- Email: [email protected]
The effects of exercise and calm interactions on in-kennel behavior of shelter dogs
Accepted Manuscript Title: The Effects of Exercise and Calm Interactions on In-kennel Behavior of Shelter Dogs Authors: Alexandra Protopopova, Hagar Hauser, Kissel J. Goldman, Clive D.L. Wynne PII: DOI: Reference:
S0376-6357(17)30443-6 https://doi.org/10.1016/j.beproc.2017.11.013 BEPROC 3546
To appear in:
Behavioural Processes
Received date: Revised date: Accepted date:
20-9-2017 13-11-2017 16-11-2017
Please cite this article as: Protopopova, Alexandra, Hauser, Hagar, Goldman, Kissel J., Wynne, Clive D.L., The Effects of Exercise and Calm Interactions on In-kennel Behavior of Shelter Dogs.Behavioural Processes https://doi.org/10.1016/j.beproc.2017.11.013 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Running Head: EXERCISE OR CALM INTERACTION
1
The Effects of Exercise and Calm Interactions on In-kennel Behavior of
SC RI PT
Shelter Dogs
Alexandra Protopopova1*, Hagar Hauser2, Kissel J. Goldman3, Clive D. L. Wynne4
of Animal and Food Sciences, Texas Tech University, Lubbock, TX, USA
U
1 Department
College of Veterinary Sciences, University of Florida
3
Department of Psychology, University of Florida, Gainesville, FL, USA
4
Department of Psychology, Arizona State University, Tempe, AZ, USA.
A
N
2
M
*Correspondence concerning this article should be addressed to Alexandra Protopopova,
D
Department of Animal and Food Sciences, Texas Tech University, Lubbock, TX 79415-2141.
TE
Contact: [email protected] Highlights
Some behaviors may be unattractive to adopters and an indicator of poor welfare in
EP
kenneled dogs.
CC
We assessed the effects of providing calm interaction and exercise on in-kennel behavior
A
of 16 shelter dogs.
These interventions result in both a reduction of some, but also in an increase of other unattractive behavior.
EXERCISE OR CALM INTERACTION
2
Over-activity, or excessive locomotion and barking in the kennel, may be unattractive to adopters and an indicator of poor welfare of kenneled dogs. The study assessed the efficacy of
SC RI PT
two common enrichment strategies, providing calm interaction and additional exercise, on inkennel behavior in 16 shelter dogs. Both interventions resulted in appropriate behavior just prior to the sessions (t = 2.10, df = 7, p = .03 and F [2, 216] = 7.58, p = .0007, respectively), but both also resulted in an increase of some undesirable behaviors immediately after the dogs were taken back to their kennels (F [3, 216] = 7.77, p = .0001 and F [5, 216] = 10.1, p < .0001 respectively).
U
Right after receiving additional exercise, the dogs spent more time in back and forth motion in
N
the kennel. Right after receiving the calm interaction, the dogs spent less time in the front of the
A
kennel, less time facing forward, and more time engaging in back and forth motion. However,
M
dogs also spent less time barking and jumping on the kennel door right after the calm interaction.
D
The results suggest that both interventions may be useful, but shelter administrators and
EP
interventions.
TE
volunteers must take all of the behavioral changes into account when administering these
CC
Keywords: Behavior, Dog, Enrichment, Exercise, Learning
A
Animal shelters take in large quantities of surrendered, stray, and confiscated dogs. In the
United States (US) alone, approximately 3.9 million dogs enter animal shelters yearly (American Society for the Prevention of Cruelty to Animals [ASPCA], 2015a). Shelter housing is likely to be stressful, at least initially, (Hennessy, Davis, Williams, Mellot, & Douglas, 1997; Hiby, Rooney, & Bradshaw, 2006; Shiverdecker, Schiml, & Hennessy, 2013; Stephen & Ledger, 2006)
EXERCISE OR CALM INTERACTION
3
and many dogs may stay at the shelter for a prolonged period of time or are euthanized for lack of interested adopters. One determinant of adoptability is the behavior of the animal while at the shelter
SC RI PT
(Luescher & Medlock, 2009; Protopopova, Mehrkam, Boggess, & Wynne, 2014; Protopopova & Wynne, 2014; Weiss, Miller, Mohan-Gibbons, & Vela, 2012; Wells & Hepper, 1992). Previous research has suggested that dogs that engage in excessive back and forth motion in the kennel,
rub or lean on kennel walls, spend a prolonged period of time in the back of the kennel, or face backwards (Protopopova et al., 2014), and engage in barking (Wells & Hepper, 1992) are less
U
attractive to potential adopters. These behaviors may be indicative of stress (e.g., dogs in social
N
isolation had higher locomotion; Hetts et al., 1992) or reactions to a novel environment (activity
A
was found to decrease across time spent in the kennel, Rooney, Gaines, & Bradshaw, 2007;
M
Titulaer, Blackwell, Mendl, & Casey, 2013; but see Protopopova et al., 2014). Furthermore, dogs
D
that are perceived as overly active may further be avoided; 27% of people who decided not to
TE
adopt a dog at a shelter claimed it was because the dog was too active (Protopopova & Wynne, 2014). Since potential adopters take an average of only 20-70 s to look at a kenneled dog prior to
EP
their decision, inappropriate in-kennel behaviors may pose a serious risk of adversely affecting
CC
adoption rates (Wells & Hepper, 2001). Therefore, improving in-kennel behavior of shelter dogs
A
may serve to improve their likelihood of adoption. One common-sense approach to reducing over-activity of shelter dogs is providing
supplemental exercise. Several organizations within the US recommend exercise to maintain wellbeing and behavioral health for kenneled dogs (ASPCA, “Position Statement on Euthanasia”, 2015b; Newbury et al., 2010) and the US Animal Welfare Act (7 U.S.C., 2143 section 13[a][2][B]) specifically addresses exercise requirements for laboratory-housed dogs.
EXERCISE OR CALM INTERACTION
4
Previous research suggests that dogs receiving exercise, play, and human interaction had decreased stress and improved some behaviors in a kenneled environment (Bergamasco, Osella, Savarino, Larosa, Ozella et al., 2010; Coppola, Grandin, & Enns, 2006; Normando, Corain,
SC RI PT
Salvadoretti, Meers, & Valsecchi, 2009; Shiverdecker, Schiml, & Hennessy, 2013). However, research by Hetts et al. (1992) and Campbell, Hughes, Griffin, Landi, and Mallon (1988) found that exercise did not improve behavior in kenneled dogs. In fact, Normando et al. (2009) and
Clark, Rager, Crowell-Davis, and Evans (1997) found that additional exercise actually increased active behaviors in dogs and Clark et al. (1997) found that it also did not prevent the
U
development of abnormal in-kennel behaviors.
N
An alternate point of view rejects additional exercise as an effective treatment for over-
A
activity in kenneled dogs and suggests that calm interactions between volunteers and shelter dogs
M
may result in calmer dogs (e.g., ASPCA, “Enrichment in the Shelter”, 2015c; Sternberg, n.d.).
D
This intervention is based on the premise that dogs are able to learn, through Pavlovian
TE
conditioning, that people are a predictor of excitement. If staff and volunteers only appear in front of the dog’s kennel to take the dog for an exercise session, the visual presentation of a
EP
person may become a conditioned stimulus eliciting excitement (a conditioned response). In
CC
fact, seeing a person resulted in higher activity and approach behavior in kenneled dogs (Arhant & Troxler, 2014; Wells & Hepper, 2000). When visitors were prevented from visiting kenneled
A
shelter dogs, dogs spent more time resting, and less time barking and engaging in stereotypic behaviors (Hewison, Wright, Zulch, & Ellis, 2014). Thus, providing additional calm interactions between people and dogs, might result in the extinction of the conditioned excitable response and subsequently in calmer dogs.
EXERCISE OR CALM INTERACTION
5
These two interventions, additional exercise or calm interactions, are particularly interesting because they directly oppose each other in methodology, yet promise the same outcome. We assessed how these interventions affect behavior immediately before and after the
SC RI PT
sessions; we predicted that, due to learning effects throughout the calm interaction sessions, dogs would appear calm when they first see the experimenter, or right before the treatment sessions. However, due to the same learning effects, dogs may engage in inappropriate behavior right before the sessions during the additional exercise treatment. Alternatively, dogs may appear
calmer directly following the additional exercise treatments due to fatigue, but this effect would
U
not be observed after calm interventions. Therefore, the aim of this study was to compare two
N
opposing, yet currently popular, interventions for counteracting hyperactivity and undesirable in-
A
kennel behavior in dogs housed in shelters.
TE
Animals and housing
D
M
Materials and Methods
Sixteen dogs at the XXX Humane Society (location of Humane Society removed),
EP
located in the United States, were enrolled into the study. All dogs in the shelter at the start of the study were enrolled. Dogs were included in the study regardless of physical characteristics such
CC
as breed, age, or in-kennel behavior. However, dogs that the staff deemed dangerous and/or
A
unpredictable were excluded from the study due to safety concerns. Dogs were also excluded from the study if they were adopted out during the study. Dogs were housed in rows of adjacent 2-sided, indoor kennels with a guillotine door between each side, and glass doors. Each kennel was 1.2 m wide x 2.4 m long x 3.0 m high. The floor material was epoxy over cement. Dogs were housed mostly singly (with 3 dogs housed in
EXERCISE OR CALM INTERACTION
6
pairs). The dogs could be viewed by the public from either side of the run when the guillotine door was open. The standard adoption fee for an adult dog was 100 USD. Each kennel contained a water dish and food dish that hung on the doors as well as a Kuranda bed (Kuranda USA,
SC RI PT
Annapolis, MD, USA) on one side of the kennel. Staff fed the dogs and cleaned kennels daily before 11:00 h. Volunteers at the shelter exercised, trained, and played with the dogs
approximately once per day on the shelter premises. Each kennel card noted the dog’s name and sometimes, a few words on the dog’s personality as perceived informally by shelter staff and
volunteers. Dogs were marketed by the shelter staff and volunteers on their website and through
U
popular online social networking sites.
A
N
Procedure
M
In a counterbalanced mixed-group design, dogs were randomly assigned into the calm interaction or exercise condition (between-subjects factor). The dogs in each condition were
D
further randomly assigned to experience either the baseline or the treatment phase first (within-
TE
subjects factor). Thus, four dogs experienced 2 weeks of the baseline phase followed by 2 weeks
EP
of an exercise phase, four dogs experienced 2 weeks of the exercise phase followed by 2 weeks of the baseline phase, four dogs experienced 2 weeks of the baseline phase followed by 2 weeks
CC
of a calm interaction phase, and four dogs experienced 2 weeks of the calm interaction phase
A
followed by 2 weeks of the baseline phase. Because we aimed to assess the effect of both treatment conditions just prior to treatment
and immediately after treatment, the in-kennel behavior of the dogs was sampled twice during each session in the experimental conditions: once right before the treatment session and once right after the treatment session. Every day, an experimenter conducted an assessment of each
EXERCISE OR CALM INTERACTION
7
dog’s in-kennel behavior prior to the treatment session, conducted the treatment session, and conducted an assessment of in-kennel behavior after the treatment session. If a dog was in the baseline phase of the experiment, the experimenter conducted only a single in-kennel assessment
SC RI PT
daily. Staff and volunteers were aware of the general purposes of the study, but were blind to
the specific conditions and condition assignment of each dog. Also, the staff and volunteers were asked to remain consistent with their routines for the dogs in our study, including the number of
U
times they were walked each day.
N
Intervention conditions
A
The exercise intervention consisted of daily 15 min periods of active toy play and/or
M
running on a leash with the experimenter, whichever the dog appeared to prefer. These sessions
D
were conducted between 9:00 and 18:00. The exercise took place in an indoor playpen located
TE
near the dog kennels or in an outdoor yard. The toys offered to the dogs included Frisbees, tennis balls, squeaky plastic balls, or a ball at the end of a flirt pole. The leashes used while running the
EP
dogs were 1.2 m long slip leads provided by the shelter.
CC
The calm interaction intervention consisted of leading the dog to a quiet indoor area for 15 min daily. These sessions were conducted between 9:00 and 18:00. During this time, the
A
experimenter sat in a chair, read out loud in a calm tone and normal volume, while ignoring the dog. The experimenter took care not to speak to the dog, engage in petting, or in any other way provide attention to the dog. In-kennel assessment
EXERCISE OR CALM INTERACTION
8
Each day, just prior to and just after the experimental condition (or once daily during the baseline condition), the dogs were videotaped for 30 s with the dogs inside their kennels and the experimenter directly outside. The experimenter approached the kennel casually, as a potential
SC RI PT
adopter would, and recorded the dog using a small handheld camera or phone with video capability.
The videos were analyzed using partial 5-s interval recording method by additional experimenters to record the different behaviors that the dogs exhibited (Table 1). These
behaviors were chosen to capture the activity levels of the dogs and to monitor their levels of
U
adoptability. Prior research has shown that the location of the dog in the kennel, face orientation,
N
back and forth motion in the kennel, and leaning or rubbing of the body on kennel walls predicts
A
length of stay in the shelter (Protopopova et al., 2014). Barking has been shown to affect health
M
outcomes in shelter dogs and shelter staff (Coppola et al., 1996; Sales et al., 1997) as well as
D
having been implicated in affecting adoption rates (Wells & Hepper, 2001). Barking was
TE
therefore aggregated with the other identified undesirable behaviors, into one overall measure of
EP
“undesirable behavior”. The ethogram with operational definitions appear in Table 1. ---Table 1 here---
CC
Statistical Analysis
A
Multivariate regression models with treatment as the independent variable (IV) and all
behaviors (except “undesirable behavior”) entered into the model as dependent variables (DV) were conducted for each condition (exercise and calm interaction) and time point (before and after treatment). The effects of condition (IV) and time point (IV) on the aggregate measure of “undesirable behavior” (DV) were further explored with paired-samples t-tests. The impact of
EXERCISE OR CALM INTERACTION
9
the order of baseline and treatment (IV) on all individual behaviors (seven DVs) were analyzed using separate multivariate regression models and descriptive statistics. Statistical significance was determined at p-value less than .05.
SC RI PT
Inter-observer agreement (IOA) was calculated by dividing the number of intervals in which an agreement occurred (both observers agreed whether the behavior occurred or not) by
the sum of agreements and disagreements and converting the result to a percentage in each video. The percentages were then averaged across videos and reported individually by each behavior.
U
Ethical Statement
N
All procedures were approved by the University of Florida Institutional Animal Care and
M
A
Use Committee.
TE
Descriptive Statistics
D
Results
Sixteen dogs completed the study with eight dogs experiencing the calm condition and
EP
eight experiencing the additional exercise with the experimenter condition. Within each group,
CC
half (n = 4), experienced the baseline first and the remaining half experienced the treatment first.
A
Table 2 lists the sex, approximate age, breed, and length of stay at the shelter of each dog. ---Table 2 here--In the large majority of observation periods collected during baseline and just prior to both interventions, the dogs remained facing forward and in the front of the kennel (Table 3). In many observation periods dogs were recorded engaging in back and forth motion, jumping on
EXERCISE OR CALM INTERACTION
10
the kennel door, and barking. Very few observation periods saw dogs jumping straight up in the kennel and rubbing their bodies on the kennel walls.
SC RI PT
---Table 3 here--Exercise Just prior to treatment
A multivariate regression model was conducted with treatment as the independent
U
variable and facing forward, being in the front of the kennel, engaging in back and forth motion,
N
jumping on the kennel door, jumping straight up in kennel, rubbing the body on the wall, and
A
barking as the dependent variables. The model revealed that only face forward and back and
M
forth motion contributed significantly (p < .05) and thus were retained in the final model (F [2, 216] = 7.58, p = .0007). Table 4 lists the predictors. Additional exercise increased the proportion
D
of time the dogs spent facing forward when a person approached (baseline: mean = .88, SD =
TE
.25; treatment: mean = .96, SD = .13) and decreased the proportion of time walking back and
EP
forth in the kennel (baseline: mean = .10, SD = .23; treatment: mean = .04, SD = .10).
CC
---Table 4 here ---
However, receiving exercise did not alter the rate of undesirable behavior in general (as
A
defined above: engaging in at least one of a) Front of kennel not noted, b) Facing forward not noted; c) Back and forth motion noted, d) Barking noted, or e) Leaning/ rubbing body on kennel wall noted; t = 0.06, df = 7, p > .05). The dogs engaged in undesirable behavior on average .40 (SD = .04) proportion of time in baseline and .39 (SD = .04) in treatment.
EXERCISE OR CALM INTERACTION
11
Immediately after treatment An additional multivariate regression model containing all seven DVs revealed that back and forth motion, jumping on door, and wall rubbing were the only variables which contributed
SC RI PT
significantly (p < .05) to the model and thus were retained (F [3, 216] = 7.77, p = .0001). Table 5 lists the predictors and all values. Additional exercise increased back and forth motion of the
dogs (baseline: mean = .10, SD = .23; treatment: mean = .17, SD = .24), decreased jumping on
the door (baseline; mean = .05, SD = .12; treatment; mean = .004, SD = .04), and eliminated wall
U
rubbing (baseline; mean = .04, SD = .18; treatment: behavior never observed).
N
---Table 5 here---
A
Receiving exercise did not alter the rate of undesirable behavior (t = 0.26, df = 7, p >
M
.05). The average proportion of time that the dogs engaged in undesirable behavior in baseline
TE
Calm Interaction
D
was .40 (SD = .04) and after exercise was .37 (SD = .04).
EP
Just prior to treatment
CC
A multivariate regression model, identical to the model applied for the exercise treatment, revealed that no behavioral variables, out of the seven, were influenced by the calm interaction
A
(all p > .05).
A two-tail paired t-test found no statistically significant effect of calm interaction on the
rate of undesirable behavior (t = 2.10, df = 7, p > .05). However, it is interesting to note that a one-tail t-test revealed that receiving calm interactions reduced the rate of undesirable behavior (t
EXERCISE OR CALM INTERACTION
12
= 2.10, df = 7, p = .04, Cohen's d = 3.64; Figure 1). The average proportion of time that the dogs engaged in undesirable behavior in baseline was .36 (SD = .04) and before calm interactions was .28 (SD = .07).
SC RI PT
---Figure 1 here--Immediately after treatment
An additional multivariate regression model containing seven DVs revealed that facing forward, front of kennel, back and forth motion, jumping on door, and barking contributed
U
significantly (p < .05) and thus were retained in the final model (F [5, 216] = 10.1, p < .0001).
N
Table 5 lists the relevant variables. Calm interaction resulted in a decrease in facing forward
A
(baseline: mean = .93, SD = .18; treatment: mean = .82, SD = .31), being in the front of the
M
kennel (baseline: mean = .91, SD = .24; treatment: mean = .87, SD = .27), jumping on kennel
D
door (baseline: mean = .08, SD = .17; treatment: mean = .006, SD = .04), and barking (baseline:
TE
mean = .22, SD = .32; treatment: mean = .05, SD = .16), and an increase in back and forth motion (baseline: mean = .07, SD = .16; treatment: mean = .16, SD = .21).
EP
---Table 6 here---
CC
Receiving calm interactions did not alter the rate of undesirable behavior in general
immediately after treatment (t = 0.58, df = 7, p > .05). The average proportion of time that the
A
dogs engaged in undesirable behavior in baseline was .36 (SD = .04) and .33 (SD = .03) after treatment. Order of Treatment
EXERCISE OR CALM INTERACTION
13
An additional multivariate regression model containing seven DVs revealed that several behaviors were influenced by the order of the treatment conditions. Dogs which received the baseline condition followed by the calm interaction intervention decreased their time in the front
SC RI PT
of kennel prior to the calm interaction (baseline: mean = .99, SD = .05; treatment: mean = .94, SD = .23), but dogs which received the calm interaction first showed an increase in front of
kennel behavior prior to the treatment (baseline: mean = .85, SD = .32; treatment: mean = .92,
SD = .22; t = -2.53,df = 219, p = .012). Furthermore, dogs which received the calm interaction first showed a larger increase in back and forth motion after the calm interaction treatment
U
(baseline: mean = .08, SD = .18; treatment: mean = .22, SD = .24 and baseline: mean = .05, SD =
N
.13; treatment: mean = .09, SD = .15 respectively; t = 3.35, df = 219, p = .001).
A
Dogs which received the baseline first, increased their jumping on the door prior to
M
additional exercise, but the dogs which received treatment first, decreased their jumping prior to
D
exercise (baseline: mean = .006, SD = .03; treatment: mean = .02, SD = .06 and baseline: mean =
TE
.09, SD = .16; treatment: mean = .04, SD = .08 respectively; t = 4.42, df = 216, p < .001). Jumping in the kennel was never seen prior to exercise when the treatment came first (t = -2.48,
EP
df = 216, p = .014), but rubbing the body on the wall was likewise never seen in dogs prior to
CC
exercise which experienced the baseline first (t = 3.79, df = 216, p < .001). Back and forth motion and jumping on the kennel door after the exercise condition were
A
at overall higher levels for dogs which received the exercise first rather than the baseline. The dogs that began in treatment exhibited a higher overall proportion of back and forth motion (mean = .18, SD = .27) compared to those that started in the baseline condition (mean = .09, SD = .18; t = 3.30, df = 216, p = .001). Similarly, the dogs that began in treatment exhibited a higher
EXERCISE OR CALM INTERACTION
14
overall proportion of jumping on the door (mean = .05, SD = .12) compared to the dogs that started in the baseline condition (mean = .004, SD = .03; t = 4.02, df = 216, p < .001).
SC RI PT
Inter-Observer Agreement Approximately a tenth of randomly selected videos (81 out of 725) were coded by an
additional observer. The average IOA was 96.2% (SD = 4.2%). The minimum average IOA for any one behavior was 88.3% (aggregate undesirable behavior) and the maximum was 99.8% (jumping in kennel).
N
U
Discussion
A
Both interventions had complex effects on the behaviors of the dogs, supporting some of
M
our hypotheses but not others. Dogs that received additional exercise spent more time facing forward and less time in back and forth motion just prior to treatment. These findings contradict
D
the hypothesis that providing additional exercise results in increased hyperactive behavior when
TE
a person is first seen by the dog. Instead, dogs were alert and stationary when they saw a person, who was previously paired with exercise. However, dogs did not seem fatigued after the
EP
exercise. Right after receiving additional exercise, dogs increased their back and forth motion in
CC
the kennel. Jumping on the door and wall rubbing decreased, but this decrease was largely driven by an order effect: only dogs which received treatment prior to baseline showed a decrease in
A
jumping on the kennel. Differences in rubbing on the kennel wall were likewise driven by an order effect: only dogs which received baseline first showed a decrease in this behavior. These findings suggest either that exercise does not function to decrease active behaviors of the dogs or that our exercise sessions were too brief to result in fatigue. It is possible that a longer exercise session would result in calmer behavior of the dogs; alternatively, fatigue from extensive
EXERCISE OR CALM INTERACTION
15
physical activity could produce undesirable behaviors, such as appearing lethargic or uninterested. It would be useful for future research to assess varying lengths of exercise sessions to find an optimal duration to elicit quieter behavior in the kennel without exhausting the dogs.
SC RI PT
Our data corresponds in part with previous research. Normando et al. (2009) found that a combination of 15 min of petting and play weekly served to increase locomotion and time spent in sight of kenneled dogs. Normando et al. (2009) reported that this increase in locomotion and
staying in sight was evident both just prior to the interaction as well as immediately afterwards. Perhaps the difference in our data was due to differences in our exercise components and the
U
amount of interaction; Normando et al. (2009) combined petting and play and administered the
N
treatment weekly, whereas our intervention included play and on-leash running and was
A
conducted daily. Hetts et al. (1992) did not detect changes in in-kennel behavior in laboratory
M
dogs that experienced a forced exercise on a treadmill intervention compared to dogs that were not exercised. Furthermore, Campbell et al. (1998) also did not find differences in in-kennel
D
behavior in laboratory dogs that were exercised through access to a larger enclosure nor through
TE
access to conspecifics, compared to dogs that were not exercised. Clark et al. (1997) found that
EP
dogs barked in the kennels more when exercised with conspecifics as compared to dogs that did not receive exercise. These results may be substantially different to our findings as no person
CC
was specifically paired with the exercise component. Dogs that received additional calm interaction displayed less undesirable behavior in
A
general, but did not show differences in any individual behaviors, just prior to calm interaction. Due to our small sample size, we may not have detected the influence of calm interactions on individual behaviors, but we did find an effect when these behaviors were combined. These findings support the learning hypothesis that pairing the visual presentation of a person with
EXERCISE OR CALM INTERACTION
16
calm interactions serves to increase calm, desirable behavior in these dogs. Furthermore, order effects played a role: dogs which received the calm interaction first and baseline second, spent more time in the front of the kennel, whereas the dogs that received the baseline first spent less
SC RI PT
time in the front of the kennel. It is therefore possible that time in front of the kennel was related to the novelty of the experimenter. However, we found that following the calm interaction
intervention, the dogs behaved in a mix of desirable and undesirable ways. Increased undesirable behaviors included spending less time in the front of the kennel, less time facing forward, and more time engaging in back and forth motion. Decreased undesirable behaviors included less
U
time spent barking and less time jumping on the door. Taken together, it may be inferred that the
N
dogs perceived the person as uninteresting right after the intervention and thus were reacting
A
asocially.
M
Whilst we did not record the dog’s behavior during the interactions, anecdotally, dogs in the calm interaction condition spent a large portion of their time laying down. Shiverdecker et al.
D
(2013) found that dogs spent less time barking during passive interactions with a person, in
TE
which a person sat quietly in a chair and ignored the dog. However, they did not find a
EP
statistically different proportion of time spent laying down as compared to other conditions in which an experimenter played with the dogs, or in which the dogs remained isolated in a separate
CC
room. One difference is that the dogs in our study were restrained with a leash, whereas the dogs in the Shiverdecker et al. study were left unrestrained in an enclosure. As previous research has
A
found that laying down next to potential adopters increases the likelihood of adoption (Protopopova & Wynne, 2014), future research can directly evaluate the effect of leash-restraint on out-of-kennel behavior.
EXERCISE OR CALM INTERACTION
17
To summarize, we found that dogs displayed appropriate behavior when anticipating either calm interaction or additional exercise. However, right after either of the interventions, the dogs displayed undesirable behaviors. Immediately after exercise, the dogs exhibited high
SC RI PT
locomotion in their kennels, and immediately after calm interactions, the dogs exhibited behaviors that may be interpreted as asocial (e.g., spent less time in the front of the kennel, less time facing forward, not barking). In order to increase adoptability, our results suggest that it is best to administer both or either of these interventions at the end of the day when all adopters
have already left the shelter. If a shelter has a “quiet hour” in which no adopters are allowed to
U
visit and the dogs are encouraged to lie down quietly in their kennels, it may be advisable to
N
administer the calm interactions right before this, as this intervention seems to promote quiet and
A
asocial behavior. Additionally, it may be best to tailor interventions to specific dogs. For
M
example, the dogs that engage in high rates of barking, the calm interaction may serve to reduce this unwanted behavior. However, for dogs that do not exhibit excessive barking, but instead
D
engage in high levels of wall rubbing, additional exercise may be warranted. These data may be
TE
an important first step in developing tailored interventions to individual dogs in the shelter.
EP
An alternative interpretation of our results is that the two conditions were not sufficiently different from each other to provoke different behavioral responses from the dogs. It is possible
CC
that the relatively similar changes in behavior in response to the two conditions were in response simply to being taken out of the kennel, and not related to the different interactions that occurred
A
afterwards.
Overall, the dogs, regardless of the condition, spent practically all of their time facing
forward and being in the front of the kennel during the observations. Dogs barked in a third of the observations and engaged in barking behavior for approximately a fifth of the observation
EXERCISE OR CALM INTERACTION
18
time. Jumping on the door and engaging in back and forth motion was recorded in a fifth of the observations, but dogs only spent on average 6% and 7% of their time engaging in these behaviors respectively. Leaning and rubbing the body on the enclosure wall and jumping straight
SC RI PT
up in the kennel were rarely seen. Our results suggest that whereas increasing appropriate behavior (e.g., increasing time spent in the front of the kennel and facing forward) may not
provide significant effects on adoptability, as all dogs are engaging in these behaviors frequently and for long durations already, decreasing inappropriate behavior (e.g., time spent in the back of the kennel, facing backwards, back and forth motion, etc.) may serve to improve adoption
U
outcomes. Previous research suggests that visitors pay attention more to negative behaviors than
N
to positive ones (Protopopova et al., 2014).
A
One limitation of this study was that the dogs’ behavior was only recorded for a short
M
period of time. This length of time (30 s) was chosen based on previous research that suggested that adopters only observe the behavior of the dogs for less than 1 min and interact with kenneled
D
dogs on average for only 20 s (Wells & Hepper, 2001). It is possible that the different
TE
enrichment strategies resulted in differences in behavior over longer time periods at the shelter.
EP
Future research could examine the effect of various enrichments on longer periods of in-kennel behavior. Furthermore, all of the behavioral assessments were conducted in response to a
CC
familiar person standing in front of the kennel. As potential adopters are unfamiliar to the dogs, future research should investigate the effects of various human interactions on in-kennel
A
behavior in response to seeing unfamiliar people. Another methodological limitation was the limited sample size and the non-selective
nature of the dogs. Because we required the dogs to stay at the shelter for four weeks in order to include their data, all dogs regardless of breed, size, and temperament were enrolled in the study
EXERCISE OR CALM INTERACTION
19
in order to reach a sufficient sample size. To control for individual differences among dogs, we used a within-subject comparison to determine a treatment effect in each condition. We did not compare calm interactions and exercise to each other directly as different dogs experienced the
SC RI PT
two different conditions. Because we wanted to assess the learning effects that may have occurred when a dog was exposed to the two conditions, we needed to provide each treatment
repeatedly and for a sufficient number of days. And given that the average length of stay at the
shelter used was relatively short, we were not able to subject each dog to both conditions. Thus, our current methodology involved a mixed design; a future within-subject comparison of
U
conditions in a shelter that houses dogs for a much longer period of time may be interesting in
N
order to compare the conditions more directly as well as assess any individual differences. There
A
is increasing evidence that dogs, like other animals, display differences in coping with stressors
M
in their environment. Several studies have now shown that dogs display individual differences, with dogs exhibiting a “proactive” temperament (Koolaas et al., 1999) not showing a significant
D
change in cortisol levels in response to environmental stressors (Horváth, Igyártó, Magyar, &
TE
Miklósi, 2007; Wood, de Bie, & Clarke, 2014).
EP
Another limitation was that we did not utilize the commonly deployed method of calm interaction currently performed in shelters. The more typical method involves reading to the dog
CC
inside their own kennel. The dogs in the present study were housed in kennels that could only be opened from the outside, so for the safety of the experimenters and research assistants, calm
A
interactions were conducted in another room. Whereas the results presented provide empirical support for one type of calm interaction procedure, future research should investigate the differences across commonly used calm interactions and compare the relative effectiveness of each.
EXERCISE OR CALM INTERACTION
20 Conclusion
Our data suggests that both kinds of intervention, calm interactions and additional exercise, provide some benefits, but also result in some undesirable behaviors. Both
SC RI PT
interventions decrease undesirable behaviors just prior to the sessions, but both also result in undesirable behaviors immediately after the dogs were taken back to their kennels. Thus, it may be possible to harness the benefits by administering either or both interventions at the end of the day so that no potential adopters will be dissuaded by the dog’s undesirable behavior after the
enrichment sessions. Alternatively, animal shelters may decide which behaviors they would like
U
to see in their population of dogs and provide these enrichment strategies accordingly.
N
Acknowledgements: This research was supported in part by a grant from the Humane Society
A
Veterinary Medical Association and Merial Veterinary Scholars Program. The authors thank
M
Kelly Cathey, Avi Hershkowitz, Justyna Resztak, Jennifer Taylor, Jessica Vondran, Courtney
D
Alexander, Kaila Ames, Rachel Bradley, Lauren Burstein, Sarah Weinsztok, Nancy Ordax,
TE
Jonathan Pruitt, Jennifer Higgins, Devin Bogart, Estefania Junco, Austin Folger, and Jessica Jeong for assisting in data collection. The authors also thank the staff and the dogs of Alachua
A
CC
EP
County Humane Society.
References
The American Society for the Prevention of Cruelty to Animals (2015a). Pet statistics. Available at: https://www.aspca.org/about-us/faq/pet-statistics. Accessed October 21, 2015.
EXERCISE OR CALM INTERACTION
21
The American Society for the Prevention of Cruelty to Animals (2015b). Position Statement on Euthanasia. Available at: https://www.aspca.org/about-us/aspca-policy-and-positionstatements/position-statement-on-euthanasia. Accessed October 21, 2015.
SC RI PT
The American Society for the Prevention of Cruelty to Animals (2015c). Read and Relax (R&R) Kennel Enrichment Program. Available at:
http://www.aspcapro.org/sites/default/files/mym_enrichment_r_r_program.pdf. Accessed October 21, 2015.
U
Arhant, C., & Troxler, J. (2014). Approach behaviour of shelter dogs and its relationships with
N
the attitudes of shelter staff to dogs. Applied Animal Behaviour Science, 160, 116-126.
A
Bergamasco, L., Osella, M.C., Savarino, P., Larosa, G., Ozella, L., Manassero, M., Badino, P.,
M
Odore, R., Barbero, R., Re, G., 2010. Heart rate variability and saliva cortisol assessment
D
in shelter dog: human-animal interaction effects. Appl. Anim. Behav. Sci. 125, 56–68.
TE
Clark, J. D., Rager, D. R., Crowell-Davis, S., & Evans, D. L. (1997). Housing and exercise of dogs: effects on behavior, immune function, and cortisol concentration. Comparative
EP
Medicine, 47(5), 500-510.
CC
Campbell, S. A., Hughes, H. C., Griffin, H. E., Landi, M. S., & Mallon, F. M. (1988). Some
A
effects of limited exercise on purpose-bred beagles. American journal of veterinary research, 49(8), 1298-1301.
Coppola, C. L., Enns, R. M., & Grandin, T. (2006). Noise in the animal shelter environment: Building design and the effects of daily noise exposure. Journal of Applied Animal Welfare Science, 9, 1–7.
EXERCISE OR CALM INTERACTION
22
Hiby, E. F., Rooney, N. J., & Bradshaw, J. W. (2006). Behavioural and physiological responses of dogs entering re-homing kennels. Physiology & behavior, 89(3), 385-391. Hennessy, M. B., Davis, H. N., Williams, M. T., Mellott, C., & Douglas, C. W. (1997). Plasma
SC RI PT
cortisol levels of dogs at a county animal shelter. Physiology & Behavior, 62(3), 485-490. Hennessy, M. B., Williams, M. T., Miller, D. D., Douglas, C. W., & Voith, V. L. (1998).
Influence of male and female petters on plasma cortisol and behaviour: can human interaction reduce the stress of dogs in a public animal shelter? Applied Animal
U
Behaviour Science, 61(1), 63-77.
N
Hetts S, Clark JD, Calpin JP, Arnold CE, Mateo JM (1992). Influence of housing conditions on
M
A
beagle behaviour. Appl Anim Beh Sci 34: 137–155.
Hewison, L. F., Wright, H. F., Zulch, H. E., & Ellis, S. L. (2014). Short term consequences of
D
preventing visitor access to kennels on noise and the behaviour and physiology of dogs
TE
housed in a rescue shelter. Physiology & Behavior, 133, 1-7. Horváth, Z., Igyártó, B. Z., Magyar, A., & Miklósi, Á. (2007). Three different coping styles in
EP
police dogs exposed to a short-term challenge. Hormones and Behavior, 52(5), 621-630.
CC
Luescher, A., Medlock, R., (2009). The effects of training and environment alterations on
A
adoption success of shelter dogs. Appl. Anim. Behav. Sci. 117, 63–68.
Newbury, S., Blinn, M. K., Bushby, P. A., Cox, C. B, Dinnage, J. D. et al. (2010). “Guidelines for Standards of Care in Animal Shelters”. The Association of Shelter Veterinarians. Available at: http://www.sheltervet.org/assets/docs/shelter-standards-oct2011wforward.pdf. Accessed on October 21, 2015.
EXERCISE OR CALM INTERACTION
23
Normando, S., Corain, L., Salvadoretti, M., Meers, L., & Valsecchi, P. (2009). Effects of an Enhanced Human Interaction Program on shelter dogs’ behaviour analysed using a novel nonparametric test. Applied Animal Behaviour Science, 116(2), 211-219.
SC RI PT
Protopopova, A., Mehrkam, L. R., Boggess, M. M., & Wynne, C. D. L. (2014). In-kennel behavior predicts length of stay in shelter dogs. PLoS ONE, 9, e114319. doi: 10.1371/journal.pone.0114319
Protopopova, A., & Wynne, C. D. L. (2014). Adopter-dog interactions at the shelter: Behavioral
U
and contextual predictors of adoption. Applied Animal Behaviour Science, 157, 109-116.
N
Rooney, N. J., Gaines, S. A., & Bradshaw, J. W. (2007). Behavioural and glucocorticoid
A
responses of dogs (Canis familiaris) to kennelling: investigating mitigation of stress by
M
prior habituation. Physiology and Behavior, 92(5), 847-854.
D
Sales, G., Hubrecht, R., Peyvandi, A., Milligan, S., & Shield, B. (1997). Noise in dog kennelling:
TE
Is barking a welfare problem for dogs? Applied Animal Behaviour Science, 52, 321–329.
EP
doi: 10.1016/S0168-1591(96)01132-X Shiverdecker, M. D., Schiml, P. A., & Hennessy, M. B. (2013). Human interaction moderates
CC
plasma cortisol and behavioral responses of dogs to shelter housing. Physiology &
A
Behavior, 109, 75-79.
Stephen, J. M., & Ledger, R. A. (2006). A longitudinal evaluation of urinary cortisol in kennelled dogs, Canis familiaris. Physiology & Behavior, 87(5), 911-916.
EXERCISE OR CALM INTERACTION
24
Sternber, S. (n.d.). “The Deterioration of a Shelter Dog”. Association of Professional Dog Trainers Newsletter. Available at: https://www.petfinder.com/pro/forshelters/deterioration-shelter-dog/. Accessed on October 21, 2015.
SC RI PT
Titulaer, M., Blackwell, E. J., Mendl, M., & Casey, R. A. (2013). Cross sectional study comparing behavioural, cognitive and physiological indicators of welfare between short and long term kennelled domestic dogs. Applied Animal Behaviour Science, 147(1), 149158.
U
Weiss, E., Miller, K., Mohan-Gibbons, H., Vela, C., (2012). Why did you choose this pet?
N
Adopters and pet selection preferences in five animal shelters in the United States.
A
Animals, 2, 1–17.
M
Wells, D.L., Hepper, P.G., (1992). The behaviour of dogs in a rescue shelter. Animal Welfare, 1,
D
171–186.
TE
Wells, D. L., & Hepper, P. G. (2000). The influence of environmental change on the behaviour
EP
of sheltered dogs. Applied Animal Behaviour Science, 68, 151–162. Wells, D.L., Hepper, P.G., (2001). The behavior of visitors towards dogs housed in an animal
CC
rescue shelter. Anthrozoos, 14, 12–18.
A
Wood, P. A., de Bie, J., & Clarke, J. A. (2014). Behavioural and physiological responses of domestic dogs (Canis familiaris) to agonistic growls from conspecifics. Applied Animal Behaviour Science, 161, 105-112.
Figures
EXERCISE OR CALM INTERACTION
25
Figure 1. The average proportion of time spent in undesirable behavior in general in baseline and
D
M
A
N
U
SC RI PT
prior to the calm/exercise interaction treatments. Means and individual data are shown.
TE
Table 1. Ethogram of all in-kennel behaviors. Behavior
Located between front of cage, and up to and including the midpoint of kennel
CC
EP
Front of kennel
Operational definition
Head oriented such that the observer is able to
Back and forth motion
Distance between the dog and the observer is
A
Facing forward
see more than the side profile of face
either increased, decreased, or both. This change in the distance to the observer occurs at
EXERCISE OR CALM INTERACTION
26 least once (in either direction) during the observation interval.
Jumping on door
Dog on its hind legs with at least one paw on
Jumping in kennel
SC RI PT
the kennel door Dog on its hind legs or jumps off the floor, but not touching the kennel door Barking
Vocalization of very short duration and a low pitch
Prolonged (> .1 sec) contact with the cage wall
U
Leaning/ rubbing body on kennel wall
Scored when at least one of following criteria
A
Undesirable behavior (aggregate measure)
N
by pushing side of body against the cage wall
A
CC
EP
TE
D
M
is satisfied: a)
Front of kennel not noted
b)
Facing forward not noted
c)
Back and forth motion noted
d)
Barking noted
e)
Leaning/ rubbing body on
kennel wall noted
Table 2. Age, breed, and length of stay at the shelter prior to the beginning of the study of
the dogs in both conditions. Age and breed are approximations determined by shelter staff. Name
Approximate age
Breed
Length of
Condition (Calm
stay (days)
interaction [C] or
EXERCISE OR CALM INTERACTION
27 Exercise [E]) and order of Baseline (B)
4 y, 2 m
American Pitbull
54
Terrier mix 3 y, 1 m
Bulldog mix
1
Casey
1 y, 11 m
Retriever mix
275
Gus
1 y 2 mo
Terrier/Pitbull mix
150
Helen
2 y, 1 m
Terrier/Pitbull mix
Maya
2y4m
Mia
2 y, 0 m
Catahoula Leopard
BC EB EB
CB
Terrier/Pitbull mix
4
CB
1
CB
1
BE
77
BC
1
EB
D
A
N
1
M
U
Brandy
BE
SC RI PT
Arnold
TE
Dog mix
A
CC
Norman
4 y, 0 m
EP
Mildred
Pamela
4 y, 0 m
American Bulldog/Boxer mix American Pitbull Terrier mix
2 y, 1 m
Terrier/Retriever mix
EXERCISE OR CALM INTERACTION
Porscha
7 y, 1 m
28
Catahoula Leopard
242
BC
30
BC
Rodeo
2 y, 1 m
American Pitbull Terrier mix
Rosco
3 y, 0 m
Terrier/Pitbull mix
1
Samuel
1 y, 3 m
American Pitbull
82
2 y, 3 m
American Pitbull
BE
BE
1
EB
4 y, 0 m
Hound mix
D
Troy
M
A
Terrier mix
CB
12
N
Simba
U
Terrier mix
SC RI PT
Dog/Boxer mix
Mean proportion of
observations in which
observation period
behavior was exhibited
devoted to behavior
Facing forward
1.00
0.93
0.18
Front of kennel
0.96
0.93
0.22
A
Proportion of
CC
EP
both interventions.
TE
Table 3. The incidence of behaviors collected during control sessions and just prior to
SD
EXERCISE OR CALM INTERACTION
29
0.33
0.20
0.33
Jumping on door
0.20
0.06
0.15
Back and forth motion
0.20
0.07
0.16
Leaning/ rubbing body on
0.04
0.03
0.15
0.02
0.00
0.03
SC RI PT
Barking
kennel wall
U
Jumping in kennel
N
Table 4. Output of the multivariate reduced regression model on the effects of exercise on
A
the behavior of kenneled dogs prior to the treatment condition. The coefficient (B), standard
M
error (S.E.), t statistic (t), P-value (P), and the confidence interval (C.I.) for each variable are
D
shown below.
t
P
0.16
2.65
.009
0.11
0.74
-0.50
0.19
-2.65
.009
-0.87
-0.13
0.15
0.15
0.98
0.33
-0.15
0.45
forwards
CC
Back and
0.43
EP
Facing
S.E.
TE
B
95% C.I.
A
forth motion Constant
Table 5. Output of the multivariate reduced regression model on the effects of exercise on the behavior of kenneled dogs after the treatment condition. The coefficient (B), standard error
EXERCISE OR CALM INTERACTION
30
(S.E.), t statistic (t), P-value (P), and the confidence interval (C.I.) for each variable are shown
t
P
0.30
0.14
2.10
.037
-1.28
0.37
-3.49
.001
-0.81
0.30
-2.66
0.51
0.02
0.58
-2.01
-0.56
-1.41
-0.21
< .001
0.44
0.59
forth motion Jumping on door .008
N
Leaning/
95% C.I.
SC RI PT
S.E.
U
Back and
B
D
below.
A
rubbing
M
body on wall 0.04
13.12
EP
TE
Constant
CC
Table 6. Output of the multivariate reduced regression model on the effects of calm interaction on the behavior of kenneled dogs after the treatment condition. The coefficient (B),
A
standard error (S.E.), t statistic (t), P-value (P), and the confidence interval (C.I.) for each variable are shown. B
S.E.
t
P
95% C.I.
EXERCISE OR CALM INTERACTION
Back and
31
0.42
0.17
2.39
0.017
0.07
0.76
-0.74
0.27
-2.75
0.006
-1.27
-0.21
-0.46
0.17
-2.78
0.006
-0.79
-0.13
0.34
0.17
2.01
0.045
0.007
0.68
Barking
-0.35
0.13
-2.72
0.007
-0.60
-0.10
Constant
0.63
0.13
< .001
0.37
0.90
Jumping on
SC RI PT
forth motion
door Facing
Front of
U
forward
A
M
D TE EP CC A
N
kennel
4.75
S0376-6357(17)30443-6 https://doi.org/10.1016/j.beproc.2017.11.013 BEPROC 3546
To appear in:
Behavioural Processes
Received date: Revised date: Accepted date:
20-9-2017 13-11-2017 16-11-2017
Please cite this article as: Protopopova, Alexandra, Hauser, Hagar, Goldman, Kissel J., Wynne, Clive D.L., The Effects of Exercise and Calm Interactions on In-kennel Behavior of Shelter Dogs.Behavioural Processes https://doi.org/10.1016/j.beproc.2017.11.013 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Running Head: EXERCISE OR CALM INTERACTION
1
The Effects of Exercise and Calm Interactions on In-kennel Behavior of
SC RI PT
Shelter Dogs
Alexandra Protopopova1*, Hagar Hauser2, Kissel J. Goldman3, Clive D. L. Wynne4
of Animal and Food Sciences, Texas Tech University, Lubbock, TX, USA
U
1 Department
College of Veterinary Sciences, University of Florida
3
Department of Psychology, University of Florida, Gainesville, FL, USA
4
Department of Psychology, Arizona State University, Tempe, AZ, USA.
A
N
2
M
*Correspondence concerning this article should be addressed to Alexandra Protopopova,
D
Department of Animal and Food Sciences, Texas Tech University, Lubbock, TX 79415-2141.
TE
Contact: [email protected] Highlights
Some behaviors may be unattractive to adopters and an indicator of poor welfare in
EP
kenneled dogs.
CC
We assessed the effects of providing calm interaction and exercise on in-kennel behavior
A
of 16 shelter dogs.
These interventions result in both a reduction of some, but also in an increase of other unattractive behavior.
EXERCISE OR CALM INTERACTION
2
Over-activity, or excessive locomotion and barking in the kennel, may be unattractive to adopters and an indicator of poor welfare of kenneled dogs. The study assessed the efficacy of
SC RI PT
two common enrichment strategies, providing calm interaction and additional exercise, on inkennel behavior in 16 shelter dogs. Both interventions resulted in appropriate behavior just prior to the sessions (t = 2.10, df = 7, p = .03 and F [2, 216] = 7.58, p = .0007, respectively), but both also resulted in an increase of some undesirable behaviors immediately after the dogs were taken back to their kennels (F [3, 216] = 7.77, p = .0001 and F [5, 216] = 10.1, p < .0001 respectively).
U
Right after receiving additional exercise, the dogs spent more time in back and forth motion in
N
the kennel. Right after receiving the calm interaction, the dogs spent less time in the front of the
A
kennel, less time facing forward, and more time engaging in back and forth motion. However,
M
dogs also spent less time barking and jumping on the kennel door right after the calm interaction.
D
The results suggest that both interventions may be useful, but shelter administrators and
EP
interventions.
TE
volunteers must take all of the behavioral changes into account when administering these
CC
Keywords: Behavior, Dog, Enrichment, Exercise, Learning
A
Animal shelters take in large quantities of surrendered, stray, and confiscated dogs. In the
United States (US) alone, approximately 3.9 million dogs enter animal shelters yearly (American Society for the Prevention of Cruelty to Animals [ASPCA], 2015a). Shelter housing is likely to be stressful, at least initially, (Hennessy, Davis, Williams, Mellot, & Douglas, 1997; Hiby, Rooney, & Bradshaw, 2006; Shiverdecker, Schiml, & Hennessy, 2013; Stephen & Ledger, 2006)
EXERCISE OR CALM INTERACTION
3
and many dogs may stay at the shelter for a prolonged period of time or are euthanized for lack of interested adopters. One determinant of adoptability is the behavior of the animal while at the shelter
SC RI PT
(Luescher & Medlock, 2009; Protopopova, Mehrkam, Boggess, & Wynne, 2014; Protopopova & Wynne, 2014; Weiss, Miller, Mohan-Gibbons, & Vela, 2012; Wells & Hepper, 1992). Previous research has suggested that dogs that engage in excessive back and forth motion in the kennel,
rub or lean on kennel walls, spend a prolonged period of time in the back of the kennel, or face backwards (Protopopova et al., 2014), and engage in barking (Wells & Hepper, 1992) are less
U
attractive to potential adopters. These behaviors may be indicative of stress (e.g., dogs in social
N
isolation had higher locomotion; Hetts et al., 1992) or reactions to a novel environment (activity
A
was found to decrease across time spent in the kennel, Rooney, Gaines, & Bradshaw, 2007;
M
Titulaer, Blackwell, Mendl, & Casey, 2013; but see Protopopova et al., 2014). Furthermore, dogs
D
that are perceived as overly active may further be avoided; 27% of people who decided not to
TE
adopt a dog at a shelter claimed it was because the dog was too active (Protopopova & Wynne, 2014). Since potential adopters take an average of only 20-70 s to look at a kenneled dog prior to
EP
their decision, inappropriate in-kennel behaviors may pose a serious risk of adversely affecting
CC
adoption rates (Wells & Hepper, 2001). Therefore, improving in-kennel behavior of shelter dogs
A
may serve to improve their likelihood of adoption. One common-sense approach to reducing over-activity of shelter dogs is providing
supplemental exercise. Several organizations within the US recommend exercise to maintain wellbeing and behavioral health for kenneled dogs (ASPCA, “Position Statement on Euthanasia”, 2015b; Newbury et al., 2010) and the US Animal Welfare Act (7 U.S.C., 2143 section 13[a][2][B]) specifically addresses exercise requirements for laboratory-housed dogs.
EXERCISE OR CALM INTERACTION
4
Previous research suggests that dogs receiving exercise, play, and human interaction had decreased stress and improved some behaviors in a kenneled environment (Bergamasco, Osella, Savarino, Larosa, Ozella et al., 2010; Coppola, Grandin, & Enns, 2006; Normando, Corain,
SC RI PT
Salvadoretti, Meers, & Valsecchi, 2009; Shiverdecker, Schiml, & Hennessy, 2013). However, research by Hetts et al. (1992) and Campbell, Hughes, Griffin, Landi, and Mallon (1988) found that exercise did not improve behavior in kenneled dogs. In fact, Normando et al. (2009) and
Clark, Rager, Crowell-Davis, and Evans (1997) found that additional exercise actually increased active behaviors in dogs and Clark et al. (1997) found that it also did not prevent the
U
development of abnormal in-kennel behaviors.
N
An alternate point of view rejects additional exercise as an effective treatment for over-
A
activity in kenneled dogs and suggests that calm interactions between volunteers and shelter dogs
M
may result in calmer dogs (e.g., ASPCA, “Enrichment in the Shelter”, 2015c; Sternberg, n.d.).
D
This intervention is based on the premise that dogs are able to learn, through Pavlovian
TE
conditioning, that people are a predictor of excitement. If staff and volunteers only appear in front of the dog’s kennel to take the dog for an exercise session, the visual presentation of a
EP
person may become a conditioned stimulus eliciting excitement (a conditioned response). In
CC
fact, seeing a person resulted in higher activity and approach behavior in kenneled dogs (Arhant & Troxler, 2014; Wells & Hepper, 2000). When visitors were prevented from visiting kenneled
A
shelter dogs, dogs spent more time resting, and less time barking and engaging in stereotypic behaviors (Hewison, Wright, Zulch, & Ellis, 2014). Thus, providing additional calm interactions between people and dogs, might result in the extinction of the conditioned excitable response and subsequently in calmer dogs.
EXERCISE OR CALM INTERACTION
5
These two interventions, additional exercise or calm interactions, are particularly interesting because they directly oppose each other in methodology, yet promise the same outcome. We assessed how these interventions affect behavior immediately before and after the
SC RI PT
sessions; we predicted that, due to learning effects throughout the calm interaction sessions, dogs would appear calm when they first see the experimenter, or right before the treatment sessions. However, due to the same learning effects, dogs may engage in inappropriate behavior right before the sessions during the additional exercise treatment. Alternatively, dogs may appear
calmer directly following the additional exercise treatments due to fatigue, but this effect would
U
not be observed after calm interventions. Therefore, the aim of this study was to compare two
N
opposing, yet currently popular, interventions for counteracting hyperactivity and undesirable in-
A
kennel behavior in dogs housed in shelters.
TE
Animals and housing
D
M
Materials and Methods
Sixteen dogs at the XXX Humane Society (location of Humane Society removed),
EP
located in the United States, were enrolled into the study. All dogs in the shelter at the start of the study were enrolled. Dogs were included in the study regardless of physical characteristics such
CC
as breed, age, or in-kennel behavior. However, dogs that the staff deemed dangerous and/or
A
unpredictable were excluded from the study due to safety concerns. Dogs were also excluded from the study if they were adopted out during the study. Dogs were housed in rows of adjacent 2-sided, indoor kennels with a guillotine door between each side, and glass doors. Each kennel was 1.2 m wide x 2.4 m long x 3.0 m high. The floor material was epoxy over cement. Dogs were housed mostly singly (with 3 dogs housed in
EXERCISE OR CALM INTERACTION
6
pairs). The dogs could be viewed by the public from either side of the run when the guillotine door was open. The standard adoption fee for an adult dog was 100 USD. Each kennel contained a water dish and food dish that hung on the doors as well as a Kuranda bed (Kuranda USA,
SC RI PT
Annapolis, MD, USA) on one side of the kennel. Staff fed the dogs and cleaned kennels daily before 11:00 h. Volunteers at the shelter exercised, trained, and played with the dogs
approximately once per day on the shelter premises. Each kennel card noted the dog’s name and sometimes, a few words on the dog’s personality as perceived informally by shelter staff and
volunteers. Dogs were marketed by the shelter staff and volunteers on their website and through
U
popular online social networking sites.
A
N
Procedure
M
In a counterbalanced mixed-group design, dogs were randomly assigned into the calm interaction or exercise condition (between-subjects factor). The dogs in each condition were
D
further randomly assigned to experience either the baseline or the treatment phase first (within-
TE
subjects factor). Thus, four dogs experienced 2 weeks of the baseline phase followed by 2 weeks
EP
of an exercise phase, four dogs experienced 2 weeks of the exercise phase followed by 2 weeks of the baseline phase, four dogs experienced 2 weeks of the baseline phase followed by 2 weeks
CC
of a calm interaction phase, and four dogs experienced 2 weeks of the calm interaction phase
A
followed by 2 weeks of the baseline phase. Because we aimed to assess the effect of both treatment conditions just prior to treatment
and immediately after treatment, the in-kennel behavior of the dogs was sampled twice during each session in the experimental conditions: once right before the treatment session and once right after the treatment session. Every day, an experimenter conducted an assessment of each
EXERCISE OR CALM INTERACTION
7
dog’s in-kennel behavior prior to the treatment session, conducted the treatment session, and conducted an assessment of in-kennel behavior after the treatment session. If a dog was in the baseline phase of the experiment, the experimenter conducted only a single in-kennel assessment
SC RI PT
daily. Staff and volunteers were aware of the general purposes of the study, but were blind to
the specific conditions and condition assignment of each dog. Also, the staff and volunteers were asked to remain consistent with their routines for the dogs in our study, including the number of
U
times they were walked each day.
N
Intervention conditions
A
The exercise intervention consisted of daily 15 min periods of active toy play and/or
M
running on a leash with the experimenter, whichever the dog appeared to prefer. These sessions
D
were conducted between 9:00 and 18:00. The exercise took place in an indoor playpen located
TE
near the dog kennels or in an outdoor yard. The toys offered to the dogs included Frisbees, tennis balls, squeaky plastic balls, or a ball at the end of a flirt pole. The leashes used while running the
EP
dogs were 1.2 m long slip leads provided by the shelter.
CC
The calm interaction intervention consisted of leading the dog to a quiet indoor area for 15 min daily. These sessions were conducted between 9:00 and 18:00. During this time, the
A
experimenter sat in a chair, read out loud in a calm tone and normal volume, while ignoring the dog. The experimenter took care not to speak to the dog, engage in petting, or in any other way provide attention to the dog. In-kennel assessment
EXERCISE OR CALM INTERACTION
8
Each day, just prior to and just after the experimental condition (or once daily during the baseline condition), the dogs were videotaped for 30 s with the dogs inside their kennels and the experimenter directly outside. The experimenter approached the kennel casually, as a potential
SC RI PT
adopter would, and recorded the dog using a small handheld camera or phone with video capability.
The videos were analyzed using partial 5-s interval recording method by additional experimenters to record the different behaviors that the dogs exhibited (Table 1). These
behaviors were chosen to capture the activity levels of the dogs and to monitor their levels of
U
adoptability. Prior research has shown that the location of the dog in the kennel, face orientation,
N
back and forth motion in the kennel, and leaning or rubbing of the body on kennel walls predicts
A
length of stay in the shelter (Protopopova et al., 2014). Barking has been shown to affect health
M
outcomes in shelter dogs and shelter staff (Coppola et al., 1996; Sales et al., 1997) as well as
D
having been implicated in affecting adoption rates (Wells & Hepper, 2001). Barking was
TE
therefore aggregated with the other identified undesirable behaviors, into one overall measure of
EP
“undesirable behavior”. The ethogram with operational definitions appear in Table 1. ---Table 1 here---
CC
Statistical Analysis
A
Multivariate regression models with treatment as the independent variable (IV) and all
behaviors (except “undesirable behavior”) entered into the model as dependent variables (DV) were conducted for each condition (exercise and calm interaction) and time point (before and after treatment). The effects of condition (IV) and time point (IV) on the aggregate measure of “undesirable behavior” (DV) were further explored with paired-samples t-tests. The impact of
EXERCISE OR CALM INTERACTION
9
the order of baseline and treatment (IV) on all individual behaviors (seven DVs) were analyzed using separate multivariate regression models and descriptive statistics. Statistical significance was determined at p-value less than .05.
SC RI PT
Inter-observer agreement (IOA) was calculated by dividing the number of intervals in which an agreement occurred (both observers agreed whether the behavior occurred or not) by
the sum of agreements and disagreements and converting the result to a percentage in each video. The percentages were then averaged across videos and reported individually by each behavior.
U
Ethical Statement
N
All procedures were approved by the University of Florida Institutional Animal Care and
M
A
Use Committee.
TE
Descriptive Statistics
D
Results
Sixteen dogs completed the study with eight dogs experiencing the calm condition and
EP
eight experiencing the additional exercise with the experimenter condition. Within each group,
CC
half (n = 4), experienced the baseline first and the remaining half experienced the treatment first.
A
Table 2 lists the sex, approximate age, breed, and length of stay at the shelter of each dog. ---Table 2 here--In the large majority of observation periods collected during baseline and just prior to both interventions, the dogs remained facing forward and in the front of the kennel (Table 3). In many observation periods dogs were recorded engaging in back and forth motion, jumping on
EXERCISE OR CALM INTERACTION
10
the kennel door, and barking. Very few observation periods saw dogs jumping straight up in the kennel and rubbing their bodies on the kennel walls.
SC RI PT
---Table 3 here--Exercise Just prior to treatment
A multivariate regression model was conducted with treatment as the independent
U
variable and facing forward, being in the front of the kennel, engaging in back and forth motion,
N
jumping on the kennel door, jumping straight up in kennel, rubbing the body on the wall, and
A
barking as the dependent variables. The model revealed that only face forward and back and
M
forth motion contributed significantly (p < .05) and thus were retained in the final model (F [2, 216] = 7.58, p = .0007). Table 4 lists the predictors. Additional exercise increased the proportion
D
of time the dogs spent facing forward when a person approached (baseline: mean = .88, SD =
TE
.25; treatment: mean = .96, SD = .13) and decreased the proportion of time walking back and
EP
forth in the kennel (baseline: mean = .10, SD = .23; treatment: mean = .04, SD = .10).
CC
---Table 4 here ---
However, receiving exercise did not alter the rate of undesirable behavior in general (as
A
defined above: engaging in at least one of a) Front of kennel not noted, b) Facing forward not noted; c) Back and forth motion noted, d) Barking noted, or e) Leaning/ rubbing body on kennel wall noted; t = 0.06, df = 7, p > .05). The dogs engaged in undesirable behavior on average .40 (SD = .04) proportion of time in baseline and .39 (SD = .04) in treatment.
EXERCISE OR CALM INTERACTION
11
Immediately after treatment An additional multivariate regression model containing all seven DVs revealed that back and forth motion, jumping on door, and wall rubbing were the only variables which contributed
SC RI PT
significantly (p < .05) to the model and thus were retained (F [3, 216] = 7.77, p = .0001). Table 5 lists the predictors and all values. Additional exercise increased back and forth motion of the
dogs (baseline: mean = .10, SD = .23; treatment: mean = .17, SD = .24), decreased jumping on
the door (baseline; mean = .05, SD = .12; treatment; mean = .004, SD = .04), and eliminated wall
U
rubbing (baseline; mean = .04, SD = .18; treatment: behavior never observed).
N
---Table 5 here---
A
Receiving exercise did not alter the rate of undesirable behavior (t = 0.26, df = 7, p >
M
.05). The average proportion of time that the dogs engaged in undesirable behavior in baseline
TE
Calm Interaction
D
was .40 (SD = .04) and after exercise was .37 (SD = .04).
EP
Just prior to treatment
CC
A multivariate regression model, identical to the model applied for the exercise treatment, revealed that no behavioral variables, out of the seven, were influenced by the calm interaction
A
(all p > .05).
A two-tail paired t-test found no statistically significant effect of calm interaction on the
rate of undesirable behavior (t = 2.10, df = 7, p > .05). However, it is interesting to note that a one-tail t-test revealed that receiving calm interactions reduced the rate of undesirable behavior (t
EXERCISE OR CALM INTERACTION
12
= 2.10, df = 7, p = .04, Cohen's d = 3.64; Figure 1). The average proportion of time that the dogs engaged in undesirable behavior in baseline was .36 (SD = .04) and before calm interactions was .28 (SD = .07).
SC RI PT
---Figure 1 here--Immediately after treatment
An additional multivariate regression model containing seven DVs revealed that facing forward, front of kennel, back and forth motion, jumping on door, and barking contributed
U
significantly (p < .05) and thus were retained in the final model (F [5, 216] = 10.1, p < .0001).
N
Table 5 lists the relevant variables. Calm interaction resulted in a decrease in facing forward
A
(baseline: mean = .93, SD = .18; treatment: mean = .82, SD = .31), being in the front of the
M
kennel (baseline: mean = .91, SD = .24; treatment: mean = .87, SD = .27), jumping on kennel
D
door (baseline: mean = .08, SD = .17; treatment: mean = .006, SD = .04), and barking (baseline:
TE
mean = .22, SD = .32; treatment: mean = .05, SD = .16), and an increase in back and forth motion (baseline: mean = .07, SD = .16; treatment: mean = .16, SD = .21).
EP
---Table 6 here---
CC
Receiving calm interactions did not alter the rate of undesirable behavior in general
immediately after treatment (t = 0.58, df = 7, p > .05). The average proportion of time that the
A
dogs engaged in undesirable behavior in baseline was .36 (SD = .04) and .33 (SD = .03) after treatment. Order of Treatment
EXERCISE OR CALM INTERACTION
13
An additional multivariate regression model containing seven DVs revealed that several behaviors were influenced by the order of the treatment conditions. Dogs which received the baseline condition followed by the calm interaction intervention decreased their time in the front
SC RI PT
of kennel prior to the calm interaction (baseline: mean = .99, SD = .05; treatment: mean = .94, SD = .23), but dogs which received the calm interaction first showed an increase in front of
kennel behavior prior to the treatment (baseline: mean = .85, SD = .32; treatment: mean = .92,
SD = .22; t = -2.53,df = 219, p = .012). Furthermore, dogs which received the calm interaction first showed a larger increase in back and forth motion after the calm interaction treatment
U
(baseline: mean = .08, SD = .18; treatment: mean = .22, SD = .24 and baseline: mean = .05, SD =
N
.13; treatment: mean = .09, SD = .15 respectively; t = 3.35, df = 219, p = .001).
A
Dogs which received the baseline first, increased their jumping on the door prior to
M
additional exercise, but the dogs which received treatment first, decreased their jumping prior to
D
exercise (baseline: mean = .006, SD = .03; treatment: mean = .02, SD = .06 and baseline: mean =
TE
.09, SD = .16; treatment: mean = .04, SD = .08 respectively; t = 4.42, df = 216, p < .001). Jumping in the kennel was never seen prior to exercise when the treatment came first (t = -2.48,
EP
df = 216, p = .014), but rubbing the body on the wall was likewise never seen in dogs prior to
CC
exercise which experienced the baseline first (t = 3.79, df = 216, p < .001). Back and forth motion and jumping on the kennel door after the exercise condition were
A
at overall higher levels for dogs which received the exercise first rather than the baseline. The dogs that began in treatment exhibited a higher overall proportion of back and forth motion (mean = .18, SD = .27) compared to those that started in the baseline condition (mean = .09, SD = .18; t = 3.30, df = 216, p = .001). Similarly, the dogs that began in treatment exhibited a higher
EXERCISE OR CALM INTERACTION
14
overall proportion of jumping on the door (mean = .05, SD = .12) compared to the dogs that started in the baseline condition (mean = .004, SD = .03; t = 4.02, df = 216, p < .001).
SC RI PT
Inter-Observer Agreement Approximately a tenth of randomly selected videos (81 out of 725) were coded by an
additional observer. The average IOA was 96.2% (SD = 4.2%). The minimum average IOA for any one behavior was 88.3% (aggregate undesirable behavior) and the maximum was 99.8% (jumping in kennel).
N
U
Discussion
A
Both interventions had complex effects on the behaviors of the dogs, supporting some of
M
our hypotheses but not others. Dogs that received additional exercise spent more time facing forward and less time in back and forth motion just prior to treatment. These findings contradict
D
the hypothesis that providing additional exercise results in increased hyperactive behavior when
TE
a person is first seen by the dog. Instead, dogs were alert and stationary when they saw a person, who was previously paired with exercise. However, dogs did not seem fatigued after the
EP
exercise. Right after receiving additional exercise, dogs increased their back and forth motion in
CC
the kennel. Jumping on the door and wall rubbing decreased, but this decrease was largely driven by an order effect: only dogs which received treatment prior to baseline showed a decrease in
A
jumping on the kennel. Differences in rubbing on the kennel wall were likewise driven by an order effect: only dogs which received baseline first showed a decrease in this behavior. These findings suggest either that exercise does not function to decrease active behaviors of the dogs or that our exercise sessions were too brief to result in fatigue. It is possible that a longer exercise session would result in calmer behavior of the dogs; alternatively, fatigue from extensive
EXERCISE OR CALM INTERACTION
15
physical activity could produce undesirable behaviors, such as appearing lethargic or uninterested. It would be useful for future research to assess varying lengths of exercise sessions to find an optimal duration to elicit quieter behavior in the kennel without exhausting the dogs.
SC RI PT
Our data corresponds in part with previous research. Normando et al. (2009) found that a combination of 15 min of petting and play weekly served to increase locomotion and time spent in sight of kenneled dogs. Normando et al. (2009) reported that this increase in locomotion and
staying in sight was evident both just prior to the interaction as well as immediately afterwards. Perhaps the difference in our data was due to differences in our exercise components and the
U
amount of interaction; Normando et al. (2009) combined petting and play and administered the
N
treatment weekly, whereas our intervention included play and on-leash running and was
A
conducted daily. Hetts et al. (1992) did not detect changes in in-kennel behavior in laboratory
M
dogs that experienced a forced exercise on a treadmill intervention compared to dogs that were not exercised. Furthermore, Campbell et al. (1998) also did not find differences in in-kennel
D
behavior in laboratory dogs that were exercised through access to a larger enclosure nor through
TE
access to conspecifics, compared to dogs that were not exercised. Clark et al. (1997) found that
EP
dogs barked in the kennels more when exercised with conspecifics as compared to dogs that did not receive exercise. These results may be substantially different to our findings as no person
CC
was specifically paired with the exercise component. Dogs that received additional calm interaction displayed less undesirable behavior in
A
general, but did not show differences in any individual behaviors, just prior to calm interaction. Due to our small sample size, we may not have detected the influence of calm interactions on individual behaviors, but we did find an effect when these behaviors were combined. These findings support the learning hypothesis that pairing the visual presentation of a person with
EXERCISE OR CALM INTERACTION
16
calm interactions serves to increase calm, desirable behavior in these dogs. Furthermore, order effects played a role: dogs which received the calm interaction first and baseline second, spent more time in the front of the kennel, whereas the dogs that received the baseline first spent less
SC RI PT
time in the front of the kennel. It is therefore possible that time in front of the kennel was related to the novelty of the experimenter. However, we found that following the calm interaction
intervention, the dogs behaved in a mix of desirable and undesirable ways. Increased undesirable behaviors included spending less time in the front of the kennel, less time facing forward, and more time engaging in back and forth motion. Decreased undesirable behaviors included less
U
time spent barking and less time jumping on the door. Taken together, it may be inferred that the
N
dogs perceived the person as uninteresting right after the intervention and thus were reacting
A
asocially.
M
Whilst we did not record the dog’s behavior during the interactions, anecdotally, dogs in the calm interaction condition spent a large portion of their time laying down. Shiverdecker et al.
D
(2013) found that dogs spent less time barking during passive interactions with a person, in
TE
which a person sat quietly in a chair and ignored the dog. However, they did not find a
EP
statistically different proportion of time spent laying down as compared to other conditions in which an experimenter played with the dogs, or in which the dogs remained isolated in a separate
CC
room. One difference is that the dogs in our study were restrained with a leash, whereas the dogs in the Shiverdecker et al. study were left unrestrained in an enclosure. As previous research has
A
found that laying down next to potential adopters increases the likelihood of adoption (Protopopova & Wynne, 2014), future research can directly evaluate the effect of leash-restraint on out-of-kennel behavior.
EXERCISE OR CALM INTERACTION
17
To summarize, we found that dogs displayed appropriate behavior when anticipating either calm interaction or additional exercise. However, right after either of the interventions, the dogs displayed undesirable behaviors. Immediately after exercise, the dogs exhibited high
SC RI PT
locomotion in their kennels, and immediately after calm interactions, the dogs exhibited behaviors that may be interpreted as asocial (e.g., spent less time in the front of the kennel, less time facing forward, not barking). In order to increase adoptability, our results suggest that it is best to administer both or either of these interventions at the end of the day when all adopters
have already left the shelter. If a shelter has a “quiet hour” in which no adopters are allowed to
U
visit and the dogs are encouraged to lie down quietly in their kennels, it may be advisable to
N
administer the calm interactions right before this, as this intervention seems to promote quiet and
A
asocial behavior. Additionally, it may be best to tailor interventions to specific dogs. For
M
example, the dogs that engage in high rates of barking, the calm interaction may serve to reduce this unwanted behavior. However, for dogs that do not exhibit excessive barking, but instead
D
engage in high levels of wall rubbing, additional exercise may be warranted. These data may be
TE
an important first step in developing tailored interventions to individual dogs in the shelter.
EP
An alternative interpretation of our results is that the two conditions were not sufficiently different from each other to provoke different behavioral responses from the dogs. It is possible
CC
that the relatively similar changes in behavior in response to the two conditions were in response simply to being taken out of the kennel, and not related to the different interactions that occurred
A
afterwards.
Overall, the dogs, regardless of the condition, spent practically all of their time facing
forward and being in the front of the kennel during the observations. Dogs barked in a third of the observations and engaged in barking behavior for approximately a fifth of the observation
EXERCISE OR CALM INTERACTION
18
time. Jumping on the door and engaging in back and forth motion was recorded in a fifth of the observations, but dogs only spent on average 6% and 7% of their time engaging in these behaviors respectively. Leaning and rubbing the body on the enclosure wall and jumping straight
SC RI PT
up in the kennel were rarely seen. Our results suggest that whereas increasing appropriate behavior (e.g., increasing time spent in the front of the kennel and facing forward) may not
provide significant effects on adoptability, as all dogs are engaging in these behaviors frequently and for long durations already, decreasing inappropriate behavior (e.g., time spent in the back of the kennel, facing backwards, back and forth motion, etc.) may serve to improve adoption
U
outcomes. Previous research suggests that visitors pay attention more to negative behaviors than
N
to positive ones (Protopopova et al., 2014).
A
One limitation of this study was that the dogs’ behavior was only recorded for a short
M
period of time. This length of time (30 s) was chosen based on previous research that suggested that adopters only observe the behavior of the dogs for less than 1 min and interact with kenneled
D
dogs on average for only 20 s (Wells & Hepper, 2001). It is possible that the different
TE
enrichment strategies resulted in differences in behavior over longer time periods at the shelter.
EP
Future research could examine the effect of various enrichments on longer periods of in-kennel behavior. Furthermore, all of the behavioral assessments were conducted in response to a
CC
familiar person standing in front of the kennel. As potential adopters are unfamiliar to the dogs, future research should investigate the effects of various human interactions on in-kennel
A
behavior in response to seeing unfamiliar people. Another methodological limitation was the limited sample size and the non-selective
nature of the dogs. Because we required the dogs to stay at the shelter for four weeks in order to include their data, all dogs regardless of breed, size, and temperament were enrolled in the study
EXERCISE OR CALM INTERACTION
19
in order to reach a sufficient sample size. To control for individual differences among dogs, we used a within-subject comparison to determine a treatment effect in each condition. We did not compare calm interactions and exercise to each other directly as different dogs experienced the
SC RI PT
two different conditions. Because we wanted to assess the learning effects that may have occurred when a dog was exposed to the two conditions, we needed to provide each treatment
repeatedly and for a sufficient number of days. And given that the average length of stay at the
shelter used was relatively short, we were not able to subject each dog to both conditions. Thus, our current methodology involved a mixed design; a future within-subject comparison of
U
conditions in a shelter that houses dogs for a much longer period of time may be interesting in
N
order to compare the conditions more directly as well as assess any individual differences. There
A
is increasing evidence that dogs, like other animals, display differences in coping with stressors
M
in their environment. Several studies have now shown that dogs display individual differences, with dogs exhibiting a “proactive” temperament (Koolaas et al., 1999) not showing a significant
D
change in cortisol levels in response to environmental stressors (Horváth, Igyártó, Magyar, &
TE
Miklósi, 2007; Wood, de Bie, & Clarke, 2014).
EP
Another limitation was that we did not utilize the commonly deployed method of calm interaction currently performed in shelters. The more typical method involves reading to the dog
CC
inside their own kennel. The dogs in the present study were housed in kennels that could only be opened from the outside, so for the safety of the experimenters and research assistants, calm
A
interactions were conducted in another room. Whereas the results presented provide empirical support for one type of calm interaction procedure, future research should investigate the differences across commonly used calm interactions and compare the relative effectiveness of each.
EXERCISE OR CALM INTERACTION
20 Conclusion
Our data suggests that both kinds of intervention, calm interactions and additional exercise, provide some benefits, but also result in some undesirable behaviors. Both
SC RI PT
interventions decrease undesirable behaviors just prior to the sessions, but both also result in undesirable behaviors immediately after the dogs were taken back to their kennels. Thus, it may be possible to harness the benefits by administering either or both interventions at the end of the day so that no potential adopters will be dissuaded by the dog’s undesirable behavior after the
enrichment sessions. Alternatively, animal shelters may decide which behaviors they would like
U
to see in their population of dogs and provide these enrichment strategies accordingly.
N
Acknowledgements: This research was supported in part by a grant from the Humane Society
A
Veterinary Medical Association and Merial Veterinary Scholars Program. The authors thank
M
Kelly Cathey, Avi Hershkowitz, Justyna Resztak, Jennifer Taylor, Jessica Vondran, Courtney
D
Alexander, Kaila Ames, Rachel Bradley, Lauren Burstein, Sarah Weinsztok, Nancy Ordax,
TE
Jonathan Pruitt, Jennifer Higgins, Devin Bogart, Estefania Junco, Austin Folger, and Jessica Jeong for assisting in data collection. The authors also thank the staff and the dogs of Alachua
A
CC
EP
County Humane Society.
References
The American Society for the Prevention of Cruelty to Animals (2015a). Pet statistics. Available at: https://www.aspca.org/about-us/faq/pet-statistics. Accessed October 21, 2015.
EXERCISE OR CALM INTERACTION
21
The American Society for the Prevention of Cruelty to Animals (2015b). Position Statement on Euthanasia. Available at: https://www.aspca.org/about-us/aspca-policy-and-positionstatements/position-statement-on-euthanasia. Accessed October 21, 2015.
SC RI PT
The American Society for the Prevention of Cruelty to Animals (2015c). Read and Relax (R&R) Kennel Enrichment Program. Available at:
http://www.aspcapro.org/sites/default/files/mym_enrichment_r_r_program.pdf. Accessed October 21, 2015.
U
Arhant, C., & Troxler, J. (2014). Approach behaviour of shelter dogs and its relationships with
N
the attitudes of shelter staff to dogs. Applied Animal Behaviour Science, 160, 116-126.
A
Bergamasco, L., Osella, M.C., Savarino, P., Larosa, G., Ozella, L., Manassero, M., Badino, P.,
M
Odore, R., Barbero, R., Re, G., 2010. Heart rate variability and saliva cortisol assessment
D
in shelter dog: human-animal interaction effects. Appl. Anim. Behav. Sci. 125, 56–68.
TE
Clark, J. D., Rager, D. R., Crowell-Davis, S., & Evans, D. L. (1997). Housing and exercise of dogs: effects on behavior, immune function, and cortisol concentration. Comparative
EP
Medicine, 47(5), 500-510.
CC
Campbell, S. A., Hughes, H. C., Griffin, H. E., Landi, M. S., & Mallon, F. M. (1988). Some
A
effects of limited exercise on purpose-bred beagles. American journal of veterinary research, 49(8), 1298-1301.
Coppola, C. L., Enns, R. M., & Grandin, T. (2006). Noise in the animal shelter environment: Building design and the effects of daily noise exposure. Journal of Applied Animal Welfare Science, 9, 1–7.
EXERCISE OR CALM INTERACTION
22
Hiby, E. F., Rooney, N. J., & Bradshaw, J. W. (2006). Behavioural and physiological responses of dogs entering re-homing kennels. Physiology & behavior, 89(3), 385-391. Hennessy, M. B., Davis, H. N., Williams, M. T., Mellott, C., & Douglas, C. W. (1997). Plasma
SC RI PT
cortisol levels of dogs at a county animal shelter. Physiology & Behavior, 62(3), 485-490. Hennessy, M. B., Williams, M. T., Miller, D. D., Douglas, C. W., & Voith, V. L. (1998).
Influence of male and female petters on plasma cortisol and behaviour: can human interaction reduce the stress of dogs in a public animal shelter? Applied Animal
U
Behaviour Science, 61(1), 63-77.
N
Hetts S, Clark JD, Calpin JP, Arnold CE, Mateo JM (1992). Influence of housing conditions on
M
A
beagle behaviour. Appl Anim Beh Sci 34: 137–155.
Hewison, L. F., Wright, H. F., Zulch, H. E., & Ellis, S. L. (2014). Short term consequences of
D
preventing visitor access to kennels on noise and the behaviour and physiology of dogs
TE
housed in a rescue shelter. Physiology & Behavior, 133, 1-7. Horváth, Z., Igyártó, B. Z., Magyar, A., & Miklósi, Á. (2007). Three different coping styles in
EP
police dogs exposed to a short-term challenge. Hormones and Behavior, 52(5), 621-630.
CC
Luescher, A., Medlock, R., (2009). The effects of training and environment alterations on
A
adoption success of shelter dogs. Appl. Anim. Behav. Sci. 117, 63–68.
Newbury, S., Blinn, M. K., Bushby, P. A., Cox, C. B, Dinnage, J. D. et al. (2010). “Guidelines for Standards of Care in Animal Shelters”. The Association of Shelter Veterinarians. Available at: http://www.sheltervet.org/assets/docs/shelter-standards-oct2011wforward.pdf. Accessed on October 21, 2015.
EXERCISE OR CALM INTERACTION
23
Normando, S., Corain, L., Salvadoretti, M., Meers, L., & Valsecchi, P. (2009). Effects of an Enhanced Human Interaction Program on shelter dogs’ behaviour analysed using a novel nonparametric test. Applied Animal Behaviour Science, 116(2), 211-219.
SC RI PT
Protopopova, A., Mehrkam, L. R., Boggess, M. M., & Wynne, C. D. L. (2014). In-kennel behavior predicts length of stay in shelter dogs. PLoS ONE, 9, e114319. doi: 10.1371/journal.pone.0114319
Protopopova, A., & Wynne, C. D. L. (2014). Adopter-dog interactions at the shelter: Behavioral
U
and contextual predictors of adoption. Applied Animal Behaviour Science, 157, 109-116.
N
Rooney, N. J., Gaines, S. A., & Bradshaw, J. W. (2007). Behavioural and glucocorticoid
A
responses of dogs (Canis familiaris) to kennelling: investigating mitigation of stress by
M
prior habituation. Physiology and Behavior, 92(5), 847-854.
D
Sales, G., Hubrecht, R., Peyvandi, A., Milligan, S., & Shield, B. (1997). Noise in dog kennelling:
TE
Is barking a welfare problem for dogs? Applied Animal Behaviour Science, 52, 321–329.
EP
doi: 10.1016/S0168-1591(96)01132-X Shiverdecker, M. D., Schiml, P. A., & Hennessy, M. B. (2013). Human interaction moderates
CC
plasma cortisol and behavioral responses of dogs to shelter housing. Physiology &
A
Behavior, 109, 75-79.
Stephen, J. M., & Ledger, R. A. (2006). A longitudinal evaluation of urinary cortisol in kennelled dogs, Canis familiaris. Physiology & Behavior, 87(5), 911-916.
EXERCISE OR CALM INTERACTION
24
Sternber, S. (n.d.). “The Deterioration of a Shelter Dog”. Association of Professional Dog Trainers Newsletter. Available at: https://www.petfinder.com/pro/forshelters/deterioration-shelter-dog/. Accessed on October 21, 2015.
SC RI PT
Titulaer, M., Blackwell, E. J., Mendl, M., & Casey, R. A. (2013). Cross sectional study comparing behavioural, cognitive and physiological indicators of welfare between short and long term kennelled domestic dogs. Applied Animal Behaviour Science, 147(1), 149158.
U
Weiss, E., Miller, K., Mohan-Gibbons, H., Vela, C., (2012). Why did you choose this pet?
N
Adopters and pet selection preferences in five animal shelters in the United States.
A
Animals, 2, 1–17.
M
Wells, D.L., Hepper, P.G., (1992). The behaviour of dogs in a rescue shelter. Animal Welfare, 1,
D
171–186.
TE
Wells, D. L., & Hepper, P. G. (2000). The influence of environmental change on the behaviour
EP
of sheltered dogs. Applied Animal Behaviour Science, 68, 151–162. Wells, D.L., Hepper, P.G., (2001). The behavior of visitors towards dogs housed in an animal
CC
rescue shelter. Anthrozoos, 14, 12–18.
A
Wood, P. A., de Bie, J., & Clarke, J. A. (2014). Behavioural and physiological responses of domestic dogs (Canis familiaris) to agonistic growls from conspecifics. Applied Animal Behaviour Science, 161, 105-112.
Figures
EXERCISE OR CALM INTERACTION
25
Figure 1. The average proportion of time spent in undesirable behavior in general in baseline and
D
M
A
N
U
SC RI PT
prior to the calm/exercise interaction treatments. Means and individual data are shown.
TE
Table 1. Ethogram of all in-kennel behaviors. Behavior
Located between front of cage, and up to and including the midpoint of kennel
CC
EP
Front of kennel
Operational definition
Head oriented such that the observer is able to
Back and forth motion
Distance between the dog and the observer is
A
Facing forward
see more than the side profile of face
either increased, decreased, or both. This change in the distance to the observer occurs at
EXERCISE OR CALM INTERACTION
26 least once (in either direction) during the observation interval.
Jumping on door
Dog on its hind legs with at least one paw on
Jumping in kennel
SC RI PT
the kennel door Dog on its hind legs or jumps off the floor, but not touching the kennel door Barking
Vocalization of very short duration and a low pitch
Prolonged (> .1 sec) contact with the cage wall
U
Leaning/ rubbing body on kennel wall
Scored when at least one of following criteria
A
Undesirable behavior (aggregate measure)
N
by pushing side of body against the cage wall
A
CC
EP
TE
D
M
is satisfied: a)
Front of kennel not noted
b)
Facing forward not noted
c)
Back and forth motion noted
d)
Barking noted
e)
Leaning/ rubbing body on
kennel wall noted
Table 2. Age, breed, and length of stay at the shelter prior to the beginning of the study of
the dogs in both conditions. Age and breed are approximations determined by shelter staff. Name
Approximate age
Breed
Length of
Condition (Calm
stay (days)
interaction [C] or
EXERCISE OR CALM INTERACTION
27 Exercise [E]) and order of Baseline (B)
4 y, 2 m
American Pitbull
54
Terrier mix 3 y, 1 m
Bulldog mix
1
Casey
1 y, 11 m
Retriever mix
275
Gus
1 y 2 mo
Terrier/Pitbull mix
150
Helen
2 y, 1 m
Terrier/Pitbull mix
Maya
2y4m
Mia
2 y, 0 m
Catahoula Leopard
BC EB EB
CB
Terrier/Pitbull mix
4
CB
1
CB
1
BE
77
BC
1
EB
D
A
N
1
M
U
Brandy
BE
SC RI PT
Arnold
TE
Dog mix
A
CC
Norman
4 y, 0 m
EP
Mildred
Pamela
4 y, 0 m
American Bulldog/Boxer mix American Pitbull Terrier mix
2 y, 1 m
Terrier/Retriever mix
EXERCISE OR CALM INTERACTION
Porscha
7 y, 1 m
28
Catahoula Leopard
242
BC
30
BC
Rodeo
2 y, 1 m
American Pitbull Terrier mix
Rosco
3 y, 0 m
Terrier/Pitbull mix
1
Samuel
1 y, 3 m
American Pitbull
82
2 y, 3 m
American Pitbull
BE
BE
1
EB
4 y, 0 m
Hound mix
D
Troy
M
A
Terrier mix
CB
12
N
Simba
U
Terrier mix
SC RI PT
Dog/Boxer mix
Mean proportion of
observations in which
observation period
behavior was exhibited
devoted to behavior
Facing forward
1.00
0.93
0.18
Front of kennel
0.96
0.93
0.22
A
Proportion of
CC
EP
both interventions.
TE
Table 3. The incidence of behaviors collected during control sessions and just prior to
SD
EXERCISE OR CALM INTERACTION
29
0.33
0.20
0.33
Jumping on door
0.20
0.06
0.15
Back and forth motion
0.20
0.07
0.16
Leaning/ rubbing body on
0.04
0.03
0.15
0.02
0.00
0.03
SC RI PT
Barking
kennel wall
U
Jumping in kennel
N
Table 4. Output of the multivariate reduced regression model on the effects of exercise on
A
the behavior of kenneled dogs prior to the treatment condition. The coefficient (B), standard
M
error (S.E.), t statistic (t), P-value (P), and the confidence interval (C.I.) for each variable are
D
shown below.
t
P
0.16
2.65
.009
0.11
0.74
-0.50
0.19
-2.65
.009
-0.87
-0.13
0.15
0.15
0.98
0.33
-0.15
0.45
forwards
CC
Back and
0.43
EP
Facing
S.E.
TE
B
95% C.I.
A
forth motion Constant
Table 5. Output of the multivariate reduced regression model on the effects of exercise on the behavior of kenneled dogs after the treatment condition. The coefficient (B), standard error
EXERCISE OR CALM INTERACTION
30
(S.E.), t statistic (t), P-value (P), and the confidence interval (C.I.) for each variable are shown
t
P
0.30
0.14
2.10
.037
-1.28
0.37
-3.49
.001
-0.81
0.30
-2.66
0.51
0.02
0.58
-2.01
-0.56
-1.41
-0.21
< .001
0.44
0.59
forth motion Jumping on door .008
N
Leaning/
95% C.I.
SC RI PT
S.E.
U
Back and
B
D
below.
A
rubbing
M
body on wall 0.04
13.12
EP
TE
Constant
CC
Table 6. Output of the multivariate reduced regression model on the effects of calm interaction on the behavior of kenneled dogs after the treatment condition. The coefficient (B),
A
standard error (S.E.), t statistic (t), P-value (P), and the confidence interval (C.I.) for each variable are shown. B
S.E.
t
P
95% C.I.
EXERCISE OR CALM INTERACTION
Back and
31
0.42
0.17
2.39
0.017
0.07
0.76
-0.74
0.27
-2.75
0.006
-1.27
-0.21
-0.46
0.17
-2.78
0.006
-0.79
-0.13
0.34
0.17
2.01
0.045
0.007
0.68
Barking
-0.35
0.13
-2.72
0.007
-0.60
-0.10
Constant
0.63
0.13
< .001
0.37
0.90
Jumping on
SC RI PT
forth motion
door Facing
Front of
U
forward
A
M
D TE EP CC A
N
kennel
4.75