**Salivary Vasopressin as a Potential Non–Invasive Biomarker of Anxiety in Dogs Diagnosed with Separation–Related Problems**

### **Federica Pirrone 1,\*, Ludovica Pierantoni 2, Andrea Bossetti 1, Stefania Uccheddu <sup>3</sup> and Mariangela Albertini <sup>1</sup>**


Received: 24 October 2019; Accepted: 19 November 2019; Published: 26 November 2019

**Simple Summary:** Oxytocin and vasopressin have been shown to have opposite effects on the expression of anxiety and fear responses in rodents. In the present study, we analyzed the salivary fluctuations of these neuropeptides in both behaviorally normal dogs and dogs with separation distress in response to a three-minute separation from the owner, in a new environment. Dogs with a previous diagnosis of separation distress showed more anxiety-related behaviors and higher concentrations of vasopressin than control dogs when separated from the owner. Further research is needed on the potential use of salivary vasopressin as an early, non-invasive biomarker of anxiety-related disorders in pet dogs.

**Abstract:** Physiological biomarkers of canine anxiety have not been extensively investigated to date. To identify new biomarkers in dogs, we compared behaviorally normal dogs (Control group, N = 13) to dogs diagnosed with separation problems (Case group, N = 13) as they were introduced into a novel environment in the presence of two strangers and subjected to a short episode of separation and reunion with the owner. During the separation phase, dogs in the Case group explored significantly less than controls and were significantly more persistent in expressing passive stress-coping strategies aimed at seeking proximity to their owners. When the owners returned, dogs with separation distress spent significantly more time jumping up on the strangers than control dogs did. Salivary oxytocin and vasopressin concentrations did not differ between samples taken before and after the separation. However, vasopressin concentrations immediately after separation were significantly higher in the Case than in the Control group and remained higher, although not significantly so, 10 min later. These results indicated that dogs with separation distress became more anxious than typical dogs when separated from their owner in an unfamiliar environment and provided preliminary support for the use of salivary vasopressin as a possible biomarker for anxiety-related responses in dogs.

**Keywords:** dogs; separation distress disorder; vasopressin; oxytocin; anxiety

#### **1. Introduction**

Both humans and dogs are highly social mammals who develop selective forms of sociality, in which are embedded lasting relationships defined as social bonds or attachments [1]. An attachment bond may also link dogs and owners [2], the latter becoming the animals' reference point in the environment, influencing their welfare [3]. The attachment relationship between dogs and owners is one of the primary reasons why people keep dogs as companion animals [4].

Studies throughout the last decade have explored dogs' attachment-related behavioral reactions to separation from and reunion with the owner [2,3,5], while little is known about concomitant physiological reactions. Physiological changes, which include increased secretion of glucocorticoids and/or adrenocorticotropic hormone (ACTH), as well as altered cardiovascular or immune parameters, have been described in dogs during laboratory testing [4,6]. Increased circulating levels of nerve growth factor (NGF) have been directly associated with psychosocial stress in human and animal models [7,8]. However, in both rodents and humans, the closely related neuropeptides oxytocin (OT) and vasopressin (AVP) have been associated with distinct, partly opposite roles in social behavior, as well as in stress, fear, and anxiety responsiveness following social separation [9–11]. In particular, OT has been found to attenuate anxiety, central fear responses, and neuroendocrine reactivity [12], while stimulation of the AVP system has been shown to lead to augmented anxiety and fear expression [10] and increased neuroendocrine stress response [12].

Recently, some studies have begun to focus on OT in affiliative interactions in dog-human dyads [13,14]. We are also aware of two studies investigating the relationships between AVP and behavior in dogs, both of which revealed positive associations with fear and aggression [15,16].

To the best of our knowledge, there are no published data on the roles of these neuropeptides in anxiety-like behavior in dogs.

This study sought to examine the behavior and OT and AVP fluctuations in dogs in response to a commonly occurring human-dog interaction that might potentially induce anxiety in dogs. Specifically, we compared two groups of pet dogs that were separated from their owners for 3 min and then reunited with them, all while they were in a novel environment and the presence of two unfamiliar people. The Control group consisted of behaviorally normal dogs, while dogs diagnosed with separation-related problems (SRP) formed the Case group. Separation-related problems are described as physical, physiological, and/or behavioral signs of the distress exhibited by the dog only in the absence of, or lack of access to, the owner [17].

Approximately 14%–20% of dog patients [18,19] from general veterinary practices show signs of SRP in their owners' absence [20], and the anxiety emotional system is recognized as one possible cause. However, as with other anxiety disorders in veterinary patients, underdiagnosis of anxiety-dependent separation problems has been previously reported [21] due to misinterpretation of normal and pathological anxiety [22] and the lack of recognition of mild clinical signs by owners [23]. Therefore, unsurprisingly, according to Karagiannis et al. [24], it is suspected that up to 50%–56% of the overall dog population may actually display clinical symptoms of SRP at some point in their life [25,26], which, out of the total population of approximately 160 million dogs in the US and Europe [27,28] represents approximately 85 million dogs with SRP on the two continents. Underdiagnosis of SRP suggests a need to identify putative measurable markers that are specific to these behaviors, which would allow early diagnosis and intervention. Since separation-related problems are one of the primary cited reasons for the relinquishment of dogs to animal shelters [29,30], the relevant practical implications make research in this direction unquestionably worthwhile.

To assess the levels of relevant hormones, we examined their concentrations in saliva. A recent study by MacLean [31] validated salivary measures of OT and AVP in dogs. Because saliva collection is simple and painless, it is better suited for assessments in dogs than other methods, such as using plasma or urine, which present a host of challenges related to design, analysis, and welfare [31]. However, it is currently unknown whether these measures respond dynamically to anxiety-related aspects of human-dog interaction.

From previous findings [4], we hypothesized that separation from the owner in our experimental setting would not lead to different behavioral responses from dogs in the two groups. In fact, using an experimental protocol similar to ours, Parthasarathy and Crowell–Davis [4] found that, regardless of their separation anxiety status, all of the dogs displayed behavioral signs of increased anxiety when their owners left the room. Conversely, from observations in rodent and human models, we expected differences in endocrine responses during the separation phase, with a lower level of OT and a higher concentration of AVP in the Case group than in the Control group.

#### **2. Materials and Methods**

The Animal Welfare Committee (OPBA) of the University of Milan approved this research (OPBA\_106\_2018). All methods were performed in accordance with the current European and Italian legislation.

#### *2.1. Participants and Study Setting*

This study was part of a research project designed to discover novel potential biomarkers of stress and anxiety in pet dogs brought by their owner to the Veterinary Behavior and Consulting Services at CAN (Comportamento Animale Napoli) Training Center in Naples, Italy, for signs of fear or anxiety-related conditions. Initially, the owners of all dogs were asked, through a brief telephone interview, the reason for requesting the behavioral visit. Dogs were then tested, and, for this study, 13 dogs were selected by simple random sampling from those who had SRP according to the owners and received a diagnosis that confirmed SRP based on the outcome of the behavioral visit (Case group).

We also selected 13 controls from dogs whose owners attended the training center for the first time to get information on the service of dog's night board and daycare. The target sample size for the study was determined on the basis of a power analysis with 80% power, a large effect size (0.50), and an α value (error rate) of 0.05. The selected controls, who were rated as behaviorally typical based on the outcome of the visit purposely performed to include them in the present study, were similar to the cases with respect to the owners' gender and the dogs' sex, age, origin, and breed type (Supplementary Table S1). The dogs were 12 females (of which 4 were spayed), and 14 males (of which 6 were neutered), either pure or mixed breed. The mean age of the dogs, in years, was 3.7 (SE = 0.4, range: 1–8). Moreover, cases and controls had belonged to their current owners for at least one year to ensure that they had a reasonable amount of time to form a relationship with the owners and were an adult at the time of the study (1–10 years). All dogs underwent a physical examination. The Case group also had blood tests done. Those who were healthy for the veterinarian and not in treatment for behavioral or physical problems at the time of data collection were included in the study. Exclusion criteria included estrus, pregnancy, and nursing, as well as a history or present signs of stranger-directed aggression or fear of strangers or novel environments.

The experimenter explained the general objective of the research to all owners, who signed written informed consent and provided their assent to participate in the study. By agreeing to participate, owners confirmed that they were informed of a cameraperson's presence for the videotaping procedure and acknowledged that, in compliance with the General Data Protection Regulation (GDPR) (EU) 2016/679, their data and video recordings would be stored on the principal investigator's computer at the Department of Veterinary Medicine at the University of Milan and otherwise processed by the investigators of this study, with all ensuring an adequate level of data protection. Informed consent was obtained from the owners of dogs whose information, images, and/or videos would be published in an online journal, which could lead to their identification. Testing took place during the behavioral visit in a 300 m2 outdoor arena (Figure 1), containing three chairs (for the owner, the veterinary behaviorist, and a cameraperson), a bowl for fresh water, and some toys. The arena had four sides and was enclosed with chain-link fencing approximately 2 m tall; one side of the arena included an entrance gate. The duties of the veterinary behaviorist and the cameraperson were always performed by the same two women, who had never met the dogs before. The arena was located in a larger area. For standardization, and to minimize distractions and background noises, which could have acted as confounders, all dogs were tested in mild weather conditions and the late afternoon (between 1630 and 1830), after the center's planned closure, so no other social stimuli were present.

**Figure 1.** The spatial arrangement of the test. A moment during the session with one of the dogs. The owner and the two strangers (the veterinary behaviorist and the camerawoman) are visible. Drawing by Valentina Sammartano.

#### *2.2. Study Protocol*

Owners were instructed not to provide their dogs with food or exercise 1.5 hours prior to the start of the test. A simplified version of the Ainsworth's strange situation test (ASST) was employed. During the test, the owner and veterinary behaviorist could talk to each other and interact with the dog only if he/she was seeking their attention. The cameraperson was instructed not to interact with the dog or with the veterinary behaviorist and owners to avoid reinforcing attention-seeking behaviors and to control for the possibility that the strangers would act differently around different dogs [4]. The owners were told that if at any point they were concerned about their dogs' welfare or safety, they could stop the test. The entire procedure comprised three steps: An acclimatization phase, a separation phase, and a reunion phase.

*Acclimatization phase (10 min).* Prior to the short separation task, the owner, the dog, the veterinary behaviorist, and the camerawoman entered the arena. The dog was left unleashed and free to explore the environment, while the owner and the strangers remained seated in the chairs. Saliva was collected from the dog at the end of the tenth minute (T0).

*Separation phase (3 min).* Immediately afterward, the owner left the arena, where the dog remained in the company of the two strangers. During the ASST, the veterinary behaviorist attempted to engage the dog in friendly interaction, including gently petting him/her and speaking to him/her in a calm tone (Supplementary Video S1). However, the veterinary behaviorist allowed the dog to lead these interactions, and dogs were always free to disengage and move away from her. If a dog exhibited signs of severe distress or anxiety, the owner was asked to come back, and the testing stopped (Supplementary Video S2). At the end of the third minute, the owner returned to the arena, sat in the same chair as before, and made conversation with the veterinary behaviorist, and the second sample of saliva was collected (T1).

*Reunion phase (10 min).* Immediately upon returning, the owner was allowed to respond to his/her dog's greeting by interacting both verbally and physically in a calm way. After ten minutes, saliva was collected (T2), and the test ended.

#### *2.3. Parameters Recorded: Behavioral Responses*

Observer-blind analysis of behavior was carried out with focal animal sampling and continuous recording using the Observer XT software package (Noldus Information Technology, 6702 EA Wageningen, The Netherlands). Another coder, expert in animal behavior but unfamiliar with the aims and conditions of the study, verified the reliability of the coding in 20% of the videos. A Cronbach's α of 0.80 or higher was considered acceptable for this study. According to a study by Mariti et al. [2], the behaviors were divided into social and non–social, and each social behavior towards the owner and the strangers was analyzed (Table 1). Behavior definitions were formulated on the basis of a literature review [2–4,32–37]. As for social behaviors, attention-seeking and proximity were grouped to create the category spontaneous interactions. During the separation phase, we also recorded the interactions solicited by the veterinary behaviorist from the dogs, which included talking to and petting the dogs to comfort them if they showed signs of distress. Behavioral variables were measured in terms of relative frequency (the number of occurrences per minute) and/or duration (time spent on a behavior, expressed in seconds) of occurrence during each observation period.


**Table 1.** List of behaviors and definitions used in the study. F = frequency (number of occurrences); D = duration (s).

#### *2.4. Parameters Recorded: Endocrine Responses*

We collected saliva samples from dogs using commercially available swabs (SalivaBio Children's Swab, Salimetrics, Carlsbad, CA, USA). All samples were taken by the veterinary behaviorist. The swab was gently placed into the cheek or under the tongue of the dog for approximately 60 s, without the restraint of the animal. The dog's salivation was stimulated by presenting the odor of food treats. The dog received a food treat only after the last saliva sample was taken because the consumption of

food immediately before sample collection has been found to affect OT and AVP measurements [31]. Each sample was replaced in the device tube and closed with a plastic stopper to prevent evaporation. The collected material was refrigerated at 4 ◦C and then stored at −20 ◦C immediately after it arrived at the laboratory. At the time of analysis, the samples were thawed at room temperature and centrifuged according to the protocol for salivary samples. The laboratory technician who performed these analyses was blinded to the hypotheses and conditions. All samples were analyzed by enzyme-linked immunosorbent assay (ELISA) following previously validated protocols [31,34]. For the measurement of OT and AVP, we used commercially available enzyme-linked immunosorbent assay kits from Arbor Assays (Ann Arbor, MI, USA) and MyBiosource Inc. (San Diego, CA, USA). Each sample was prepared in duplicate, and concentrations were calculated using a Labisystem Multiskan Ex (Nepean, ON, Canada) microplate reader according to the relevant standard curves.

The mean recovery was 102.8% ± 10.8 for OT and 94.3 ± 2.2% for AVP. The average intra- and inter-assay coefficients of variation, respectively, were 4.7% and 8.8% for OT and 5.7% and 6.5% for AVP. The assay sensitivity was 17 pg/mL and 1 pg/mL for OT and AVP, respectively.

The laboratory technician was blinded to the hypotheses and conditions.

#### *2.5. Statistical Analysis*

Due to the number of animals and the distribution of the data, non-parametric statistics were used to analyze the behavioral and hormonal data [38,39]. Differences between groups were analyzed using the Mann–Whitney *U* test, while the Kruskal–Wallis test for multiple comparisons was conducted to compare behaviors within each group. A post hoc Mann–Whitney *U* test with the Bonferroni correction followed the Kruskal–Wallis test in case a significant effect was detected. The Friedman test for paired samples was used to test the difference in endocrine parameters among time points. In addition, OT and AVP concentrations in the two groups of dogs were compared using Mann–Whitney *U* tests. The OT and AVP concentrations and the duration and relative frequency of behaviors were presented as median. *p*-values ≤ 0.05 were deemed statistically significant. Statistical analyses were performed with IBM SPSS Statistics 25.0.

#### **3. Results**

#### *3.1. Behavioral Responses*

Intra-observer and inter-observer reliability were confirmed, with a Cronbach's α of 0.995 and 0.997, respectively. No significant differences were found in behavioral responses during the acclimatization phase between groups (Mann–Whitney *U* test, *p* > 0.05). As shown in Figures 2 and 3, in this phase, attention directed towards the fence and exploring were the most frequent behaviors and had the greatest total duration in both Case (duration: Kruskal–Wallis test, χ<sup>2</sup> = 52.803, *p* = 0.001; relative frequency: Kruskal–Wallis test, χ<sup>2</sup> = 93.552, *p* = 0.001) and Control dogs (duration: Kruskal–Wallis test, χ<sup>2</sup> = 37.895, *p* = 0.001; relative frequency: Kruskal–Wallis test, χ<sup>2</sup> = 97.349, *p* = 0.001). Behaviors oriented towards the fence and standing by the fence were not sufficiently expressed to be analyzed in this phase.

**Figure 2.** Duration and relative frequency, expressed as the number of occurrences per minute, of the behaviors observed during the acclimatization period in the Case group (N = 13). FENCE: Attention oriented to the fence; IPLAY: Individual play; JUMPSONO: Jumps on owner; JUMPSONS: Jumps on the stranger; SPINTO: Spontaneous interactions with the owner; SPINTS: Spontaneous interactions with a stranger.

**Figure 3.** Duration and relative frequency, expressed as the number of occurrences per minute, of the behaviors observed during the acclimatization period in the Control group (N = 13). FENCE: Attention oriented to the fence; IPLAY: Individual play; JUMPSONO: Jumps on owner; JUMPSONS: Jumps on the stranger; SPINTO: Spontaneous interactions with the owner; SPINTS: Spontaneous interactions with a stranger.

The groups differed significantly during the separation phase (Figures 4 and 5). Dogs in the Case group showed significantly less exploration (duration: 0 vs. 6, Case vs. Control; Mann–Whitney U = 122.500, *p* = 0.05; relative frequency: 0 vs. 0.33, Mann–Whitney U = 125.000, *p* = 0.039) and spent significantly more time standing by the fence (duration: 260 vs. 151, Case vs. Control; Mann–Whitney U = 36.000, *p* = 0.012, Figure S1) than Control dogs. In addition, at T1, we recorded significantly more frequent attempts by the veterinary behaviorist to interact with Case dogs (relative frequency: 2 vs. 0.33, Case vs. Control; Mann–Whitney U = 40.000, *p* = 0.022) than with controls. During dog-owner reunion, Case dogs were significantly more persistent than dogs from the Control group in jumping up on the stranger (duration: 10 vs. 0, Case vs. Control; Mann–Whitney U = 185.000, *p* = 0.004; Supplementary Video S3), with the veterinary behaviorist as the only target of this behavior.

**Figure 4.** Duration and relative frequency, expressed as the number of occurrences per minute, of the behaviors observed during the separation period in the Case group (N = 13). Fence: Attention oriented to the fence; IPLAY: Individual play; JUMPSONO: Jumps on owner; JUMPSONS: Jumps on the stranger; SPINTO: Spontaneous interactions with the owner; SPINTS: Spontaneous interactions with a stranger.

**Figure 5.** Duration and relative frequency, expressed as the number of occurrences per minute, of the behaviors observed during the separation period in the Control group (N = 13). FENCE: Attention oriented to the fence; IPLAY: Individual play; JUMPSONO: Jumps on owner; JUMPSONS: Jumps on the stranger; SPINTO: Spontaneous interactions with the owner; SPINTS: Spontaneous interactions with a stranger.

#### *3.2. Endocrine Responses*

As shown in Table 2 and Figure 6, the analysis of salivary OT and AVP revealed non-significant differences among time points within each group. However, at T1, AVP concentrations were significantly higher in the Case group than in the Control group. In addition, there was a trend, although not statistically significant, towards lower OT concentrations at T1 and T2 and higher AVP concentrations at T2 in Case dogs than in controls.

**Figure 6.** Concentrations of oxytocin (OT) and vasopressin (AVP) measured in saliva before (T0), immediately (T1), and 10 min after (T2) separation from the owner. CA: Case group, N = 13. CO: Control group, N = 13. Salivary AVP concentrations showed a statistically significant difference between groups at T1 (#, *p*<0.05).


**Table 2.** Salivary concentrations of oxytocin (OT) and vasopressin (AVP).

#### **4. Discussion**

This study aimed to determine whether there were differences in specific behaviors and endocrine responses between dogs affected by separation distress at home and behaviorally normal dogs while experiencing a potentially anxiogenic situation (separation from the owner in a novel environment and the presence of two strangers). We found that during the initial acclimatization phase, when the owner was still present, dogs with SRP behaved similarly to normal dogs. They were mainly explorative, as reflected by the higher frequency and longer duration of time spent in investigating the physical environment investigation time than in the other behaviors. Thus, dogs in both groups exhibited a similar attachment style, showing that they viewed the attachment figure (the owner) as a secure base for exploration of the novel environment [2]. This finding did not conform to the traditional vision, which considers dogs with separation distress as having excessive attachment towards their owners [40,41]. However, it agreed with what was more recently reported by Parthasarathy and Crowell–Davis [4], namely, that separation distress is not correlated with hyper attachment, confirming that these dogs were confident with the novel environment and strangers. Regarding the behavioral responses observed during the separation episodes, contrary to what was expected based on the results by Parthasarathy and Crowell–Davis [4], the patterns of behavior were different for the two groups in our study. Dogs in the Case group were significantly less explorative and more persistent in passive behaviors aimed at maintaining proximity to the owners, such as staying near the fence by the exit, than controls. According to Topál et al. [37], who analyzed the attachment relationship of behaviorally typical dogs with their owners, the fact that, during the separation phase, this behavior was not reduced by the presence of a stranger, despite her attempts to relate positively, might suggest that dogs with SRP have an especially strong preference for their owners in stress situations. This searching response has been observed in children [42] and non-human primates [43,44] and is regarded as aimed at maintaining the comforting bond of attachment. Nevertheless, dogs in the study by Topál et al. [45] showed a tendency to seek and maintain contact with the returning owner but not the stranger. Conversely, in our study, dogs with separation distress tended to show greater persistence in investigating the veterinary behaviorist (as suggested by more time spent jumping up) during post-separation reunion than did Control (CO) dogs, and this could also be seen as indirect evidence that these dogs experienced a higher level of anxiety during the separation. In fact, although we could not exclude the possibility that this was a side effect of dogs with separation distress needing to spend more time in proximity to the owner, who was sitting near the stranger, another possible explanation could be put forward. The veterinary behaviorist was the stranger who had made attempts to relate to dogs during the separation phase when proximity to their attachment figure was no longer possible. Those efforts might have made her a target of exploration for the SRP dogs in the reunion phase when contact with the attachment figure was re-established. This would also explain why the dogs' jumping up behavior was directed only towards the veterinary behaviorist and not the camerawoman. Indeed, even if our experimental design

was supposed to lead to calmer dogs compared to that used by Topal et al. [45] (e.g. a 10 vs. 4 minutes acclimatization period; dog vs. stranger initiated contact in the separation phase), dogs in the study by Topal et al. [45] were all behaviorally normal, and therefore less prone to anxiety than our Case dogs. More research is needed to understand whether behavioral reactions of a dog while alone (such as reduced exploration and increased seeking of proximity to the owner), or even after the owner's return, might be used as effective indicators of the presence of separation distress disorders.

Notably, salivary AVP concentrations were significantly higher in Case dogs than Control dogs immediately after the end of the separation period (T1). The timing with which OT and AVP reach saliva is not well understood, but they appear faster than other salivary hormones (e.g., cortisol), which reach peak concentrations in saliva ∼10 min after those in blood [31]. Previous studies have shown effects at a minimum time delay of 10 min [13,31]. However, other studies revealed early increases in salivary OT concentrations in nursing mothers and dams [31,46]. In the study by MacLean et al. [31], in particular, a large and statistically significant increase in salivary OT was detected in dams from baseline, at the end of a pre-test separation from their litters, to nursing 3 min later. This effect has been interpreted as reflecting an anticipatory rise in salivary OT. Similarly, it is possible that, in our study, the significant between-group difference observed in salivary AVP at 3 min partially reflected a rapid anticipatory response in SRP dogs. Given that non-social fears, such as fear of novel situations/environments, are common comorbidities of separation anxiety in dogs [17,47], we chose only dogs that did not present signs of stranger-directed aggression or fear of strangers or novel environments. Therefore, jointly with the fact that all dogs acted normal during the acclimatization period, the difference at 3 min was unlikely to be the result of stress at the start of the experiment in a novel environment. It is worth noting that, at T2, Case dogs still had higher concentrations of AVP than Control dogs, although the difference was not statistically significant due to the relatively high standard error of the mean. Although we could not be certain, it is possible that separation from the owner accounted also for this difference in salivary concentrations.

This was an intriguing outcome because central AVP, particularly that released within the amygdala, has been shown to be involved in the generation of passive coping strategies for acute stress in rodent models [37,48] through processes that would be mediated by the V1a and V1b receptors [48].

Both OT and AVP are synthesized in the hypothalamus, primarily in large magnocellular neurons situated in the supraoptic and paraventricular nuclei, and secreted from their axons, which are projected to the neurohypophysis, into the general circulation (for example, during labor or imbalance of water homeostasis) [48]. Vasopressin molecules that have been released in this way are, for the most part, prevented from re-entering the central nervous system via the blood-brain barrier [49]. In parallel, AVP and OT are also secreted within the brain, from the dendrites of the same neurons, in a manner regulated semi-independently of axonal release [50]. Following secretion, these peptides diffuse throughout the extracellular space, serving as neuromodulators for surrounding brain tissue [51]. As above, dendritic secretion of AVP has been shown to be of central importance in animal models of anxiety disorders [48], while axonal secretion has been shown to affect fear responses in mice [52], probably by regulating stress responses through the hypothalamic–pituitary–adrenal axis [53].

In our study, we could detect only an increased peripheral AVP level immediately after separation-induced social stress in SRP dogs compared to normal dogs, but central concentrations of AVP were likely increased as well. In fact, plasma and salivary AVP, measured using ELISA kits, were found to be moderately correlated in humans in one study [54], and plasma AVP concentrations significantly and positively predicted cerebrospinal fluid AVP concentrations in human neonates in another study [55]. Thus, these two studies provided preliminary support for the use of salivary AVP ELISA measurement as a proxy for brain AVP activity, at least, in humans. Future studies are now required to determine the relationship between behavioral measures and AVP concentrations in both the central and peripheral compartments in dogs.

It is worth mentioning that interest in using V1b antagonism to treat anxiety disorders has been investigated. Clinical trials in humans failed, with the V1b receptor antagonist SSR149415 not being useful for the treatment of generalized anxiety disorder [56]. However, two newly synthesized V1b receptor antagonists—TASP0233278 and TASP0390325—have shown potential benefits in rodent models [57]. Future studies could examine whether this therapeutic approach might also benefit dogs diagnosed with SRP. The antidepressants—clomipramine and fluoxetine—which act primarily as serotonin reuptake inhibitors, are currently approved for the treatment of canine separation distress [58]. As more evidence is collected on the role of AVP in the pathophysiology of SRP, future drugs targeting the vasopressinergic system would offer treatment options for canine separation distress therapy.

Finally, although AVP is often anxiogenic, the closely related nonapeptide OT often has anxiolytic effects [59] and may reduce the stress of negative social interactions [60]. The concentrations of OT that we reported here with a sample size of 26 dogs (13 behaviorally normal, 13 with SRP) did not reach statistical significance over time, although a trend towards a reduction (and an increase in AVP concentrations) was observed during and even after the separation from the owner in dogs from the Case group. This could also be a type II error-related false-negative result due to the small sample size. Future research, exploring in a larger sample size whether both OT and AVP respond dynamically to this potentially anxiogenic situation, might help detect a significant effect.

#### **5. Conclusions**

In conclusion, this study showed that, when placed in a novel environment, dogs presented different behavioral and endocrine responses to a short separation from the owner, followed by the reunion, depending on whether they suffered from separation distress at home. Dogs with SRP became more distressed than CO dogs when the owner was gone for a short time and left them in the company of two unfamiliar persons. They were less able to mediate their reactions in such a stressful situation, showing more passive coping strategies aimed at seeking proximity to the owner while he/she was absent, and had significantly higher salivary concentrations of AVP at the very end of this phase, which are two responses that have been previously associated in other animal models of social separation. Although these results are preliminary and should, therefore, be interpreted with caution, the differences observed between groups in both behavioral and endocrine responses during and after the separation lead the way to further exploration of the use of salivary AVP as an early, non-invasive biomarker of canine anxiety-related disorders and support AVP antagonism as a potential new mechanism-based therapeutic approach. According to Thielke and Udell [61], improved methods of treating SRP in dogs would not only benefit human-dog relationships but also potentially contribute to the decreased surrender of dogs to shelters by reducing the effort needed to modify this behavior problem successfully.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-2615/9/12/1033/s1, **Figure S1:** A photo of a dog from the Case group who is standing by the fence during the 3-minute separation phase (T1). Photo by Valentina Sammartano. **Table S1:** Main characteristics of the dogs involved in the study. **Video S1:** A dog is showing signs of stress while the owner is out of the arena and the veterinary behaviorist is engaging the dog in friendly interaction, including gently petting her and speaking to her in a calm tone. **Video S2:** A dog is exhibiting signs of severe distress and anxiety; consequently, the owner is asked to come back, and the test is stopped. **Video S3.** A dog in the Case group persists in jumping up on the veterinary behaviorist during the post-separation reunion phase.

**Author Contributions:** First Author (F.P.) was responsible for conceptualization, data curation, funding acquisition, investigation, and writing (original draft + review and editing). The second author (L.P.) was responsible for conceptualization and data collection. The third author (A.B.) was responsible for data analysis. Forth author (S.U.) was responsible for conceptualization, reviewing, and editing. The last author (M.A.) was responsible for conceptualization, methodology, interpretation of data, and supervision.

**Funding:** This research was funded by the UNIVERSITY OF MILAN (LINEA2\_FPIRR\_AA\_2017 grant number 24066).

**Acknowledgments:** We thank Carla Colombani for her support in conducting laboratory analyses in this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Disaster Preparedness among Service Dog Puppy-Raisers (Human Subject Sample)**

### **Sarah E. DeYoung 1,\*, Ashley K. Farmer 2, Zoe Callaro <sup>3</sup> and Shelby Naar <sup>4</sup>**


Received: 20 December 2019; Accepted: 1 February 2020; Published: 4 February 2020

**Simple Summary:** Before service animals are matched with clients, they undergo training programs for increasing the dogs' ability to navigate public spaces. Increasingly, service dog training programs recruit young adults from universities and college campuses. Little is known, however, how these students prepare for disasters and the ways in which they create plans to keep service dogs in training safe during hazard events. We collected data from service dog puppy raisers in a hurricane-prone region of the United States to understand their concepts and actions of disaster preparedness. People who were raising a service puppy for the first time were more likely to consider evacuating from Hurricane Irma in 2017 than people who had participated in the program before. Additionally, over half of the respondents did not have a disaster preparedness kit. Finally, many respondents in this study indicated that their service dog in training provides a sense of safety and security.

**Abstract:** Little is known about the ways in which puppy raisers engage in disaster preparedness for their puppies (or "guide dogs in training"). The aim of this research is to understand disaster preparedness among service dog puppy raisers. A web-based survey was distributed to people raising puppies in a service dog training program (*n* = 53 complete survey responses). Questions in the survey included items about disaster preparedness and plans for canine safety in hazards events. Out of those who said they had an evacuation plan for their puppy in training, 59% stated they would put the dog in their vehicles for evacuating to safety in the event of a hurricane or other disaster. The odds of first-time puppy raisers who considered evacuation for Hurricane Irma in 2017 was 15.3 times the odds of repeat raisers. Over half the raisers reported that they did not have a disaster kit. Additionally, 82% of respondents indicated that having a service puppy in training makes them feel safer. These results can be used as a foundation for service dog organizations in disaster preparedness among their puppy raiser volunteers and in designing recruitment messages for new volunteers.

**Keywords:** disaster; preparedness; puppy raiser; service dogs; working dogs

#### **1. Introduction**

The Pets Evacuation and Transportation Standards (PETS) Act was passed on 6 October 2006 after category five Hurricane Katrina devastated the coast of Louisiana and Texas. This new act amended the Robert T. Stafford Disaster Relief and Emergency Assistance Act to address the needs of household companion animal and service animals [1]. However, preparedness information typically disseminated to communities through outreach programs or county emergency managers encourages companion animal guardians to prepare not only for themselves but also for their animals [2]. Individual preparedness—such as being able to care for household animals in the event of an emergency—helps reduce losses in disasters and puts less strain on first responders and resources [3]. Although the PETS Act was an important policy for creating plans for managing pets in disasters, individual preparedness remains extremely important for all companion animal guardians, service dog users, and those with service dogs in training [4].

Globally, pets and companion animals are a critical part of the human landscape. According to 2016 companion animal guardianship statistics, out of Australia's 9.2 million households, 5.7 million housed a companion animal [5]. Companion animal guardianship has grown tremendously among the United States population, serving many different forms of companionship. According to the 2017–2018 American Companion Pet Products Association (APPA) National Companion Pet Guardians Survey, 68% of households within the U.S. own a pet. For context, 68% is equivalent to about 84.6 million homes [6]. The National Pet Survey was first conducted in the year 1988, where it found that about 56% of households owned a companion animal at that time. Besides the typical dog and cat, the survey now includes numbers for birds, horses, freshwater fish, saltwater fish, and even reptiles. In 1994, APPA calculated total U.S. companion animal industry expenditures to be \$17 billion [6]. In 2016, that number soared to almost \$67 billion, showing the bond between guardian and companion animal has strengthened over the past few decades [6].

While expenditures alone may not capture the overall social landscape of humans and their pets, news media coverage of humans and their pets in disasters is present in almost every disaster [7]. The ongoing concern about pets in disasters reflects the notion that pets are important in emergencies. Less attention has been given to the topic of service animals in disasters and the ways in which people who rear service animals engage in disaster preparedness. In the United States, college students are increasingly volunteering to train guide dog puppies—and many of the campuses in which they live are prone to hurricanes and other hazards. This research is an attempt at understanding the ways in which guide dog trainers prepare their puppies for disasters and emergency scenarios. Specifically, this research focuses on the emerging population of college student puppy raisers and the ways in which they perceive preparedness, safety, and risk for themselves and their puppies in training. We begin the paper by describing the context of the human-animal bond because this may influence decision-making such as disaster preparedness. We then describe research related to college students and the ways in which they interpret risk, followed by literature on the social landscape of animals in disasters. We then provide the method for our study followed by the results, discussion, and implications for future research.

#### *1.1. Human-Animal Bond*

Companion animals provide a multitude of benefits for their guardians. Although emotional benefits may be most obvious, companion animals can also have social, physiological, and physical health benefits [8–10]. Companion animal guardianship directly affects cardiovascular health, increases physical activity, enhances the welfare of the elderly, and even serves as a source of support for people experiencing crisis and hardship. These hardships can include homelessness, family stressors, or some type of disaster [9].

Therapeutic organizations have used animals in several areas, including animal-assisted therapy, animal-assisted crisis response, and child psychotherapy [8,11]. The role of pets is crucial, as reflected in some reports, where companion animals were the main reason individuals survived personal crises such as loss of a loved one, a family health crisis, or periods following substance abuse [9]. Benefits created by companion animals are not unique to one specific age, gender, or ethnicity; evidence suggests companion animals to help hospitalized children recover, ease suffering for those in hospice care, and alleviate depression among Acquired Immune Deficiency Syndrome (AIDS) patients [11].

Dogs, although popular companion animals, are also popular in occupational roles. The roles include guiding for the blind, search and rescue, assistance for people with disabilities, and police detection work [12,13]. Despite working dogs serving a different role than companion animals, they may have similar attachment levels to their handlers as pets do to their guardians [14]. Assistance dogs serve important roles for their handlers, especially for handlers who cannot see or are visually impaired [12].

#### *1.2. History and Context of the Guide Dog Occupation*

The origins of the guide dog can be traced back to Germany after the end of World War I and have now become more and more common throughout the U.S. and countries around the world [15]. Guide dogs play an important role for their handlers, serving as companions, security detail, and safety officers. After World War I, many German troops returned home with no physical injuries besides the loss of their eyesight [14]. In the U.K., many guide dog users expressed social and psychological benefits from having a guide dog, as well as increased confidence and socialization with other people [16].

The way other people treat those who are visually impaired has also been stated as a benefit among guide dog users; out in public people tend to be friendlier towards those individuals when the dog is present [16]. This particular benefit has been very important because individuals with visual impairment are already prone to social stigma and even discrimination [17]. In South Africa, guide dog users explained the dogs were safer and faster than using a cane, they provided more confidence regarding mobility, and enhanced independence [17]. Japanese guide dog teams expressed that their dogs reduced tension and strain when traveling and had positive effects on mobility and quality of life [18]. Many of these benefits originate from the strong bond and attachment that forms between handler and dog. Additionally, research on guide dog training programs suggests that existence of service animals can be linked with the perceived validity of a persons' functional and access needs status [19].

#### *1.3. Guide Dog-Handler Bond*

Guide dogs go through many bonding stages during their training and placement process with a handler. They live and stay with a puppy raiser, which is typically a volunteer through the guide dog school, get assigned to a trainer once they are around sixteen months old, and then are eventually matched with a person who is visually impaired. Each of these stages has been isolated and studied to understand the difference in bond strength between the dogs and their raisers, the dogs and their trainers, and the dogs and their handlers [20,21]. For example, after one year of partnership, guide dogs showed all signs of attachment when separated from their handler, even in the presence of another friendly individual, and remained oriented to the door for their handler to return [20,21].

Some evidence suggests that because handlers who are blind have different situational awareness, some dogs developed signals to more effectively communicate with their handler, such as maintaining physical contact [21]. Because of the process used to train guide dogs, bonds are formed and technically lost after the dogs leave their puppy raisers and then again when they leave their trainers. However, the use of puppy raisers is still important in ensuring the dogs reach their end goal of helping individuals dependent upon their services [22].

#### *1.4. Puppy Raisers*

Guide dog schools utilize volunteer puppy raisers as an essential part of their process [22]. Puppy raisers typically have the dogs from around eight weeks old until a little over a year old or until the dog has matured. Because the puppy raisers have the dogs for quite a while, it is understandable that bonds form between the trainer and puppy. Puppy raisers can help contribute to a dog's success or more accurately predict the ability of the dog they are raising to become a guide dog. For example, assessments or questionnaires done by puppy raisers concerning their dogs' chances of success have served as important predictors for the respective guide dog schools [22–24]. This suggests that puppy raisers bond with and know their dog very well over just a year of time. Other benefits can also come

with being a puppy raiser, including improved mental, social, and physical health; increased awareness of responsibility; and new friendships with other raisers [25].

Depending on the reach and location of each guide dog school, a diverse array of people volunteer as puppy raisers. In many instances, college students tend to be the most common group of volunteers. To become successful guide dogs, raisers take the puppies with them everywhere they go so that the dogs grow accustomed early on to what they will be doing as working canines. For working adults with full-time jobs, it is harder for them to be accompanied by young puppies every day, and the puppies are also big commitments. College students can more easily take a puppy in training to class, grocery shopping, or to social events [26]. Interestingly, the popularity of having a guide dog puppy is expressed as a trend among college students in some areas of the United States, like fashion and technology branding [27] that signifies identity, social status, and possible wealth. In fact, it is common to see students walking on campuses with the training dogs that wear the signature yellow vests. For the broader puppy-raiser population, training of the dogs is likely to take place by able-bodied persons [28]. In the context of this research, it is important to consider how the able-bodiedness of the handlers influences their ability to engage in activities such as disaster preparedness, as well as perceptions about ease of preparedness by able-bodied handlers [29].

#### *1.5. College Students: Risk Perception*

Risk perception is defined as a "highly personal process of decision making, based on an individual's frame of reference developed over a lifetime" [30] (p. 1) and can be directly linked to protective actions taken [30,31] In other words, perception of risk is a very individualistic concept and will differ from person to person [32]. This important aspect of risk perception becomes a challenge in many risk communication strategies, particularly when risk perception is low.

Source and repetition of information also contribute to decreased risk perception. If the source is not trusted or people do not trust current science or technology in general, there will be lower general risk perception [33]. If information regarding a specific risk has become the social norm and is common knowledge, risk communicators will have a hard time heightening risk perception when necessary [34,35]. A major example of this would be safe driving practices for younger drivers (i.e., college students), which is a very popular risk communication topic. Adults between the age 18–29 are more likely to drive while drowsy compared to other age groups, and drivers in their 20's make up 27% of distracted drivers in fatal crashes; however, college students have grown accustomed to car-related risk information and accept it as common knowledge [36,37]. Other factors contributing to low risk perception include optimistic bias, self-efficacy, perception of benefits, and perceived barriers to preventative action [35,38,39].

In general, higher perceived severity of a risk (understanding the risk has serious consequences) will lead to higher risk perception and cause an individual to engage in more physical protective behavior [40,41]. Higher perceived severity is common in those who have personally experienced certain risks in their area (i.e., tornadoes, earthquakes, wildfires), so it is understandable that heightened risk perception differs from region to region [42,43]. Among college students, local hazards, as well as a risk being uncontrollable or unknown, will contribute to higher risk perceptions [43].

#### *1.6. College Students: Risk Information*

The sources of risk information have changed as technology has advanced over the years [44]. For example, in 2010 students were more likely to receive information from family, friends, and television and less likely to receive it from university flyers, professors, or course content [45]. However, in 2012 students preferred to receive emergency preparedness information directly from their school via texts and emails [44]. These results were proven a few years prior when a tornado touched down near the Mississippi State University campus, and most of the students stated they first heard of the tornado through university alert messages [46]. This may be because students indirectly expect their university to "take care of them" in the event of an emergency or disaster [42].

#### *1.7. Service Animals & Disasters*

Because there were multiple instances of people being forcefully separated from their animals during Hurricane Katrina [47] there has been growing attention on the issue of animals and disasters. While the PETS Act was passed in 2006, the Act has limitations in terms of breed, disaster phases (e.g., recovery and long-term housing is not considered in the PETS Act, such as rental fees for pets of displaced persons), and refusal of evacuation and reunification issues remain a concern in disasters [48]. Simultaneously, the body of research on pets and disasters is growing– especially in the areas of understanding attachment [49,50] and issues related to grief [51,52]. There is also a growing body of research on the needs of functional-access individuals in disaster planning [53,54]. However, there is a scarcity of research on guide dog trainers and disaster planning and preparedness. The primary research aim for this paper is to identify how guide dog puppy-raisers engage in and perceive preparedness for their guide dog puppies. We focus on college students as a population of puppy raisers because college students are at a unique developmental phase in which perceptions of risk and safety, as well as independence, begin to take on a new role in their lives [55]. Additionally, it is important to understand the ways in which college students prepare for hazards [56]. If being a guide dog puppy volunteer is associated with lower or higher levels of preparedness, this has important implications for higher education institutions in terms of planning and outreach. No studies exist that examine the ways in which college students as puppy-raisers prepare for disasters. Studies so far on dogs on college campuses focus on dogs as stress relief for students [57,58] rather than how having a guide dog puppy in training changes the students' level of preparedness.

#### **2. Method & Materials**

#### *2.1. Participants*

The researchers gathered data using a Qualtrics survey platform to reach volunteer puppy raisers within a southeastern college town. Qualtrics is an electronic survey platform that collects responses and can be automatically uploaded through the internet or in which responses can be uploaded onto a hard drive. The participants were contacted through an email list serve composed of the guide dog puppy raisers. There were approximately 300 people on the list-serve. The puppy raisers targeted for this survey are volunteers through an organization based in the northeastern United States that recruits' college students throughout the US for puppy-training. Specifically, as part of their puppy-raising program, this school has volunteer groups throughout the east coast, with a lot of groups in southeastern states. One of the researchers for this project knew the leader of this foundation from her personal social network, which enabled the research team to distribute the survey through the email list serve. In other words, the recruitment was through convenience sampling [59] but with the support of the leader of this organization that trains guide dogs. While richness of data is beneficial from open-ended interviews [60], we chose a survey format because it facilitates rapid data collection and is convenient for respondents to take a web-based survey rather than coming to a specific location at a specific time for data collection.

Because of the nature of being a puppy raiser (i.e., taking care of a young puppy, taking it with you everywhere you go), groups usually center around universities, as many of their volunteers are college students. Although the guide dog school is headquartered in the northeast, they have a southeastern coordinator who maintains the southern groups. A reminder email was sent two weeks after the initial survey recruitment and notification.

#### *2.2. Procedure*

The survey (Appendix A) was conducted under a university Institutional Review Board in December 2017 and January 2018. The survey link contained a consent form, an introduction to the study and allowing participants to consent and continue on with the questionnaire. The responses were automatically and anonymously uploaded into the Qualtrics system after each submission and transferred to SAS statistical software for data cleaning and analysis.

#### *2.3. Data Analysis*

The results from the Qualtrics survey had 84 responses, not including pilot responses. Analyses were conducted on the sample size of *N* = 84 (response rate = 28%), but there were missing data for 31 respondents. Of those 31, 84% did not answer past the first two questions and were not included in the final analyses, and the remaining 16% left partial responses and were included. There were a total of 53 respondents who completed the entire survey. SAS version 9.4 was used to code the survey responses, assess frequencies, and conduct chi-square analyses.

#### **3. Results**

Out of the 53 participants who completed the entire survey, the majority were female (*n* = 48, 91%) with a mean age of 28 years (*SD* = 14). The respondents were predominantly White (*n* = 47, 88%). The remaining respondents were Hispanic (*n* = 3, 1%), Asian (*n* = 2, <1%), and African American (*n* = 1, <1%). 40 out of the 58 participants identified as college students. Seniors made up most of the college students (*n* = 19, 48%), followed by graduate and dual-degree students (*n* = 9, 22%), juniors (*n* = 8, 20%), and sophomores (*n* = 4, 10%). It is likely that the high number of responses from college students is because most of the people on the list-serve are students and because they may be more likely to check email frequently because of schoolwork and communication. There were no freshman participants; however, it is less likely for first year students to be puppy raiser volunteers as it takes time to learn about the program, apply, and go through the volunteer process. Among the puppy raisers, 65% of them were repeat raisers, meaning they had already raised one or more dogs, while 35% were first time raisers.

The first items in the survey measured evacuation plans, locations, and logistics (Refer to Appendix A). Among the puppy raisers who said they had an evacuation plan for their puppy in training (*n* = 17), many stated they would just put the dog in the car with them and leave. Some raisers specifically added they would bring food, water, toys, a dog bed, or a crate (*n* = 8). Only two raisers said anything about bringing a preparedness or disaster kit for their dog, and another two mentioned bringing papers and vaccine records. One survey item inquired about where participants would go if needing to evacuate. Many respondents (*n* = 33) stated they would go stay with family, including parents, siblings, or even extended family. Family was also a popular response among college students when asked where they would go if they had to evacuate their parents' house.

Four survey questions specifically asked participants about plans relating to Tropical Storm Irma of 2017 since the storm had hit southeastern states a few months prior to the survey. An analysis of the crude relationship between puppy raiser experience (first time raiser versus repeat raiser) and consideration to evacuate for Irma gave an odds ratio of 15.3. This reflects that the odds of first-time puppy raisers considering evacuation for Irma was 15.3 times the odds of repeat puppy raisers, revealing that first-time raisers were much more likely to consider evacuation, *X*<sup>2</sup> (1, *N* = 39) = 11.37, *p* < 0.05. Also, of the 33 participants (77%) that said they prepared supplies specifically for their animal before Irma, 32 (74%) mentioned essentials such as food and water, 4 (9%) said they gathered medical supplies or a first aid kit for their animal, 5 (12%) mentioned outside gear for the dogs when going to the bathroom or a way for the dogs to go to the bathroom inside, and a final 2 (5%) mentioned getting together papers or vaccine records. It should be noted that some respondents were included in more than one category as they mentioned more than one of the above answers.

In response to the item inquiring about having a disaster kit at home, over half (58%) of the participants said no. Most participants (82%) said that their companion animal or puppy in training makes them feel safe daily as opposed to the 2% who said their companion animal or puppy in training makes them feel unsafe. Table 1 shows frequencies of the risk perception responses, reflecting what participants view as the highest risk to their companion animal or puppy in training daily versus what they view as the highest risk to themselves.


**Table 1.** Survey responses to "highest risk to self" and "highest risk to pet".

#### **4. Discussion**

The primary aim of this study was to explore levels of preparedness and risk perception in college students with service dogs (puppies) in training. We expected volunteer raisers who had raised more than one dog to be more prepared for sudden emergencies or disasters (i.e., have a disaster kit, have a plan for their dog). However, over half of the puppy raisers surveyed did not have a disaster kit of any kind and although many prepared food and water for their dog prior to Irma, very few mentioned vaccine records or anything medical-related for the dog. Additionally, first-time puppy-raisers were more prepared than people who had raised multiple puppies. This may be because of the influence of having a "new animal" and potential feelings of protectiveness, like findings in which families with children are more likely to have higher levels of household preparedness [61].

Vaccine records are extremely important to have on hand in case of an emergency as they would be needed when taking an animal to a boarding facility, vet office, or even a shelter [62]. Being turned away at any of those facilities due to lack of records can also play a role in people not wanting to evacuate, especially when they are unable to locate a pet-friendly shelter to begin with [48].

In response to the item inquiring about having an evacuation plan for their puppy in training, over half stated not having a plan at all. Among those that did have a plan, the majority just mentioned evacuating with the dog in the car. Two participants mentioned bringing a crate, and although research suggests transportation challenges as the main cause of evacuation failure for companion animal guardians, it should be noted that these dogs are trained to ride in the footwell of cars (not in crates) and that every puppy raiser should be issued a crate when they receive their puppy in training. Again, few respondents mentioned vaccine records or a disaster kit, with only one participant stating they always kept a kit in the car for their dog. These results support existing literature on preparedness among college students in general, with the most common preparedness activities found to be storing water or food, having a working flashlight, and taking a Cardiopulmonary Resuscitation (CPR) class [63]. In a similar way that students take few protective actions because they assume the university will take care of them in the event of a disaster, puppy raisers may expect the same resources and guidance from the organization for which they volunteer [42].

Many participants said in the event of an evacuation, they would evacuate to a family or relative's place of residence. No specific logistical issues were mentioned; however, that may be due to the socioeconomic status of the survey sample. Most respondents have access to cars—the dogs have been trained and accustomed to riding in a car—and therefore most participants would not need the use of local shelters. Previous hurricane evacuation data reflects that most evacuees stay with relatives or friends [64]. Many college students mentioned evacuating to where their parents live and is important to note since students may in some cases rely on parents for risk-related information more so than media or other interpersonal sources [44].

There was a difference in likelihood of evacuating or considering evacuating prior to Irma between first time raisers and repeat raisers, with first time raisers more likely to do so. This could be due to repeat raisers being more confident about their responsibility towards their puppy in training as they have been through the raising process before. Of the college student participants, many reported being enrolled in majors relating to animals or other involved sciences. The confidence among puppy raisers concerning the safety of their dog could also stem from their area of study being closely related to the risk, a reason many college students tend to have low risk perception [32,33]. It should also be noted, again, that the samples for this study were not diverse in terms of sex and ethnicity, which creates limitations for the implications regarding additional analyses to compare risk perception and preparedness among respondents. Additional research in the area of guide dog puppy raisers should include a sample that has more ethnic and gender diversity, especially since minority groups more susceptible to adverse outcomes in disasters [65].

Many of the participants stated that their companion animal or puppy in training made them feel safe. These results are consistent with other findings that suggest pets help enable a sense of security and serve as a haven and secure base for their guardians [66,67]. This sense of safety from the human-animal bond also suggests that companion animal guardianship can serve as a protective factor during evacuations where guardians are more likely to get to safety for the sake of their companion animal [68]. College students also directly benefit from pets, where just a simple twenty-minute canine therapy session reduced stress and even homesickness [69].

However, despite the sense of security the puppies in training and pets bring to their guardians, there are still daily risk perceptions that differ from person to person. Table 1 reflects that most participants viewed the highest risk to themselves and their companion animal to be an auto accident. Despite the number of young drivers involved in car accidents every year, some participants viewed the highest risk to their companion animal to be an event of lower probability, such as severe weather, structural fire, or an active shooter [36]. Because individuals have less control over these types of events, it is possible that fear caused a heightened risk perception despite an auto accident being a more significant risk [70].

#### **5. Conclusions**

This study took place in the United States and therefore the results are not generalizable beyond the specific population of college students with guide dog puppies. Also, as we mention in the broader discussion, the homogenous sample (white, able-bodied, and female) is a limitation. Future research should include diverse groups of volunteers. Additionally, the small sample size of survey responses is a limitation of this study. Although there were differences between first time and repeat puppy raisers, the small sample did not yield more than one statistically significant finding. Future research should also focus on the potential difference in preparedness practices and activities between companion animal guardians and puppy raisers since puppy raisers take their dog most places with them as opposed to a companion animal guardian. Future research should also include in-depth qualitative data in which students as puppy-raisers share their thoughts on safety, risk, and preparedness. This would add richness to the data regarding motivations, perceptions, and processes for preparedness. This might also uncover new information about levels of bonding with their guide dog puppies and how levels of bonding relate to preparedness.

Finally, it is also important for future research to explore the issue of branding or the "trendiness" of having a guide dog puppy among college students, especially in the context of understanding how this experience is different for the person training the dog compared to the person who is ultimately paired with and requires the dog because of a functional-access need. The college students in this sample were white and able-bodied [71] whereas this may not be the case for functional access individuals in which having a service dog may be another form of labeling the person as "disabled" [72,73] and in which preparedness may entail different meanings and processes. In other words, it may be important to understand how the accompaniment of a service-dog elicit stigma from others and the ways in which this impacts functional and access needs individuals in hazards events.

The results from this study can be used as a foundation in future research for service dog organizations into hazard and disaster preparedness among their puppy raiser volunteers. Because service dog volunteer groups are sometimes spread several states away from the headquarters of the guide dog organization, accountability and tracking the dogs could become difficult in the event of an emergency or disaster. For some volunteer groups, there are hierarchical leadership roles in place to help keep track of the raisers and all the dogs. These leadership positions could incorporate potential preparedness protocols into their volunteer group meetings or even just going over what to include in a pet-related disaster kit. Since most groups are required to attend monthly meetings, those are already designated times when many puppy raisers and volunteers come together where preparedness practices could easily be discussed among the group.

Additionally, since having a service dog in training made students feel safer, this information could be used in developing recruitment messages for new volunteers. Having service puppies in training makes students feel safer and possibly less stressed. A key message for college student volunteers could be something along the lines of, "Your service puppy in training helps you to feel safe. Learn how to keep your puppy in training safe during disasters". In general, puppy raisers are a distinct and dynamic group of companion animal guardians that are most likely not accounted for in many companion animal management plans, so it is important to practice individual and community-level preparedness for the safety of both people and animals. Finally, it is also important to consider that preparedness strategies and meanings will likely be different for trainers than they are for people who are later paired with the dogs and this may impact implications for experiences of both the dogs and humans. This discrepancy should be accounted for when organizations are creating training protocols for raising guide dog puppies (i.e., considering a more inclusive and aggressive outreach of puppy-raisers who have functional-access needs, people of color, and more diverse groups of people). This is also reflected as a guideline for best practices in preparedness and disaster planning [74].

**Author Contributions:** Conceptualization, S.E.D. and S.N.; Formal analysis, S.N.; Investigation, S.N.; Methodology, S.N.; Supervision, S.E.D.; Writing—original draft, S.N.; Writing—review & editing, S.E.D., A.K.F. and Z.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We thank Curt Harris at the University of Georgia for his comments on the early draft of this paper.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A Survey Instrument**

#### *Preparedness among Puppy Raisers*

	- a. Yes
	- b. No
	- a. Yes \_\_\_\_\_\_\_\_\_\_\_\_\_
	- b. No (Skip to Question 6)
	- a. Car/auto accident
	- b. Disease or illness
	- c. Severe weather
	- d. Structural fire
	- e. Other
	- a. Car/auto accident
	- b. Flu, illness, or communicable disease
	- c. Severe weather
	- d. Active shooter
	- e. Structural fire
	- f. Sport or other physical injury
	- g. Other
	- a. Internet news
	- b. Television news stations
	- c. Mobile news apps
	- d. Social media
	- e. Word of mouth on campus


	- a. Very unsafe
	- b. Moderately unsafe
	- c. Slightly unsafe
	- d. Neutral
	- e. Slightly safe
	- f. Moderately safe
	- g. Very safe
	- h. Do not know
	- i. N/A (No pets)
	- a. Internet news
	- b. Television news stations
	- c. Mobile news apps
	- d. Social media
	- a. Yes \_\_\_\_\_\_\_\_\_\_\_\_\_

b. No

	- a. Yes

b. No

14. If yes, what was your evacuation destination and why did you choose that destination?

	- a. Black or African American

\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_

	- a. Female
	- b. Male
	- c. Transgender
	- d. Other
	- e. Prefer not to answer
	- a. Freshman
	- b. Sophomore
	- c. Junior
	- d. Senior
	- e. Grad/professional student
	- f. Dual degree
	- g. N/A (Skip to question 21)

#### **References**

1. American Veterinary Medical Association. PETS Act (FAQ). Available online: https://www.avma.org/KB/ Resources/Reference/disaster/Pages/PETS-Act-FAQ.aspx (accessed on 3 May 2018).


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Back to the Future: A Glance Over Wolf Social Behavior to Understand Dog–Human Relationship**

#### **Giada Cordoni <sup>1</sup> and Elisabetta Palagi 1,2,\***


Received: 25 October 2019; Accepted: 11 November 2019; Published: 18 November 2019

**Simple Summary:** Wolves, the ancestors of dogs, are one of the most cooperative canine species. This cooperative propensity derives from the fact that each subject needs other group members to obtain resources and increase survival. The pack functions as a unit in which each individual collaborates in territory defence, hunting, and rearing of offspring. For this reason, even though a clear hierarchy exists among wolves, subordinates can provide help to dominants to obtain social tolerance in a sort of commodity exchange. Wolves can make peace after aggression, console victims of a conflict, and calm down the aggressors. This set of behaviors, also called post-conflict strategies, requires a social attentiveness towards others' emotional state and the ability to coordinate appropriate reactions. Adult wolves also play. They engage in play fighting, which strongly resembles real fighting, by finely modulating their motor actions and quickly interpreting playmates' *intentions*, thus maintaining the non-serious playful mood. All these cognitive and social skills were a fertile ground for the artificial selection operated by humans to redirect the cooperative propensity of wolves towards dog–human affective relationship.

**Abstract:** This review focuses on wolf sociobiology to delineate the traits of *cooperative baggage* driven by natural selection (wolf-wolf cooperation) and better understand the changes obtained by artificial selection (dog-human cooperation). We selected some behaviors of the dog's ancestors that provide the basis for the expression of a cooperative society, such as dominance relationships, leverage power, post-aggressive strategies, and playful dynamics between pack members. When possible, we tried to compare the data on wolves with those coming from the dog literature. Wolves can negotiate commodities when the interacting subjects occupy different ranking positions by bargaining social tolerance with helping and support. They are able to manage group disruption by engaging in sophisticated post-conflict maneuvers, thus restoring the relationship between the opponents and reducing the spreading of aggression in the group. Wolves engage in social play also as adults to manipulate social relationships. They are able to flexibly adjust their playful interactions to minimize the risk of escalation. Complex cognitive abilities and communicative skills are probably the main proximate causes for the evolution of inter-specific cooperation in wolves.

**Keywords:** social tolerance; social attentiveness; reconciliation; consolation; appeasement; play fighting; leverage; behavioural plasticity; *Canis lupus lupus*

#### **1. Introduction**

*"Domestication is a sustained multigenerational, mutualistic relationship in which one organism assumes a significant degree of influence over the reproduction and care of another organism in order to secure a more predictable supply of a resource of interest, and through which the partner organism gains advantage over individuals that remain outside this relationship, thereby benefitting and often increasing the fitness of both the domesticator and the target domesticate"*. [1]

Even though we intuitively understand what domestication is, there is a surprising lack of consensus on its definition. Beyond the agreement that domestication involves a relationship between a *domesticator* and a *domesticated organism*, there are many debates on what this relationship entails and how and when it occurs. Many definitions of this process take into account only the perspective of the *domesticator*, emphasizing the impact of humans in this role. Historically, humans deliberately and opportunistically select a wild species for creating its "domesticated counterpart" of whom they control all aspects of its life cycle [1]. According to "*domestication syndrome*" [2], a set of morphological, physiological, reproductive, and behavioral traits can be observed in domesticated species but not in their wild ancestors. In animals, these traits can include, for example, increased fecundity, altered coat color, reduced body size, facial neoteny, increased docility, and hypersociability [3–5].

Dog, the oldest domesticated animal by humans, certainly shows many traits of the "*domestication syndrome*" such as, reduced body size and snoutlength andincreased docility, tameness, and playfulness [5,6]. Many researchers suggest that the beginning of dog domestication took place in the Early Upper Paleolithic (~30,000 years ago), when people still lived in small groups as hunter–gatherers and agriculture was not yet practiced [7–10]. Nevertheless, the fossil remains confidently indicate the appearance of dog in Europe ~15,000 years ago [11]. Basing on morphological and genetic analyses, wolves (*Canis lupus lupus*) are undoubtedly the ancestors of modern dogs [9,11–13]; while the wild wolf phenotype changed markedly, the genotype changed only minimally, leaving domestic dogs, genetically speaking, still as wolves [10]. The domestic relationship between people and dogs is the result of a wolf ecological strategy to cope successfully with the Late Pleistocene environmental changes due to the increased human population. The plasticity characterizing wolves permitted them modifying their ecological niche by joining the human niche; people possibly facilitated this change by incorporating some young wolves into their groups and by selecting over time the more docile and tameness subjects [10,13,14].

According to the *Domestication Hypothesis*, it has been proposed that through an evolutionary and ontogenetic positive feedback processes, dogs have become more socially tolerant and attentive than wolves, two characteristics that are crucial for cooperation to occur [15,16]. However, the studies supporting the *Domestication Hypothesis* were mainly based on wolf–dog behavioral difference in relation to their interactions with humans [15–21]. Range & Virányi [20,21] proposed an alternative (but not exclusive) theory, the *Canine Cooperation Hypothesis*, according to which these differences may reflect only an improved capacity of dogs to accept humans as social partners instead of an increased in their general tolerance, attention, and cooperation degree. Some cognitive studies have shown that wolves can attentively use the information provided by a familiar human to solve a task [22], can follow human gaze as readily as conspecific gaze [23] and are more successful than dogs in copying the actions of conspecifics [24,25]. Moreover, when wolves and dogs (reared under the same conditions) were faced with a series of object-choice tasks, wolves showed similar results to dogs in responding to communicative and behavioral cues, but they outperformed dogs in their ability to follow causal cues [26]. All these authors suggested that high level of cooperativeness characterizing wolf society may come together with a high propensity to pay close attention to others' actions. Also, by moving the focus from human–wolf to wolf–wolf interactions, researchers have highlighted the high cooperativeness and cohesiveness characterizing wolf packs [20,26–28].

This review aims at delineating the possible pathways of the behavioral changes that, over the time, have led from wolves to dogs and, consequently, to the strong dog-human relationship. To address this issue, we deal with different aspects of wolf sociality. In particular, we "take a glance" to dominance, post-aggressive and playful dynamics between pack members by comparing the findings on wolves with those on dogs [29]. Can the wolf-dog behavioral difference be credited only to the domestication process? Can the domestication process have induced a shift of social tolerance and attentiveness from conspecifics to humans, thus leading dogs towards an exclusive inter-specific cooperation?

#### **2. Social Tolerance by Dominants, Leverage Power by Subordinates, and Peaceful Strategies by All Group Members**

Wolf pack is defined as a cohesive family group, including a long-term bond breeding pair, mature offspring, and pups; occasionally, an unrelated individual may join the group [30,31]. All wolves participate in pack life by creating a system of division of *labour* in which individuals cooperatively hunt and defend their territories and collectively rear the pups [30,32]. The socialization begins at around four months of age when cubs start to follow the adults on hunting trips ("hunting school"). During this period, puppies improve their motor and perception skills and perfect mutual interaction and coordination with conspecifics. The affinitive relationships develop during puberty, when maturing individuals are slowly integrated into the daily life of the group [30–32]. The strict social association between pack members finds support in a study of Cassidy and McIntyre [33]. The authors recorded 121 territorial inter-pack conflicts in Yellowstone National Park, and in 17.6% of cases, wolves engaged in aggression to defend their pack fellows.

Within the pack, puppies generally occupy lower ranking positions compared to their parents and older siblings. When wolves reach sexual maturity (~2 years), they disperse from their natal group, attempt to pair with other dispersed wolves and start their own packs, thus avoiding competing for dominant-breeder *status* with natal group members [30,32]. However, under some conditions, both in the wild and in captivity, mature individuals delay dispersal or do not disperse at all; in these cases, competition for dominant-rank may be stronger [32,34,35]. In captive packs, wolves often have a linear hierarchy in which all males are dominant over females [32,36,37]. Nevertheless, the more appropriate term to define dominance relationships in a typical wolf pack (nuclear, extended or complex families) is "age-graded dominance hierarchy" [31,36]. Moreover, the subordinate individuals can sometimes oppose their leader's actions; for this reason, Zimen [38] defined the leadership in wolf packs as a "qualified democracy", in which no subject decides alone to carry out activities that are crucial for the group survival. Recently, Range and colleagues [21] found that, under feeding conditions, captive wolves are more tolerant compared to dogs. Indeed, high-ranking dogs monopolized the resources while low-ranking individuals showed deference by staying apart without trying to obtain food from the dominant subjects. Conversely, subordinate wolves overtly challenged the dominant ones to subtract food from them. Wolves are cooperative hunters [34] and, in term of mutual beneficial exchanges, all members of the pack have the possibility to access food, independently of their ranking position. On the contrary, domestic dogs rely on humans for food and feral dogs are solitary scavengers, hence, both do not depend on conspecifics' support for obtaining the resource [39]. These findings suggest that, in the wolf society, the power is not entirely "in the hands" of the physically stronger subjects. The subordinates can exert leverage power [40,41] because of their support to the pack life and their cooperation is gained by high-ranking individuals through peaceful sharing instead of aggressive coercion [42].

Nevertheless, despite this cooperating social system, the presence of aggression is the other inevitable side of the coin that leads to a temporary interruption of the inter-individual relationships [29,43]. To cope with aggression and the consequent social damage, as it occurs in many social mammals (human primates [44]; non-human primates, [45–49]; dolphins [50]; spotted hyenas, [51]; red-necked wallaby [52], wolves engage in post-conflict contacts such as reconciliation (i.e., the first affinitive contact exchanged by the former opponents relatively shortly after a conflict [53]).

By analyzing 3344 conflicts, Cordoni and Palagi [37] provided the first evidence for the occurrence of reconciliation in wolves by observing the pack hosted at the Pistoia Zoo (Italy), which was categorized as a "disrupted family" due to the absence of the alpha female. The high level of conciliatory contacts was uniformly distributed across the different sex–class combinations. Interestingly, reconciliation was not linked with rank distance between opponents but it positively correlated with coalitionary support (defined as a third party joining an ongoing conflict by attacking one of the opponents in support of the other, [54]). Generally, in social mammals, the high level of support can unveil high level of cooperation. In wolves, alliances [55] and reconciliation [37] act as diffuse non-dispersive mechanisms that concur in strengthening group cohesiveness. The occurrence of reconciliation was confirmed by other studies both in wild [56] and captive wolves [57]. Also, in the wild condition, wolves showed high level of conciliatory contacts and, once more, the finding was explained by the authors in the light of the strong cohesion between pack members [56].

Even though few doubts remain for the occurrence of reconciliation in wolves [37,56,57], contrasting results derive from canine reconciliation. Cools and co-workers [58] demonstrated reconciliation in small groups of dogs sharing a pen. Recently, Cafazzo and colleagues [57] by studying four captive small packs of dogs and wolves provided evidence for reconciliation in wolves, but not in dogs. Indeed, in this study, the dogs avoided affiliating with their opponents after conflicts. The social repairing function ascribed to the reconciliation mechanism [53] is probably useless for dogs. The difference in socio-ecological habits between dogs and wolves may cause the difference in their conciliatory tendency. In dogs, the absence of cooperative hunting and collective rearing of offspring limits the need to maintain friendly and peaceful social relationships with conspecifics [57].

Beyond reconciliation, other types of post-conflict interactions can occur. Group members not involved in aggression (bystanders) can spontaneously offer friendly contacts to both victims ("consolation"; [53]) and aggressors ("appeasement"; [59]). Even though the cognitive and emotional skills underpinning these contacts are still under debate, "consolation" and "appeasement" seem to serve different functions.

In wild and captive wolves both "consolation" and "appeasement" are present [43,54,56]. In the Pistoia pack [43,54], the two post-conflict behavioral strategies occur within two minutes after the end of the aggression and are performed with comparable levels. Despite these similarities, "consolation" and "appeasement" seem to play different roles. In the Pistoia wolves, in about 45% of cases, bystanders offer affinitive contacts to aggressors, which generally occupy high-ranking positions. Such calming interactions have the immediate effect to reduce the likelihood of renewed aggression toward other group members by the previous aggressor. On the other hand, affiliation that bystanders direct towards the victims follows the relationship quality linking the subjects more than their hierarchical positions: the stronger the bonding, the higher the frequency of affinitive contacts. Furthermore, these contacts protect the victim against the reiterated attacks from the previous aggressor. In sum, "consolation" may represent a "victim protection" strategy and "appeasement" a "bystander protection" strategy, thus highlighting the functional dichotomy characterizing these two behaviors [43,54]. The Figure 1 illustrates a reciprocal muzzle licking during a post-conflict triadic affiliation also named "consolation".

**Figure 1.** Reciprocal muzzle licking during a post-conflict triadic affiliation ("consolation") between an adult (the consoler) (**a**) and an immature subject (the victim) (**b**) Photos by Elisabetta Palagi.

Although bystander post-conflict affiliation has been demonstrated in dogs [58], it does not seem that familiarity between interacting subjects significantly affects the behaviour. Cools and colleagues [58] showed that bystander affinitive contacts were more frequently directed to victims than to aggressors. Moreover, victim-directed affiliation was markedly higher than affinitive interactions between the former opponents. The scarcity of data did not permit the authors to investigate the possible functions of dog "consolation" and "appeasement" leaving the question still open.

#### **3. The Playful Wolf: Tactics and Plasticity for a Deeper Knowledge of Others**

Beyond dominance, affiliation, and conflict management, social play is another type of interaction that can affect the social dynamics within a group [60]. Immature individuals of many mammalian species engage in different types of playful activities [61]. The pervasive distribution of play suggests that the core neural circuitry underpinning the modulation of this behavior may have evolved early in mammals and it may be shared by different species [62].

During play, animals execute motor action patterns that are recruited from "serious" contexts such as agonistic, anti-predatory and sexual [61]. However, these patterns are performed in an exaggerated, incomplete, repeated, mixed and unexpected manner [63–65].

The first playful experience occurs between mothers and infants [66]; these sessions represent a good training ground for preparing infants to the future interactions with peers [67,68]. During ontogeny, play concurs in implementing physical, cognitive, emotional, and social skills of individuals by guaranteeing a safe and fruitful "environment" in which making training [45,63,69–73]. Held and Špinka [74] suggested that play could provide individuals with psychological benefits by promoting opioid-mediated pleasurable experiences: the better individuals feel, the more individuals play. The resultant positive feedback between play and animal well-being becomes an important tool for improving and ameliorating the management of animals in captivity.

Although benefits provided by social play can vary depending on context (e.g., pre-feeding, mating period), habitat, playmate characteristics (e.g., sex, age, and rank) and species sociality (e.g., tolerant versus despotic), the intrinsic nature of play requires an implicit agreement, cooperation, and negotiation between players, who have to trust one another [65,75–77]. Playmates reach this goal by fine-tuning their contact interactions and by avoiding the performance of behaviors that might be misinterpreted. By playing fairly, animals may acquire the social competence and rules that are at the basis of a peaceful coexistence [60].

Peaceful cohesiveness is a feature of wolf society that is guaranteed also by playful activity [54,60,78]. Although most studies have focused on play in domestic dogs [6,79–81], this behavior does not appear as an artefact of domestication since also wolves play even during adulthood [78,82].

It has been suggested that, through play, adult subjects can evaluate and manipulate the social relationships with group-members [71]. Cafazzo and colleagues [82] investigated play behavior in four captive wolf packs, two composed by immature peers and the other two by mixed-age subjects (puppies and adults). A positive linkage between play frequency and relationship quality was found; indeed, those dyads that spent more time in relaxed play (defined as play sessions involving a limited number of offensive patterns) engaged in more interactions belonging to the affinitive domain. Moreover, in mixed-age groups, but not in peer groups, the frequencies of aggression are negatively correlated with play levels. Interpreting this result, the authors suggested that play can limit aggressiveness between group-members only after the establishment of clear hierarchical positions.

The observation of the wolf pack hosted at the Pistoia Zoo (Italy) led to contrasting results compare to those of Cafazzo and colleagues [82], although direct comparison can be misleading due to the different conditions under which the wolves were reared. In the Pistoia group, composed by adult related individuals, playful activity was not affected by both social relationship quality and aggression level, but it was strictly linked to dominance. Play was negatively correlated to the rank distance between fellows; in other words, subjects with closer ranking position played more than subjects differing greatly in rank [78]. In another study on the same pack, Cordoni and Palagi [54] compared the level of aggressiveness and steepness of hierarchy in two different periods (Sample 1 and Sample 2). During Sample 1, two adult females died and this event probably provoked delayed social effects which were manifest during the Sample 2 characterized by higher aggressiveness and steeper linear hierarchy. The re-arrangement of the dominance relationships within a pack occurs gradually and it is

usually manifest later in time respect to the occurrence of the perturbing event (e.g., the removal of particular subjects; [83]). Intriguingly, in the Pistoia pack, the hierarchical difference between the two Samples had a significant effect on playful but not on affiliative dynamics. During the riskier period (Sample 2), wolves consistently reduced their playful activity, avoided playing during high-tension contexts (e.g., pre-feeding time) and limited the number of players per session by preferring dyadic (only two players) at polyadic interactions [54]. As a whole, these findings suggest that "rank rules dictate play rules". By playing, wolves can acquire information about physical and cognitive skills of fellows with whom they can compete in the future for dominant position. Nevertheless, in order to maintain a "not serious" mood, wolves had to manage play in a flexible manner by place the interaction in the right time, in the right context, and with the right players [45,54,72,73,78].

Pal [84] observed 24 free-ranging dog puppies from birth to 13 weeks of age. In each litter, there were particular subjects that initiated offensive play more often than the other siblings did. Moreover, the same-sex offensive play was the predominant interaction between puppies. These findings suggest that, starting from five weeks of age, free-ranging dogs can employ playful activity to acquire information about individuals with whom they may compete in the future for reaching dominance positions, as it occurs in wolves [84].

In domestic dogs, adult-adult play does not follow the 50–50 rule that is a balance between offensive ("dominant") and defensive ("subordinate") patterns exchanged by playmates during the interaction. In play, adult dogs seem to hold the same dominance position they have outside the playful context [79]. Also, dog puppies do not adhere to the 50–50 rule by performing offensive playful patterns much more frequently than defensive ones [80]. Play becomes more asymmetrical as the puppies matured. Along with dog developmental pathway, the winning and the losing positions during playful interactions mirror the dominance relationships between individuals. Intriguingly, Cordoni and colleagues [6], by studying 49 domestic dogs in an off-leash dog-park, evidenced that the level of play asymmetry did not differ between "friends" (i.e., dogs that lived together or regularly interacted) and "strangers" (i.e., dogs that have never interacted before the observation). In the light of this result, dog play can be not only predictive of the dominance relationships between playmates, but it may serve an important function in maintaining good social bonds with specific partners. Nevertheless, independently of the inter-individual relationship, the playful sessions characterized by high asymmetry, and, consequently, by high competition, had a shorter duration compared to the more balanced sessions. The decrease in play duration and the use of clear communication can represent strategies used by dogs for overcoming the risk of escalation during very asymmetric playful interactions [6].

In wolf puppies, social play seems to be well-balanced with both immature partners performing a similar amount of self-handicapping behaviors, this reciprocity decreases when one of the players is an adult. When play involves two mismatched wolves, the session generally becomes more asymmetric [85]. The flexibility in managing the playful arousal is evident in the study by Cordoni & Palagi [54]. The adult wolves of the Pistoia group changed their play modality according to the period of observation. In Sample 1, the period characterized by low level of hierarchical steepness and aggressiveness, subjects engaged in more self-handicapping and role reversal manoeuvres, thus making their playful sessions more symmetric. This suggests that animals are able to flexibly adjust their playful tactics according to the social circumstances and that the play asymmetry is not always predictive of the dominance status of the players. In this view, the analysis of some social factors such as the exact quantification of (i) the hierarchical steepness, (ii) the level of affiliation and (iii) the bidirectionality of agonistic conflicts is mandatory before beginning a study on play in adult wolves.

#### **4. Conclusion**

*"Can the wolf-dog behavioral di*ff*erence be credited only to the domestication process? Can the domestication process have induced a shift of social tolerance and attentiveness from conspecifics to humans thus leading dogs towards an exclusive inter-specific cooperation?"*

Through a systematic comparison between data coming from the available studies on wolves and dogs we propose possible answers to these questions.

A group of wolves moves as a unique entity, with each subject relying on others' support to gain benefits and increase survival and fitness. In this network of cooperating individuals, some are dominants and some subordinates. These hierarchical relationships are not based on a mere deference of subordinates towards the dominants but on an exchange of services that both counterparts seem to put in act. In the wolf cooperative system, the social tolerance provided by dominants to subordinates can be repaid by subordinates with their help and support in the group maintenance activities. This does not necessarily mean that individuals cannot use their high-ranking position to obtain priority in certain domains. Indeed, a dominant can exert a strong control over resources to the detriment of subordinates. However, keeping everything under control through threats and overt aggression is energetically demanding, and, for this reason, trading can be also present when coercion by dominants is possible. In these situations, also the subordinates can exert their little amount of power over the dominants, that is called leverage. This delicate equilibrium permits and sustains the development of cooperation in social groups.

A further prerequisite for the development of a cooperative society is social attentiveness, which occurs when a subject is sensitive to others' behaviors and needs. The level of attentiveness can change according to different factors such as contexts, kinship, relationship quality shared by the interacting subjects. When an individual is attentive to fellows, he/she can adjust and coordinate his/her competitive or cooperative reactions. The link between attention and action coordination appears clear in the "consolation" and "appeasement" dynamics occurring after a conflict. The ability to perceive the emotional mood of the victim (anxiety) or the aggressor (arousal) is highly beneficial to the subject who can reduce, through the post conflict affiliative interaction, the probability of further aggression thus concurring in the maintenance of group cohesion.

An important mix of cooperation and social attentiveness is mandatory during play fighting, an activity which is complex to manage and that can lead easily to misinterpretation. Animals have to finely read and rapidly interpret each single pattern of the playmate in order to react in an appropriate manner. This needs a profound knowledge of the partner, because the reaction can change as a function of the relationship linking the two players. This is a highly cognitive demanding behavior which comes into place when animals are strongly motivated to cooperate.

In conclusion, the domestication process has acted over *cooperative baggage* already present in the sociobiology of the dog's ancestors. In wolves, such *baggage* has been driven by natural selection towards conspecifics, while in dogs, this *baggage* has been redirected by the artificial selection towards humans.

**Author Contributions:** Conceptualization, G.C. and E.P.; Writing—original draft preparation, G.C. and E.P.; Writing—review and editing, G.C. and E.P.

**Acknowledgments:** Thanks are due to the Pistoia Zoo (Italy), its Director Paolo Cavicchio, the wolf keepers for allowing and facilitating the work; Maria Vittoria Ulivi, Iolanda Ferranti, Chiara Stefani and Federica Pacini for their collaboration in the data collection.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Erratum*
