Evaluating the effects of arousal and emotional valence on performance of racing greyhounds

The racing greyhound industry in Australia has come under scrutiny in recent years due to animal welfare concerns, including so-called behavioural wastage whereby physically sound greyhounds are removed from the racing industry because of poor performance. The non-medical reasons why greyhounds perform poorly at the racetrack are not well understood, but may include insufficient reinforcement for racing, or negative affective states associated with the context of racing. This study sought evidence for the affective states of greyhounds (n=525) at race meets and associations of those states with performance. It collected demographic, behavioural and performance data, along with infrared thermographic images of greyhounds at race-meets to investigate whether arousal influenced performance. It also collected behavioural data in the catching pen at the completion of races to examine possible evidence of frustration that may reflect sub-optimal behavioural reinforcement. Linear regression models were built to determine factors affecting greyhound performance. Increasing mean eye temperature after the race and increasing greyhound age both had a statistically significant, negative effect on performance. The start box number also had a significant effect, with boxes 4, 5 and 7 having a negative effect on performance. There was a significant effect of track on mean eye temperatures before and after the race, suggesting that some tracks may be inherently more stressful for greyhounds than others. Behaviours that may indicate frustration in the catching pen were extremely common at two tracks, but much less common at the third, where play objects in motion were used to draw greyhounds into the catching pen. The study provides evidence for the use of eye temperature in predicting performance, guidance for assessment of poor performance in greyhounds and suggested approaches to the management of frustration in racing greyhounds.


35
Greyhound racing in Australia is a sport supported largely by revenue from betting on the 36 outcome of races. As such, there may be pressure on owners of racing greyhounds to race 37 their dogs as often as they are physically able so that race winnings support the greyhounds' 38 upkeep and further racing activities. Recent scrutiny into the greyhound racing industry in 39 Australia, and in particular, the state of New South Wales (NSW), has raised questions about 40 the level of so-called wastage of greyhounds within the racing industry, which is where 41 greyhounds are discarded from racing because they do not perform the task they were bred 42 for, i.e., racing [1]. Wastage can be physical e.g., from lameness, or behavioural e.g., from 43 relative disinterest in running. The ultimate fate of discarded dogs is unknown, but may 44 include rehoming to a pet home, being retired but retained by the original owner or trainer, or 45 euthanasia. Failing to chase a lure is considered a form of behavioural wastage, which is 46 where otherwise physically healthy and sound animals are removed from a role because they 47 are unable to perform it adequately due to behavioural unsuitability [2]. It is therefore critical 48 to understand why greyhounds may fail to adequately perform the activity they were 49 specifically bred to perform. This is a multi-faceted issue with many potential contributing 50 factors, and little research has been conducted on it to date. Although it is likely that most 51 behavioural wastage has taken place before greyhounds reach the track [1], a key component 52 of understanding why greyhounds may fail to chase is in understanding their experience of 53 race meets. 54 Greyhound races in Australia begin with a so-called stir-up, which is where greyhounds may 55 watch the lure traverse the track usually twice while they are in a pen outside the track. Pre-56 stir-up occurs approximately ten minutes before the start of each greyhound race, and five 57 minutes before the stir-up under Australian greyhound racing rules [3]. Pre-stir-up involves 5 58 the collection of the dogs from their kennel and walking them to a grass area next to the track 59 where they may eliminate, and fitting them with a racing rug and any additional pre-race 60 preparations such as taping body parts for protection against injury. After witnessing the lure 61 traverse the track, the greyhounds are walked to their starting boxes, loaded into the boxes, 62 and then released from the boxes and chase the lure for a set distance. At the end of the race, 63 a gate is swung across the track immediately behind the lure to stop the greyhounds from 64 chasing it,. The lure itself draws away from the greyhounds and passes through a small flap in 65 this gate, and the greyhounds are diverted into the catching pen alongside the track, where 66 they are caught by handlers and then led from the track. The catching pen is unique to 67 Australia, and, it is argued by industry participants [4], may be a source of frustration for 68 greyhounds, as they are unable to capture the lure and there is rarely an object in the catching 69 pen for them to interact with in lieu of capturing the lure. The consequences of frustration 70 may be subsequent failure to chase or redirection of frustration onto nearby dogs, both of 71 which attract penalties if they occur during races rather than in the catching pen. Furthermore, 72 the risk of injury in the catching pen may be greater if greyhounds redirect frustration onto 73 nearby conspecifics as race participants are decelerating at different rates. 74 Many factors influence performance of racing animals. Previous research on racehorses has 75 shown that horses that finished as winners (top 20% of finishers) tended to be less aroused in 76 the mounting yard immediately before the race than the losers (bottom 20% of finishers) [5], 77 with arousal being determined by behavioural indicators. High arousal may lead to a 78 reduction in fine motor control [6], and may also compromise judgement and cognitive 79 processing [7]. These outcomes may manifest in racing greyhounds that show poor cornering 80 or manoeuvring around other dogs, starting a race too fast or expending excessive energy in 81 the stir-up and fatiguing early, or interfering with other dogs during the race. However, it is 82 impossible to tell whether a dog is performing according to a typical pattern they employ on 6 83 the racetrack, or if they are performing differently from usual, so it is necessary to seek 84 indicators of sub-optimal arousal. We therefore employed the use of an infrared 85 thermographic (IRT) camera to record the surface temperature of greyhound eyes before and 86 after the race. IRT detects infrared radiation, providing a pictorial representation of surface Heightened arousal prior to the race in greyhounds may be caused by distress related to the 98 racetrack environment including kennelling. Anticipation may also heighten arousal levels 99 [16,18,19], which may in turn be influenced by how long the dog has been kennelled for at 100 the race meet, or how many days it has been since the dog last raced, or how experienced the 101 dog is with the procedure at race-meets. A previous study on racing greyhounds revealed an 102 increase in arousal in dogs that race as well as those that have merely watched racing [20], 103 suggesting greyhound arousal increases with anticipation of an opportunity to race. Road 104 transport over an hour in duration is regarded as distressing for livestock animals [21], and 105 studies on air travel in dogs show it increases behavioural and physiological signs of distress 106 [22,23].

107
The current study aimed to determine possible effects of arousal and frustration on 108 performance in racing greyhounds at race-meets. As well as obtaining IRT images of 109 greyhounds before and after races, behavioural observations were collected of greyhounds 110 during the stir-up immediately prior to racing, and in the catching pen at the conclusion of 111 races to explore putative behavioural indicators of increased arousal before racing and signs 112 of frustration in the catching pen associated with being thwarted in capturing the lure.

114
The University of Sydney Animal Ethics Committee approved the current study (Approval The study was conducted at three greyhound racetracks in NSW over a period of 6 months.

119
The tracks were Richmond and Wentworth Park in the Sydney metropolitan area in June and 120 July 2017 respectively, and Gosford on the New South Wales Central Coast, approximately 121 80km north of Sydney, in October and November 2017. Data were collected from 3 race 122 meets at Richmond, with 11 races per meet, 2 race meets at Wentworth Park with 10 races 123 per meet, and 3 race meets at Gosford with 8, 10 and 11 races respectively.

124
Each track was configured differently (see supplemental material for diagrams). Minimum 125 distances between features of the track and where on the grounds greyhounds were subject to 126 potentially arousing stimuli were measured using the measurement tool in Google Earth Pro 127 (Google Earth Pro version 7.3.2.5776, Google LLC 2019).

128
During the period of data collection, the Richmond racetrack was trialling a bungee teaser in 129 the catching pen. Teasers consisted of two toys made of synthetic fur attached to one bungee 130 line each that was in turn anchored at the back fence of the catching pen. To offer teasers, the 8 131 track steward operating the catching pen gate walks onto the track with the teasers, stretching 132 both bungee lines taut. The steward releases the teasers as the dogs approach the catching 133 pen, providing a moving stimulus across the track and into the catching pen. The teasers 134 come to rest in the sand trap of the catching pen and the dogs are able to interact with them.

135
All greyhounds racing are muzzled, so interactions with the teaser are restricted. This system 136 was in place for all race-meets where data were collected at Richmond.

138
A total of 525 greyhounds were recruited to this study over the 8 race meets at 3 racetracks.

139
The races included were for both male and female greyhounds aged 1-6 years old, and dogs  Physiological data collection 149 IRT data were collected twice from each dog during each race meet, with the first IR 150 thermograph being taken during pre-stir-up. The second IR was taken 15 minutes after the 151 race while the greyhound was kennelled. Post-race kennelling proceeds after greyhounds are 152 hosed down and offered water to drink upon finishing the race.

153
IRT images were captured using a FLIR T640 Professional Thermal Imaging camera (T640, 154 FLIR Systems Inc. Danderyd, Sweden) at an 80⁰ or 100⁰ angle and 1m distance from the dog.

155
The FLIR ResearchIR Max software program was used to calculate the average eye 156 temperature under the 1234 palette because it best exposed the circumference of the eye.

157
Greyhound eye temperature was calculated by tracing the eye lids of the greyhounds using 158 the Stats tool then using Statistics Viewer to calculate the mean and max temperature inside 159 the traced area.  (Table 1 and 2) were recorded 166 for each dog. Greyhounds whose trainers excluded them from the optional stir-up event were 167 not analysed with an ethogram, and were recorded as being absent.

168
The ethograms were informed in part by Travain et al 2015 [8], who used an ethogram to 169 estimate distress in a group of 14 dogs, with behaviours being considered indicative of 170 distress when they were accompanied by a significant increase in eye temperature (detected 171 by IRT) [8]. No racing-specific ethogram for dogs has been developed before, so several 172 behaviours were added to the ethograms that were considered good candidates for detecting 173 high arousal, frustration, or fixation on the lure.

Behaviour Description Frequency
Rising (  in the stir-up were rarely more than 5 counts. The only exception was barking, which is 202 energetically a much less costly behaviour and is also much quicker to perform than other 203 behaviours in the ethograms. All behaviours were scaled using the max-min method to a 204 scale of 0-5 counts to avoid the inflation of results in dogs prone to vocalisation. Counts for 205 pooled behaviours were then rounded to the nearest whole number to allow for a negative 206 binomial model to be fitted. This step was relevant only for Aroused_S behaviours, as 207 behaviours in other categories did not need to be scaled.

208
An ordinal linear regression model was used to determine factors influencing performance.

209
Generalised linear models with a quasi-poisson distribution due to over-dispersion in count 210 data were used from the lme4 package using the glm function in RStudio to determine factors 211 that have a significant effect on Aroused_S behaviours and Mean ET Before races and Mean 212 ET After races. The final models were built using the stepwise method and the AIC number 213 to determine the model of best fit.

214
Pearson's Correlation tests using the cor function were performed on factors that were not 215 included in models or for which models were difficult to resolve.

218
Track configuration in terms of where the kennel block, stir-up yard and catching pen were 219 located in relation to the track differed between tracks, as summarised in Table 3   Box 1 showed the strongest association with good performance while, in comparison, Boxes 240 4, 5 and 7 had a significantly negative effect on performance (see Table 4 for figures).  Table 3.  and Aroused_S, and a summary of the model is shown in Table 6  showing the relationship between ambient temperature and Mean Eye Temperature Before 280 the race is shown in Figure 8. Mean Eye Temperature After the race was, as expected, 281 influenced by ambient temperature, and the relationship is shown in Figure 9.    There was a significant, negative correlation between Teaser-related behaviour and Mean ET pattern existed in the current study, but was not statistically significant.

371
. Sampling in the kennels prior to racing would likely provide a better comparison of before 372 and after race as the act of taking the dogs out of the kennels may elevate their aorusal. There was a statistically significant, positive effect of race number on mean eye temperatures 416 before the races, suggesting that greyhounds at the race-meet grew increasingly aroused as 417 the race-meet progressed. All greyhounds racing must be kennelled 30 minutes before the 418 first race. They are undisturbed in the time between kennelling closing and the first race, but 419 once races are underway, a steady stream of trainers enter the kennels to collect dogs and 420 return dogs that have just raced. The kennelled dogs are therefore exposed to ongoing Wentworth Park at least [32]. Box 1 appears to offer an advantage, as noted by The

460
Greyhound Recorder [33] and ,in the current study, Boxes 4, 5 and 7 conferred a significant 461 disadvantage when compared to Box 1. Greyhounds that prefer to run close to the rail are 462 likely to perform better regardless of starting box because they must cover less ground over 463 the course of the race than greyhounds that prefer to run on the outside of the pack. As such, 25 464 greyhounds that prefer to run close to the rail and that also start from Box 1, 2 or 3 are likely 465 to cover less ground than greyhounds with this preference that start from boxes farther from 466 the rail. This issue may be best addressed by adopting track safety recommendations for a 467 lure system that places the lure closer to the centre of the track [34].

469
Male greyhounds in the current study were significantly more likely to place favourably than 470 females. However, this was complicated by an interaction between sex and days since last 471 raced. Whereas females showed no clear pattern in their performance regardless of how long 472 it had been since they were last raced, males were more likely to place poorly the longer it 473 had been since they last raced. There was no significant difference between males and 474 females in latency since the previous race. The effect of this interaction on performance is 475 intriguing, but small and difficult to interpret. The significant negative correlation between 476 mean eye temperatures before the races and latency since the previous race is at odds with a 477 more intense stress response to the race-meet environment after longer rest periods as a 478 potential explanation. It is possible that increasing latency since the previous race 479 compromises race fitness, as extant data on racehorses shows an increase in the likelihood of 480 sustaining a serious injury during a race with increasing days since last racing [35]. It is also 481 possible that anticipation of racing is diminished by the lack of recent associations with track-482 related stimuli if the dog has not raced for more than a week or so, and this compromises 483 performance. This is consistent with the current finding of a negative correlation with mean 484 eye temperatures before the races.

485
Why latency since the previous race should affect male performance more than female 486 performance is unclear and, to the authors' knowledge, has not previously been reported in 487 animal performance studies. As always, it is possible this reflects a statistical anomaly, and 488 that simply increasing the sample size would verify the strength of this relationship.

489
There was no significant difference in mean eye temperatures between sexes before or after 490 the race and, to the authors' knowledge, no sex differences in response to arousal have 491 previously been reported in dogs. That said, there may be differences in how individuals of 492 either sex behave in response to different levels of arousal.

493
Catching pens 494 One of the goals of the current study was to investigate whether greyhounds in NSW races 495 were being sufficiently rewarded for racing despite being unable to access the lure at the 496 conclusion of races. We may assume that if greyhounds are finishing races without any 497 penalty for failure to chase the lure, they are being sufficiently rewarded for racing at the time 498 of observing them race. However, recording one race per greyhound cannot demonstrate that 499 greyhounds are being sufficiently rewarded to continue racing indefinitely. As such, we initiated the direction change rather than following another dog that had changed direction).

508
We found these behaviours in 59.1% of greyhounds. the dogs, or how quickly they enter the catching pen and restrain the dogs, or it may be the 558 tracks are particularly attractive to owners of greyhounds that are more or less prone to 559 distress. As noted in the previous section, this effect may also to some extent be attributable 560 to the presence of teasers in the catching pen. Further investigation into which track attributes 561 may influence greyhound stress before racing is important for the integrity of the sport so that 29 562 vetting, kennelling, and pre-stir-up procedures can be designed to support greyhounds 563 equitably.

564
There were significantly more behaviours indicative of arousal in the stir-up at Wentworth 565 Park compared to Gosford. This may reflect the effects of a track design factor or a 566 operational factor. The track at Wentworth Park has its catching pen adjacent to its stir-up 567 pen and flush against the track itself, and directly in front of the kennels. In contrast, at 568 Gosford, the stir-up pen is a considerable distance from the catch-pen, and even farther from 569 the kennels. Richmond has the greatest distance between the stir-up pen and the track by a 570 small margin, with the catching pen separated, and the kennels behind the stir-up pen.

571
Handlers may also behave differently between tracks and may encourage more behaviours 572 indicative of arousal at some tracks more than at others. This may reflect the handlers 573 awareness of the available prize money or prestige of the races, the importance of a particular 574 day of the week on which the races are held, or it may be entirely sub-conscious.

576
The need to understand behavioural wastage in racing greyhounds is clear. Ambient 577 temperature, race number, age and experience, start box, sex of dog, (frustration and teasers) 578 in catching pens and track effects all affect the performance of greyhounds on tracks.

579
This is the first published study t of racing greyhound behaviour at race-meets and the first, 580 for any racing code, to use IRT at racetracks to assess arousal. The results describe modest 581 relationships between eye temperature and performance, but these effects are significant and 582 may assist in the development of more detailed studies to identify specific factors that 583 compromise performance and establish how they can be modified to reduce their negative 584 effects. IRT before races may be more revealing than IRT after races due to the influence of 585 core body temperature that may reflect the legacy effects of physical effort more than arousal, 30 586 and the influence of ambient temperature. Attempts to use IRT before races bring significant 587 timing and logistical challenges, but this study showed there is promise in eye temperature 588 measurements before races to reveal the effects of experience at racetracks in the long-and 589 short-term on behaviour and performance. For clearer results, it would be worthwhile 590 investigating how IRT images may be collected closer to the races 591 This study also offers insights into how individuals within the racing greyhound population 592 respond differently to the anticipation of racing, and how this might correlate with their 593 preferences for rewards (e.g. teaser or lure) upon conclusion of any given race. Clearly, the 594 use of teasers in the catching pen and the track effects on behaviour and arousal are both 595 areas that merit further research to makes race-meets optimally arousing for racing 596 greyhounds, and to improving reward availability and disengagement from the lure at the end 597 of racse. The large percentage of dogs t showing signs of frustration and continuing to search 598 for the lure when in the catching pen raises concern for both the physical and emotional 599 wellbeing of greyhounds. The catching pen system has been in operation for many years, but 600 may not support all racing greyhounds equitably in avoiding wastage. The period during 601 which greyhounds are kennelled before their race may contribute to their stress at race-meets, 602 so any means by which kennelling time can be reduced, especially for greyhounds in races 603 late in the days' program, are worth further investigation and suitable trials.