Maximum and minimum peaks in rein tension within canter strides

Journal of Veterinary Behavior 13 (2016) 63e71

Contents lists available at ScienceDirect

Journal of Veterinary Behavior journal homepage: www.journalvetbehavior.com

Research

Maximum and minimum peaks in rein tension within canter strides Agneta Egenvall a, *, Lars Roepstorff b, Marie Rhodin a, Marie Eisersiö a, Hilary M. Clayton c a

Department of Clinical Sciences, Faculty of Veterinary Medicine and Animal Husbandry, Swedish University of Agricultural Sciences, Uppsala, Sweden Department of Anatomy, Physiology and Biochemistry, Unit of Equine Studies, Faculty of Veterinary Medicine and Animal Husbandry, Swedish University of Agricultural Sciences, Uppsala, Sweden c Department of Large Animal Clinical Sciences, College of Veterinary Medicine, Michigan State University, East Lansing, Michigan b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 6 January 2016 Received in revised form 15 March 2016 Accepted 25 March 2016 Available online 6 April 2016

Equestrians use reins to communicate with the horse. The aim of this study was to describe the amplitudes of rein tension oscillations at canter. Rein tension data were collected from 8 professional riders each riding 2-3 horses during a normal dressage training session using rein tension meters (128 Hz, logged by an inertial measurement unit sensor). Data were stride-split at the maximal positive vertical poll acceleration. Strides were categorized by canter lead, rider position (sitting/2 point), corners, circles, lateral movements, and stride length (collected/working/lengthened). Changes in head angle were determined from gyroscopic sensor data. Dependent data extracted from each stride and rein were maximal tension (MAX), minimal tension (MIN), and the absolute difference between them (CHANGE). Square-root transformed data were analyzed using mixed models with stride categorizations as fixed effects, and rider and horse included as random effects. Findings for rein tension were considered borderline if 0.05

0.001, but significant if P < 0.001. For the rider’s position, the magnitudes were higher in sitting canter than 2-point seat (P < 0.0001), except for inside rein MIN value (n ¼ 21,548 strides). For MAX (both reins), MIN (inside), and CHANGE (outside), the right circle had lower values than the left circle or no circle. For the outside rein, MAX and MIN values showed borderline differences with higher values for lengthened strides than working canter (P ¼ 0.03/0.0014). Inside rein values in right half pass were significantly or borderline higher than left half pass or baseline, and for MIN values, this was found for both inside/outside reins. Both group effects and all pairwise comparisons evaluated were significant for MAX and CHANGE, except the comparison between inside and outside rein in right canter. MAX/MIN tensions were higher if the nose was moving caudally relative to poll at the MAX/MIN event, respectively. Young horses had the largest MAX and CHANGE values, whereas advanced horses had the highest MIN values. The horse contributed 7%, 27%, and 29% of the variation to MIN, MAX, and CHANGE models, respectively. The rider contributed 19% of the variation to the MIN value models but 0% to the MAX and CHANGE models, suggesting that the horse or the dyad (not statistically separable) is responsible for the basic rein tension pattern at canter. Overall results indicate that asymmetry, of riders and/or horses, plays a role in rein tension. Ó 2016 Published by Elsevier Inc.

Keywords: inertial measurement unit rein tension canter signal analysis horse

Introduction There are many styles of horseback riding that use different equipment and various criteria for assessing performance. One of the features that is common to most riding styles is the use of reins as a means of communicating signals from the rider to the horse. * Address for reprint requests and correspondence: Agneta Egenvall, Department of Clinical Sciences, Swedish University of Agricultural Sciences, P.O. Box 7054, SE750 07 Uppsala, Sweden. Tel: þ46 70 3799544. E-mail address: [email protected] (A. Egenvall). http://dx.doi.org/10.1016/j.jveb.2016.03.007 1558-7878/Ó 2016 Published by Elsevier Inc.

The practice of inserting a bit into the oral diastema has been used as an integral part of a rein-based communication since around 3,500 BC (Anthony and Brown, 1991). Over the intervening years, horseback riding has evolved from the utilitarian needs of transportation and warfare into sophisticated sports in which the bit and the reins offer a means of subtle communication between rider and horse. The effect of rein tension is relevant in relation to equine welfare because it is assumed that excessive tension may be deleterious to the horse’s welfare (Ödberg and Bouissou, 1999). Therefore, scientists are interested in measuring rein tension both as a variable that influences success in equestrian sporting performance

64

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

and in relation to the development of oral pathologies (Ödberg and Bouissou, 1999; Tell et al., 2008). Studies have shown that rein tension varies between riding sessions, riders, horses, reins and other equipment, gaits, exercises, and also between and within-strides (von Borstel and Glißman, 2014; Clayton et al., 2003; Clayton et al., 2005, 2011; Egenvall et al., 2012; Egenvall et al., 2015; Egenvall et al., 2016; Eisersiö et al., 2013; 2015; Heleski et al., 2009; Kuhnke et al., 2010; Warren-Smith et al., 2007). When horses are ridden on the bit, which implies that a contact is maintained between the rider’s hand and the bit, rein tension fluctuates considerably over the stride cycle (Clayton et al., 2003; 2011; Egenvall et al., 2015; Egenvall et al., 2016; Eisersiö et al., 2013) and shows a gaitspecific pattern of oscillations (Clayton et al., 2003). These oscillations are thought to reflect the cyclic movements of the horse’s head and neck relative to the trunk that are characteristic of each gait (Clayton et al., 2011). For example, in trot, the head and neck undergo 2 motion cycles per stride in which they descend under the influences of gravity and inertia to reach their lowest point just after the middle of the diagonal stance phases and are then raised to reach their highest point during the suspension phases (Hobbs et al., 2014). Lowering of the neck coincides with an increase in rein tension (Clayton et al., 2011). Rein tension during the stride cycle has mainly been studied using average values calculated over multiple strides and using variables such as minimal, maximal, and mean tension to represent the cyclic nature of the data (e.g., von Borstel and Glißman, 2014; Clayton et al., 2011; Warren-Smith et al., 2007). Data presented by Clayton et al. (2011), as well as Eisersiö et al. (2015), suggested that using mean rein tension over the entire stride to discriminate between different experimental conditions may be difficult and that minimal and maximal values could be more useful. A complete description of the oscillations in rein tension should include variables describing the magnitude of the peaks and troughs in rein tension, and the amplitude, duration, and rate of tension change during the rise and fall phases. In addition, summation of the tension values (area under the curve) during the complete oscillation may be useful (Clayton et al., 2011). The timing of these events, for example, when minima and maxima occur, relative to the horse’s stride cycle provides relevant information. Factors that affect rein tension include the horse’s gait, the rider’s ability to follow the movements of the horse, the rein aids given by the rider to communicate with the horse, and the horse’s response to the rein aids (von Borstel and Glißman, 2014; Clayton et al., 2003; Egenvall et al., 2012; Egenvall et al., 2015; Eisersiö et al., 2013, 2015; Heleski et al., 2009). Oral lesions can be due to abrasion of soft tissues, such as the cheeks or tongue, by contact with sharp edges of the teeth or pressure applied by the bit to the oral tissues (Björnsdottir et al., 2014; Tell et al., 2008). However, the susceptibility of different oral tissues to pressure of different magnitudes and durations is not known. Because increased pressure on the oral tissues is a direct consequence of an increase in rein tension, studies of rein tension during riding are required to provide normative values as a prerequisite to seeking information regarding the magnitude and effects of excessive rein tension. This study analyzes rein tension within strides of canter performed by professional riders mounted on horses in their own training. The long-term goal is to further our understanding of how the interaction between the horse and riders affects forces between the rein and the bit, and the bit and the mouth, to improve both equine welfare and riding pedagogy. The aim is to quantify the patterns and magnitudes of the applied rein tension by measuring variables that describe the minima, maxima, and the absolute difference between them together with the horse’s sagittal-plane head rotation. These variables were analyzed in relation to rein (inside/outside), the rider’s position in

saddle (sitting, 2-point seat), passing through corners, circling, performing lateral movements, and riding with shorter or longer strides (collected/working/lengthened). Our hypotheses were that rein tension is equal in the left and right reins regardless of direction of travel or movements performed and that rein tension differs when shortening versus lengthening the stride, and rein tension differs when the rider uses a 2-point seat versus sitting in the saddle. Materials and methods Riders and horses Rein tension data were collected from 8 professional riders (mean  STD height 173  6 cm and weight 65.5  10 kg) riding horses they had trained regularly over a period of one month to 22 years (median 24 months). The horses were classified according to their level of training as advanced, medium, basic, or young. Seven riders rode 3 horses each and 1 rider rode 2 horses (n ¼ 23). All horses wore their own well-fitting saddle (as assessed by the riders) and a bridle with a snaffle bit. Fifteen of the snaffles were double jointed, 2 of which had fixed rings and 1 had a small port. Six bits were single jointed, 2 of which were full cheek. Two bits were unjointed, 1 with rigid rings and 1 with rubber rings. One rider was left handed, the others were right handed. When asked the riders assessed 7 horses as being easier to bend to the left, 15 horses were easier to bend to the right, 1 horse was equally easy to bend left and right, and 1 horse was easier to bend right at the trot and left at the canter. Equipment Data collection took place at each horse’s current stable in an outdoor (n ¼ 4 riders, gravel-based, the smallest 23  62 m and the largest 40  80 m) or indoor (n ¼ 4 riders, 2 sand-fiber arenas and 2 sand-wood chip arenas, the smallest 20  50 m and the largest 24  62 m), riding arena depending on the weather conditions. Before mounting by the rider, custom-made rein tension meters (sampling rate: 128 Hz; measuring range: 0-500 N; resolution: 0.11 N), were fitted onto leather reins. A cable from each tension meter ran forward along the rein and up the cheek piece of the bridle, to an inertial measurement unit (IMU, x-io Technologies Limited, UK) fastened at the browband of the bridle using hook and loop fasteners. The rein tension meter (Eisersiö, 2013) was successfully tested in a tensile test machine for stability and repeatability (e.g., see Egenvall et al., 2016) and was calibrated before starting the riding sessions for each rider by suspension of 13 known weights ranging from 0 to 20 kg. It took about 10 minutes to fit the equipment onto the horse and synchronize the rein tension meters with the videos. Video recordings (Canon Legria HF200, 25 Hz) of the entire riding session were made from the middle of one of the long sides of the arena. All horses were judged to be free from lameness by a veterinarian who visually evaluated the videos. Study design The riders were asked to perform a flatwork/dressage training session that was appropriate for the educational level of each horse including periods of walk, trot, and canter. The whole riding arena was used, and the duration of the riding session and the exercise performed was determined by the rider. Synchronization of equipment After the rider had mounted at the start of the session and before dismounting at the end, the rein tension meter was synchronized with the video recordings by pulling the right rein to apply the

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

tension meter 5 times while counting aloud in front of the camera, then repeating the process a second time. The video ran continuously through the calibration procedure and exercise session so the elapsed time was comparable between the video and the IMU/rein tension meter system. Variables and data management For this study, rein tension data were downloaded to a personal computer and analyzed using custom scripts in MATLAB (MathWorks Inc., Natick, Massachusetts, USA). One investigator (Marie Eisersiö) scrutinized the videos and categorized the data according to rider position (sitting/2 point), the presence of corners (left/right/no corner), the presence of circles (left/right/no circle), lateral movements (half pass left/half pass right/ no lateral movements), and stride length (collected/working/ lengthened based on whether the horses performed a working canter or cantered with lengthened or shortened (collected) strides). The evaluator-determined gait categorizations based on video evaluations were compared with head acceleration data from the IMUs and were further validated by a second researcher (A. Egenvall) during the data preparation phase. The direction of head rotation was classified according to the whether the nose was moving cranially or caudally relative to the poll as determined from Euler angles, derived from the gyroscopic IMU signal from the sensor on the poll. An increase in head angle indicated that the nose was moving cranially relative to the poll and, conversely a decrease in head angle indicated that the nose was moving caudally relative to the poll. Data for the canter were stride-split at the maximal positive vertical acceleration signal from the poll, which is a highly consistent event. In canter, it occurs 24%-27% of stride duration after first contact of the trailing hind limb (unpublished data). Within each stride, the maximal and minimal rein tension peaks were identified using a peakfinder (the MATLAB tool peakfinds, with [“meanpeakdistance,” 1] as added option). Statistical modeling The statistical analysis was conducted in SAS (SAS Institute Inc., Cary, NC 27513). Strides in which the difference in magnitude differences between MAX and MIN was <1N were deleted (n ¼ 106) and strides with a length 60 samples (n ¼ 400) and over 100 samples (n ¼ 548) were deleted. The outcome was tension in the left and right reins during canter. Dependent data from each stride and rein were maximal rein tension (MAX), minimal rein tension (MIN), and the absolute change between MAX and MIN (CHANGE). Each observation belonged to a “compound category” (e.g., left rein, half pass to the left, working canter, rider sitting). The outcomes were transformed along the ladder of powers to achieve the transformation closest to normality, checking that the means and medians were deemed close (in this case mean differing from median by not more than 10% of the median), the standard deviations were “small” (50% of the mean), and skewness and kurtosis close to zero. Lead (whether cantering with left or right limb as the leading fore), rider position (sitting/2 point), corners, circles, lateral movements, and stride length (collected/working/lengthened) were treated as fixed-effects variables. The direction of movement of the horse’s nose relative to the poll (cranial, caudal) at the identified maxima and minima was also modeled. Training level of the horse (advanced, medium, basic, young horse) was included as a fixed effect. To evaluate the controlling variables, nose direction, and horse level of training, models with data for both reins combined were evaluated and reduced. To determine statistical significance between gaits and reins, exercises and rider’s position

65

models were stratified by rein and each variable was evaluated while controlling for relative nose direction and horse level. Inside/ outside rein, defined according to whether the rein was on the same (inside) or opposite (outside) side as the leading limb in the canter, was forced in as a fixed effect. The interaction between rein and canter lead was the only interaction tested both in the overall models, and in the rein and canter lead specific models. For pairwise comparisons, only the interaction terms within each lead or within sides were compared, for example, the inside rein in left canter was not compared to the outside rein in right canter. Random effects were rider, horse, category within horse, category within horse-side (only overall models), deemed by the Akaike criterion. The percent of the variation contributed by horse and rider was estimated, dividing by the sum of all sources of variation. Model reduction, where applicable, was based on the type III sums of squares to P < 0.05. The correlation structure was variance component. PROC MIXED (SAS Institute Inc., Cary, NC 27513) was used for modeling. Pairwise comparisons were conducted. Variables and pairwise comparisons were considered borderline if 0.05

0.001, but significant if P < 0.001. Descriptive data are presented as boxplots for gait and rein, rider position, exercises and horse. The distributions of the dependent data are presented, including skewness and kurtosis statistics. In general, back-transformed least square means have been used to demonstrate the magnitude of the modeled variables. Results General descriptive results In total, MAX and MIN values were determined for 21,548 rein tension cycles that had a complete set of data. Figure 1 illustrates 12 examples of minima and maxima identified from stride-split raw data. The median length of the strides was 83 samples (0.65 seconds) with a range of 61-100 samples (0.48-0.78 seconds). Table 1 summarizes the general distributions and timing of MAX, MIN, and CHANGE. Magnitudes of these variables are illustrated in boxplots by combining gait and rein, by rider position and exercises (Supplementary information 1-2) and by horse (Figure 2). The 3 outcome variables were deemed best as square-root transformed (Table 1). The nose was moving cranially relative to the poll during 48% of the strides at MAX and 67% of the strides at MIN. Rein, seat, exercises, and gaits Figure 3 demonstrates least square estimates for inside/outside rein combined with canter lead. There were no statistical difference for MIN, comparing inside/outside rein combined with canter lead, whereas both group effects and all pairwise comparisons evaluated were significant for MAX and CHANGE, except the comparison between inside and outside reins for CHANGE in right canter (P ¼ 0.13). Figures 4-6 show least square estimates for rider position and exercises identified by inside (n ¼ 10,697) and outside rein (n ¼ 10,851). For the rider’s position, the magnitudes of all variables were higher (Figures 4-6) in sitting canter compared to the 2-point seat (group-P-values are the same as those from pairwise comparisons in this case), except for inside rein MIN value (P ¼ 0.27). The right circle had lower values for all variables than either the left circle or no circle when including both significant and borderline findings (Figures 4-6). For CHANGE in the outside rein (P ¼ 0.10) and MIN for the inside rein, the group P-values are high (P ¼ 0.42). For the corners, no variables had P-values <0.05. Outside rein values for MAX and MIN were higher when the strides were lengthened compared to working canter (Figures 4-6; MAX/MIN group P-values 0.03 and 0.0014). For MIN, there was a

66

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

Figure 1. Each row shows 3 strides from a horse, 1 horse per row, for the identification of minima and maxima of rein tension. A maximum (MAX: red bar) and a minimum (MIN: magenta bar). In the example, the first row is a left rein in left canter, the second a right rein in left canter, the third a left rein in right canter, and the fourth a right rein in right canter. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

borderline difference in the outside rein with lengthening having a higher value than collection. Half pass to the right had higher values than half pass to the left or no half pass for MAX (group P-value < 0.0001) and CHANGE (P ¼ 0.0004) in the inside rein and for MIN values in both reins (inside/ outside rein P < 0.0001/P ¼ 0.0001, Figures 4-6).

Table 1 Distribution of the dependent variables, their square-root transforms, and timing data describing rein tension in 23 horses ridden by 8 riders at canter Variable

N

Mean SD

Min Median Max

MAX (N) MAX square-root transform Percent in stride where MAX found MIN (N) MIN square-root transform Minimum (N) if found before MAX Minimum (N) if found after MAX Change MAX  MIN (N) Change square-root transform Percentage of stride where MIN found, if MIN found before MAX Percentage of stride where MIN found, if MIN found after MAX

21,548 56.1 21,548 7.2

33.0 1.5 2.2 1.2

48.9 7.0

21,548 45.9

35.0 1.0

39.2

SKEW KURT

284 1.9 16.9 0.1

1.2 0.3

99.0

Effect of nose angle MAX tension was higher when the horse’s nose was moving cranially relative to the poll, whereas MIN was higher when the nose was moving caudally. For CHANGE, the highest values were associated with the nose moving caudally at both MIN (borderline P ¼ 0.01) and MAX (P < 0.0001). Horse level For horse level, 6 pairwise borderline significances were found for the rein tension variables (Table 2). Scrutinizing the point estimates, MAX and CHANGE values were highest for the young horses, whereas advanced horses had the highest MIN values. The variation

21,548 21,548

8.5 2.6

8.3 0.0 1.3 0.0

5.9 2.4

50 3.6 7.1 0.1

1.8 0.7

9,994

8.5

7.8 0.0

6.3

49.9 3.7

1.7

11,554

8.5

8.8 0.0

5.6

50

3.4

1.8

The horse contributed 7%, 27%, and 29% of the variation to the MIN, MAX, and CHANGE models, respectively, whereas the rider contributed 19%, 0%, and 0% of the variation to the MIN, MAX, and CHANGE models, respectively.

21,548 47.6

30.4 1.0

40.4

271

2.3

1.3

Discussion

6.6

2.1 1.0

6.4

16.5 0.1

0.4

9,994 35.9

15.7 1.0

33.3

88.9

11,554 60.0

18.7 7.0

58.8

98.9

21,548

Zero values for skewness (SKEW) and kurtosis (KURT) are optimal.

To date, there have been only a few studies describing rein tension during cantering (Clayton et al., 2003; Clayton et al., 2005; Egenvall et al., 2015; Eisersiö et al., 2015; Kuhnke et al., 2010). Comparing to previous studies, the present study reports data from a considerably larger number of strides (21,548) in 23 horserider dyads and includes data describing the minimal and maximal rein tension values and the changes between them. These variables have not been reported previously in canter, although they have been reported for unridden horses trotting

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

67

Figure 2. Box plot of rein tension parameters for 23 horses (no data from horse 6). Each of the 8 riders rode 3 sequentially; horses 1, 2, and 3 were ridden by rider 1; horses 4, 5, and 6 to rider 2, etc.). Parameters shown are MAX (maximum value at peak rein tension), MIN (the minimum peak value), and CHANGE (the absolute difference MAX  MIN). The boxes show medians (red lines) and 25th and 75th percentiles (black lines). The whiskers contain data not considered outliers and outliers (defined as greater than 1.5 interquartile range above the upper quartile, or less than 1.5 interquartile range below the lower quartile) are shown as red crosses. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

with side-reins (Clayton et al., 2011), However, they are likely to be important in relation to welfare concerns as they are variables that, to a large extent, determine the pressure the oral tissues are

subjected to. The relationship between external pressures and interstitial stresses is influenced by the tissue characteristics overlying the bone. Given the pressure-time dependency in the

Figure 3. Least square means estimates from modeling the rein (in-inside, out-outside) and canter lead (left canter [LC], right canter [RC]) gait interaction (23 horses, 8 riders, 21,548 observations). The overall evaluation of statistical significance was significant for MAX and CHANGE (P < 0.0001). For the MAX variable, all evaluated pairwise comparisons were significant at P < 0.0001, except inside and outside right canter (P ¼ 0.010). For CHANGE, all evaluated pairwise comparisons were significant at P < 0.0001, except inside and outside right canter which was nonsignificant (P ¼ 0.13).

68

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

Figure 4. Least square means estimates modeling MAX values (in/out inside/outside rein). Exercises and seat position variables are modeled and those with a group P-value <0.05 are shown. Significant pairwise comparisons are shown, white bars indicate P < 0.0001 and black bars 0.05

0.0001. Rider position, 2-point seat versus Sit; Sit, sitting; Cil, circle in left direction; Cir, circle in right direction; Nc, no circle; Col, collection; Len, lengthening; Work, working canter, neither collection/lengthening; Hpl, half pass to the left; Hpr, half pass to the right; Nl, no lateral movements.

etiology of pressure sores (Reswick and Rogers, 1976), periodic release of pressure has been recommended so as to spare the tissues from ischemic damage (Thomas, 2001).

Rein sensors cannot discriminate tension caused by the rider pulling the reins from tension due to the horse pushing against the bit. Clayton et al. (2011) suggested that the rider establishes a

Figure 5. Least square means estimates modeling MIN values (in/out inside/outside rein). Exercises and seat position variables are modeled and those with a group P-value <0.05 are shown. Significant pairwise comparisons are shown, white bars indicate P < 0.0001 and black bars 0.05

0.0001. Rider position, 2-point seat versus Sit; Sit, sitting; Cil, circle in left direction; Cir, circle in right direction; Nc, no circle; Col, collection; Len, lengthening; Work, working canter, neither collection/lengthening; Hpl, half pass to the left; Hpr, half pass to the right; Nl, no lateral movements.

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

69

Figure 6. Least square means estimates modeling CHANGE values (in/out inside/outside rein). Exercises and seat position variables are modeled and those with a group P-value <0.05 are shown. Significant pairwise comparisons are shown, white bars indicate P < 0.0001. Rider position, 2-point seat versus Sit; Sit, sitting; Cil, circle in left direction; Cir, circle in right direction; Nc, no circle; Hpl, half pass to the left; Hpr, half pass to the right; Nl, no lateral movements.

contact with the horse’s mouth that is the baseline tension represented by the minima in the tension recording. This likely explains why statistical analysis indicated that the rider has more influence on MIN values than the horse. An intermittent pressure pattern with periodic release of pressure has a sparing effect on tissues in relation to the development of ischemic damage (Thomas, 2001). Therefore, it is suggested that the intermittent pattern of rein tension is potentially less damaging than continuous, steady pressure. The regularly repeated peaks in rein tension are due to the mechanical effects of the movements of the horse’s neck and head under the influence of gravity and inertia when moving. The trot has been studied in more detail than the canter in this regard. During trotting, unridden trot and horses ridden in a free head and neck position show 2 rein tension peaks per stride that occur in the later part of the diagonal stance phase coinciding with the time when the head nods down to its lowest point (Clayton et al., 2011; Eisersiö et al., 2013). In horses ridden on the bit at trot, additional peaks may be present in the suspension phase (Eisersiö et al., 2013; Egenvall et al., 2016) which are thought to represent the rider’s application of half halts that act to maintain the horse’s balance and self carriage. In canter, there is 1 primary tension peak per stride which occurs around midstance of the diagonally-synchronized pair of limbs (Egenvall et al., 2015) and coincides with lowering of the head and neck. The effect of the head-neck oscillations on rein tension is thought to explain why the statistical models suggest that the horse contributes more of the variation than the rider to MAX and CHANGE. Superimposed on the rein, tension peaks associated with the horse’s head and neck mechanics are the tension [increase or decrease] caused by the movements the rider makes to follow the motion of the horse’s neck and head, the effects of the rider’s aids, such as to achieve the half halt, and the horse’s response to

those aids. The effects of the rider’s half halts are likely to be considerably more variable than the mechanical effects of the horse’s body motion both in magnitude and duration in contributing to the variation than the horse per se. In this study, the horses and riders were accustomed to each other which could explain why the dyad and the horses were not statistically separable. Perhaps, a rider can actually remodel the horse-effect over time. To elucidate the real question of whether it is the horse or the rider that influences the rein tension data from different riders riding several horses over time with washout periods in between are needed. In this study, MIN value (8.8  9.2 N) is a little higher and MAX value (56.8  33.5 N) is considerably lower than the tension values reported by Eisersiö et al. (2015). However, the values were determined differently; the maximum values presented by Eisersiö et al. (2015) were extracted from the entire data string without stride segmentation (crude level) versus the stride-by-stride analysis in the present study (stride level) so the values cannot be directly compared. Some of the reported parameters show considerable variability which reflects the fact that the horses were being ridden as in a normal schooling session rather than under strictly controlled experimental conditions. Rein tension can also vary greatly when the same horse is ridden by different riders with comparable levels of experience (Christensen et al., 2015). This may be due to differences in riding styles, ability to perceive changes in the horse’s performance or subtlety in applying corrective aids. In young horses, changes in rein tension can be used to achieve control of speed and direction and the horse is encouraged to seek contact with the bit. As the level of training increases, these functions are taken over by the rider’s legs and seat allowing rein tension to play a more subtle role in maintaining suppleness of the horse’s jaw and in fine control of the horse’s head and neck position. Young horses that are being ridden actively forward have a

70

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

Table 2 Least square means estimates (estimate and SE, including the back-transformed estimate) for the controlling variables, nose angle decrease/increase, and horse level Outcome: independent variables

Estimate

MAX (N) Nose angle direction Caudal 7.4 At MAX Cranial 7.3 Horse level Advanced 7.7 Medium 6.6 Basic 6.2 Young 9.0 MIN (N) Nose angle direction Caudal 3.1 At MIN Cranial 3.3 Horse level Advanced 3.9 Medium 3.0 Basic 2.7 Young 3.0 CHANGE (N) Nose angle direction Caudal 6.8 At MIN Cranial 6.7 Nose angle direction Caudal 6.8 At MAX Cranial 6.6 Horse level Advanced 7.0 Medium 5.9 Basic 5.8 Young 8.3

SE

Back-transformed estimate

P-value

0.3

55.0

<0.0001

0.3

53.2

<0.0001

0.5 0.5 0.3 0.8

58.8 43.3 38.7 80.2

0.02

0.3

9.6

<0.0001

0.3

10.7

<0.0001

0.4 0.3 0.3 0.4

15.5 9.3 7.3 9.3

0.002

0.3

45.9

0.01

0.3

45.0

0.01

0.3

46.8

<0.0001

0.3

44.2

<0.0001

0.6 0.5 0.3 0.8

48.8 35.3 33.2 68.7

0.01 0.02 0.01

0.002 0.002

0.002

0.01 0.01

0.004 0.004

Significance of pairwise comparisons of categories within the variables are shown, where the “same” P-value within a column demonstrates P < 0.05 for this comparison.

large bounding stride at canter with an exaggerated rocking motion of the trunk, neck and head, which is thought to be, at least partly, responsible for the low MIN and high MAX values. As the horse’s balance and self carriage improve, the rocking motion diminishes and MAX values become smaller. At the same time, the horse becomes more confirmed in seeking contact with the rider’s hand so MIN values increase. As in the study of Eisersiö et al. (2015), the 2-point seat was associated with lower rein tension than sitting canter, which supports the experimental hypothesis. It has been demonstrated that when trotting in a 2-point seat, the rider has a different balance in the saddle (Münz et al., 2013), the center of pressure, calculated from the forces applied to an electronic saddle pressure mat, is further forward, which indicates that the horse’s back is loaded more cranially (Geser-von Peinen et al., 2013). In addition, the rider’s position is more stable in the 2-point position (Peham et al., 2010). The lower rein tension values and their later occurrence when cantering in a 2-point seat may accordingly be related to lower forces on the horse’s back, greater stability of the rider, and/ or to the fact that riding in a 2-point seat is regarded as a time of relaxation and finding harmony that is often preferably used in young horses, and during which the rider imposes less influence on the horse (Eisersiö et al., 2015). Comparing inside and outside reins for left and right canter leads without turns or circles and when controlling for exercises and other variables, all groupwise comparisons for MAX and CHANGE, and many pairwise comparisons were significantly different (Figure 3). A previous study reported that tension in the

outside rein was lower in right than left canter (Egenvall et al., 2015), which does not agree with the current findings. However, the statistical differences in MAX for reins within canter were not directly studied by Egenvall et al. (2015). The present study, which is more targeted toward these specific results, had higher MAX tension in the outside rein in left canter, whereas the inside rein had higher tension in the right canter, that is, tension was always higher in the right rein than the left regardless of canter lead. The same effect was seen in half pass to the right in which the right (inside) rein had significantly higher tension than the left (outside) reins for all parameters analyzed. The tendency for rein tension to be higher in the right rein may be a consequence of laterality of horse and/or rider (seven of eight riders claimed to be right handed in the current study) that warrants further study. The fact that there were few significant results when turning corners was likely due to the small number of strides and consequently low statistical power. During circling, both within-inside (MAX and CHANGE) and within-outside (MAX and MIN) reins showed higher values for most parameters when circling to the left, which may be another manifestation of asymmetry, for example, laterality in horses or riders. For left- and right-lateralized horses, rein tensions were not usually mirrored on left and right sides (Kuhnke et al., 2010) which supports our suggestion that laterality in horses and humans is a contributor to rein tension asymmetry. In earlier analyses, however, the inclusion of rider or horse laterality did not improve model fit (Egenvall et al., 2015; Eisersiö et al., 2015). Lengthened strides had higher values for MIN, MAX, and CHANGE compared with collected or working canter strides although most comparisons had borderline significance. The differences may be a consequence of increased head acceleration at faster speed or riders may choose to take a stronger contact when lengthening the strides. Because only a few horses performed shortened canter strides that could be classified as collection, more studies should be done on this subject. MAX and CHANGE had higher values when the nose was moving caudally relative to the poll (at MAX for MAX and at both MAX and MIN for CHANGE), whereas for MIN, the values were larger when moving cranially at MIN. The most likely explanation is that the higher tension was due to the rider using rein signals to influence the positioning of the horse’s head or to adjust the horse’s balance or speed. The collecting or balancing aids are applied early in the canter stride when both horse and rider are rotating backward (nose up trunk rotation in the horse, posterior trunk rotation in the rider). The goal of the collecting aids is to discourage nose-down trunk rotation when the forelimbs are grounded which would put the horse “on the forehand.” The present study selected peaks and troughs in “chaotic” rein tension data within each stride. In some strides, it proved difficult to find seemingly relevant occurrences of both events. Graphical scrutinization suggests that the extracted data do indeed represent the highest and lowest events of the rein tension signal within each stride (Figure 1) although the inherent variability reflects the influences of 2 somewhat independent contributors (the horse and the rider). Other approaches for selecting peaks were tried, but the method presented here was deemed to be more repeatable, transparent, and parsimonious. However, the first aim (using other algorithms) was to study magnitudes and durations (and average rates) of “rise” and “fall” of rein tension suggesting these as variables that may be directly associated with the actual events of riding at the bit-horse interface and likely also with behavioral response (only nose angle measured in this regard). For example, descriptive results from these data showed that the durations of rise and fall were not systematically different, with substantial between horse and rider variation (data not shown). Based on our work, we suggest that to gain systematic results for finer

A. Egenvall et al. / Journal of Veterinary Behavior 13 (2016) 63e71

parameters, rein tension during riding should likely first be studied using strict experimental settings or more limited data collection, so that transparent findings can be achieved even if the possibility for extrapolation may be compromised. (The last 2 parameters in Table 1 relate to the timing of MIN and MAX, but because of the systematic position of the stride-split, they should not be regarded as reflecting durations of rise and fall.) Determining the stride-split based on data from the poll could be problematical in that unsteadiness of the head sometimes made it difficult to determine an exact stride split. In future studies, it might be preferable to base the stride-split on croup accelerations or a combination of poll and croup accelerations which would likely enhance the accuracy and have the added benefit of decreasing the likelihood that a somewhat imprecise stride-split would add difficulties in identifying all relevant MAX (and MIN) peaks. In addition, even in a study where riders were told to ride normally, there is likely a spectator effect which could be expressed by a change in the rider’s behavior that affected rein tension. A further limitation is that exercises were defined by 1 person which involved an element of subjectivity, for example, in distinguishing between working and collected canter. Conclusion The models indicate that the horse contributes most of the variation in rein tension and this supports the suggestion that the horse or the rider-horse dyad is responsible for the basic rein tension pattern during canter. Some consistent left/right asymmetries emerged indicate that asymmetry of horse and rider plays a role in rein tension. Hypotheses that were supported indicated that rein tension was higher in the sitting canter than in the 2-point seat and when the stride was lengthened. The rider contributes more to the variation of MIN which is regarded as a baseline tension, whereas the horse makes a larger contribution MAX and CHANGE as a consequence of the mechanical movements of the head and neck relative to the trunk. Overall, the highest MAX values were recorded in young horses and these values decreased with training. Acknowledgments The study was funded by the Swedish Research Council Formas (grant number: 2011-561). The funding body had no influence on the study design, in the collection, analysis, and interpretation of data; in the writing of the article; or in the decision to submit the article for publication. The authors thank the riders for their contributions. The idea for the article was conceived by Agneta Egenvall, Marie Eisersiö, and Hilary M. Clayton. The experiments were designed by Agneta Egenvall and Marie Eisersiö. The experiments were performed by Marie Eisersiö. The data were analyzed by Agneta Egenvall. The article was written by Agneta Egenvall, Hilary M. Clayton, Marie Rhodin, and Lars Roepstorff. Ethical considerations According to the Swedish legislation, it was not necessary to have an ethical permit for this study. Conflict of interest None of the authors have any potential conflicts of interest, including any financial, personal, or other relationships with other

71

people or organizations, within 3 years of beginning the submitted work that could inappropriately influence, or be perceived to influence, the work. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jveb.2016.03.007. References Anthony, D., Brown, D., 1991. The origins of horseback riding. Antiquity 65, 22e38. Björnsdottir, S., Frey, R., Kristjansson, T., Lundström, T., 2014. Bit-related lesions in Icelandic competition horses. Acta. Vet. Scand 56, 40. von Borstel, U.K., Glißman, C., 2014. Alternatives to conventional evaluation of rideability in horse performance tests: suitability of rein tension and behavioural parameters. PLoS One 1, 1e9. Christensen, J.W., Munk, R., Hawson, L., Koenig von Borstel, U., Roepstorff, L., Egenvall, A., 2015. Subjective scoring of rideability by professional ridersdis it linked to rein tension and occurrence of conflict behavior? International Society for Equitation Science 2015, Vancouver. Abstract 9. Clayton, H.M., Larson, B., Kaiser, L.J., Lavagnino, M., 2011. Length and elasticity of side reins affect rein tension at trot. Vet. J. 188, 291e294. Clayton, H.M., Singleton, W.H., Lanovaz, J., Cloud, G.L., 2003. Measurement of rein tension during horseback riding using strain gage transducers. Exp. Tech. 27, 34e36. Clayton, H.M., Singleton, W.H., Lanovaz, J.L., Cloud, G.L., 2005. Strain gauge measurement of rein tension during riding: a pilot study. Eq. Comp. Exerc. Physiol. 2, 203e205. Egenvall, A., Eisersiö, M., Rhodin, M., van Weeren, R., Roepstorff, L., 2015. Rein tension during canter. Comp. Exerc. Physiol. 11, 107e117. Egenvall, A., Roepstorff, L., Eisersiö, M., Rhodin, M., van Weeren, R., 2016. Rein tension in walk and trot. Acta. Vet. Scand 57, 89. Egenvall, A., Eisersiö, M., Roepstorff, L., 2012. Pilot study of behavior responses in young riding horses using 2 methods of making transitions from trot to walk. J. Vet. Behav.: Clin. Appl. Res. 7, 157e168. Eisersiö, M., 2013. How to Build a Rein Tension Meter. Degree project in Biology, Swedish University of Agricultural Science, Available at: http://stud.epsilon.slu.se. Accessed January 1, 2016. Eisersiö, M., Roepstorff, L., Rhodin, M., Egenvall, A., 2015. Rein tension in eight professional riders during regular training sessions. J. Vet. Behav.: Clin. Appl. Res. 10, 419e426. Eisersiö, M., Roepstorff, L., Weishaupt, M.A., Egenvall, A., 2013. Movements of the horse’s mouth in relation to the horse-rider kinematic variables. Vet. J. 198, e33ee38. Geser-von Peinen, K., Latif, S.N., Wiestner, T., Bitschnau, C., Renk, B., Weishaupt, M.A., 2013. Applied load on the horse’s back under racing conditions. Vet. J. 198, e88ee92. Heleski, C.R., McGreevy, P.D., Kaiser, L.J., Lavagnino, M., Tans, E., Bello, N., Clayton, H.M., 2009. Effects on behaviour and rein tension on horses ridden with and without martingales and rein inserts. Vet. J. 181, 56e62. Hobbs, S.J., Baxter, J., Broom, L., Rossell, L., Sinclair, J., Clayton, H.M., 2014. Posture, flexibility and grip strength in horse riders. J. Hum. Kinet. 42, 113e125. Kuhnke, S., Dumbell, L., Gauly, M., Johnson, J.L., McDonald, K., von Borstel, U.K., 2010. A comparison of rein tension of the rider’s dominant and non-dominant hand and the influence of the horse’s laterality. Comp. Exerc. Phys. 7, 57e63. Münz, F., Eckardt, F., Heipertz-Hengst, C., Peham, C., Witte, K., 2013. A preliminary study of an inertial sensor-based method for the assessment of human pelvis kinematics in dressage riding. J. Equine Vet. Sci. 33, 950e955. Ödberg, F.O., Bouissou, M.F., 1999. The development of equestrianism from the baroque period to the present day and its consequences for the welfare of horses. Equine Vet. J. 31, 26e30. Peham, C., Kotschwar, A.B., Borkenhagen, B., Kuhnke, S., Molsner, J., Baltacis, J., 2010. A comparison of forces acting on the horse’s back and the stability of the rider’s seat in different positions at the trot. Vet. J. 184, 56e59. Reswick, J.B., Rogers, J., 1976. Experience at Rancho Los Amigos Hospital with devices and techniques to prevent pressure sores. In: Kenedi, R.M., Cowden, J.M., Scales, J.T. (Eds.), Bedsore Biomechanics. University Park Press, Baltimore, MD, pp. 301e310. Tell, A., Egenvall, A., Lundström, T., Wattle, O., 2008. The prevalence of oral ulceration in Swedish horses when ridden with bridle and bit. Vet. J. 178, 405e410. Thomas, D.R., 2001. Issues and dilemmas in the prevention and treatment of pressure ulcers: A review. J. Gerontol. 56A, M328eM340. Warren-Smith, A.K., Curtis, R.A., Greetham, L., McGreevy, P.D., 2007. Rein contact between horse and handler during specific equitation movements. Appl. Anim. Behav. Sci. 108, 157e169.