- Email: [email protected]
Effects of human contact and toys on the fear responses to humans of shelter-housed dogs
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS Applied Animal Behaviour Science xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Applied Animal Behaviour Science journal homepage: www.elsevier.com/locate/applanim
Effects of human contact and toys on the fear responses to humans of shelter-housed dogs Melanie J. Conley a,∗ , Andrew D. Fisher b , Paul H. Hemsworth a a
Animal Welfare Science Centre Melbourne School of Land and Environment, University of Melbourne, Parkville, VIC 3010, Australia Animal Welfare Science Centre, Faculty of Veterinary Science, The University of Melbourne, 250 Princes Highway, Werribee, VIC 3030, Australia b
a r t i c l e
i n f o
Article history: Accepted 20 March 2014 Available online xxx
Keywords: Human interaction Stress Fear Dog Cortisol Animal shelter
a b s t r a c t This study examined the effects of human contact and toys on fear responses to humans in small breed, shelter-housed dogs. Ninety dogs were assigned to one of three treatments: “control” (control), comprising routine husbandry performed by shelter staff; “human contact” (HC), where dogs additionally received 2 min positive contact daily with an experimenter; or “human contact + toys” (HCT), where the additional human contact included the opportunity to interact with toys. Treatments were implemented daily from day 2 (second day in the shelter) until day 6. On day 7, the fear response towards the experimenter was assessed using salivary cortisol concentrations and a human avoidance test. The behavioural parameters measured were approach and withdrawal responses to the experimenter and time spent in each section of the pen while the experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). On day 8, dogs were assessed by the shelter veterinarian for their suitability for adoption by the public. Treatment had no effect on salivary cortisol concentrations or the proportion of dogs selected for adoption. The results indicated that dogs in the HC and HCT treatments spent more time (P = 0.024) at the front of the pen when the experimenter was at position 3 than control dogs (HC 8.8 s, HCT 8.7 s, control 6.6 s from a possible 10 s). For those dogs that were not at the front of the pen when the experimenter was in position 3, a higher (P < 0.05) percentage of HCT dogs (100%) but not HC dogs (86%, P > 0.05) approached the experimenter compared with control dogs (45%). A second experiment on another 40 dogs examined if dogs, based on their behaviour in the human avoidance test, discriminate between familiar and unfamiliar humans. All dogs received the HC protocol as in Experiment 1 and were observed in the human avoidance test, however, all dogs were tested twice, once with the ‘familiar’ experimenter and once with an ‘unfamiliar’ experimenter. There were no significant (P > 0.05) familiarity or testing order effects, except that dogs approached the unfamiliar experimenter more (60%) than the familiar experimenter (29%) at position 2 (P < 0.05). The results indicate that additional positive human contact other than that associated with routine husbandry reduced the behavioural fear response of dogs to humans, but did not affect the proportion of dogs selected for adoption. The results also indicate that the shelter dogs did not discriminate between familiar and unfamiliar humans, in the handling context used in this study, suggesting a degree of stimulus generalisation may occur. © 2014 Elsevier B.V. All rights reserved.
∗ Corresponding author. Tel.: +61 422141809. E-mail address: [email protected] (M.J. Conley). http://dx.doi.org/10.1016/j.applanim.2014.03.008 0168-1591/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
2
1. Introduction Fear can be considered as an undesirable emotional state of suffering that functions to protect the animal through defensive behaviour or escape and is elicited by the perception of actual or perceived danger (Toates, 1980; Jones and Waddington, 1992; Hemsworth and Coleman, 2011). The behavioural responses associated with fear include withdrawal, or avoidance responses, as well as immobility, such as freezing or crouching (Hemsworth and Barnett, 1987; Jones, 1987; Mills and Faure, 1990). The physiological responses associated with fear include responses of the sympatho-adrenal medullary system (e.g. secretions of adrenalin and noradrenalin) and the hypothalamo-pituitary adrenal system (HPA axis e.g. secretions of corticosteroid hormones, cortisol and corticosterone) (Hemsworth and Coleman, 2011). In Victoria, the Code of Practice for the Management of Dogs and Cats in Shelters and Pounds requires that dogs are housed during the first 8 days in shelters in quarantine in individual pens with no tactile contact with other dogs. This requirement often results in dogs also having no visual contact with other dogs during this period. Standard operating procedures in other Australian states impose similar “quarantine” periods, ranging from 3 days to a week in animal shelters. Surprisingly little research has been conducted on stress in shelter-housed dogs. Animal shelters are a stressful environment for dogs, possibly because of the novel surroundings, social restrictions and reduced human contact. Dogs have increased cortisol concentrations during the first 3 days in an animal shelter, declining slowly before reaching a plateau around day 9–10 (Hennessy et al., 1997). Even so, the authors reported that this plateau was above the average basal cortisol levels found in domestic dogs. An initial stress response can prepare an animal to adapt to a change in its environment; however, a longer-term activation of the stress response can lead to behavioural and physiological disturbances such as stereotypies, prolonged elevation of corticosteroids and immunosuppression (Johnson et al., 1992; Broom and Johnson, 1993), with welfare implications. Dogs are social animals, needing contact with both conspecifics (Fox, 1965; Fox and Stelzner, 1967) and, in dogs socialised to humans, contact with humans (Freedman et al., 1961; Wolfe, 1990). In the presence of a human, kennelled dogs typically become more active and spend more time at the front of their cage, reducing the human–animal distance and facilitating interaction (Campbell et al., 1988; Hughes et al., 1989; Wells and Hepper, 2000). Surveys have found that people looking to adopt a shelter dog prefer dogs located at the front of the cage rather than at the back, and dogs that are alert rather than non-alert (Wells and Hepper, 2000). The restricted human contact that typically occurs in shelters may increase fear of humans in shelter dogs and thus reduce their attractiveness to people looking to adopt a dog. While the history of interactions between a human and a domesticated animal will affect the behavioural response of the animal to humans, it has been suggested that the behavioural response of farm animals to an individual human, through stimulus generalisation, can extend to
other humans (Hemsworth and Coleman, 2011). Stimulus generalisation can be defined as a stimulus similar in nature to an original stimulus that produces the same behavioural response in a learning situation (Reber, 1988) but there has been very little research conducted on stimulus generalisation in dogs housed at a shelter. Hubrecht (1993, 1995) examined stimulus generalisation in laboratory-housed beagles and found that as little as 30 s of human handling daily for 2 months resulted in the beagles displaying more approach to both familiar and unfamiliar humans. Very little research has been conducted examining the effects of combining human contact with the opportunity to play in shelter dogs. The dog is one of few species to engage in interspecific play with humans (Russell, 1936). Introducing play into a shelter environment may be beneficial for socialisation and preparation for rehousing by providing appropriate dog–human relationships (Wells, 2004). The first experiment in this study examined the effects of daily human contact and the opportunity to interact with toys during the first week of housing in an animal shelter on the behaviour and welfare of small breed dogs. Furthermore, treatment effects on the percentage of dogs selected as suitable for adoption at the completion of the 8-day quarantine period was also examined. A second experiment in this study examined whether the effects of human contact were specific to the human that provided the contact, or were generalised to other humans. 2. Materials and methods 2.1. Ethical approval of animal experimentation The protocol and conduct of the study were approved by the University of Melbourne, Melbourne School of Land and Environment Animal Ethics Committee (Ethics # 0703341.1 and 0911103). 2.2. Animals and facilities The study was conducted at a large animal shelter in Victoria, Australia receiving at least 20,000 dogs per year. During the 8-day quarantine period, dogs at the shelter were individually housed in pens (approximately 1 m wide × 2.5 m long × 2.5 m high) consisting of a concrete floor with a raised bed at the back of the pen, solid walls and a wire mesh front gate. Dogs remained in their pens during the 8-day quarantine period and were visually isolated from all other dogs in the shelter. Routine care by shelter staff was restricted to morning cleaning, whereby dogs remained in their pens while the pens were hosed clean and food and water provided. Tactile contact with the dogs by the staff was rare. 2.3. Experiment 1 Ninety small breed dogs over a 7-week period were studied on admittance to the shelter. The dogs varied in age, sex and breed and their past human and housing experiences were unknown.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
2.3.1. Treatments and measurements Five rows of nine pens were used to house the study dogs with pens 2, 3, 5, 6, 8 and 9 in each row allocated. As dogs of a suitable size (small breed) entered the shelter (referred to as day 1), they were allocated to the next empty pen in a row. To deal with possible position effects within rows and between rows, dogs were allocated to treatment in a pseudo-randomised manner so that each pair of pens in each row was represented at least once by each treatment. The following treatments were imposed from days 2–6: 1. “Control”: routine cleaning and feeding was performed by shelter staff. 2. “Human Contact” (HC): in addition to the routine husbandry by staff, the experimenter (MJC) provided 2 min daily positive human contact. The experimenter crouched down inside the pen with her back against the front gate, then interacted with the dog by talking in soft tones, presenting her hand for the dog to sniff, and patted the dog if it approached. On completion of the 2 min, the experimenter quietly stood and exited the pen. 3. “Human Contact & Toys” (HCT): in addition to the routine husbandry, the experimenter imposed the HC treatment and at the same time also offered one of two toys for the dog to interact with during the 2 min period. On days 2 and 3, the toy was a rope toy, and for days 4, 5 and 6, the toy was a plush squeaky toy in the shape of a bone. After 2 min, the experimenter quietly exited the pen. Treatments were conducted daily from 07:00 h on weekends and 08:00 h on weekdays except for days in which saliva was collected and the behavioural response of dogs to the experimenter was observed (day 7, 06:00 h on weekends and 07:00 h on weekdays), when treatment imposition followed. After the imposition of all treatments, staff cleaned pens and fed the dogs. The accommodation area was locked for the day after feeding. 2.3.2. Saliva sampling Saliva samples were collected by the experimenter on day 7 with the dogs in their pens. When preparing to collect the saliva samples, any dogs considered likely to be aggressive, based on warning signs pinned on the pens by shelter staff or previously recorded behaviours by the experimenter such as baring of teeth and/or extreme submissive crouching, were not sampled. Seventy-four dogs from the 90 were sampled. Two saliva samples were collected 10 min apart from each dog tested on that day. The first was collected to estimate basal cortisol concentration and the second to estimate the cortisol response to human handling associated with the first sampling (i.e. the acute response to handling) (Broom and Johnson, 1993). Salivary cortisol provides an accurate estimate of plasma cortisol (Negrao et al., 2004; Vincent and Michell, 1992). An experienced volunteer aided the experimenter during the collection by holding the dog on their knee. All samples were collected using a salivette (Sarstedt Inc. Sarstedtstraße Postfach, Nümbrecht, Germany) and were collected within 3 min of the experimenter and volunteer entering the pen. Dogs were encouraged to chew the swab, as chewing increases salivation. The swab was
3
then placed into a labelled salivette tube and placed in a rack on ice. Within 2 h of collection, the salivettes were removed from ice and centrifuged at 3500 rpm at 10 ◦ C for 45 min. The centrifuged saliva was then transferred into labelled 2mL freezing tubes and stored at −20 ◦ C. Saliva cortisol was measured using a Human/Sheep Cortisol ELISA Quantification Kit (MyBioSource Inc., San Diego) that also measures cortisol concentrations in dog saliva (Dreschel and Granger, 2005). The collection and storage procedure has been successfully used to collect saliva from dogs (Dreschel and Granger, 2005). 2.3.3. Human avoidance test The behavioural response of the dogs to the experimenter was assessed on day 7, immediately following collection of the saliva samples. In this test, the experimenter placed a video camera (Panasonic PV-L352, Osaka, Japan) mounted on a tripod 2 m from the pen gate (position 1), pressed record, then removed herself from the dog’s sight for 10 s. After this time, the experimenter returned, stood next to the camera at position 1 for 5 s, took one step towards the pen gate to position 2 which was 1 m from the gate for 5 s, then took one step to the front of the pen gate where she crouched down for 10 s at position 3. At no point during the test did the experimenter make eye contact with the test dog. For this test, the pen was nominally sectioned into quarters (S1–S4, representing the four areas from the front of the pen) to measure the time spent in the four locations of the pen. The test was recorded in real time on video, read in real time and from the video recordings, the following was calculated for each dog when the experimenter was in each position: time spent in each area; and time to entry and exit from each area of the pen. Location in each area of the pen was assessed on the basis of the dog’s head location. In order to assess the approach response when the experimenter was in each position, those dogs not initially at the front of the pen when the experimenter was either in position 1, position 2 or position 3 were classified as either approaching within the 5- or 10-s observation period (first step taken was towards the front of the pen and the experimenter) or not (no step or the first step taken was towards the back of the pen). Data for dogs initially at the front of the pen that withdrew before the 5th or 10th s of the 5- or 10-s observation periods were also analysed. These dogs were classified as either approaching within the 5-s or 10-s observation period (after withdrawing, the first step taken was towards the front of the pen) or not (after withdrawing, no step forward was taken). For each position, those dogs initially at the front of the pen that did not withdraw before the 5th or 10th s of the 5or 10-s observation periods were excluded from analyses as they were not able to approach so would not be given a score to represent their true behaviour. The corresponding method was used for assessing withdrawals. For example, for each position, those dogs not initially at the back of the pen (e.g. anywhere else in the pen except the back) were classified as either withdrawing
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
4
Table 1 The effects of handling and time on dog salivary cortisol concentrations in Experiment 1. Sample 2 was taken 10 min after Sample 1. Means (and standard errors of the mean in parentheses) are presented. Item
Sample 1 (nmol/L) Sample 2 (nmol/L) Overall (nmol/L)
Treatment
Time
Control
HC
HCT
P
Sample 1
Sample 2
P
46.3 (15.37) 39.0 (11.70) 42.7 (13.09)
53.0 (14.19) 40.8 (10.80) 46.9 (12.08)
73.3 (15.05) 64.8 (11.45) 69.1 (12.81)
0.393 0.210 0.889
57.5 (8.59)
48.2 (6.54)
0.037
(first step taken within the 5- or 10-s observation period was towards the back of the pen) or not (no step taken for the 5- or 10-s observation period or the first step taken was towards the front of the pen). 2.3.4. Assessment of suitability of the dogs for adoption On completion of the 8-day quarantine period at the shelter, dogs that had not been collected by their owners underwent behaviour (e.g., aggression, fear) and health (e.g., old age, ongoing skin allergies) assessments by the shelter veterinarian, determining if these dogs were suitable for adoption by the public. Shelter records were collected on whether or not the study dogs were assessed as suitable for adoption. 2.3.5. Statistical analysis Some variables did not conform to a normal distribution as indicated by visual methods (Q–Q plots and histograms, SPSS statistical package, SPSS 17.0, SPSS Inc., Chicago, IL, USA). A Repeated Measures Analysis of Variance (SPSS) was conducted to examine the effects of time (sample) and treatment on salivary cortisol concentration. Univariate Analysis of Variance (SPSS) was used to examine effects of treatment, row and week on dog behaviour. Comparisons between treatments were tested using post hoc l.s.d. tests. To analyse approach responses, separate analyses were performed for dogs not initially at the front of the pen and for dogs initially at the front of the pen that first withdrew. Similarly, probabilities of withdrawing were analysed for dogs not initially at the back of the pen and for dogs initially at the back of the pen but then approached. Effects of treatment on approach and withdrawal responses were assessed using overall and pair-wise Fisher’s exact tests, calculated using the Compare2 module (version 2.75) in WinPepi version 11.25 (Abramson, 2011). As the data from the suitability for adoption assessments were not normally distributed, a non-parametric Kruskal–Wallis Analysis of Variance (SPSS) was used to examine treatment effects on the fate of the animals. 2.4. Experiment 2 Another 40 small breed dogs were used to determine if the dogs discriminated between familiar and unfamiliar humans. All 40 dogs were exposed to the same conditions as in treatment 2 of Experiment 1 for 5 days, with the same
experimenter exposing all 40 dogs to the human contact treatment in order to become the ‘familiar’ human.
2.4.1. Human avoidance test The same method used in Experiment 1 was implemented again in Experiment 2, however, all dogs were tested twice, once with the ‘familiar’ experimenter (MJC) who had been imposing the human contact, and once with an ‘unfamiliar’ experimenter with whom the dogs had not had contact. Both experimenters were female, similar in height and wore matching blue overalls. The order in which individual dogs were tested with the familiar and unfamiliar experimenters was randomised and the interval between the two tests was 30–40 min. For this experiment, to improve ease of video analysis, the pen was nominally sectioned into thirds to measure time spent in different locations in the pen. From the video observations, the time that each dog spent in each area and their approach and avoidance responses when the experimenter was in each position was calculated as in Experiment 1. 2.4.2. Statistical analysis As in Experiment 1, some variables did not conform to a normal distribution as indicated by visual methods (Q–Q plots and histograms, SPSS statistical package, SPSS 17.0, SPSS Inc., Chicago, IL, USA). Analysis of Variance (SPSS) was used to examine the experimenter effects and testing order effects on dog behaviour. To analyse approach behaviours (as in Experiment 1), separate analyses were performed for dogs not initially at the front of the pen and for dogs initially at the front of the pen that first withdrew. Similarly, probabilities of withdrawing were analysed for dogs not initially at the back of the pen and for dogs initially at the back of the pen that first approached. Effects of experimenter (‘familiar’ or ‘unfamiliar’) and testing order were assessed using a method for partially (i.e. incompletely) paired data. The data were partially paired because some dogs contributed data for both experimenters while others contributed data for only one of the two experimenters, as dogs that were initially at the front of the pen that never withdrew to allow an approach, and dogs that were initially at the back of the pen and never approached to allow a withdrawal could not contribute data. We used the method reported by Tang et al. (2009), implemented using the PAIRS etc. module (version 3.13) in WinPepi. This method assumed that exclusions were independent of experimenter.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
Test phase Position 1 Treatment
xy
HC
HCT
3.9 (0.37) 0.0 (0.19) 0.2 (0.10) 1.0 (0.32)
4.3 (0.04) 0.3 (0.19) 0.0 (0.10) 0.3 (0.32)
3.8 (0.37) 0.4 (0.19) 0.0 (0.10) 0.8 (0.33)
P 0.488 0.306 0.410 0.297
Position 3
Control
HC
HCT
P
3.3 (0.37) 0.6 (0.21) 0.0 (0.06) 1.0 (0.32)
4.4 (0.36) 0.2 (0.20) 0.1 (0.06) 0.3 (0.32)
3.8 (0.37) 0.3 (0.21) 0.1 (0.06) 0.8 (0.33)
0.110 0.349 0.815 0.285
Control
HC
HCT
P
6.6x (0.64) 0.3 (0.21) 0.8 (0.25) 2.3 (0.56)
8.8y (0.63) 0.4 (0.21) 0.1 (0.25) 0.7 (0.55)
8.7y (0.65) 0.3 (0.22) 0.1 (0.26) 0.9 (0.57)
0.024 0.965 0.089 0.093
Means without a common superscript are significantly different (P < 0.05).
Table 3 The effects of handling on the approach and withdrawal responses to humans of dogs in the human avoidance test in Experiment 1. The experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented. Item
Test phase Position 1 Treatment
Approached (not initially at front) (%) Approached (initially at front) (%) Withdrew (not initially at front) (%) ab
Position 2
Position 3
Control
HC
HCT
P
Control
HC
HCT
P
13 (8) 33 (3) 13 (23)
38 (8) 0 (1) 3 (29)
14 (7) 100 (2) 15 (26)
0.557 0.400 0.262
20 (10) 67 (3) 22 (23)
44 (9) 67 (3) 21 (29)
33 (9) 0 (2) 24 (25)
0.536 45 (11)a 0.486 50 (4) 1.000 35 (23)
Control
HC 86 (7)ab 100 (2) 14 (29)
HCT 100 (8)b 100 (1) 12 (26)
P 0.025 0.619 0.110
Means without a common superscript are significantly different (P < 0.05).
ARTICLE IN PRESS
Time spent in S1 (s) Time spent in S2 (s) Time spent in S3 (s) Time spent in S4 (s)
Position 2
Control
G Model
APPLAN-3883; No. of Pages 8
Item
M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
Table 2 Effects of handling on the location of dogs in their pens during the human avoidance test in Experiment 1. S1 represents the section closest to the front of the pen and S4 being farthest from the front of the pen. The experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented.
5
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
6
Table 4 Results examining treatment effects on the assessment of suitability of the dogs for adoption in Experiment 1. Item
Proportion of dogs euthanased for behavioural reasons (%) Proportion of dogs euthanased for health reasons (%) Proportion of dogs assessed as suitable for adoption (%)
Treatment Control
HC
HCT
Chi-square
P
32 29 39
32 23 45
34 35 31
0.044 1.033 1.251
0.978 0.596 0.535
3. Results 3.1. Experiment 1 There were no significant (P > 0.05) effects of pen location (block) or week of the study on the behavioural or physiological variables. 3.1.1. Cortisol concentrations There was a significant (P < 0.05) effect of time but not of treatment (P > 0.05) on salivary cortisol concentrations, with lower cortisol concentrations in sample 2 (48.2) than in sample 1 (57.5) across all treatments (Table 1). 3.1.2. Human avoidance test The effects of treatment on dog behaviour are shown in Table 2. When the experimenter was in positions 1 and 2 in the human avoidance test, there were no significant (P > 0.05) treatment effects on the total time dogs spent in any quarter of the pen. However, when the experimenter was in position 3 of the test, dogs in the HC and HCT treatments spent more time (P < 0.05) in the front quarter of the pen (S1) than control dogs. The effects of treatment on approach and withdrawal responses are shown in Table 3. For dogs that were not initially at the front of the pen when the experimenter was in position 3, a higher (P < 0.05) percentage of HCT dogs approached than control dogs. There were no significant (P > 0.05) treatment effects on withdrawal responses. 3.1.3. Assessment of suitability of the dogs for adoption Treatment did not significantly (P < 0.05) influence the percentages of dogs assessed as unsuitable for adoption based on behavioural or health reasons, or those assessed as suitable (Table 4). 3.2. Experiment 2 There were no significant (P > 0.05) effects of pen location (block) or week of the study on the behavioural variables. Furthermore, there were no significant (P > 0.05) familiarity or testing order effects on the time dogs spent in the three areas of the pen during the human avoidance test (Table 5). The effects of a familiar and unfamiliar experimenter on approach and withdrawal response are shown in Table 6. While there were no effects of familiarity when the experimenter was in positions 1 and 3, a higher (P < 0.05) percentage of dogs approached the unfamiliar experimenter than the familiar experimenter during position 2.
There were no significant (P > 0.05) experimenter effects on withdrawal, and no significant (P > 0.05) testing order effects on approach or withdrawal responses.
4. Discussion The results of Experiment 1 indicate that dogs that receive positive human contact and positive human contact and toys increased the time that they spent in the front quarter of the pen (S1) when the experimenter was very close to the front of the pen (position 3 of the human avoidance test) in comparison to the control dogs. Furthermore, for dogs that were not at the front of the pen when the test commenced, more of the dogs that received positive human contact and toys approached the experimenter than the control dogs. The dogs that received human contact were intermediate. There were no treatment effects on the location of the dogs in the pen when the experimenter was 1 and 2 m from the front of the pen. Avoidance of an approaching experimenter and approach to a stationary experimenter are two of the most common and well-accepted measures used to assess fear of humans in domestic animals (Waiblinger et al., 2006; Hemsworth and Coleman, 2011) and thus in the human avoidance test, dogs that are less fearful of humans would be expected to show more approach to the experimenter and consequently spend more time at the front of the pen, particularly when the experimenter is stationary and close to the front of the pen. While there is evidence of reduced fear of humans in dogs that received positive human contact and positive human contact and toys, this evidence is limited. The treatment effects on time spent in the front of the pen when the experimenter was very close were relatively small and only the treatment consisting of positive human contact and toys increased approach in those dogs that were not at the front of the pen when the test commenced. However, it is important to recognise that this latter variable applies only to a subset of dogs, that is those that were not initially at the front of the pen, and this result needs to be interpreted cautiously. A necessary limitation of the study is that some of the obvious factors that may markedly affect the opportunity for brief imposition of the two intervention treatments to affect fear, such as previous human and housing experiences of the dogs, were not controlled. Further research on large numbers of shelter dogs on the effects of positive human contact and the opportunity to interact with toys on the behavioural response of dogs is warranted because of the implications on fear and welfare of shelter dogs.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS 0.394 0.359 0.752 8.3 (0.47) 0.9 (0.31) 0.8 (0.39)
Test order 1 Test order 2 P
8.9 (0.47) 0.5 (0.31) 0.6 (0.39) 0.629 0.111 0.499 8.8 (0.47) 8.5 (0.47) 0.4 (0.31) 1.1 (0.31) 0.9 (0.39) 0.5 (0.39) 0.127 0.077 0.641 3.8 (0.28) 0.6 (0.18) 0.6 (0.23) 4.4 (0.28) 0.2 (0.18) 0.4 (0.23) 0.798 0.428 0.353 4.1 (0.28) 4.2 (0.28) 0.3 (0.18) 0.5 (0.18) 0.6 (0.23) 0.3 (0.23) 0.346 0.449 0.582 4.0 (0.28) 0.5 (0.19) 0.6 (0.22) 4.4 (0.28) 0.3 (0.19) 0.4 (0.22) 0.950 0.570 0.582
Unfamiliar P
Time spent in S1 (s) 4.2 (0.28) 4.2 (0.28) Time spent in S2 (s) 0.3 (0.19) 0.4 (0.19) Time spent in S3 (s) 0.6 (0.22) 0.4 (0.22)
Familiar Test order 1 Test order 2 P Familiar
Familiar
Position 2
Treatment Test order Treatment
Unfamiliar P
Test order 1 Test order 2 P
Position 3
Test order
Treatment
Unfamiliar P
Test order
7
Position 1
Test phase Item
There were no treatment effects on salivary cortisol concentrations, however, there was a significant time effect. In the present study, cortisol concentrations overall were lower in sample 2 than sample 1. Handling associated with the first saliva sample was expected to result in a shortterm increase in cortisol concentration within 5–15 min (Hemsworth et al., 1981), particularly in fearful dogs. One explanation is that differences in the fear responses between treatments were insufficient to produce treatment effects on the cortisol response to handling. Furthermore, there is evidence that the secretion of cortisol in dogs coincides with activity (Kempanien and Sartin, 1984; Orth et al., 1988; Castillo et al., 2009). In the present study, the dogs may have been aroused at the sound of a human entering the pen blocks and perhaps expected to be fed, which may have masked any treatment effects of fear of humans. There were no treatment effects on dogs assessed as suitable for adoption. Luescher and Medlock (2009) found that trained dogs available for adoption were 1.4 times more often adopted than untrained dogs. These dogs received one 20 min training session daily over a 2 week period where they were desensitised to a Gentle Leader, taught to come forward in their cage when approached, walk on a leash, sit and to not jump on people. In the present study, treatment effects may not have directly impacted on the dog’s behaviour during assessment, however, the intervention treatments may have enhanced the behaviour of dogs that were always destined to pass, but may not have had sufficient effect on those that were always destined to fail. There were a number of dogs euthanased for health reasons, which may have resulted in an insufficient sample size to demonstrate an effect on suitability of dogs for adoption. The results of Experiment 2 indicate that there were no effects of familiarity with the experimenter on the location of the dogs in the pen when the experimenter was close to the front of the pen or 1 and 2 m from the front of the pen. There were also no effects of familiarity with the experimenter on the avoidance responses of the dogs that were not initially at the front of the pen. However, for dogs that were not initially at the front of the pen when the experimenter was 1 m from the front of the pen, more dogs approached the unfamiliar experimenter than the familiar experimenter. There is no obvious explanation for this effect since there were no treatment effects when the experimenter was either 2 m from the pen or very close to the pen. As discussed earlier in Section 4, dogs that are less fearful of humans would be expected to show more approach to the experimenter and dogs that are more fearful would be expected to show the opposite. Thus these results suggest that, in the handling context used in this study, a degree of stimulus generalisation in shelter housed dogs may occur whereby the response of an animal towards a familiar human who administered positive handling produced the same or a similar response to an unfamiliar human with the same physical characteristics. Discrimination between people by domesticated animals is obviously easier if the animals have some distinct cues on which they can discriminate, such as size, behaviour, colour of clothing or location of handling (Hemsworth and Coleman, 2011).
Table 5 The effects of unfamiliar and familiar testers and testing order on the location of dogs in the human avoidance test in Experiment 2. S1 represents the section closest to the front of the pen and S3 being farthest from the front of the pen. The familiar or unfamiliar experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented.
M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
8
Table 6 The effects of familiar and unfamiliar experimenters on the approach and withdrawal responses to humans of dogs in the human avoidance test in Experiment 2. The Familiar or Unfamiliar experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented. Item
Test phase Position 1
Approached (not initially at front) (%) Withdrew (not initially at front) (%)
Position 2
Position 3
Familiar
Unfamiliar
P
Familiar
Unfamiliar
P
Familiar
Unfamiliar
P
15 (13) 3 (35)
7 (15) 5 (37)
0.451 0.592
29 (14) 29 (35)
60.0 (15) 27 (37)
0.016 0.858
75 (16) 40 (35)
88 (17) 45 (38)
0.263 0.15
5. Conclusion The present study provides limited evidence that positive human contact may reduce the fear response of shelter dogs towards humans during the first week of admittance to a shelter. However, further research is required to determine the effects of human handling at a shelter on dog welfare. Acknowledgements The technical support of M. Rice and L. Edwards, John Morton for assisting with the statistical analysis, Sally Haynes for comments on the draft manuscript and the participation of the shelter staff are gratefully acknowledged. References Abramson, J.H., 2011. WINPEPI updated: computer programs for epidemiologists, and their teaching potential. Epidemiol. Perspect. Innov. 8, 1. Broom, D.M., Johnson, K.G., 1993. Stress and Animal Welfare. Chapman and Hall, London, UK. Campbell, S.A., Hughes, H.C., Griffin, H.E., Landi, M.S., Mallon, F.M., 1988. Some effects of limited exercise on purpose bred Beagles. Am. J. Vet. Res. 49, 1298–1301. Castillo, V.A., Cabrera Blatter, M.F., Gomez, N.V., Sinatra, V., Gallelli, M.F., 2009. Diurnal variations in healthy dogs and those with pituitarydependent Cushing’s syndrome before and after treatment with retinoic acid. Res. Vet. Sci. 86 (2), 223–229. Dreschel, N.A., Granger, D.A., 2005. Physiological and behavioral reactivity to stress in thunderstorm-phobic dogs and their caregivers. Appl. Anim. Behav. Sci. 95 (3-4), 153–168. Fox, M.W., 1965. Environmental factors influencing stereotyped and allelomimetic behaviour in animals. Lab. Anim. Care 15, 363–370. Fox, M.W., Stelzner, D., 1967. The effects of early experience on the development of inter and intraspecies social relationships in the dog. Appl. Anim. Behav. Sci. 15, 377–386. Freedman, D.G., King, J.A., Elliot, O., 1961. Critical periods in the social development of the dog. Science 133, 1016. Hemsworth, P.H., Barnett, J.L., Hansen, C., 1981. The influence of handling by humans on behaviour, growth and corticosteroids in the juvenile female pig. Horm. Behav. 15, 396–403. Hemsworth, P.H., Barnett, J.L., 1987. Human–animal interactions. Veterinary clinics of North America. Food Anim. Pract. 3, 339–356. Hemsworth, P.H., Coleman, G.J., 2011. Human–Livestock Interactions. The Stockperson and the Productivity and Welfare of Intensively Farmed Animals. CAB International, Wallingford, UK. Hennessy, M.B., Davis, H.N., Williams, M.T., Mellott, C., Douglas, C.W., 1997. Plasma cortisol levels of dogs at a county animal shelter. Physiol. Behav. 62, 485–490.
Hubrecht, R.C., 1993. A comparison of social and laboratory environmental enrichment methods for laboratory housed dogs. Appl. Anim. Behav. Sci. 37, 345–361. Hubrecht, R.C., 1995. Enrichment in puppyhood and its effects on later behaviour of dogs. Lab. Anim. Sci. 45, 70–75. Hughes, H.C., Campbell, S., Kenney, C., 1989. The effects of cage size and pair housing on exercise of Beagle dogs. Lab. Anim. Sci. 39, 302–305. Johnson, E.O., Kamilaris, T.C., Crousos, G.P., Gold, P.W., 1992. Mechanisms of stress: a dynamic overview of hormonal and behavioral homeostasis. Neurosci. Biobehav. Rev. 16, 115–130. Jones, R.B., 1987. Fear and fear responses: a hypothetical consideration. Med. Sci. Res. 15, 1287–1290. Jones, R.B., Waddington, D., 1992. Modifications of fear in domestic chicks Gallus gallus domesticus, via regular handling and early environmental enrichment. Anim. Behav. 43, 1021–1033. Kempanien, R.J., Sartin, R.L., 1984. Evidence for episodic but not circadian activity in plasma concentrations of adrenocorticotrophin, cortisol and thyroxine in dogs. J. Endocrinol. 103, 219–225. Luescher, A.U., Medlock, 2009. The effects of training and environmental alterations on adoption success of shelter dogs. Appl. Anim. Behav. Sci. 117, 63–68. Mills, A.D., Faure, J.M., 1990. Panic and hysteria in domestic fowl: a review. In: Zayan, R., Dantzer, R. (Eds.), Social Stress in Domestic Animals. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 248–272. Negrao, J.A., Porcionato, M.A., de Passille, A.M., Rushen, J., 2004. Cortisol in saliva and plasma of cattle after ACTH administration and milking. J. Dairy Sci. 87, 1713–1718. Orth, D.N., Peterson, M.E., Drucker, W.D., 1988. Plasma immunoreactive proopiomelanocortin peptides and cortisol in normal dogs and dogs with Cushing’s syndrome: diurnal rhythm and responses to various stimuli. Endocrinology 122, 1250–1262. Reber, A.S., 1988. Dictionary of Psychology. Penguin Books, London. Russell, E.S., 1936. Playing with a dog. Q. Rev. Biol. 11, 1–15. Tang, M.L., Ling, M.H., Tian, G.L., 2009. Exact and approximate unconditional confidence intervals for proportion difference in the presence of incomplete data. Stat. Med. 28, 625–641. Toates, F., 1980. Aggression and fear. In: Animal Behaviour—A Systems Approach. John Wiley and Sons, Chichester, pp. 193–220. Vincent, I.C., Michell, A.R., 1992. Comparison of cortisol concentrations in saliva and plasma of dogs. Res. Vet. Sci. 53 (3), 342–345. Waiblinger, S., Boivin, X., Pedersen, V., Tosi, M.-V., Janczak, A.M., Visser, K., Jones, R.B., 2006. Assessing the human–animal relationship in farmed species: a critical review. Appl. Anim. Behav. Sci. 101 (3–4), 185–242. Wells, D.L., Hepper, G.L., 2000. The influence of length of time spent in a rescue shelter on the behaviour of kennelled dogs. Anim. Welf. 11, 317–325. Wells, D.L., 2004. The influence of toys on the behaviour and welfare of kennelled dogs. Anim. Welf. 13 (3), 367–373. Wolfe, T.L., 1990. Policy, program and people: the three P’s to well-being. In: Mench, J.A., Krulisch, L. (Eds.), Canine Research Environment. Scientists Centre for Animal Welfare, Bethesda, pp. 41–47.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
ARTICLE IN PRESS Applied Animal Behaviour Science xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Applied Animal Behaviour Science journal homepage: www.elsevier.com/locate/applanim
Effects of human contact and toys on the fear responses to humans of shelter-housed dogs Melanie J. Conley a,∗ , Andrew D. Fisher b , Paul H. Hemsworth a a
Animal Welfare Science Centre Melbourne School of Land and Environment, University of Melbourne, Parkville, VIC 3010, Australia Animal Welfare Science Centre, Faculty of Veterinary Science, The University of Melbourne, 250 Princes Highway, Werribee, VIC 3030, Australia b
a r t i c l e
i n f o
Article history: Accepted 20 March 2014 Available online xxx
Keywords: Human interaction Stress Fear Dog Cortisol Animal shelter
a b s t r a c t This study examined the effects of human contact and toys on fear responses to humans in small breed, shelter-housed dogs. Ninety dogs were assigned to one of three treatments: “control” (control), comprising routine husbandry performed by shelter staff; “human contact” (HC), where dogs additionally received 2 min positive contact daily with an experimenter; or “human contact + toys” (HCT), where the additional human contact included the opportunity to interact with toys. Treatments were implemented daily from day 2 (second day in the shelter) until day 6. On day 7, the fear response towards the experimenter was assessed using salivary cortisol concentrations and a human avoidance test. The behavioural parameters measured were approach and withdrawal responses to the experimenter and time spent in each section of the pen while the experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). On day 8, dogs were assessed by the shelter veterinarian for their suitability for adoption by the public. Treatment had no effect on salivary cortisol concentrations or the proportion of dogs selected for adoption. The results indicated that dogs in the HC and HCT treatments spent more time (P = 0.024) at the front of the pen when the experimenter was at position 3 than control dogs (HC 8.8 s, HCT 8.7 s, control 6.6 s from a possible 10 s). For those dogs that were not at the front of the pen when the experimenter was in position 3, a higher (P < 0.05) percentage of HCT dogs (100%) but not HC dogs (86%, P > 0.05) approached the experimenter compared with control dogs (45%). A second experiment on another 40 dogs examined if dogs, based on their behaviour in the human avoidance test, discriminate between familiar and unfamiliar humans. All dogs received the HC protocol as in Experiment 1 and were observed in the human avoidance test, however, all dogs were tested twice, once with the ‘familiar’ experimenter and once with an ‘unfamiliar’ experimenter. There were no significant (P > 0.05) familiarity or testing order effects, except that dogs approached the unfamiliar experimenter more (60%) than the familiar experimenter (29%) at position 2 (P < 0.05). The results indicate that additional positive human contact other than that associated with routine husbandry reduced the behavioural fear response of dogs to humans, but did not affect the proportion of dogs selected for adoption. The results also indicate that the shelter dogs did not discriminate between familiar and unfamiliar humans, in the handling context used in this study, suggesting a degree of stimulus generalisation may occur. © 2014 Elsevier B.V. All rights reserved.
∗ Corresponding author. Tel.: +61 422141809. E-mail address: [email protected] (M.J. Conley). http://dx.doi.org/10.1016/j.applanim.2014.03.008 0168-1591/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
2
1. Introduction Fear can be considered as an undesirable emotional state of suffering that functions to protect the animal through defensive behaviour or escape and is elicited by the perception of actual or perceived danger (Toates, 1980; Jones and Waddington, 1992; Hemsworth and Coleman, 2011). The behavioural responses associated with fear include withdrawal, or avoidance responses, as well as immobility, such as freezing or crouching (Hemsworth and Barnett, 1987; Jones, 1987; Mills and Faure, 1990). The physiological responses associated with fear include responses of the sympatho-adrenal medullary system (e.g. secretions of adrenalin and noradrenalin) and the hypothalamo-pituitary adrenal system (HPA axis e.g. secretions of corticosteroid hormones, cortisol and corticosterone) (Hemsworth and Coleman, 2011). In Victoria, the Code of Practice for the Management of Dogs and Cats in Shelters and Pounds requires that dogs are housed during the first 8 days in shelters in quarantine in individual pens with no tactile contact with other dogs. This requirement often results in dogs also having no visual contact with other dogs during this period. Standard operating procedures in other Australian states impose similar “quarantine” periods, ranging from 3 days to a week in animal shelters. Surprisingly little research has been conducted on stress in shelter-housed dogs. Animal shelters are a stressful environment for dogs, possibly because of the novel surroundings, social restrictions and reduced human contact. Dogs have increased cortisol concentrations during the first 3 days in an animal shelter, declining slowly before reaching a plateau around day 9–10 (Hennessy et al., 1997). Even so, the authors reported that this plateau was above the average basal cortisol levels found in domestic dogs. An initial stress response can prepare an animal to adapt to a change in its environment; however, a longer-term activation of the stress response can lead to behavioural and physiological disturbances such as stereotypies, prolonged elevation of corticosteroids and immunosuppression (Johnson et al., 1992; Broom and Johnson, 1993), with welfare implications. Dogs are social animals, needing contact with both conspecifics (Fox, 1965; Fox and Stelzner, 1967) and, in dogs socialised to humans, contact with humans (Freedman et al., 1961; Wolfe, 1990). In the presence of a human, kennelled dogs typically become more active and spend more time at the front of their cage, reducing the human–animal distance and facilitating interaction (Campbell et al., 1988; Hughes et al., 1989; Wells and Hepper, 2000). Surveys have found that people looking to adopt a shelter dog prefer dogs located at the front of the cage rather than at the back, and dogs that are alert rather than non-alert (Wells and Hepper, 2000). The restricted human contact that typically occurs in shelters may increase fear of humans in shelter dogs and thus reduce their attractiveness to people looking to adopt a dog. While the history of interactions between a human and a domesticated animal will affect the behavioural response of the animal to humans, it has been suggested that the behavioural response of farm animals to an individual human, through stimulus generalisation, can extend to
other humans (Hemsworth and Coleman, 2011). Stimulus generalisation can be defined as a stimulus similar in nature to an original stimulus that produces the same behavioural response in a learning situation (Reber, 1988) but there has been very little research conducted on stimulus generalisation in dogs housed at a shelter. Hubrecht (1993, 1995) examined stimulus generalisation in laboratory-housed beagles and found that as little as 30 s of human handling daily for 2 months resulted in the beagles displaying more approach to both familiar and unfamiliar humans. Very little research has been conducted examining the effects of combining human contact with the opportunity to play in shelter dogs. The dog is one of few species to engage in interspecific play with humans (Russell, 1936). Introducing play into a shelter environment may be beneficial for socialisation and preparation for rehousing by providing appropriate dog–human relationships (Wells, 2004). The first experiment in this study examined the effects of daily human contact and the opportunity to interact with toys during the first week of housing in an animal shelter on the behaviour and welfare of small breed dogs. Furthermore, treatment effects on the percentage of dogs selected as suitable for adoption at the completion of the 8-day quarantine period was also examined. A second experiment in this study examined whether the effects of human contact were specific to the human that provided the contact, or were generalised to other humans. 2. Materials and methods 2.1. Ethical approval of animal experimentation The protocol and conduct of the study were approved by the University of Melbourne, Melbourne School of Land and Environment Animal Ethics Committee (Ethics # 0703341.1 and 0911103). 2.2. Animals and facilities The study was conducted at a large animal shelter in Victoria, Australia receiving at least 20,000 dogs per year. During the 8-day quarantine period, dogs at the shelter were individually housed in pens (approximately 1 m wide × 2.5 m long × 2.5 m high) consisting of a concrete floor with a raised bed at the back of the pen, solid walls and a wire mesh front gate. Dogs remained in their pens during the 8-day quarantine period and were visually isolated from all other dogs in the shelter. Routine care by shelter staff was restricted to morning cleaning, whereby dogs remained in their pens while the pens were hosed clean and food and water provided. Tactile contact with the dogs by the staff was rare. 2.3. Experiment 1 Ninety small breed dogs over a 7-week period were studied on admittance to the shelter. The dogs varied in age, sex and breed and their past human and housing experiences were unknown.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
2.3.1. Treatments and measurements Five rows of nine pens were used to house the study dogs with pens 2, 3, 5, 6, 8 and 9 in each row allocated. As dogs of a suitable size (small breed) entered the shelter (referred to as day 1), they were allocated to the next empty pen in a row. To deal with possible position effects within rows and between rows, dogs were allocated to treatment in a pseudo-randomised manner so that each pair of pens in each row was represented at least once by each treatment. The following treatments were imposed from days 2–6: 1. “Control”: routine cleaning and feeding was performed by shelter staff. 2. “Human Contact” (HC): in addition to the routine husbandry by staff, the experimenter (MJC) provided 2 min daily positive human contact. The experimenter crouched down inside the pen with her back against the front gate, then interacted with the dog by talking in soft tones, presenting her hand for the dog to sniff, and patted the dog if it approached. On completion of the 2 min, the experimenter quietly stood and exited the pen. 3. “Human Contact & Toys” (HCT): in addition to the routine husbandry, the experimenter imposed the HC treatment and at the same time also offered one of two toys for the dog to interact with during the 2 min period. On days 2 and 3, the toy was a rope toy, and for days 4, 5 and 6, the toy was a plush squeaky toy in the shape of a bone. After 2 min, the experimenter quietly exited the pen. Treatments were conducted daily from 07:00 h on weekends and 08:00 h on weekdays except for days in which saliva was collected and the behavioural response of dogs to the experimenter was observed (day 7, 06:00 h on weekends and 07:00 h on weekdays), when treatment imposition followed. After the imposition of all treatments, staff cleaned pens and fed the dogs. The accommodation area was locked for the day after feeding. 2.3.2. Saliva sampling Saliva samples were collected by the experimenter on day 7 with the dogs in their pens. When preparing to collect the saliva samples, any dogs considered likely to be aggressive, based on warning signs pinned on the pens by shelter staff or previously recorded behaviours by the experimenter such as baring of teeth and/or extreme submissive crouching, were not sampled. Seventy-four dogs from the 90 were sampled. Two saliva samples were collected 10 min apart from each dog tested on that day. The first was collected to estimate basal cortisol concentration and the second to estimate the cortisol response to human handling associated with the first sampling (i.e. the acute response to handling) (Broom and Johnson, 1993). Salivary cortisol provides an accurate estimate of plasma cortisol (Negrao et al., 2004; Vincent and Michell, 1992). An experienced volunteer aided the experimenter during the collection by holding the dog on their knee. All samples were collected using a salivette (Sarstedt Inc. Sarstedtstraße Postfach, Nümbrecht, Germany) and were collected within 3 min of the experimenter and volunteer entering the pen. Dogs were encouraged to chew the swab, as chewing increases salivation. The swab was
3
then placed into a labelled salivette tube and placed in a rack on ice. Within 2 h of collection, the salivettes were removed from ice and centrifuged at 3500 rpm at 10 ◦ C for 45 min. The centrifuged saliva was then transferred into labelled 2mL freezing tubes and stored at −20 ◦ C. Saliva cortisol was measured using a Human/Sheep Cortisol ELISA Quantification Kit (MyBioSource Inc., San Diego) that also measures cortisol concentrations in dog saliva (Dreschel and Granger, 2005). The collection and storage procedure has been successfully used to collect saliva from dogs (Dreschel and Granger, 2005). 2.3.3. Human avoidance test The behavioural response of the dogs to the experimenter was assessed on day 7, immediately following collection of the saliva samples. In this test, the experimenter placed a video camera (Panasonic PV-L352, Osaka, Japan) mounted on a tripod 2 m from the pen gate (position 1), pressed record, then removed herself from the dog’s sight for 10 s. After this time, the experimenter returned, stood next to the camera at position 1 for 5 s, took one step towards the pen gate to position 2 which was 1 m from the gate for 5 s, then took one step to the front of the pen gate where she crouched down for 10 s at position 3. At no point during the test did the experimenter make eye contact with the test dog. For this test, the pen was nominally sectioned into quarters (S1–S4, representing the four areas from the front of the pen) to measure the time spent in the four locations of the pen. The test was recorded in real time on video, read in real time and from the video recordings, the following was calculated for each dog when the experimenter was in each position: time spent in each area; and time to entry and exit from each area of the pen. Location in each area of the pen was assessed on the basis of the dog’s head location. In order to assess the approach response when the experimenter was in each position, those dogs not initially at the front of the pen when the experimenter was either in position 1, position 2 or position 3 were classified as either approaching within the 5- or 10-s observation period (first step taken was towards the front of the pen and the experimenter) or not (no step or the first step taken was towards the back of the pen). Data for dogs initially at the front of the pen that withdrew before the 5th or 10th s of the 5- or 10-s observation periods were also analysed. These dogs were classified as either approaching within the 5-s or 10-s observation period (after withdrawing, the first step taken was towards the front of the pen) or not (after withdrawing, no step forward was taken). For each position, those dogs initially at the front of the pen that did not withdraw before the 5th or 10th s of the 5or 10-s observation periods were excluded from analyses as they were not able to approach so would not be given a score to represent their true behaviour. The corresponding method was used for assessing withdrawals. For example, for each position, those dogs not initially at the back of the pen (e.g. anywhere else in the pen except the back) were classified as either withdrawing
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
4
Table 1 The effects of handling and time on dog salivary cortisol concentrations in Experiment 1. Sample 2 was taken 10 min after Sample 1. Means (and standard errors of the mean in parentheses) are presented. Item
Sample 1 (nmol/L) Sample 2 (nmol/L) Overall (nmol/L)
Treatment
Time
Control
HC
HCT
P
Sample 1
Sample 2
P
46.3 (15.37) 39.0 (11.70) 42.7 (13.09)
53.0 (14.19) 40.8 (10.80) 46.9 (12.08)
73.3 (15.05) 64.8 (11.45) 69.1 (12.81)
0.393 0.210 0.889
57.5 (8.59)
48.2 (6.54)
0.037
(first step taken within the 5- or 10-s observation period was towards the back of the pen) or not (no step taken for the 5- or 10-s observation period or the first step taken was towards the front of the pen). 2.3.4. Assessment of suitability of the dogs for adoption On completion of the 8-day quarantine period at the shelter, dogs that had not been collected by their owners underwent behaviour (e.g., aggression, fear) and health (e.g., old age, ongoing skin allergies) assessments by the shelter veterinarian, determining if these dogs were suitable for adoption by the public. Shelter records were collected on whether or not the study dogs were assessed as suitable for adoption. 2.3.5. Statistical analysis Some variables did not conform to a normal distribution as indicated by visual methods (Q–Q plots and histograms, SPSS statistical package, SPSS 17.0, SPSS Inc., Chicago, IL, USA). A Repeated Measures Analysis of Variance (SPSS) was conducted to examine the effects of time (sample) and treatment on salivary cortisol concentration. Univariate Analysis of Variance (SPSS) was used to examine effects of treatment, row and week on dog behaviour. Comparisons between treatments were tested using post hoc l.s.d. tests. To analyse approach responses, separate analyses were performed for dogs not initially at the front of the pen and for dogs initially at the front of the pen that first withdrew. Similarly, probabilities of withdrawing were analysed for dogs not initially at the back of the pen and for dogs initially at the back of the pen but then approached. Effects of treatment on approach and withdrawal responses were assessed using overall and pair-wise Fisher’s exact tests, calculated using the Compare2 module (version 2.75) in WinPepi version 11.25 (Abramson, 2011). As the data from the suitability for adoption assessments were not normally distributed, a non-parametric Kruskal–Wallis Analysis of Variance (SPSS) was used to examine treatment effects on the fate of the animals. 2.4. Experiment 2 Another 40 small breed dogs were used to determine if the dogs discriminated between familiar and unfamiliar humans. All 40 dogs were exposed to the same conditions as in treatment 2 of Experiment 1 for 5 days, with the same
experimenter exposing all 40 dogs to the human contact treatment in order to become the ‘familiar’ human.
2.4.1. Human avoidance test The same method used in Experiment 1 was implemented again in Experiment 2, however, all dogs were tested twice, once with the ‘familiar’ experimenter (MJC) who had been imposing the human contact, and once with an ‘unfamiliar’ experimenter with whom the dogs had not had contact. Both experimenters were female, similar in height and wore matching blue overalls. The order in which individual dogs were tested with the familiar and unfamiliar experimenters was randomised and the interval between the two tests was 30–40 min. For this experiment, to improve ease of video analysis, the pen was nominally sectioned into thirds to measure time spent in different locations in the pen. From the video observations, the time that each dog spent in each area and their approach and avoidance responses when the experimenter was in each position was calculated as in Experiment 1. 2.4.2. Statistical analysis As in Experiment 1, some variables did not conform to a normal distribution as indicated by visual methods (Q–Q plots and histograms, SPSS statistical package, SPSS 17.0, SPSS Inc., Chicago, IL, USA). Analysis of Variance (SPSS) was used to examine the experimenter effects and testing order effects on dog behaviour. To analyse approach behaviours (as in Experiment 1), separate analyses were performed for dogs not initially at the front of the pen and for dogs initially at the front of the pen that first withdrew. Similarly, probabilities of withdrawing were analysed for dogs not initially at the back of the pen and for dogs initially at the back of the pen that first approached. Effects of experimenter (‘familiar’ or ‘unfamiliar’) and testing order were assessed using a method for partially (i.e. incompletely) paired data. The data were partially paired because some dogs contributed data for both experimenters while others contributed data for only one of the two experimenters, as dogs that were initially at the front of the pen that never withdrew to allow an approach, and dogs that were initially at the back of the pen and never approached to allow a withdrawal could not contribute data. We used the method reported by Tang et al. (2009), implemented using the PAIRS etc. module (version 3.13) in WinPepi. This method assumed that exclusions were independent of experimenter.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
Test phase Position 1 Treatment
xy
HC
HCT
3.9 (0.37) 0.0 (0.19) 0.2 (0.10) 1.0 (0.32)
4.3 (0.04) 0.3 (0.19) 0.0 (0.10) 0.3 (0.32)
3.8 (0.37) 0.4 (0.19) 0.0 (0.10) 0.8 (0.33)
P 0.488 0.306 0.410 0.297
Position 3
Control
HC
HCT
P
3.3 (0.37) 0.6 (0.21) 0.0 (0.06) 1.0 (0.32)
4.4 (0.36) 0.2 (0.20) 0.1 (0.06) 0.3 (0.32)
3.8 (0.37) 0.3 (0.21) 0.1 (0.06) 0.8 (0.33)
0.110 0.349 0.815 0.285
Control
HC
HCT
P
6.6x (0.64) 0.3 (0.21) 0.8 (0.25) 2.3 (0.56)
8.8y (0.63) 0.4 (0.21) 0.1 (0.25) 0.7 (0.55)
8.7y (0.65) 0.3 (0.22) 0.1 (0.26) 0.9 (0.57)
0.024 0.965 0.089 0.093
Means without a common superscript are significantly different (P < 0.05).
Table 3 The effects of handling on the approach and withdrawal responses to humans of dogs in the human avoidance test in Experiment 1. The experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented. Item
Test phase Position 1 Treatment
Approached (not initially at front) (%) Approached (initially at front) (%) Withdrew (not initially at front) (%) ab
Position 2
Position 3
Control
HC
HCT
P
Control
HC
HCT
P
13 (8) 33 (3) 13 (23)
38 (8) 0 (1) 3 (29)
14 (7) 100 (2) 15 (26)
0.557 0.400 0.262
20 (10) 67 (3) 22 (23)
44 (9) 67 (3) 21 (29)
33 (9) 0 (2) 24 (25)
0.536 45 (11)a 0.486 50 (4) 1.000 35 (23)
Control
HC 86 (7)ab 100 (2) 14 (29)
HCT 100 (8)b 100 (1) 12 (26)
P 0.025 0.619 0.110
Means without a common superscript are significantly different (P < 0.05).
ARTICLE IN PRESS
Time spent in S1 (s) Time spent in S2 (s) Time spent in S3 (s) Time spent in S4 (s)
Position 2
Control
G Model
APPLAN-3883; No. of Pages 8
Item
M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
Table 2 Effects of handling on the location of dogs in their pens during the human avoidance test in Experiment 1. S1 represents the section closest to the front of the pen and S4 being farthest from the front of the pen. The experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented.
5
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
6
Table 4 Results examining treatment effects on the assessment of suitability of the dogs for adoption in Experiment 1. Item
Proportion of dogs euthanased for behavioural reasons (%) Proportion of dogs euthanased for health reasons (%) Proportion of dogs assessed as suitable for adoption (%)
Treatment Control
HC
HCT
Chi-square
P
32 29 39
32 23 45
34 35 31
0.044 1.033 1.251
0.978 0.596 0.535
3. Results 3.1. Experiment 1 There were no significant (P > 0.05) effects of pen location (block) or week of the study on the behavioural or physiological variables. 3.1.1. Cortisol concentrations There was a significant (P < 0.05) effect of time but not of treatment (P > 0.05) on salivary cortisol concentrations, with lower cortisol concentrations in sample 2 (48.2) than in sample 1 (57.5) across all treatments (Table 1). 3.1.2. Human avoidance test The effects of treatment on dog behaviour are shown in Table 2. When the experimenter was in positions 1 and 2 in the human avoidance test, there were no significant (P > 0.05) treatment effects on the total time dogs spent in any quarter of the pen. However, when the experimenter was in position 3 of the test, dogs in the HC and HCT treatments spent more time (P < 0.05) in the front quarter of the pen (S1) than control dogs. The effects of treatment on approach and withdrawal responses are shown in Table 3. For dogs that were not initially at the front of the pen when the experimenter was in position 3, a higher (P < 0.05) percentage of HCT dogs approached than control dogs. There were no significant (P > 0.05) treatment effects on withdrawal responses. 3.1.3. Assessment of suitability of the dogs for adoption Treatment did not significantly (P < 0.05) influence the percentages of dogs assessed as unsuitable for adoption based on behavioural or health reasons, or those assessed as suitable (Table 4). 3.2. Experiment 2 There were no significant (P > 0.05) effects of pen location (block) or week of the study on the behavioural variables. Furthermore, there were no significant (P > 0.05) familiarity or testing order effects on the time dogs spent in the three areas of the pen during the human avoidance test (Table 5). The effects of a familiar and unfamiliar experimenter on approach and withdrawal response are shown in Table 6. While there were no effects of familiarity when the experimenter was in positions 1 and 3, a higher (P < 0.05) percentage of dogs approached the unfamiliar experimenter than the familiar experimenter during position 2.
There were no significant (P > 0.05) experimenter effects on withdrawal, and no significant (P > 0.05) testing order effects on approach or withdrawal responses.
4. Discussion The results of Experiment 1 indicate that dogs that receive positive human contact and positive human contact and toys increased the time that they spent in the front quarter of the pen (S1) when the experimenter was very close to the front of the pen (position 3 of the human avoidance test) in comparison to the control dogs. Furthermore, for dogs that were not at the front of the pen when the test commenced, more of the dogs that received positive human contact and toys approached the experimenter than the control dogs. The dogs that received human contact were intermediate. There were no treatment effects on the location of the dogs in the pen when the experimenter was 1 and 2 m from the front of the pen. Avoidance of an approaching experimenter and approach to a stationary experimenter are two of the most common and well-accepted measures used to assess fear of humans in domestic animals (Waiblinger et al., 2006; Hemsworth and Coleman, 2011) and thus in the human avoidance test, dogs that are less fearful of humans would be expected to show more approach to the experimenter and consequently spend more time at the front of the pen, particularly when the experimenter is stationary and close to the front of the pen. While there is evidence of reduced fear of humans in dogs that received positive human contact and positive human contact and toys, this evidence is limited. The treatment effects on time spent in the front of the pen when the experimenter was very close were relatively small and only the treatment consisting of positive human contact and toys increased approach in those dogs that were not at the front of the pen when the test commenced. However, it is important to recognise that this latter variable applies only to a subset of dogs, that is those that were not initially at the front of the pen, and this result needs to be interpreted cautiously. A necessary limitation of the study is that some of the obvious factors that may markedly affect the opportunity for brief imposition of the two intervention treatments to affect fear, such as previous human and housing experiences of the dogs, were not controlled. Further research on large numbers of shelter dogs on the effects of positive human contact and the opportunity to interact with toys on the behavioural response of dogs is warranted because of the implications on fear and welfare of shelter dogs.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS 0.394 0.359 0.752 8.3 (0.47) 0.9 (0.31) 0.8 (0.39)
Test order 1 Test order 2 P
8.9 (0.47) 0.5 (0.31) 0.6 (0.39) 0.629 0.111 0.499 8.8 (0.47) 8.5 (0.47) 0.4 (0.31) 1.1 (0.31) 0.9 (0.39) 0.5 (0.39) 0.127 0.077 0.641 3.8 (0.28) 0.6 (0.18) 0.6 (0.23) 4.4 (0.28) 0.2 (0.18) 0.4 (0.23) 0.798 0.428 0.353 4.1 (0.28) 4.2 (0.28) 0.3 (0.18) 0.5 (0.18) 0.6 (0.23) 0.3 (0.23) 0.346 0.449 0.582 4.0 (0.28) 0.5 (0.19) 0.6 (0.22) 4.4 (0.28) 0.3 (0.19) 0.4 (0.22) 0.950 0.570 0.582
Unfamiliar P
Time spent in S1 (s) 4.2 (0.28) 4.2 (0.28) Time spent in S2 (s) 0.3 (0.19) 0.4 (0.19) Time spent in S3 (s) 0.6 (0.22) 0.4 (0.22)
Familiar Test order 1 Test order 2 P Familiar
Familiar
Position 2
Treatment Test order Treatment
Unfamiliar P
Test order 1 Test order 2 P
Position 3
Test order
Treatment
Unfamiliar P
Test order
7
Position 1
Test phase Item
There were no treatment effects on salivary cortisol concentrations, however, there was a significant time effect. In the present study, cortisol concentrations overall were lower in sample 2 than sample 1. Handling associated with the first saliva sample was expected to result in a shortterm increase in cortisol concentration within 5–15 min (Hemsworth et al., 1981), particularly in fearful dogs. One explanation is that differences in the fear responses between treatments were insufficient to produce treatment effects on the cortisol response to handling. Furthermore, there is evidence that the secretion of cortisol in dogs coincides with activity (Kempanien and Sartin, 1984; Orth et al., 1988; Castillo et al., 2009). In the present study, the dogs may have been aroused at the sound of a human entering the pen blocks and perhaps expected to be fed, which may have masked any treatment effects of fear of humans. There were no treatment effects on dogs assessed as suitable for adoption. Luescher and Medlock (2009) found that trained dogs available for adoption were 1.4 times more often adopted than untrained dogs. These dogs received one 20 min training session daily over a 2 week period where they were desensitised to a Gentle Leader, taught to come forward in their cage when approached, walk on a leash, sit and to not jump on people. In the present study, treatment effects may not have directly impacted on the dog’s behaviour during assessment, however, the intervention treatments may have enhanced the behaviour of dogs that were always destined to pass, but may not have had sufficient effect on those that were always destined to fail. There were a number of dogs euthanased for health reasons, which may have resulted in an insufficient sample size to demonstrate an effect on suitability of dogs for adoption. The results of Experiment 2 indicate that there were no effects of familiarity with the experimenter on the location of the dogs in the pen when the experimenter was close to the front of the pen or 1 and 2 m from the front of the pen. There were also no effects of familiarity with the experimenter on the avoidance responses of the dogs that were not initially at the front of the pen. However, for dogs that were not initially at the front of the pen when the experimenter was 1 m from the front of the pen, more dogs approached the unfamiliar experimenter than the familiar experimenter. There is no obvious explanation for this effect since there were no treatment effects when the experimenter was either 2 m from the pen or very close to the pen. As discussed earlier in Section 4, dogs that are less fearful of humans would be expected to show more approach to the experimenter and dogs that are more fearful would be expected to show the opposite. Thus these results suggest that, in the handling context used in this study, a degree of stimulus generalisation in shelter housed dogs may occur whereby the response of an animal towards a familiar human who administered positive handling produced the same or a similar response to an unfamiliar human with the same physical characteristics. Discrimination between people by domesticated animals is obviously easier if the animals have some distinct cues on which they can discriminate, such as size, behaviour, colour of clothing or location of handling (Hemsworth and Coleman, 2011).
Table 5 The effects of unfamiliar and familiar testers and testing order on the location of dogs in the human avoidance test in Experiment 2. S1 represents the section closest to the front of the pen and S3 being farthest from the front of the pen. The familiar or unfamiliar experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented.
M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008
G Model APPLAN-3883; No. of Pages 8
ARTICLE IN PRESS M.J. Conley et al. / Applied Animal Behaviour Science xxx (2014) xxx–xxx
8
Table 6 The effects of familiar and unfamiliar experimenters on the approach and withdrawal responses to humans of dogs in the human avoidance test in Experiment 2. The Familiar or Unfamiliar experimenter was situated at each of three positions outside the pen: 2 m (position 1) or 1 m away from the pen gate (position 2) or crouched against the pen gate (position 3). Means (and standard errors of the mean in parentheses) are presented. Item
Test phase Position 1
Approached (not initially at front) (%) Withdrew (not initially at front) (%)
Position 2
Position 3
Familiar
Unfamiliar
P
Familiar
Unfamiliar
P
Familiar
Unfamiliar
P
15 (13) 3 (35)
7 (15) 5 (37)
0.451 0.592
29 (14) 29 (35)
60.0 (15) 27 (37)
0.016 0.858
75 (16) 40 (35)
88 (17) 45 (38)
0.263 0.15
5. Conclusion The present study provides limited evidence that positive human contact may reduce the fear response of shelter dogs towards humans during the first week of admittance to a shelter. However, further research is required to determine the effects of human handling at a shelter on dog welfare. Acknowledgements The technical support of M. Rice and L. Edwards, John Morton for assisting with the statistical analysis, Sally Haynes for comments on the draft manuscript and the participation of the shelter staff are gratefully acknowledged. References Abramson, J.H., 2011. WINPEPI updated: computer programs for epidemiologists, and their teaching potential. Epidemiol. Perspect. Innov. 8, 1. Broom, D.M., Johnson, K.G., 1993. Stress and Animal Welfare. Chapman and Hall, London, UK. Campbell, S.A., Hughes, H.C., Griffin, H.E., Landi, M.S., Mallon, F.M., 1988. Some effects of limited exercise on purpose bred Beagles. Am. J. Vet. Res. 49, 1298–1301. Castillo, V.A., Cabrera Blatter, M.F., Gomez, N.V., Sinatra, V., Gallelli, M.F., 2009. Diurnal variations in healthy dogs and those with pituitarydependent Cushing’s syndrome before and after treatment with retinoic acid. Res. Vet. Sci. 86 (2), 223–229. Dreschel, N.A., Granger, D.A., 2005. Physiological and behavioral reactivity to stress in thunderstorm-phobic dogs and their caregivers. Appl. Anim. Behav. Sci. 95 (3-4), 153–168. Fox, M.W., 1965. Environmental factors influencing stereotyped and allelomimetic behaviour in animals. Lab. Anim. Care 15, 363–370. Fox, M.W., Stelzner, D., 1967. The effects of early experience on the development of inter and intraspecies social relationships in the dog. Appl. Anim. Behav. Sci. 15, 377–386. Freedman, D.G., King, J.A., Elliot, O., 1961. Critical periods in the social development of the dog. Science 133, 1016. Hemsworth, P.H., Barnett, J.L., Hansen, C., 1981. The influence of handling by humans on behaviour, growth and corticosteroids in the juvenile female pig. Horm. Behav. 15, 396–403. Hemsworth, P.H., Barnett, J.L., 1987. Human–animal interactions. Veterinary clinics of North America. Food Anim. Pract. 3, 339–356. Hemsworth, P.H., Coleman, G.J., 2011. Human–Livestock Interactions. The Stockperson and the Productivity and Welfare of Intensively Farmed Animals. CAB International, Wallingford, UK. Hennessy, M.B., Davis, H.N., Williams, M.T., Mellott, C., Douglas, C.W., 1997. Plasma cortisol levels of dogs at a county animal shelter. Physiol. Behav. 62, 485–490.
Hubrecht, R.C., 1993. A comparison of social and laboratory environmental enrichment methods for laboratory housed dogs. Appl. Anim. Behav. Sci. 37, 345–361. Hubrecht, R.C., 1995. Enrichment in puppyhood and its effects on later behaviour of dogs. Lab. Anim. Sci. 45, 70–75. Hughes, H.C., Campbell, S., Kenney, C., 1989. The effects of cage size and pair housing on exercise of Beagle dogs. Lab. Anim. Sci. 39, 302–305. Johnson, E.O., Kamilaris, T.C., Crousos, G.P., Gold, P.W., 1992. Mechanisms of stress: a dynamic overview of hormonal and behavioral homeostasis. Neurosci. Biobehav. Rev. 16, 115–130. Jones, R.B., 1987. Fear and fear responses: a hypothetical consideration. Med. Sci. Res. 15, 1287–1290. Jones, R.B., Waddington, D., 1992. Modifications of fear in domestic chicks Gallus gallus domesticus, via regular handling and early environmental enrichment. Anim. Behav. 43, 1021–1033. Kempanien, R.J., Sartin, R.L., 1984. Evidence for episodic but not circadian activity in plasma concentrations of adrenocorticotrophin, cortisol and thyroxine in dogs. J. Endocrinol. 103, 219–225. Luescher, A.U., Medlock, 2009. The effects of training and environmental alterations on adoption success of shelter dogs. Appl. Anim. Behav. Sci. 117, 63–68. Mills, A.D., Faure, J.M., 1990. Panic and hysteria in domestic fowl: a review. In: Zayan, R., Dantzer, R. (Eds.), Social Stress in Domestic Animals. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 248–272. Negrao, J.A., Porcionato, M.A., de Passille, A.M., Rushen, J., 2004. Cortisol in saliva and plasma of cattle after ACTH administration and milking. J. Dairy Sci. 87, 1713–1718. Orth, D.N., Peterson, M.E., Drucker, W.D., 1988. Plasma immunoreactive proopiomelanocortin peptides and cortisol in normal dogs and dogs with Cushing’s syndrome: diurnal rhythm and responses to various stimuli. Endocrinology 122, 1250–1262. Reber, A.S., 1988. Dictionary of Psychology. Penguin Books, London. Russell, E.S., 1936. Playing with a dog. Q. Rev. Biol. 11, 1–15. Tang, M.L., Ling, M.H., Tian, G.L., 2009. Exact and approximate unconditional confidence intervals for proportion difference in the presence of incomplete data. Stat. Med. 28, 625–641. Toates, F., 1980. Aggression and fear. In: Animal Behaviour—A Systems Approach. John Wiley and Sons, Chichester, pp. 193–220. Vincent, I.C., Michell, A.R., 1992. Comparison of cortisol concentrations in saliva and plasma of dogs. Res. Vet. Sci. 53 (3), 342–345. Waiblinger, S., Boivin, X., Pedersen, V., Tosi, M.-V., Janczak, A.M., Visser, K., Jones, R.B., 2006. Assessing the human–animal relationship in farmed species: a critical review. Appl. Anim. Behav. Sci. 101 (3–4), 185–242. Wells, D.L., Hepper, G.L., 2000. The influence of length of time spent in a rescue shelter on the behaviour of kennelled dogs. Anim. Welf. 11, 317–325. Wells, D.L., 2004. The influence of toys on the behaviour and welfare of kennelled dogs. Anim. Welf. 13 (3), 367–373. Wolfe, T.L., 1990. Policy, program and people: the three P’s to well-being. In: Mench, J.A., Krulisch, L. (Eds.), Canine Research Environment. Scientists Centre for Animal Welfare, Bethesda, pp. 41–47.
Please cite this article in press as: Conley, M.J., et al., Effects of human contact and toys on the fear responses to humans of shelter-housed dogs. Appl. Anim. Behav. Sci. (2014), http://dx.doi.org/10.1016/j.applanim.2014.03.008