Equipment and training risk factors associated with ridden behaviour problems in UK leisure horses

Equipment and training risk factors associated with ridden behaviour problems in UK leisure horses

Applied Animal Behaviour Science 137 (2012) 36–42 Contents lists available at SciVerse ScienceDirect Applied Animal Behaviour Science journal homepa...

242KB Sizes 1 Downloads 27 Views

Applied Animal Behaviour Science 137 (2012) 36–42

Contents lists available at SciVerse ScienceDirect

Applied Animal Behaviour Science journal homepage: www.elsevier.com/locate/applanim

Equipment and training risk factors associated with ridden behaviour problems in UK leisure horses Jo Hockenhull a,b,∗ , Emma Creighton a,c a b c

Department of Biological Sciences, University of Chester, Chester CH1 4BJ, UK School of Veterinary Sciences, University of Bristol, Langford House, Langford, Bristol BS40 5DU, UK School of Agriculture, Food and Rural Development, Newcastle University, Newcastle upon Tyne NE1 7RU, UK

a r t i c l e

i n f o

Article history: Accepted 15 January 2012 Available online 11 February 2012 Keywords: Horse Problem behaviour Equitation Risk factors

a b s t r a c t Ridden behaviour problems are prevalent in the UK leisure horse population and may have implications for horse welfare and rider safety. This study aimed to identify risk factors associated with ridden behaviour problems in UK leisure horses from the training approaches and equipment used with them. An Internet survey was used to collect data on 1326 horses from a convenience sample of leisure horse owners. The survey asked owners to report the frequency their horses displayed fifteen ridden behaviour problems over the previous week. Data on the frequency of occurrence of behaviour in four components of related ridden behaviour problems were explored for association with details of the horse’s working life, including the type of tack, equipment and training used, and the frequency the professional services of saddlers and farriers were employed using logistic regression analyses. Behaviour data were generated for 791 individual horses. Risk factors associated with the ridden behaviour problems emerged as three themes. One related to the design and fit of the saddle, with dressage and working hunter saddles associated with a reduced risk of ridden behaviour problems compared to general purpose saddles. The horse’s footcare and shoeing regime was associated with three of the four groups of behaviour problems. An extended interval (seven weeks or more) between farrier visits was associated with an increased risk of discomfort behaviour. Taking an outcome-centred approach to training, for example through the use of artificial training aids, was associated with an increased risk of behaviour problems while spending more time with the horse outside of training situations, a more horse-centred approach, was associated with a reduced risk of problems. Further research is required to understand the causal relationships behind these associations, with the aim of improving the welfare of the horse and the well-being and safety of its rider. © 2012 Elsevier B.V. All rights reserved.

1. Introduction The aetiology of ridden behaviour problems is likely to be multifactorial (Casey, 1999). Inappropriate behaviour under saddle may be a learned behaviour that has been

∗ Corresponding author. Present address: School of Veterinary Sciences, University of Bristol, Langford House, Langford, Bristol BS40 5DU, UK. Tel.: +44 01173 319309. E-mail addresses: [email protected] (J. Hockenhull), [email protected] (E. Creighton). 0168-1591/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.applanim.2012.01.007

unintentionally taught to the horse by its rider, or may develop in response to pain, conflict, frustration or hyperreactivity (McLean, 2003; McGreevy, 2004; Zeitler-Feicht, 2004; McGreevy and McLean, 2010). Yet, while the origins of ridden behaviour problems are not necessarily clearcut, the welfare implications of their occurrence mean that they represent a significant cause for concern, be it due to the pain or conflict underlying their performance, or the negative consequences of actions taken to resolve them. It has been suggested that pain may underlie many equine behavioural problems (Casey, 1999). Common

J. Hockenhull, E. Creighton / Applied Animal Behaviour Science 137 (2012) 36–42

sources of pain in ridden horses are mouth pain due to poor dental maintenance (Dixon et al., 2004) or bit induced trauma (Cook, 1999); back pain arising from badly fitted saddles (De Cocq et al., 2004, 2006); and pain from inadequate or inappropriate shoeing (Cook, 2003). The use of artificial aids such as draw reins and running martingales, has also been implicated as a potential source of pain (Rhodin et al., 2005) and thus a welfare concern (McGreevy, 2007; von Borstel et al., 2009). The use of artificial riding aids is widespread in the leisure horse population, with 78% of respondents using one or more on their horses (Hockenhull and Creighton, in press-b). Understanding the risk associated with using certain equipment and training approaches would enable riders to avoid these risks by using alternative approaches when riding their horse. Principal components analysis (PCA) has been used to identify groups of related ridden behaviour problems from fifteen problems included in a survey of UK leisure horses (Hockenhull and Creighton, 2010, in press-a). Overall, 91% of UK leisure horses displayed one or more of the fifteen ridden behaviour problems (Hockenhull and Creighton, 2010), demonstrating the need to improve our understanding of their associated risk factors. This study aimed to identify the equipment and training approaches used with horses that were associated with each the four ridden behaviour problem components. While causal relationships cannot be inferred from survey data, recognising the practices that are potential risk factors for behaviour problems provides data and hypotheses upon which future empirical studies can be founded. 2. Materials and methods An Internet survey was used to collect horse-level data on ridden behaviour problems and the equipment and training approaches used with the horse, from a convenience sample of UK leisure horse carers. The survey was part of a larger study exploring husbandry and welfare (Hockenhull, 2010) and was online for 12 months between 2006 and 2007. The larger study comprised of two further surveys covering the horse’s day-to-day management (stabling, opportunity for social contact and turn-out) and feeding routine (forage, concentrates and supplements). Consequently these factors were not included in the survey reported here. The survey contained sixteen questions regarding the type of work undertaken with the horse, the types of equipment and training methods used on it and the regularity that professional services (farriers, saddlers and dentistry professionals) were employed. Respondents were also asked to rate the frequency that their horse performed fifteen different behaviour problems over the last week it was ridden. The behaviour problems were identified as the most common problems featuring in equestrian magazines and online forums. The individual behaviour problems were not defined in the survey. The rating scale respondents were asked to use ran from 1 to 5 and was anchored only at the endpoints (1 = never, 5 = often). The behaviour items were presented in two matrix formatted questions and these were each followed by an open-ended text box to allow respondents to elaborate on the circumstances behind the

37

behaviour reported should they feel this was necessary. Pilot testing had revealed that respondents were more likely to report their horse’s behaviour if they could rate it rather than commit to a binary present/absent answer and if they could explain the circumstances involved. Consequently the survey was structured in this way to reduce nonresponse. Principal components analysis was used to identify groups of related ridden behaviour problems from the fifteen problems included in the survey (Hockenhull and Creighton, 2010, in press-a). Four components were extracted, their prevalence within the sample and composite problems were: (1) Not slowing/resistance, 84% (662/791): resist slowing down when asked, jog when asked to walk, shy, move off when rider mounts, pull/lean on the bit. (2) Discomfort, 61% (482/790): resist turning when asked, buck, trip over its own feet, canter on the wrong leg, refuse to move forward when asked. (3) Jumping issues, 36% (165/455*): run out when jumping, stop at jumps, rush over jumps. *Only data from horses that had the opportunity to jump over the previous week were included in the analyses for this component. (4) Extreme conflict, 22% (175/789): bolt, rear, buck. 2.1. Statistical analyses Data were filtered for analysis to include only horses that were in ridden work at the time the survey was completed to prevent redundancies in the data affecting the outcome of the models. All statistical analyses were conducted in SPSS version 14 for Windows (SPSS Inc., USA). Associations between the four ridden behaviour problem components (dependent variables) and the equipment and training risk factors (independent variables) were explored using univariate logistic regression analyses to determine which variables would be entered in to multivariate binary logistic regression models (Dohoo et al., 2003) for each of the four behaviour problem components (generating four groups of independent variables). Independent variables associated with a behaviour problem component at P < 0.25 (Hosmer and Lemeshow, 2000) were tested for multicollinearity and rejected where Spearman’s correlation matrices were r ≥ 0.4 or collinearity diagnostics showed tolerance values ≤0.1 and variance inflation factor values ≥10 (Field, 2005; Pallant, 2007). Where covariance was found, summary variables and those more directly relating to the horse’s working life were preferentially retained. The selected independent variables were entered into multivariate logistic regression models for each behaviour problem component using the forced entry method and goodness-of-fit was assessed using the Omnibus Test of Model Coefficients (P ≤ 0.05) and the Hosmer and Lemeshow Test (P ≥ 0.05). The findings are presented as odds ratios (OR (95% CI)) for each independent variable. Independent variables that contribute significantly to the model have P values ≤0.05. The OR is an indicator of the unit change in odds resulting from a unit change in predictor (Dohoo et al., 2003; Field,

38

J. Hockenhull, E. Creighton / Applied Animal Behaviour Science 137 (2012) 36–42

2005; Tabachnick and Fidell, 2007). Odds ratios greater than one reflect the increase in odds (the likelihood) of an outcome (i.e. the horse has a behaviour problem) with a one-unit increase in the predictor; odds ratios less than one reflect the decrease in odds of that outcome with one-unit change (Tabachnick and Fidell, 2007). 3. Results The survey generated data for 1326 individual horses. However, due to a faulty software upgrade, responses to the behaviour matrix questions were not recorded for a three-month period between March and June 2007, resulting in usable behaviour data for 791 individual horses. The sample sizes of the multivariate logistic regression models varied due to item non-response by respondents. Only horses with full datasets for all the variables within the model were included in the analysis. The lower sample size for jumping issues reflects the number of respondents who jumped their horses during the last week the horse was ridden. 3.1. Not slowing/resistance – reportedly displayed by 84% (662/791) of horses sampled Nine equipment and training risk factors were significantly associated with the behaviour component not slowing down/resistance in the univariate logistic regression analyses. A reduced likelihood of these problems was associated with using a hackamore or bitless bridle (OR 0.370 (0.187–0.729); P = 0.004) compared to using a snaffle bit, as was using a saddle falling into the category other (OR 0.563 (0.321–0.988); P = 0.045) rather than a general purpose saddle. An increased likelihood of not slowing down/resistance problems was associated with using traditional shoes on all feet (OR 2.586 (1.676–3.989); P < 0.001) and using other shoes (not barefoot or traditional shoes) (OR 2.133 (1.066–4.271); P = 0.032) compared with unshod or barefoot horses. An increased likelihood of problems was associated with using 1 (OR 2.235 (1.403–3.560); P = 0.001), 2 (OR 3.026 (1.780–5.146); P < 0.001), or 3+ (OR 3.624 (1.883–6.793); P < 0.001) artificial riding aids of whip, spurs, running martingale, standing martingale, running/draw reins, drop/flash nosebands; as was using 1 or both of twitch or pressure halter as artificial aids for controlling the horse from the ground (OR 2.112 (1.174–3.797); P = 0.013) compared to using none of these riding or ground control aids. An increased likelihood of problems was associated with using traditional training techniques often compared to not often using Natural Horsemanship techniques, Parelli, Clicker training or TTouch (OR 1.801 (1.164–2.784); P = 0.008). When the independent variables identified as potential risk factors were combined in the multivariate model, five significant associations between the occurrence of not slowing down/resistance behaviour and equipment and training risk factors were identified (Table 1). A reduced likelihood of not slowing/resistance problems were associated with using a working hunter saddle, while an increased likelihood of problems was associated

Table 1 Significant equipment and training risk factors for the performance of not slowing/resistance problems identified using a forced entry multivariate logistic regression model (N = 686). Equipment and training variable

Odds ratio

Type of saddle used 1 General purpose (Ref) 0.336 Working hunter Type of shoe used 1 Barefoot/no shoes (Ref) 1.830 Traditional shoes all round Number of artificial riding aids used 1 Not used (Ref) 1.906 1 item 2.246 2 items 2.333 3+ items

95%CI

– 0.137–0.823 –

P value

– 0.017 –

1.037–3.229

0.037

– 1.083–3.355 1.173–4.298 1.077–5.053

– 0.025 0.015 0.032

Ref = Reference category; 95% CI = 95% confidence interval. The model classified 84.8% of cases correctly.

with using traditional shoes on all feet and using artificial riding aids. 3.2. Discomfort – reportedly displayed by 61% (482/790) of horses sampled Fifteen equipment and training risk factors were significantly associated with discomfort behaviour in the univariate analysis. There was a reduced likelihood of these problems associated with using a dressage (OR 0.563 (0.383–0.828); P = 0.003) or other saddle (OR 0.492 (0.308–0.787); P = 0.003) compared to using a GP saddle. There was also a reduced likelihood of discomfort behaviour if the respondent used a treeless or adjustable saddle (OR 0.452 (0.239–0.854); P = 0.014), checked the saddle fit themselves (OR 0.268 (0.108–0.665); P = 0.005), or got the saddle fit checked when there was a problem or the horse’s shape changed (OR 0.462 (0.229–0.928); P = 0.030) compared to respondents who never or rarely had the fit of their horse’s saddle checked professionally. Other factors associated with a reduced likelihood of discomfort problems were using a hackamore or bitless bridle (OR 0.478 (0.253–0.904); P = 0.023) rather than a snaffle bit and spending more time in close contact with the horse outside of work or training situations over the previous week (OR 0.965 (0.935–0.995); P = 0.024). An increased likelihood of problems was associated with using 1 (OR 1.500 (1.027–2.190); P = 0.036), 2 (OR 1.863 (1.245–2.788); P = 0.0021), or 3+ (OR 2.694 (1.671–4.344); P < 0.001) artificial riding aids; as was using 1 or both artificial aids for controlling the horse from the ground (OR 2.213 (1.477–3.315); P < 0.001) compared to using none of these control aids. Using traditional shoes on all feet was associated with increased likelihood of problems compared to leaving the horse unshod or barefoot (OR 1.623 (1.147–2.299); P = 0.006). There was also an increased likelihood of problems if the horse was last seen by a farrier 0–3 weeks ago (OR 3.129 (1.560–6.275); P = 0.001), 4–6 weeks ago (OR 3.162 (1.540–6.494); P = 0.002) or 7+ weeks ago (OR 4.396 (1.754–11.013); P = 0.002) compared with those

J. Hockenhull, E. Creighton / Applied Animal Behaviour Science 137 (2012) 36–42

39

Table 2 Significant equipment and training risk factors for the performance of behavioural expressions of physical discomfort identified using a forced entry multivariate logistic regression model (N = 790). Equipment and training variable Type of saddle used General purpose (Ref) Dressage Frequency saddle fit is professionally checked Never/only when new (Ref) Check it myself/been on a course Time since feet last professionally checked Do not use farrier/manage feet myself (Ref) 7+ weeks ago Number of artificial riding aids used Not used (Ref) 3+ items Ground control equipment used Not used (Ref) Used (1 or both of twitch and pressure halter) Number of hours spent in close contact outside work per week (continuous variable)

Odds ratio

95%CI

P value

1 0.650

– 0.424–0.997

– 0.048

1 0.360

– 0.136–0.954

– 0.040

1 4.634

– 1.502–14.301

– 0.008

1 2.439

– 1.369–4.345

– 0.002

1 2.111 0.960

– 1.352–3.296 0.926–0.995

– 0.001 0.026

Ref = Reference category; 95% CI = 95% confidence interval. The model classified 66.4% of cases correctly.

horses that were never seen by a farrier/whose owner managed their feet themselves. When these independent variables identified as potential risk factors were combined in the multivariate logistic regression model, six equipment and training risk factors significantly associated with the occurrence of behaviour problems potentially indicative of physical discomfort were identified (Table 2). Increased likelihood of discomfort problems was associated with the horse last being seen by a farrier 7+ weeks ago, where 3+ artificial riding aids were used, and where equipment to control the horse from the ground was used. A reduced likelihood of these problems was associated with the horse being regularly ridden in a dressage saddle and the respondent checking the fit of the saddle themselves. 3.3. Jumping issues – reportedly displayed by 36% (165/455) of horses sampled that had been jumped in the last week they were ridden The univariate analysis identified ten equipment and training risk factors significantly associated with jumping

issues. An increased likelihood of these problems was associated with the more time the horse spent in work over the previous week (OR 1.040 (1.003–1.078); P = 0.035); where traditional training was used often (OR 1.806 (1.116–2.924); P = 0.016) compared to often using none of the training techniques listed; and where 2 (OR 2.364 (1.299–4.301); P = 0.005) or 3+ (OR 2.876 (1.546–5.349); P = 0.001) artificial riding aids were used compared to not using any of these aids. An increased likelihood of jumping issues was also associated with using traditional shoes on all feet (OR 2.988 (1.630–5.478); P < 0.001), using traditional shoes on the front feet and no shoes behind (OR 3.400 (1.421–8.133); P = 0.006), and using other shoe types (OR 3.264 (1.431–7.444); P = 0.005) compared to keeping the horse unshod or barefoot. Reduced likelihood of jumping issues was associated with using a dressage (OR 0.431 (0.227–0.818); P = 0.010) or other (OR 0.398 (0.167–0.948); P = 0.037) saddle regularly when ridden rather than a GP saddle, and where the fit of the saddle was professionally checked if there was a problem or the horse’s shape changed (OR 0.332 (0.125–0.883); P = 0.027) compared to those horses whose saddles were rarely or never professionally checked.

Table 3 Significant equipment and training risk factors for the occurrence of jumping issues identified using a forced entry multivariate logistic regression model (N = 410). Equipment and training variable Type of saddle used General purpose (Ref) Dressage Frequency saddle fit is professionally checked Never/only when new (Ref) Only when problem/horse’s shape changes Type of shoeing used Barefoot/no shoes (Ref) Traditional front feet + barefoot back feet Type of training used often No training technique used often (Ref) Traditional only Number of hours spent in close contact outside work per week (Continuous variable)

Odds ratio

95%CI

P value

1 0.409

– 0.203–0.824

– 0.012

1 0.260

– 0.085–0.801

– 0.019

1 4.106

– 1.389–12.136

– 0.011

1 1.838 0.943

– 1.050–3.220 0.895–0.994

– 0.033 0.029

Ref = Reference category; 95% CI = 95% confidence interval. The model classified 68.8% of cases correctly.

40

J. Hockenhull, E. Creighton / Applied Animal Behaviour Science 137 (2012) 36–42

When these independent variables identified as potential risk factors were combined in the multivariate model, five risk factors significantly associated with jumping issues were identified (Table 3). An increased likelihood of jumping issues was associated with the horse often being trained using traditional techniques and being shod using traditional shoes on its front feet and no shoes behind. A decreased likelihood of these problems was associated with increased time the horse spent in close contact with people outside of work situations, where the horse was regularly ridden in a dressage saddle, and where the fit of the saddle was professionally checked when there was a problem or the horse’s shape changed. 3.4. Extreme conflict – reportedly displayed by 22% 175/789) of horses sampled The univariate analysis identified ten equipment and training risk factors significantly associated with extreme conflict behaviour. A reduced likelihood of these problems (OR 0.391 (0.169–0.908); P = 0.029) was associated with using a treeless or adjustable saddle compared to respondents that never or rarely got the fit of their horse’s saddle professionally checked. An increased likelihood of extreme conflict behaviour was associated with having three rather than one regular rider (OR 2.501 (1.228–5.094); P = 0.012), shoeing the horse traditionally (OR 1.606 (1.033–2.496); P = 0.035) rather than keeping it unshod or barefoot and using 1 (OR 2.334 (1.362–4.000); P = 0.002), 2 (OR 2.719 (1.568–4.716); P < 0.001), or 3+ (OR 3.670 (2.049–6.573); P < 0.001) artificial riding aids and using artificial aids to control the horse from the ground (OR 1.616 (1.078–2.422); P = 0.020). An increased likelihood of extreme conflict behaviour was also associated with often using traditional training (OR 2.091 (1.348–3.244); P = 0.001) or a combination of traditional and Natural Horsemanship techniques (OR 2.120 (1.012–4.442); P = 0.046) compared with not often using any of the training techniques listed in the survey. When these independent variables identified as potential risk factors were combined in the multivariate model six equipment and training risk factors were significantly associated with extreme conflict behaviour (Table 4). An increased likelihood of extreme conflict behaviour was associated with the horse having three regular riders, where artificial riding aids were used and where it was often trained using traditional techniques or a combination of traditional and Natural Horsemanship techniques. 4. Discussion The analyses identified training and equipment risk factors for each of the four ridden behaviour problem components. While it should be remembered that the data generated by the survey relied on owner self-report, every effort was made during the design and the administration of the survey to minimise recall bias and to promote accurate reporting. Although patterns of risk factors were identified for each of the behaviour problem components, three themes common to all four components emerged and

Table 4 Significant equipment and training risk factors for the occurrence of extreme conflict behaviour identified using a forced entry multivariate logistic regression model (N = 687). Equipment and training variable

Odds ratio

Number of riders 1 1 person (Ref) 2.645 3 people Number of artificial riding aids used 1 Not used (Ref) 1.925 1 item 1.922 2 items 2.267 3+ items Type of training used often 1 No training technique used often (Ref) 2.152 Traditional only Tradi1.303 tional + Natural Horsemanship

95%CI

P value

– 1.197–5.845

– 0.016

– 1.029–3.600 1.006–3.672 1.113–4.618

– 0.040 0.048 0.024



1.273–3.639 0.999–5.583

0.004 0.050

Ref = Reference category; 95% CI = 95% confidence interval. The model classified 77.6% of cases correctly.

the discussion is structured around these themes. Each of the themes identified raise further questions as to causal relationships that can be investigated empirically by future research studies.

4.1. Saddle design and fit Using a dressage saddle was associated with a reduced risk of discomfort behaviours and jumping issues, and using a working hunter saddle was associated with a reduced risk of slowing/resistance problems. One interpretation of this association is that saddle designs that allow greater freedom of movement through the shoulder than general purpose saddles (Hartley Edwards, 1990) may decrease the likelihood of ridden behaviour problems. However, the choice of saddle may equally reflect the type of ridden work the horse undertakes and potentially the skill level of the rider. It may be that both explanations have some degree of truth. Empirical studies are required to establish the causal relationship(s) underlying this association. A reduced likelihood of discomfort problems was also associated with respondents who reported checking the fit of the saddle themselves compared to those only having the fit checked when the saddle was new or not having it professionally checked at all. A similar association was identified in relation to jumping issues where the risk of problems decreased if the respondent had the saddle checked when there was a problem or the horse changed shape. Although industry guidelines recommend regular professional saddler checks (Auty, 1998; DEFRA, 2008), the findings suggest that owners who take responsibility for their saddle fit may have greater vigilance and awareness of how the saddle is affecting their horse and the condition of its back, leading to earlier detection of problems.

J. Hockenhull, E. Creighton / Applied Animal Behaviour Science 137 (2012) 36–42

4.2. Footcare and shoeing The type of shoeing used was associated with an increased likelihood of three of the four behaviour problem components. Pain arising from the feet has been implicated as a cause of jumping-related behaviour problems in the BHS Equitation Manual (Auty, 2001) and related publications (Dillon and Revington, 2000), suggesting that footcare/shoeing is a recognised risk factor in the equestrian industry. However, there is little published research on the impact of shoeing and barefoot practices on horse performance (Creighton and Jones, 2008) and empirical studies are needed to further explore this relationship. Horses that had last seen the farrier seven or more weeks prior to the respondent completing the survey were associated with a four-fold increase in the likelihood of expressing physical discomfort behaviour. This is consistent with industry recommendations of four-six week intervals between farrier visits to keep feet healthy (Auty, 1998; DEFRA, 2008). Similar to the reduced risk of problems seen when owners check the saddle fit themselves, taking responsibility for the horse’s footcare may increase owner awareness of potential problems and increase their vigilance, allowing them to be proactive when issues arise (La Pierre, 2004). 4.3. Rider approach – outcome vs. horse-centred The third risk factor theme to emerge concerns the overall approach of the rider. The use of artificial riding aids was associated with an increased likelihood of problems on all four components. Using equipment to control the horse from the ground was also associated with an increased risk of discomfort issues. The use of potentially aversive artificial riding and ground aids to control the horse suggests an outcome-centred approach to behavioural or training issues, where achieving a goal takes precedence over the means used to achieve it. However, it is unclear from this research whether these devices were being used to regain control over a horse exhibiting problematic behaviour or whether their use led to the development of the behaviour. The use of traditional training techniques as the only often used training method was associated with an increased risk of jumping issues. Traditional training is one of the only training techniques that instructs riders in how to jump their horse, however, the technique is often highly focused on outcome. For example, the BHS Manual of Equitation (Auty, 2001) gives advice on resolving jumpingrelated problems where the horse is presumed to be acting out of low confidence, bad riding, mischief or pain. Yet no information is offered to help riders distinguish between these or address the underlying causes, and only details of the actions needed to improve the outcome of the jump are described. Often using traditional training techniques and often using a combination of traditional and Natural Horsemanship (NH) techniques were both associated with an increased risk of extreme conflict behaviour. It has been suggested that these problems may arise through riders missing early warning signs that the horse was experiencing difficulties or even responding to them by employing

41

artificial aids to restrict their performance (Widdicombe, 2008). This may reflect the outcome-focused approach of traditional techniques as suggested above. However, the nature of the relationship between training approach and extreme conflict behaviour is speculative whilst so little research on the impact of different horse training approaches exists. That the more horse-centred NH techniques are associated with these behaviours, even if only when used in combination with traditional techniques, raises questions. NH techniques emphasise the importance of communicating with the horse and understanding its behaviour (Roberts, 1997; Birke, 2007) suggesting that those using them would be aware of the horse’s difficulties before the behavioural warnings escalate. Using NH techniques alone was not identified as risk factor, implying that the risk may lie in using a combination of techniques, possibly due to the inclusion of traditional methods, as the combination of traditional and NH techniques was associated with a smaller increase in risk than when traditional techniques alone were used. One hypothesis present in the limited research literature on comparative horse training methods centres upon inconsistency within training as a source of stress and conflict for the horse (Hall et al., 2008), and riders switching between these two very different approaches (Birke, 2007) may somehow contribute to these behavioural problems. An alternative hypothesis is that riders experiencing problems with their horse are more likely to try a combination of approaches, using NH techniques alongside traditional techniques, in their training. Again, further work is needed to explore the causal relationship underlying this association. The increased risk of extreme conflict behaviour when the horse was ridden regularly by three different riders may also reflect the detrimental effects of inconsistency; this time arising when the horse is trained by multiple riders rather than multiple approaches. Alternatively, increasing the number of riders may be in response to the performance of these problems rather than their cause, if the regular rider can no longer cope with the horse’s behaviour. It is possible that the quality of the primary horse–rider relationship suffers if the horse is ridden by a number of different riders. Yet, despite the importance of the horse–rider relationship in both competition and recreational riding, there is little scientific literature on the various facets of this association and further research is needed to explore the repercussions of any communication breakdown (Hausberger et al., 2008). The relationship between the horse and rider is an important aspect of a horse-centred training approach (Waran, 2005), and spending more time in contact with the horse outside of work or training situations was here associated with a reduction in risk of discomfort behaviour and jumping issues. The increased understanding of an individual horse’s behaviour and reactions that can develop from spending time with the horse may benefit the detection of pain-related behaviour in an animal adept at concealing behavioural indicators of pain (Ashley et al., 2005) and lead to the rider taking actions to alleviate it. It may also help develop the trust needed for a successful jumping partnership (Widdicombe, 2008).

42

J. Hockenhull, E. Creighton / Applied Animal Behaviour Science 137 (2012) 36–42

The risk factors discussed above in association with ridden behaviour problems are unlikely to be the only factors associated with these problems, the horse’s daily management and feeding routine are also likely to be influential. Rider-based parameters, such as skill level and the consistency, strength and accuracy of the application of riding aids, are also highly relevant. Such factors were not included in the survey due to the difficulties in generating data requiring accurate rider self-assessment using an online survey, when a rider is often unaware of how they are interacting with their horse when riding. Rider-based parameters would best be explored using observational studies. Empirical studies would have more scope to explore potential risk factors from rider-based parameters and to improve our understanding of the causal relationships underlying the associations identified in this study. 5. Conclusion Here we identified three themes within the risk factors associated with ridden behaviour problems. While causal relationships cannot be inferred from survey data, the study findings highlight the lack of empirical research in this area and suggest factors that would benefit from future experimental research exploring causality and the aetiology of ridden behaviour problems. Acknowledgements Jo Hockenhull received financial support from Bransby Home of Rest for Horses, the University of Chester and Funds for Women Graduates. The authors would like to thank Duncan Brown at the University of Liverpool for computer support and all survey respondents who volunteered their time to report in detail on the behaviour and management of their horse. References Ashley, F.H., Waterman-Pearson, A.E., Whay, H.R., 2005. Behavioural assessment of pain in horses and donkeys: application to clinical practice and future studies. Equine Vet. J. 37, 565–575. Auty, I., 1998. The British Horse Society Complete Manual of Stable Management. Kenilworth Press, Addington. Auty, I., 2001. The British Horse Society Manual of Equitation. Kenilworth Press, Shrewsbury. Birke, L., 2007. Learning to speak horse”: the culture of “natural horsemanship”. Soc. Anim. 15, 217–239. Casey, R.A., 1999. Recognising the importance of pain in the diagnosis of equine behaviour problems. In: Harris, P.A., Gomarsall, G.M., Davidson, H.P.B., Green, R.E. (Eds.), Proceedings of the BEVA Specialist Days on Behaviour and Nutrition. , pp. 25–28. Cook, W.R., 1999. Pathophysiology of bit control in the horse. J. Equine Vet. Sci. 19, 196–204. Cook, W.R., 2003. Professional dismissiveness of equine barefootedness. J. Equine Vet. Sci. 23, 564–566. Creighton, E., Jones, K., 2008. Horse owners’ experiences of (horses) going barefoot. In: Murphy, J., Hennessy, K., Wall, P., Hanly, P. (Eds.), International Society for Equitation Science 4th International Conference. Dublin, p. 39.

De Cocq, P., van Weeren, P.R., Back, W., 2004. Effects of girth, saddle and weight on movements of the horse. Equine Vet. J. 36, 758–763. De Cocq, P., van Weeren, P.R., Back, W., 2006. Saddle pressure measuring: validity, reliability and power to discriminate between saddle fits. Vet. J. 172, 265–273. DEFRA, 2008. In: DEFRA (Ed.), Consultation on Code of Practice for the Welfare of Equines. DEFRA, London. Dixon, P.M., R, A., Brannon, H., Burgess, R., Gibson, A., Little, J.C., Orange, B., Rudolph, T., Shaw, D.J., 2004. Survey of the provision of prephylactic dental care for horses in Great Britain and Ireland between 1999 and 2002. Vet. Rec. 155, 692–698. Dillon, E., Revington, H., 2000. Show Jumping for Fun or Glory. Kenilworth Press, Addington. Dohoo, I.R., Martin, W., Stryhn, H., 2003. Veterinary Epidemiologic Research. AVC Inc., Prince Edward Island. Field, A., 2005. Discovering Statistics using SPSS, 2nd ed. Sage Publications Ltd., London. Hall, C., Goodwin, D., Heleski, C., Randle, H., Waran, N., 2008. Is there evidence of learned helplessness in horses? J. Appl. Anim. Welf. Sci. 11, 249–266. Hausberger, M., Roche, H., Henry, S., Visser, E.K., 2008. A review of the human–horse relationship. Appl. Anim. Behav. Sci. 109, 1–24. Hockenhull, J., 2010. The Epidemiology of Behaviour Problems and Risk Factors for Poor Welfare in UK Leisure Horses. Biological Sciences, University of Liverpool. Hockenhull, J., Creighton, E., 2010. Equipment and training risk factors associated with ridden behaviour problems in UK leisure horses. J. Vet. Behav.: Clin. Appl. Res. 5, 52. Hockenhull, J., Creighton, E. The strengths of statistical techniques in identifying patterns underlying apparently random behavioral problems in horses. J. Vet. Behav.: Clin. Appl. Res., in press. Hockenhull, J., Creighton, E. The use of equipment and training practices and the prevalence of owner-reported ridden behaviour problems in UK leisure horses, Equine Vet. J., in press. Hartley Edwards, E., 1990. The Saddle in Theory and Practice. J.A. Allen, London. Hosmer, D.W., Lemeshow, S., 2000. Applied Logistic Regression, 2nd ed. John Wiley & Sons, Inc., New York. La Pierre, K.C., 2004. The Chosen Road: Achieving High Performance through Applied Equine Podiatry. The Naked Greyhound Press, Ocala. McGreevy, P., McLean, A., 2010. Equitation Science. Wiley-Blackwell, Chichester. McGreevy, P.D., 2004. Miscellaneous unwelcome behaviors, their causes and resolution. In: McGreevy, P.D. (Ed.), Equine Behavior: A Guide for Veterinarians and Equine Scientists. Saunders, Edinburgh, pp. 331–345. McGreevy, P.D., 2007. The advent of equitation science. Vet. J. 174, 492–500. McLean, A., 2003. The Truth about Horses. David and Charles, Newton Abbot. Pallant, J., 2007. SPSS Survival Manual, 3rd edn. McGraw-Hill Education, Maidenhead: Open University Press. Rhodin, M., Johnston, C., Roethlisberger Holm, K., Wennerstrand, J., Drevemo, S., 2005. The influence of head and neck position on kinematics of the back in riding horses at the walk and trot. Equine Vet. J. 37, 7–11. Roberts, M.C., 1997. The Man Who Listens to Horses. Arrow Books Ltd., London. Tabachnick, B.G., Fidell, L.S., 2007. Using Multivariate Statistics. Pearson Education Inc., Boston. von Borstel, U.U., Duncan, I.J.H., Shoveller, A.K., Merkies, K., Keeling, L.J., Millman, S.T., 2009. Impact of riding in a coercively obtained Rollkur posture on welfare and fear of performance horses. Appl. Anim. Behav. Sci. 116, 228–236. Waran, N., 2005. Equestrianism and horse welfare: the need for an ‘equine-centred’ approach to training. In: McGreevy, P., McLean, A., Warren-Smith, A., Goodwin, D., Waran, N. (Eds.), Proceedings of the 1st International Equitation Science Symposium. Melbourne, pp. 67–74. Widdicombe, S., 2008. The British Horse Society book of the Natural Horse. David and Charles Ltd., Cincinnati. Zeitler-Feicht, M.H., 2004. Horse Behaviour Explained. Manson Publishing, London.