Effects of Olfactory and Auditory Enrichment on the Behaviour of Shelter Dogs.

Shelter environments are stressful for dogs, as they must cope with many stimuli over which they have little control. This can lead to behavioural changes, negatively affect their welfare and downgrade the human-animal bond, affecting re-homing success. Arousal is evident in their behaviour, particularly increased activity and frequent vocalisation. Environmental enrichment plays an important role in reducing arousal behaviour, either through direct physiological effects or by masking stressful stimuli. The present study focused on sensory environmental enrichment, using olfactory and auditory stimuli under shelter conditions. Sixty dogs were allocated to one of four treatments: three types of enrichment, Lavender, Dog appeasing pheromone (DAP) and Music, and a Control group. Stimuli were applied for 3 h/d on five consecutive days. Dogs exposed to DAP lay down more, and those exposed to Music lay down more with their head down, compared to the Control. Those in the Control stood more on their hind legs with their front legs on the exit door, compared to those exposed to Music and DAP, particularly if they had only been in the shelter for a short time. They also panted and vocalised much more than dogs in the three enrichment treatments, which tended to persist during the 4 h period post treatment, and in the case of vocalisation into the subsequent night. The study suggests that all three enrichments had some positive benefits for dogs in shelters, as well as being non-invasive and easy to apply in the shelter environment.


Introduction
Behaviour problems are a major reason for dogs to be relinquished to rescue shelters [1]. Even though shelter staff attempt to give these animals a good quality of life, shelters are inherently stressful environments [2][3][4]. Stress is defined as a state in which homeostasis is threatened by intrinsic or extrinsic negative effects known as stressors [5,6], examples of which are noise pollution, novelty, example, when challenged with stressful situations, male dogs show more tail wagging, higher posture but less paw lifting and changes of locomotion and posture than females [51]. When approached by a familiar person, female dogs show less ambivalent postures than males, and when undisturbed, they wag their tails more than male dogs [52]. Similarly, age may affect responses. Puppies have reduced cortisol responses and are more relaxed than juvenile/adult dogs when challenged [53]. Older dogs are less likely to explore their environment [48], sleep more during the day and less at night [54] and urinate more frequently [54]. However, some studies have not found differences in dog behaviour due to their sex and age [22,55,56]. It is also important to consider the lengths of dogs' stay in shelters, since they may adapt and the environment may become familiar over time [57][58][59], with some [21,58,60] but not all [55,57] studies showing a reduction in activity over time.
Although the stimuli discussed above have been demonstrated to have positive effects on the behaviour of shelter dogs, there has been no study that systematically compares different types of sensory stimuli in a shelter context. Therefore, the aim of this study was to compare the effects of music, lavender and DAP as enrichments on the behaviour of shelter dogs, and particularly to examine which reduces arousal the most.

Kennel Environment
This study was conducted at the Royal Society for the Prevention of Cruelty to Animals Queensland's (RSPCA Qld) Animal Care Campus at Wacol, Brisbane, Australia, between August and November 2017. It occurred concurrently with regular shelter activities, such as routine cleaning, feeding and walking, and staff members and volunteers were always present in the kennel blocks. Each block consisted of 16 kennels (Figure 1), each divided into two rooms of 8 kennels (two rows of four) separated by a door. Dogs were individually housed in kennels, which had dimensions 1.6 × 4 m, and included a crate measuring 0.72 × 1.55 m and a bed. Both sides had plastic walls that prevented dogs from seeing each other. The back of the kennel had thin metallic bars from roof to floor, which permitted dogs to look outside and air to circulate. The front door had a solid section at the bottom and the same metallic bars from the top of the solid section to the top of the door. Each kennel had two water bowls that were refilled during the day when necessary. The dogs were taken for walks during the morning cleaning and in the afternoon while their kennels were spot cleaned. They were fed dry food twice a day. They were walked twice a day by volunteers for 10 min each time, and had occasional contact with volunteers at other times, except for the 3 h when they were exposed to the stimuli. Kennels used for the measurement of the behaviour of dogs (n = 60) exposed to, Music, DAP, Lavender stimuli or a Control. K7 and K8 were the two blocks of dog kennels used for the study. 1: kennels for the Lavender treatment; 2: kennels for the Music treatment (separated from the other treatments by solid dividing walls); 3: kennels for the Control treatment; 4: kennels for the DAP treatment; *: kennels not holding a study dog.
On entry, all dogs had a veterinary clinical examination and a standardised behaviour assessment as described in Clay et al. [61]. Dogs with high arousal-related behaviours, such as air snapping, mouthing, attempts to bite lead or handler, excessive activity, constant vocalisation and over-reaction to other dogs, were selected for our study weekly by the RSPCA Qld Behaviour Team, based on their kennel behaviour and on information given by shelter staff working with the dogs. Dogs in need of immediate behaviour intervention as a result of hyperarousal, according to RSPCA Qld protocols, were not selected. RSPCA Qld staff were responsible for selecting the participant dogs and placing them in the study kennels; they were blind to the treatments and assigned dogs at random to each kennel as they became available.

Study Design
Three forms of enrichment were applied in this study: Music (n = 14), Lavender (n = 15), Dog appeasing pheromone (DAP) (n = 16) and compared to a Control (n = 15). The kennels used for the Lavender treatment were in a separate kennel block to avoid other dogs in the study being influenced by the lavender, which potentially could spread to surrounding kennels. The other two enrichment treatments and the Control were set up in adjacent kennel areas divided by a wall and door, DAP and the Control in one half and Music in the other half, to prevent dogs in other treatments hearing the music. All kennels in the block were occupied throughout the study period. Dogs were exposed to the stimuli in their kennel for 3 h/d on 5 consecutive days. This commenced between 10.30 and 13.30 h, depending when their walk, breakfast and cleaning of the kennels finished. Dog behaviour was also observed in a post-treatment observation period from 14.00 to 17.00 h. At 16.30 h kennel duties ended. A final night period was observed between 18.00 and 09.00 h on the following day, except for day 5 when recoding finished at midnight. Our behavioural observations used cameras, detailed below, and were therefore completed without disturbing the dogs [62], providing a measure of the animal's response to its surroundings, indicative of its stress response [22].
For the Music treatment, a solo instrument, the pianoforte (hereafter piano), was selected because it needs less neurological processing than multiple instruments [63], and a wide variety of tracks are available. The piano was the sole instrument, except in 6 tracks in which there was accompaniment by violins for part of the tracks. For the music selection, we used an automated filtering mechanism based on an audio analysis algorithm [64]. The programme employed a psychoacoustic approach to assess audio features of the tracks [65], generating a metadata file with the required track structure and music attributes, including rhythm, pitch and timbre. First a large number of songs (301) were downloaded from a common music platform (Spotify, www.spotify.com/, see Appendix A for list of tracks). In this overall set of tracks we applied filters -one readily quantifiable parameter (tempo, in beats per minute) and two qualitative parameters (valence and energy, both on scales of 0-1.0) describing the characteristics of the overall track's sound. Only tracks with a tempo of 70 or fewer beats per minute were selected, in an attempt to entrain the dogs' normal resting heart rate of 70-120 bpm [66] to a lower value. Only tracks with a valence from 0 to 0.5 were selected, thus the musical positiveness was less euphoric. Low valence is associated with low heart rate, respiratory frequency and electrodermal activity, all physiological markers of parasympathetic activity [16]. Tracks were selected with an energy rating of less than 0.2, reflecting low intensity and activity. Energetic tracks feel fast, loud and noisy, as is the case with Death Metal, compared with, for example, a Bach prelude [67]. While these parameters are based in human psychoacoustics, we extrapolated them to our canine population on the assumption that music has similar effects in dogs as in humans.
A selection of51 tracks that fit the above criteria were played for 183 min each day with random track selection order on a Motorola ® mobile phone connected to a Logitech ® speaker. Both were placed in a plastic holder hung on the crate's door (in the middle of the kennel) to make sure the dog could hear the music. The music was played at 70 dBA, measured using a Digitech ® mini sound level meter at the beginning of the Treatment, as used previously [68,69]. Sound was also recorded in the kennels used for other treatments to ensure that the music was not audible, recognising that there is better auditory acuity in dogs, compared with humans, at high frequencies [70].
For the Lavender treatment, two ultrasonic diffusers (Select Botanicals, Gladesville, New South Wales, Australia) were placed in each kennel, one in the crate and one at the back of the kennel, to make sure the dog was exposed to the odour, regardless of its location in the kennel. Both diffusers were placed under milk crates to avoid being damaged or tipped over by the dog. The dilution was 4 drops of 100% organic Bulgarian lavender (Lavandula Angustifolia) (Select Botanicals, Gladesville, New South Wales, Australia) in 60 ml of water.
For the DAP treatment, 3 and 5 pumps of a synthetic analogue of the canine appeasing pheromone (15.72 mg/mL; Adaptil ® , Ceva, Glenorie, New South Wales, Australia) were sprayed on a bandana worn by the dog, the dog's bedding as recommended by the manufacturer and at three different point of the kennel's floor (2 back corners and front door).
The Control did not receive any extra sensory stimulus.

Data Collection and Analysis
Mini cameras with charge-coupled devices and infra-red facility (Signet ® , Electus Distribution Pty. Ltd., New South Wales, Australia) were fitted at the front and back of the kennels to record behaviour continuously (24 h/d during the 5 d of stimuli exposure). Behaviour was recorded and observed in three periods of the day: the Treatment period (3 h), 5 min observed every 15 min, i.e., 12 separate observations lasting 3600 s in total; the Post-treatment period (4 h), 5 min observations every 30 min, i.e., 8 separate observations lasting 2400 s in total; and the Night period, 5 min of each hour were observed, i.e., 16 separate observations lasting 4800 s in total. Behaviour coding software (Boris ® version 6.0.4 for Windows [71]) was used to record behaviours in an ethogram (Table 1), developed using previous literature and the researchers' experiences of dogs' behaviour in the shelter. Time values were then transformed into % values (duration of behaviour/total observation time × 100, in s). Some behavioural observations were missed because of technical issues with a camera, a dog was out of camera frame or absent when out for a walk (18 during the Treatment period, 0.5% of total observations, 283 in the Post-treatment, 11.8% of total observations,) and 783 in the Night, 16.3% of total observations). These were treated as missing values and taken into account when calculating the percentage of time that the behaviours were performed for.

Statistical Analysis
The behaviour data were statistically analysed using Minitab 18 software. A Principle Component Analysis was initially performed to inspect behaviours for similarity, which were combined if appropriate, as detailed in the Results section.
Each of the three observation periods (Treatment, Post-treatment and Night) was analysed separately. Treatment effects on behaviour were analysed using a Mixed Effects Model, which was constructed using dog as random factor and dog number (entry time to the study), treatment and day as fixed factors. A more complex model including age and sex and their interactions with treatment and treatment x day was tested, but no significant interactions of these two factors with treatment were found, and therefore these factors were removed from the model. Length of stay was also tested but no significant treatment effects on behaviour were found, with the exception of standing at exit door, which is detailed in the results. Residuals were inspected for conformity to normality using the Anderson-Darling test. Square root transformations were used if necessary, to secure normal distribution of residuals. Differences between individual treatments were examined using Tukey's test when treatment differences were detected.
Even after creating combinations of behaviours, some still did not have enough data for statistical analysis. For these, the data from the five days were collapsed and treatment effects determined using a General Linear Model (GLM) constructed with dog number and treatment as fixed factors. The same normality-testing steps mentioned above were followed in the analysis of the data. The following behaviours were very rarely observed and were not analysed statistically: paw lift, lick nose/lip and play bounce.

Results
Using the PCA results, similar behaviours were combined as follows: pace repetitively + spin, pawing + scrabble at bars/door/wall, chew bedding + chew objects (bandana, monitors and milk crates/diffuser wires), roll on ground + circling, and tail medium + tail high.

Treatment Effects on Behaviour During the 3 H Treatment Period
Dogs exposed to DAP spent more time lying down, compared to the Control, with those in the Music and Lavender treatments intermediate ( Table 2). Animals exposed to Music spent more time lying down with their head down, compared to the Control, with those in the DAP and Lavender treatments intermediate. There were no treatment effects on the time that dogs spent lying down with their head up. There was a trend for dogs in the Music and DAP treatments to stand and walk less than those in the Control (p = 0.08 and 0.09, respectively). Dogs in the Control spent more time standing on their hind legs with their front legs resting against the exit ('standing exit door' in the ethogram), compared to the dogs exposed to Music and DAP, with those exposed to Lavender intermediate. There was a significant length of stay x treatment effect for standing at the exit door ( Figure 2). In the Control, dogs that had been in the shelter for only a short time spent longer performing this behaviour, whereas those in the enrichment treatments did not. Dogs in the Control spent approximately three to four times as much time vocalising as dogs in the three enrichment treatments. Dogs in the DAP treatment shook themselves the most, and they sniffed the ground more than the dogs exposed to Music, with those in the Lavender and Control intermediate. The dogs in the Control panted much more than dogs exposed to any of the enrichment treatments. They also excreted and moved their tail more, compared to Music and DAP, with those exposed to Lavender intermediate. There were no effects of treatment on position in the kennel.

Residual Treatment Effects on Behaviour during A 4 H Post-Treatment Period
Dogs that had been in the Control tended to spend less time lying down compared with dogs that had been in the three enrichment treatments (Table 3). They also tended to spend more time standing at the exit door (p = 0.08). They vocalised for longer, when compared to those that had been exposed to Lavender and DAP, with those exposed to Music intermediate. Dogs that had been exposed to DAP spent more time sniffing the ground when compared to those that had been exposed to Music, with Lavender and Control intermediate. Dogs that had been in the Control spent more time drinking water than those that had been in the Music treatment, with those that had been in the Lavender and DAP treatments intermediate. Those that had been in the Control also spent much more time panting than those exposed to Lavender, with those that had received Music and DAP intermediate. There were no effects of treatment on tail position/movement or dogs' position in the kennel in the Post-treatment period.

Residual Treatment Effects on Behaviour during the Night Period
During the Night period, there was a trend for dogs that had been in the Control treatment to still vocalise more frequently than those that had been in the three enrichment treatments (p = 0.06) ( Table 4). Dogs in the Lavender treatment groomed themselves more when compared to those that had been in the Music treatment, with those that had been in the DAP and Control treatments intermediate.

Treatment Period
The results of this study are in line with other studies where shelter dogs were exposed to similar stimuli: classical music [20][21][22], lavender [39] and DAP [46]. Dogs exposed to these stimuli displayed more restful behaviours and vocalised less in all the studies. The results from the present study suggest that sensory environmental enrichment (auditory and olfactory) can help to reduce arousal-in shelter dogs.
During the Treatment period, when exposed to DAP dogs spent more time lying down compared to the Control, and when exposed to Music they spent more time lying down with their head down compared to the Control. These resting behaviours are associated with increased relaxation and lower arousal [73] and stress levels [20,21,39], and lying with their head down is particularly likely to be indicative of relaxation. Considering the busy shelter environment during the day, being able to rest more might be indicative of improved welfare [73].
Dogs in the Control treatment spent more time vocalising, a possible indicator of stress [74,75]. They also spent more time standing on hind legs with front legs resting against the exit (standing exit door) than dogs in the Music and DAP treatments, particularly if they had not been in the shelter for long. This behaviour may be escape motivated due to an interest in events happening outside of the kennel, but could also indicate boredom [76] or a quest for someone to come and see them. Dogs in this treatment also panted more, a behaviour that may suggest an elevated stress response in dogs [51,77,78]. It could also be related to the higher activity levels, or physiological responses to arousal increasing body temperature [51], with the dogs panting as a cooling mechanism [78]. Control dogs also excreted more than those exposed to Music; it has been reported that dogs kept under austere conditions and presumably experiencing higher stress levels show increased excretion [75,79].
Dogs in the DAP treatment shook themselves most, a behaviour that has been associated with acute stress [80] and the release of tension [75]. But considering that the other behaviours performed by dogs in this treatment are associated with more relaxing behaviours (more resting and less barking), the shaking could be due to some dogs finding the DAP bandana uncomfortable. Dogs from this treatment also sniffed the ground more than dogs exposed to Music; this is considered an investigatory behaviour [81] and they were possibly following traces of the pheromones as this was spread around the kennel's floor.
The dogs' location in kennel was not influenced by any of the stimuli, and nor was it affected by music in the work of Wells et al. [21], suggesting that they did not actively seek out the source of the auditory stimulation.
Dogs in the Control treatment spent more time wagging their tails in comparison to dogs in the Music and DAP treatments. Tail wagging in dogs can be motivated by different things, such as play and appeasement, but other body signals can also infer the dog's emotional state [82]. In this study, the increase appeared related to other behaviours connected to arousal, which is supported by other studies in dogs [9,72,75,83]. Beerda et al. [75] found that when dogs were challenged by situations such as poor housing, they increased tail wagging and had more changes in locomotion.
Dogs exposed to Lavender showed the least behavioural differences when compared to the Control. This could be due to the lavender oil being very volatile and easily escaping the kennels, losing concentration and producing fewer anxiolytic effects in the dogs. It may be that concentrations near the floor were particularly low, diminishing any effects on the time that they lay down. Exposure to lavender oil, as a novel scent, can increase rather than decrease arousal in anxious animals, as documented by Hawken et al. [38] in a study in which nervous sheep exposed to lavender oil showed increased anxiety (more vocalisation, escape attempts and higher cortisol concentrations). In their study, calm sheep exposed to lavender showed behaviours related to reduced anxiety, such as less activity and vocalisation. Another possibility is that the lavender oil was not very effective producing behaviours associated with relaxation in the dogs tested.
The lower levels of tail movement (primarily wagging) and standing at the exit door in the Music and DAP treatments and the lower levels of vocalisation in the three enrichment treatments suggests that even small changes to their environment can help reduce arousal. Furthermore, the trend towards increased walking and standing in Control dogs suggests that they are less relaxed.

Post-Treatment Period and the Night Period
During the Post-treatment period, dogs from the Control treatments drank more water than those exposed to Music. Increased water intake has been correlated with higher activity levels [51] which were performed by the Control dogs.
Dogs exposed to Lavender and DAP vocalised less than the Control treatments, indicating persistence of this behaviour, even into this Night period (Figure 3). High noise levels can affect the dogs' auditory system and other physiological systems, such as the immune and endocrine systems [84]. They can also be detrimental for shelter workers' auditory systems [84]. This major effect on vocalisation, a common stress response, could be utilised as a key stress indicator in shelters. Automatic detection may be possible, allowing attendants to monitor changes in stress levels in dogs in their charge. For the other behaviours that appeared to respond to stimuli, lie down total, stand, walk and standing at the exit door, the post-treatment effects were similar between the Treatment and Post-treatment periods, indicating some persistence of the beneficial effects of the enrichment, but these had mostly disappeared by the night time ( Figure 3).
During the Night period, dogs from the four treatments spent most of the time lying down and dogs in the Control barked less than during the other two observation periods. During this period the shelter has very little or no human presence, therefore there are not many external noises or stimuli to startle the dogs and arouse them. It seems that by this observation period most treatment effects had disappeared.

Other Influences on Behaviour
The dogs' responses to the stimuli remained similar during the five days of the trial, as there were no day x treatment interactions, suggesting a lack of habituation. Previous studies also failed to find habituation when dogs were exposed to lavender and classical music [21,39]. With music the extent of habituation may depend on the repetition frequency. As we had 51 tracks in our playlist, there was no repetition on the day, only across days and even then, the sequence was varied. Hence we can assume that there was insufficient repetition to produce habituation, as shown by Bowman et al. [22].
Even though there is evidence of age and sex affecting behaviour in dogs, we did not find that treatment effects on dog behaviour correlated to these variables, in line with other studies [22,55,56]. Length of stay evaluations have suggested that long shelter stays may produce changes in behaviour that could deter adopters, increasing the chances of further undesirable behaviours [85]. Contrastingly, Titulaer et al. [60] suggest that the dogs' individual shelter experience may influence their behaviour, rather than the amount of time spent in care. Bowman et al. [22] found a relationship between time spent sitting and length of stay in a shelter, with dogs kenneled for more than three months spending more time sitting down when exposed to an auditory stimulus. However, as they did not find associations between length of stay and more relevant activities, such as lying down and standing, they concluded that behaviour responses were not influenced by length of stay.
External variables must also be considered when trying to help shelter dogs to be more relaxed. Volunteers are extremely important for shelters, as the number of animals in care can be very large and to be able to give the animals the best care possible, many people are needed. This also means that there is a lot of human movement noise in the kennels, which can contribute to high arousal [86].
In this study, it was observed that when dogs were resting, noises like slamming doors would make them react immediately: walking around the kennel, jumping at the door, circling and barking. Dogs from all treatments reacted to this external stimulus, but these behaviours were exacerbated in the Control group. In Beerda et al.'s study [79], dogs in the most austere conditions reacted more actively to disturbances such as slamming doors. It is possible that some beneficial effects could be achieved in a shelter just by limiting the external stimulation that the dogs experience. For example, movement of people in and around kennels could be forbidden for several hours every day. The effect of such 'quiet time' would be worth investigating.

Limitations of the Study
We utilised only single levels of lavender release, a single volume for the music, and a single administration of DAP. It is quite possible that we did not use the optimal levels of any one of these treatments. Shelters are already noisy environments and if the extra acoustic stimulation is too loud, it could instead have a negative effect for the animals [87]. This could be equally true for lavender, as moderate doses are considered anxiolytic, but high doses cause sedation [88]. However, the lower incidence of vocalisation in all three treatments, along with behaviours associated with reduced arousal, e.g., lying down in the Music and DAP treatments, suggest that enrichment type may be more important than the level chosen.
In future studies an initial period could be recorded, that could be used as a covariate to account for differences between dogs before treatment. However, our study took place in a working shelter, and having the dogs in the same kennel for 5 days was already challenging as it meant the dogs were unavailable for adoption during this time. The random distribution of dogs between treatments in our study could have produced biases in terms of age and gender distribution, however, as these were not related to treatment effects, it seems unlikely that this affected our results.

Conclusions
Shelters can be very stressful environments and high arousal is a major issue that can stimulate undesirable behaviours in dogs in care. In our study, reductions in arousal-related behaviours were evident for dogs exposed to music and DAP for a 3 h period each day, and to a lesser extent dogs exposed to lavender for the same period. There was some persistence of responses over the 4 h post treatment, but in the subsequent Night period most responses had waned. There was no evidence of habituation to the enrichments over the 5 days of exposure for each dog. Shelters could consider using our music enrichment primarily, as is it the easiest and cheapest to apply. If music is used, the volume, type of music, extent of repetition and effects on care staff all need careful consideration. Even though DAP is expensive, it could still be justified for incoming animals showing high levels of stress, while they get used to their new environment. Lavender can only be recommended for more enclosed environments so that it persists, and dogs can get the full anxiolytic benefits of the essential oil. These types of stimuli are not invasive and easy to apply in a shelter environment, potentially having many positive effects in dogs experiencing high arousal.