Published November 20, 2014

Genetic analysis of a temperament test

as a tool to select against everyday life fearfulness in Rough Collie1
P. Arvelius,2 H. Eken Asp, W. F. Fikse, E. Strandberg, and K. Nilsson
Department of Animal Breeding and Genetics,
Swedish University of Agricultural Sciences, PO Box 7023, S-750 07 Uppsala, Sweden

ABSTRACT: Fear-related problems are common Assessment and Research Questionnaire. Each dogs
among Rough Collies in Sweden. Annually, on average, questionnaire result was condensed into 18 underly-
>200 Rough Collies are subjected to the dog mentality ing behavioral subscales. Genetic parameters for the
assessment (DMA), a temperament test during which subscales were estimated using a linear animal model,
33 behavioral reactions are rated. Previous research has including a fixed effect of sex and random genetic effect
shown that a dogs DMA result can be condensed into 5 of the individual and residual. Age when the question-
underlying personality traits. The aim of the study was naire was completed was included as linear and qua-
to evaluate if it is possible to use the DMA for selec- dratic regressions. Heritability estimates for the ques-
tion for temperament in Swedish Rough Collies, in par- tionnaire subscales were 0.06 to 0.36. There were high
ticular to decrease everyday life fearfulness. We also and significant genetic correlations between DMA per-
wanted to compare 2 methods to compute the personal- sonality traits and questionnaire subscales. For instance,
ity traits: summated scales (SS) and factor scores (FS). the DMA personality trait Curiosity/Fearlessness cor-
The DMA data for 2,953 Rough Collies were used to related strongly genetically to the questionnaire sub-
estimate genetic parameters for the 5 personality traits scale Non-social fear (0.70), DMA Sociability to
(both SS and FS), using a linear animal model including Stranger-directed interest (0.87) and Stranger-directed
fixed effects of sex, year and month of test, and random fear (0.80), DMA Playfulness to Human-directed play
effects of litter, judge, test occasion, genetic effect of interest (0.63), and DMA Chase-proneness to Chasing
the individual, and residual. Age at test was included as (0.73). We could not detect any obvious difference in
linear and quadratic regressions. The DMA personality validity between DMA SS and FS. We conclude that the
trait heritability estimates ranged from 0.13 to 0.25. The DMA is an effective tool for selection of breeding ani-
SS showed greater or equal heritability estimates com- mals with the goal to decrease everyday life fearfulness
pared with the FS. To validate the DMA, data on every- among Swedish Rough Collies. The DMA can also be
day life behavior of 1,738 Rough Collies were collected used for breeding for other traits. The SS method seems
using an extended version of the Canine Behavioral to perform at least as good the FS method.
Key words: behavior, breeding, Collie, dog, fearfulness, temperament

2014 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2014.92:48434855
doi:10.2527/jas2014-8169
INTRODUCTION

1We wish to thank J. Serpell and University of Pennsylvania for High levels of fear and anxiety in dogs cause dif-
permission to use the C-BARQ questionnaire. The Swedish Collie ficulties both for the dogs, from a welfare perspective,
Club and the Swedish Kennel Club are acknowledged for financial
support and the Swedish Kennel Club for permission to use DMA
and for the dog owners by inflicting limitations in their
and pedigree data and for sending letters to Rough Collie owners everyday life. There are several strong indications that
encouraging them to fill out our questionnaire. We would also like fear-related behaviors, especially in nonsocial situa-
to thank K. Svartberg who gave valuable suggestions on content and
structure of the article. Finally, we wish to express our gratitude to all
tions, are common among Rough Collies in Sweden
Rough Collie owners who answered the questionnaire. (elaborated on in Results and Discussion).
2Corresponding author: [email protected]
Since 1997, 25 to 50% of the Rough Collies regis-
Received June 10, 2014. tered in the Swedish Kennel Club each year are subjected
Accepted September 3, 2014.
4843
4844 Arvelius et al.

to the temperament test dog mentality assessment (DMA), Table 1. Parameters describing the Swedish Kennel
during which 33 behavioral reactions are rated. A dogs Club (SKC) pedigree for 3,688 Rough Collies subjected
DMA result can be condensed into 5 underlying person- to the dog mentality assessment (DMA) and/or with
ality traits: Playfulness, Curiosity/Fearlessness, Chase- questionnaire information
proneness, Sociability, and Aggressiveness (Svartberg Parameter Gen. Value
and Forkman, 2002). These traits have been shown to Number of animals in the SKC pedigree1 8,443
correlate phenotypically with, for example, fearfulness Percentage of known ancestors in generation 3 98.7
5 87.6
in everyday life situations (Svartberg, 2005) but not spe-
7 74.9
cifically for Collie. Strandberg et al. (2005) found heri- 9 57.6
tabilities of 0.10 to 0.25 for DMA personality traits for 11 33.7
Rottweiler and German Shepherd Dog. If DMA traits 13 12.6
and everyday life behavior also are genetically correlated 15 2.6
and if DMA traits have reasonably high heritabilities also Complete generation equivalent (CGE)2 8.9
Average number of generations to oldest ancestor 18.0
for Rough Collie, then DMA traits can be used as indica- Number of founders3 1,636
tors of the breeding goal trait everyday life behavior in Number of half-founders4 171
Rough Collie. 1All dogs with DMA and/or questionnaire information and their known
Our main objective was to study if the DMA can ancestors. N nj
be used for selection for temperament in the Swedish 2CGE ( )
= (1/ N ) 1/ 2 , in which N is the number of dogs given
gij

Rough Collie, in particular to decrease everyday life =j 1 =i 1

DMA, nj is the total number of ancestors of animal j, and gij is the number of
fearfulness. We did this by defining underlying DMA generations between j and its ancestor i (Boichard et al. 1997).
traits and estimating heritabilities for these and their ge- 3Both parents unknown.
4One parent unknown; the unknown parent is considered a founder.
netic correlations to relevant breeding goal traits. These
in turn were defined in an extended version of a frequent-
ly used dog owner questionnaire, the Canine Behavioral Impossible observations, such as score of 6 on a scale
Assessment and Research Questionnaire (C-BARQ). from 1 to 5, were considered as missing.
We also compared 2 different methods to calculate the The SAS software (release 9.2; SAS Inst. Inc., Cary,
underlying DMA traits, to see if either method resulted NC) was used for data editing, factor analysis, and calculat-
in larger heritabilities or stronger genetic correlations to ing Cronbachs values, and the DMU software (Madsen
the breeding goal traits. and Jensen, 2010) was used for all genetic analyses.

MATERIALS AND METHODS Dog Mentality Assessment Description and Data

Temperament test data used in this study were ob- The DMA is a test battery of 10 standardized subtests,
tained from preexisting databases. Therefore, animal during which the intensities of 33 behavioral reactions
care and use committee approval was not necessary. displayed by the tested dog are rated according to a stan-
The analyses were based on DMA and question- dardized score sheet. The subtests are performed consecu-
naire data from 3,688 Rough Collies, of which 1,003 tively with only short breaks in between, and the dog is
had both DMA and questionnaire data, and pedigrees handled by its owner during the test. The test is given at
for these dogs. Pedigree and DMA data were received 1 of approximately 200 official test sites, and the whole
from the Swedish Kennel Club. Questionnaire data were test typically takes around 45 min per dog. The score sheet
collected by us. Pedigree data contained records from consists of 5-step scales for all 33 items, and all steps of
Rough Collies born from 1965 to 2011, and after edit- the scales for all items contain short descriptions of typi-
ing (duplicate identity numbers were removed; parents cal behaviors. The intention when constructing the scales
who did not occur as individuals were added as indi- was to define each step of a scale as objectively as possible
viduals with unknown parents; and dogs with different and to arrange the steps from low to high intensity of the
ID numbers but with the same name, mother, and birth behavioral reactions; that is, a low rating corresponds to
date were considered as the same individual and only 1 low intensity of the reaction. No judgment is made during
of the records was kept) and excluding all dogs except the test whether a dog shows preferred behaviors or not.
those with DMA or questionnaire information and their The DMA was developed during the 1980s by peo-
ancestors, 8,443 individuals remained. Pedigree depth ple from the Swedish Working Dog Association, the
and completeness is described in Table 1. Questionnaire governmental Swedish Dog Training Centre, Stockholm
and DMA records belonging to dogs that did not exist University, and the Swedish University of Agricultural
in pedigree data were deleted, and so were records with Sciences. In 1989 the test became official, in the sense that
missing values for all DMA and questionnaire items. since then all results have been registered by the Swedish
 Options for selection against fearfulness 4845

Table 2. Information on systematic environmental of the DMA test procedure, behavioral definitions in the
effects for 2,953 Rough Collies subjected to the dog score sheets, etc., see Svartberg and Forkman (2002).
mentality assessment (DMA) and for 1,738 Rough The DMA data contained records from 2,953 iden-
Collies with questionnaire information tifiable Rough Collies tested from January 1997 to
Systematic N November 2010 (Tables 2 and 3). In addition to the be-
environmental effect Class DMA Questionnaire havioral ratings, the data included information on test
Sex Male 1,364 878 date, judge, and test occasion (the combination of test
Female 1,589 860
date and test site). The corresponding Swedish Kennel
Test year (DMA) and birth 1997 111 0
year (questionnaire) Club pedigree included information on sex, birth date,
1998 139 0 and litter identity (mother identity combined with birth
1999 117 125 date). The sex ratio of males to females for DMA re-
2000 108 126 cords was close to even. The average test age was 21
2001 191 123 mo (SD of 10, minimum of 12, and maximum of 120),
2002 225 165
2003 217 161
and 50% of the dogs were younger than 18 mo and 78%
2004 248 182 were younger than 24 mo when tested.
2005 272 138
2006 275 172 Dog Mentality Assessment Definitions
2007 250 211
of Personality Traits
2008 299 176
2009 254 159
2010 247 0
In a previous study, 5 so-called personality traits
Test month (DMA) January 4 have been defined using factor analysis of the DMA
February 10 items (Svartberg and Forkman, 2002): Playfulness,
March 179 Curiosity/Fearlessness, Chase-proneness, Sociability, and
April 439 Aggressiveness. Svartberg (2005) used 23 of the 33 DMA
May 522
June 245
items to compute summated scales (SS) for the personal-
July 86 ity traits (a SS was computed as the average of the rep-
August 333 resentative variables for a factor). The Swedish Kennel
September 527 Club publishes SS for these 5 personality traits on their
October 494 web page (www.skk.se) for all dogs subjected to the
November 110
DMA. Today, based on more recent studies (K. Svartberg,
December 4
Litters1 1,234 928
Svartbergs Hundkunskap, Alunda, Sweden, personal
DMA judges2 236 communication), the Swedish Kennel Club includes only
DMA occasions3 1,101 22 of the items when calculating the SS (Table 4).
1Average number of DMA-tested dogs per litter = 2.39 (SD = 1.70, mini- The DMA personality trait definitions have not been
mum = 1, and maximum = 10). Average number of dogs with questionnaire made specifically for Collie but more, in general, for the
information per litter = 1.87 (SD = 1.08, minimum = 1, and maximum = 7). dog as a species. Furthermore, the definitions used today,
2Average number of tested dogs per judge = 12.52 (SD = 23.12, mini-
mum= 1, and maximum = 225).
where 22 of the DMA items are used for calculating the
3Average number of tested dogs per occasion = 2.68 (SD = 2.41, mini- SS, have not been published in a peer-reviewed journal.
mum= 1, and maximum = 14). As a pilot study, following Hair et al. (1998), we therefore
performed a factor analysis of the DMA data for Rough
Collie, which resulted in 5 factors (latent root criterion;
Kennel Club. After revising the test procedure as well as Table 4). The factor analysis technique applied was prin-
the score sheet, a second version of the DMA was launched cipal axis factoring (Sharma, 1996), and it was done us-
in 1997. The present study is based solely on the latter ver- ing Proc Factor (METHOD = PRINIT) in SAS (release
sion. The DMA was designed with the main purpose to 9.2). Sampling adequacy was examined using the Keyser-
help breeders select breeding animals with respect to tem- Meyer-Olkin (KMO) measure as described by Sharma
perament traits of importance to working dogs. Today the (1996); KMO was 0.84, suggesting that the correlation
test is considered and used as a general temperament test, matrix was appropriate for factoring. After orthogonal
not only for working dog breeds. All judges, as well as varimax rotation, the factor loading pattern indicated that
test leaders responsible for the practical aspects of the test 22 of the 33 items would be appropriate to use when com-
and others involved in the actual testing procedure, have puting 5 composite scores following the concept of SS. A
been trained and certified by the Swedish Working Dog SS was calculated as an average of the standardized (mean
Association. To be allowed to participate in the DMA, a of 0 and SD of 1) values for the items judged to be good
dog has to be at least 1 yr old. For a detailed description representatives for that factor, defined as showing a factor
4846 Arvelius et al.

Table 3. Descriptive statistics on Rough Collie results loading with absolute value >0.4 for the factor and >0.25
from the dog mentality assessment (DMA) and the dog difference between the greatest and the second greatest
owner questionnaire loading. Between 3 and 7 itemsexactly the same as used
Variable N Mean1 SD by the Swedish Kennel Club todaywere included in
DMA item each SS. Furthermore, the loadings from the rotated solu-
SOCIAL CONTACT Greeting 2,953 3.59 0.62 tion were used for computing factor scores (FS). To com-
SOCIAL CONTACT Cooperation 2,953 3.29 0.68
pute a FS, all 33 items were included, weighed with their
SOCIAL CONTACT Handling 2,953 3.18 0.71
PLAY 1 Interest in play 2,947 3.10 1.00
respective factor loadings for that factor. Thus, composite
PLAY 1 Grabbing 2,947 2.59 0.97 scores were constructed both as SS and as FS, and both
PLAY 1 Tug of war 2,945 2.00 1.10 types of scores were used in the further analyses. When
CHASE Following 1 2,941 1.73 1.00 computing DMA SS and FS, no missing value for any
CHASE Following 2 2,940 1.71 1.09 item used for the calculation was tolerated.
CHASE Grabbing 1 2,940 1.20 0.55
CHASE Grabbing 2 2,938 1.29 0.77
Our factor analysis gave a similar pattern of fac-
PASSIVE SITUATION Activity 2,937 3.24 0.88 tor loadings as the study by Svartberg and Forkman
DISTANCE PLAY Interest 2,935 2.69 0.69 (2002). Therefore, the factors were considered as per-
DISTANCE PLAY Aggression 2,936 1.47 0.93 sonality traits and given the same names: Playfulness,
DISTANCE PLAY Exploration 2,936 1.92 1.27 Curiosity/Fearlessness, Chase-proneness, Sociability,
DISTANCE PLAY Tug-of-war 2,935 1.80 1.14
and Aggressiveness. The traits are briefly described in
DISTANCE PLAY Play invitation 2,935 1.81 1.21
SUDDEN APPEARANCE Startle reaction 2,915 3.49 1.06
Table 5. Out of the 5 DMA personality traits (calculated
SUDDEN APPEARANCE Aggression 2,892 1.58 0.78 as SS), 4 showed Cronbachs values > 0.70 (Table 6),
SUDDEN APPEARANCE Exploration 2,879 2.02 0.98 indicating sufficient internal consistency (Hair et al.,
SUDDEN APPEARANCE Remaining avoidance 2,848 2.29 1.20 1998). For Aggressiveness, the Cronbachs value was
SUDDEN APPEARANCE Remaining approach 2,838 1.62 0.77 0.64, suggesting slightly weaker correlations among the
METALLIC NOISE Startle reaction 2,808 3.56 1.23
METALLIC NOISE Exploration 2,774 2.45 1.29
representative items for this trait. Our results regarding
METALLIC NOISE Remaining avoidance 2,727 2.04 1.15 internal consistencies for the 5 personality traits agree
METALLIC NOISE Remaining approach 2,720 1.57 0.76 well with those shown by Svartberg (2005).
GHOSTS Aggression 2,682 1.82 0.88 The DMA item Gunshot avoidance describes a dogs
GHOSTS Attention 2,682 3.35 1.01 reaction when a functionary fires a starters gun (9 mm) at
GHOSTS Avoidance 2,682 2.78 1.41
a 20 m distance from the dog, during both activity (play-
GHOSTS Exploration 2,678 2.55 1.44
GHOSTS Greeting 2,677 3.01 1.02
ing tug-of-war with the owner) and passivity. The intention
PLAY 2 Interest in play 2,668 2.77 1.13 is to measure fear-related behaviors, and the 5-step inten-
PLAY 2 Grabbing 2,667 2.26 1.09 sity scale goes from Not affected, quick control (1) to
GUNSHOT Avoidance 2,666 2.62 1.58 Affected, scared, or wants to leave the area, tries to escape
Questionnaire behavioral subscale (5). Gunshot avoidance does not, however, fulfill the criteria
Attachment/attention seeking 1,728 1.81 0.67
Chasing 1,720 1.53 1.04
to be included in the personality trait Curiosity/Fearlessness
Dog-directed aggression/fear 1,713 0.41 0.45 (or in any of the other personality traits), because the fac-
Dog-directed aggression 1,717 0.53 0.63 tor loading on this factor is only 0.33. But Gunshot avoid-
Dog-directed fear 1,714 0.30 0.51 ance clearly intends to capture a type of nonsocial fear and
Dog-directed interest 1,709 2.48 0.86 is therefore still of potential interest for Collie breeders.
Dog rivalry 1,149 0.42 0.57
Therefore, we decided to include Gunshot avoidance in the
Energy 1,724 1.95 0.91
Excitability 1,729 1.61 0.69 further analyses in addition to the 5 personality traits.
Human-directed play interest 1,728 2.68 0.85
Non-social fear 1,732 0.62 0.66 Questionnaire Description, Data,
Owner-directed aggression 1,725 0.10 0.22 and Definitions of Behavior Subscale Scores
Pain sensitivity 1,725 0.40 0.55
Stranger-directed aggression 1,715 0.29 0.41
Stranger-directed fear 1,730 0.31 0.62
Data on dog behavior in everyday life situations were
Stranger-directed interest 1,731 2.74 0.93 collected by asking approximately 3,600 owners of Rough
Separation-related behavior 1,727 0.19 0.35 Collies aged from 6 mo to 10 yr to fill out a web-based
Trainability 1,729 2.99 0.50 questionnaire. Data were collected during the period June
1Minimum and maximum ratings/scores are 1 and 5 for the DMA items through October 2010 and included records from 1,738
and 0 and 4 for the questionnaire behavioral subscales. All ratings and scores identifiable dogs with a close to even sex ratio (Tables 2
occurred in the data, except the following questionnaire behavioral subscale
scores: >3.50 for Dog rivalry, >2.88 for Owner-directed aggression, >3.75 for
and 3). The average age of the dog when the questionnaire
Pain sensitivity, >3.30 for Stranger-directed aggression, >3.13 for Separation- was answered was 68 mo (SD of 37, minimum of 6, and
related behavior, and <0.88 for Trainability. maximum of 140) and 92% of the dogs were still alive.
 Options for selection against fearfulness 4847

Table 4. Factor loadings1 for 33 dog mentality assessment (DMA) items on 5 orthogonally rotated (varimax) factors
(N = 2,666 to 2,953), and indications according to the Swedish Kennel Club (SKC) definition on which items that are
included when they compute summated scales for the personality traits Sociability (Soc), Playfulness (Play), Chase-
proneness (Chase), Aggressiveness (Agg), and Curiosity/Fearlessness (Cur/Fear)
Factors SKC
DMA item Playfulness Curiosity/Fearlessness Chase-proneness Sociability Aggressivness definition
SOCIAL CONTACT Greeting 0.15 0.04 0.01 0.65 0.07 Soc
SOCIAL CONTACT Cooperation 0.15 0.05 0.04 0.64 0.03 Soc
SOCIAL CONTACT Handling 0.14 0.08 0.01 0.61 0.01 Soc
PLAY 1 Interest in play 0.76 0.04 0.12 0.21 0.11 Play
PLAY 1 Grabbing 0.77 0.01 0.12 0.08 0.08 Play
PLAY 1 Tug of war 0.67 0.04 0.16 0.18 0.10 Play
CHASE Following 1 0.14 0.17 0.72 0.03 0.12 Chase
CHASE Following 2 0.17 0.10 0.73 0.01 0.09 Chase
CHASE Grabbing 1 0.12 0.16 0.68 0.04 0.02 Chase
CHASE Grabbing 2 0.17 0.09 0.66 0.02 0.05 Chase
PASSIVE SITUATION Activity 0.14 0.01 0.08 0.10 0.06
DISTANCE PLAY Interest 0.19 0.11 0.20 0.12 0.21
DISTANCE PLAY Aggression 0.01 0.08 0.00 0.01 0.51 Agg
DISTANCE PLAY Exploration 0.36 0.24 0.39 0.31 0.01
DISTANCE PLAY Tug-of-war 0.53 0.18 0.37 0.33 0.00
DISTANCE PLAY Play invitation 0.48 0.19 0.37 0.33 0.01
SUDDEN APPEARANCE Startle reaction 0.00 0.60 0.08 0.01 0.01 Cur/Fear
SUDDEN APPEARANCE Aggression 0.10 0.09 0.09 0.00 0.57 Agg
SUDDEN APPEARANCE Exploration 0.05 0.57 0.10 0.00 0.02 Cur/Fear
SUDDEN APPEAR. Remaining avoidance 0.07 0.58 0.01 0.08 0.00 Cur/Fear
SUDDEN APPEAR. Remaining approach 0.05 0.09 0.12 0.10 0.18
METALLIC NOISE Startle reaction 0.05 0.64 0.09 0.05 0.12 Cur/Fear
METALLIC NOISE Exploration 0.11 0.60 0.11 0.03 0.15 Cur/Fear
METALLIC NOISE Remaining avoidance 0.07 0.60 0.01 0.02 0.06 Cur/Fear
METALLIC NOISE Remaining approach 0.01 0.13 0.09 0.10 0.11
GHOSTS Aggression 0.06 0.10 0.01 0.03 0.76 Agg
GHOSTS Attention 0.11 0.20 0.07 0.07 0.44
GHOSTS Avoidance 0.03 0.35 0.08 0.09 0.06
GHOSTS Exploration 0.08 0.48 0.07 0.19 0.12 Cur/Fear
GHOSTS Greeting 0.13 0.36 0.07 0.37 0.00
PLAY 2 Interest in play 0.83 0.18 0.10 0.08 0.10 Play
PLAY 2 Grabbing 0.79 0.18 0.09 0.01 0.06 Play
GUNSHOT Avoidance 0.09 0.33 0.07 0.07 0.12
Variance explained after rotation (%) 12 9 8 6 5
1Bold indicates |factor loading| > 0.4 and bold and italic indicates <0.25 difference between the greatest loading (if >0.4) and the second greatest loading for a rating.

In the questionnaire, the dog owner was asked to 22 of which can be removed without potentially reduc-
rate their dogs typical behavior in the recent past on ing reliability and/or validity (J.Serpell, University of
a 5-step scale;: either the frequency of certain behav- Pennsylvania, personal communication); we removed 21
iors (Never, Seldom, Sometimes, Usually, or Always) or of these, leaving 80 items. To these original C-BARQ
the intensity of the behavior in defined situations (e.g., items, we added 15 questions regarding playfulness and
No aggression: No visible signs of aggressive behav- sociability according to Svartberg (2005). It has been
iors to Serious aggression: Snaps, bites or attempts shown that it is possible to condense these 80 and 15
to bite). The major part of the questionnaire was a items into 15 and 3 underlying behavioral subscales, re-
Swedish translation of the C-BARQ. The C-BARQ was spectively, by computing so-called behavioral subscale
developed and validated by Hsu and Serpell (2003) and scores (Tables 5 and 6; Hsu and Serpell, 2003; Svartberg,
later modified to improve reliability of some existing 2005; Duffy and Serpell, 2012). Between 2 and 10 items
factors, and to include new Dog rivalry and Energy fac- were included in each score, which was calculated as
tors (J. Serpell, University of Pennsylvania, personal an average of the included items. In this study, no miss-
communication). The C-BARQ contains 101 questions, ing values were accepted for scores based on 2 question-
4848 Arvelius et al.

Table 5. Summary description of the dog mentality assessment (DMA) personality traits and questionnaire behav-
ioral subscales, based on behaviors mentioned in the score sheet/questionnaire
Variable Summary description:
The personality trait or behavioral subscale relates to/describes
DMA personality trait
Playfulness a dogs interest in, and intensity when, playing with an object (tug-of-war) together with a human.
Curiosity/Fearlessness the intensity of flight behaviors displayed by a dog when exposed to suddenly appearing human-like objects or loud noises,
and thereafter the degree of difficulties to return to a normal state of mind and the signs of remaining fearfulness.
Chase-proneness a dogs interest in, and intensity when, following a rag quickly moving away in a zigzag pattern, and engagement in the rag if
catching it.
Sociability a dogs eagerness to reject or initiate contact with, and handling by, an unknown friendly person.
Aggressiveness the frequency and severity of aggressive behaviors displayed by a dog when confronted with a suddenly appearing, or slowly
approaching, human or human-resembling object.
Questionnaire behavioral subscale
Attachment/attention seeking1 a dogs propensity to stay close to the owner and to seek attention, and how agitated it becomes when the owner directs his/
her attention towards someone else.
Chasing1 a dogs tendency to chase after cats, birds and other small animals (if given the opportunity).
Dog-directed aggression/fear1 a dogs tendency to display aggressive or fearful reactions when approached by unfamiliar dogs.
Dog-directed aggression2 a dogs tendency to display aggressive reactions when approached by unfamiliar dogs.
Dog-directed fear2 a dogs tendency to display fearful reactions when approached by unfamiliar dogs.
Dog-directed interest3 a dogs eagerness to greet, approach and play with unfamiliar dogs.
Dog rivalry2 aggressive response towards other dogs in the household, especially when approached when in possession of resources such
as food, toy or favorite sleeping place.
Energy2 how energetic, active and playful the dog is.
Excitability1 how strong reaction the dog displays to potentially exciting or arousing events, for example, going for a walk or a car ride, or
the doorbell signal.
Human-directed play interest3 a dogs eagerness to play with an object (e.g., a ball, stick, etc.) together with familiar and unfamiliar humans.
Non-social fear1 a dogs tendency to show fearful or wary responses to sudden or loud noise, in heavy traffic, to unfamiliar situations and
objects, during thunderstorms, and to wind or wind-blown objects.
Owner-directed aggression1 threatening or aggressive behavior towards owner or other household member when verbally corrected or punished,
challenged, stepped over, during handling or when approached while in possession of food or objects.
Pain sensitivity1 fearful or wary reaction to potentially painful situations, for example, when being groomed, bathed, or examined by a
veterinarian.
Stranger-directed aggression1 threatening or aggressive reaction towards unfamiliar persons approaching or invading the dog, the owner or the dogs or the
owners territory.
Stranger-directed fear1 the degree of fearful reactions when approached by unfamiliar persons.
Stranger-directed interest3 a dogs willingness and eagerness to greet and to approach unfamiliar persons.
Separation-related behavior1 a dogs tendency to vocalize, shake, tremble, salivate, scratch at doors, etc., when left, or about to be left, alone.
Trainability1 a dogs willingness to pay attention to and obey the owner, and its ability to learn new tasks and to ignore distracting stimuli.
1Behavioral subscale defined in Hsu and Serpell (2003)
2Behavioral subscale defined in Duffy and Serpell (2012)
3Behavioral subscale defined in Svartberg (2005)

naire items, scores based on 4 to 5 items could have 1 value indicated at least as good internal consistency as in
missing value, and scores based on 6 items could have previous studies (0.75 compared with 0.74 and 0.71).
2 missing values. For dogs with information from both the DMA and
Eight of the 18 behavioral subscales showed the C-BARQ (N = 1,003), the DMA took place first for
Cronbachs values < 0.70 (Table 6), indicating poor in- 931 dogs. For 71 dogs, the C-BARQ was answered be-
ternal consistencies (Hair et al., 1998). Compared with fore the DMA test had been taken, and for 1 dog the
results from previous studies, our Cronbachs values questionnaire was answered on the day of DMA testing.
were generally lower. Hsu and Serpell (2003) obtained The average time period from the DMA to the question-
adequate Cronbachs values (>0.70) for 10 of 11 exam- naire for all 1,003 dogs was 48 mo (SD of 34, minimum
ined subscales and Svartberg (2005) for 9 of (the same) 11 of 5, and maximum of 123).
C-BARQ subscales and for all the 3 additional subscales
regarding playfulness and sociability. In our study, only 4 Genetic Parameters
of these 11 C-BARQ subscales and 2 of the 3 additional
subscales showed Cronbachs values > 0.70. However, Dog Mentality Assessment. Model choice for ana-
for the most important questionnaire subscale given the lyzing DMA items and personality traits was based on
aim of our studyNon-social fearour Cronbachs previous model comparisons for DMA personality traits
 Options for selection against fearfulness 4849

Table 6. Internal consistencies (Cronbachs values [CA]) and variance components1 for additive genetic ( s2a ), lit-
ter ( sl2 ), judge ( s2j ), occasion ( so2 ), and residual ( se2 ) effects and heritabilities1 for dog mentality assessment (DMA)
personality traits computed as summated scales and for questionnaire behavioral subscales

Variable N No. of items2 CA s2a sl2 s2j so2 se2 h2 (SE)

DMA personality trait
Playfulness 2,667 5 0.90 0.14 0.01 0.01 0.03 0.43 0.25 (0.04)
Curiosity/Fearlessness 2,678 7 0.78 0.07 0.01 0.02 0.02 0.28 0.20 (0.04)
Chase-proneness 2,938 4 0.82 0.09 0.02 0.01 0.02 0.48 0.16 (0.03)
Sociability 2,953 3 0.75 0.12 0.01 0.03 0.03 0.44 0.22 (0.04)
Aggressiveness 2,682 3 0.64 0.08 0.03 0.02 0.02 0.43 0.14 (0.04)
Questionnaire behavioral subscale
Attachment/attention seeking 1,728 6 0.53 0.04 0.40 0.10 (0.04)
Chasing 1,720 4 0.71 0.15 0.91 0.14 (0.04)
Dog-directed aggression/fear 1,713 8 0.69 0.02 0.17 0.12 (0.04)
Dog-directed aggression 1,717 4 0.73 0.03 0.33 0.09 (0.04)
Dog-directed fear 1,714 4 0.64 0.03 0.23 0.11 (0.04)
Dog-directed interest 1,709 5 0.62 0.09 0.49 0.15 (0.04)
Dog rivalry 1,149 4 0.72 0.05 0.27 0.16 (0.06)
Energy 1,724 2 0.77 0.04 0.70 0.06 (0.03)
Excitability 1,729 6 0.62 0.05 0.43 0.10 (0.04)
Human-directed play interest 1,728 5 0.70 0.18 0.48 0.28 (0.05)
Non-social fear 1,732 6 0.75 0.15 0.28 0.36 (0.06)
Owner-directed aggression 1,725 8 0.69 0.00 0.04 0.10 (0.03)
Pain sensitivity 1,725 4 0.59 0.06 0.23 0.21 (0.05)
Stranger-directed aggression 1,715 10 0.85 0.04 0.13 0.24 (0.05)
Stranger-directed fear 1,730 4 0.74 0.10 0.29 0.25 (0.05)
Stranger-directed interest 1,731 5 0.74 0.24 0.57 0.29 (0.05)
Separation-related behavior 1,727 8 0.75 0.02 0.11 0.14 (0.05)
Trainability 1,729 8 0.69 0.04 0.21 0.15 (0.04)
1Bold indicates variance component or heritability estimate significantly different from 0 (P < 0.05) defined as >1.645 SE, that is, a 1-sided t test. For DMA
personality traits, = =
h 2 s2a / (s2a + sl2 + se2 ) . For questionnaire behavioral subscales, h 2 sa2 / (s2a + se2 ) .
2Number of items included when computing a summated scale/behavioral subscale score for the variable.

in German Shepherd Dog by Strandberg et al. (2005). j (j= 1997, , 2010); month is the fixed effect of test
They used a model including fixed effects of sex, age, month k (k = January, , December); b1 age and b2
test year, test month, and judge and random effects of lit- age2 are the linear and quadratic regressions on the
ter, additive genetic effect of the individual, and residual. animals age at test (in days); litter is the random effect
In our data, 104 judges (out of 236) were represented of litter m~ND(0, s2L ), in which ND is normally dis-
with 5 records and 155 with 10. Therefore, we chose tributed, s2L is the litter variance; judge is the random
to treat the effect of judge as random rather than fixed. effect of judge n ~ND(0, s2J ), in which s2J is the judge
Based on DMA analyses for 13 breeds (Arvelius et al., variance; occasion is the random effect of test occasion
2
2010), we also included test occasion as a random ef- o ~ND(0, sO2 ), in which sO is the occasion variance;
fect. After studying plots of test result for the 5 DMA SS ap is the random additive genetic effect (i.e., breeding
against age at test, we decided to include age (days) at value) of animal p ~ND(0, A s2A ), in which A is the ad-
2
test as linear and quadratic regressions. Variance compo- ditive genetic relationship matrix and s A is the additive
nents for the 33 DMA items (N = 2,666 to 2,953) were genetic variance; and eijklmnop is the random residual
estimated in univariate analyses using the following effect related to observation yijklmnop ~ND(0, s2E ), in
mixed linear animal model: which s2E is the residual variance. The effects are fur-
ther defined in Table 2. The heritability was defined as
yijklmnop =  + sexi + yearj + monthk + b1 =h 2 s2A / (s2L + s2A + s2E ) .
agel + b2 agel2 + litterm + judgen + Genetic parameters for the DMA personality traits
occasiono + ap + eijklmnop, [1] Playfulness, Curiosity/Fearlessness, Chase-proneness,
Sociability, and Aggressiveness were estimated in 2
in which yijklmnop is the rating for an item for animal multivariate analyses1 for traits calculated as SS (N=
p; is the overall mean; sex is the fixed effect of sex i 2,667 to 2,953) and 1 for FS traits (N = 2,664)using
(i = male, female); year is the fixed effect of test year model [1] in a multivariate setting. The expectations of
 4850 Arvelius et al.

random effects were all 0 with the following variance ditive relationship matrix and s2A is the additive genetic
covariance matrices: variance; and eilp is the random residual effect related to
a1 s2A ... s A A observation yilp ~ND(0, s2E ), in which s2E is the residual
1 1 5
variance. The effects are further defined in Table 2. The
V ...
= ... ... A ,
sym. heritability was defined= as h 2 s2A / (s2A + s2E ) .
a
5 s2A
5 Correlations between Dog Mentality Assessment
and Questionnaire. Genetic correlations between ques-
litter1 Is L1 ... Is L L
2

= 1 5
tionnaire behavioral subscales and the 2 versions of the
V ... ... ... , 5 DMA personality traits and the DMA item Gunshot
litter sym. Is2L
5 5
avoidance were estimated in bivariate analyses applying
judge1 Is J1 ...
2
Is J J the models in Eq. [1] and [3] and the variance covari-
= , ance structure in Eq. [2] with the appropriate covari-
1 5

V ... ... ...
judge sym.
ances deleted. For 1 of the analyses, Gunshot avoidance
5 Is2J
5 with Non-social fear, we had to relax the convergence
occasion1 IsO1
2
... IsO O criteria for norm of the update vector from default value
= 1 5
, and 1.0 107 to 5.0 106 for the analysis to converge.
V ... ... ...
Corresponding phenotypic correlations were computed
occasion sym. IsO2 1/2
5 5 as rp = (s A + s E ) / (s2A + s2L + s2E ) (s2A + s2E ) , in which i
i, j i, j i i i j j
and j are the DMA personality trait (or the item Gunshot
e1 Is ... Is E E
2

E5 1 5
avoidance) and the questionnaire behavioral subscale,
V ... = ... ... , [2] respectively.
e sym. Is2E Because of the large variation in time between when
5 5

DMA and questionnaire data were collected, one concern

in which sym. indicates that the matrix is symmetric. was that it would not be optimal to use all dogs with data for
Genetic correlations between the DMA item Gunshot estimating genetic correlations between DMA personality
avoidance and each of the 5 personality traits were esti- traits and questionnaire subscales. To investigate this fur-
mated in bivariate analyses using Eq. [1] and [2]. To study ther, we plotted Pearson correlations between phenotypic
how similar corresponding DMA SS and FS traits were scoresquestionnaire behavioral subscale scores vs. DMA
(SS Playfulness vs. FS Playfulness, etc.), correlations be- SSagainst number of years from the DMA to the ques-
tween all corresponding random effects were estimated in tionnaire. We could not detect any systematic differences in
bivariate analyses using Eq. [1] and [2]. correlations depending on time between the DMA and the
Questionnaire. All dog owner questionnaire questionnaire. Therefore, all available records were used to
data were collected the same year during only 5 mo. estimate genetic correlations.
Therefore, effects of year or month were not regarded
as meaningful to include in the model when analyzing RESULTS AND DISCUSSION
the questionnaire behavioral subscales. Also, litter was
excluded as random effect, because each litter was rep-
resented by, on average, only 1.87 dogs (Table 2). After Heritabilities and Correlations among Dog Mentality
studying plots of results for the 18 questionnaire behav- Assessment Items and Personality Traits
ioral subscale scores against age of the dogs, we decided All 5 SS and 5 FS DMA personality traits showed sig-
to include age (days) when the questionnaire was com- nificant heritabilities (h2 = 0.14 to 0.25 [Table 6] and h2
pleted as linear and quadratic regressions. = 0.13 to 0.21 [not shown], respectively). For most traits,
Variance components for all 18 questionnaire behavior- a majority of the random environmental factors (litter,
al subscales (N = 1,149 to 1,732) were estimated in univari- judge, and occasion) were significant. Heritabilities for
ate analyses using the following mixed linear animal model: the 33 individual DMA items varied between 0.03 and
0.30 (SE = 0.02 to 0.04) with an average of 0.14, and all
yilp = + sexi + b1 agel + b2 agel2 + ap + eilp, [3] but 2 (Distance play aggression and Sudden appearance
remaining approach) were significantly different from 0.
in which yilp is a behavioral subscale score for animal For the DMA item Gunshot avoidance, the heritability
p; is the overall mean; sex is the fixed effect of sex i was estimated at 0.10 (SE = 0.04). The item heritabili-
(i = male, female); b1 age and b2 age2 are the linear ties were, in general, lower than for the personality traits.
and quadratic regressions on the animals age (in days); This is likely because the items included in a personal-
ap is the random additive genetic effect (i.e., breeding ity trait were highly genetically correlated to each other
value) of animal p ~ND(0, A s2A ), in which A is the ad- and therefore can be regarded as repeated measurements
 Options for selection against fearfulness 4851

Table 7. Genetic correlations1 among dog mentality assessment (DMA) personality traits computed as summated
scales and the DMA item Gunshot avoidance (N = 2,667 to 2,953) and their SE (in parentheses), obtained from 1
multivariate analysis (all 5 summated scales) and 5 bivariate (Gunshot avoidance vs. each of the summated scales)
DMA personality trait
DMA personality trait Playfulness Curiosity/fearlessness Chase-proneness Sociability Aggressiveness
Curiosity/Fearlessness 0.47 (0.12)
Chase-proneness 0.67 (0.10) 0.74 (0.11)
Sociability 0.69 (0.10) 0.06 (0.14) 0.40 (0.13)
Aggressiveness 0.29 (0.15) 0.33 (0.16) 0.20 (0.16) 0.16 (0.16)
Gunshot avoidance 0.50 (0.17) 0.54 (0.17) 0.63 (0.17) 0.26 (0.18) 0.15 (0.22)
1Bold indicates genetic correlation significant from 0 (P < 0.05) defined as >1.645 SE, that is, a 1-sided t test.

of the same underlying personality trait, leading to de- due to greater residual variance for the FS (not shown).
creased random error variance. On the one hand, inclusion of all 33 original DMA items
Our heritability estimates for the DMA items and to calculate all 5 FS could have been expected to reduce
personality traits are well in concordance with results residual variance with greater heritabilities as a result
from previous studies. Saetre et al. (2006) analyzed (when calculating SS, only 3 to 7 items were used to
DMA results from 5,964 German Shepherd Dogs and calculate each SS). On the other hand, many items are
4,589 Rottweilers. Their heritability estimates for the only weakly correlated to each other and inclusion of
items were quite similar for the 2 breeds and varied be- all items when calculating FS apparently increased the
tween 0.04 and 0.19 and a majority were significantly residual variances and thus had a negative influence on
different from 0 (P< 0.01). Strandberg et al. (2005) esti- the FS heritabilities. These results have practical impli-
mated heritabilities for 4 of the 5 DMA personality traits cations. In our opinion, SS are simpler both to compute
to 0.09 to 0.26 ( s2a was significantly different from 0 and to explain. Therefore, we recommend choosing the
[P< 0.05] for all traits) using results from 2,959 German SS method over the FS method when computing DMA
Shepherd Dogs. personality trait scores for Rough Collie.
All 5 DMA personality traits (as well as most of the The DMA SS Playfulness, Curiosity/Fearlessness,
33 DMA items) were clearly influenced by genetic fac- Chase-proneness, and Sociability were rather strongly
tors and can therefore be selected for. They were also genetically intercorrelated (rg = 0.40 to 0.74), with 1
influenced by several systematic environmental effects, exception: Curiosity/Fearlessness with Sociability (rg=
which therefore should be taken into account when select- 0.06; Table 7). The fifth SS, Aggressiveness, showed
ing breeding animals. Selection on phenotype is the most weaker correlations (|rg| = 0.16 to 0.33). Previous find-
common method in dog breeding today. Using a BLUP ings suggest that these first 4 personality traits can be
animal model to estimate breeding values for relevant combined into 1 more basic personality trait, called
temperament traits for the Swedish Rough Collie popula- Boldness (Svartberg and Forkman, 2002; Strandberg
tion would potentially increase the annual genetic gain. et al., 2005; Saetre et al., 2006). This is not evident
Compared with the previous situation, it therefore would from our results, due to the weak correlation between
be possible to accomplish a faster improvement in the ge- Curiosity/Fearlessness and Sociability.
netic predisposition among Rough Collies regarding, for Both Curiosity/Fearlessness and Gunshot avoidance
example, Curiosity/Fearlessness and Gunshot avoidance. showed their strongest genetic correlations to Chase-
There were strong correlations for genetic, litter, proneness (Table 7). It can be hypothesized that these
judge, occasion, and residual effects between corre- correlations reflect an indirect link between the traits,
sponding SS and FS (SS Playfulness vs. FS Playfulness, rather than that they are biochemically controlled by the
SS Curiosity/Fearlessness vs. FS Curiosity/Fearlessness, same genes. Chasing after an object involves leaving the
etc.; rg = 0.95 to 1.00, SE = 0.00 to 0.06; rl = 0.97 to safeness of being close to the owner and to become ex-
1.00, SE = 0.02 to 0.15; rj = 0.85 to 0.99, SE = 0.01 to posed to more unknown and unpredictable, potentially
0.09; ro = 0.87 to 0.99, SE = 0.02 to 0.08; and re = 0.91 to harmful, conditions. Therefore, chasing requires some
0.97, SE = 0.00 to 0.01), indicating that they can be con- degree of courage. Therefore, when measuring Chase-
sidered as more or less the same traits. For all pairwise proneness in a breed such as Rough Collie with a high
comparisons of heritability estimates, the SS method re- proportion of fearful dogs, one might actually be mea-
sulted in equal or greater estimates compared with the suring whether or not the dog is courageous enough to
FS method, with the exception for Aggressiveness (h2 = chase, rather than the eagerness for chasing in itself.
0.14 and h2 = 0.15, SE = 0.04 for both estimates), mainly
4852 Arvelius et al.

Questionnaire Heritabilities
ity traits, a justified question is if it would not be better to
The 18 questionnaire behavioral subscale heritabili- select directly on the highly heritable breeding goal trait
ties (0.06 to 0.36, all significant) were of similar size as Non-social fear rather than on correlated DMA traits.
the DMA personality trait heritabilities (Table 6). One However, if a questionnaire was used as a basis for
could have expected that trained judges and standardized routine genetic evaluation, we see a substantial risk that
implementation of the DMA would reduce the measure- the reliability of the answers with time would become
ment error and thereby generate more reliable ratings of compromised. Basically, we think it would be easier and
the dogs behavior compared with a dog owner question- more tempting for breeders to manipulate the breeding
naire. One explanation for the similarities between DMA values of their dogs by convincing their puppy buyers to
and questionnaire heritabilities could be that the owner give certain answers in the questionnaire than to bring
compensates for lack of training and standardization by about improved behavioral reactions in the DMA.
having the opportunity of observing a dogs behavior dur-
ing a significant portion of its life, in contrast to the DMA Genetic Correlations between Dog Mentality
where the judge observes the dog only for around 45 min. Assessment and Questionnaire
The C-BARQ has been widely used in studies related
to dog behavior (e.g., Svartberg, 2005; Duffy et al., 2008; There were significant genetic correlations between
Duffy and Serpell, 2008; Barnard et al., 2012). There are, to both versions of the DMA personality traits (SS and FS)
our knowledge, very few studies where genetic parameters and everyday life behavior of the dogs as described by
have been estimated for the C-BARQ items or subscales. the owners in the questionnaire. The correlation pattern
Liinamo et al. (2007) present highly varying heritability esti- was similar for both methods to calculate DMA person-
mates, some of them extremely high, for different C-BARQ ality trait scores, and neither method resulted in system-
scores related to aggressiveness in Golden Retriever dogs. atically greater correlations than the other (not shown).
However, their analyses included relatively few individu- Based on similarities in behaviors indicating the
als (N = 115 to 316), which, in addition, were preselected; different DMA personality traits and questionnaire sub-
the subjects had been recruited to the study either because scales (Table 5), we had expected the following genetic
they had shown aggressive behavior or because they were correlations to be strong and significant:
closely related to an aggressive dog. Several of the herita-
bility estimates were 0.00 or 1.00, and for roughly half of DMA Playfulness with questionnaire Human-
the analyses, no SE could be obtained. These circumstances directed play interest,
make it very difficult to evaluate which of the results can be DMA Curiosity/Fearlessness with questionnaire
trusted. This is also pointed out by the authors, who empha- Non-social fear,
size that the results should be approached with caution and DMA Chase-proneness with questionnaire Chasing,
that the conclusions that can be drawn from the study are DMA Sociability with questionnaire Stranger-
limited. In a master thesis study, Schiefelbein (2012, 2013) directed interest and Stranger-directed fear,
collected C-BARQ data on Labrador Retrievers, Golden DMA Aggressiveness with questionnaire
Retrievers, and German Shepherd Dogs that were 6 and 12 Stranger-directed aggression, and
mo old. The dogs had been bred at 2 U.S. guide dog schools. DMA Gunshot avoidance with questionnaire Non-
Heritabilities for the subscales were estimated at 0.00 to 0.47 social fear.
(SE = 0.01 to 0.12). Only every seventh estimate was >0.10, Each of the 5 DMA personality traits (computed
and therefore the heritabilities were, in general, lower com- as SS), as well as the DMA item Gunshot avoidance,
pared with our results. This could be because Schiefelbeins showed significant genetic correlations to at least 2 of
study was done on very young dogs, whereas most dogs in the 18 questionnaire subscales (Table 8). All expected
our study were adults (90% were 18 mo or older). One can correlations were confirmed for DMA Playfulness,
also hypothesize that the guide dog populations from which Curiosity/Fearlessness, Chase-proneness, Sociability,
the samples were collected may be more genetically ho- and Gunshot avoidance. Furthermore, each of these ex-
mogenous than the Swedish Rough Collie population and pected genetic correlations was greater than all other ge-
that this is a reason for the difference in heritabilities. netic correlations, in the way that the greatest correlation
Our main objective was to study if it is possible to involving Playfulness was with Human-directed play
use the DMA for selection for temperament. To investi- interest, and the greatest correlation involving Human-
gate this, we wanted to estimate heritabilities for the DMA directed play interest was with Playfulness, and so on.
personality traits and their genetic correlations to the The only DMA personality trait for which we did not get
breeding goal traits, the main one being the questionnaire the expected result was Aggressiveness, which was not
subscale Non-social fear. Because Non-social fear shows significantly genetically correlated to Stranger-directed
a higher heritability (0.36) than any of the DMA personal- aggression. These results are very similar to the phe-
 Options for selection against fearfulness 4853

Table 8. Genetic correlations1 and their SE (in parentheses) between dog mentality assessment (DMA) personality
traits computed as summated scales (N = 2,667 to 2,953) and questionnaire behavioral subscales (N = 1,149 to 1,732),
obtained from 108 bivariate analyses
Questionnaire DMA personality trait
behavioral subscale Playfulness Curiosity/fearlessness Chase- proneness Sociability Aggressivness Gunshot avoidance2
Attachment/attention seeking 0.14 (0.19) 0.30 (0.20) 0.31 (0.18) 0.14 (0.19) 0.26 (0.23) 0.24 (0.25)
Chasing 0.54 (0.15) 0.59 (0.16) 0.73 (0.13) 0.31 (0.17) 0.42 (0.19) 0.48 (0.21)
Dog-directed aggression/fear 0.38 (0.17) 0.33 (0.19) 0.20 (0.18) 0.50 (0.16) 0.26 (0.22) 0.68 (0.28)
Dog-directed aggression 0.42 (0.20) 0.34 (0.22) 0.25 (0.21) 0.49 (0.18) 0.28 (0.23) 0.43 (0.32)
Dog-directed fear 0.31 (0.18) 0.35 (0.19) 0.13 (0.19) 0.39 (0.18) 0.16 (0.23) 0.73 (0.23)
Dog-directed interest 0.07 (0.16) 0.09 (0.17) 0.07 (0.16) 0.45 (0.16) 0.40 (0.19) 0.47 (0.23)
Dog rivalry 0.39 (0.24) 0.51 (0.24) 0.40 (0.24) 0.26 (0.24) 0.05 (0.28) 0.50 (0.27)
Energy 0.08 (0.18) 0.08 (0.19) 0.03 (0.19) 0.36 (0.18) 0.29 (0.22) 0.13 (0.25)
Excitability 0.01 (0.18) 0.05 (0.19) 0.07 (0.18) 0.12 (0.18) 0.12 (0.22) 0.28 (0.22)
Human-directed play interest 0.63 (0.11) 0.30 (0.14) 0.40 (0.14) 0.19 (0.14) 0.18 (0.16) 0.58 (0.16)
Non-social fear 0.52 (0.11) 0.70 (0.10) 0.43 (0.12) 0.46 (0.11) 0.05 (0.16) 1.003 (0.12)
Owner-directed aggression 0.32 (0.18) 0.24 (0.20) 0.13 (0.19) 0.48 (0.17) 0.18 (0.22) 0.42 (0.28)
Pain sensitivity 0.34 (0.14) 0.31 (0.16) 0.29 (0.15) 0.45 (0.14) 0.23 (0.18) 0.55 (0.23)
Stranger-directed aggression 0.23 (0.14) 0.08 (0.15) 0.00 (0.15) 0.58 (0.12) 0.20 (0.17) 0.08 (0.20)
Stranger-directed fear 0.40 (0.13) 0.44 (0.14) 0.30 (0.14) 0.80 (0.09) 0.07 (0.18) 0.53 (0.20)
Stranger-directed interest 0.28 (0.13) 0.09 (0.14) 0.12 (0.14) 0.87 (0.08) 0.16 (0.16) 0.09 (0.19)
Separation-related behavior 0.54 (0.16) 0.43 (0.18) 0.44 (0.18) 0.51 (0.17) 0.14 (0.22) 0.69 (0.22)
Trainability 0.27 (0.16) 0.36 (0.18) 0.32 (0.18) 0.03 (0.16) 0.13 (0.20) 0.66 (0.21)
1Underlined value indicates expected strong and significant correlation and bold indicates genetic correlation significant from 0 (P < 0.05) defined as >1.645
SE, that is, a 1-sided t test.
2Gunshot avoidance is a DMA item, not a personality trait.
3Relaxed convergence criteria for norm of the update vector to 5.0 106 (default value, 1.0 107).

notypic correlations studied by Svartberg (2005). The as chasing after small animals in everyday life situations,
only obvious difference is for DMA Chase-proneness, an existing genetic correlation would not be reflected by
which in our study correlated strongly with the question- a corresponding phenotypic correlation.
naire subscale Chasing (rg= 0.73, SE= 0.13), whereas Selecting against everyday life fearfulness in non-
Svartberg (2005) could not show a significant (pheno- social situations is a high priority when it comes to the
typic) correlation. When we estimated the phenotypic Swedish Rough Collie population. Therefore, the ques-
correlation based on our data and compared it to the tionnaire subscale Non-social fear is important as a breed-
genetic correlation, the genetic correlation was more ing goal trait. The genetic correlation to Non-social fear
than 4 times greater. For the remaining 5 correlations was high and significant for the DMA personality trait
between DMA personality traits and questionnaire sub- Curiosity/Fearlessness (rg = 0.70, SE = 0.10), and even
scales, which we had expected to be highly associated, greater for the DMA item Gunshot avoidance (rg = 1.00,
the genetic correlations were 1.5 to 2.5 times greater SE = 0.12). Therefore, selecting breeding animals with
than the corresponding phenotypic correlations. The the aim to increase the Curiosity/Fearlessness score, or to
reason behind the big difference between phenotypic decrease the Gunshot avoidance rating, will result in less
and genetic correlations for DMA Chase-proneness with Non-social fear in everyday life situations as described
questionnaire Chasing would be interesting to study fur- by the dog owners. However, both Curiosity/Fearlessness
ther. One hypothesis could be that the genetic correla- and Gunshot avoidance are significantly genetically cor-
tion between proneness to chase a rag in the test and related not only to Non-social fear but also to, for ex-
to chase after animals in everyday life is masked on a ample, Chasing, Dog rivalry, Stranger-directed fear, and
phenotypic level by, on average, greater training skills Separation-related behaviors. In other words, breeding for
among owners of dogs with high proneness for chas- less fearful dogs based on the DMA will also result in
ing. The reason could be that a dogs chase-proneness dogs that have a greater tendency to chase after small ani-
is valuable for trainers of working dogs, who therefore mals, are more likely to respond aggressively if provoked
tend to obtain dogs who are genetically predisposed to by other dogs in the household, are less afraid of unfamil-
show high degrees of this trait. If working-dog trainers, iar persons, and show less problem behaviors when left
at the same time, are more capable than companion-dog alone. The common approach to avoid undesired selec-
owners to limit unwanted behaviors in their dogs, such tion responses for some traits is to use a selection index
4854 Arvelius et al.

with several index traits and have a breeding goal with ap- We found nothing in either the questionnaire or the
propriate weights. It could be of interest to develop such DMA data that contradicted the indications that the prob-
a selection criterion and breeding goal specifically for the lem is real and substantial. Rather, the frequency of dogs
Rough Collie, but that is outside the scope of this paper. that seem to be encumbered with high levels of fear and
Our results indicate that the DMA item Gunshot anxiety gives reason for concern. For example, for each
avoidance is genetically identical to Non-social fear in of the following situations, approximately half (45 to
everyday life (rg = 1.00). Could one not, then, breed 55%)of the Rough Collies described by their owners in
only for Gunshot avoidance? Is there any point paying our questionnaire study typically displayed some degree
attention to Curiosity/Fearlessness (rg = 0.70) when se- of fear-related behaviors: in response to sudden or loud
lecting breeding animals? We would recommend selec- noises such as vacuum cleaner or road drills, during thun-
tion primarily based on Curiosity/Fearlessness but also der storms, when exposed to fireworks or the sound of
to consider Gunshot avoidance. There are 2 reasons for gun shots, when walking on slippery floors, when walk-
this. First, the heritability is twice as high for Curiosity/ ing on unstable surfaces, when walking in an unfamiliar
Fearlessness (h2 = 0.20, SE = 0.04) as for Gunshot avoid- stair, and/or when riding an elevator. Among the 2,666
ance (h2 = 0.10, SE = 0.04), resulting in greater accuracy Rough Collies subjected to the Gunshot test situation in
of selection and a potentially faster genetic gain. Second, the DMA, 1 out of 5 was categorized according to the
the genetic correlation between Gunshot avoidance and scales greatest intensity of fearful behaviors: Affected,
Non-social fear may be overestimated because this bi- scared, or wants to leave the area, tries to escape.
variate analysis had problems to converge (we had to
relax the convergence criteria). We believe that the cor- Valid BLUP Breeding Values with Sufficient
relation is indeed very high (for example, the variance or Accuracies is Not Enough to Achieve Genetic Progress
covariance components were stable over iterations), but
until this has been confirmed in further analyses it is still There are several possible explanations as to why
prudent to be careful. Swedish Collie breeders so far have experienced little
The 5 DMA factors together explain only 39% of the success in decreasing the problems with fearfulness
total variation (Table 4). It is possible that some of the in the breed. One reason might be different opinions
questionnaire subscales, for example, Non-social fear, among breeders about the breeding goal and how to
would be better captured if DMA data were treated differ- prioritize among various traits. A contributing reason
ently. For example, one could try extracting more than 5 could be that modern methods for genetic evaluation of
factors or performing the factor analysis using the genetic potential breeding animals, such as BLUP, still, today,
correlations matrix rather than phenotypic records. In a are rarely used by dog breeders (discussed by Arvelius
future study we intend to test more advanced methods for and Klemetsdal, 2013).
defining and computing composite DMA traits. No matter how good the tools the breeders are provided
with, no genetic progress will happen unless the tools are
Collie and Fearfulness widely used. Besides trusting and wanting to use estimat-
ed breeding values over phenotypic records, the average
One important reason for conducting this study was the Swedish Rough Collie breeder also has to put much more
strong indications that fear-related behaviors are common weight on behavior than on appearance when selecting
among Swedish Rough Collies. Ranking the 75 breeds breeding animals. Previous studies describe how pedigree
with the greatest number of DMA tested dogs in Sweden (N dog breeders, to gain success at dog shows, tend to empha-
= 114 to 17,143) based on their average score for the per- size appearance and conformation rather than behavioral
sonality trait Curiosity/Fearlessness, Rough Collie shows traits (Willis, 1995; McGreevy and Nicholas, 1999; Mki
the lowest score of all. Ranking the breeds for the remain- et al., 2005; Svartberg, 2006; McGreevy, 2008; Rooney,
ing 4 traits, from lowest (rank 1) to greatest average score 2009). In a simulation study on the Finnish Rottweiler
(rank 75), Rough Collie has rank 2 for Chase-proneness, population, Mki et al. (2005) compared genetic respons-
rank 9 for Playfulness and Aggressiveness, and rank 27 for es for hip and elbow dysplasia, behavior, and appearance
Sociability. Concerns repeatedly expressed by the Swedish when different breeding schemes were applied. They con-
Collie Club about the breeds temperament status are cluded that to achieve genetic improvement for health and
thus supported by DMA results. Furthermore, it has been behavioral traits, changes were required in the current dog
shown by Svartberg (2006) and K. Svartberg (Svartbergs breeding programs; emphasis on selection for health and
Hundkunskap, Alunda, Sweden, personal communication) behavior had to be increased at the expense of appearance.
that Rough Collies, on average, show high degrees of fear- Since March 2012, BLUP breeding values for, for ex-
related behaviors in nonsocial situations compared with ample, DMA Curiosity/Fearlessness and Gunshot avoid-
dogs from other breeds. ance are published openly on the Swedish Collie Club
 Options for selection against fearfulness 4855
Duffy, D. L., and J. A. Serpell. 2008. Behavioral assessment of guide
web page (www.svenskacollieklubben.se/mentalindex/ and service dogs. J. Vet. Behav. 3(4):186188.
mentala_index.htm). The breed club works to inform Duffy, D. L., and J. A. Serpell. 2012. Predictive validity of a method
and educate breeders and puppy buyers about breeding for evaluating temperament in young guide and service dogs.
and temperament, how to interpret and use breeding val- Appl. Anim. Behav. Sci. 138:99109.
Hair, J. F., R. E. Anderson, R. L. Tatham, and W. C. Black. 1998. Factor
ues, etc. In the near future, a system for certifying litters
Analysis. In: Multivariate data analysis. 5th ed. Prentice Hall, Upper
with good chances of becoming less fearful than Rough Saddle River, NJ. p. 87-140.
Collies in general will be launched. A breeder with a Hsu, Y. Y., and J. A. Serpell. 2003. Development and validation of a
litter for sale, where the average breeding value of the questionnaire for measuring behavior and temperament traits in
parents for Curiosity/Fearlessness (and maybe also for pet dogs. J. Am. Vet. Med. Assoc. 223:12931300.
Gunshot avoidance) is better than the breed average, will Liinamo, A.-E., L. van den Berg, P. A. J. Leegwater, M. B. H. Schilder,
J. A. M. van Arendonk, and B. A. van Oost. 2007. Genetic varia-
be allowed to market the litter as Mentally Sound Collie tion in aggression-related traits in Golden Retriever dogs. Appl.
Certified by the Swedish Collie Club. The goal is that Anim. Behav. Sci. 104:95106.
with time it will become natural for puppy buyers to re- Madsen, P., and J. Jensen. 2010. A users guide to DMU A pack-
quest only certified puppies, thereby putting pressure on age for analysing multivariate mixed models. Version 6, release
the breeders to breed less fearful Rough Collies. 5.0. University of Aarhus, Faculty Agricultural Sciences (DJF),
Dep. of Genetics and Biotechnology, Research Centre Foulum,
Tjele, Denmark.
Conclusions Mki, K., A.-E. Liinamo, A. F. Groen, P. Bijma, and M. Ojala. 2005.
The effect of breeding schemes on the genetic response of ca-
We conclude that the temperament test DMA could nine hip dysplasia, elbow dysplasia, behaviour traits and ap-
be an effective tool for selection of breeding animals pearance. Anim. Welf. 14:117124.
with the goal to decrease everyday life fearfulness in the McGreevy, P. 2008. Comment: We must breed happier, healthier
dogs. New Sci. 200(2677):18.
Swedish Rough Collie population. The DMA can also McGreevy, P. D., and F. W. Nicholas. 1999. Some practical solutions
be used for breeding for, for example, Human-directed to welfare problems in dog breeding. Anim. Welf. 8:329341.
play interest, Chasing, Stranger-directed fear, and Rooney, N. J. 2009. The welfare of pedigree dogs: Cause for concern
Separation-related behavior. The SS method to compute (Editorial). J. Vet. Behav. 4:180186.
DMA personality trait scores seems to perform at least Saetre, P., E. Strandberg, P.-E. Sundgren, U. Pettersson, E. Jazin, and
T. F. Bergstrm. 2006. The genetic contribution to canine per-
as good the FS method; heritabilities and genetic cor-
sonality. Genes Brain Behav. 5:240248.
relations between DMA results and everyday life behav- Schiefelbein, K. M. 2012. Estimation of genetic parameters for be-
ior as described by dog owners are generally equal or havioral assessment scores in Labrador Retrievers, German
greater for the SS. Because they are also easier to use for Shepherd Dogs, and Golden Retrievers. MSc thesis. Kansas
practical breeding purposes, SS are the first choice in a State University, Manhattan, Kansas.
breeding program for Rough Collie based on DMA data. Schiefelbein, K. M. 2013. Estimation of genetic parameters for be-
havioral assessment scores in Labrador Retrievers, German
Shepherd Dogs, and Golden Retrievers. In: Abstracts from the
LITERATURE CITED 8th Int. Working Dog Conf., San Antonio, Texas. p. 6.
Arvelius, P., F. Fikse, and E. Strandberg. 2010. Bilaga 5 Genetiska Sharma, S. 1996. Factor Analysis. In: Applied multivariate tech-
analyser av MH. In: C. Blixt, P. Arvelius, K. Svartberg, and niques. John Wiley & Sons, New York. p. 90-143.
S. Trenkle Nyberg, editors, Utarbetande av ett beteende- och Strandberg, E., J. Jacobsson, and P. Saetre. 2005. Direct genetic, ma-
personlighetstest fr avelsndaml. (In Swedish.) Report for ternal and litter effects on behaviour in German shepherd dogs
the Swedish Kennel Club. www.skk.se/Global/Dokument/Om- in Sweden. Livest. Prod. Sci. 93:3342.
SKK/BPH/BPH-rapport-2010.pdf. (Accessed 21 May 2014.) Svartberg, K. 2005. A comparison of behaviour in test and in every-
Arvelius, P., and G. Klemetsdal. 2013. How Swedish breeders can day life: Evidence of three consistent boldness-related personal-
substantially increase the genetic gain for the English Setters ity traits in dogs. Appl. Anim. Behav. Sci. 91:103128.
hunting traits. J. Anim. Breed. Genet. 130:142153. Svartberg, K. 2006. Breed-typical behaviour in dogs Historical rem-
Barnard, S., C. Siracusa, I. Reisner, P. Valsecchi, and J. A. Serpell. nants or recent constructs? Appl. Anim. Behav. Sci. 96:293313.
2012. Validity of model devices used to assess canine tempera- Svartberg, K., and B. Forkman. 2002. Personality traits in the domes-
ment in behavioral tests. Appl. Anim. Behav. Sci. 138:7987. tic dog (Canis familiaris). Appl. Anim. Behav. Sci. 79:133155.
Boichard, D., L. Maignel, and . Verrier. 1997. The value of using Willis, M. B. 1995. Genetic aspects of dog behaviour with particular
probabilities of gene origin to measure genetic variability in a reference to working ability. In: J. Serpell, editor, The domes-
population. Genet. Sel. Evol. 29:523. tic dog Its evolution, behaviour and interactions with people.
Duffy, D. L., Y. Y. Hsu, and J. A. Serpell. 2008. Breed differences in Cambridge Univ. Press, Cambridge, UK. p. 5164.
canine aggression. Appl. Anim. Behav. Sci. 114:441460.