. Gentle feather
pecking causes little damage and may be directed at food or dust particles, seldom eliciting a
reaction from the recipient (Savory, 1995). In contrast, severe feather pecking occurs when a bird
forcefully pecks at and pulls feathers from another bird, resulting in feather damage and feather
loss, and causing the recipient to react or move away (Savory, 1995). Initially, birds that are

21

recipients of feather pecking may respond “by moving away or by confronting the pecker”, but
birds may stop reacting to being pecked as pecking continues (Rodenburg et al., 2013). Severe
feather pecking may lead to skin and tissue damage and cannibalism, which in turn may lead to
mortality (Savory, 1995; reviewed in Rodenburg et al., 2013). Feather pecking is recorded in one
of two ways, either by recording plumage damage (indirectly), or by directly observing bird
behavior to identify individuals performing feather pecking. Plumage scores are often used
because plumage damage is easier to examine than feather pecking behavior (Nicol et al., 2013).

Feather pecking of laying hens
Feather pecking occurs widely among egg laying hen flocks, and feather pecking and
cannibalism are the major causes of mortality in laying hen flocks housed in non-cage systems
(Nicol et al., 2013). Gentle feather pecking is observed as early as one day of age and remains
stable or decreases in prevalence after the rearing period, whereas severe feather pecking is
observed during the laying period (after 18 weeks of age) and usually increases throughout lay
(Nicol et al., 1999; Pötzsch et al., 2001; Lambton et al., 2010; reviewed in Nicol et al., 2013).
Because gentle feather pecking is not associated with much plumage damage (Rodenburg et al.,
2013), most research has been aimed at understanding the causes and development of severe
feather pecking.
In a recent review, Rodenburg et al. (2013) concluded that severe feather pecking is
related to foraging and feeding behavior, in line with the redirected foraging and ground pecking
hypothesis proposed by Blokhuis (1989). Indeed, the importance of access to litter as a foraging
substrate in reducing feather pecking has been confirmed by several studies (reviewed in
Rodenburg et al., 2013). Severe feather pecking may also be related to dietary factors such as the

22

presence of loose feathers in the litter, which leads to feather eating and further feather pecking;
or to the amount of fiber, magnesium, protein and amino acids (lysine, cystine and methionine)
present in the laying hen diet (Al Bustany and Elwinger, 1987a; 1987b; Elwinger et al., 2002;
reviewed in Rodenburg et al., 2013). The maternal environment may also influence feather
pecking behavior in the offspring through the deposition of maternal hormones in the eggs,
which affect development of the brain and behavior of the chick (Freire et al., 2006; de Haas et
al., 2014; reviewed in Rodenburg et al., 2013). Indeed, high maternal levels of corticosterone and
whole-blood serotonin are associated with increased levels of severe feather pecking in chicks
(de Haas et al., 2014). Lastly, individual and genetic factors play important roles in the
development of feather pecking behavior. Feather pecking is known to differ among genetic
strains of hens; indeed, most research regarding laying hen feather pecking is conducted with
two strains that have been divergently selected for high (HFP) and low (LFP) feather pecking,
respectively (Kjaer and Sørensen, 1997; Kjaer et al., 2001).
Several studies have confirmed a link between feather pecking and temperament
characteristics; specifically, fearfulness and the hen’s ability to cope with stress (e.g. Blokhuis
and Beutler, 1992; Jones et al., 1995; Rodenburg et al., 2004; reviewed in Rodenburg et al.,
2013). In general, the performance of feather pecking and the severity of feather damage are
associated with increased levels of fearfulness in various genetic lines of chickens, including in
chickens from high and low feather pecking lines (e.g. Jones et al., 1995; Rodenburg et al., 2010;
reviewed in Rodenburg et al., 2013). Furthermore, genetic lines differing in feather pecking
behavior also differ in physiological responses to stress (hens from the HFP line appear to be
more susceptible to stress), and in serotinergic and dopaminergic neurotransmission in the brain
(van Hierden et al., 2002; 2004; 2005; reviewed in Rodenburg et al., 2013). Indeed, hens from

23

the high feather pecking line have lower serotonin turnover vs. hens from the lower feather
pecking line (van Hierden et al., 2002), and dopamine turnover is lower in chicks from high vs.
low feather pecking lines (van Hierden et al., 2002; 2005; reviewed in Rodenburg et al., 2013).
In conclusion, feather pecking is a multifactorial problem related to early experience,
feeding and foraging behavior, maternal effects, genetics, and the ability to cope with fear and
stress (Rodenburg et al., 2013). Some other factors related to environment and management that
are associated with increased risk of feather pecking in commercial facilities include not
providing access to perches, increased stocking density and group size, more than one or two
dietary changes during the rearing period and more than three dietary changes during the laying
period (reviewed in Nicol et al., 2013).

Feather pecking of turkeys
In contrast to chickens, very little is known about the development and causes of feather
pecking of turkeys, but feather pecking has long been recognized as an important welfare
problem (Hocking, 1993; Marchewka et al., 2013). Feather pecking of turkeys occurs as early as
5 days of age (Moinard et al., 2001) and has been reported to account for up to 58% of culls and
mortalities in a commercial facility (Duggan et al., 2014).
Few studies have systematically examined feather pecking of turkeys, and only two
studies have been conducted since 2006. Most recently, a study examining injurious pecking
among turkeys in a commercial facility (Duggan et al., 2014) demonstrated that the housing
environment significantly affects the prevalence of injurious pecking. Indeed, mortality and
culling was higher and feather condition was worse in turkeys housed in curtain-sided barns
compared to turkeys housed in mechanically ventilated barns where the light intensity was lower

24

(Duggan et al., 2014). Results from another study revealed that feather pecking is affected by
genetic strain and sex (Busayi et al., 2006). Specifically, the frequency of gentle and “strong”
feather pecking was higher among randombred turkeys than among turkeys of a commercial
strain, but damaging pecking was more severe among commercial turkeys. Furthermore, there
were age-dependent sex differences in feather pecking, and more injuries in males than females
(Busayi et al., 2006). Older studies examining feather pecking among turkeys have found that
other factors such as environmental enrichment, light intensity, light source (incandescent vs.
fluorescent) and diet affect feather pecking (reviewed in Dalton et al., 2013).
As with chickens, it has been concluded that injurious pecking of turkeys is consistent
with the redirected foraging hypothesis of feather pecking of chickens, and is a multifactorial
problem that is associated with genetic, environmental and nutritional factors (Dalton et al.,
2013). There is no published literature examining the relationship between fearfulness, stress
reactivity and feather pecking of turkeys.

Stress, temperament and meat quality
Meat quality concerns the attributes of meat that influence consumer acceptance,
including appearance, tenderness, juiciness, flavor (Wilson, 1960) and functionality (Fletcher,
2002). The most important attributes determining meat quality are appearance which includes
color and defects, and texture which is affected by processing procedures and deboning
(Fletcher, 2002). The major factors affecting poultry meat color include heme pigments, such as
myoglobin and haemoglobin; pre-slaughter factors, such as genetic strain, stress before slaughter,
and feed; and the slaughter, chilling and processing procedures (reviewed in Fletcher, 2002).
Color is also affected by pH such that meat with a high pH is darker in color whereas a lower pH

25

is associated with a lighter color (Fletcher, 2002). Color is measured using one of three methods,
including visual assessment, spectrophotometry and reflectance calorimetry (Castigliego et al.,
2012). The CIE (International Commission on Illumination) is the most common calorimetric
method and is used to determine L* which refers to lightness, a* which refers to redness or
greenness, and b* which refers to yellowness or blueness (Castigliego et al., 2012). Texture, on
the other hand, is determined by the “contractile state of the myofibrillar proteins” which is a
factor of the rate at which rigor mortis develops (Fletcher, 2002). Rigor mortis development
affects the ultimate pH and color (Castigliego et al., 2012) and is indirectly measured using the
R-value which is an indicator of the level of adenosine-triphosphate (ATP) breakdown (Calkins
et al., 1982).

Pre-slaughter stress and meat quality
Two conditions associated with stress and meat quality are pale, soft, exudative (PSE)
meat in pigs and dark, firm, dry (DFD) meat in pigs and cattle. PSE meat in pigs results when
pigs experience stress before slaughter, which results in rapid glycolysis, increased lactic acid
production and a decrease in muscle pH (reviewed in Broom and Johnson, 1993) and accelerated
rigor mortis (Sosnicki et al., 1998). These factors result in water loss and pale muscle color. PSElike meat has also been documented in chickens and turkeys (Fletcher, 2002), and has increased
over the years in turkeys (Sosnicki et al., 1998) as turkeys have been selectively bred for higher
growth rates and breast muscle yield (Strasburg and Chaing, 2009).
In contrast to PSE meat, DFD meat results when there is little or no glycogen in the
muscle before slaughter, resulting in low lactic acid production and high pH (see Broom and
Johnson, 1993). High levels of physical activity before slaughter result in DFD meat because

26

glycogen reserves are depleted (see Terlouw, 2005). DFD meat has been extensively studied in
cattle, but DFD-like meat has also been reported in chickens (Fletcher, 2002).

Temperament and meat quality
The relationship between temperament and meat quality is based on the premise that
animals that differ in temperament also differ in stress reactivity, and therefore differ in meat
quality because conditions at slaughter are associated with a number of stressful events,
including catching and crating, transportation, separation from conspecifics and unfamiliarity.
Numerous studies using different breeds of cattle have confirmed that cattle with poor
temperaments (more fearful, flightier animals) have higher stress-susceptibility to procedures
before and at slaughter (e.g. Petherick et al., 2002) and produce meat with a higher pH (e.g. Cafe
et al., 2011; Ribeiro et al., 2012), and darker color (e.g. Cafe et al., 2011; Ribeiro et al., 2012;
Voisinet et al., 1997). Similarly, research with pigs has found that temperament characteristics
such as aggression are correlated with pigs’ responses to a novel object, and pigs that are more
aggressive and that show less fear of a novel object are more reactive to pre-slaughter
procedures, resulting in increased glycolysis and reduced pH (Terlouw, 2005). In contrast, pigs
that approach and touch a human more often during a “human exposure test” administered
during the rearing period exhibit reduced stress reactivity to pre-slaughter procedures, resulting
in lower muscle metabolic activity and reduced glycolysis (Terlouw, 2005).
Very few studies have been conducted to assess whether meat quality is related to
temperament in poultry, and there are no such studies with turkeys. Remingnon et al. (1998)
assessed meat quality (pH and drip loss) in the LTI and STI lines of quail that had been subjected
to physical restraint before slaughter. Their results demonstrated that pH and drip loss were

27

higher in quail from the LTI line. In contrast, Debut et al. (2003) found no relationship between
meat quality and TI of quail assessed one week before slaughter.

CHAPTER OVERVIEW
The chapters of this dissertation examine fearfulness, feather pecking and meat quality of
turkeys and aim to determine whether fearfulness is associated with feather pecking and meat
quality. Chapter 2 examines the behavior of male commercial turkeys in four commonly used
tests of fear. The reliability of each fear test is also evaluated as well as whether behavior
changes over time. In order to examine test reliability, turkeys were repeatedly tested in each fear
test. Chapter 3 addresses the validity of the four fear tests used in Chapter 2. A strategy for
assessing test validity is to determine whether animals behave in the same way across different
test situations; in other words, whether animals that are classified as being more fearful in one
test are also more fearful in other tests relative to other animals. Therefore, test validity was
assessed using inter-test correspondence. Chapter 4 examines feather pecking behavior in male
and female turkeys of a commercial and randombred strain, and examines whether feather
pecking behavior is associated with fearfulness in an open field test. Open field test reliability for
males and females of both strains is examined and discussed in relation to the results presented in
Chapter 2. Lastly, Chapter 5 examines differences in meat quality between commercial and
randombred turkey males, and examines whether meat quality is associated with differences in
open field behavior.

28

REFERENCES

29

REFERENCES
Al Bustany, Z., and K. Elwinger. 1987a. Comparison between barley/fish meal- and
maize/soybean meal-based diets with various lysine and protein levels fed to different
strains of laying hens. Acta Agric. Scand. 37: 41-49.
Al Bustany, Z., and K. Elwinger. 1987b . Response of laying hens to different dietary lysine
intakes. A comparison of some commercial hybrids with strains selected on a low protein
diet. Acta Agric. Scand. 37: 27-40.
Appleby, M. C., J. A. Mench, and B. O. Hughes, eds. 2004. Origins. Pages 5-8 in: Poultry
Behaviour and Welfare. CABI Publishing, Cambridge, MA, USA.
Archer, J. 1973. Tests for emotionality in rats and mice - review. Anim. Behav. 21: 205-235.
Armario, A., N. Daviu, C. Munoz-Abellan, C. Rabasa, S. Fuentes, X. Belda, H. Gagliano, and R.
Nadal. 2012. What can We Know from Pituitary-Adrenal Hormones About the Nature
and Consequences of Exposure to Emotional Stressors? Cell. Mol. Neurobiol. 32: 749758.
Blokhuis, H.J. 1989. The development and causation of feather pecking in the domestic fowl.
Vakgroep-Ethologie, Thesis Landbouw Universiteit Wageningen, Wageningen.
http://edepot.wur.nl/202211. Accessed Jun. 2014.
Blokhuis, H.J. and A. Beutler. 1992. Feather pecking damage and tonic immobility response in
two lines of white leghorn hens. J. of Anim. Sci. 70: 170.
Boissy, A. Fear and fearfulness in animals. 1995. Q. Rev. Biol. 70: 165-191.
Bourin, M., B. Petit-Demoulière, B. N. Dhonnchadha, and M. Hascöet. 2007. Animal models of
anxiety in mice. Fundam. Clin. Pharm. 21: 567–574.
Brambell, F. W. R. 1965. Report of the technical committee to enquire into the welfare of
animals kept under intensive livestock husbandry systems. (Command Rep. 2836).
London: Her Majesty's Stationery Office.
Broom, D. M. 1986. Indicators of poor welfare. Br. Vet. J. 142: 524-526.
Broom, D. M., and K. G. Johnson. 1993. Chapter 5 Assessing animal welfare: short term
responses. Page 106 in Stress and animal welfare. Broom, D. M., and K. G. Johnson, eds.
Kluwer Academic Publishers, Cambridge, MA, USA.
Brown, J. A., C. Dewey, C. F. M. Delange, I. B. Mandell, P. P. Purslow, J. A. Robinson, E. J.
Squires, and T. M. Widowski. 2009. Reliability of temperament tests on finishing pigs in
group-housing and comparison to social tests. Appl. Anim. Behav. Sci. 118: 28-35.
30

Busayi, R. M., C. E. Channing, and P. M. Hocking. 2006. Comparisons of damaging feather
pecking and time budgets in male and female turkeys of a traditional breed and a
genetically selected male line. Appl. Anim. Behav. Sci. 96: 281-292.
Buss, E. G. 1989. Genetics of turkeys: Origin and development. World. Poult. Sci. J. 45:27-52.
Cafe, L. M., D. L. Robinson, D. M. Ferguson, G. H. Geesink, and P. L. Greenwood. 2011.
Temperament and hypothalamic-pituitary-adrenal axis function are related and combine
to affect growth, efficiency, carcass, and meat quality traits in Brahman steers. Dom.
Anim. Endocrinol. 40: 230-240.
Calkins, C. R., T. R. Dutson, G. C. Smith, and Z. L. Carpenter. 1982. Concentration of creatinephosphate, adenine-nucleotides and their derivatives in electrically stimulated and nonstimulated beef muscle. J. Food Sci. 47: 1350-1353.
Cannon, W.B. 1914. The emergency function of the adrenal medulla in pain and the major
emotions. Am. J. Physiol. 33: 356– 372.
Carcia, R. V. and S. Harvey. 2000. Chapter 19 Adrenals. Pages 500-501 in Sturkie’s Avian
Physiology, 5th rev. ed. Whittow, G. C., ed. Academic Press, New York, NY, USA.
Carere, C., D. Caramaschi, and T. W. Fawcett. 2010. Covariation between personalities and
individual differences in coping with stress: Converging evidence and hypotheses. Curr.
Zool. 56: 728-740.
Castanheira, M. F., M. Herrera, B. Costas, L. E. C. Conceicao, and C. I. M. Martins. 2013. Can
We Predict Personality in Fish? Searching for Consistency over Time and across
Contexts. PLoS One 8: 9.
Castigliego, L., A. Armani, and A. Guidi. 2012. Meat color. Pages 82-90 in Handbook of Meat
and Meat Processing, 2nd rev. ed. Hui, Y. H., ed. CRC Press, New York, NY, USA.
Clarke, C. H., and R. B. Jones. 2000. Effects of prior video stimulation on open-field behaviour
in domestic chicks. Appl. Anim. Behav. Sci. 66: 107-117.
Clerici, C. A, and L. Veneroni (eds.) 2011. Chapter I. Pages 13-25 in: The impossible escape:
studies on the tonic immobility in animals from a comparative psychology perspective
(Animal science, issues and professions). Nova Science Publishers Inc. Hauppage, NY,
USA. Available online: http://www.scribd.com/doc/103780538/The-impossible-escapestudies-on-the-tonic-immobility-in-animals-from-a-comparative-psychology-perspective
Cockrem, J. F. 2007. Stress, corticosterone responses and avian personalities. J Ornithol. 148:
S169–S178.

31

Cockrem, J. F. 2013. Corticosterone responses and personality in birds: Individual variation and
the ability to cope with environmental changes due to climate change. Gen. Comp.
Endocr. 190: 156-163.
Coppens, C. M., S. F. de Boer, and J. M. Koolhaas. 2010. Coping styles and behavioural
flexibility: towards underlying mechanisms. Philos. T. Roy. Soc. B. 365: 4021-4028.
Crawford, R. D. 1992. Introduction to Europe and diffusion of domesticated turkeys from the
Americas. Arch. Zootec. 41: 307-314.
Crowe, R., and J. M. Forbes. 1999. Effects of four different environmental enrichment treatments
on pecking behaviour in turkeys. Br. Poult. Sci. 40: S1, 11-12.
Dalton, H. A., B. J. Wood, and S. Torrey. 2013. Injurious pecking in domestic turkeys:
development, causes and potential solutions. World’s Poult. Sci. J. 69: 865-875.
da Silva, L. F. S., and L. Menescal-de-Oliveira. 2006. Cholinergic modulation of tonic
immobility and nociception in the NRM of guinea pig. Physiol. & Behav. 87: 821-827.
Davis, M. 1992. The role of the amygdala in fear and anxiety. Ann. Rev. Neurosci. 15: 353-375.
Debut, M., C. Berri, E. Baeza, N. Sellier, C. Arnould, D. Guemene, N. Jehl, B. Boutten, Y. Jego,
C. Beaumont, and E. Le Bihan-Duval. 2003. Variation of chicken technological meat
quality in relation to genotype and preslaughter stress conditions. Poult. Sci. 82: 1829–
1838.
de Haas, E. N., J. E. Bolhuis, B. Kemp, T. G. G. Groothuis, and T. B. Rodenburg. 2014. Parents
and Early Life Environment Affect Behavioral Development of Laying Hen Chickens.
PLoS One 9: 12.
Dennis, R. L., A. G. Fahey, and H. W. Cheng. 2013. Alterations to embryonic serotonin change
aggression and fearfulness. Aggressive Behav. 31: 91-98.
Donatti A. F., and C. R. A. Leite-Panissi. 2009. GABAergic antagonist blocks the reduction of
tonic immobility behavior induced by activation of 5-HT2 receptors in the basolateral
nucleus of the amygdala in guinea pigs. Brain. Res. Bull. 79: 358–64.
Donatti, A. F., and C. R. A. Leite-Panissi. 2011. Activation of corticotropin-releasing factor
receptors from the basolateral or central amygdala increases the tonic immobility
response in guinea pigs: An innate fear behavior. Biobehav. Brain Res. 225: 23-30.
Duggan, G., T. Widowski, M. Quinton, and S. Torrey. The development of injurious pecking in a
commercial turkey facility. 2014. J. Appl. Poult. Res. 23: 1-11.
Elwinger, K., R. Tauson, M. Tufvesson, and C. Hartmann. 2002. Feeding of layers kept in an
organic feed environment. 11th. European Poultry Conference, Bremen.
32

Farabollini F., F. Facchinetti, C. Lupo, and G. Carli. 1990. Time course of opioid and pituitary–
adrenal hormone modification during the immobility reaction in rabbits. Physiol. Behav.
47: 337–41.
Farm Animal Welfare Council. 1979. Press Statement.
http://www.fawc.org.uk/pdf/fivefreedoms1979.pdf. Accessed Jun. 2014.
Fletcher, D. L., M. Qiao, and D. P. Smith. 2000. The relationship of raw broiler breast meat color
and pH to cooked meat color and pH. Poult. Sci. 79:784–788.
Forkman, B., A. Boissy, M. –C. Maunier-Salaün, E. Canali, and R. B. Jones, R. B. 2007. A
critical review of fear tests used on cattle, pigs, sheep, poultry and horses. Physiol.
Behav. 92: 340-374.
Freire, R., P. C. Glatz, and G. Hinch. 2008. Self-administration of an analgesic does not alleviate
pain in beak trimmed chickens. Asian-Aust. J. Anim. Sci. 21: 443-448.
Freire, R., S. van Dort, and L. J. Rogers. 2006. Pre- and post-hatching effects of corticosterone
treatment on behavior of the domestic chick. Horm. Behav. 49: 157-165.
Gallup, G. G. 1974. Animal hypnosis - factual status of a fictional concept. Psychol. Bull. 81:
836-853.
Gentle, M. J., R. B. Jones, and S. C. Woolley. 1989. Physiological changes during tonic
immobility in Gallus gallus var domesticus. Physiol. & Behav. 46: 843-847.
Giannico, A. T., L. Lima, R. R. Lange, T. R. Froes, and F. Montiani-Ferreira. 2014. Proven
cardiac changes during death‑feigning (tonic immobility) in rabbits (Oryctolagus
cuniculus). J. Comp. Physiol. A. 200:305–310.
Gosling, S. D. 2001. From mice to men: what can we learn about personality from animal
research? Psychol. Bull. 127: 45-86.
Graml., C., S. Waiblinger, and K. Niebuhr. 2008. Validation of tests for on-farm assessment of
the hen-human relationship in non-cage systems. Appl. Anim. Behav. Sci. 111: 301-310.
Gray, J. A. 1979. Emotionality in male and female rodents. A reply to Archer. Br. J. Psychol. 70:
425-440.
Groothuis, T. G. G., and C. Carere. 2005. Avian personalities: characterization and epigenesis.
Neurosci. Biobehav. Rev. 29: 137-150.
Hennig, C. W., J. F. McIntyre, D. D. Moriarty, J. M. Picerno, and J. L. Allen. 1988. Differential
cholinergic influences on the immobility response in various strains of domestic fowl.
Pharmacol. Biochem. Behav. 30: 625-634.

33

Hill, W. L., T. M. Fleming, and E. M. Shrier. 1994. Tonic immobility and high-intensity calls in
a precocial chick as a function of age, diet and time of day. Dev. Psychobiol. 27: 331342.
Hoagland, H. 1928. On the mechanism of tonic immobility in vertebrates. J. Gen. Phys. 11: 715741.
Hocking, P. 1993. Welfare of turkeys. Pages 125-138 in Proceedings of the Fourth European
Symposium on Poultry Welfare. Savory C J and B. O. Hughes, eds. Universities
Federation for Animal Welfare, Potters Bar, UK.
Hughes, B. O., and P. N. Grigor. 1996. Behavioural time-budgets and beak related behaviour in
floor-housed turkeys. Anim. Welf. 5:189-198.
Jacobs, D. S. 2009. Animal personality and biological markets: rise of the individual. Afr. Zool.
44: 271-282.
Jensen, P., M. Johnsson, and D. Wright. 2014. Domestication related anxiety genes identified in
chickens. 48th Congress for the International Society for Applied Ethology. VitoriaGasteiz, Spain. Pp. 134 (Abstr.).
Jones, B. E. 2008. Modulation of cortical activation and behavioral arousal by cholinergic and
orexinergic systems. Ann. N.Y. Acad. Sci. 1129: 26–34.
Jones, R. B. 1986. The tonic immobility reaction of the domestic-fowl - a review. Worlds Poult.
Sci. J. 42: 82-96.
Jones, R. B. 1996. Fear and adaptability in poultry: insights, implications and imperatives.
World’s Poultry Science Journal. 52: 131-174.
Jones, B., and A. Boissy. 2011. Fear and other negative emotions. Page 81 in Animal Welfare.
2nd rev. ed. Appleby, M. C, J. A. Mench, I. A. S. Olsson, and B. O. Hughes, eds. CABI
International, Cambridge, MA, USA.
Jongman, E.C., P. C. Glatz, and J. L. Barnett. 2008. Changes in behaviour of laying hens
following beak trimming at hatch and re-trimming at 14 weeks. Asian-Aust. J. Anim. Sci.
21: 291-298.
Kjaer, J.B., and P. Sørensen. 1997. Feather pecking behaviour in White Leghorns, a genetic
study. Br. Poult. Sci. 38: 333–341.
Kjaer, J.B., P. Sørensen, and G. Su. 2001. Divergent selection on feather pecking behaviour in
laying hens (Gallus gallus domesticus). Appl. Anim. Behav. Sci. 71: 229–239.
Klemm, W. R. 1976. Identity of sensory and motor systems that are critical to the immobility
reflex (“animal hypnosis”). J. Neurosci. Res. 2: 57-69.
34

Koolhaas, J. M., S. F. de Boer, C. M. Coppens, and B. Buwalda. 2010. Neuroendocrinology of
coping styles: Towards understanding the biology of individual variation. Front.
Neuroendocrinol. 31: 307-321.
Koolhaas, J. M., S. M. Korte, S. F. De Boer, B. J. Van Der Vegt, C. G. Van Reenen, H. Hopster,
I. C. De Jong, M. A. W. Ruis, and H. J. Blokhuis. 1999. Coping styles in animals: current
status in behavior and stress-physiology. Neurosci. Biobehav. Rev. 23: 925-935.
Kralj-Fiser, S., B. M. Weiss, and K. Kotrschal. 2010. Behavioural and physiological correlates of
personality in greylag geese (Anser anser). J. Ethol. 28: 363-370.
Lambton, S.L., T. G. Knowles, C. Yorke, and C. J. Nicol. 2010. The risk factors affecting the
development of gentle and severe feather pecking in loose housed laying hens. Appl.
Anim. Behav. Sci. 123: 32-42.
LeDoux, J. 2003. The Emotional Brain, Fear, and the Amygdala. Cellular and Molecular
Neurobiology. 23: 727-738.
Leite-Panissi, C. R. A., N. C. Coimbra, and L. Menescal-de-Oliveira. 2003. The cholinergic
stimulation of the central amygdala modifying the tonic immobility response and
antinociception in guinea pigs depends on the ventrolateral periaqueductal gray. Brain
Res. Bull. 60: 167–178
Leite-Panissi, C. R. A., A. A. Ferrarese, A. L. B. Terzian, and L. Menescal-de-Oliveira. 2006.
Serotoninergic activation of the basolateral amygdala and modulation of tonic immobility
in guinea pig. Brain Res. Bull. 69: 356-364.
Leite Panissi C. R. A., and L. Menescal-de-Oliveira. 2002. Central nucleus of the amygdala and
the control of tonic immobility in guinea pigs. Brain Res. Bull. 58: 13–9.
Lewis, P. D., G. C. Perry, and C. M. Sherwin. 1998. Effect of photoperiod and light intensity on
the performance of intact male turkeys. Anim. Sci. 66: 759-767.
Marchewka, J., T. T. N. Watanabe, V. Ferrante, and I. Estevez. 2013. Review of the social and
environmental factors affecting the behavior and welfare of turkeys (Meleagris
gallopavo). Poult. Sci. 92:1467-1473.
Martrenchar, A., D. Huonnic, and J. P. Cotte. 2001. Influence of environmental enrichment on
injurious pecking and perching behaviour in young turkeys. Br. Poult. Sci. 42: 161-170.
Martin, P., and P. Bateson, eds. 1993. The reliability and validity of measures. Pages 114-116 in
Measuring behaviour: an introductory guide. 2nd rev. ed. Cambridge University Press,
New York, NY, USA.

35

Martins, C. I. M., P. I. M. Silva, L. E. C. Conceicao, B. Costas, E. Hoglund, O. Overli, and J. W.
Schrama. 2011. Linking Fearfulness and Coping Styles in Fish. PLoS One 6: 8.
Matteri, R. L., J. A. Carroll, and C. J. Dyer. 2000. Neuroendocrine responses to stress. Pages 4376 in Biology of Animal stress - Basic Principles and Implications for Animal Welfare.
Moberg, G. P., and J. A. Mench, J. A., eds. CABI Publishing, New York, NY, USA.
Mazurek, M., M. McGee, M. A. Crowe, D. J. Prendiville, X. Boivin, and B. Earley. 2011.
Consistency and stability of behavioural fear responses of heifers to different feareliciting situations involving humans. Appl. Anim. Behav. Sci. 131: 21-28.
Mellor, D. J., C. J. Cook, and K. J. Stafford. 2000. Quantifying some responses to pain as a
stressor. Page 174 in Biology of Animal stress - Basic Principles and Implications for
Animal Welfare. Moberg, G. P., and J. A. Mench, J. A., eds. CABI Publishing, New
York, NY, USA.
Moinard, C., P. D. Lewis, G. C. Perry, and C. M. Sherwin. 2001. The effects of light intensity
and light source on injuries due to pecking of male domestic turkeys (Meleagris
gallopavo). Anim. Welfare 10: 131-139.
Monassi, C. R., A. Hoffman, and L. Menescal-de-Oliveira. 1997. Involvement of the cholinergic
system and periaqueductal gray matter in the modulation of tonic immobility in the
guinea pig. Physiol. & Behav. 62: 53-59.
Monassi, C. R., C. R. Leite-Panissi, and L. Menescal-de-Oliveira. 1999.
Ventrolateralperiaqueductal gray matter and the control of tonic immobility. Brain Res.
Bull. 50: 201-8.
Moruzzi, G. and H. W. Magoun. 1949. Brain stem reticular formation and activation of the EEG.
Electroencephalogr. Clin. Neurophysiol. 1: 455–473.
Nash, R. F., G. G. Gallup, and D. A. Czech1976. Pshychophysiological correlates of tonic
immobility in the domestic chicken (Gallus gallus). Physiol. & Behav. 17: 413-418.
Neumann, I. D., G. Wegener, J. R. Homberg, H. Cohen, D. A. Slattery, J. Zohar, J. D. A. Olivier,
and A. A. Mathe. 2011. Animal models of depression and anxiety: What do they tell us
about human condition? Prog. Neuro-Psychoph. 35: 1357-1375.
Nickla, D.L., C. F. Wildsoet, and D. Troilo. 2001. Endogenous rhythms in axial length and
choroidal thickness in chicks: Implications for ocular growth regulation. Invest.
Ophthalmol. Vis. Sci. 42: 584-588.
Nicol, C.J., N. G. Gregory, T. G. Knowles, I. D. Parkman, and L. J. Wilkins. 1999. Differential
effects of increased stocking density, mediated by increased flock size, on feather
pecking and aggression in laying hens. Appl. Anim. Behav. Sci. 65: 137-152.

36

Nicol, C. J., M. Bestman, A. M. Gilani, E. N. De Haas, I. C. De Jong, S. Lambton, J. P.
Wagenaar, C. A. Weeks, and T. B. Rodenburg. 2013. The prevention and control of
feather pecking: application to commercial systems. Worlds Poult. Sci. J. 69: 775-788.
Noble, D. O., J. W. Anderson, and K. E. Nestor. 1996. Range and confinement rearing of four
genetic lines of turkeys. 2. Effects on behavior and tonic immobility. Poult. Sci. 75: 165171.
Oliveira L, A. Hoffmann, and L. Menescal-de-Oliveira. 1997. Participation of the medial and
anterior hypothalamus in the modulation of tonic immobility in guinea pigs. Physiol.
Behav. 62: 1171–8.
Owens, C. M., and A. R. Sams. 2000. The influence of transportation on turkey meat quality.
Poult. Sci. 79: 1204-1207.
Petherick, J. C., R. G. Holroyd, V. J. Doogan, and B. K. Venus. 2002. Productivity, carcass and
meat quality of lot-fed Bos indicus cross steers grouped according to temperament. Aust.
J. Exp. Agr. 42: 389-398.
Pötzsch, C.J., K. Lewis, C. J. Nicol, and L. E. Green. 2001. A cross-sectional study of the
prevalence of vent pecking in laying hens in alternative systems and its associations with
feather pecking, management and disease. Appl. Anim. Behav. Sci. 74: 259-272.
Raubek, J., K. Niebuhr, and S. Waiblinger. 2007. Development of on-farm methods to assess the
animal-human relationship in laying hens kept in non-cage systems. Anim. Welfare. 16:
173-175.
Reiner, A., D. J. Perkel, L. L. Bruce, A. B. Butler, A. Csillag, W. Kuenzel, L. Medina, G.
Paxinos, T. Shimizu, G. Striedter, M. Wild, G. F. Ball, S. Durand, O. Gunturkun, D. W.
Lee, C. V. Mello, A. Powers, S. A. White, G. Hough, L. Kubikova, T. V. Smulders, K.
Wada, J. Dugas-Ford, S. Husband, K. Yamamoto, J. Yu, C. Siang, and E. D. Jarvis. 2004.
Revised nomenclature for avian telencephalon and some related brainstem nuclei. J.
Comp. Neurol. 473: 377-414.
Remignon, H., A. D. Mills, D. Guemene, V. Desrosiers, M. Garreau-Mills, M. Marche, and G.
Marche. 1998. Meat quality traits and muscle characteristicsin high or low fear lines of
Japanese quails (Coturnix japonica) subjected to acute stress. Br. Poult. Sci. 39: 372-378.
Ribeiro, J. d. S., T. d. M. Goncalves, M. M. Ladeira, R. R. Tullio, F. R. Campos, J. A. Garcia
Bergmann, O. R. Machado Neto, and J. R. Reis de Carvalho. 2012. Reactivity,
performance, color and tenderness of meat from Zebu cattle finished in feedlot. Rev.
Bras. Zootecn. 41: 1009-1015.
Rodenburg, T.B., A. J. Buitenhuis, B. Ask, K. A. Uitdehaag, P. Koene, J. J. Van Der Poel, J. A.
M. Van Arendonk, and H. Bovenhuis. 2004. Genetic and phenotypic correlations

37

between feather pecking and open-field response in laying hens at two different ages.
Behav. Genet. 34: 407-415.
Rodenburg, T.B., E. N. de Haas, B. L. Nielsen, and A. J. Buitenhuis. 2010. Fearfulness and
feather damage in laying hens divergently selected for high and low feather pecking.
Appl. Anim. Behav. Sci. 128: 91-96.
Rodenburg, T. B., M. M. van Krimpen, I. C. de Jong, E. N. de Haas, M. S. Kops, B. J. Riedstra,
R. E. Nordquist, J. P. Wagenaar, M. Bestman, and C. J. Nicol. 2013. The prevention and
control of feather pecking in laying hens: identifying the underlying principles. Worlds
Poult. Sci. J. 69: 361-373.
Rodrigues, S. M., J. E. LeDoux, and R. M. Sapolsky. 2009. The influence of stress hormones on
fear circuitry. Annu. Rev. Neurosci. 32: 289–313.
Sarica, M., N. Ocak, S. Turhan, C. Kop, and U. S. Yamak. 2011. Evaluation of meat quality from
3 turkey genotypes reared with or without outdoor access. Poult. Sci. 90: 1313-1323.
Savory, C. J. 1995. Feather pecking and cannibalism. Worlds Poult. Sci. J. 51: 215-219.
Selye, H., 1932. The general adaptation syndrome and the diseases of adaptation. J. Clin.
Endocrin. 6: 117–152.
Sherwin, C. M., P. D. Lewis, and G. C. Perry. 1999. Effects of environmental enrichment,
fluorescent and intermittent lighting on injurious pecking amongst male turkey poults.
British Br. Poult. Sci. 40:592-598.
Shin, L. M., and I. Liberzon. 2010. The neurocircuitry of fear, stress and anxiety disorders.
Neuropsychopharm. Rev. 35: 169-191.
Sih, A., A. Bell, and J. C. Johnson. 2004. Behavioral syndromes: an ecological and evolutionary
overview. Trends Ecol. Evol. 19: 372-378.
Sosnicki, A. A., M. L Greaser, M. Pietrzak, E. Pospiech, and V. Sante.1998. PSE-like syndrome
in breast muscle of domestic turkeys: a review. J. Muscle Foods. 9: 13-23.
Stamps, J., and T. G. G. Groothuis. 2010. The development of animal personality: relevance,
concepts and perspectives. Biol. Rev. 85: 301-325.
Strasburg, G. M. and W. Chiang. 2009. Pale, soft, exudative turkey-the role of ryanodine
receptor variation in meat quality. Poult. Sci. 88: 1497-1505.
Suarez, S.D. and G. G. Gallup. 1983. Social reinstatement and open field testing in chickens.
Anim. Learn. Behav. 11: 119-126.

38

Terlouw, C. 2005. Stress reactions at slaughter and meat quality in pigs: genetic background and
prior experience - A brief review of recent. Livest. Prod. Sci. 94: 125-135.
Thompson, R. W.; J. Piroch, D. Fallen, and D. Hatton. 1974. A central cholinergic inhibitory
system as a basis for tonic immobility (animal hypnosis) in chickens. J. Comp. Physiol.
Psychol. 87: 507-512.
United States Department of Agriculture National Agricultural Statistics Service. 2014. Poultry –
production and value. USDA, Washington, DC, USA.
http://www.nass.usda.gov/Publications/Todays_Reports/reports/plva0414.pdf. Accessed
Jun. 2014.
Updike, M. S., H. N. Zerby, M. S. Sawdy, G. Kaletunc, and M. P. Wick. 2005. Turkey breast
meat functionality differences among turkeys selected for body weight and/or breast
yield. Meat Sci. 71:706–712.
van Hierden, Y.M., J. M. Koolhaas, and S. M. Korte. 2004. Chronic increase of dietary tryptophan decreases gentle feather pecking behaviour. Appl. Anim. Behav. Sci. 89: 7184.
van Hierden, Y.M., J. M. Koolhaas, L. U. Kost'al, P. Vyboh, M. Sedlackova, M. Rajman, M.
Jurani, and M. S. Korte. 2005. Chicks from a high and low feather pecking line of laying
hens differ in apomorphine sensitivity. Physiol. .Behav. 84: 471-477.
van Hierden, Y.M., S. M. Korte, E. W. Ruesink, C. G. van Reenen, B. Engel, G. A. H.
Kortebouws, J. M. Koolhaas, and H. J. Blokhuis. 2002. Adrenocortical reactivity and
central serotonin and dopamine turnover in young chicks from a high and low featherpecking line of laying hens. Physiology & Behavior 75: 653-659.
Voisinet, B. D., T. Grandin, S. F. O’Connor, J. D. Tatuma, and M. J. Deesing. 1997. Bos
Indicus-cross feedlot cattle with excitable temperaments have tougher meat and a higher
incidence of borderline dark cutters. Meat Sci. 4: 367-377.
Waiblinger, S., X. Boivin, V. Pedersen, M. V. Tosi, A. M. Janczak, E. K. Visser, and R. B.
Jones. 2006. Assessing the human-animal relationship in farmed species: A critical
review. Appl. Anim. Behav. Sci. 101: 185-242.
Wallnau, L. B., and G. G. Gallup. 1977. A serotonergic, midbrain-raphe model of tonic
immobility. Biobehav. Rev. 1: 35-43.
Wechsler, B. 1995. Coping and coping strategies: a behavioural review. Appl. Anim. Behav. Sci.
43: 123-134.
Welfare Quality®. 2009a. Welfare Quality® Assessment Protocol for Poultry. Welfare Quality
Consortium, Lelystad, Netherlands.

39

Welfare Quality®. 2009b. Welfare Quality® Assessment Protocol for Pigs. Welfare Quality
Consortium, Lelystad, Netherlands.
Welfare Quality®. 2009c. Welfare Quality® Assessment Protocol for Cattle. Welfare Quality
Consortium, Lelystad, Netherlands.
Werner, C., J. Riegel, and M. Wicke. 2008. Slaughter performance of four different turkey
strains, with special focus on the muscle fiber structure and the meat quality of the breast
muscle. Poult. Sci. 87:1849-1859.
Wilson, G. D. 1960. Chapter 8 quality factors in meat and meat foods. Page 259 in The Science
of Meat and Meat Products. Salisbury, G. W., and E. W. Crampton, eds. W. H. Freeman
and Company, San Francisco, CA, USA.

40

CHAPTER 2: TEMPERAMENTAL TURKEYS: RELIABILITY OF BEHAVIORAL
RESPONSES TO FOUR TESTS OF FEAR
ABSTRACT
Open field (OF), tonic immobility (TI), voluntary approach (VA) and novel object (NO)
tests are used to assess fear responses, activity levels and coping styles of poultry. Fear tests are
also used as part of welfare assessment programs. Little is known about fear responses of
turkeys. Specifically, it is unknown whether turkeys’ fear responses are reliable which is
important when developing tests for assessing welfare. This study examined the short- (between
days) and long-term (between weeks) changes in, and test-retest reliability of, turkeys’ fear
responses. Male commercial turkeys were housed in groups of four to six in 16 pens. Turkeys
were individually tested in OF (n = 60) and TI (n = 66) tests. VA and NO tests were
administered to groups of turkeys in their pens (n = 16). Turkeys were tested three times in each
test. The first (Period 1) and second (Period 2) tests were administered on consecutive days
between 4 and 6 weeks, and the third test (Period 3) between 8 and 10 weeks. Other than
increased sitting and reduced standing during OF testing and more birds approaching and
pecking the observer during VA testing (Period 2 vs. 1), frequencies of responses did not differ
between test periods. However, test responses (e.g. latencies to ambulate, vocalize, approach and
peck) differed between periods depending on the fear test used. All VA test measures differed
between periods, whereas most TI test measures did not. Most OF and NO test measures differed
between weeks, but not between days. Except for the number of vocalizations (rS = 0.39), most
OF test responses were moderately reliable (rS > 0.40). The latency to vocalize (rS ≥ 0.51) and
number of vocalizations (rS ≥ 0.59) were the most reliable TI test measures, whereas reliability

41

of TI duration (rS ≤ 0.31) was low. All VA test measures were moderately to highly repeatable
between days and weeks (e.g. latency to peck: rS ≥ 0.67). Reliability was lowest for the NO test,
where only the latency to peck was moderately repeatable (rS = 0.61). These findings suggest
that although some fear responses of turkeys change over time and with repeated testing, most
OF, TI and VA responses are reliable between days and weeks. However, few NO test measures
were reliable under the conditions of this study. Further research is needed to assess the validity
of OF, TI, VA and NO tests for assessing fear responses of turkeys.

INTRODUCTION
Fear and fearfulness have been major topics of poultry behavior and welfare research
since 1965 with the release of the Brambell Report (Brambell, 1965) and the development of the
Five Freedoms (Farm Animal Welfare Council, 1979). Fear is defined as the “reaction to the
perception of actual danger” (Forkman et al., 2007), and fearfulness is defined as the
susceptibility of an individual to being frightened (Boissy, 1995; Jones, 1996). Not only is fear a
negative affective state that is indicative of suffering, but it is associated with numerous adverse
effects on productivity and welfare (reviewed in Jones, 1996), such as increased injury (Reed et
al., 1993), reduced production (Hemsworth et al., 1994; Cransberg et al., 2000; de Haas et al.,
2013), depressed growth (Jones, 2002; Schütz et al., 2004), and feather pecking (Vestergaard et
al., 1993; de Haas et al., 2010).
Underlying fearfulness cannot be measured directly because it is a subjective state.
Therefore, inferences about a bird’s level of fearfulness are based on the bird’s responses to
situations believed to induce fear. Fearful birds exhibit more intense fear responses compared to

42

less fearful birds (Jones, 1996). Moreover, birds that show greater fear responses have greater
physiological responses to other stressors (Jones, 1989; Calandreau et al., 2011), which affects
the bird’s ability to cope with its environment. Fearfulness is considered by some to be a
characteristic associated with coping style (Cockrem, 2007), which is defined as “a coherent set
of behavioral and physiological stress responses which is consistent over time and which is
characteristic to a certain group of individuals” (Koolhaas et al., 1999). However, others consider
fearfulness and coping styles to be separate dimensions of personality (Jones and Boissy, 2011),
where coping styles are discussed in relation to stress physiology (Groothuis & Carrere, 2005)
and fear is associated with emotional reactivity which is distinct from coping style (Koolhaas et
al., 1999). Irrespective of whether fearfulness is a component of coping style or an independent
dimension associated with emotion, how the animal responds to environmental challenges affects
the animal’s welfare (Broom, 1988). To this end, fear tests have been incorporated into welfare
assessment programs, such as the Welfare Quality® Assessment Protocols for poultry (Welfare
Quality®, 2009a), pigs (Welfare Quality®, 2009b) and cattle (Welfare Quality®, 2009c).
The most common types of fear tests for poultry involve exposure to predatory stimuli
(e.g. human approach or avoidance), restraint (e.g. tonic immobility), or novelty (e.g. novel
object), or all three. However, fear tests are only usable if they are reliable. Reliable tests have
few random errors and are consistent over repeated applications (Martin and Bateson, 1993).
Reliability is distinct from validity, which refers to accuracy, freedom from systematic errors;
specificity, whether the measurement measures what it is supposed to; and scientific validity,
whether the measurement provides information relevant to the questions being investigated
(Martin and Bateson, 1993). Forkman et al. (2007) assessed the reliability and validity of a
number of fear tests used for various domestic farm animal species, including hens and quail.
43

Open field, tonic immobility and novel object tests are reliable and valid tests of fear for hens
and quail (Forkman et al., 2007). Less information is available regarding birds’ fear responses to
humans, but Raubek et al. (2007) concluded that repeatability between consecutive days is high
(rS > 0.80) for the avoidance distance, stationary person and the touch test (performed in the litter
area). Furthermore, Graml et al. (2008) demonstrated the validity of the tests used by Raubek et
al. (2007) for assessing the reactions of hens toward humans.
In contrast to chickens and quail, there is scant research examining the welfare of turkeys
(Marchewka et. al., 2013). Research examining fear or fearfulness per se is nearly nonexistent,
even though turkey production has increased dramatically over the past three decades. Fear
responses of turkeys have not been described or characterized, there is no information available
regarding the reliability of any fear tests for turkeys, and results from studies with other poultry
species may not be transferrable to turkeys. With increasing public interest in animal welfare and
development of animal welfare assessment programs, there is an urgent need for information
regarding fear and fearfulness of turkeys. The objectives of this study were to describe turkeys’
responses to four tests of fear (tonic immobility, open field, novel object and voluntary approach
tests) and to assess the short- and long-term reliability of the four tests.

MATERIALS AND METHODS
All procedures were approved by the Institutional Animal Care and Use Committee of
Michigan State University.

44

Animals and housing
Commercial male turkeys (Hybrid Converter) were brooded in groups of 24 in one of
four littered (wood shavings) pens (2.40 m x 3.05 m) from 1 day to 2 weeks of age. At 2 weeks,
birds were distributed throughout the barn and randomly assigned to one of 16 littered pens (2.40
m x 3.05 m) in the same barn and housed in groups of four to six where they remained until
study completion at 14 weeks. Poults were brooded under heat lamps (35 °C to 37 °C) for the
first 7 days. After heat lamps were removed, the temperature was reduced to, and maintained at
30 °C for 7 days. Thereafter, the temperature was gradually reduced by 1 to 3 °C per week over a
period of 6 weeks to reach a final temperature of between 13 and 18 °C. A photoperiod of
24L:0D was applied for the first 7 days. Thereafter, the photoperiod was reduced by 1 hr per day
for 7 days. A photoperiod of 16L:8D was maintained from 14 days to 14 weeks. Lights came on
at 06:00 h and were turned off at 22:00 h. Light intensity was 20 lx for the first 10 days and then
reduced to 5 lx for the remainder of the study.
Turkeys were fed a commercial starter diet (Blue Seal® Multi Flock Game Starter, Kent
Nutrition Group, Muscatine, IA, USA) from 1 to 8 weeks and a commercial grower diet (Blue
Seal® Multi Flock Starter/Grower, Kent Nutrition Group, Muscatine, IA, USA) from 8 to14
weeks. At 2 weeks, turkeys were individually identified using blue non-toxic livestock marker
(Prima Tech marking stick, Neogen Corporation, Lansing, MI, USA) which was applied to a
different area for each bird in the pen (e.g. shoulders, left or right wing) and was visible on
overhead cameras. Marker was reapplied biweekly to ensure markings remained visible. Body
weight was recorded the day prior to testing at 4 weeks and 8 weeks.

45

Test procedures
For TI and OF tests, birds were individually removed from their home pens and tested in
another room. VA and NO tests were conducted with groups of turkeys in their home pens to
reduce confounds associated with environmental novelty, and because NO tests are used to
assess welfare in groups of birds as part of welfare assessment programs (e.g. Welfare Quality®,
2009a). The same observer administered NO, VA and TI tests. Three different observers
conducted OF tests during Periods 1 and 2. The same observer conducted all OF tests during
Period 3. All observers wore blue coveralls and blue shoe covers. For OF and TI testing, turkeys
TM

were placed in plastic Rubbermaid® Roughneck

bins (Rubbermaid®, Mogadore, OH, USA)

measuring 67.3 (length) x 47.2 (width) x 43.4 (height) or 81.3 (length) x 50.8 (width) x 47.8
(height) containing a small amount of litter on the bottom and carried to and from testing areas.

Test schedule
Turkeys in each pen were tested three times (three test periods) in all four tests to
examine test-retest reliability. Pens were randomly assigned to one of four test groups consisting
of four pens. To control for possible carryover effects and to balance test order across test
groups, each test group was randomly assigned to a particular test sequence (Table 2.1). Each
test group was tested in only one test on a particular test day. In order to examine repeatability
between days (Period 1 vs. 2), birds were tested in the same test (e.g. TI) over 2 consecutive days
(Period 1 = day 1, Period 2 = day 2), followed by 2 days of no testing, followed by the next type
of test (e.g. OF test) which was administered twice over 2 consecutive days (day 5 and day 6).
This pattern of testing was used to control for possible carryover effects between test types.
Therefore, birds were tested in all four tests over a period of 2 weeks (4 to 6 weeks of age).
46

Testing was then repeated 2 weeks later (Period 3) but birds were then tested only once in each
test (3 days between test types), so that the first and third tests (Period 1 vs. 3) could be
compared to assess repeatability between weeks. For example, birds in Group 1 were tested in
the VA test at 28 days (Period 1) and 29 days (Period 2) to examine repeatability between days,
and again at 56 days (Period 3) to examine repeatability between weeks (Period 1 vs. 3), and
then in the NO test at 32 days (Period 1), 33 days (Period 2), and 60 days (Period 3, Table 2.1).
Test order for each bird was randomized for each day of testing. All testing took place between
07:45 h and 17:45 h.

Test measures
Open field (OF) test. Turkeys were individually moved to an empty room adjacent to the
room in which their home pens were located for testing. The OF test arena consisted of a square
arena (length: 2.74 m, width: 2.74 m, height: 1.83 m) with a concrete floor. The testing arena
was enclosed by solid black walls (1.22 m high) with grey mesh netting (0.61 m high) above the
2

solid black sections. Black electrical tape was used to create a grid of 81 squares (each 0.30 m )
on the floor of the test arena. Each bird was placed in the centre of the arena for 10 min. and
behavior was recorded in real time (60i fields/s) from two overhead high definition camcorders
(VIXIA HF M41, Canon USA, Inc. Melville, NY, USA). Behavioral data (latency to ambulate
(s), latency to vocalize (s), number of vocalizations, number of squares entered, number of steps
taken and escape attempts) were collected from video recordings. Ambulation was defined as
two or more steps in rapid (within 4 s) succession. A bird was considered to have entered a
square if at least 67% of both of the bird’s feet were in the same square. An escape attempt was

47

defined as the bird attempting to jump and/or fly out of the test arena. Test arena walls were too
high to permit successful escapes.
Tonic immobility (TI). Turkeys were individually tested in an empty pen in an empty
room adjacent to the room in which their home pens were located. The observer restrained each
turkey using methods similar to those of Noble et al. (1996a). Briefly, the turkey was placed on a
table covered with a cloth sheet. The observer restrained the turkey on its left side by placing the
left hand over the turkey’s right wing and gently holding the turkey’s legs with the right hand.
The turkey was restrained in this manner for 15 s (induction period) and then released. If the
turkey righted itself within 15 s after being released, the turkey was restrained twice more at
most. If tonic immobility was not induced after a third attempt, the test was terminated and the
turkey was recorded as not being inducible. If the turkey remained on its side after being
induced, it was observed until it righted itself, or up to a maximum of 13 min. after which the test
was terminated and the turkey was returned to its home pen. The number of inductions, number
of vocalizations, latency to vocalize (s) and TI duration (s) were recorded.
Voluntary approach (VA) test. Turkeys’ responses to an observer in their home pen were
recorded for 15 min. The same observer conducted all tests. The observer entered the pen and
stood against the wall across from the pen doors and remained motionless for the duration of the
test. Turkeys’ behavior during testing was recorded in real time (60i fields/s) with an overhead
camera (2MB-70IR42L210 Jaguar Series infrared CCTV camera, 2M® Technology, Grand
Prairie, TX, USA). Turkeys’ responses, including latencies (s) to move within two body lengths,
one body length, to approach, and to peck the person, were determined from video analyses. A
body length was defined as the distance from the bird’s shoulders to the base of its tail

48

(excluding tail feathers), and an approach was defined as the bird being less than one body length
away and within pecking distance.
Novel object (NO) test. Procedures were similar to those for the VA test, with the exception
that one of four novel objects, including a deflated beach ball, yellow bucket, wooden rod with
colored tape similar to that used in the Welfare Quality® Assessment Protocol for Poultry
(Welfare Quality®, 2009a), and six plastic balls strung together (two each painted red, white
and blue), was placed in the center of the pen. Similar to studies with cattle (Gibbons et al.,
2009), quail (Miller et al., 2005) and chickens (Hocking et al., 2001), different objects were used
because objects could not be considered novel upon repeated exposure. Therefore, objects were
balanced across test repeats to prevent habituation and maintain object novelty. Novel objects
were balanced within test groups so that each pen in a particular test group was tested with a
different object in each test period. In this manner, every object was represented on every test
day, and each pen was only exposed to a particular novel object once. Each pen was therefore
tested with three of the four objects (one object for each test period). Behavior during testing was
recorded with an overhead camera from which behavioral data were collected. Cameras and
behavioral measures that were recorded were the same as for VA testing.

Statistical analysis
All statistical analyses were performed using SAS version 9.2 (SAS Institute Inc., Cary,
NC). For OF (n = 60) and TI (n = 66) tests, the experimental unit was the individual turkey. For
NO and VA tests, the experimental unit was the pen (n = 16). Data were not normally distributed
and transformations were ineffective. Levene’s test for homogeneity of variance indicated that

49

variance was not equal. Body weight was included as a covariate in all analyses. Data consisted
of binomial responses and frequencies, counts and latencies.
Binomial (yes/no) responses in the OF and TI tests were analyzed using the GENMOD
procedure for a binomial distribution with the logit link function (e.g. Rodenburg et al., 2010).
Turkey nested within pen was included as a repeated measure. Frequencies of birds in each pen
that moved within two body lengths, one body length and that approached and pecked during NO
and VA testing were compared between test periods using the GENMOD procedure for a
poisson distribution. Pen was included as a repeated measure. For the NO test, object was also
included as a covariate to account for differences between novel objects.
A characteristic of count data (e.g. number of steps or vocalizations) is the occurrence of
an excess of zeros. In order to account for excess zeros, a zero-inflated Poisson model (Littell et
al., 2006) was used to compare count variables between test periods (NLMIXED procedure). The
parameter corresponding to pen was included in the model to account for similarities among
turkeys from the same pen. The random intercept corresponding to the individual turkey was
included in the model to account for repeated measures on the same individuals. For the TI test,
the number of inductions was compared between test periods using the GENMOD procedure for
a poisson distribution. Turkey nested within pen was included as a repeated measure.
Because fear test responses were only observed and recorded within a finite period of
time (OF test: 10 min., TI: 13 min., and NO and VA tests: 15 min.), data were censored
(truncated) such that many observations had values equal to the maximum test time allotted. For
example, birds that did not ambulate or vocalize during OF testing were assigned a value of 600
s for latency to ambulate/vocalize. In order to account for the censored, non-normal data, the
nonparametric LIFETEST procedure was used to compare latency measures (e.g. latency to

50

ambulate or vocalize) between test periods. Similar analyses have recently been used to examine
censored observations in grey parrots (Van Zeeland et al., 2013) and grasshoppers (Karpestam et
al., 2012). For the TI and OF tests, latencies from each individual bird were included in the
analysis. Pen was specified in the STRATA statement, period was specified as the GROUP
variable to compare latencies between periods while adjusting for pen differences, and weight
was included as a covariate in the TEST statement. For the VA and NO tests, pen averages were
used because pen was the experimental unit. Therefore, the average latencies to move within two
body lengths, one body length, to approach and peck the object were calculated for each pen.
Period was specified in the STRATA statement and weight was included as a covariate in the
TEST statement. In order to determine whether responses during NO testing depended on which
object was used, the LIFETEST procedure was used to compare latency measures between
objects. Period was specified in the STRATA statement and object was specified as the GROUP
variable to compare latencies between objects while adjusting for test period differences. The
LIFETEST procedure calculates two test statistics (Log-rank, which places greater emphasis on
differences in latencies occurring later in time, and Wilcoxon, which places greater emphasis on
differences in latencies occurring earlier in time; SAS/STAT User’s Guide, 2012) to test the
effect of test period on the response variable (e.g. latency to ambulate). Both test statistics are
2

reported. Results are presented as medians, 25-75% quartiles, χ values for the log-rank and
Wilcoxon test statistics, and the P value.
To assess test-retest reliability within each fear test, the Spearman rank correlation
coefficient was used to assess correlations between Period 1 vs. Period 2 and Period 1 vs. Period
3. Correlation coefficients were defined as low: 0.2 – 0.4, moderate: 0.4 – 0.7, high: 0.7 – 0.9,

51

and very high: 0.9 – 1.0 (Martin and Bateson, 1993). P values < 0.05 were regarded as
significant.

RESULTS
Mean (± SE) body weights were 1.68 ± 0.03 kg for Period 1 and 6.15 ± 0.08 kg for
Period 3. For each type of fear test, only birds from which data for all three test periods had been
collected were included in analyses.

Changes in fear responses over time and with repeated testing
Open field test. Turkeys’ OF responses did not differ between observers. Analyses of
binomial responses indicated that the percentages of turkeys sitting and standing differed
between days (Period 1 vs. 2), but there were no other differences between test periods in any
other OF responses (Table 2.2). Some turkeys attempted to escape the testing arena during
Period 1 (13%) and Period 2 (8%), but no attempts were made in Period 3 (Table 2.2). The
numbers of steps taken, squares entered and vocalizations produced increased over time and with
repeated testing (Table 2.3). The latencies to ambulate and vocalize did not differ between days
(Table 3), but were shorter for Period 3 than Period 1 (Table 2.3).
Tonic immobility. There was a trend for a higher percentage of turkeys to defecate in
Period 3 (18%) vs. Period 1 (14%, Table 2.2). The number of vocalizations produced differed
between test periods (Table 2.3), but no differences were found in the duration of TI or number
of inductions required to induce TI.
Voluntary approach test. The percentage of turkeys that approached and pecked the
person increased from Period 1 to Period 2 (Table 2.2) and latencies to move within two body

52

lengths, one body length, approach and peck decreased with each consecutive test period (Table
2.3).
Novel object test. NO test responses did not differ between objects for Period 1 and Period 2.
However, when responses were compared between Period 1 and Period 3, latency to move
within two body lengths (LIFETEST; Log-rank = 8.8, P = 0.03, Wilcoxon = 5.5, P = 0.1)
differed depending on which object was used, and there were tendencies for the latency to move
within one body length (LIFETEST; Log-rank = 7.1, P = 0.07, Wilcoxon = 5.3, P = 0.2) and
latency to approach (LIFETEST; Log-rank = 7.5, P = 0.06, Wilcoxon = 7.6, P = 0.06) to differ
between Periods 1 and 3 depending on the object used.
There was a tendency for the percentages of turkeys that moved within one body length
and that approached the novel object to be higher for Period 3 than Period 1 (Table 2.2). No
differences in latencies were found between Period 1 and Period 2 (Table 2.3). Latencies to
move within two body lengths and one body length were shorter for Period 3 than Period 1
(Table 2.3).

Test-retest reliability
Spearman rank correlation coefficients for correlations between days and weeks are
presented in Table 2.4. Significant, moderate correlations were found between Period 1 and
Period 2 for all OF test measures. With the exception of a low correlation for the number of
vocalizations, and no significant correlation for latency to vocalize, moderate correlations were
also found between Period 1 and Period 3 for all other OF test measures. Moderate correlations
were found between Period 1 vs. 2 and Period 1 vs. 3 for TI vocalization latency and number of
vocalizations, whereas low correlations were found for TI duration. The number of inductions

53

was not significantly correlated between Period 1 and 3. Moderate to high correlations were
found for all VA test measures. For the NO test, the highest correlation was found for latency to
peck (Period 1 vs. 3). No other NO test measures were significantly correlated.

DISCUSSION
Previous experiments using fear tests for turkeys have used fear tests in a variety of
applications, ranging from comparing fear responses of turkeys housed in different rearing
systems (TI: Noble et al., 1996a), to examining the effects of dietary ascorbic acid on TI
reactions (Konca et al., 2009), assessing motor activity (OF test: Kowalski et al., 2002) and
activity level (OF test: Huff et al., 2007), and examining the effects of beak trimming on
tameness (NO test: Noble et al., 1996b). Notably, fear responses of turkeys had not been
described, characterized or assessed for test-retest reliability.
Our results indicated that responses were highly variable among turkeys with some turkeys
displaying very little, if any reaction during testing, whereas other turkeys were extremely active
and vocal. For example, between 27% (Period 1) and 53% (Period 3) of turkeys ambulated, with
the number of steps taken reaching as high as 382 steps when tested in the OF. Few other studies
have reported OF behavior of turkeys, and those that have are not directly comparable to our
study due to differences in methodology and how test measures were defined and recorded.
Kowalski et al. (2002) examined OF activity in two strains of commercial turkeys (BIG-6 and
BUT-9) as a measure of motor activity and emotional reactivity. Results were reported as means
(± SD) of activity levels in different areas of the OF (center, periphery), and episodes of
grooming, vocalization, sitting, and defecation. It is not stated how episodes were defined and
quantified. Another study examined “freezing and active behaviors” of male and female turkeys

54

of three different genetic lines in an OF test at 8 days (Huff et al., 2007). Birds were also tested
for the latency to reach an area of the OF adjacent to another pen of turkeys that was visible to
the test bird. OF test results were used to compare activity levels between sexes and strains (Huff
et al., 2007), and not to examine fear per se.
Similar to the OF test, few studies have examined TI responses of turkeys. Tonic
immobility was induced in most turkeys following only one induction, and a large percentage of
turkeys (39% in Period 1 and 58% in Period 3) remained immobile for the entire 780 s testing
period. Previous studies stopped TI testing at 60 s (Noble et al., 1996a; 1996b) and 600 s (Konca
et al., 2009). Mean TI durations in four different turkey lines tested at 20 weeks ranged from 6 to
20 s (Noble et al., 1996a) and 278.0 ± 38.3 to 332.9 ± 39.3 s in commercial (BUT-6) turkeys
tested at 16 weeks (Konca et al., 2009). Because data were not normally distributed, median TI
durations were reported here, and results are not directly comparable to those of Noble et al.
(1996a) and Konca et al. (2009). In an early study with chickens, it is stated that TI is
characterized by a period of reduced arousal, although some birds may vocalize (Gentle et al.,
1989). In contrast, as many as 42% (Period 1) and 26% (Period 3) of turkeys vocalized during
TI. Interestingly, TI measures associated with vocalization were more highly correlated between
test periods than TI duration. However, vocalizations are not often measured or reported as part
of TI, because unlike TI duration, vocalization during TI has not been shown to be an indicator
of the level of fear (see Forkman et al., 2007). Similar to OF and TI test responses, NO and VA
responses varied greatly among individual turkeys, with some birds pecking within 8 s (VA test)
and 18 s (NO test) and other birds not pecking at all. On average, the majority of birds (≥ 69%)
pecked the person during the VA test, and between 37% (Period 1) and 62% (Period 3) of birds
pecked the novel object during NO testing.

55

Measures that are closely spaced in time are more likely to have a greater degree of
relatedness than measures that are more distantly spaced in time. Therefore, little difference
would be expected between test measures from Period 1 vs. Period 2, whereas a larger difference
would be expected between Period 1 and Period 3. In general, results were reflective of a greater
difference between Period 1 and Period 3 test measures, with the exception of the VA test, where
all measures were significantly different between days and between weeks. For most test
measures, the likelihood of a behavioral response occurring (binomial responses) and frequencies
of responses did not change between test periods, but the rate at which behavioral responses were
performed changed. Specifically, the number of steps (OF test) and vocalizations (TI and OF
tests) changed between test periods, indicating that these behavioral responses change over time
and with repeated testing. Unfortunately, it was not possible to record vocalizations during VA
and NO testing due to background noise from fans, heaters and other birds in the barn. Further
research is needed to examine vocalization during NO and VA tests. Latencies to ambulate and
vocalize during OF testing, and latencies to move within two and one body lengths during NO
testing were all significantly shorter in Period 3 than Period 1, whereas there were no differences
in these test measures between Period 1 vs. 2. Contrarily, TI test responses (except for the
number of vocalizations) did not differ significantly between days and between weeks. Some
level of habituation is to be expected with repeated testing. Therefore, changes in test responses
may be indicative of habituation or age-related changes in behavior, or both, but it is not possible
to distinguish between changes due to age or due to habituation.
For most test measures, test repeatability was higher between days than between weeks,
as indicated by higher Spearman rank correlation coefficients for Period 1 vs. 2 than for Period 1
vs. 3. Moderately high (rS > 0.4) correlations were found for most TI, OF, and VA test measures.
56

The highest correlations were found for VA test measures. The levels of correlation found
among VA test measures are comparable to those from a study examining human-animal
interactions on commercial turkey farms (Botheras et al., 2008). They reported that the number
of turkeys within a prescribed distance from a person who was at times moving and at times
stationary was significantly, positively correlated (r = 0.67, P < 0.01) between 8 and 12 weeks
(during the stationary phase of the test). However, the numbers of turkeys within proximity of
the person were not correlated between 4 and 8 weeks or 4 and 12 weeks (Botheras et al., 2008),
whereas our results indicated moderate to high correlations between Period 1 (4 to 6 weeks) and
Period 3 (8 to 10 weeks). Turkeys in our study received frequent human contact because they
were tested in multiple tests and because marker was reapplied bi-weekly to enable individual
identification of birds. The amount of human contact likely affected their response to the
observer during the VA test such that few turkeys were actually fearful of the observer. Indeed,
Botheras et al. (2008) stated that turkeys receiving more human contact were less fearful of
humans. In our study, latencies to approach and peck the observer were shorter with each
consecutive VA test, indicating reduced fearfulness of the observer over time. However,
responses were highly correlated, suggesting that although fearfulness of the observer may have
been reduced, turkeys’ responses were reliable. It is possible that increased human contact may
influence reliability of test responses, but further research is necessary to evaluate whether
responses of animals experiencing more, or consistent, human contact are more reliable than
responses of animals receiving little, or inconsistent, human contact.
In contrast to the VA test, the only NO test measure that was significantly correlated was the
latency to peck the NO (Period 1 vs. 3). These results are in agreement with those from a study
with quail, which indicated that the latency to peck the NO had the highest correlation coefficient

57

of all NO test measures recorded (Miller et al., 2005). In both our study and that of Miller et al.
(2005), different novel objects were used and objects were balanced across test repeats. Low
repeatability of the NO test probably resulted from using different objects for each test repeat.
Indeed, the latency to move within two body lengths of the NO differed depending on the object
used when Period 1 responses were compared to those of Period 3, indicating that some objects
may have been perceived as more frightening than others. However, exposing birds to the same
object repeatedly would result in the object no longer being novel, creating a dilemma between
test repeatability and object novelty. As discussed by Miller et al. (2005) using quail, and
Gibbons et al. (2009) using cattle, using the same object repeatedly would have increased
habituation and the likelihood of turkeys responding to specific features of the object, which in
turn reduces generalizability of the results.
According to Cohen (1988), a Pearson’s product moment correlation coefficient of 0.1 is
considered a small effect size, 0.3 is a medium effect size, and a large effect size is indicated by a
correlation coefficient of 0.5 or larger. If Cohen’s large effect size criterion is applied to
correlations between Period 1 and 3, some measures from all four tests meet the criterion (OF
test: latency to ambulate, number of steps, number of squares, and number of defecations; TI
test: latency to vocalize and number of vocalizations; VA test: all measures; NO test: latency to
peck). Cohen’s large effect size criterion has been used by others to select fear tests measures
that were reliable enough for testing the validity of fear tests in quail (Miller et al., 2006).
Most studies of test repeatability expose animals to the same conditions repeatedly and
then assess the degree of relatedness (correlation) between test repeats. However, it is possible
that fear responses may be consistent, not because they are characteristic of a stable personality
trait, but because the conditions under which responses are elicited and conditions under which

58

animals are reared are kept constant (Chesler et al., 2003; Crabbe et al, 1999; Miller et al., 2005).
Because no research has previously examined fear test reliability using turkeys, a first step is to
determine whether test responses are consistent under stable conditions, and then further research
can be conducted to examine how varying environmental conditions affect test reliability.
Further research is needed to examine to what degree other variables may influence results when
they are not held constant over test situations. Research is also needed to assess the validity of
the four tests for assessing fear.
In their review of fear tests for hens and quail, Forkman et al. (2007) concluded that the
OF, TI and NO tests are reliable for testing fear in hens and quail. Ambulation in the open field
is heritable and there is a QTL related to OF behavior, suggesting that the response is stable
within individuals (Forkman et al., 2007). Similarly, there is a QTL for TI and previous research
has shown TI to be repeatable (reviewed in Forkman et al., 2007). The NO test is repeatable
between days and weeks (Forkman et al., 2007). Repeatability of hens’ responses to humans was
demonstrated in a study by Raubek et al. (2007), who concluded that hens’ responses to the
stationary person test, avoidance distance test and touch test are repeatable between days. Results
from our study are generally in agreement with results from studies with hens and quail that
responses to TI, OF and VA tests are repeatable between days and weeks. However, the only NO
measure that was reliable was the latency to peck.
In conclusion, fear responses of turkeys are not necessarily stable over time, but
responses to TI, OF and VA tests are repeatable because most test measures are moderately
correlated between days and weeks. Caution is needed when using the NO test as a welfare
assessment tool, because under the conditions of this study, only the latency to peck the object

59

was a reliable measure. These results may aid in the development of measures for assessing
welfare of turkeys, but further research examining fear test validity is needed.

ACKNOWLEDGEMENTS
We thank Hybrid Turkeys for supplying the poults, Dr. Darrin Karcher, and Erika First,
Caitlin Hummel, Sarah Jager, Cheyenne Ryzenga and Shannon Williams for assistance with data
collection. This work was supported by a Natural Sciences and Engineering Research Council of
Canada postgraduate scholarship and Michigan State University AgBioResearch.

60

APPENDIX

61

Table 2.1. Testing schedule and test order for each group of four pens. Each test group was tested
three times in four tests of fear (VA = voluntary approach test, NO = novel object test, OF =
open field test, TI = tonic immobility).
Test used and order of testing
Test Period Age (d) Group 1 Group 2 Group 3 Group 4
1
28
VA
TI
NO
OF
2
29
VA
TI
NO
OF
1
32
NO
VA
OF
TI
2
33
NO
VA
OF
TI
1
36
OF
NO
TI
VA
2
37
OF
NO
TI
VA
1
40
TI
OF
VA
NO
2
41
TI
OF
VA
NO
3
3
3
3

56
60
64
68

VA
NO
OF
TI

TI
VA
NO
OF

62

NO
OF
TI
VA

OF
TI
VA
NO

Table 2.2. Differences in percentages of turkeys performing a particular behavior in the open
field, tonic immobility, voluntary approach and novel object tests between days (test Period 1 vs.
2) and between weeks (Period 1 vs. 3).
Period 1
Standing
Sitting
Ambulation
Vocalization
Escape
Defecation
One induction
Vocalization
Immobile for full
duration
Defecation
Within two BL
Within one BL
Approach
Peck
Within two BL
Within one BL

Period 2

χ

2

Open Field (n = 60)
88.0
78.0
4.5*
20.0
28.0
4.5*
27.0
40.0
1.0

Period 3

χ

93.0
47.0
53.0

2

57.0
63.0
2.7
13.0
8.0
0.1
87.0
78.0
1.6
Tonic Immobility (n = 66)
77.0
70.0
0.1
42.0
38.0
0.1
39.0
52.0
1.6

83.0
0
95.0

0.02
0.1
T
2.8
1.4
.
0.1

71.0
26.0
58.0

0.3
1.6
0.2

14.0
17.0
1.0
Voluntary Approach (n = 16)
84.3
95.0
1.7
79.0
92.2
2.5
72.9
89.6
5.2*
69.2
88.0
5.1*
Novel Object (n = 16)
78.1
67.5
2.1
61.0
55.0
0.5

18.0

3.0T

97.7
95.2
93.6
91.1

0.5
1.1
0.9
1.2

95.0
81.3

0.2
T
3.1
3.5
2.2

Approach

50.4

50.0

0

73.8

Peck

37.1

47.7

1.2

62.1

Note: BL = body length.
T
P < 0.10, *P < 0.05.

63

T

Table 2.3. Differences in median (25-75% range) values of open field, tonic immobility, voluntary approach and novel object test
measures between Period 1 and Period 2 and Period 1 and Period 3. Testing occurred at 28 to 40 days (Period 1), 29 to 41 days (Period
2), and 56 to 68 days (Period 3).
Test measure

Period 1

Ambulate (s)
Step no.
Square no.
Vocalize (s)
Vocalize no.
Defecate no.

600 (487-600)
0 (0-7)
0 (0-1)
381 (127-600)
6 (0-168)
2 (1-3)

Dur. (s)
Vocalize (s)

589 (321-780)
780 (189-780)

Test Statistic
Period 3
Period 1 vs. 2
Open field test (n = 60)
600 (22-600)
L = 1.7, W = 1.9
550 (183-600)
0 (0-18)
t = 6.3***
7 (0-21)
0 (0-1)
t = 3.4**
0 (0-3)
285 (73-600)
L = 1.4, W = 1.9
52 (10-163)
19 (0-197)
t = 5.6***
99 (10-240)
2 (1-3)
t = 0.5
2 (1-3)
Tonic immobility test (n = 66)
2
1 (1-2)
1 (1-2)
χ = 1.5
780 (365-780) L = 0.6, W = 0.3
780 (493-780)
780 (65-780)
L = 0.1, W = 0
780 (630-780)

Vocalize no.

0 (0-21)

0 (0-4)

Induction no. 1 (1-1)

Period 2

Two BL (s)
103 (59-369)
One BL (s)
244 (133-518)
Approach (s) 315 (181-559)
Peck (s)
449 (293-817)

t = 10.3***
0 (0-2)
Voluntary approach test (n = 16)
24 (18-32)
L = 7.6*, W = 10.9** 29 (24-35)
32 (22-46)
L = 7.4*, W = 11.6** 48 (36-61)
52 (32-75)
L = 7.0*, W = 9.7*
66 (44-194)
75 (41-133)
L = 8.6*, W = 12.4** 115 (78-259)

Two BL (s)
One BL (s)
Approach (s)

69 (18-474)
426 (33-702)
563 (113-900)

143 (9-634)
323 (2-900)
601 (79-900)

Peck (s)

764 (450-900)

680 (254-900)

Novel object test (n = 16)
L = 0.001, W = 0.3
9 (3-57)
L = 0.4, W = 0.1
24 (12-388)
L = 0.2, W = 0.1
153 (24-730)
L = 0.02, W = 0.2

Note: Lat. = latency, BL = body length, L = log-rank, W = Wilcoxon.
T
P < 0.10, * P < 0.01, **P < 0.001, *** P < 0.0001.
64

440 (97-736)

Test Statistic
Period 1 vs. 3
L = 8.3*, W = 11.5**
t = 8.0***
t = 3.2*
L = 22.5***, W = 29.7***
t = 11.6***
t = 1.5
2

χ = 1.4
L = 2.4, W = 1.1
T
L = 3.3 , W = 2.5
t = -12.5***
L = 11.7**, W = 10.6*
L = 10.2*, W = 9.5*
L = 9.8*, W = 8.9*
L = 12.2**, W = 10.7*
L = 11.6**, W = 10.3**
L = 6.3*, W = 7.3*
T
L = 1.4, W = 2.9
L = 2.3, W = 2.4

Table 2.4. Spearman rank correlation (rS) of test measures between days (Period 1 vs. 2) and
weeks (Period 1 vs. 3) for four tests of fear (open field, tonic immobility, voluntary approach,
and novel object tests).
Test
Open Field
(n = 60)

Behavioural response
Lat. ambulate
Lat. vocalize
Step no.
Square no.
Vocalization no.
Defecation no.

Period 1 vs.2
0.62****
0.47***
0.65****
0.71****
0.51****
0.47***

Period 1 vs. 3
0.66****
0.17
0.58****
0.59****
0.39**
0.50****

Tonic
Immobility
(n = 66)

Duration
Lat. vocalize
Induction no.
Vocalization no.

0.35**
0.57****
0.39**
0.67****

0.31*
0.51****
0.08
0.59****

Voluntary
Approach
(n = 16)

Lat. two BL
Lat. one BL
Lat. approach
Lat. peck

0.63**
0.62*
0.81***
0.91****

0.62*
0.62**
0.87****
0.67**

Novel Object
(n = 16)

Lat. two BL
Lat. one BL
Lat. approach
Lat. peck

0.39
0.38
0.29
0.35

0.34
0.23
0.15
0.61*

Note: Lat. = latency, BL = body length.
T
P*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

65

REFERENCES

66

REFERENCES
Boissy, A. Fear and fearfulness in animals. 1995. Q. Rev. Biol.. 70: 165-191.
Botheras, N. A., J. A. Pempek, and D. E. Enigk. 2008. Human-animal interactions in the turkey
industry. Midwest Poultry Research Program Project Database. Available online:
http://www.mwpoultry.org/ProjectPDFs/08-09.pdf
Brambell, F. W. R. (1965). Report of the technical committee to enquire into the welfare of
animals kept under intensive livestock husbandry systems. (Command Rep. 2836).
London: Her Majesty's Stationery Office.
Broom, D. M. 1988. The scientific assessment of animal welfare. Appl. Anim. Behav. Sci. 20: 519.
Calandreau L., A. Favreau-Peigné, A. Bertin, P. Constantin, C. Arnould, A. Laurence, S.
Lumineau, C. Houdelier, M. A. Richard-Yris, A. Boissy, and C. Leterrier. 2011. Higher
inherent fearfulness potentiates the effects of chronic stress in the Japanese quail. Behav.
Brain Res. 225: 505-510.
Chesler, E. J.,S. G. Wilson, W. R. Lariviere, S. L. Rodriguez-Zas, and J. S. Mogil. 2003.
Identification and ranking of genetic and laboratory environment factors influencing a
behavioral trait, thermal nociception, via computational analysis of a large data archive.
Neurosci. Biobehav. Rev. 26: 907–923.
Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences. Hillsdale, New Jersey: L.
Erlbaum. Pp 79-81.
Cockrem, J. F. 2007. Stress, corticosterone responses and avian personalities. J Ornithol. 148
(Suppl 2): S169–S178.
Crabbe, J.C., D. Wahlsten, and B. C. Dudek. 1999. Genetics of mouse behavior: interactions
with laboratory environment. Science. 284: 1670–1672.
Cransberg, P. H., P. H. Hemsworth, and G. J. Coleman. 2000. Human factors affecting the
behavior and productivity of commercial broiler chickens. Br. Poult. Sci. 41: 272-9.
de Haas, E. N., B. L. Nielsen, A. J. (Bart) Buitenhuis, T. B. Rodenburg. 2010. Selection on
feather pecking affects response to novelty and foraging behavior in laying hens. Appl.
Anim. Behav. Sci. 124: 90-96.
de Haas, E. N., B. Kemp , J. E. Bolhuis , T. Groothuis , and T. B. Rodenburg. 2013. Fear, stress,
and feather pecking in commercial white and brown laying hen parent-stock flocks and
their relationships with production parameters. Poult. Sci. 92: 2259-2269.

67

Farm Animal Welfare Council. 1979. Press Statement. Available Online:
http://www.fawc.org.uk/pdf/fivefreedoms1979.pdf
Forkman, B., A. Boissy, M. –C. Maunier-Salaün, E.Canali, and R. B. Jones. 2007. A critical
review of fear tests used on cattle, pigs, sheep, poultry and horses. Physiol. Behav. 92:
340-374.
Gentle, M. J., R. B. Jones, and S. C. Woolley. 1989. Physiological changes during tonic
immobility in Gallus gallus var domesticus. Physiol. Behav. 46: 843-847.
Gibbons, J., A. Lawrence, and M. Haskell. 2009. Responsiveness of dairy cows to human
approach and novel stimuli. Appl. Anim. Behav. Sci.116: 163-173.
Graml., C., S. Waiblinger, and K. Niebuhr. 2008. Validation of tests for on-farm assessment of
the hen-human relationship in non-cage systems. Appl. Anim. Behav. Sci. 111: 301-310.
Groothuis, T. G. G., and C. Carere. 2005. Avian personalities: characterization and epigenesis.
Neurosci. and Biobehav. Rev. 29: 137-150.
Hemsworth, P. H., G. J. Coleman, J. L. Barnett, and R. B. Jones. 1994. Behavioral responses to
humans and the productivity of commercial broiler chickens. Appl. Anim. Behav. Sci. 41:
101-114.
Hocking, P. M., C. E. Channing, D. Waddington, and R. B. Jones. 2001. Age-related changes in
fear, sociality and pecking behaviors in two strains of laying hen. Br. Poult. Sci. 42: 414423.
Huff, G., W. Huff, N. Rath, A. Donoghue, N. Anthony, and K. Nestor. 2007. Differential effects
of sex and genetics on behavior and stress response of turkeys. Poult. Sci. 86: 1294-1303.
Jones, R. B. 1989. Chronic stressors, tonic immobility and leucocytic responses in the domestic
fowl. Physiol. Behav. 46: 439-442.
Jones, R. B. 1996. Fear and adaptability in poultry: insights, implications and imperatives.
World’s Poult. Sci. J. 52: 131-174.
Jones, R. B. 2002. Role of comparative psychology in the development of effective
environmental enrichment strategies to improve poultry welfare. Int. J. Comp. Psychol.
15:77-106.
Jones, B., and A. Boissy. 2011. Fear and other negative emotions. In: Animal welfare. Second
ed. Appleby, M. C, J. A. Mench, I. A. S. Olsson, and B. O. Hughes (eds.). pp 81.
Karpestam, E., S. Merilaita, and A. Forsman. 2012. Reduced predation risk for melanistic pygmy
grasshoppers in post-fire environments. Ecol. Evol. 2: 2204-2212.

68

Konca, Y., F. Kirkpinar, and M. Çabuk. 2009. Effects of dietary ascorbic acid on blood
haemoatological profile, serumbiochemical components and tonic immobility reaction of
male turkeys under summer condition. J. Poult. Sci. 46: 105-111.
Koolhaas, J. M., S. M. Korte, S. F. De Boer, B. J. Van Der Vegt, C. G. Van Reenen, H. Hopster,
I. C. De Jong, M. A. W. Ruis, and H. J. Blokhuis. 1999. Coping styles in animals: current
status in behavior and stress-physiology. Neurosci. Biobehav. Rev. 23: 925–935.
Kowalski, A., P. Mormède, K. Jakubowski, and M. Jedlińska-Krakowska. Comparison of
susceptibility to stress in two genetic lines of turkey broilers, BUT-9 and Big-6. Polish J.
Vet. Sci. 5: 145-150.
Littell, R. C., G. A. Milliken, W. W. Stroup, R. D. Wolfinger, and O. Schabenberger. 2006.
SAS® for mixed models (2nd ed.)SAS Institute Inc., North Carolina, USA, pp. 589-594.
Marchewka, J., T. T. N. Watanabe, V. Ferrante, and I. Estevez. 2013. Review of the social and
environmental factors affecting the behavior and welfare of turkeys (Meleagris
gallopavo). Poult. Sci. 92:1467–1473.
Martin, P., Bateson, P., 1993. Measuring Behavior, An Introductory Guide. Second ed.
Cambridge University Press, New York, USA, pp. 1–222.
Miller, K. A., J. P. Garner, and J. A. Mench. 2005. The test-retest reliability of four behavioral
tests of fearfulness for quail: a critical evaluation. Appl. Anim. Behav. Sci. 92:113-127.
Miller, K. A., Garner, J. P., and J. A. Mench. 2006. Is fearfulness a trait that can be measured
with behavioral tests? A validation of four fear tests for Japanese quail. Anim. Behav. 71:
1323-1334.
Noble, D. O., J. W. Anderson, and K. E. Nestor. 1996a. Range and confinement rearing of four
genetic lines of turkeys. 2. Effects on behavior and tonic immobility. Poult. Sci. 75: 165171).
Noble, D. O., K. E. Nestor, and K. K. Krueger. 1996b. The effect of beak trimming on two
strains of commercial turkeys. 2. Behavior traits. Poult. Sci. 75: 1468-1471.
Raubek, J., K. Niebuhr, and S. Waiblinger. 2007. Development of on-farm methods to assess the
animal-human relationship in laying hens kept in non-cage systems. Anim. Welfare. 16:
173-175.
Reed, H. J., L. J. Wilkins, S.D. Austin, and N. G. Gregory. 1993. The effect of environmental
enrichment during rearing on fear reactions and depopulation trauma in adult caged hens.
Appl. Anim. Behav. Sci. 36: 39-46.

69

Rodenburg, T. B., de Haas, E. N., Nielsen, B. L., and A. J. Buitenhuis. 2010. Fearfulness and
feather damage in laying hens divergently selected for high and low feather pecking.
Appl. Anim. Behav. Sci.128: 91-96.
SAS/STAT User’s Guide. 2012. SAS/STAT 12.1 user’s guide the LIFETEST procedure
(chapter). Available online:
http://support.sas.com/documentation/onlinedoc/stat/121/lifetest.pdf
Schütz, K. E., S.Kerje, L.Jacobsson, B. Forkman, O. Carlborg, L. Andersson, and P. Jensen.
2004. Major growth QTLs in fowl are related to fearful behavior: possible genetic links
between fear responses and production traits in a red junglefowl x white leghorn
intercross. Behav Genet. 34: 121-30.
van Zeeland, Y. R. A., M. M .J. A. van der Aa, C. M. Vinke, J. T. Lumeij, and N. J. Schoemaker.
2013. Behavioral testing to determine differences between coping styles in Grey parrots
(Psittacus erithacus erithacus) with and without feather damaging behavior. Appl. Anim.
Behav. Sci. 148: 218-231.
Vestergaard, K. S., J. P. Kruijt, and J. A. Hogan. 1993. Feather pecking and chronic fear in
groups of red junglefowl: their relations to dustbathing, rearing environment and social
status. Anim. Behav. 45: 1127-1140.
Welfare Quality®. 2009a. Welfare Quality® Assessment Protocol for Poultry. Welfare Quality
Consortium, Lelystad, Netherlands.
Welfare Quality®. 2009b. Welfare Quality® Assessment Protocol for Pigs. Welfare Quality
Consortium, Lelystad, Netherlands.
Welfare Quality®. 2009c. Welfare Quality® Assessment Protocol for Cattle. Welfare Quality
Consortium, Lelystad, Netherlands.

70

CHAPTER 3: USING INTER-TEST CORRESPONDENCE TO ASSESS THE VALIDITY
OF FOUR TESTS OF FEAR FOR TURKEYS
ABSTRACT
Fear and fearfulness are the most widely studied aspects of poultry temperament, but
little is known regarding turkeys’ fear responses. If fearfulness is part of temperament, fear tests
should be reliable and valid. This study evaluated inter-test correspondence as a measure of
validity among four tests (tonic immobility (TI), open field (OF), voluntary approach (VA), and
novel object (NO) tests). Male turkeys (Meleagris gallopavo) were housed in groups of four to
six in 16 pens (N = 16). Turkeys were tested in all four tests between 4 and 6 weeks (Age 1) and
8 and 10 weeks (Age 2). Test correspondence was assessed using pair-wise correlations and by
examining whether turkeys showing extreme responses (high (HR) vs. low (LR) responders) in
each test situation differed across other test situations. At Age 1, the VA and OF tests tended to
be correlated (rS = 0.43, P < 0.10), and turkeys classified as HR in the VA test (VA-HR) tended
to be more active in the OF test vs. VA-LR (S = 141, P < 0.10). At Age 2, OF responses were
moderately correlated with TI vocalizations (rS = -0.50, P < 0.05) and tended to be correlated
with NO test behavior (rS = -0.44, P < 0.10). Compared to NO-LR, NO-HR were significantly
more active in the OF and took less time to peck when tested in the VA test. Overall, results
reveal a weak association between turkeys’ willingness to move toward an observer during the
VA test and their ambulatory behavior in an OF test at Age 1. There may be stronger
associations among tests at Age 2, but age was confounded with test habituation. Results did not
validate that OF, TI, VA and NO tests measure the same types of fear responses. Fear responses
may therefore be situation-dependent.

71

INTRODUCTION
Individual differences in behavior are widely recognized in different vertebrate species,
ranging from fish to birds and mammals. Research into individual differences continues to
increase and has widespread applications, including applications to human health and disease,
food quality, and animal well-being. For instance, rodents are used as models for human
disorders such as post traumatic stress disorder and anxiety (reviewed in Bourin et al., 2007, and
Neumann et al., 2011) because individual differences in behavior facilitate the development and
testing of drugs for treating these disorders. In domestic animal species, individual differences in
behavior are used in the development of selection and breeding programs to increase favorable
traits or reduce harmful or unwanted traits. Indeed, individual behavioral differences are
associated with differences in disease susceptibility in rats, calves and pigs (reviewed in
Koolhaas et al., 1999), and production and meat quality in steers (e.g. Petherick et al., 2002;
Voisinet et al., 1997) and quail (e.g. Remignon et al., 1998).
Different terminology is used in the context of individual differences in behavior,
including the terms personality, temperament, coping styles, coping strategies, behavioral
syndromes and behavioral profiles, among others. Several review papers discuss the definitions
of, and differences among, these terms (Cockrem, 2007; Gosling, 2001; Groothuis & Carere,
2005; B. Jones & Boissy, 2011; A. C. Jones & Gosling, 2005; Koolhaas et al., 1999; Stamps &
Groothuis, 2010). In general, temperament, personality, and coping styles are considered to have
similar meanings (Cockrem, 2007; Stamps & Groothuis, 2010). However, “dimensions of
personality are generally described in terms of coping styles or fearfulness” (B. Jones and
Boissy, 2011), where coping styles concern stress physiology (Groothuis & Carrere, 2005), and
fear is a characteristic associated with emotional reactivity and not coping style (Koolhaas et al.,

72

1999). However, some descriptions of coping styles also include fearfulness as a coping style
characteristic (e.g. Cockrem, 2007).
If fearfulness is part of temperament or coping styles, then fearfulness should be stable in
individuals over time and tests used to assess fear responses should be reliable and valid even if
tests differ in the specific fear stimuli used. Correspondence among fear tests can be used as a
method of validating that tests are measuring fear responses, and not some other response.
Indeed, Forkman et al. (2007) state that “validation through evaluation of inter-test consistency
should be given more attention in future studies”. In their review of fear tests for chickens and
quail, Forkman et al. (2007) conclude that the open field and tonic immobility tests are reliable
and valid, and the novel object test is a reliable and practical test of fear for chickens and quail.
Graml et al. (2008) concluded that the voluntary approach test (stationary person test) is a valid
test for assessing the hen-human relationship. However, there is conflicting evidence regarding
fear test correspondence. Older studies appear to indicate some correspondence among fear tests
(e.g. R. B. Jones, 1996), whereas more recent studies have failed to show correspondence,
especially among tests assessing novelty and tests assessing predatory-type stimuli (e.g. Miller et
al., 2006).
Several strategies have been used to validate and assess correspondence among tests
intended to assess fear or other components of animal temperament. One strategy is to examine
whether differences between behaviorally distinct groups of animals remain consistent across
various test situations. Animals are tested in one type of fear test, such as a tonic immobility (TI)
test, so that they can be assigned to behaviorally distinct groups, such as short TI and long TI
durations. The groups are then tested in another test, such as a novel object test, and their
behavior is compared to assess whether TI and novel object (NO) tests correspond. If tests

73

correspond, short TI and long TI groups will differ in their responses to the NO. This strategy
has been used to assess tests of coping styles in pigs (Spake et al., 2012). A related method is to
compare animals from genetic lines that have been bidirectionally selected for a particular
behavioral or physiological response to ascertain whether their responses differ in other
situations as well. For example, genetically selecting for animals with short and long TI
durations and then comparing NO test responses between genetic lines (e.g. Hazard et al., 2008;
Richard et al., 2008). A third method is to use correlations to assess the strength of the
relationship between pairs of test measures from different tests (e.g. Albentosa et al., 2003).
Similarly, factor analyses can also be used to evaluate whether measures from different tests are
related (e.g. Mazurek et al., 2011; Miller et al., 2006). Another strategy is to use drugs, such as
anti-anxiolytics, to determine whether fear and anxiety-related behavior of treated and untreated
animals differ when tested in fear tests (e.g. Marín et al., 1997). Lastly, animals can be assigned
to different treatments designed to increase or reduce fear (e.g. increased positive human contact
to reduce fear), and responses to other fear tests can be compared among the treatment groups to
determine whether groups that display reduced fear as a result of the treatment also exhibit
reduced fear under other fear-eliciting conditions (e.g. Graml et al., 2008).
Most research concerning fearfulness of domestic poultry has used chickens and quail,
but there is scant research examining fear or fearfulness of turkeys. Chickens are one of the most
widely studied production animal species, and quail are an important species used in laboratory
research. However, domestic turkey production has increased greatly in recent decades, yet there
is very little research examining behavior in this species compared to chickens and quail, despite
the importance of behavior to turkey welfare, management and temperament. We recently
demonstrated that some measures of tonic immobility, open field, voluntary approach and novel

74

object tests are reliable for assessing fear responses of turkeys (Erasmus & Swanson, 2014). This
study builds upon our previous results. The objectives were to assess inter-test correspondence
among tonic immobility, open field, voluntary approach, and novel object tests for turkeys using
two strategies, including 1) calculating correlations between test measures, and 2) evaluating
whether differences among behaviorally distinct groups of turkeys remain consistent across test
situations.

MATERIALS AND METHODS
All procedures were approved by the Institutional Animal Care and Use Committee of
Michigan State University.

General procedures
A full description of materials and methods is presented elsewhere (Erasmus & Swanson,
2014). Briefly, commercial male turkeys (Hybrid Converter) were brooded in four littered (wood
shavings) pens (2.4 m x 3.05 m) in groups of 24 from 1 day to 2 weeks of age. At 2 weeks,
turkeys were randomly assigned to one of 16 littered pens (2.4 m x 3.05 m) in the same barn and
housed in groups of four to six until 14 weeks.
Turkeys were tested three times (Period 1: 28 to 41 d, Period 2: 29 to 42 d, and Period 3:
56 to 68 d) in each of four tests of fear (tonic immobility (TI), open field (OF), voluntary
approach (VA) and novel object (NO)). Only data from Period 1 and Period 3 were used here to
assess inter-test correspondence at two ages (Period 1 and Period 3). Test periods will
consequently be referred to as Age 1 (28 to 41 d; formerly Period 1) and Age 2 (56 to 68 d;
formerly Period 3) throughout.

75

For TI and OF tests, birds were individually removed from their home pens and tested in
another room. Voluntary approach and NO tests were conducted with groups of turkeys in their
home pens to reduce confounds associated with environmental novelty, and because NO tests are
used to assess welfare in groups of birds as part of welfare assessment programs (e.g. Welfare
Quality®, 2009). The TI test lasted until the turkey righted itself, or for a maximum of 13 min.
The number of inductions, number of vocalizations, latency to vocalize (s) and TI duration (s)
were recorded. Turkeys were individually tested in the OF, consisting of a square arena (length:
2.74 m, width: 2.74 m, height: 1.83 m) with a concrete floor that was divided into 81 squares
2

(each 0.30 m ) for 10 min. OF test measures included the latency to ambulate (s), latency to
vocalize (s), number of vocalizations, number of squares entered, number of steps taken and
escape attempts. Turkeys were tested in VA and NO tests for 15 min. Four different objects were
used to assess NO test responses to avoid habituation. Test measures that were recorded included
latencies to move within two body lengths, one body length, approach and peck.
Based on results from Erasmus & Swanson (2014), test measures that had between-week
(Period 1 vs. Period 3) test-retest correlation coefficients of rS ≥ 0.5 were retained for analysis of
inter-test correspondence. The following test measures were thus included in the analyses: OF
test latency to ambulate (rS = 0.66), number of steps (rS = 0.58), number of squares (rS = 0.59),
and number of defecations (rS = 0.50); TI test latency to vocalize (rS = 0.51) and number of
vocalizations (rS = 0.59); all VA test measures (latency to move within two body lengths: rS =
0.62, latency to move within one body length: rS = 0.62, latency to approach: rS = 0.87, and

76

latency to peck: rS = 0.67); and the latency to peck the object during the NO test (rS = 0.61)
(Erasmus & Swanson, 2014).
Two strategies were used to assess correspondence among fear tests. Test measures were
correlated to examine pair-wise associations between measures from different tests. Secondly,
the two turkeys from each pen showing the highest and lowest extremes in responses to fear test
measures, respectively, were selected for further analysis to determine whether birds showing
extremes in their behavioral responses in one test also differed in responses to other tests. These
birds were classified as either a low responder (LR; fewest vocalizations, longest latency to
ambulate, approach or peck) or a high responder (HR; most vocalizations, shortest latency to
ambulate, approach or peck) in each test. In a study with pigs, the top and bottom 10% of high
and low responding pigs were selected for further study. Here, we selected one HR and one LR
from each pen. Each HR and LR therefore represented 17-25% of turkeys in the pen.
For each fear test, the test measure that was most reliable when test-retest reliability was
assessed between weeks (Erasmus & Swanson, 2014) was used as the basis for choosing HR and
LR for each test. For the OF test, the two turkeys from each pen with the shortest (open field
high responder; OF-HR) and longest (open field low responder; OF-LR) latencies to ambulate
were selected. For the TI test, turkeys were chosen based on the number of vocalizations (TI-HR
had the most vocalizations and TI-LR had the fewest vocalizations). For the VA test, turkeys
were selected based on the latency to approach the observer (VA-HR: fastest approach vs. VALR: slowest approach), and for the NO test turkeys were selected based on the latency to peck to
NO (NO-HR: shortest peck latency vs. NO-LR: longest peck latency). Similar to methods of
Spake et al. (2012), we verified that turkeys that were classified as HR and LR in each fear test
differed significantly. Specifically, OF-HR and OF-LR differed significantly in the latency to
77

ambulate (Age 1: U = 53.0, P < 0.0001; Age 2: U = 106.0, P < 0.0001). Turkeys classified as TIHR and TI-LR differed significantly in the number of vocalizations produced (Age 1: U = 5.09,
P < 0.0001; Age 2: U = 5.03, P < 0.0001), VA-HR and VA-LR differed in the latency to
approach (Age 1: U = 109.0, P = 0.0002; Age 2: U = 332.0, P = 0.01) and NO-HR and NO-LR
differed in the latency to peck (Age 1: U = 73.0, P < 0.0001; Age 2: U = 91.0, P < 0.0001).
There were several pens in which none of the birds responded (all low responders). In these
cases, data from only one randomly selected bird was used in further analyses and the bird was
classified as LR. Turkeys classified as HR and LR in each test were not necessarily the same
birds that were classified as HR or LR in other tests. Therefore, sample sizes varied for each test
(OF-HR: N = 9, OF-LR: N = 15; TI-HR: N = 14, TI-LR: N = 15; VA-HR: N = 12, VA-LR: N =
15; NO-HR: N =10, NO-LR: N = 15).

Statistical analysis
All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary,
NC). Pen (N = 16) was the experimental unit. For pair-wise correlations, data from each test
measure were averaged across birds in each pen to obtain a single value for each test measure for
each pen. Data were not normally distributed and transformations were ineffective. Therefore,
test measures were correlated across all four tests of fear using the Spearman rank correlation
coefficient. Correlation coefficients were defined as low: 0.2 – 0.4, moderate: 0.4 – 0.7, high: 0.7
– 0.9, and very high: 0.9 – 1.0 (Martin & Bateson, 1993). The Mann-Whitney U test (Wilcoxon
rank sum test; SAS PROC NPAR1WAY) was used to assess differences in test responses
between high and low responders. The exact Wilcoxon test was specified. The Wilcoxon rank

78

sum test was used to verify that behavior of turkeys that were classified as LR and HR in each
test differed significantly.

RESULTS

Pair-wise correlations of test measures
Spearman rank correlation coefficients for pair-wise comparisons among test measures
are presented in Table 3.1.
Age 1. There was a tendency for the number of steps taken and number of squares entered
in the OF test to be moderately, positively correlated with the latency to peck the observer during
the VA test. No other significant correlations were found.
Age 2. A greater number of significant correlations were observed between test measures
at Age 2 compared to Age 1. The number of vocalizations during TI testing was moderately
correlated with all OF test measures. The number of steps taken and number of squares entered
in the OF tended to be correlated with the latency to peck the NO. There were no significant
correlations between TI and NO, or TI and VA test measures.

Differences between high (HR) and low (LR) responders
Age 1. Turkeys classified as OF-LR and OF-HR did not differ in their responses to any
other fear tests (Table 3.2). Similarly, TI-HR and TI-LR, and NO-HR and NO-LR did not differ
in their responses to other tests. However, there was a tendency for VA-HR to take less time to
ambulate and to enter more squares when tested in the OF test compared to VA-LR (Table 3.2).
Age 2. A greater number of differences were observed between HR and LR turkeys at
Age 2 (Table 3.3). Turkeys classified as OF-HR tended to take less time to move within two
79

body lengths and approach the observer, and took significantly less time to move within one
body length of the observer in the VA test compared to OF-LR, but OF-HR and OF-LR did not
differ in TI or NO behavior. There was a tendency for TI-HR to defecate more when tested in
the OF test compared to TI-LR. Turkeys classified as VA-HR tended to take less time to peck the
NO during NO testing compared to VA-LR. Turkeys classified as NO-HR were more active in
the OF test and took significantly less time to peck the observer during the VA test compared to
NO-LR.

DISCUSSION
Two strategies were used to assess inter-test correspondence between four tests of fear
for turkeys. Results from comparisons between high and low responders were in agreement with
results from pair-wise correlations at both ages, indicating that these two strategies yielded
similar results. Specifically, the tendency for a positive correlation between the latency to peck
the observer during the VA test and the numbers of squares entered and steps taken during OF
testing at Age 1 were in agreement with the tendency for VA-HR and VA-LR to differ in the
latency to ambulate and number of squares entered when tested in the OF test. At Age 2,
significant correlations were found between the TI and OF tests, and TI-HR and TI-LR tended to
differ in the number of defecations produced in the OF test. Furthermore, the numbers of steps
taken and squares entered in the OF test and latency to peck the NO tended to be correlated, and
turkeys that took less time to peck the NO (NO-HR) took less time to ambulate, took more steps
and defecated more in the OF than NO-LR at Age 2.
Interestingly, there were cases where HR and LR in one type of test differed in their
responses in another test type, but the reverse was not observed. For example, at Age 1 VA-HR

80

and VA-LR tended to differ in the latency to ambulate in the OF test, but turkeys classified OFHR and OF-LR did not differ in their responses to the VA test. Similarly, NO-HR and NO-LR
differed in OF behavior at Age 2, but OF-HR and OF-LR did not differ in their NO responses.
Because high and low responding individuals were selected separately for each test, HR and LR
birds were not the same birds in one test as HR and LR birds in another test. In other words,
there was some overlap between tests in the individual turkeys that were represented, but VAHR turkeys for example were not all the same turkeys as OF-HR turkeys. The reason for
differences in which turkeys were used to represent HR and LR is that individual turkeys’
behavior varied within pens from test situation to test situation, such that the same two birds
were not always the highest and lowest responders in the pen across all test situations. A possible
explanation for the inconsistency in individual turkeys’ behavior is that there may have been
different floor (high numbers of birds showing little or no reaction) and ceiling (high numbers of
birds showing the maximum response) effects depending on the test used. Indeed, a greater
number of turkeys displayed the same behavior in the NO test (e.g. did not peck the NO) than in
the OF test, making it impossible in some cases to separate high and low responders in the NO
test. Increasing NO test duration may have lead to greater observed differences in behavioral
responses, but further research is needed to understand floor and ceiling effects of various fear
tests.
We previously reported that turkeys’ behavior in fear tests changed over time and with
repeated testing (Erasmus & Swanson, 2014). For example, activity levels in the OF increased
over time (shorter latency to ambulate and greater numbers of steps taken), whereas latencies to
approach and peck during VA and NO testing decreased over time (Erasmus & Swanson, 2014).
Similarly, inter-test correspondence appears to change over time and with repeated testing. In

81

contrast to results from Age 1, there were more instances of inter-test correspondence and
stronger relationships between different test situations when turkeys were tested at Age 2.
Results from pair-wise comparisons and differences between HR and LR suggest that there may
be correspondence between turkeys’ responses in a VA test and an OF test between 4 and 6
weeks, but the relationship between these two tests is not very strong. Conversely, there appear
to be strong relationships between turkeys’ behavior in OF and NO tests, and OF and TI tests,
and a weak relationship between VA and NO tests at 8 to 10 weeks of age.
A study assessing the validity of four tests of fear for quail also reported age-related
differences in test correspondence: correlations between test measures were different when quail
were tested at 42 to 49 days of age compared to tests at 61 to 65 days of age (Miller et al., 2006).
Other researchers have only assessed inter-test correspondence at one age (e.g. Albentosa et al.,
2003) or have taken the average of fear responses across different ages (e.g. Hocking et al.,
2001). Our results suggest that not only does behavior in each fear test change over time
(Erasmus & Swanson, 2014), but relationships between tests also change over time and with
repeated testing. The differences in test correspondence between the two ages may have been
affected by repeated testing of turkeys, and by the amount of human contact that turkeys received
in this study. Unfortunately, the effects of repeated testing and habituation could not be separated
from the effects of age. Additional research examining test correspondence in different groups of
turkeys at different ages is necessary to fully understand how habituation and repeated testing
may influence inter-test correspondence.
Correspondence between tests, or lack thereof, may be explained by similarities and
differences in what the tests are actually measuring. Because fear is a subjective, complex
emotional state that is difficult to quantify, test measures are by necessity directed at specific

82

components of animals’ behavioral reactions that are quantifiable, such as the numbers of
vocalizations produced and steps taken. Consequently, behavioral reactions, and thus test
measures, may differ depending on the test used, making it challenging to interpret and examine
associations among different types of fear tests. Alternatively, test measures may assess similar
underlying behavioral patterns that are consistent within individuals and may therefore be more
likely to correspond. For example, the general activity level of an animal, manifested as its
locomotor behavior, may be a consistent characteristic of that animal, and therefore, it may be
expected that a test measure in one test would correspond to a measure in another test if both
assess locomotor behavior. It is possible that the correspondence that was observed among VA,
NO and OF tests may be due to all three of these tests assessing components of turkeys’ activity
levels and motor patterns. Contrarily, test measures that assess vastly different behavioral
patterns, such as vocalization vs. number of steps taken, may be less likely to correspond among
tests.
Another explanation for a lack of test correspondence is that animals’ responses to fear
tests may be reflective of other motivations or behavioral patterns and not necessarily fear per se
(Boissy, 1995). Some tests may measure exploratory motivation, social reinstatement motivation
or locomotion/general activity level (e.g. open field test) whereas other tests may only measure
fear of novelty (e.g. novel object test conducted in home pen). Indeed, the OF test not only
measures fear of novelty, but it also subjects animals to social isolation (Forkman et al., 2007)
and has been used to assess activity levels of turkeys (Huff et al., 2007).
Reviews specifically examining fear and fearfulness in poultry have concluded that there
are associations among certain fear tests, which indicates that these fear tests are assessing the
same underlying responses (R. B. Jones, 1996) and are valid (Forkman et al., 2006). A number of

83

studies with birds and other animals have shown that animals maintain their position in a fear
hierarchy, indicating stability of the fear response across situations (R. B. Jones, 1996).
Furthermore, birds in top tiers of battery cages had greater avoidance of a NO, longer TI
durations and increased latencies to approach humans (reviewed in Forkman, 2007). However,
more recent research has cast doubt upon the association among fear tests and the notion that
different tests measure the same, single, underlying fear state. For example, chickens that
received more human contact displayed shorter avoidance distances in an approaching human
test, but did not differ in their responses to a novel object compared to chickens receiving less
human contact (Graml et al., 2008). These results suggest that fear of humans and fear of novel
objects are independent. Similarly, Miller et al. (2006) reported poor correspondence among four
different tests of fear in quail (emergence test, predator test, novel object test, and novel food
test), indicating that measures in one fear test do not correspond with measures recorded in other
fear tests. When genetic selection for TI duration was used as a different strategy to examine the
correspondence between fear tests, quail selected for long (LTI) and short (STI) TI durations did
not differ in responses to novel objects, suggesting that TI duration is not associated with fear of
novelty (Richard et al., 2008).
Our results reveal a complex relationship among TI, OF, NO and VA tests. When results
from Age 1 are considered, our results are in agreement with others (Graml et al., 2008; Miller et
al., 2006; and Richard et al., 2008) that there is poor correspondence among fear tests. However,
there may be stronger associations among fear test measures as animals age. The lack of
agreement in the literature and the contradictory findings regarding fear test correspondence may
be due to differences in methodologies and strategies used. Alternatively, the contradictory
findings and lack of agreement may indicate that tests are measuring specific responses to

84

specific components or stimuli present during testing, and are not capturing a larger, general fear
response. Indeed, several researchers have stated that fear responses may depend largely on the
context (Miller et al., 2006) and that fear is multidimensional (Richard et al., 2008; 2010).
In conclusion, there appears to be a weak association between turkeys’ willingness to
move toward an observer in their pen and their ambulatory behavior in an open field test between
4 and 6 weeks of age. There may be stronger associations among test situations between 8 and 10
weeks of age, but further research is needed to examine age-related differences in fear test
correspondence independent of test habituation and repeated testing. Results did not validate that
the OF, TI, VA and NO tests are measuring underlying fearfulness as a unitary construct, or that
these tests are measuring the same types of fear responses. This study only assessed convergent
validity (whether there is agreement among tests intended to measure similar behavioral
responses) and not discriminant validity (whether tests disagree with other tests that are intended
to measure different responses; Cozby, 1993). In addition to evaluating discriminant validity,
additional research is needed to examine how external environmental conditions, genetics, and
sex influence turkeys’ OF, TI, VA and NO responses.

ACKNOWLEDGEMENTS
We thank Hybrid Turkeys for supplying the poults, and Dr. Darrin Karcher and Angelo
Napolitano for their assistance with this project. We also thank Erika First, Caitlin Hummel,
Sarah Jager, Cheyenne Ryzenga and Shannon Williams for assistance with data collection. M.E.
was supported by a Natural Sciences and Engineering Research Council of Canada postgraduate
scholarship and this work was supported by Michigan State University AgBioResearch.

85

APPENDIX

86

Table 3.1. Spearman rank correlation coefficients (rS) of open field (OF), tonic immobility (TI),
voluntary approach (VA), and novel object (NO) test measures.
_NO test_
Lat.
peck

____________VA test____________
Lat. Two Lat. one Lat.
Lat.
BL
BL
approach peck
Age 1 (N = 16)

___TI test___
Lat.
Voc.
voc.
no.

OF test
Lat. amb.
Step no
Square no.
Def. no.

-0.31
-0.14
-0.1
0.13

-0.31
0.35
0.35
-0.2

0.14
0.29
0.28
-0.33

0.06
0.42
0.37
-0.29

-0.01
0.46 T
0.43 T
-0.22

-0.03
-0.36
-0.35
-0.09

0.06
0.32
0.3
0.04

TI test
Lat. voc.
Voc. no.

-0.02
0.04

0.07
-0.12

0.1
-0.02

0.14
-0.1

0.006
-0.01

VA test
Lat. two BL
Lat. one BL
Lat. approach
Lat. peck

-0.21
-0.2
-0.23
-0.15
-0.42
0.25
0.38
0.34

0.50*
-0.51*
-0.56*
-0.50*

Age 2 (N = 16)
OF test
Lat. amb.
Step no
Square no.
Def. no.

0.28
-0.50T
-0.44 T
-0.36

0.15
-0.29
-0.37
-0.33

0.11
-0.27
-0.22
-0.17

0.1
-0.007
0.02
0.01

0.17
0.1
0.15
0.15

TI test
Lat. voc.
Voc. no.

-0.04
0.05

-0.24
0.18

-0.17
0.19

-0.13
0.02

-0.19
-0.04

VA test
Lat. two BL
Lat. one BL
Lat. approach
Lat. peck

0.39
0.26
0.04
0.22

Note: Lat. = latency, BL = body length, amb = ambulate, voc. = vocalize.
T
P < 0.10, *P < 0.05.

87

Table 3.2. Wilcoxon rank sum test differences in median (25-75% quartiles) values of open field (OF), tonic immobility (TI),
voluntary approach (VA) and novel object (NO) test measures for turkeys classified as high (HR) or low (LR) responders in each test
at Age 1 (4 to 6 weeks).
__OF test classification__ __TI test classification__
OF-HR
OF-LR
TI-HR
TI-LR
(N = 9)
(N = 15)
(N = 14)
(N = 15)
OF test responses
Lat. amb. (s)

600
(376-600)

600
(513-600)

Step no.
Square no.

0 (0-8)
0 (0-1)

0 (0-2)
0 (0-0)

Def. no.

3 (1-4)

2 (1-3)

TI test responses
Voc. Lat. (s)
780
(482-780)
Voc. no.
0 (0-21)

780
(320-780)
0 (0-13)

VA test responses
Lat. two BL (s) 37
(25-87)
Lat. one BL (s) 90
(27-688)
Lat. app. (s)
91
(44-698)
Lat. peck (s)
340
(53-900)

88
(44-327)
243
(89-773)
244
(116-776)
370
(221-777)

74
(25-390)
118
(29-461)
369
(91-900)
451
(172-900)

74
(35-387)
246
(77-688)
363
(103-858)
477
(157-900)

__VA test classification__
VA-HR
VA-LR
(N = 12)
(N = 15)

__NO test classification__
NO-HR
NO-LR
(N = 10)
(N = 15)

600
a
(358-600)
0 (0-63)

600
b
(600-600)
0 (0-2)

600
(513-600)

600
(358-600)

0 (0-16)
2 (1-4)

0 (0-0)
1 (1-3)

0 (0-14)
0 (0-3)

0 (0-18)
0 (0-1)

2 (1-3)

2 (1-3)

780
(303-780)
0 (0-36)

680
(137-680)
2 (0-27)

430
(132-780)
2 (0-27)

780
(140-780)
0 (0-39)

203
(71-900)
292
(114-900)
400
(124-900)
463
(221-900)

74
(15-387)
220
(21-688)
225
(62-858)
354
(227-900)

a

b

Note: amb. = ambulate, app. = approach, lat. = latency, def. = defecation, voc. = vocalization, BL = body length.
a, b
Rows within test classifications lacking common superscripts differ (P < 0.10).

88

Table 3.2. (cont’d).
__OF test classification__ __TI test classification__
OF-HR
OF-LR
TI-HR
TI-LR
(N = 9)
(N = 15)
(N = 14)
(N = 15)
NO test responses
Lat. peck (s)
900
(339-900)

900
(283-900)

899
(115-900)

900
(353-900)

__VA test classification__
VA-HR
VA-LR
(N = 12)
(N = 15)
900
(416-900)

__NO test classification__
NO-HR
NO-LR
(N = 10)
(N = 15)

900
(44-900)

Note: amb. = ambulate, app. = approach, lat. = latency, def. = defecation, voc. = vocalization, BL = body length.
a, b
Rows within test classifications lacking common superscripts differ (P < 0.10).

89

Table 3.3. Wilcoxon rank sum test differences in median (25-75% quartiles) values of open field (OF), tonic immobility (TI),
voluntary approach (VA) and novel object (NO) test measures for turkeys classified as high (HR) or low (LR) responders in each test
at Age 2 (8 to 10 weeks).
__OF test classification__
OF-HR
OF-LR
(N = 13)
(N = 17)
OF test responses
Lat. amb. (s)

__TI test classification__
TI-HR
TI-LR
(N = 13)
(N = 16)

__VA test classification__
VA-HR
VA-LR
(N = 15)
(N = 15)

__NO test classification__
NO-HR
NO-LR
(N = 13)
(N = 16)

600
(117-600)

528
(178-600)

410
(103-600)

600
(256-600)

508
e
(74-600)

600
f
(600-600)

Step no.

7 (1-8)

6 (0-37)

5(2-37)

1 (0-10)

8(7-10)

0 (0-3)

Square no.

0 (0-1)

1 (0-5)

0 (0-5)

0 (0-1)

1 (0-2)

0 (0-0)

Def. no.

2 (1-3)

1 (1-2)

1 (1-2)

2 (1-3)

1 (1-2)

780
(184-780)
0 (0-4)

780
(780-780)
0 (0-0)

780
(310-780)
0 (0-102)

780
(780-780)
0 (0-0)
29
b
(24-45)
47
b
(37-61)
56
(37-101)
80
b
(58-167)

TI test responses
Voc. Lat. (s)
780
(630-780
Voc. no.
0 (0-2)
VA test responses
17
Lat. two BL
a
(s)
(12-26)
Lat. one BL (s) 26
c
(20-40)
Lat. app. (s)
37
a
(26-55)
Lat. peck (s)
55
(36-78)

a

b

2 (1-2)

780
(262-780)
0 (0-1)
34
b
(23-43)
49
d
(34-79)
89
b
(38-116)
91
(55-178)

e

a
c

23
(13-37

26
(15-59)

36
(29-46)

40
(21-70)

45
(37-112)

53
(25-85)

17
a
(11-26)
29
a
(21-40)
38
(26-46)

83
(55-133)

65
(35-137)

51
a
(29-83)

Note: amb. = ambulate, app. = approach, lat. = latency, def. = defecation, voc. = vocalization, BL = body length.
a-f
a, b
c, d
e, f
Rows within test classifications lacking common superscripts differ ( P < 0.10,
P < 0.05, P < 0.01).

90

f

b
d

Table 3.3. (cont’d).
__OF test classification__
OF-HR
OF-LR
(N = 13)
(N = 17)
NO test response
Lat. peck (s)
46
(25-432)

407
(32-900)

__TI test classification__
TI-HR
TI-LR
(N = 13)
(N = 16)

__VA test classification__
VA-HR
VA-LR
(N = 15)
(N = 15)

80
(25-900)

38
a
(18-900)

863
(61-900)

__NO test classification__
NO-HR
NO-LR
(N = 13)
(N = 16)

407
b
(83-900)

Note: amb. = ambulate, app. = approach, lat. = latency, def. = defecation, voc. = vocalization, BL = body length.
a-f
a, b
c, d
e, f
Rows within test classifications lacking common superscripts differ ( P < 0.10,
P < 0.05, P < 0.01).

91

REFERENCES

92

REFERENCES
Albentosa, M. J., Kjaer, J. B., & Nicol, C. J. (2003). Strain and age differences in behaviour, fear
response and pecking tendency in laying hens. British Poultry Science, 44, 333-344.
Boissy, A. Fear and fearfulness in animals. (1995). The Quarterly Review of Biology, 70, 165191.
Bourin, M., Petit-Demoulière, B., Dhonnchadha, B. N., & Hascöet, M. (2007). Animal models of
anxiety in mice. Fundamental & Clinical Pharmacology, 21, 567–574.
Cockrem, J. F. (2007). Stress, corticosterone responses and avian personalities. Journal of
Ornithology, 148 (Suppl 2), S169–S178.
Cozby, P. C. (1993). Methods in Behavioral Research. Mountain View, California: Mayfield.
Erasmus, M. A., & Swanson, J. C. (2014) Temperamental turkeys: reliability of behavioural
responses to four tests of fear. Applied Animal Behaviour Science, 157: 100-108.
Forkman, B., Boissy, A., Maunier-Salaün, M. -C., Canali, E, & Jones, R. B. (2007). A critical
review of fear tests used on cattle, pigs, sheep, poultry and horses. Physiology &
Behavior, 92, 340-374.
Gosling, S. D. (2001). From mice to men: what can we learn about personality from animal
research? Psychological Bulletin, 127, 45-86.
Graml., C., Waiblinger, S., & Niebuhr, K. (2008). Validation of tests for on-farm assessment of
the hen-human relationship in non-cage systems. Applied Animal Behaviour Science,
111, 301-310.
Groothuis, T. G. G., & Carere, C. (2005). Avian personalities: characterization and epigenesis.
Neuroscience and Biobehavioral Reviews, 29, 137-150.
Hazard, D., Leclaire, S. Couty, M. & Guémené, D. (2008). Genetic differences in coping
strategies in response to prolonged and repeated restraint in Japanese quail divergently
selected for long or short tonic immobility. Hormones and Behavior, 54, 645-653.
Hocking , P. M., Channing , C. E., Waddington, D. & Jones, R. B. (2001). Age-related changes
in fear, sociality and pecking behaviours in two strains of laying hen. British Poultry
Science, 42, 414–423.
Huff, G., Huff, W., Rath, N., Donoghue, A., Anthony, N., & Nestor, K. (2007). Differential
effects of sex and genetics on behavior and stress response of turkeys. Poultry Science,
86, 1294-1303.

93

Jones, A. C., & Gosling, S. D. (2005). Temperament and personality in dogs (Canis familiaris): a
review and evaluation of past research. Applied Animal Behaviour Science, 95, 1-53.
Jones, R. B. (1996). Fear and adaptability in poultry: insights, implications and imperatives.
World’s Poultry Science Journal, 52, 131-174.
Jones, B., & Boissy, A. (2011). Fear and other negative emotions. In M. C. Appleby, J. A.
Mench, I. A. S. Olsson, & B. O. Hughes (Eds.), Animal welfare (2nd ed.). (pp. 81).
Cambridge, MA, USA: CAB International North America.
Koolhaas, J. M., Korte, S. M., De Boer, S. F., Van Der Vegt, B. J., Van Reenen, C. G., Hopster,
H. et al. (1999). Coping styles in animals: current status in behavior and stressphysiology. Neuroscience and Biobehavioral Reviews, 23, 925–935.
Marín, R. H., Martijena, I. D., & Arce, A. (1997). Effect of Diazepam and a β-Carboline on
open-field and T-maze behaviors in 2-day-old chicks. Pharmacology Biochemistry and
Behavior, 58, 915-921.
Martin, P., & Bateson, P. (1993). Measuring behaviour: an introductory guide (2nd ed.). New
York, USA: Cambridge University Press.
Miller, K. A., Garner, J. P., & Mench, J. A. (2006). Is fearfulness a trait that can be measured
with behavioral tests? A validation of four fear tests for Japanese quail. Animal
Behaviour, 71, 1323-1334.
Mazurek, M., McGeeb, M., Crowe, M. A., Prendiville, D. J., Boivin, X., & Earley, B. (2011).
Consistency and stability of behavioural fear responses of heifers to different feareliciting situations involving humans. Applied Animal Behaviour Science, 131, 21-28.
Neumann, I. D. Wegener, G., Homberg, J. R, Cohen, H., Slattery, D. A., Zohar, J., et al. (2011).
Animal models of depression and anxiety: What do they tell us about human condition?
Progress in Neuro-Psychopharmacology & Biological Psychiatry, 35, 1357-1375.
Petherick, J. C., Holroyd, R. G., Doogan, V. J., & Venus, B. K. (2002). Productivity, carcass and
meat quality of lot-fed Bos indicus cross steers grouped according to temperament.
Australian Journal of Experimental Agriculture, 42, 389-398.
Remignon, H., Mills, A. D., Guémené, D., Desrosiers, V., Garreau-Mills, M., Marche, M. et al.
(1998). Meat quality traits and muscle characteristics in high or low fear lines of Japanese
quails (Coturnix japonica) subjected to acute stress. British Poultry Science, 39, 372–
378.
Richard, S., Wacrenier-Ceré, N., Hazard, D., Saint-Dizier, H., Arnould, C., & Faure, J. M.
(2008). Behavioral and endocrine fear responses in Japanese quail upon presentation of a
novel object in the home cage. Behavioural Processes, 77, 313-319.

94

Richard, S., Land, N., Saint-Dizier, H., Leterrier, C., & Faure, J. M. (2010). Human handling and
presentation of a novel object evoke independent dimensions of fear in Japanese quail.
Behavioural Processes, 85, 18-23.
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.
Stamps, J., & Groothuis, T. G. G. (2010). The development of animal personality: relevance,
concepts and perspectives. Biological Reviews, 85, 301-325.
Suarez, S.D. & Gallup, G. G. (1983). Social reinstatement and open field testing in chickens.
Animal Learning & Behavior, 11, 119-126.
Voisinet, B. D., Grandin, T., O’Connor, S. F., Tatuma, J. D., & Deesing, M. J. (1997). Bos
Indicus-cross feedlot cattle with excitable temperaments have tougher meat and a higher
incidence of borderline dark cutters. Meat Science, 4, 367-377.
Welfare Quality®. 2009. Welfare Quality® Assessment Protocol for Poultry. Welfare Quality
Consortium, Lelystad, Netherlands.

95

CHAPTER 4: FEATHER PECKING AND OPEN FIELD BEHAVIOR OF TURKEYS
ABSTRACT
Feather pecking among commercial turkeys is an economic and welfare concern, but
there has been scant research on this topic. Research with chickens has provided evidence that
feather pecking may be associated with differences in temperament characteristics, such as
fearfulness. The objectives of this research were to 1) examine fear responses using the open
field (OF) test in males and females of a commercial (COMM) and randombred (RB) turkey
strain, 2) compare feather pecking between sexes and strains, and 3) ascertain whether
behavioral and physiological OF responses are associated with feather pecking. Turkeys were
housed in same-sex, same-strain groups of 14-20 in 16 pens. Turkeys were individually tested
three times in an OF test (n = 16 per pen). Males were tested at 1 (Test 1), 4 (Test 2), and 11
(Test 3) wk; females were tested at 1 (Test 1), 5 (Test 2) and 12 (Test 3) wk. Pre-and post-test
plasma corticosterone levels (n = 6 per pen) were assessed during Test 3. Prior to Test 3,
behavioral observations were conducted to identify birds that developed feather pecking so that
OF responses could be compared between birds that pecked and birds that did not. Results
revealed some sex differences in OF behavior between COMM males and females in Tests 2 and
3 and between RB males and females in Test 3. Strain differences in OF behavior were found in
all tests. Corticosterone levels following testing were higher in RB females compared to COMM
females. There were no differences in the frequency of feather pecking, but feather damage at 13
wk was worse in COMM compared to RB birds. No significant relationships were found
between feather pecking and behavioral or physiological OF responses. Overall, these results
indicate some sex and strain differences in OF behavior, but the nature of the differences
depended on when birds were tested. Our results do not support the hypothesis that feather
96

pecking is associated with differences in temperament measured using the OF test, and these
findings provide further evidence of the complex, multifactorial nature of feather pecking of
turkeys.

INTRODUCTION
Global demand for food continues to increase, and the projected increase in the world’s
population will double the current need for food (Federation of Animal Science Societies, 2012).
Therefore, there is pressure to develop faster growing, higher yielding and more efficient
production animals. Concurrently, public interest in animal welfare continues to grow, and the
welfare state of an animal can have significant and direct impacts on food quality (Blokhuis et
al., 2008). Injurious pecking, including head pecking, feather pecking and cannabalism, is an
important factor affecting food quality and animal welfare among intensively housed turkeys
(Hocking et al., 1993). Indeed, injurious pecking results in culling, mortality, and carcass
damage. However, research examining injurious pecking among turkeys is scarce, unlike
research regarding injurious pecking of chickens. In a survey of commercial turkey producers in
Canada, Erasmus (2009) reported that 13-14% of culls on farms were due to injuries. More
recently, injuries associated with severe pecking have been reported to account for 58% of culls
and mortalities at a commercial turkey facility (Duggan et al., 2014). The increasing demand for
animal products and more efficient animal production increases concerns regarding animal
welfare, making research examining injurious pecking among turkeys paramount.
In their review of injurious pecking of turkeys, Dalton et al. (2013) state that severe
feather pecking and canibalism are “multi-factorial products of genetics, environment, and
nutrition”. However, not all birds develop feather pecking. Indeed, individual differences are

97

known to exist between laying hens that develop feather pecking and hens that do not
(Rodenburg et al., 2008). For example, several studies have confirmed the relationship between
temperament (coping style and fearfulness) and feather pecking in two lines of laying hens
selected for high (HFP) and low (LFP) feather pecking, respectively (e.g. Korte et al., 1997; Van
Hierden et al, 2002; Rodenburg et al., 2004; Jensen et al., 2005).
Coping style is defined as a consistent set of physiological and behavioral responses to
stressors (Koolhaas et al., 1999). Proactive coping styles are characterized by an active
behavioral response (fight or flight) and low hypothalamo-pituitary-adrenal (HPA) axis reactivity
(low corticosterone levels) whereas a reactive coping style is characterized by reduced activity
(immobility) and high HPA axis reactivity (Koolhaas et al., 1999). As stated by de Haas et al.
(2010), chickens from the HFP line exhibit characteristics consistent with proactive coping
styles, whereas chickens from the LFP line exhibit characteristics consistent with reactive coping
styles (Korte et al., 1997; Van Hierden et al, 2002; Rodenburg et al., 2004; Jensen et al., 2005).
Specifically, HFP chickens had lower basal and post-test corticosterone levels following a
manual restraint test compared to LFP chickens (Korte et al., 1997; Van Hierden et al., 2002;
reviewed in Rodenburg et al., 2008). There is also evidence that chickens that differ in feather
pecking performance differ in fearfulness, which some scientists regard as being part of coping
styles (Cockrem et al., 2007), whereas others regard fearfulness as an altogether separate
dimension of temperament (Jones and Boissy, 2011; Koolhaas et al., 1999).
Fearfulness is often tested using the open field test (Forkman et al., 2007). Chickens from
the LFP line were less fearful when tested in an open field test (Jones et al., 1995). Moreover,
open field behavior of chicks was associated with their feather pecking behavior as adults in the
F2 generation of a cross between the HFP and LFP lines (Rodenburg et al., 2004). In a recent

98

study, anxiety (tested using a social isolation test) at 1 wk was found to be associated with
feather damage at 5 wk (de Haas et al., 2014). Therefore, individual differences in temperament
may be important indicators of, or contributors to, the likelihood of developing feather pecking,
but this has not been examined in turkeys.
Our study examined the relationship between fearfulness and feather pecking in turkeys
of two genetic strains. Specifically, we examined whether fearfulness (tested using the open field
test) and feather pecking differ between randombred (RB) and commercial (COMM) turkey
strains and between sexes within strains. We tested the hypothesis that behavioral and
physiological responses to open field testing differ between turkeys that develop feather pecking
and turkeys that do not.

MATERIALS AND METHODS
All procedures were approved by the Institutional Animal Care and Use Committee of
Michigan State University.

Animals and housing
A total of 80 male and 80 female commercial (COMM) turkeys (Hybrid Converter strain,
Hybrid Turkeys, Kitchener, On., Canada) and 62 male and 62 female randombred (RB) turkeys
(RBC2 line established in 1966 (Nestor et al., 1969) and maintained at the The Ohio Agricultural
Research and Development Center (OARDC) of The Ohio State University, Wooster, OH) were
used in this study. In total, 62 male and female RB turkeys were obtained. The RB turkeys were
incubated and hatched at the OARDC of the Ohio State University. COMM poults were obtained
from a commercial hatchery and placed 1 d prior to placement of RB poults. Poults were

99

brooded in same-sex, same-strain groups in eight littered (wood shavings) pens (2.40 m x 3.05
m) in the same barn from 1 d to 2 wk of age. At 2 wk, birds were distributed throughout the barn
to one of 16 littered pens (2.4 m x 3.05 m) and housed in same-sex, same-strain groups (14 to 20
birds per pen) where they remained until study completion at 14 wk. sexes and strains were
randomly assigned to pens such that the location of males and females of each sex was random
within the barn. At 7 wk, some COMM turkeys were removed from their pens for use in another
study and the number of COMM turkeys in each pen was reduced to, and maintained at 16.
Therefore, group sizes were similar across sexes and strains from 7 wk onward (14-16
birds/pen).
Temperature and diet were the same as described in Erasmus and Swanson (2014). A
photoperiod of 24L: 0D was applied for the first 7 d, and then the light period was reduced by 1
hr/d for 7 d. A photoperiod of 16L: 8D was maintained from 14 d - 14 wk. Lights came on at
06:00 h and were turned off at 22:00 h. Light intensity was maintained at 20 lx.
At 1 wk, turkeys were individually identified using colored (green, black, purple, and
orange) non-toxic livestock marker (Prima Tech Marking Stick, Neogen Corporation, Lansing,
MI, USA) which was applied to different areas and in different color combinations for each bird
in the pen (e.g. black and/or green on the shoulders, left or right wing) and was visible on an
overhead camera. Marker was reapplied biweekly to ensure markings remained visible. Turkeys
were weighed prior to OF testing at 23 and 73 d of age (one day older for RB birds).

Open field testing procedures
We previously reported that the open field (OF) test is reliable for assessing fear
responses of commercial male turkeys (Erasmus and Swanson, 2014). In order to examine

100

whether OF test responses at different ages were associated with feather pecking behavior, 14-16
turkeys from each pen of male and female COMM and RB turkeys were tested in the OF test at
three ages (three tests). Behavior in the OF was recorded during Test 1 (1 wk of age for males
and females), Test 2 (males: 4 wk, females: 5 wk), and Test 3 (males: 11 wk, females: 12 wk).
During the first test week, testing occurred over two consecutive days for males, followed by two
consecutive days for females. During all other test weeks, testing occurred over four consecutive
days so that a total of 24 birds could be tested each day. Each bird was only tested once during
each test week. Test order was randomized for each day of testing so that COMM and RB birds
from each sex were tested in random order. All testing took place between 07:45 h and 17:45 h.
Testing procedures were the same as described in Erasmus and Swanson (2014). Briefly,
turkeys were individually tested in an OF test arena (2.74 m (length) x 2.74 m (width) x 1.83 m
(height)). Each turkey was placed in the center of the arena for 10 min and behaviour was
recorded using two overhead cameras (VIXIA HF M41, Canon USA, Inc. Melville, NY, USA)
from which behavioral data (latency to ambulate, number of steps taken, number of squares
entered, latency to vocalize, number of vocalizations, and number of defecations) were collected.
A vocalization was defined as a sound emitted by the bird being observed. Vocalizations were
distinguished from one another by pauses (periods of no sound) between vocalizations and each
vocalization was counted separately. Some males gobbled during OF test 3. A sequence of
gobbling sounds was counted as one vocalization if there was no pause between sounds.

Feather pecking behavior
Behavior of turkeys was video recorded on three days to examine feather pecking
interactions. Video was recorded from 07:30 h to 20:30 h at 74, 75 and 81 d of age for COMM

101

males (one day older for RB males) and 74, 81 and 82 d for COMM females (one day older for
RB females). This pattern of video recording was used so that behavior could be examined
within three days prior to the last OF test, and so that there were two days of overlapping video
between males and females. Four pens were recorded using one type of camera (2MB70IR42L210 Jaguar Series infrared CCTV camera, 2M® Technology, Grand Prairie, TX, USA)
and the other 12 pens were recorded using a similar camera but from a different manufacturer
(Clinton Electronics CE-VF540 CCTV camera, Clinton Electronics Corp., IL, USA).
A pecking interaction was defined as “repeated pecks to the same individual” (HuberEicher and Sebö, 2001). The interaction was considered to have ended if there were no further
pecks within a 5 s period. The frequency of feather pecking interactions was calculated for each
pen as the number of feather pecking interactions per individual per day.
Individuals performing repeated (three or more pecks in succession) gentle or severe
feather pecking (PECK) were identified from video recordings by recording feather pecking
every 10 min. Briefly, the video was stopped every 10 min and watched for 10 s before and 10 s
after the 10 min time point to verify that feather pecking behavior was occurring and to score
feather pecking as gentle or severe. Because feather pecking occurs in bouts, feather pecking
behavior was recorded as having occurred if 1) it was observed at exactly the 10 min time point,
and 2) if the bird performing the pecking was observed to peck immediately before and
immediately after the 10 min time point, pausing for 5 s or less around the 10 min time point
without performing any other behavior. Sometimes birds remove and eat feathers from other
birds and some birds may peck at a slower rate than others. Therefore, scoring feather pecking
behavior within 5 s surrounding the 10 min time point enables feather pecking bouts to be more

102

fully represented because the bird has not changed behavior and can be considered to still be
engaged in feather pecking behavior.
Gentle and severe feather pecking were defined according to its definition for chickens
(Savory, 1995). Briefly, gentle feather pecking was defined as pecking that did not elicit a
response from the recipient. Severe feather pecking included forceful pecking, pulling and/or
removal of feathers resulting in feather damage and causing the recipient to react or move away.
Turkeys were classified as PECK only if they were observed to feather peck on two or more
occasions to reduce risks of misclassifying birds (Jensen et al., 2005). Turkeys that were not
observed to perform feather pecking were classified as NPECK. For each feather pecking
interaction, the behavior of the recipient bird at the time of feather pecking was also noted.
Behavioral categories of the recipient bird included resource (the recipient was eating or
drinking), preen, dustbathe, or other (sitting, standing, walking and environmental pecking).
The frequency of severe feather pecking interactions and the frequency of interactions
occurring for each behavioral classification of the recipient bird were calculated as the
percentage of interactions that were observed in each pen divided by the total number of feather
pecking observations, and then divided by the number of birds in the pen. The number of birds in
each pen was included in calculations to account for differences in feather pecking due to group
size. Percentages are reported as the mean of the 4 pens for males and females of each strain.
Reliability of feather pecking video analysis was r ≥ 0.8 between observers. The same
observer further examined all feather pecking data that were collected and verified identities of
PECK birds, gentle vs. severe pecking and behavior of the recipient during pecking.
Turkeys were examined daily for pecking wounds. Turkeys with wounds that had penetrated
the skin were treated with Rooster Booster® Pick-no-more lotion (TDL Industries, Inc. Fallon,

103

NV, USA) to prevent further pecking. If the pecking injury worsened, the turkey was moved to a
hospital pen to prevent further injury. In severe cases of head pecking, turkeys were culled if
necessary.
Turkeys were scored for feather pecking damage by scoring feather condition at 13 wk
using the method of Bilčík and Keeling (1999) to score the back, neck, tail, and primary left and
right wing feathers. A score of 0 represented intact feathers and little to no damage, whereas a
score of 5 indicated complete or almost complete feather loss.

Plasma corticosterone levels
Blood samples were collected from six birds in each pen (n = 24 birds/sex-strain
combination) during the second and third test periods, but only samples from Test 3 were
assayed. Blood samples were collected one day prior to OF testing in order to obtain pre-test
corticosterone levels. Samples were collected 15 min (± 10 min) after the time at which testing
was set to start the following day. Blood samples were then collected the following day 15 min
after OF testing commenced in order to examine the magnitude of corticosterone increase in
response to testing. Birds were gently restrained on their side with one person holding the bird’s
legs and wing and the other person collecting blood from the brachial vein using a 25 gauge
needle. No more than 2.2 ml of blood was collected each time. Blood was placed into EDTAcoated Vacutainer tubes (Fisher Scientific, NH, USA) and kept on ice until blood could be
centrifuged (1000 rpm). After centrifugation, plasma was collected and stored at -20 °C until
assayed. The same competitive enzyme immunoassay kit (Corticosterone HS, IDS Inc., Fountain
Hills, AZ) used by Huff et al. (2007) was used to determine plasma corticosterone levels.

104

Statistical analysis
All data were analyzed using SAS 9.4 (SAS Institute Inc., Cary, NC.). Behavioral OF test
responses were not normally distributed. Test-retest reliability of OF responses was analyzed
using Spearman rank correlation. Correlation coefficients were defined according to the
definition of Martin and Bateson (1993) (low: 0.2 – 0.4, moderate: 0.4 – 0.7, high: 0.7 – 0.9, and
very high: 0.9 – 1.0). Sex and strain differences in behavioral responses in the OF test that
consisted of counts (e.g. number of vocalizations) were analyzed using the GLIMMIX procedure
for a negative binomial distribution with the log link function. Data were analyzed separately for
each OF test period. The main effects of sex, strain and their interaction were examined.
Differences in latency measures (e.g. latency to ambulate) between sexes within strains, and
between males or females of different strains were tested using the LIFETEST procedure for
each test period. The variable sex or strain was specified in the STRATA statement when sex or
strain differences, respectively, were examined.
The MIXED procedure was used to compare the frequency of feather pecking
interactions, the percentage of birds that were classified as PECK, and the percentage of feather
pecking interactions occurring for each behavioral category of the recipient bird. The main
effects of sex, strain and their interaction were included in the model. The percentage of birds in
each pen that were classified as PECK was log(+ 1)-transformed, and the mean percentage of
birds that performed feather pecking was cosine-transformed to meet normality assumptions.
Raw means are reported.
Feather scores were compared between sexes and strains using the GLIMMIX procedure
for a negative binomial distribution with the log link function. The main effects of sex, strain and
their interaction were included in the model.

105

Differences in OF responses consisting of counts were compared between PECK and
NPECK turkeys within males and females of each strain using the GLIMMIX procedure for a
negative binomial distribution with the log link function. Separate analyses were conducted for
each OF test period to determine whether OF behavior at any of the ages tested was associated
with feather pecking behavior. The main effects of sex, strain, classification (PECK or NPECK)
and their three-way interaction were tested. Body weight was included as a covariate for OF Test
2 and OF Test 3 models (body weights had not been recorded for OF Test 1). Latency measures
were compared between PECK and NPECK birds using the LIFETEST procedure. Pen was
specified in the STRATA statement and classification was specified as the GROUP variable.
Body weight was included as a covariate in the TEST statement for OF Test 2 and Test 3.
Differences in pre- and post-test corticosterone levels, and in the change in
corticosterone level due to testing (post-test level – pre-test level) were compared between sexes
and strains using the MIXED procedure. The main effects of sex, strain and their interaction
were tested. Pen was included as a random effect. Results are reported as raw means. Similarly,
differences in pre- and post-test corticosterone levels were compared between PECK and
NPECK birds using the MIXED procedure. The main effects of sex, strain, pecking classification
and the interaction between sex, strain and pecking classification were included in the model.
Pen was included as a random effect. Pre- and post-test corticosterone levels were logtransformed to achieve normality. The change in corticosterone level was transformed using log
+ 1 to achieve normality.
All post-hoc analyses were conducted using Tukey’s test for multiple comparisons. P
values < 0.05 were considered significant.

106

RESULTS
At 23 d of age, body weights (mean ± SE) were 0.9 ± 0.02 kg (COMM males), 0.6 ±
0.007 kg (RB males), 0.8 ± 0.02 kg (COMM females), and 0.5 ± 0.006 kg (RB females). Body
weights were 8.4 ± 0.2 kg (COMM males), 4.4 ± 0.07 kg (RB males), 6.4 ± 0.2 kg (COMM
females) and 3.2 ± 0.06 kg (RB females) at 73 d of age. At 7 wk, one RB male was moved to a
hospital pen to prevent further injury from pecking. Between 11 and 13 wk, 2 RB males from the
same pen and 3 COMM females (2 from the same pen) were moved to the hospital pen. Two RB
males from another pen were culled at 10 and 13 wk, respectively, due to head pecking injuries.
Due to problems with video color, identities of PECK birds could not reliably be
determined on 1 d of video recording for one RB female pen, one COMM male pen, and one
COMM female pen, but the number of pecking interactions and behavior of recipients could still
be recorded.

Test-retest reliability of OF behavior
Spearman rank correlation coefficients for correlations between OF tests are presented in
Table 4.1. In general, a greater number of significant correlations were found, and correlation
coefficients were larger between Test 2 and Test 3 than between Tests 1 and 2, or between Tests
1 and 3.
For COMM males, the only significant correlation that occurred between Test 1 and Test
2 was a tendency for the number of squares entered to be positively correlated. Significant,
moderate correlations were found between Test 1 and Test 3 for the latency to ambulate and
number of steps taken, and there was a tendency for a low correlation for the latency to vocalize
and the number of squares entered. Significant, moderate correlations were found between Test 2

107

and Test 3 for all OF test responses except for the latency to ambulate and the number of squares
entered, which had low, positive correlation coefficients.
There was a tendency for a low, positive correlation between Tests 1 and 2 for the
number of defecations of RB males, but no other significant correlations were found between
Tests 1 and 2, or Tests 1 and 3. There were significant, moderate correlations between Tests 2
and 3 for the latency to ambulate, number of steps and number of squares, and low correlations
for the latency to vocalize and numbers of vocalizations and defecations.
For COMM females, the number of squares entered was significantly, moderately
correlated between Tests 1 and 2, and there was a low correlation for the latency to ambulate.
Additionally, the numbers of steps and vocalizations tended to be positively correlated between
Tests 1 and 2. There were moderate correlations between Tests 1 and 3 for the latency to
ambulate and number of defecations, and low correlations were found for the number of squares
entered. The latency to vocalize tended to be correlated between Tests 1 and 3. Significant,
moderate correlations were found between Test 2 and Test 3 for all OF test responses.
Significant correlations were found between all test periods for almost all behavioral OF
responses of RB females. For Test 1 vs. 2, there were moderate correlations for the latency to
ambulate and number of defecations, and low correlations for all other OF responses. Significant
moderate correlations were found between Test 1 and Test 3 for all OF responses. There were
significant, high correlations between Tests 2 and 3 for the numbers of steps and squares, and
moderate correlations for all other OF responses.

Sex and strain differences in OF test responses
Behavioral OF responses for all 3 OF test periods are presented in Table 4.2.

108

Test 1. There was a significant sex-strain interaction for the number of steps taken and
number of squares entered. Post-hoc comparisons revealed that the number of steps taken and
squares entered were significantly higher for female RB turkeys compared to male RB turkeys,
and there was a tendency for the number of vocalizations to differ between RB males and
females (t156 = 2.5, P = 0.07). Both the latency to ambulate and the latency to vocalize were
significantly shorter for RB females compared to COMM males and females.
Test 2. The numbers of steps taken and squares entered did not differ between sexes or
strains. However, the main effect of sex affected the numbers of vocalizations and defecations.
Post-hoc comparisons revealed that the numbers of vocalizations were higher for females
compared to males, and COMM and RB females defecated more compared to COMM males.
COMM males took longer to ambulate compared to COMM and RB females, and COMM males
took significantly longer to vocalize compared to females of both strains and RB males.
Test 3. No sex or strain differences were found in the number of steps taken, but RB
females entered more squares compared to COMM males. The main effect of sex significantly
affected the number of vocalizations, but post-hoc comparisons only revealed a tendency for
vocalizations to differ between COMM males and females (t193 = 2.4, P = 0.08). Both sex and
strain significantly affected the number of defecations. COMM males defecated less compared to
females of both strains and RB males, and RB males defecated less compared to RB females.
COMM males took longer than all other turkeys to ambulate. COMM and RB females took less
time to vocalize compared to COMM and RB males, and COMM females took less time to
vocalize compared to RB females.
Corticosterone level. Post-test (Test 3) corticosterone levels were significantly higher
compared to pre-test levels for turkeys of either sex within both strains (Fig. 4.1). Pre-test
109

corticosterone levels tended to differ between female COMM and female RB turkeys (t72 = -2.4,
P = 0.08), and post-test corticosterone levels differed significantly between COMM and RB
females (Fig. 4.1). The change in corticosterone levels from pre- to post-test differed
significantly between COMM and RB females (Fig. 4.1), and tended to differ between RB males
and females (t72 = 0.4, P = 0.06).

Sex and strain differences in feather pecking
Descriptive statistics of feather pecking behavior are presented in Table 4.3.
Feather pecking behavior. The main effect of sex tended to affect the lsmean (± SE)
frequency of feather pecking (males: 0.56 ± 0.06 interactions per bird per day, females: 0.75 ±
0.06, F12 = 4.21, P = 0.06). The main effects of sex and strain significantly affected, and the
interaction between sex and strain tended to affect, the percentage of birds that were observed to
perform feather pecking. Post-hoc analyses revealed that a significantly higher percentage of RB
females were observed to perform feather pecking compared to RB males and COMM males and
females. Sex significantly affected the percentage of feather pecking interactions that were
classified as severe. However, post-hoc comparisons did not reveal differences in the percentage
of feather pecking interactions or the percentage of feather pecking interactions that were
classified as severe, between sexes within strains, or between males or females of different
strains.
Behavior of the recipient when feather pecking occurred. The percentage of feather
pecking interactions/bird occurring at a resource differed between sexes (males: 0.23 ± 0.19,
females: 1.56 ± 0.19, F12 = 26.1, P = 0.0003) and between males of different strains (Table 4.3).
110

The percentage of feather pecking interactions/bird occurring when the recipient was dustbathing
differed significantly between strains (COMM: 0.69 ± 0.22, RB: 0.73 ± 0.22, F12 = 17.8, P =
0.001). Post-hoc comparisons revealed that the percentage of interactions/bird occurring during
dustbathing tended to be higher for RB females compared to COMM females ( t12 = -2.61, P =
0.09), and COMM males (t12 = 2.90, P = 0.06), and was significantly higher for RB males
compared to COMM males, and COMM females. The percentage of feather pecking
interactions/bird that occurred when the recipient was preening did not differ between sexes,
strains or sexes within strains. The main effects of sex significantly affected, and the main effect
of strain tended to affect, the percentage of feather pecking interactions that occurred when the
recipient was engaged in “other” behavior. Post-hoc analyses indicated that COMM females
performed more feather pecking/bird when the recipient was engaged in “other” behavior
compared to RB females and RB males.

Feather scores
Feather scores are presented in Table 4.4. Only 5 turkeys had feather scores > 0 (0
indicated little or no feather damage) in the back region. Therefore, data from back feather scores
could not be compared between sexes and strains.
Tail feather scores tended to differ between sexes (F240 = 3.4, P = 0.06), and differed
significantly between strains. Tail feather scores were significantly higher (indicating worse
feather condition) for female COMM compared to female RB turkeys, and for male COMM
compared to female RB turkeys. Neck feather scores differed significantly between strains and
between sexes of different strains, but not between sexes of the same strain (Table 4.4). Neck
111

feather scores of male and female COMM turkeys were higher compared to male and female RB
turkeys.

Feather pecking and open field responses
Open field test responses of PECK and NPECK birds of males and females of both
strains are presented in Table 4.5. In Test 1, the latency to vocalize tended to differ between
COMM female PECK and NPECK birds (log-rank = 3.6, P = 0.06; Wilcoxon = 2.1, P = 0.1),
and between male RB PECK and NPECK birds (log-rank = 3.5, P = 0.06; Wilcoxon = 3.2, P =
0.07). No differences in OF Test 1 responses were found between male COMM or female RB
PECK and NPECK birds. There were no differences in any OF responses in Test 2 or Test 3
between male or female PECK and NPECK birds of either strain. Similarly, pre- and post-test
corticosterone levels did not differ between PECK and NPECK birds (Fig. 4.2).

DISCUSSION
Before any test can be used to examine differences in temperament or coping style, the
reliability (test-retest repeatability; assessed in Erasmus and Swanson, 2014) and validity (intertest correspondence, assessed in Chapter 3) of the test need to be ascertained. Previously, we
reported that responses of male COMM turkeys were repeatable between 4 and 10 wk of age
(Erasmus and Swanson, 2014), and that although fear responses appear to be context dependent,
there may be an association between OF and tonic immobility behavior at 8 to 10 wk of age
(Chapter 3).
Consistent with our previous results (Erasmus and Swanson, 2014), the majority of
behavioral OF responses of COMM male turkeys were repeatable between 4 and 11 wk. The

112

majority of responses of COMM females, and RB males and females also had moderate
correlation coefficients between 4 and 11 wk (males) and 5 and 12 wk (females). However,
repeatability of OF behavior was low between 1 and 4 (males) or 5 (females) wk, indicating that
OF behavior changes between 1 and 5 wk, and with repeated testing. There does not appear to be
any published research examining changes in turkeys’ fear responses with age, but early research
with chickens suggests that behavior associated with predator avoidance increases as birds age
(Suarez and Gallup, 1983). Therefore, OF and other fear-related behavior may be more
consistent at later ages due to the development of this behavior over time.
Interestingly, in our previous study with COMM males the latency to ambulate had the
highest correlation coefficient (rS = 0.66) and the latency to vocalize had the lowest correlation
coefficient (rS = 0.17; Erasmus and Swanson, 2014). However, results reported here revealed
that the latency to vocalize (rS = 0.47) had a higher correlation coefficient than the latency to
ambulate (rS = 0.35). The difference in test-retest repeatability may be due to the ages at which
birds were tested (and body weight differences), or differences in environmental conditions.
Turkeys in the first study (Erasmus and Swanson, 2014) were reared in smaller groups (4-6 birds
per pen) and under a much lower light intensity (5 lx) compared to turkeys in this study, where
both the group size was larger (14-16 birds per pen) and the light intensity was higher (20 lx).
Furthermore, turkeys in the present study were tested later and males had higher body weights
(6.2 ± 0.08 kg (Erasmus and Swanson, 2014) vs. 8.5 ± 0.13 kg), which may have reduced their
physical ability to move around the OF testing arena. In contrast, COMM females and RB
turkeys had much lower body weights compared to COMM males, and therefore their
ambulatory ability may not have been affected.
113

Open field behavior may not only vary due to environmental and management factors,
but sex and genetic strain may also be important factors affecting behavior. Due to genetic
selection for increased feed efficiency and increased production, the domestic turkey has
changed dramatically in the past 30 years. Therefore, we wanted to examine whether behavioral
and physiological responses in the OF differed between COMM and RB turkeys, and between
males and females of the two strains. Differences were found for some responses at each age of
testing, but differences were not the same across test periods. There were fewer strain differences
than there were sex differences. In general, COMM males defecated less and ambulated and
vocalized later than COMM females when tested at 4 and 5 wk, and 11 and 12 wk, respectively.
Similarly, RB males defecated less and vocalized later than RB females when tested at 11 and 12
wk, respectively. Strain differences were limited to differences in vocalization latency (males:
Test 2, females: Test 1 and Test 3), ambulation latency (males: Test 3, females: Test 1), and
number of defecations (males: Test 3). Results reveal that sex and strain differences in OF
behavior appear to depend on the age at which birds are tested, and may be affected by repeated
testing. Sex differences may have been confounded by the age at testing because females were
tested one week later in Tests 2 and 3 compared to males. Males and females were tested one
week apart because it was not possible to test all birds in the same week. Furthermore, blood
samples were collected before and after testing and in order to prevent reducing the sample size
for blood samples, we elected to test all birds from the same sex in the same week rather than
confound corticosterone levels and age. Further research is needed to examine age-related effects
on fear responses and how male and female turkeys differ in the rate of development of fear
responses.

114

All turkeys experienced significantly elevated post-test corticosterone levels compared to
pre-test levels, indicating activation of the hypothalamic-pituitary-adrenal (HPA) axis. The HPA
axis is activated when animals experience fear, and corticosterone levels are higher in birds when
they are exposed to a threatening situation (Cockrem, 2007). Research with chickens and quail
have provided evidence that corticosterone levels are higher in birds that show behavioral
responses consistent with increased fear in tonic immobility and OF tests (reviewed in Cockrem,
2007). Unlike in the studies with chickens and quail, we did not observe a clear relationship
between behavioral and physiological responses of turkeys to OF testing. Corticosterone levels
following testing were higher in RB females compared to COMM females, but behavioral
responses were not significantly different. If corticosterone levels were associated with fear, and
if lower activity levels are indicative of greater levels fear, birds with lower activity levels may
be expected to have higher corticosterone levels following OF testing. In fact, the opposite
occurred: the number of steps taken and squares entered were numerically, although not
statistically, higher for RB vs. COMM females, which may suggest lower fear in RB females.
Alternatively, corticosterone levels may reflect differences in activity levels and not fear per se.
Furthermore, inactivity in the OF may not necessarily be indicative of a high level of fear
because some birds may be inactive without being fearful. Another possible explanation for the
discrepancy between OF behavior and corticosterone level is that the physiological stress
responses that were observed were due to capture and handling by observers, rather than
conditions associated with the OF test environment. Capture and handling by human observers
may be more threatening than the OF test itself, thereby masking the physiological response to
OF testing. In agreement with this explanation, Hazard et al. (2008) reported that quail selected
for high and low durations of tonic immobility experienced greater increases in corticosterone

115

levels at the beginning of the tonic immobility test when they were being restrained by an
observer, but no relationship was found between tonic immobility duration (measure of fear) and
corticosterone level. Another explanation is that behavioral differences in fear behavior are not
consistently reflected by differences in HPA axis activity (Armario et al., 2012). In their review
of emotional stressors and hormones, Armario et al. (2012) state that “the relationship between
HPA activation and anxiety is far from being adequately characterized”. Further research
examining the relationship between OF behavior and corticosterone is needed to understand
whether corticosterone levels are related to differences in activity levels in the OF.
Our second objective was to examine whether feather pecking behavior differed among
sexes and strains. To date, there has been very little research examining feather pecking among
turkeys, and ours is the first study to directly examine feather pecking among turkeys at the
individual level. As many as 42% of COMM males (39% RB males), and 57% of COMM
females (63% RB females) were observed to perform feather pecking across a 3-day period. The
percentages of birds that performed feather pecking was higher overall for females, and was
highest for RB females on a per bird basis. In contrast, plumage scores in the neck region were
higher for COMM turkeys, and tail scores were higher for COMM vs. RB females, indicating
that pecking damage was greater in COMM birds.
Plumage damage was scored at 13 wk (91 d), whereas observations of feather pecking
were conducted earlier, at 75 to 83 d, which may explain why behavioral observations were not
consistent with plumage scores. Another possible explanation is that some feather pecking
interactions from the video may have been scored as gentle because the recipient bird did not
react, and classification of the feather pecking interaction depends, in part, on the reaction of the
recipient. Furthermore, Busayi et al. (2006) postulated that turkeys of different strains may differ

116

behaviorally in their responses to feather pecking such that turkeys of a particular strain move
away faster, thereby reducing pecking damage, or birds of different strains may differ in feather
density, and therefore in susceptibility to pecking damage. Further research is necessary to
examine behavior of the recipient, and whether turkeys of different strains react differently to
feather pecking.
Feather pecking behavior may have been influenced by the colors used to individually
identify birds in the pen. Using colored livestock marker to distinguish between individuals was
necessary so that markings were visible from overhead cameras. Marker color, color
combinations and location were randomly assigned to birds in each pen. Color was therefore
confounded with the location on the turkey’s body where the marker was applied. Because each
turkey in the pen was marked differently, and because feather pecking behavior may differ
among pens, it was not possible to assess the effects of color on pecking behavior. Further
research is needed to determine whether color markings of turkeys affect feather pecking
behavior by including appropriate control groups that are not marked.
In addition to examining the frequency of feather pecking, we examined when feather
pecking occurred in relation to the behavior of the recipient. Results revealed that feather
pecking occurred when the recipient birds were performing a wide range of behaviors. However,
the majority (> 40%) of feather pecking interactions occurred when the recipient was engaged in
“other” behavior, which included sitting, standing, walking and environmental pecking.
Interestingly, there were strain differences in when feather pecking occurred, with RB males
pecking more when the recipient had been dustbathing, compared to COMM males, and a similar
tendency between RB and COMM females. None of the COMM male recipients of feather
pecking were observed to dustbathe, indicating that some behavior of turkeys may have changed

117

as a result of genetic selection for increased feed efficiency and faster growth. However, further
research is needed to verify whether dustbathing behavior is affected by selection for faster
growth rates and increased feed efficiency because this was not examined in our study.
The considerable amount of scientific literature concerning feather pecking of laying
hens provides evidence of a relationship between fear responses and the propensity to develop
feather pecking (reviewed in Rodenburg et al., 2013). Most recent studies have been conducted
with laying hens that have been divergently selected for high or low feather pecking behavior,
respectively, but an earlier study with red jungle fowl also found evidence that feather pecking
and fear responses differ between birds that develop feather pecking and birds that do not.
Vestergaard et al. (1993) found that the rate of pecks delivered was correlated with the duration
of tonic immobility in red jungle fowl. In another study with the F2 generation of a cross
between the high and low feather pecking lines, OF behavior at a young age (5 wk) was
genetically correlated with pecking behavior in adulthood (Rodenburg et al., 2004). The only
association between OF responses and feather pecking that were found here was a tendency for
the latency to vocalize to be shorter for COMM female and RB male turkeys that were not
observed to perform feather pecking compared to turkeys that developed feather pecking, when
tested at 1 wk. Furthermore, there were no differences in corticosterone levels between PECK
and NPECK turkeys. As discussed above, the corticosterone levels may not have been reflective
of fear in the OF test, but it is possible that the OF test assesses other behavioral responses or
motivations, such as social reinstatement motivation, and not just fear, as discussed in Chapter 3.
Research examining the relationships between feather pecking and responses in other tests of
fear, such as tonic immobility, may provide further insight into whether turkeys that differ in
fearfulness differ in feather pecking.

118

In conclusion, results from this study support our previous work that turkeys’ responses
in the OF test are repeatable from 4 to 12 wk of age. Furthermore, there were some differences in
OF behavior between COMM males and females at 4-5 and 11-12 wk of age, and between RB
males and females at 1 and 11-12 wk of age, indicating that age, sex and strain differences
should be taken into account when OF test results are interpreted. Our results did not support the
hypothesis that feather pecking is associated with differences in temperament measured using the
OF test. Additionally, our results provide further evidence of the complex, multifactorial nature
of feather pecking among domestic turkeys, which remains poorly understood. Several areas of
future research will provide insight into the physiology and behavior turkeys, including research
examining the relationship between corticosterone levels and behavioral responses to various
other fear tests, and research examining the behavior of the recipient bird during a feather
pecking interaction to ascertain whether there are strain differences in how recipients react.

ACKNOWLEDGEMENTS
We thank Dr. Darrin Karcher, and Hybrid Turkeys and Cuddy farms for supplying the
poults, and Angelo Napolitano and the staff at the Michigan State University Poultry Research
and Teaching Center for their assistance. We would also like to express our gratitude to Janet
Ireland for conducting corticosterone assays, and to the student members of the Michigan State
University Animal Behavior and Welfare Group for the invaluable assistance with data
collection. M.E. was supported by a Natural Sciences and Engineering Research Council of
Canada postgraduate scholarship and this work was supported in part by Michigan State
University AgBioResearch.

119

APPENDIX

120

Table 4.1. Spearman rank correlation (rS) of open field (OF) behavior among test periods for
male and female commercial (COMM) and randombred (RB) turkeys.
Test 1 vs. 2

Lat. ambulate
Lat. vocalize
Step no.
Square no.

N = 31
0.12
0.22
0.30

Vocalization no.
Defecation no.

0.31T
0.26
0.08

Lat. ambulate
Lat. vocalize
Step no.
Square no.
Vocalization no.
Defecation no.

N = 34
0.19
0.02
0.24
0.23
0.16
T
0.33

Lat. ambulate
Lat. vocalize
Step no.
Square no.
Vocalization no.
Defecation no.

Lat. ambulate
Lat. vocalize
Step no.
Square no.
Vocalization no.
Defecation no.

N = 32
0.38*
0.04
T
0.32
0.43*
T
0.32
0.23

Test 1 vs. 3
COMM males
N = 31
0.40*
T
0.35
0.40*
T
0.35
0.24
0.02
RB males
N = 34
0.01
0.02
0.10
-0.05
0.05
0.17
COMM females
N = 32
0.46*
T
0.33
0.27
0.37*
0.23
0.40*
RB females
N = 29
0.41*

N = 29
0.59***
T
0.34
0.39*
0.37*
0.37*
0.52**

0.52**
0.55**
0.57**
0.49**
0.41*

T

Test 2 vs. 3
N = 47
0.35*
0.47***
0.41**
0.31*
0.69****
0.58****
N = 47
0.42**
0.37*
0.46**
0.51***
0.34*
0.33*
N = 47
0.41**
0.47***
0.54****
0.55****
0.64****
0.56****
N = 47
0.60****
0.43**
0.73****
0.71****
0.54****
0.42**

Note: Lat. = latency. P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

121

Table 4.2. Behavioral responses of commercial (COMM) and randombred (RB) male and female turkeys to open field testing at 3 ages
- Test 1: wk 1; Test 2: wk 4 (males) and wk 5 (females); Test 3: wk 11 (males) and wk 12 (females).
Step no.

Square no.

Voc. no.

Def. no.

Lat. amb.

ab

4.8 ± 0.4

0.05 ± 0.2

a

502 (236-600)

1.5 ± 0.5

4.7 ± 0.4

-0.2 ± 0.2

4.0 ± 0.4

ab

600 (214-600)

2.6 ±0.5

5.2 ± 0.4

0.07 ± 0.2

RB females

5.2 ± 0.4

b

600 (223-600)

3.6 ± 0.5

6.0 ± 0.4

0.2 ± 0.2

216 (47-600)

Sex

F156 = 1.8

F156 = 3.3

F156 = 5.2*

F156 = 1.3

F156 = 0.02

F156 = 0.3

F156 = 0.7

F156 = 0.08

F156 = 6.9**

F156 = 7.4**

F156 = 1.5

F156 = 1.0

Test 2
COMM males

3. 9 ± 0.2

2.6 ± 0.2

5.5 ± 0.2

0.8 ± 0.1

321 (80-600)

RB males

4.3 ± 0.2

2.9 ± 0.2

5.9 ± 0.2

1.0 ± 0.1

143 (55-600)

COMM females

4.3 ± 0.2

2.8 ± 0.2

6.2 ± 0.2

1.3 ± 0.1

112 (56-464)

RB females

4.7 ± 0.2

3.1 ± 0.2

6.3 ± 0.2

1.3 ± 0.1

76 (34-195)

Sex

F221 = 2.8

F221 = 0.9

F221 = 9.7** F221 = 15.9****

F221 = 1.7

F221 = 2.3

F221 = 0.7

F221 = 0.01

F221 = 1.1

F221 = 1.1

Test 1
COMM males

4.5 ± 0.4

3.0 ±0.4

RB males

3.5 ± 0.4

COMM females

Strain
Sex x strain

Strain
Sex x strain

ab
a
ab
b

F221 = 3.8
F221 = 0.0

T
T

T

a
ab
b
b

a
ab
b
b

Lat. voc.

b

a
ab
a

b

a
ab
b

246 (45-600)
522 (72-600)

a
ab

a

414 (5-600)
b

7 (2-520)

a

69 (4-216)
b

7 (2-178)
b

4 (1-28)

b

3 (1-49)

Note: Lat. = latency, amb. = ambulate, voc. = vocalize, L = Log-rank, W = Wilcoxon. Numbers of steps and squares entered,
vocalizations and defecations are presented as lsmean ± SE. Latencies to ambulate and vocalize are presented as median (25-75%
a-c
quartiles). For each test, values within columns lacking common superscripts differ significantly (P < 0.05).
T

P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

122

Table 4.2. (cont’d).
Step no.

Square no.

Voc. no.

Def. no.

Test 3
COMM males

3.0 ± 0.3

RB males

1.0 ± 0.3

a

4.6 ± 0.3

600 (163-600)

3.5 ± 0.3

ab

0.3 ± 0.1

1.8 ± 0.3

4.7 ± 0.3

337 (83-600)

COMM females

3.4 ± 0.3

ab

0.6 ± 0.1

1.4 ± 0.3

5.5 ± 0.3

268 (132-600)

RB females

3.9 ± 0.3

b

0.8 ± 0.1

2.1 ± 0.3

5.4 ± 0.3

1.0 ± 0.1

226 (78-550)

Sex

F193 = 2.0

F193 = 1.8

F193 = 8.8** F193 = 17.4****

F193 = 6.1*

F193 = 0.0

F193 = 4.3*

F193 = 0.02

F193 = 0.2

F193 = 0.04

Strain
Sex x strain

F193 = 3.3
F193 = 0.0

T

Lat. amb.
a
c
bc
b

Lat. voc.
a

b
b

b

102 (15-600)
141 (16-600)

a
a

b

26 (7-117)

c

84 (14-158)

Note: Lat. = latency, amb. = ambulate, voc. = vocalize, L = Log-rank, W = Wilcoxon. Numbers of steps and squares entered,
vocalizations and defecations are presented as lsmean ± SE. Latencies to ambulate and vocalize are presented as median (25-75%
a-c
quartiles). For each test, values within columns lacking common superscripts differ significantly (P < 0.05).
T

P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

123

Table 4.3. Sex and strain differences in feather pecking (FP) behavior (n = 4 pens of males and females per strain). Behavior was
recorded over 3 days. Raw data (mean ± SE) are presented.
B

FP at
B
resource (%)

Birds
performing
B
FP (%)

Severe FP
(%)

COMM males

Frequency
of FP
A
interactions
0.9 ± 0.1

42.2 ± 3.9

a

0.8 ± 0.1

RB males

0.7 ± 0.06

a

0.4 ± 0.3

39.1 ± 6.6

1.1 ± 0.2

COMM females

1.0 ± 0.2

57.8 ± 10.0

a

RB females

1.3 ± 0.4

b

Sex

F12 = 4.2

Strain
Sex x strain

FP while
B
other (%)

0.0 ± 0.0

a

FP while
B
preening
(%)
1.1 ± 0.06

0.04 ± 0.04

1.5 ± 0.4

b

0.8 ± 0.1

1.5 ± 0.3

1.3 ± 0.4

a

4.2 ± 0.2

0.1 ± 0.08

1.1 ± 0.3

63.3 ± 10.0

1.6 ± 0.2

1.8 ± 0.2

ab

3.7 ± 0.1

1.3 ± 0.4

0.8 ± 0.1

2.8 ± 0.5

F12 = 4.8*

F12 = 7.4*

F12 = 26.1***

F12 = 0.0

F12 = 0.04

F12 = 16.0**

F12 = 0.0

F12 = 12.1**

F12 = 1.1

F12 = 0.1

F12 = 17.8**

F12 = 3.1

F12 = 4.7

F12 = 1.9

F12 = 4.2

F12 = 0.1

F12 = 2.6

F12 = 0.3

F12 = 0.01

F12 = 0.6

T

T

ab
a

bc
c

A

FP while
B
dustbathing
(%)

a

4.7 ± 0.3

a
ab
b
T

The frequency of feather pecking interactions was calculated as the mean across 4 pens of the total number of feather pecking
interactions per individual per day in each pen.
B
The percentage was calculated as the mean across 4 pens of the total percentage of feather pecking interactions occurring over 3 d
for each behavior of the recipient bird / the number of birds in the pen.
a, b
Values within rows lacking common superscripts differ significantly (P < 0.05).
T

P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001.

124

Table 4.4. Feather scores (lsmean ± SE) from the neck, tail, and right and left flight feathers of
male and female commercial (COMM) and randombred (RB) turkeys.
Neck

Right flight
0.53 ± 0.10

Left flight
0.58 ± 0.09

0.74 ± 0.09

-1.46 ± 0.34

0.71 ± 0.09

0.65 ± 0.09

0.53 ± 0.10

COMM females

0.30 ± 0.23

0.66 ± 0.09

0.63 ± 0.09

0.66 ± 0.09

RB females

-0.75 ± 0.28

0.64 ± 0.09

0.63 ± 0.09

0.24 ± 0.12

Sex

F240 = 2.1

F240 = 0.1

F240 = 0.03

F240 = 3.4

F240 = 25.0****

F240 = 0.7

F240 = 0.13

F240 =10.2**

F240 = 1.3

F240 = 1.2

F240 = 0.09

F240 =1.1

0.23 ± 0.23

RB males

Strain
Sex x strain
a, b
T

a

COMM males

b

a
b

Tail

ab
a
b

Values within rows lacking common superscripts differ significantly (P < 0.05).

P < 0.10, *P < 0.05, **P < 0.01.

125

a

T

Table 4.5. Behavioral open field test responses of male and female turkeys that were observed to perform feather pecking (PECK) and
that did not perform feather pecking (NPECK) of a commercial (COMM) and randombred (RB) strain. Open field testing was
conducted at 3 ages - (Test 1: wk 1; Test 2: wk 4 (males) and wk 5 (females); Test 3: wk 11 (males) and wk 12 (females).

Test 1
COMM males
RB males
COMM females
RB females
Test 2
COMM males
RB males
COMM females
RB females

Step no.

Square no.

Voc. no.

Def. no.

Lat. amb.

Lat. voc.

PECK (n = 14)
NPECK (n = 26)
PECK (n = 15)

4.4 ± 0.7
4.6 ± 0.5
2.8 ± 0.7

2.9 ± 0.7
3.1 ± 0.5
0.8 ± 0.7

4.8 ± 0.6
4.9 ± 0.5
3.7 ± 0.6

0.1 ± 0.3
0.1 ± 0.2
-0.4 ± 0.4

504 (336-600)
494 (219-600)
600 (381-600)

365 (12-600)
233 (58-600)
a
600 (324-600)

NPECK (n = 24)

3.7 ± 0.5

1.8 ± 0.6

5.0 ± 0.5

-0.1 ± 0.3

600 (144-600)

183 (58-600)

PECK (n = 24)

3.5 ± 0.5

2.0 ± 0.6

4.8 ± 0.5

0.0 ± 0.2

600 (131-600)

600 (4-600)

NPECK (n = 14)

4.6 ± 0.7

3.2 ± 0.7

5.8 ± 0.6

0.3 ± 0.3

443 (129-600)

PECK (n = 24)
NPECK (n = 14)

5.1 ± 0.6
5.4 ± 0.7

3.5 ± 0.6
3.9 ± 0.8

6.1 ± 0.5
6.1 ± 0.7

0.3 ± 0.3
0.1 ± 0.3

173 (39-600)
199 (58-600)

44 (4-450)
6 (2-600)
7 (2-159)

PECK (n = 23)
NPECK (n = 33)
PECK (n = 16)
NPECK (n = 38)
PECK (n = 35)
NPECK (n = 22)
PECK (n = 34)
NPECK (n = 20)
NPECK (n = 31)

4.0 ± 0.4
3.8 ± 0.5
4.4 ± 0.4
4.2 ± 0.3
4.4 ± 0.3
4.4 ± 0.4
4.6 ± 0.4
4.4 ± 0.5
4.0 ± 0.5

2.5 ± 0.4
2.5 ± 0.5
3.0 ± 0.4
2.7 ± 0.3
2.9 ± 0.3
2.8 ± 0.4
3.1 ± 0.4
2.8 ± 0.5
2.4 ± 0.5

5.4 ± 0.3
5.5 ± 0.3
6.0 ± 0.3
5.9 ± 0.2
6.1 ± 0.2
6.2 ± 0.3
6.2 ± 0.3
6.4 ± 0.3
4.7 ± 0.5

0.5 ± 0.2
0.8 ± 0.2
1.1 ± 0.2
1.0 ± 0.1
1.2 ± 0.1
1.2 ± 0.2
1.4 ± 0.2
1.3 ± 0.2
0.5 ± 0.2

238 (46-600)
321 (151-600)
136 (51-540)
158 (62-600)
112 (41-343)
127 (57-464)
75 (28-498)
94 (47-168)
248 (43-600)

69 (4-256)
63 (4-213)
9 (2-259)
9 (2-174)
4 (1-28)
3 (1-40)
3 (1-95)
5 (2-7)
137 (7-600)

b

a

b

Note: Lat. = latency, amb. = ambulate, voc. = vocalize, L = Log-rank, W = Wilcoxon. Numbers of steps and squares entered,
vocalizations and defecations are presented as lsmean ± SE. Latencies to ambulate and vocalize are presented as median (25-75%
a-c
quartiles). For each test, values within columns lacking common superscripts tended to differ significantly (P < 0.10).
T

P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

126

Table 4.5. (cont’d).

Test 3
COMM males
RB males
COMM females
RB females
Sex
Strain
Classification
Sex x strain x
classification

PECK (n = 20)
NPECK (n = 30)
PECK (n = 19)
PECK (n = 28)
NPECK (n = 19)
PECK (n = 29)
NPECK (n = 21)

Step no.

Square no.

Voc. no.

Def. no.

Lat. amb.

Lat. voc.

1.8 ± 0.9
1.6 ± 1.0
4.0 ± 0.6
3.2 ± 0.4
2.6 ± 0.6
4.9 ± 0.8
4.7 ± 0.8
F211 = 1.6

-0.6 ± 0.9
-0.5 ± 0.9
2.3 ± 0.6
1.3 ± 0.4
0.3 ± 0.6
3.3 ± 0.7
3.0 ± 0.8
F211 = 0.7

4.1 ± 0.9
4.4 ± 0.9
4.9 ± 0.6
5.5 ± 0.4
5.3 ± 0.5
5.6 ± 0.8
5.6 ± 0.8
F211 = 8.1**

0.4 ± 0.3
0.2 ± 0.3
0.6 ± 0.2
0.8 ± 0.1
0.6 ± 0.2
1.0 ± 0.3
1.1 ± 0.3
F211 = 15.1***

600 (113-600)
590 (89-600)
166 (57-600)
193 (131-600)
268 (131-600)
214 (72-600)
253 (131-544)

82 (12-600)
91 (5-600)
63 (15-600)
16 (5-87)
26 (8-117)
112 (14-162)
82 (49-131)

F211 = 0.3

F211 = 0.2

F211 = 0.7

F211 = 1.5

F211 = 0.4

F211 = 0.3

F211 = 0.1

F211 = 0.3

F211 = 0.06 F211 = 0.1

F211 = 0.3

F211 = 1.0

Note: Lat. = latency, amb. = ambulate, voc. = vocalize, L = Log-rank, W = Wilcoxon. Numbers of steps and squares entered,
vocalizations and defecations are presented as lsmean ± SE. Latencies to ambulate and vocalize are presented as median (25-75%
a-c
quartiles). For each test, values within columns lacking common superscripts tended to differ significantly (P < 0.10).
T

P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

127

a

a

Figure. 4.1. Plasma corticosterone levels (pre-test , post-test and change from pre- to post-test)
of male and female turkeys of a commercial (COMM) and randombred (RB) strain tested in an
open field test at 11 (males) and 12 (females) weeks of age. Raw data (mean ± SE) are presented.
Statistical analyses were conducted on log-transformed data.

*

20

b

Corticosterone level (ng/ml)

18
16

*

b b

14
12

b

10
8
6
4
2

a

a

a
a

0
Pre-test

Post-test

a

Change

Pre-test blood samples were collected one day prior to testing at the same time that post-test
samples were to be collected the following day. Post-test samples were collected 15 min
following the start of the test.
Different letters (a, b) indicate significant differences between pre- and post-test corticosterone
levels of males and females within strains (P < 0.05).
*Asterisks indicate significant differences between sexes of different strains (P < 0.05).

128

a

a

Figure. 4.2. Plasma corticosterone levels (pre-test , post-test and change from pre- to post-test) of male and female turkeys of a
commercial (COMM) and randombred (RB) strain that were observed to perform feather pecking (PECK) and that did not perform
feather pecking (NPECK). Turkeys were tested in an open field test at 11 (males) and 12 (females) wk of age.
Raw data (mean ± SE) are presented. Statistical analyses were conducted on log-transformed data. No significant differences were
found.
Pre-test

Corticosterone level (ng/ml)

25

Post-test

Change

20

15

10

5

0

COMM male
PECK
NPECK

RB male
PECK
NPECK

a

COMM female
PECK
NPECK

RB female
PECK
NPECK

Pre-test blood samples were collected one day prior to testing at the same time that post-test samples were to be collected the
following day. Post-test samples were collected 15 min following the start of the test.
129

REFERENCES

130

REFERENCES
Armario, A., N. Daviu, C. Muñoz-Abellán, C. Rabasa, S. Fuentes, X. Belda, H. Gagliano, and R.
Nadal. 2012. What can we know from pituitary–adrenal hormones about the nature and
consequences of exposure to emotional stressors? Cell. Mol. Neurobiol. 32: 749–758.
Bilčík, B., and L. J. Keeling. 1999. Changes in feather condition in relation to feather pecking
and aggressive behaviour in laying hens. Br. Poult. Sci. 40: 444-451.
Blokhuis, H. J., L. J. Keeling, A. Gavinelli, and J. Serratosa. 2008. Animal welfare’s impact on
the food chain. Trends Food Sci. Tech. 19: S79-S87.
Busayi, R. M., C. E. Channing, and P. M. Hocking. 2006. Comparisons of damaging feather
pecking and time budgets in male and female turkeys of a traditional breed and a
genetically selected male line. Appl. Anim. Behav. Sci. 96: 281-292.
Cockrem, J. F. 2007. Stress, corticosterone responses and avian personalities. J Ornithol. 148:
S169–S178.
Dalton, H. A., B. J. Wood, and S. Torrey. 2013. Injurious pecking in domestic turkeys:
development, causes and potential solutions. World’s Poult. Sci. J. 69: 865-875.
de Haas, E. N., J. E. Bolhuis, B. Kemp, T. G. G. Groothuis, and T. B. Rodenburg. 2014. Parents
and early life environment affect behavioral development of laying hen chickens. PLoS
ONE. 9: 1-12.
de Haas, E. N., B. L. Nielsen, A. J. Buitenhuis, and T. B. Rodenburg. 2010. Selection on feather
pecking affects response to novelty and foraging behaviour in laying hens. Appl. Anim.
Behav. Sci. 124:90-96.
Duggan, G., T. Widowski, M. Quinton, and S. Torrey. The development of injurious pecking in a
COMM turkey facility. 2014. J. Appl. Poult. Res. 23: 1-11.
Erasmus, M. A. 2009. Examining physical methods for on-farm killing of turkeys. MS Thesis,
University of Guelph, Guelph, ON, Canada.
Erasmus, M. A., and J. C. Swanson. 2014. Temperamental turkeys: reliability of behavioral
responses to four tests of fear. Appl. Anim. Behav. Sci. 157: 100-108.
Federation of Animal Science Societies, 2012. Farm Animal Integrated Research. Available
Online: http://www.fass.org/docs/FAIR2012_Summary.pdf
Forkman, B., A. Boissy, M. –C. Maunier-Salaün, E.Canali, and R. B. Jones. 2007. A critical
review of fear tests used on cattle, pigs, sheep, poultry and horses. Physiol. Behav. 92:
340-374.
131

Hocking, P. 1993. Welfare of turkeys. Page 125-138 in Proceedings of the Fourth European
Symposium on Poultry Welfare. Savory C J and B. O. Hughes, eds. Universities
Federation for Animal Welfare, Potters Bar, UK.
Hazard, D., S. Leclaire, M. Couty, and D. Guémené. 2008. Genetic differences in coping
strategies in response to prolonged and repeated restraint in Japanese quail divergently
selected for long or short tonic immobility. Horm. Behav. 54: 645-653.
Huber-Eicher, B., and F. Sebö. 2001. The prevalence of feather pecking and development in
COMM flocks of laying hens. Appl. Anim. Behav. Sci. 74: 223-231.
Huff, G., W. Huff, N. Rath, A. Donoghue, N. Anthony, and K. Nestor. 2007. Differential effects
of sex and genetics on behavior and stress response of turkeys. Poult. Sci. 86:1294-1303.
Jensen, P., L. Keeling, K. Schutz, L. Andersson, P. Mormede, H. Brandstrom, B. Forkman, S.
Kerje, R. Fredriksson, C. Ohlsson, S. Larsson, H. Mallmin, and A. Kindmark. 2005.
Feather pecking in chickens is genetically related to behavioral and developmental traits.
Physiol. & Behav. 86:52-60.
Jones, R. B., H. J. Blokhuis, and G. Beuving. 1995. Open-field and tonic immobility responses in
domestic chicks of 2 genetic lines differing in their propensity to feather peck. Br. Poult.
Sci. 36:525-530.
Jones, B., and A. Boissy. 2011. Fear and other negative emotions. Page 81 in Animal Welfare.
2nd rev. ed. Appleby, M. C, J. A. Mench, I. A. S. Olsson, and B. O. Hughes, eds. CABI
International, Cambridge, MA., USA.
Koolhaas, J. M., S. M. Korte, S. F. De Boer, B. J. Van Der Vegt, C. G. Van Reenen, H. Hopster,
I. C. De Jong, M. A. W. Ruis, and H. J. Blokhuis. 1999. Coping styles in animals: current
status in behavior and stress-physiology. Neurosci. Biobehav. Rev. 23: 925–935.
Korte, S. M., G. Beuving, W. Ruesink, and H. J. Blokhuis. 1997. Plasma catecholamine and
corticosterone levels during manual restraint in chicks from a high and low feather
pecking line of laying hens. Physiol. & Behav. 62: 437-441.
Martin, P., and P. Bateson. 1993. The reliability and validity of measures. Pages 114-116 in
Measuring behaviour: an introductory guide. 2nd rev. ed. Cambridge University Press,
New York, NY., USA.
Nestor, K. E., M. G. McCartney, and N. Bachev. 1969. Relative contributions of genetics and
environment to turkey improvement. Poult. Sci. 48:1944-1949.
Rodenburg, T. B., A. J. Buitenhuis, B. Ask, K. A. Uitdehaag, P. Koene, J. J van der Poel, J. A.
M. van Arendonk, and H. Bovenhuis. 2004. Genetic and phenotypic correlations between
feather pecking and open-field response in laying hens at two different ages. Behav.
Genet. 34: 407-415.
132

Rodenburg, T. B., Komen, H., Ellen, E. D., Uitdehaag, K. A., and J. A. M. van Arendonk. 2008.
Selection method and early life-history affect behavioral development, feather pecking
and cannibalism in laying hens: a review. Appl. Anim. Behav. Sci. 110: 217-228.
Rodenburg, T. B, M. M. van Krimpen, I. C. De Jong, E. N. de Haas, M. S. Kops, B. J. Riedstra,
R. E. Nordquist, J. P. Wagenaar, M. Bestman and C. J. Nicol. 2013. The prevention and
control of feather pecking in laying hens: identifying the underlying principles. World’s
Poult. Sci. J. 69: 361-374.
Suarez, S. S., and G. G. Gallup. 1983. Social reinstatement and open-field testing in chickens.
Anim. Learn. Behav. 11: 119-126.
van Hierden, Y. M., S. M. Korte, W. Ruesink, C. G. van Reenen, B. Engel, J. M. Koolhaas, and
H. J. Blokhuis. 2002. The development of feather pecking behavior and targeting of
pecking in chicks from a high and low feather pecking line of laying hens. Appl. Anim.
Behav. Sci. 77: 183-196.
Vestergaard, K., J. P. Kruijt, and J. A. Hogan. 1993. Feather pecking and chronic fear in groups
of red junglefowl: their relations to dustbathing, rearing environment and social status.
Anim. Behav. 45: 1127-1140.

133

CHAPTER 5: INDIVIDUAL AND STRAIN DIFFERENCES IN OPEN FIELD
BEHAVIOR AND ITS RELATIONSHIP WITH MEAT QUALITY OF TURKEYS
ABSTRACT
The occurrence of pale, soft, exudative (PSE) turkey meat has become increasingly
problematic, necessitating research examining factors that influence meat quality. Differences in
meat quality are known to exist among pigs and cattle differing in temperament characteristics
such as fearfulness. However, associations between temperament and meat quality of turkeys
have not been examined. The objectives of this research were to 1) examine differences in meat
quality characteristics (pH, R-value, and L*, a* and b* color values) between a commercial
(COMM) and randombred (RB) turkey strain, and 2) examine the relationship between meat
quality and behavioral and physiological responses in an open field (OF) test, which is used to
assess fear in poultry. Male COMM and RB turkeys were each housed in groups of 16 in 4 pens.
Turkeys were individually tested in an OF (2.74 x 2.74 m, divided into 81 squares) at 11 weeks
and birds were grouped into two clusters based on cluster analysis of OF behavior. Plasma
corticosterone levels were assessed in subsamples of birds (n = 20/strain) before and after OF
testing. Turkeys were processed and meat quality characteristics were evaluated in 10 to 11 birds
per pen (n = 44/strain) at 15 to 17 weeks (COMM) and 20 to 21 weeks (RB). Results were
analyzed using a mixed model (SAS 9.3). The R-value tended to be lower (COMM: 1.18 ± 0.02,
RB: 1.24 ± 0.02, P = 0.07) for the COMM vs. the RB strain, but there were no differences in pH,
or L*, a* or b* color values. Corticosterone levels did not differ between clusters of either strain,
and no differences in any meat quality characteristics were found between clusters. Results
indicate that plasma corticosterone responses of COMM and RB turkeys to OF testing are
similar, and there is little difference in meat quality characteristics between RB and COMM male

134

turkeys. Within strains, individual differences in behavioral OF responses are not associated with
differences in meat quality.

INTRODUCTION
Annual turkey production in the United States has increased from 90 million in 1962 to
over 250 million in 2012 (USDA, 2013). Increased turkey production was facilitated by changes
in production, such as intensive housing, and selective breeding methods that reduced costs
associated with raising turkeys (Martin, 2009). Concurrent with increased production, problems
with meat quality have also increased (Petracci and Cavani, 2012). For example, the incidence of
pale, soft, exudative (PSE) turkey meat which is associated with intensive selection for increased
muscle growth (Strasburg and Chiang, 2009) has become increasingly problematic “due to the
demand for processed, value-added poultry meat” (Sosnicki et al., 1998).
Meat quality, including PSE, may be affected by genetic strain, but results are
contradictory. Research by Werner et al. (2008) failed to find a difference in meat quality
characteristics (pH and color) between four turkey strains, whereas Sarica et al. (2011) reported
differences in color parameters and pH between a fast growing, commercial strain and a slower
growing strain. Another study comparing differences in breast muscle function among four
turkey strains reported that functional characteristics that are ultimately related to meat quality
(e.g. shear force and water holding capacity) are altered in faster growing strains, such that there
is a “decrease in post-mortem functionality” relative to slower growing strains (Updike et al.,
2005).
Meat quality may also be affected by differences in individual characteristics of animals
such as stress reactivity, fearfulness, and other temperament characteristics. Research with pigs

135

has demonstrated that some muscle characteristics associated with meat quality (e.g. pH and
muscle lactate levels) are associated with the animal’s reactivity to stressors occurring months
prior to slaughter (see Terlouw, 2005). For example, pigs displaying reduced fear of humans are
less reactive to slaughter procedures, and pigs displaying reduced fear of novel objects have
higher muscle pH levels (Terlouw and Rybarczyk, 2008).
Similarly, steers with poor temperaments (measured using flight speed, defined as the
time taken to move a specified distance after being released from a weighing crate) are more
susceptible to stress before slaughter; and have lower live weights, poorer feed conversion
efficiency, lower carcass dressing percentages and lower initial muscle pH levels (Petherick et
al., 2002; reviewed in Norris et al., 2014). Temperament is also associated with production
measures such as bodyweight and time spent eating (Cafe et al., 2011; reviewed in Norris et al.,
2014) and other meat quality characteristics, including cooking loss, shear force, color (Cafe et
al., 2011) and dark cutting (Voisinet et al, 1997).
There is scant research examining temperament and meat quality in poultry, and results
are contradictory. Research with broiler chickens failed to find a relationship between fear
(measured using tonic immobility) and meat quality (Debut et al., 2003). However, meat quality
was lower in quail selected for increased fear (increased tonic immobility duration) that were
exposed to acute stress before slaughter, as indicated by higher pH following slaughter and
increased drip loss (Remignon et al., 1998). The relationship between meat quality and fear
responses has not been examined in turkeys, but results from studies with other species suggest
that animals that differ in stress reactivity and fear responses may differ in characteristics
associated with meat quality.

136

Fearfulness is one of the most widely studied aspects of poultry temperament. In addition
to the tonic immobility test, the open field test is a widely used test for assessing fear responses
of poultry. Recently, it was demonstrated that commercial male turkeys’ responses in an open
field test are repeatable between weeks (Erasmus and Swanson, 2014), and some open field
responses differ between commercial and randombred turkeys (Chapter 4). This study tested the
hypothesis that differences in behavioral and physiological open field responses are associated
with differences in meat quality characteristics of turkeys. Additionally, strain differences in
meat quality characteristics were also examined.

MATERIALS AND METHODS
All procedures were approved by the Institutional Animal Care and Use Committee of
Michigan State University. Materials and methods were the same as for Chapter 4.

Animals and housing
This study was conducted in conjunction with another study examining the effects of
different chilling treatments on meat quality of male randombred (RB) and male commercial
(COMM) turkeys. Results reported here only concern the relationship between OF behavior and
meat quality. Results concerning chilling treatments will be reported elsewhere.
For a full description of animals and housing, refer to Chapter 4 materials and methods.
In total, 62 male RB turkeys were obtained. Commercial poults were placed 1 d prior to
placement of RB poults. Male COMM (Hybrid Converter, Hybrid Turkeys, Kitchener, ON,
Canada) and male RB (RBC2 line, The Ohio Agricultural Research and Development Center of
The Ohio State University, Wooster, OH) turkeys were brooded in same-strain groups in eight

137

littered (wood shavings) pens (2.40 m x 3.05 m) in the same barn from 1 d - 2 wk of age. At 2
wk, birds were distributed throughout the barn and randomly assigned to one of 16 littered pens
(2.4 m x 3.05 m) in the same barn and housed in same-strain groups (14 to 20 birds per pen; 4
pens of males and females of each strain) where they remained until study completion at 14 wk.
At 7 wk, some COMM turkeys were removed from their pens for use in another study and the
number of COMM turkeys in each pen was reduced to, and maintained at 16. Turkeys were
individually identified using colored non-toxic livestock marker as described in Chapter 4
(materials and methods section).

Open field (OF) testing
Birds were tested in the OF test at 1 wk (Test 1), 4 wk (Test 2) and 11 wk (Test 3), but
only results from Test 3 were used here to examine the relationship between OF behavior and
meat quality. Data from Test 3 was used because reliability of OF behavior is higher at later ages
(Erasmus and Swanson, 2014; Chapter 4). Furthermore, physiological responses to OF testing
could only be analyzed for Test 3 due to time and cost constraints associated with running the
assays.
The OF test arena and testing procedures are the same as described in Erasmus and
Swanson (2014). Briefly, each bird was placed in the center of the OF test arena for 10 min. and
behaviour was recorded in real time (60i fields per second) from two overhead high definition
camcorders (VIXIA HF M41, Canon USA, Inc. Melville, NY, USA). Behavioural data,
including the latency to ambulate, number of steps taken, number of squares entered, latency to
vocalize, number of vocalizations and number of defecations, were collected from video
recordings.

138

OF responses and meat quality
To examine the relationship between meat quality and OF responses, birds were assigned
to clusters based on hierarchical cluster analysis of responses during OF Test 3. Test responses
that were used in the cluster analysis included the latency to ambulate, number of steps and
number of defecations. These test measures were selected because ambulatory behavior is
believed to be an indicator of the level of fearfulness in the OF, whereas vocalizations are
believed to be associated with social reinstatement motivation (Forkman et al., 2007). Except for
the number of defecations of RB males (rS = 0.33) and the latency of COMM males to ambulate
(rS = 0.35), all of the aforementioned OF test measures had moderate reliability (rS > 0.4) when
turkeys were tested between 4 and 11 weeks of age (Chapter 4).

Blood sample collection and corticosterone assays
Blood samples were collected from 6 birds in each pen (n = 24 birds/strain) during the
second and third test periods. Only blood samples from wk 11 were assayed due to time and cost
constraints. Furthermore, Test 3 rather than Test 2 was chosen for blood sample analysis because
Test 3 was the last test that was conducted before slaughter. Blood samples were collected
following the procedures outlined in Chapter 4. Briefly, samples were collected one day prior to
OF testing in order to obtain pre-test corticosterone levels, and again the following day 15 min.
after OF testing started. The same competitive enzyme immunoassay kit (Corticosterone HS,
IDS Inc., Fountain Hills, AZ) used by Huff et al. (2007) was used to determine plasma
corticosterone levels.

139

Processing
Males from both strains were slaughtered in three groups of up to 20 birds each. Five
birds from each pen were randomly assigned to a slaughter group. To mimic current turkey
production, male COMM turkeys (14.52 ± 0.29 kg average liveweight) were reared to 15, 16,
and 17 wk in three replicates, respectively. The first and second replicate were slaughtered one
week apart and the third replicate was slaughtered four days after the second. Considering their
small body size, RB males (11.04 ± 0.10 kg average liveweight) were reared up to 20 and 21 wk
and processed in three replicates with three to four days between processing.
After feed withdrawal for 12 h, 20 turkeys from each strain (5 turkeys per pen, randomly
selected) were individually tagged in the wing and transported for 10-15 min. to the Michigan
State University Meat Laboratory. The outside temperature at the time of transportation ranged
between 5 and 14 °C. Upon arrival at the Meat Laboratory, the turkeys were shackled and
electrically stunned for 3 s (80 mA, 60 Hz, 110 V), bled for 90 s by severing both the carotid
artery and jugular vein on one side of the neck, and scalded for 120 s at 59 ºC. After mechanical
defeathering and manual evisceration, the breast muscle (Pectoralis major) was obtained 20-25
min postmortem for meat quality analysis.

Muscle pH, R-value, and color measurements
The pH was measured following procedures of Jeong et al. (2011). Briefly, 5 g of muscle
tissue was taken from the cranial area of the left breast fillet and homogenized in 25 ml of
distilled/deionized water. The pH was then measured using a pH electrode (model 13-620-631,
Fisher Scientific Inc., Houston, TX) attached to a pH meter (Accumet AR15, Fisher Scientific
Inc., Pittsburgh, PA). The R-value (ratio of inosine:adenosine) was measured as an indicator of

140

adenosine triphosphate (ATP) depletion in the muscle using the method of Thompson et al.
(1987).
Color measurements were taken after carcasses were chilled, following the procedure
described by Jeong et al. (2011). Briefly, L* (lightness), a* (redness), and b* (yellowness) values
were measured on the skinless surface of each breast using a chromameter (CR-400, 8-mm
aperture, illuminant C; Konika Minolta Sensing Inc., Osaka, Japan) that had been calibrated with
a white plate (L*, 97.28; a*, −0.23; b*, 2.43). Six readings of CIE L*, a*, and b* were obtained
from the area that was “free of any obvious blood-related defects, such as bruises, hemorrhages,
or full blood vessels” (Fletcher et al., 2000).

Statistical analysis
All statistical analyses were performed using SAS version 9.3. Similar to a study
examining individual differences in behaviour of goats (Miranda-de la Lama et al., 2014),
hierarchical cluster analysis (PROC CLUSTER) was used to assign turkeys to a particular
cluster. Prior to cluster analysis, variables (latency to ambulate, number of steps, number of
squares and number of defecations) were standardized (mean = 0, sd = 1). Ward’s method of
clustering was specified.
The GLIMMIX procedure for a negative binomial distribution with the log link function
was used to verify that the numbers of steps and defecations differed significantly between
clusters. The LIFETEST procedure was used to verify that the latency to ambulate differed
between clusters. The variable corresponding to cluster was specified in the STRATA statement.
Corticosterone levels, carcass weights and meat quality parameters (pH, R-value, and L*,
a* and b* color values) were compared among strains and clusters using a mixed model with

141

Tukey’s adjustment for multiple comparisons. The pH, R-value and carcass weights were
assessed before chilling treatments. Therefore, the slaughter group was included as a random
effect and the lsmeans of the strain and cluster and their interaction were analyzed. Because
turkeys reared within the same pen may have been more similar to turkeys reared in other pens,
the variable corresponding to pen was also included as a random effect.
Color measurements (L*, a* and b* values) were assessed after carcasses had been
chilled. Therefore, the chilling treatment was included as a random effect nested within the
slaughter group. Pen was also included as a random effect. Pre- and post-test corticosterone
levels, carcass weight and a* values were log-transformed. The change in corticosterone level
(post-test – pre-test) was transformed using log + 1, and b* values were transformed using log +
10 to meet normality assumptions.

RESULTS

Cluster analysis
Two clusters were identified for COMM (Fig. 5.1) and RB (Fig. 5.2) turkeys. In general,
turkeys in cluster 1 did not ambulate whereas turkeys in cluster 2 ambulated. The median (2575% range) latency to ambulate (s) differed significantly between clusters of both strains
(COMM cluster 1 = 600 (600-600) vs. COMM cluster 2 = 89 (53-133), log-rank = 57.8, P <
0.0001, Wilcoxon = 52.2, P < 0.0001; RB cluster 1 = 600 (600-600) vs. RB cluster 2 = 86 (47210), log-rank = 53.2, , P < 0.0001, Wilcoxon = 42.4, P < 0.0001).

142

Similarly, the number of steps (lsmean ± SE) differed significantly between clusters
within strains (COMM cluster 1 = 0 (0-2) vs. COMM cluster 2 = 44 (36-76), F42 = 28.7, P <
0.0001; RB cluster 1 = 0 (0-0) vs. RB cluster 2 = 50 (14-119), F46 = 119.2, P < 0.0001).
However, the number of defecations differed between RB clusters but not between
COMM clusters (COMM cluster 1 = 1 (0-2) vs. COMM cluster 2 = 2 (1-2), F42 = 1.5, P = 0.2;
RB cluster 1 = 1 (0-1) vs. RB cluster 2 = 3 (1-4), F46 = 18.6, P < 0.0001). Pre- and post-test
corticosterone levels did not differ between clusters within strains (Fig. 5.3).

Strain and cluster differences in meat quality and carcass weight
A total of four breast fillet samples from each strain were classified as PSE-like, based on
pH ≤ 5.8 (Rathgeber et al., 1999).
Results of the main effects of strain, cluster and their interaction on meat quality and
carcass weight are presented in Table 5.1. The main effects of strain and cluster were significant
for carcass weight, but carcass weight did not differ between clusters within strains. No
significant differences were found for L* or pH. The interaction between strain and cluster was
significant for a*, but no post-hoc differences were found. There was a tendency for b* to differ
between clusters, and the R-value tended to differ between strains.

DISCUSSION
The only differences in meat quality characteristics that were found between strains was a
difference in carcass weight, which is to be expected because COMM turkeys have been selected
for increased body weight, and a tendency for R-values of RB turkey fillets to be higher on
143

average than R-values of COMM turkey fillets. The higher R-value in RB turkey fillets indicates
that the onset of rigor mortis may be accelerated in RB turkeys compared to COMM turkeys.
Accelerated rigor mortis along with low pH (< 5.8) and muscle temperatures above 35 °C are
associated with reduced meat quality; specifically, leading to the development of PSE meat
(Sosnicki et al., 1998). Although R-values tended to differ, L* values did not differ between
strains, and the percentage of PSE-like breast fillets was similar for both strains (COMM = 9%,
RB = 8%). Overall, results demonstrated that there is little difference in meat quality
characteristics between the slower growing RB strain and the COMM strain.
Our results are in agreement with the majority of studies comparing meat quality between
different turkey strains. Results support those of Updike et al. (2005) who compared the same
RB line that was used here with a commercial strain at 16, 18 and 22 weeks. Although strains
differed in shear force and water holding capacity, pH values measured at 2 or 20h post-slaughter
did not differ between the RB and COMM strains (Updike et al., 2005). Results from a study
comparing lightweight and heavyweight lines of British United Turkeys (BUT) similarly
reported no genotype differences in the rate of pH decline (Yost et al., 2002). Our results are also
in agreement with those of Werner et al. (2008) comparing meat quality among a faster growing,
commercial turkey strain (BUT Big 6) and three lines of slower growing broad breasted bronze
turkeys. The faster growing commercial strain did not differ from the broad breasted bronze
strains in color characteristics or in pH levels measured 20 min post-slaughter, but the pH
measured at 4 h post-slaughter was significantly higher in the commercial strain compared to two
of the three broad breasted bronze strains (Werner et al., 2008). In contrast, Sarica et al. (2011)
reported that pH and a* values were higher, and b* values were lower in breast muscle from

144

commercial Hybrid Converter turkeys (the same strain used in our study) compared to slower
growing American Bronze turkeys when birds were slaughtered at 17 and 21 weeks.
Discrepancies among studies may be due to genotype and methodological differences in
animal housing and husbandry, and conditions before and at slaughter. Sarica et al. (2011) report
genotype differences in meat quality results that were pooled across sexes and across different
housing environments, whereas other studies used only males (Updike et al., 2005; Werner et al.,
2008) or reported data separately for males and females (Yost et al., 2002). Furthermore, some
turkeys in the study by Sarica et al. (2011) had outdoor access, whereas those in the other studies
were reared indoors (Updike et al., 2005; Werner et al., 2008; Yost et al., 2002). Indeed, Sarica
et al. (2011) found differences in a* values, and protein, ash and fat content of breast meat from
birds with and without outdoor access. We did not examine strain differences in water holding
capacity or shear force, which have been reported to differ between strains (Updike et al., 2005).
Further research is needed to examine how genetics and environmental conditions influence
water holding capacity and shear force.
The majority of research concerning temperament and meat quality has been conducted
with cattle and pigs, and several studies have provided evidence of a relationship between meat
quality and behavioral temperament characteristics such as responses to fear tests, and
physiological characteristics such as stress reactivity (e.g. Petherick et al., 2002; Terlouw and
Rybarczyk, 2008). Individual animals are consistent in their stress reactivity, and the slaughter
process is associated with a number of events and conditions that may induce stress. Indeed, the
effects of pre-slaughter stress on meat quality have been well documented. Ante-mortem
stressors such as transportation, heat stress and struggling of birds before slaughter promote
faster post-mortem glycolysis which is associated with the development of PSE meat (McCurdy

145

et al., 1996; McKee and Sams, 1997) and increased ultimate pH levels (Terlouw and Rybarczyk,
2008). Therefore, it can be hypothesized that if individual animals are consistent in their stress
reactivity and the slaughter process induces a stress response, then animals that differ in stress
reactivity should differ in meat quality characteristics regardless of whether stress reactivity is
assessed immediately prior to slaughter or weeks or months before slaughter.
Most research comparing meat quality among individuals differing in stress reactivity
have done so by manipulating conditions immediately prior to slaughter. Turkeys in our study
were likely exposed to acute stress before slaughter, because preslaughter and slaughter
procedures are associated with a number of events that may elicit stress and fear responses,
including catching and crating, transportation, separation from conspecifics, and temperature
fluctuations (Terlouw et al., 2008). However, the conditions associated with our study may not
have elicited large enough stress responses for differences in meat quality between clusters to
become apparent. Indeed, turkeys were transported a distance of 5.7 km, with a duration of 10 to
15 min, and research by Owens and Sams (2000) found that transporting turkeys for as long as 3
h did not adversely affect meat quality. Further research is needed to examine the relationships
between acute stress, temperament and meat quality of turkeys. Additional research is needed to
examine whether turkeys’ physiological stress responses during various fear tests are consistent
over time.
In order to examine the relationship between turkey temperament (measured during
rearing) and meat quality, we recorded both behavioral and physiological (plasma corticosterone
levels) responses of turkeys in an OF test at 11 weeks of age. Post-test corticosterone levels were
significantly higher vs. pre-test levels in both strains, indicating that turkeys had in fact
experienced increased stress levels resulting from OF testing procedures (Chapter 4). We used

146

cluster analysis to group COMM and RB turkeys based on their behavior in an OF test, and
verified that ambulatory behavior was indeed different between clusters within strains. However,
there were no differences between clusters within strains in pH, R-value, or color characteristics.
In addition, corticosterone levels did not differ between clusters within strains, indicating that
differing behavioral responses to OF testing were not reflective of corresponding differences in
physiological responses. Results therefore do not support our hypothesis that turkeys differing in
open field behavior differ in meat quality characteristics. Because plasma corticosterone levels
could only be assessed in 8-14 turkeys of each group, the sample size may have been too low to
detect differences in corticosterone levels between clusters.
There do not appear to be any other studies examining the relationship between
temperament and meat quality of turkeys. Nonetheless, our results are in agreement with those of
Debut et al. (2003), who examined the association between meat quality and tonic immobility
responses in female broiler chickens of a fast-growing and slow-growing strain. No association
was found between fearfulness measured using the tonic immobility test and meat pH or color
values (Debut et al., 2003). Similarly, a study with quail from genetic lines selected for long
(increased fear) and short (decreased fear) durations of tonic immobility, respectively, found no
differences in meat color characteristics between short and long tonic immobility lines, but the
pH measured 24 h post-slaughter was significantly higher in quail from the long tonic immobility
line when quail were exposed to an acute stressor before slaughter (Remignon et al., 1998).
An explanation for the discrepancies between our results and those of Remignon et al.
(1998) and between those of Debut et al. (2003) and Remignon et al. (1998) may be that meat
quality differences between birds differing in temperament characteristics are only observed
when birds are exposed to acute stress before slaughter. Although Debut et al. (2003) examined

147

the effects of different pre-slaughter stressors on meat quality, they did not do so within tonic
immobility classifications. Therefore, they were not able to test whether exposing chickens to a
particular stressor before slaughter affected meat quality differently depending on the chicken’s
tonic immobility reaction. Furthermore, only TI was examined and it is possible that the
relationship between meat quality and fear responses may depend on the type of fear test used.
In conclusion, our results are generally in agreement with most other studies that there is
little difference in meat quality characteristics between randombred and commercial male
turkeys. Although studies with pigs have found differences in meat quality between pigs
differing in fear responses during rearing (Terlouw and Porcher, 2005), our results did not
support the hypothesis that there is an association between meat quality and open field responses
of turkeys.

ACKNOWLEDGEMENTS
We thank Hybrid Turkeys and Cuddy Farms for supplying the poults, and Dr. Darrin
Karcher and Angelo Napolitano and the Michigan State University Poultry Research and
Teaching Center for assistance. Furthermore, we thank Janet Ireland for conducting
corticosterone assays, and Thanikarn Sansawat, Pranjal Singh and the student members of the
Michigan State University Animal Behavior and Welfare Group for assistance with data
collection. M.E. was supported by a Natural Sciences and Engineering Research Council of
Canada postgraduate scholarship and this work was supported by Michigan State University
AgBioResearch.

148

APPENDIX

149

Table 5.1. Differences in live weight, carcass weight, pH, R-value, and L*, a*, and b* values between male turkeys of commercial
(COMM) and randombred (RB) strains.
Raw data are presented (mean ± SE). Analyses were conducted with transformed data (live weight, carcass weight, and a* and b*
values).
Carcass weight (kg)
a

14.0 ± 0.4

L*
52.5 ± 0.9

a*
4.4 ± 0.3

b*
-1.1 ± 0.4

pH
6.0 ± 0.03

R-value
1.2 ± 0.02

14.4 ± 0.6

a

52.3 ± 1.4

4.1 ± 0.5

-1.7 ±0.7

6.0 ± 0.06

1.2 ± 0.04

RB Cluster 1 (n = 25)

b

10.8 ± 0.1

51.3 ± 0.8

4.6 ± 0.3

-2.1 ± 0.6

6.0 ± 0.03

1.2 ± 0.03

RB Cluster 2 (n = 22)

11.4 ± 0.2

b

50.8 ± 0.9

4.5 ± 0.3

-2.8 ± 0.4

5.9 ± 0.03

1.3 ± 0.03

F50 = 6.3*

F53 = 0.4

F53 = 0.4

F53 = 2.1

F77 = 1.1

F77 = 3.2

F50 = 6.3*

F53 = 1.1

F53 = 0.2

F53 = 3.2

F77 = 1.2

F77 = 0.4

F50 = 0.3

F53 = 1.2

F53 = 7.7**

F53 = 2.2

COMM Cluster 1 (n = 31)
COMM Cluster 2 (n = 13)

Main effects
Strain
Cluster
Strain x Cluster
a, b
T

Means within columns lacking common superscripts differ (P < 0.10).

P < 0.10, *P < 0.05, **P < 0.01.

150

T

F77 = 0.05 F77 = 0.4

T

Figure. 5.1. Dendogram (graphical representation of cluster analysis) of clusters of commercial (COMM) male turkeys (n = 44) based
on their latency to ambulate, number of steps and number of defecations in an open field test at 11 weeks of age. Two clusters were
identified (dotted line).
0.5

Semi-partial R squared

0.4

0.3

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
0.2

0.1

Cluster 2

Cluster 1

0.0

Subject (individual turkey)

151

Figure. 5.2. Dendogram (graphical representation of cluster analysis) of clusters of randombred (RB) male turkeys (n = 47) based on
their latency to ambulate, number of steps and number of defecations in an open field test at 11 weeks of age. Two clusters were
identified (dotted line).
0.6

0.5

Semi-partial R squared

0.4

0.3

0.2

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
Cluster 2

Cluster 1
0.1

0.0

Subject (individual turkey)

152

Figure. 5.3. Plasma corticosterone levels (pre-test, post-test and change from pre- to post-test) of
male turkeys of a commercial (COMM) and randombred (RB) strain tested in an open field test
at 11 wk of age. Turkeys were grouped into two clusters based on their responses in the open
field test. Raw data (mean ± SE) are presented. Statistical analyses were conducted on logtransformed data.

18

N = 31

N = 13

N = 25

N = 22

16

Corticosterone level (ng/ml)

14
12
10
8
6
4
2
0
Pre-test

Post-test

153

Change

REFERENCES

154

REFERENCES
Cafe, L. M., D. L. Robinson, D. M. Ferguson, G. H. Geesink, and P. L. Greenwood. 2011.
Temperament and hypothalamic-pituitary-adrenal axis function are related and combine
to affect growth, efficiency, carcass, and meat quality traits in Brahman steers. Dom.
Anim. Endocrinol. 40:230-240.
Debut, M., C. Berri, E. Baeza, N. Sellier, C. Arnould, D. Guemene, N. Jehl, B. Boutten, Y. Jego,
C. Beaumont, and E. Le Bihan-Duval. 2003. Variation of chicken technological meat
quality in relation to genotype and preslaughter stress conditions. Poult. Sci. 82:1829–
1838.
Erasmus, M. A., & Swanson, J. C. 2014. Temperamental turkeys: reliability of behavioral
responses to four tests of fear. Appl. Anim. Behav. Sci. 157: 100-108.
Fletcher, D. L., M. Qiao, and D. P. Smith. 2000. The relationship of raw broiler breast meat color
and pH to cooked meat color and pH. Poult. Sci. 79:784–788.
Forkman, B., A. Boissy, M. –C. Maunier-Salaün, E.Canali, and R. B. Jones. 2007. A critical
review of fear tests used on cattle, pigs, sheep, poultry and horses. Physiol. Behav. 92:
340-374.
Huff, G., W. Huff, N. Rath, A. Donoghue, N. Anthony, and K. Nestor. 2007. Differential effects
of sex and genetics on behavior and stress response of turkeys. Poult. Sci. 86:1294-1303.
Jeong, J. Y., K. K. Janardhanan, A. M. Booren, J. B. Harte, and I. Kang. 2011. Breast meat
quality and consumer sensory properties of broiler carcasses chilled by water, air, or
evaporative air. Poult. Sci. 90:694-700.
Martin, J., 2009. The commercialisation of british turkey production. Rural Hist. 20:209-228.
McCurdy, R. D., S. Barbut, and M. Quinton. 1996. Seasonal effect on pale soft exudative (PSE)
occurrence in young turkey breast meat. Food Res. Int. 29:363–366.
McKee, S. R., and A. R. Sams. 1997. The effect of seasonal heat stress on rigor development and
the incidence of pale, exudative turkey meat. Poult. Sci. 76:1616–1620.
Nestor, K. E., M. G. McCartney, and N. Bachev. 1969. Relative contributions of genetics and
environment to turkey improvement. Poult. Sci. 48:1944-1949.
Owens, C. M., and A. R. Sams. 2000. The influence of transportation on turkey meat quality.
Poult. Sci. 79:1204-1207.

155

Petherick, J. C., R. G. Holroyd, V. J. Doogan, and B. K. Venus. 2002. Productivity, carcass and
meat quality of lot-fed Bos indicus cross steers grouped according to temperament. Aust.
J. Exp. Agric. 42:389–98.
Petracci, M. and C. Cavani. 2012. Muscle growth and poultry meat quality issues. Nutrients. 4:112.
Rathgeber, B. M., M. D. Pato, J. A., Boles, and P. J. Shand. 1999. Rapid post-mortem glycolysis
and delay chilling of turkey carcasses cause alterations to protein extractability and
degradation of breast muscle proteins. J. Agric. Food Chem. 47:2529–2536.
Remignon, H., A. D. Mills, D. Guemene, V. Desrosiers, M. Garreau-Mills, M. Marche, and G.
Marche. 1998. Meat quality traits and muscle characteristicsin high or low fear lines of
Japanese quails (Coturnix japonica) subjected to acute stress. Br. Poult. Sci. 39:372-378.
Sarica, M., N. Ocak, S. Turhan, C. Kop, and U. S. Yamak. 2011. Evaluation of meat quality from
3 turkey genotypes reared with or without outdoor access. Poult. Sci. 90:1313-1323.
Sosnicki, A. A., M. L Greaser, M. Pietrzak, E. Pospiech, and V. Sante.1998. PSE-like syndrome
in breast muscle of domestic turkeys: a review. J. Muscle Foods. 9:13-23.
Strasburg, G. M. and W. Chiang. 2009. Pale, soft, exudative turkey-the role of ryanodine
receptor variation in meat quality. Poult. Sci. 88:1497-1505.
Terlouw E. M. C., and P. Rybarczyk. 2008. Explaining and predicting differences in meat quality
through stress reactions at slaughter: the case of Large White and Duroc pigs. Meat Sci.
79:795–805.
Terlouw, E. M. C., and J. Porcher. 2005. Repeated handling of pigs during rearing. I. Refusal of
contact by the handler and reactivity to familiar and unfamiliar humans. J. Anim. Sci.
83:1653-1663.
Terlouw, E. M. C., C. Arnould, B. Auperin, C. Berri, E. Le Bihan-Duval, V. Deiss, F. Lefèvre,
B. J. Lensink, and L. Mounier. 2008. Pre-slaughter conditions, animal stress and welfare:
current status and possible future research. Animal. 2:1501-1517.
Thompson, L. D., D. M. Janky, and S. A. Woodward. 1987. Tenderness and physical
characteristics of broiler breast fillets harvested at various times from post-mortem
electrically stimulated carcasses. Poult. Sci. 66:1158–1167.
Updike, M. S., H. N. Zerby, M. S. Sawdy, G. Kaletunc, and M. P. Wick. 2005. Turkey breast
meat functionality differences among turkeys selected for body weight and/or breast
yield. Meat Sci. 71:706–712.

156

USDA-NASS. 2014. Quick Stats. Available online:
http://quickstats.nass.usda.gov/data/printable/6F84B583-B4A8-3B68-89CFD3F8CD9BFE7B
Voisinet, B. D., T. Grandin, S. F. O’Connor, J. D. Tatuma, and M. J. Deesing. 1997. Bos
Indicus-cross feedlot cattle with excitable temperaments have tougher meat and a higher
incidence of borderline dark cutters. Meat Sci. 4:367-377.
Werner, C., J. Riegel, and M. Wicke. 2008. Slaughter performance of four different turkey
strains, with special focus on the muscle fiber structure and the meat quality of the breast
muscle. Poult. Sci. 87:1849-1859.
Yost, J. K., P. B. Kenney, S. D. Slider, R. W. Russell, and J. Killefer. 2002. Influence of
selection for breast muscle mass on myosin isoform composition and metabolism of deep
pectoralis muscles of male and female turkeys. Poult. Sci. 81:911-917.

157

CHAPTER 6: GENERAL DISCUSSION AND CONCLUSIONS
Injurious pecking is an important problem affecting the well-being of commercially
farmed turkeys, yet there has been little research on this subject. The large amount of scientific
literature regarding feather pecking in chickens indicates that feather pecking is a multifactorial
problem, and the propensity to develop feather pecking differs between individual chickens and
is related to fearfulness and the ability of the hens to cope with stress (Rodenburg et al., 2013).
However, individual differences in fearfulness had not previously been studied in turkeys.
Furthermore, tests used to assess fear in hens and quail have been used to draw conclusions
regarding the well-being of turkeys (e.g. Noble et al., 1996 beak trimming), in spite of there
being no research into turkeys’ fear behavior in general, or on the reliability and validity of these
tests for turkeys. Individual differences in behavior and stress reactivity are also associated with
meat quality of other species (e.g. Terlouw, 2005). Research into meat quality has gained
importance in the turkey industry due to the increased incidence of PSE meat (Sosnicki et al.,
1998). Four studies were conducted to examine the reliability and validity of fear tests for
turkeys, genetic and sex differences in feather pecking and whether feather pecking and meat
quality are associated with differences in fear behavior.
Chapter 2 examined whether the same tests of fear that are commonly used for chickens
and quail are reliable when used to assess fear in male commercial turkeys. The changes in
turkeys’ responses over time were also discussed. Responses to TI, OF and VA tests were
reliable, as indicated by moderately high Spearman rank correlation coefficients. For the NO test,
the only test measure that was reliable was the latency to peck the object. Reliability of test
responses may have been affected by the body conformation and growth rates of the commercial
domestic turkey. Indeed, behavioral responses of the commercial turkey to fear tests may be
158

difficult to interpret due to the changes in body conformation resulting from genetic selection for
increased feed efficiency and higher body weights. For example, the larger breast muscle size of
the commercial domestic turkey may impact the turkey’s ability to right itself during a TI test
when the turkey is tested at an older age. Therefore, there may be changes in TI responses
resulting from age, development and also changes in body conformation. An important area of
future research is to examine how changes in body conformation and age affect behavior.
The same data that was used in Chapter 2 was used in Chapter 3 to examine whether
turkeys’ responses to TI, OF, VA and NO tests were correlated at two ages (4 to 6 weeks and 8
to10 weeks). Turkeys showing extreme behavioral responses were compared across test
situations to assess whether they maintained differences in their responses relative to one
another. Results were rather complicated, revealing differences in test correspondence depending
on the age at which turkeys were tested. At 4 to 6 weeks, ambulatory activity in the OF test
tended to be correlated with the latency to peck the observer during the VA test, indicating that
OF ambulation tended to increase as VA peck latency increased. At 8 to10 weeks, ambulatory
activity in the OF test increased as vocalization during TI decreased, and tended to decrease as
the latency to peck the object during the NO test increased. Overall, results suggested that
behavioral responses to fear tests appear to be specific to the test, and therefore contextdependent because results from Chapter 3 indicated little correspondence between test measures.
Fear tests may be testing different aspects or dimensions of fear and fearfulness, or they may be
testing behavior that is not necessarily associated with fear per se.
A limitation of this research is that only some aspects of reliability and validity were
assessed. Reliability in terms of consistency over repeated applications was assessed, but the
sensitivity of test measures (whether test measures detect small changes in the true value of the

159

level of fear) was not examined because additional experiments, turkeys and resources would
have been required. Additionally, the external validity (whether results are applicable to
situations outside of the experimental environment) was not tested because this would require
research to be conducted in commercial facilities. Further research is therefore needed to validate
the four fear tests. The finding that the OF test was reliable in females and in randombred turkeys
(Chapter 4) demonstrates that the OF test is generalizable across sexes and strains when these
sexes and strains are housed and tested under the same standard conditions.
Fear tests are difficult to interpret because they may have different floor and ceiling
effects. Moreover, fear tests do not necessarily measure the same motivations even though they
are believed to assess some aspects of fear (discussed in Chapter 3). For example, the OF test
assesses neophobia as well as social reinstatement motivation (Forkman et al., 2007). The TI test
on the other hand, is an unlearned response. Although TI duration is influenced by the level of
fear experienced before TI is induced, responses to OF and TI tests may differ because unlike TI,
behavior in the OF test is not the terminal response to capture by a predator. The OF test is
therefore not a “life or death” situation and may not be associated with the same degree of fear
compared to fear induced by a predatory attack. In the VA test, turkeys were repeatedly exposed
to the same observer and the latency to approach and peck the observer decreased with each test
period, indicating reduced fear and/or habituation over time. In fact, most turkeys approached
and pecked the observer; therefore, it is unlikely that the VA test was measuring fear per se.
Arguably, the NO test was most likely to induce a fear response each time the turkeys were
tested (i.e. turkeys were less likely to become habituated to NO testing) because turkeys were
exposed to a different unfamiliar object each time they were tested, and the object was placed in

160

their home pen. However, the NO test was also the least repeatable of the four fear tests, as
discussed in Chapter 2.
Caution is needed when using fear tests to draw conclusions about animal welfare. Fear
tests should not be used indiscriminately because behavior appears to be situation-dependent.
Within the scientific literature, there is debate as to whether fearfulness can be considered a
unitary construct, or whether fearfulness has different dimensions (Boissy, 1995). If each fear
test measures a different dimension of fearfulness, then it is expected that there would be less
correspondence between tests than if they were all measuring the same underlying state. A lack
of correlation between fear tests have led some researchers to conclude that different fear tests
used for poultry are multidimensional and caution is needed when interpreting fear responses
(e.g. Miller et al., 2006; Richard et al., 2008). Overall, results are in agreement with the notion
that fear responses are context-specific and that fear is multidimensional. Indeed, fear tests do
not all measure fear in the same way, some fear tests impose additional stress or situations that
are associated with other behavior and not just fear, and fear tests differ in the degree of fear that
is elicited.
Based on results from Chapters 2 and 3, the OF test was chosen to further examine the
relationship between fear-related behavior, feather pecking and meat quality. The OF test was
chosen for several reasons: turkeys’ responses were repeatable (Chapter 2), the OF test was
associated and tended to be associated with behavior in the TI and VA tests, respectively
(Chapter 3), and the OF test enables birds to be tested individually and is easier to use and
implement compared to the TI test.
In Chapters 4 and 5, statistical analysis confirmed that turkeys’ behavioral OF responses
are significantly different whether turkeys at the extreme ends of the behavior are compared

161

(Chapter 4) or whether turkeys are grouped together using cluster analysis (Chapter 5). These
statistically significant differences between extreme behavioral phenotypes and between clusters
may suggest that turkeys can be categorized into distinct coping styles. However, despite the
behavioral differences, there did not appear to be an association between behavior and
physiological responses to the OF test because corticosterone levels measured before and after
OF testing did not differ between turkeys of different clusters (Chapter 5). Similar results have
been reported when behavioral and physiological responses were compared between HFP and
LFP chicken lines (reviewed in Groothuis and Carere, 2005). This research therefore could not
determine whether coping styles exist in turkeys as they do in other species, and results raise
more questions about the link between behavior and physiology. Furthermore, the relationship
with other behavior such as aggression needs to be assessed in order to develop a clear picture of
turkey temperament.
In contrast to research with laying hens (Rodenburg et al., 2004; reviewed in Rodenburg
et al., 2013), a clear relationship between feather pecking and OF responses was not found. Most
studies examining fearfulness and feather pecking in laying hens uses the HFP and LFP selection
lines. Therefore, it may be more likely that relationships will be found between feather pecking
and fearfulness in selection lines, because variability is reduced in lines that are genetically
similar. Although feather pecking did not appear to be associated with OF behavior, Chapter 4
provides new information regarding feather pecking of turkeys. Specifically, it was interesting to
learn how many individuals were observed to perform feather pecking, and that the percentage of
birds that pecked did not differ between sexes or strains. However, no distinction was made
between turkeys that performed gentle vs. severe pecking, and this is an important avenue for

162

future research because the two forms of feather pecking may not be related (see Rodenburg et
al., 2013).
Chapter 5 examined whether OF behavior was related to meat quality in commercial and
randombred male turkeys. Strains tended to differ in the rate at which rigor mortis developed, but
no other meat quality differences were found. In contrast to research with other species (e.g.
Terlouw, 2005), there was no relationship between physiological or behavioral fear responses
and meat quality. However, turkeys were not specifically subjected to an intense stressor before
slaughter, and it is likely that pre-slaughter stress may result in observable differences in meat
quality. Further research is needed to examine whether physiological responses to fear testing are
comparable to physiological responses resulting from other stressors, and whether pre-slaughter
stress is higher, and meat quality poorer, in turkeys showing increased physiological stress
responses.
The studies presented in this dissertation provide new insights into turkeys’ behavioral
and physiological fear responses and feather pecking behavior and a number of avenues for
future research have been identified. There appear to be distinct individual differences in
behavior that are correlated over time, but the behavior depends on the situation because it is not
strongly correlated to behavior in other tests. Very little is known about turkey temperament and
individual differences in behavioral and physiological stress responses, which presents an
interesting area for future research. Furthermore, the absence of a relationship between
physiological and behavioral responses indicates that more research is needed to understand how
individual animals respond to stress and whether stress reactivity is more likely to be associated
with meat quality if animals are stressed immediately prior to slaughter. Feather pecking remains
a complex problem that is poorly understood, but results reveal that there are strain differences in

163

the severity of feather pecking damage. Further investigation of genetic differences in the
behavior of the recipient of feather pecking, and of the possible genetic differences in feather
cover and skin integrity may provide a means to reduce feather pecking damage.

164

REFERENCES

165

REFERENCES
Boissy, A. Fear and fearfulness in animals. 1995. Q. Rev. Biol. 70: 165-191.
Groothuis, T. G. G., and C. Carere. 2005. Avian personalities: characterization and epigenesis.
Neurosci. Biobehav. Rev. 29: 137-150.
Forkman, B., A. Boissy, M. –C. Maunier-Salaün, E. Canali, and R. B. Jones, R. B. 2007. A
critical review of fear tests used on cattle, pigs, sheep, poultry and horses. Physiol.
Behav. 92: 340-374.
Miller, K. A., J. P. Garner, and J. A. Mench. 2006. Is fearfulness a trait that can be measured
with behavioural tests? A validation of four fear tests for Japanese quail. Anim. Behav.
71:1323-1334.
Noble, D. O., K. E. Nestor, and K. K. Krueger. 1996b. The effect of beak trimming on two
strains of commercial turkeys. 2. Behavior traits. Poult. Sci. 75: 1468-1471.
Norris, D., J.W. Ngambi, M. Mabelebele, O.J. Alabi and K. Benyi. 2014. Genetic selection for
docility: a review. J. Anim. Plant Sci. 24: 374-379.
Richard, S., N. Wacrenier-Ceré, D. Hazard, H. Saint-Dizier, C. Arnould, and J. M. Faure. 2008.
Behavioral and endocrine fear responses in Japanese quail upon presentation of a novel
object in the home cage. Behav. Process. 77: 313-319.
Rodenburg, T. B., A. J. Buitenhuis, B. Ask, K. A. Uitdehaag, P. Koene, J. J van der Poel, J. A.
M. van Arendonk, and H. Bovenhuis. 2004. Genetic and phenotypic correlations between
feather pecking and open-field response in laying hens at two different ages. Behav.
Genet. 34: 407-415.
Rodenburg, T. B., M. M. van Krimpen, I. C. de Jong, E. N. de Haas, M. S. Kops, B. J. Riedstra,
R. E. Nordquist, J. P. Wagenaar, M. Bestman, and C. J. Nicol. 2013. The prevention and
control of feather pecking in laying hens: identifying the underlying principles. Worlds
Poult. Sci. J. 69: 361-373.
Sosnicki, A. A., M. L Greaser, M. Pietrzak, E. Pospiech, and V. Sante.1998. PSE-like syndrome
in breast muscle of domestic turkeys: a review. J. Muscle Foods. 9: 13-23.
Terlouw, C. 2005. Stress reactions at slaughter and meat quality in pigs: genetic background and
prior experience - A brief review of recent. Livest. Prod. Sci. 94: 125-135.

166