Rein tension in 8 professional riders during regular training sessions

Accepted Manuscript Rein tension in eight professional riders during regular training sessions M. Eisersiö, M. Rhodin, L. Roepstorff, A. Egenvall PII:

S1558-7878(15)00078-7

DOI:

10.1016/j.jveb.2015.05.004

Reference:

JVEB 892

To appear in:

Journal of Veterinary Behavior

Received Date: 5 January 2015 Revised Date:

12 April 2015

Accepted Date: 21 May 2015

Please cite this article as: Eisersiö, M., Rhodin, M., Roepstorff, L., Egenvall, A., Rein tension in eight professional riders during regular training sessions, Journal of Veterinary Behavior (2015), doi: 10.1016/ j.jveb.2015.05.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Rein tension in eight professional riders during regular training sessions

2 3

M. Eisersiö a, M. Rhodin a, L. Roepstorff b, A. Egenvall a *

4

a

5

Swedish University of Agricultural Sciences, Box 7054, SE-750 07 Uppsala, Sweden

6

b

7

Veterinary Medicine and Animal Husbandry, Swedish University of Agricultural Sciences,

8

Box 7046, SE-750 07 Uppsala, Sweden

Department of Clinical Sciences, Faculty of Veterinary Medicine and Animal Husbandry,

10

* Corresponding author. Tel.: +46 70 3799544.

11

E-mail address: [email protected] (A. Egenvall)

12

SC

9

RI PT

Department of Anatomy, Physiology and Biochemistry, Unit of Equine Studies, Faculty of

Abstract

14

Rein tension signals are commonly used to communicate the intended speed, direction and

15

head carriage to the horse during horseback riding. Rein tension have previously been

16

recorded relative to gait, exercises and turning maneuvers. The aim of this study was to target

17

the between-gait and between-exercise variation in rein tension, controlling for riders and

18

horses within riders, the between-rein variation and the general within-gait/exercise variation,

19

during entire riding sessions. Eight riders with 3 horses each were included in the study and

20

each horse was fitted with a custom-made rein tension meter fastened on leather reins. Rein

21

tension data and video films were collected during the riding session and the video films were

22

scrutinized and categorized according to ridden exercises. Statistics used to model rein tension

23

in mixed models were ‘median’, area under curve, averages of 2 and 25 percentiles (‘low’)

24

and of 75 and 98 percentiles (‘high’), and the difference between 98 and 2 percentiles

25

(‘range’). Fixed effects were rein, gait, rider’s position, horse level and type of ridden exercise

26

and random effects were horse-side, rider, horse and trial within horse. The analyzes

27

demonstrate substantial variation between gaits, rider position within gait and between riders

28

and horses. Considering data on short reins, the major determinants found for amount of rein

29

tension was gait (walk [median 12 N both reins] < trot [median 14-19 N left/right rein and

30

sitting/posting] < canter [median 13-24 N left/right rein and sitting/light seat]) as well as the

31

rider’s position in the saddle for trot (posting [median 14 N both reins]
32

N/19 N left/right rein]) and canter (light seat [median 13-17 N left/right rein and left/right

33

canter] < sitting [median 20-24 N left/right rein and left/right canter]). Regarding the 2 reins;

34

the right rein was highest in comparisons in the ‘high’ and ‘range’ models, while the inside

AC C

EP

TE D

M AN U

13

ACCEPTED MANUSCRIPT rein was highest in canter. Riders contributed to most of the variation in the ‘median’ and

36

‘low’ models, while horses contributed the highest relative variance estimates in the models

37

associated with high rein tension (‘high’ and ‘range’). Our results suggest that variables to

38

consider in rein tension studies are the gait of travel, the rider’s position in the saddle, the

39

ridden exercise performed, the educational level of horse, the rider and horse per se and to

40

some extent the left/right rein.

41 42

Key-words: Rein tension; Dressage; Gaits; Equine Welfare

43

RI PT

35

Introduction

45

The bit in the horse’s mouth, connected to the reins in the rider’s hand, is generally used to

46

condition the horse to respond to pressure signals i.e. negative reinforcement; pressure and

47

release of the bit against the tissues in the horse’s mouth. Applying rein tension signals is

48

commonly the means used for communicating speed, direction and head carriage to the horse

49

during horseback riding (Manfredi et al., 2010). Rein tension displays a constant variation of

50

magnitude during horseback riding, partly as a results of the rider’s cues, but also along with

51

the horse’s stride cycle (Clayton et al., 2003; 2011; Eisersiö et al., 2013), and depending on

52

the horse’s reaction to bit pressure (Clayton et al., 2011; Egenvall et al., 2011). Further, the

53

magnitude of rein tension has been found to be connected to gait (Clayton et al., 2005;

54

Kuhnke et al., 2010) and the skill level of the rider (Warren-Smith et al., 2007). Earlier rein

55

tension studies of horseback riding have recorded different gaits, exercises and turning

56

maneuvers (Clayton et al., 2003; Warren-Smith et al., 2007; Heleski et al., 2009; Egenvall et

57

al., 2012; Egenvall et al., in press) and the mean rein tension for walk, trot and canter ranged

58

from 7-43 N at the walk, 11-51 N at the trot and 16-104 N for the canter (Clayton et al., 2005;

59

Kuhnke et al., 2010).

M AN U

TE D

EP

AC C

60

SC

44

61

The Fédération Equestre Internationale (FEI) states that one of the objectives of dressage is

62

for the horse to accept the bit without any tension or resistance (Fédération Equestre

63

Internationale, 2014) and rein tension have been studied as a variable to determine rideability

64

(Borstel and Glißman, 2014). The use of tension via the rein for controlling the horse raises

65

some welfare concerns as it is likely that the pressure from the bit becomes uncomfortable for

66

the horse at some point (Christensen et al., 2011). In Ludewig et al. (2013) shortening the

67

reins by 10 cm resulted in an increase of rein tension of 10 N and the results suggested that

68

horses comply with shorter reins, i.e. more rein tension, by adapting their head position, step

ACCEPTED MANUSCRIPT length and by putting more pressure on the bit, yet at the same time these horses’ behavioral

70

expressions (open mouth, flattened ears) indicated that this posture and/or the increase of rein

71

tension was perceived as aversive. Furthermore, in Christensen et al. (2011) it was found that

72

naïve horses learn to avoid rein tension rather than habituate to it when encouraged to stretch

73

for a food reward. In addition, pressure from the bit in the horse’s mouth is associated with

74

lesions in the horse’s oral cavity (Tell et al., 2008) and conflict behavior (Egenvall et al.,

75

2012). Magnitude and loading rate of rein tension are factors that largely contribute to intra-

76

oral lesions (Clayton et al., 2011), yet there is still lacking information about these

77

relationships.

RI PT

69

SC

78

Measured rein tension is a complex signal, where a number of components contribute to the

80

resulting shape of the curve. Basically we expect between-horse, between-rider (Borstel and

81

Glißman, 2014), between-gait (Warren-Smith et al., 2007), between-exercise, between-rein,

82

between-stride variation, between-session, within-stride (Clayton et al., 2003; 2011; Eisersiö

83

et al., 2013), within-session variation and variation related to specific signaling to the horse

84

(Egenvall et al., 2012) or more extraneous circumstances. Some of these sources of variation

85

will be difficult to discern from each other. However, we should strive to quantify and

86

characterize the variation at each level.

TE D

M AN U

79

87

The aim of this study was to target the between-gait and between-exercise variation

89

controlling for riders and horses within riders, the between-rein variation, and the general

90

within-gait/exercise variation, by illustrating the distribution for rein tension during entire and

91

regular riding sessions performed by professional riders on familiar horses. Having identified

92

e.g. gaits, rider positions and ridden exercises in 8 riders riding 3 horses each (Eisersiö et al.,

93

2015), measuring rein tension simultaneously, we wanted to relate these variables to the rein

94

tension.

AC C

95

EP

88

96

Material and methods

97

Ethical permit

98

According to the Swedish legislation no specific ethical permit was necessary for this study.

99 100

Riders and horses

101

Data were collected from 8 professional riders (mean + STD height 173 + 6 cm and weight

102

65.5 + 10 kg) riding 3 horses each (n=24). The horses were either in training or owned by the

ACCEPTED MANUSCRIPT rider and had regularly been trained by the riders between 1 month and 22 years, median 24

104

months before start of the study. All horses wore their own saddle and bridle with a snaffle

105

bit. Further information on the horses and riders can be found in Eisersiö et al. (2015). The

106

riders all worked in the horse industry as riding instructors (with the exception of one, who

107

was only 14 years old, but was training horses in a professional enterprise) and/or horse

108

trainers on various levels. Two riders competed at advanced level (as classified in the national

109

Swedish system), 5 at intermediate level and 2 at basic level. One rider was left-handed, the

110

others were right-handed. Based on rider statement, 7 horses were easier to bend to the left, 15

111

horses were easier to bend right, one horse was equally easy to bend left and right, and one

112

horse was easier to bend right at the trot and to the left at the canter. The educational level of

113

the horse was stated by the riders as: basic (n=6), young horse (n=3), medium (n=5) and

114

advanced (n=4). Advanced horses had competed at Prix St. George, Intermediaire or Grand

115

Prix level; basic horses had entered low-level competitions only and medium horses were in

116

between. Young horses had been ridden for less than a year and had not competed.

M AN U

SC

RI PT

103

117

Equipment

119

Data collection took place at each horse’s current stable in an indoor (n=4 riders) or outdoor

120

(n=4 riders) riding arena, depending on the weather conditions. Each horse was fitted with a

121

custom-made rein tension meter (128 Hz), measuring range 0-500 N, resolution 0.11 N,

122

fastened on leather reins. A cable from each tension meter ran forwards along each rein and

123

up along the side piece of the bridle, behind the horse’s ear and to an Inertial Measurement

124

Unit (IMU, x-io Technologies Limited, UK) attached right below the browband of the bridle

125

using a custom made Velcro browband (Figure 1). The rein tension meter was calibrated

126

before the riding sessions started for each rider by suspending 13 known weights between 0-

127

20 kg from each meter. The rein tension meter was also successfully tested in a tensile test

128

machine for stability and repeatability of results (data not shown). Further details on the rein

129

tension meter can be found in Eisersiö (2013). All equipment was fitted on the horse in the

130

riding arena and each fitting took approximately 10 minutes including synchronization (see

131

below). Video recordings (Canon Legria HF200, 25 Hz) were made of the entire riding

132

session from the middle of one of the long sides of the arena. All horses were free from

133

lameness based on visual assessment of the videos by a veterinarian.

AC C

EP

TE D

118

134 135

Study design

ACCEPTED MANUSCRIPT 136

The riders were asked to demonstrate their normal routine with each horse for

137

flatwork/dressage and to ride in all gaits (walk, trot and canter). The whole riding arena was

138

used for the exercises and the length of the riding session was determined by the rider.

139

Synchronization of equipment

141

After the rider had mounted, and before dismounting in the end, the rein tension meter was

142

synchronized with the video recordings by manual applied tension to the right tension meter 5

143

times twice in a row while counting out loud in front of the camera.

RI PT

140

144

Data management

146

One investigator (ME) scrutinized the videos and categorized the data. Further information on

147

this protocol can be found in Eisersiö et al. (2015). In short, the categories used in this study

148

were gait (walk, trot, left lead canter, right lead canter), rider’s position (sitting, light seat,

149

posting), corners and turns (corner left/right, turn left/right), lateral movements (half-pass to

150

the left/right, shoulder-in left/right, leg-yield left/right) or riding in collection or lengthening

151

(trot, canter). The accuracy of the evaluator-determined video protocols (mainly gait) were

152

examined by comparing protocols to head acceleration data and main categorisations were

153

thus validated by a second researcher (AE) during the data management process. Rein tension

154

data were downloaded to a personal computer and processed in Matlab (MathWorks Inc.,

155

USA).

156

TE D

M AN U

SC

145

Statistics

158

Descriptive left/right rein tension statistics have been produced for the calibrated data by gait,

159

exercises within gait, exercises within rider and within horse, making description using mean

160

values of statistics (means, STDs, min, max, medians, 2 and 98% percentiles) based on horse-

161

specific data. For the outcome statistics selected for modelling descriptive statistics have been

162

done by rider position within gait and rein. The outcome in the multivariable modeling was

163

rein tension in the left and right rein, each observation was a statistic based on various activity

164

combinations in each horse. Outcome statistics used to model rein tension was median, area

165

under curve (’auc’), averages of 2 and 25 percentiles and of 75 and 98 percentiles, ‘low’ and

166

‘high’ respectively, and the difference between 98 and 2 percentiles ‘range’. The median

167

statistic was selected as a measure of average tension. ‘Low’ was selected to represent the

168

lower basic tension and ‘high’ to represent the higher pressure when the riders did the more

169

definite signaling or horses experienced the higher tension values. ‘Range’ was selected to

AC C

EP

157

ACCEPTED MANUSCRIPT demonstrate the range from ‘low’ to high’ within the same statistic, all modified from Clayton

171

et al. (2011). Models were combined for reins (both left and right rein tension in same model,

172

but separated) and data on walk, trot and canter were combined into the same model.

173

Dependent data were checked for normality, i.e. that the means and medians were deemed

174

close, the standard deviations judged as small, and skewness and kurtosis close to zero, or

175

otherwise suitably transformed. Fixed effects trial-level variables were rein (left/right), which

176

was forced in, gait [gait], whether the horse-rider combination was turning (left/right) or

177

passed through a corner (left/right) [turn/corner] compared to not doing any of this, performed

178

lateral movements [lateral] (shoulder in left/right, half-pass left/right, leg yield left/right) or

179

was riding in collection (all gaits) or lengthening (trot, canter) [collection/lengthening] versus

180

not doing those. The activity was also categorized according to [position] trot (sitting/posting)

181

and canter (sitting/light seat), which was modeled nested within gait. Horse level (young

182

horse training/basic/medium/advanced level; also see Eisersiö et al., 2015) was also included.

183

(Data from long reins were not included in modeling because it was early on found that the

184

rein tension was lower, and hence the model could be simplified.) Two-way interactions

185

between fixed effect variables were tested, after main effects had been reduced to P<0.05.

186

Two way interactions were kept at P<0.001. Random effects were horse-side (basically

187

including rein, and why rein was forced was to be able to evaluate fixed effect interactions

188

with rein), rider, horse and trial within horse, the choice of random effects guided by the

189

Akaike criterion. Models were reduced based on the type III sums of squares. Using this

190

reduced model left/right rein was exchanged for dominant rein and Akaikes criterion used to

191

verify whether the model fit the data better. The correlation structure was variance component

192

(VC) (compound symmetry (CS) was originally strived for, but either VC was deemed

193

superior based on the Akaike criterion or the CS models did not converge). For the variation

194

levels the percent of the variation contributed was estimated, dividing by the sum of all

195

sources of variation. In general the results from the median model have been presented in full,

196

while results from the main effects model has been presented for all 5 models. To study the

197

effect of the left/right rein, additional models were developed on all data, but where the

198

random effect was changed to horse within rider, rider, horse and trial within horse. In this

199

model the a priori fixed effects were rider position combined with gait, rein and their

200

interaction. Pairwise comparisons were done where interactions were involved, where P-

201

values <0.0001 were deemed significant, and 0.0001
202

(either P-values are provided or the distinction is made). Within interactions only selected

203

comparisons, deemed as useful from an equestrian perspective, were evaluated, e.g. if gait

AC C

EP

TE D

M AN U

SC

RI PT

170

ACCEPTED MANUSCRIPT was included in an interaction all comparisons were made within gait (but in this case

205

including comparisons between left and right lead canter) and changing one of the other

206

variables at a time. PROC MIXED (SAS Institute Inc., Cary, NC, 27513, USA) was used for

207

modeling. Relevant pair-wise comparisons were defined within gait, and with only one other

208

variable changing, with the exception of comparisons of categories within the combination of

209

gait and rider position in the main models.

RI PT

204

210

Results

212

Descriptive results

213

The descriptive statistics demonstrate substantial variation between gaits, rider position within

214

gait and between riders and horses (Table 1, supplementary information 1-2). Magnitude

215

differences between turns and lateral movements within gait were more difficult to confirm

216

descriptively (supplementary information 1). Note that not all horses performed all lateral

217

movements, and rather few performed lengthening (n=9) and collection (n=6). From

218

supplementary information 2 we note that data were missing from 2 categories because of

219

limited problems with the left rein tension meters.

220

M AN U

SC

211

Main effect results from all models

222

There were 1188 observations in the dataset (316 from walk, 505 from trot, 180 from left lead

223

canter and 187 from right lead canter (descriptive statistics see supplementary information 3).

224

‘Low’ and ‘auc’ rein tension were modelled as logarithm-transformed while the others as

225

square-root transformed. The main effects (Figure 2, Table 2) show that the major

226

determinants found for amount of rein tension, deemed both from statistical significances and

227

differences judged as substantial descriptively were gait; walk
228

position in the saddle had a large influence both at trot (posting
229

seat
230

right lead canter within light seat or sitting position. However comparing between positions

231

and gaits, light seat right lead canter had a slightly higher rein tension, compared to sitting

232

trot, which is reflected by light seat right lead canter and sitting trot being insignificant in all

233

but the ‘auc’ comparison, while the light seat left lead canter was significantly lower than

234

sitting trot in the ‘median’ (P=0.0002) and ‘low’ model (P<0.0001). The magnitude of rein

235

tension for walk and posting trot was close and low for all outcomes; however posting trot

236

was significantly associated with a slightly higher rein tension in all but the ‘range’ model.

237

AC C

EP

TE D

221

ACCEPTED MANUSCRIPT Horse level

239

Horse level was significant in all 5 models (P<0.007) (Figure 2). Advanced horses (n=4

240

horses at 2 riders) had the highest rein tension value in all but the ‘low’ model, in the same

241

models followed by the young horses (n=3 horses at 3 riders). In the ‘low’ model the medium

242

horses had the highest estimate while the statistical differences were actually found between

243

groups with similar point estimates, demonstrating the diverse within-category variation for

244

this variable.

RI PT

238

245

Variation from horses and riders

247

Riders and horses contributed the following proportion of the variance in the respective

248

models; in the ‘median’ model (24/13%), the ‘low’ model (33/2%), the ‘high’ model (0/35%),

249

the’ range’ model (0/28%) and the ‘auc’ model (0/5%). The horse-side effect (the rein effect)

250

was ≤12% in all but the ‘low’ model where it was 23% (suggesting a larger variation by rein

251

in the ‘low’ tension).

M AN U

SC

246

252

Ridden exercises

254

There were no significant main effects in the median model for the ridden exercises corners

255

and turns. However, significantly more rein tension was used when riding corners to the left

256

compared to corners to the right across all other models except the ‘low’ and ‘auc’ models,

257

reflecting a similar lowest left rein tension value. In contrast, turning was borderline

258

associated with a lower rein tension than not turning in the ‘low’, ‘high’ and ‘range’ models.

259

The ’auc’ model (Figure 2), where the outcome reflects the time the rider used for each

260

exercise as well the tension applied demonstrated a number of similarities with the other

261

models and all included effects, but rein, were significant (P≤0.0009).

EP

AC C

262

TE D

253

263

Lateral movements

264

In the median model half-pass to the right was associated with higher rein tension than half-

265

pass to the left and the baseline (i.e. no lateral movement, P=0.002). Half-pass to the right

266

was also associated with higher rein tension in the ‘low’ model (P=0.04), while in the ‘high’

267

the conclusion changes somewhat to that half-pass to the left is lower than the both the

268

baseline (P=0.006) and half-pass to the right (P=0.01).

269 270

Collection/lengthening

ACCEPTED MANUSCRIPT 271

For the ‘median’, ‘low’ and ‘high’ outcomes the finding was that all measures of rein tension

272

were higher at lengthening, while collection did not differ from no collection/lengthening. (In

273

the ‘range’ model collection/lengthening the group P-value was >0.05.) In the ‘auc’ model

274

the results reflected that collection and lengthening was more seldom performed, i.e. the

275

resulting impulse was highest when none of these were performed.

RI PT

276

Interaction models

278

The conclusions from the median interaction model were similar to those in the main effects

279

model (supplementary information 4-6). Another interaction results were that (sitting) right

280

lead canter in half-pass to the right (LSMEAN 32.9 N) differed from (sitting) right lead canter

281

without lateral movements (LSMEAN 27.8 N, P=0.004). From supplementary information 6

282

it is seen that the inner rein had a significantly higher tension than the outer rein in

283

sitting/light seat right lead canter and light seat left lead canter. The ’high’ and ’range’ models

284

had no interactions, while the ’low’ and ‘auc’ model had 2 and one interaction (data not

285

shown).

M AN U

SC

277

286

Rein

288

In the models dedicated to look at if the left and right rein differed, the interaction term

289

between rider’s position within gait and rein was significant in 3 models; ‘median’

290

(P<0.0001), ‘low’ (P=0.0008) and ‘range’ (P=0.04). In the ‘median’ model there were 4

291

relevant comparisons with P<0.05. In sitting trot the right rein had a slightly higher tension

292

[left/right rein 20/21 N, P=0.03]. In canter if we change nomenclature to inside/outside rein

293

we find the inside rein associated with the highest tension (one horse was ridden in counter-

294

canter, Supplementary information 1). In light seat left lead canter the inside rein was higher

295

[left/right 20/15 N, P<0.0001], and in both sitting [left/right 22/27 N, P<0.0001] and light seat

296

[left/right 17/21 N, P=0.005] right lead canter the right inner rein was higher.

EP

AC C

297

TE D

287

298

In the ‘low’ model there were 3 relevant comparisons with P<0.05. In walk the right rein had

299

a lower tension [left/right rein 5.9/5.4 N, P=0.02], in light seat left lead canter the inside rein

300

was higher [left/right 7.6/5.7 N, P=0.002], and in sitting right lead canter the inside rein was

301

higher [left/right 8.4/9.8 N, P=0.01]. In the ‘range’ model 4 relevant comparisons had P<0.05

302

and in all these the right rein was associated with the highest tension; walk [left/right rein

303

40/43 N, P=0.03], sitting trot [left/right rein 54/61 N, P=0.001], posting trot [left/right rein

ACCEPTED MANUSCRIPT 304

40/46 N, P=0.0008] and sitting left lead canter [left/right rein 65/79 N, P<0.0001], where in

305

the latter comparison the outside rein was higher.

306

In the ‘high’ and ‘auc’ models the 2 main effects, rider position within gait and rein, were

308

significant, but not their interaction. The right rein was in both these models higher than the

309

left (‘high’ [left/right rein 41/45 N, P<0.0001], ‘auc’ [left/right rein 499/550 Ns, P=0.03].

RI PT

307

310 311

Discussion

313

This study is the first to our knowledge with the aim to document the distribution of rein

314

tension during entire riding sessions. Earlier studies have mainly recorded shorter sequences

315

of rein tension during predetermined riding exercises. By letting our participating riders

316

structure their riding sessions themselves during data collection we received rein tension data

317

that likely reflect the normal situation for each horse-rider combination in terms of magnitude

318

and in relation to exercises performed.

319

M AN U

SC

312

When studying the descriptive statistics of rein tension applied by our study participants we

321

found substantial variation in magnitude (Supplementary information 1) and from the models

322

riders contributed a larger part of the variation than the horses in the ‘median’ and ‘low’

323

models, but interestingly rather low to very low proportions in the other 3 models. This

324

suggest that horses, compared to riders, contribute rather much to the found high or peak

325

tension, but less to the maintained lower tension or the total tension/impulse (‘auc’) during the

326

session. Although the participants in our study were of diverse background and educational

327

level we found certain reoccurring patterns and strong determinants with regards to amount

328

and distribution of rein tension for the variables gait, rider position in the saddle, ridden

329

exercise performed and educational level. We thus suggest that rein tension data benefit from

330

being categorised according to similar variables in studies of rein tension in relation to other

331

variables.

AC C

EP

TE D

320

332 333

Our most pronounced results were the strong connection between magnitude of rein tension

334

and gait (walk
335

Similar results for association with gait have been found by Clayton et al. (2003) and Kuhnke

336

et al. (2010). Conversely, the fact that rider position played a large part in amount of rein

337

tension used was more of a surprise. It seems that this is a factor that merits consideration

ACCEPTED MANUSCRIPT during rein tension studies. Perhaps this result is connected to the large vertical and horizontal

339

accelerations and decelerations of the horse’s trunk at the trot and canter and the rider’s ability

340

to adjust and adapt to these movements (Byström et al., 2009). Perhaps some riders support

341

their seat through use of the reins while sitting? Or the higher values of rein tension at sitting

342

trot and canter, compared to posting or light seat, may also reflect posting and light seat being

343

used during warm-up and suppling work, with less demands on the horse, while sitting to the

344

gait might be used in exercises striving to collect and ‘work’ the horse. However, it is not

345

fully in principle with riding theory that a horse that is more worked should do so with higher

346

rein tension. Though targeting the variation between and during diverse ridden exercises; we

347

selected measures associated with the most common rein tension values, instead of the

348

extreme values minimum and maximum. The maximum rein tension values were not analyzed

349

as these maximum peaks are rare and may yield an inaccurate picture of the rein tension the

350

horse has been subjected to. In addition these peaks likely don’t reflect the amount of rein

351

tension normally used by the rider or horse. The maximum rein tension peaks may also often

352

occur as the horse stumbles, shies, coughs or pulls the reins out of the rider’s hands. Even

353

though these events can be interesting to study they are not within the scope of this study. The

354

differences between the minimum value and the 2 percentile as well as the 98 percentile and

355

the maximum rein tension values per gait can be seen in Table 1. We suggest that the 98

356

percentile should be reported along with the maximum value in rein tension studies. The ‘low’

357

variable was quite similar for all categories of data ranging from 6.9 N-12.1 N, median 9.0 N,

358

(Figure 2) and may perhaps be interpreted as the contact on the reins (Anonymous, 2004).

359

This ‘low’ category also resembles the mean rein tension values found in other rein tension

360

studies where data were collected from shorter and more predetermined ridden exercises

361

(Warren-Smith et al., 2007). The ‘high’ variable on the other hand had a larger point estimate

362

range of 34-63 N, median 47 N, indicating that certain variables were connected to a higher

363

variability of rein tension than others, as also seen from the ‘range’ variable (Figure 2). For

364

example, sitting trot and sitting canter (left/right) had an equal magnitude of rein tension in

365

the ‘low’ category, but differed approximately 10 N in the ‘high’ category, suggesting that

366

rein tension vary with higher magnitudes in the canter compared to the trot.

AC C

EP

TE D

M AN U

SC

RI PT

338

367 368

Interestingly, an equal amount of rein tension was used for turning to the left and right

369

respectively (left/right rein combined) while significantly less rein tension was used when

370

riding corners to the right compared to corners to the left (Table 2, except for in the ‘auc’

371

model see discussion further down). This raises the question of laterality in horses and riders

ACCEPTED MANUSCRIPT and how it affects certain ridden exercises. In the current study a significant effect of

373

perceived laterality was not found (data not shown). Yet, ridden exercises on straight and bent

374

lines, including lateral movements, studying the rein tension used on the inside and outside

375

rein merits further investigation. Many left/right rein comparisons were significant using the

376

model that was dedicated to this. In the other models, except in the ‘range’ model, we found

377

that the rein effect was absorbed by the random effect. The right rein most often had higher

378

rein tension than the left rein, with the exception of 3 comparisons, of which 2 were from the

379

‘low’ model (the right rein had a lower tension in walk and the inside rein was higher in left

380

lead canter), suggesting that the right hand in general is associated with higher tension but

381

perhaps also somewhat associated with more release (the third one was the left rein in left

382

lead canter). This might reflect the principle that ‘every asking rein aid must be followed by a

383

yielding aid’ (Anonymous, 2002). At canter more rein tension was placed on the inside rein

384

compared to the outside rein, with the exception that range was higher on the outside rein in

385

sitting left lead canter in the rein-dedicated analysis (Figure 2, Supplementary information 6).

386

These general results are partly opposite to those of Kuhnke et al. (2010) as well as

387

handbooks on riding suggests that the outside rein should keep a more continuous contact

388

with the horse’s mouth and the inside rein should act and release according to need

389

(Anonymous, 2004). These conflicting results demonstrate the complexity of rein tension

390

measurements.

SC

M AN U

TE D

391

RI PT

372

Our results with regards to the horse’s educational level are interesting and merits

393

consideration in future rein tension studies. In the light of the small number of young horses

394

included going through young horse basic training, it is still noteworthy that they were ridden

395

with almost as much rein tension as the advanced horses and more rein tension than the basic

396

level horses. This may be explained by the fact that young horses may not yet have learnt to

397

respond correctly to the rider’s pressure signals and are perhaps going against the rein tension

398

instead of yielding and alter speed, direction or head position. This doesn’t have to be a

399

problem, as long as the rider notoriously release the rein tension as the correct response is

400

given by the horse i.e. using the principles of negative reinforcement appropriately. However,

401

the median rein tension of 24 N and the ‘high’ value of 57 N is far more rein tension than the

402

naïve horses in Christensen et al. (2011) were willing to take on in order to receive a food

403

reward, comparing to mean 10 N and max 38 N, suggesting that the young horses in our study

404

had already habituated to rein tension signals to some extent.

405

AC C

EP

392

ACCEPTED MANUSCRIPT It is more contradictory to riding handbooks that basic
407

terms of magnitude of rein tension, since lightness to the rider’s signals should be further and

408

further developed as the training progresses (Decarpentry, 1949). Given that lightness is one

409

of the cornerstones of dressage riding, repeatedly emphasised both by the Fédération Equestre

410

Internationale and riding handbooks (Fédération Equestre Internationale, 2014), a light

411

contact between the rider’s hand and the horse’s mouth should be one of the key features in

412

all ridden exercises regardless of level or difficulty. Nevertheless, the magnitude of rein

413

tension between the horse’s mouth and the rider’s hand likely also depend on the rider’s

414

purpose and intention with the ride as well as the horse’s response to the rider’s requests.

RI PT

406

SC

415

Horses are generally trained to seek contact with the bit as this is one of the objectives for

417

competitive dressage and what is sought for is the horse “accepting the bridle with a light and

418

consistent soft submissive contact” (Fédération Equestre Internationale, 2014). The reins are

419

required to be taut during dressage competition and not demonstrate any slack during the ride

420

as looseness of the rein would indicate an inconsistent contact with the horse’s mouth. While

421

the FEI advocates a continuous contact between the rider’s hand and the horse’s mouth and it

422

is described by Clayton et al. (2011) that this is done by the horse pushing against the bit to

423

some extent, it is, however, evident that rein tension always demonstrate an uneven contact as

424

the ‘range’ values (the difference between the 2-98 percentile) range from 40 N at the walk to

425

70 N at sitting right lead canter (Table 2), and this feature is further demonstrated by studies

426

on rein tension in relation to the horse’s stride cycle (Clayton et al., 2003; 2011; Eisersiö et

427

al., 2013; Egenvall et al., in press).

EP

TE D

M AN U

416

428

Correspondingly, a low rein tension and a small rein tension variance were important features

430

in Borstel and Glißman (2014) for receiving high rideability scores (i.e. the measure of how

431

comfortable it feels to ride a certain horse). They found that the lower and more steady the

432

rein tension the higher the rideability score given by the judges. In addition, variation of rein

433

tension is of great interest as pressure signals are used to communicate requests to the horse

434

and it is not yet elucidated from rein tension data what the rein tension peaks represent in

435

reality. We suggest that the resulting rein tension data seen in the data set is derived from a

436

combination of the oscillating movement of the horse’s gait, the rider’s ability to follow the

437

horse’s movement, the signaling actions of the rider’s hand and the horse’s interaction with

438

the bit. None of these factors have hitherto been well elucidated relative to rein tension in

439

riding.

AC C

429

ACCEPTED MANUSCRIPT 440

The ‘auc’ statistic produced measures that are related to how long a horse was ridden in a

442

specific exercise multiplied by the magnitude of rein tension. Because of this posting trot was

443

a determinant of high importance, because posting trot was simply very commonly performed

444

(Eisersiö et al., 2015). In lateral movements and collection/lengthening the baselines all

445

yielded high values because of the same reason. The high ‘auc’ in the advanced horses

446

reflected that they were ridden for a slightly longer time, and at gaits of higher speed,

447

compared to the other horses. Turning left had a higher value than turning right, while other

448

statistics for left/right turn showed miniscule differences, a finding that may need more fine-

449

tuned methods for elucidation. The ‘auc’ reflects the pressure from the whole riding session,

450

and if the horse eg. has a bar ulcer, a large ‘auc’ may imply a sustained discomfort for the

451

horse. Pressure sores arise from a combination of the intensity and duration of the pressure

452

applied (Chang and Seireg 1999), though specific related to the mouth of the horse has not

453

been investigated.

M AN U

SC

RI PT

441

454

The rein tension meter used in our study relied on strain gauge technique for generating rein

456

tension data (Eisersiö, 2013). Previous studies have also used strain gauge transducers for rein

457

tension measurements with reliable results (Clayton et al., 2003, 2005; Heleski et al., 2009;

458

Clayton et al., 2011). An advantage with this rein tension meter was its broad measuring

459

range of 0-500 N and resolution of 0.11 N, making sure that no peaks of tension were cut of

460

as has been a problem in previous studies with other rein tension meters (Egenvall et al.,

461

2012; Christensen et al., 2014) while still catching the small variations of rein tension that

462

occur.

EP

463

TE D

455

Currently the field of research on equine welfare with focus on horse training and rein tension

465

advocate a rigorous use of the principles of learning theory during horseback riding. Another

466

important aspect worth investigating and discussing when carrying out research on rein

467

tension is the horse’s physical and perceptual experience of the pressures applied in its mouth.

468

The rein tension meter fastened on the reins solely measure the tension applied on the leather.

469

How this tension is distributed along the bit in the horse’s mouth is unknown. The pressure

470

from the bit the horse feels in its mouth is likely affected by the shape and size of the bit, how

471

the bit and bridle is fitted on the horse, the head and neck position of the horse and how the

472

horse chooses to carry the bit in its mouth and, in particular, the anatomy of the horse’s oral

AC C

464

ACCEPTED MANUSCRIPT 473

cavity. Besides the rein tension applied on the reins by the horse and/or rider, all of these

474

variables contribute to how the pressure is applied in the horse’s mouth.

475

To document and analyze rein tension further is important both from in terms of equine

477

performance and welfare. It is highly likely that if novices can be instructed in how to use the

478

reins more explicitly, they can faster develop a more aware and precise rein handling that will

479

increase their riding abilities at a faster rate and lead to enhanced horse welfare. On

480

prerequisite for optimal usage of the reins will be that the hand is as far as possible

481

independent of the seat (Anonymous, 2004).

SC

482

RI PT

476

The detailed approach taken, including categorisation of e.g. collection and lengthening,

484

prohibited extensive and detailed data analysis of a much large sample. The limited number of

485

riders and horses leads to some problems with extrapolation, though riders were from

486

different stables without obvious problems with dependence or clustering. Our perception

487

from the study is that they produce a reasonable subsection of the targeted population,

488

however this cannot be verified. Ideally we would have wanted more participants in our study

489

and more days on each horse-rider combination as rein tension is likely affected by both the

490

rider’s and the horse’s physical and emotional state of the day, i.e. targeting also between-

491

session variability. In the design it was included speed measurements using a GPS, but data

492

collection from this unfortunately failed to a large degree (for several reasons). While our

493

study design had the advantage of capturing the ordinary situation for horse and rider in terms

494

of duration and order of exercises performed, it also meant that we didn’t receive data for all

495

exercises in all horses. A potentially more severe problem is that one evaluator classified all

496

the exercises (Eisersiö et al., 2015). For example whether a horse is in collection or not can be

497

debated. On the other hand, riding with a rein tension meter will likely always produce a

498

psychological effect on the rider, especially when done in front of a video camera. Having

499

‘good hands’ is an important feature in riding and it is highly probable that riders modify their

500

riding to some extent in order to not just have their horse perform well but to also ‘look good’

501

during data collection. We hope this effect of ‘showing off’ was limited by the data collection

502

capturing the riders’ normal training schedule. We have included descriptive statistics based

503

on the raw data as well as on a series of statistics in the multivariable analysis. This has

504

produced a somewhat different picture with regard to the descriptive statistics in some

505

instances, especially when rein tension values were close such as for left/right corners, where

506

it can be seen that the 98 percentiles of Supplementary information 1 and the results from

AC C

EP

TE D

M AN U

483

ACCEPTED MANUSCRIPT 507

Figure 2 are somewhat contradictory. As we demonstrate variation related to 2 levels of

508

variation this could not be avoided.

509

Conclusion

511

The assessment and interpretation of the rein tension signal is complex and rein tension data

512

thus need to be addressed and scrutinized in detail on several levels. Our results suggest that

513

variables to consider in rein tension studies are the gait of travel, the rider’s position in the

514

saddle, the ridden exercise performed, the educational level of horse, the rider and horse per

515

se and to some extent the left/right rein. We also suggest that the range of rein tension within

516

gait and exercise, as well as reporting the 98 percentile of data, are important features.

517

Studying e.g. more of the within-stride variation, and between-stride variation of the rein

518

tension data are next steps in receiving a more complete picture of the events taking place

519

during riding.

M AN U

SC

RI PT

510

520

Acknowledgement

522

The study was funded by The Swedish Research Council Formas. The funding body had no

523

influence on the study design, in the collection, analysis and interpretation of data; in the

524

writing of the manuscript; or in the decision to submit the manuscript for publication.We

525

thank the riders for their contributions.

526

TE D

521

Conflict of interest

528

None of the authors have any potential conflicts of interest, including any financial, personal,

529

or other relationships with other people or organisations, within 3 years of beginning the

530

submitted work that could inappropriately influence, or be perceived to influence, the work.

AC C

531

EP

527

532

Authorship statement

533

The idea for the paper was conceived by AE, ME, LR and MR. The experiments were

534

designed by AE and ME. The experiments were performed by ME. The data were analyzed

535

by AE and ME. The paper was written by ME, AE, MR and LR.

536 537 538 539 540

ACCEPTED MANUSCRIPT References

542

Anonymous. 1997. The principles of riding: the official instruction handbook of the German

543

National Equestrian Federation. Kenilworth Press, Shrewsbury, UK, pp 74, 85, 156-157.

544

von Borstel, U.K., Glißman, C. 2014. Alternatives to Conventional Evaluation of Rideability

545

in Horse Performance Tests: Suitability of Rein Tension and Behavioural Parameters, PLOS

546

ONE 1, 1-9.

RI PT

541

547

Byström, A., Rhodin, M., Von Peinen, K., Weishaupt, M.A., Roepstorff, L., 2009. Basic

549

kinematics of the saddle and rider in high-level dressage horses trotting on a treadmill. Eq.

550

Vet. J. 41, 280–284.

SC

548

M AN U

551 552

Chang, W.L., Seireg, A.A., 1999. Prediction of ulcer formation on the skin. Med. Hypotheses

553

53, 141-144.

554 555

Christensen, J.W., Zharkikh, T.L., Antoine, A., Malmkvist, J., 2011. Rein tension acceptance

556

in young horses in a voluntary test situation. Eq. Vet. J. 43, 223-228.

TE D

557 558

Christensen, J.W., Beekmans, M., van Dalum, M., Van Dierendonck, M., 2014. Effects of

559

hyperflexion on acute stress responses in ridden dressage horses. Physiol. Behav. 128, 39–45.

EP

560

Clayton, H.M., Singleton, W.H., Lanovaz, J.L., Cloud, G.L. 2003. Measurement of rein

562

tension during horseback riding using strain gage transducers. Exp. Tech. 27, 34-36.

563

AC C

561

564

Clayton, H.M., Singleton, W.H., Lanovaz, J.L., Cloud, G.L., 2005. Strain gauge measurement

565

of rein tension during riding: a pilot study. Eq. Comp. Ex. Phys. 2, 203–205.

566 567

Clayton, H.M., Larson, B., Kaiser, L.J., Lavagnino, M.m 2011. Length and elasticity of side

568

reins affect rein tension at trot. Vet. J. 188, 291–294.

569 570

Decarpentry, A., 1949. Academic equitation. A system based on the methods of D’Aure,

571

Baucher and L’Hotte (Équitation academique, translated by Nicole Bartle). Robert Hale Ltd,

572

London, UK, p 45.

ACCEPTED MANUSCRIPT 573 574

Eisersiö, M. 2013. How to build a rein tension meter. Degree project in Biology, Swedish

575

University of Agricultural Science, http://stud.epsilon.slu.se

576

Eisersiö, M., Roepstorff, L., Weishaupt, M.A., Egenvall, A. 2013. Movements of the horse’s

578

mouth in relation to horse–rider kinematic variables. Vet. J. 198, e33–e38

RI PT

577

579 580

Eisersiö, M., Roepstorff, L., Rhodin, M., Egenvall, A., 2015. A snapshot of the training

581

schedule for 8 professional riders riding dressage. Comp. Ex. Phys. 11, 35-46.

SC

582

Egenvall, A., Eisersiö, M., Roepstorff, L., 2012. Pilot study of behavior responses in young

584

riding horses using 2 methods of making transitions from trot to walk. J. Vet. Behav. 7, 157-

585

168.

586

M AN U

583

587

Egenvall, A., Eisersiö, M., Rhodin, M., van Weeren, R., Roepstorff, L., Rein tension during

588

canter. Comp. Ex. Phys. (in press)

589

Fédération

Equestre

Internationale.

2014.

Dressage

591

http://www.fei.org/fei/regulations/dressage Page visited 2014-09-16.

Rules

25th

edition.

TE D

590

592

de la Guérinière, F.T. 1994. School of Horsemanship (École de Cavalerie 1733, translated by

594

Tracy Boucher). Allen, London, UK, p. 91

EP

593

595

Heleski. C.R., McGreevy, P.D., Kaiser, L.J., Lavagnino, M., Tans, E., Bello, N., Clayton,

597

H.M., 2009. Effects on behaviour and rein tension on horses ridden with or without

598

martingales and rein inserts. Vet. J. 181, 56–62.

599

AC C

596

600

Kuhnke, S., Dumbell, L., Gauly, M., Johnson, J.L., McDonald, K., von Borstel, U.K., 2010. A

601

comparison of rein tension of the rider’s dominant and non-dominant hand and the influence

602

of the horse’s laterality. Comp. Ex. Phys. 7, 57–63.

603 604

Ludewig, A.K., Gauly, M., Von Borstel, U.K., 2013. Effect of shortened reins on rein tension,

605

stress and discomfort behavior in dressage horses. Abstracts / Journal of Veterinary Behavior

606

8, e15-e16

ACCEPTED MANUSCRIPT 607 608

Manfredi, J.M., Rosenstein, D., Lanovaz, J.L., Nauwelaerts, S., Clayton, H.M., 2010.

609

Fluoroscopic study of oral behaviours in response to the presence of a bit and the effects of

610

rein tension. Comp. Ex. Physiol. 6, 143–148.

611

Tell, A., Egenvall, A., Lundström, T., Wattle, O., 2008. The prevalence of oral ulceration in

613

Swedish horses when ridden with bridle and bit. Vet. J. 178, 405-410.

RI PT

612

614

Warren-Smith, A. K., Curtis, R. A., Greetham, L., McGreevy, P. D., 2007. Rein contact

616

between horse and handler during specific equitation movements. Appl. Anim. Beh. Sci. 108,

617

157-169.

SC

615

M AN U

618 619

AC C

EP

TE D

620

ACCEPTED MANUSCRIPT

Table 1. Descriptive statistics for rein tension (N) in all gaits for the left and right rein separating sitting from posting and light seat in trot and canter. The statistics are derived from horse- based means from raw (calibrated) data (means, STDs etc are produced as means of means, STDs etc).

Rein tension (N)

All gaits

24 24

Walk long reins

24

Walk short reins

21

Trot sitting

24

Trot posting

18

Left lead canter sitting

13

Left lead canter light seat

18

633 634 635 636 637 638 639 640

AC C

629 630 631 632

left

19

15

0

1

16

60

130

right

21

17

0

1

17

67

144

left

4

5

0

0

2

21

51

right

4

5

0

0

2

20

56

left

14

11

0

1

12

46

112

right

15

13

0

1

12

52

123

left

20

14

0

2

17

58

102

right

23

17

0

2

19

68

120

left

16

11

0

1

14

46

107

right

17

left

25

right

28

left

19

right

19

left

25

right

28

12

left

right

EP

Right lead canter light seat

Max

13

0

1

14

51

130

18

0

2

21

72

139

21

0

2

23

83

160

14

0

2

16

55

95

16

1

2

15

62

94

20

0

2

20

80

158

21

0

2

24

82

166

17

14

0

1

13

57

99

20

16

0

1

17

65

108

TE D

Right lead canter sitting

Rein Mean STD Min 2 perc Median 98 perc

SC

n

M AN U

Gait

RI PT

621 622 623 624 625 626 627 628

ACCEPTED MANUSCRIPT

c

AUC

644

S S

ns

S S BL S S S

S ns ns S S

S S ns ns

ns S S

S BL

ns

b

c

SC

RI PT

ns S S

BL S S S S S

S S S S S

ns ns S S

ns S S

BL S

ns

ns S S S S S

S S S S S

ns ns S S

ns BL S

BL S

ns

S S S S ns BL

S S S BL ns

S BL BL S

ns S S

S S

BL

L=left, R=right, AUC=area under curve

R canter sitting

BL ns BL S

L canter sitting

R canter light seat

S BL S S S

EP

AC C

Range

a

S S S S S S

TE D

High

L canter light seat

Low

Sitting trot

Walk Posting trot Sitting trot La canter light seat Rb canter light seat L canter sitting R canter sitting Walk Posting trot Sitting trot L canter light seat R canter light seat L canter sitting R canter sitting Walk Posting trot Sitting trot L canter light seat R canter light seat L canter sitting R canter sitting Walk Posting trot Sitting trot L canter light seat R canter light seat L canter sitting R canter sitting Walk Posting trot Sitting trot L canter light seat R canter light seat L canter sitting R canter sitting

M AN U

Model Median

Posting trot

Table 2. Showing the statistical significances of the comparisons for rider position within gait for the five main effects models (effect magnitude see Figure 2), S denotes P<0.0001, borderline (BL) 0.0001
Walk

641 642 643

ACCEPTED MANUSCRIPT 645

Figure legends and captions:

646

Figure 1. The rein tension meter used in the study.

647

Figure 2. Rein tension (N) evaluated from main effects models. The unit on the y-axes of the

649

red bar graphs s is N, while on the graph with green bars (area under curve (AUC)) is Ns.

650

Arrows indicate statistical differences, if denoted by ‘s’ P<0.0001 otherwise the comparisons

651

are at 0.0001
RI PT

648

652

Supplementary information 1. Rein tension (N) in all gaits comparing left/right (L/R) rein and

654

riding on one track (straight and turns) with lateral movements. The statistics are derived from

655

horse-based means from raw (calibrated) data.

656

SC

653

Supplementary information 2. Rein tension (N) for each horse and rider. Walk is on short

658

reins only. Trot is sitting and posting trot combined. Canter is left/right lead and sitting and

659

hunt seat combined. Rider-based statistics are derived from horse- based means from raw

660

(calibrated) data.

M AN U

657

661

Supplementary information 3. Descriptive statistics on the outcome statistics used in the

663

multivariable analyzes. Number of observations is derived from the number of horses

664

performing various exercises (turns, corners, lateral movements, collection, lengthening) with

665

a category.

TE D

662

EP

666

Supplementary information 4. Rein tension (N) demonstrated by lateral movements in sitting

668

trot. Blue bars indicate significant differences at 0.0001
669

The conclusions relative to posting trot have been omitted because all comparisons with

670

P<0.05 within posting trot involved half-pass to left in posting trot.

671

AC C

667

672

Supplementary information 5. Rein tension (N) in light seat and sitting canter (left/right)

673

combined with riding in lengthening or collection. Blue bars indicate significant differences at

674

0.0001
675 676

Supplementary information 6. Rein tension (N) in light seat and sitting canter (left/right)

677

combined with left/right rein. Blue bars indicate significant differences at 0.0001
678

black bars P<0.0001.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

Rein tension increases from walk, over trot and to canter Posting and light seat is associated with lower rein tension compared to sitting

AC C

EP

TE D

M AN U

SC

Lengthening is associated with high rein tension

RI PT

Rein tension varies considerably with rider and horse, slightly more with rider

ACCEPTED MANUSCRIPT

Supplementary information 1. Rein tension (N) in all gaits comparing left/right (L/R) rein and riding on one track (straight and turns) with lateral movements. The statistics are derived from horse-based means from raw (calibrated) data. Gait

Corner left

Rein

L

R

L

R

L

R

median P2, P98

16 2, 59

18 2, 66

11 2, 44

11 1, 48

15 2, 53

16 2, 60

n

24

24

24

24

24

24

median

15 4, 43 21

13 3, 44 21

13 2, 47 16

14 2, 52 16

16 4, 42 19

16 3, 48 19

R

L

R

20 2, 66

21 2, 78

18 2, 75

22 2, 77

21

23

22

24

23 4, 67 8

21 5, 73 8

10

8,16 1

3 1, 5 1

P2, P98 n

29 3, 86 19

32 4, 110 19

14 3, 44 14

16 3, 56 14

17 3, 52 17

19 3, 61 17

66 4, 184 1

70 8, 254 1

20 3, 65 13

25 3, 69 14

median

15

17

12

12

14

15

1,50 24

1,57 24

20 2, 78 23

17

1,49 24

19 2, 67 21

21 5, 66 8

P2, P98 n

2,54 24

2,63 24

1,44 24

median

16 2, 59

18 2,61

13 2, 48

13 1, 53

14 2, 53

15 1, 57

19 3, 61

23 4, 59

1,74

23 2, 77

24

24

24

24

24

24

7

8

22

24

15 4, 45

18 4, 52

13 3, 42

13 3, 45

17 6, 49

23 6, 60

-

-

-

-

-

-

-

-

12

14

6

7

9

10

-

-

-

-

19 4, 59

13 3, 38

13 2, 39

21 5, 58

25 5, 83

17 3, 61

17 4, 55

-

-

P2, P98

17 4, 52

-

-

n

14

15

8

9

11

11

3

3

-

-

22 5, 61

10 1, 35

10 2, 37

25 8, 54

26 9, 62

28 5, 84

30 4, 83

-

-

P2, P98

21 5, 57

-

-

n

13

14

6

6

8

8

8

8

-

-

median

20

12 2, 42

23 4, 61

30 5, 79

-

-

P2, P98

13 2, 44

-

-

24 3, 80

34 5, 111

8

8

-

-

P2, P98 n

Turn left

Turn right

median

P2, P98 n Shoulder-in left

median P2, P98 n

EP

median

AC C

Shoulder-in right

TE D

Corner right

Half-pass left

Half-pass right

Leg-yield left

Canter right lead

L

median

SC

Straight

Canter left lead

Trot

M AN U

Ridden exercise

Walk

RI PT

All gaits

4,58 7 18

3,62

25 4, 77

n

12

12

5

5

5

5

median

15 4,44

12 3, 40

15 2, 42

17 4, 48

18 4, 52

-

P2, P98

16 3, 47

-

-

-

-

14

15

5

6

13

14

-

-

-

-

-

ACCEPTED MANUSCRIPT

Leg-yield right

median P2, P98

20 4, 62

20 4, 52

22 4, 61

18 3, 51

19 3, 63

-

-

-

-

-

-

-

13

14

6

7

12

13

-

-

-

-

EP

TE D

M AN U

SC

RI PT

No data on exercise

AC C

-

19 4, 52

-

ACCEPTED MANUSCRIPT

Supplementary information 2. Rein tension (N) for each horse and rider. Walk is on short reins only. Trot is sitting and posting trot combined. Canter is left/right lead and sitting and hunt seat combined. Rider-based statistics are derived from horse- based means from raw (calibrated) data.

Horse 1

mean Rein (STD) left 11 (11) walk right 12 (11)

Rider Gait

walk 2

trot canter walk

3

trot canter walk

4

trot

walk 5

trot

canter walk 6 trot

34

14

(13)

range 298 perc 0

11 (10) 1 9 (11) 0

37 35

14 (13) 13 (13)

right 12 (11) 1 left 15 (17) 0

41 64

9

(11) 1

33

15 (13)

1

right 18 (17) 1 left 13 (10) 1

67

8 (10) 0 13 (12) 1

39 45

24 (27) 24 (25)

0 1

44

15 (11)

1 44

11 (8)

1

right 10 (11) 0 left 17 (12) 2 right 15 (13) 0

42

13 (10) 17 (12)

0 39 1 48

6 (8) 13 (11)

15 (11)

0 44

18 (14) 2

56

22 (13)

right 18 (14) 0 left 11 (9) 1 right 9 (8) 1

54

21 (13)

35 30

13 (10) 1

39

right 10 (9) 1 left 15 (14) 0

10

(8)

0

36 38 32

50

13 (10)

1

40

98 96

14 (14) 16 (16)

0 1

55 62

32

13 (12)

2

54

0 1

28 44

11 (15) 20 (14)

0 4

59 61

10 (11)

0

40

19 (17)

0

64

3 56

14 (14)

1

52

19 (16)

3

64

1 54

13 (13)

0

48

26 (18)

0

71

14 (11)

1 42

9

(9)

0

36

10 (7)

1

27

12 (10)

2 38

8

(8)

1

31

7

(5)

1

21

15 (13)

1 51

10 (9)

0

34

13 (8)

1

33

35

13 (11)

2 45

9

(9)

1

35

9

2

25

54

20 (19)

1 75

8

(9)

0

34

18 (13)

1

51

right 13 (11) 1 left 18 (13) 2

44

17 (16)

2 60

9

(9)

1

36

12 (9)

1

35

51

right 22 (16) 1 left 21 (13) 3 right 25 (14) 4

66

14 (10) 14 (11)

2 41 1 45

22 (15) 25 (16)

2 1

60 71

19 (14) 27 (20)

1 1

52 82

53 61

18 (10)

4 45

24 (15)

3

59

20 (13)

2

55

26 (15) 5

65

19 (12) 23 (12)

4 51 6 56

26 (15) 29 (18)

3 4

63 74

31 (16) -

6 -

71 -

left

left

12 9

range 298 perc 1 0

left

55 51

mean (STD)

(9) (8)

51 49

1 0

57

SC

39

AC C

canter

(10) 1

mean (STD)

M AN U

canter

9

range 298 p

TE D

trot

mean (STD)

3

11 (11) 0

left

EP

1

range 298 p 0 42 1 43

2

RI PT

Rider

(6)

right 36 (19) 8 left 11 (9) 0 right 9 (9) 1

84

27 (12)

8 60

37 (20)

7

88

43 (24)

8

103

37 35

14 (11)

0 44

7

(6)

0

24

12 (11)

0

44

10 (11)

1 41

6

(6)

1

23

10 (11)

1

40

left

13 (10) 1

40

17 (12)

0 45

10 (8)

0

32

13 (10)

1

42

right

9

1

35

9

(10)

1 36

8

1

30

11 (10)

1

40

left

21 (18) 1

69

26 (20)

1 77

17 (13)

1

50

22 (20)

1

81

right 17 (16) 1 left 18 (17) 1

62

19 (18)

1 66

14 (12)

1

46

19 (19)

1

75

71

right 18 (18) 1 left 21 (17) 1 right 24 (19) 1

71 70

13 (11) 11 (10)

0 46 1 43

17 (20) 20 (20)

1 1

80 83

25 (22) 24 (24)

1 1

87 87

17 (12)

0 51

17 (15)

1

61

30 (25)

1

99

15 (13)

1 52

23 (19)

2

77

33 (27)

1

106

(9)

78

(8)

32 (27) 2

104

25 (16)

right 38 (29) 2 left 13 (8) 2

108

36

right 15 (12) 2 left 19 (13) 2

52 55

22 (16)

3 66

19 (14)

1

right 21 (17) 2 left 25 (20) 2 canter right 27 (23) 1

67

22 (16) 30 (21)

2 63 2 85

25 (23) 25 (23)

1 1

28 (22)

1 82

33 (30)

left 21 (12) 3 right 25 (18) 1 left 27 (17) 3

53

29 (18) 34 (26)

3 82 3 112

15 (9) 17 (13)

right 33 (26) 0 left 33 (20) 2

103

32 (25) 40 (34)

0 94 0 132

21 (13) 25 (18)

RI PT

ACCEPTED MANUSCRIPT

80

40 (25)

0 98

right 39 (30) 1

118

46 (33)

0 129

walk 7

trot

walk 8

trot canter

left

78 86 78 69

3 64

37 (35)

1

131

35 (30)

1

118

25 (17)

2 68

39 (34)

2

125

50 (37)

1

132

14 (9)

2 37

14 (9)

3

41

10 (7)

2

31

18 (13)

1 51

17 (16)

3

76

11 (7)

3

30

60

15 (9)

1

38

97 88

16 (10) 21 (15)

1 2

41 61

1

115

19 (15)

1

60

0 0

38 52

18 (8) 24 (17)

4 1

38 69

4 2

57 75

26 (13) 33 (25)

5 0

57 102

26 (15)

4

63

-

-

-

-

28 (23)

0

89

43 (35)

2

136

SC

canter

AC C

EP

TE D

M AN U

Cells marked with – implies no data available because of problems with rein tension meter

ACCEPTED MANUSCRIPT

Supplementary information 3. Descriptive statistics on the outcome statistics used in the multivariable analyzes. Number of observations is derived from the number of horses performing various exercises (turns, corners, lateral movements, collection, lengthening) with a category.

EP

AC C

High (N)

Range (N)

Number of observations 155 161 118 131 127 129 61 64 26 29 62 65 29 31 155 161 118 131 127 129 61 64 26 29 62 65 29 31 155 161 118 131 127 129 61 64 26 29 62 65 29 31 155 161 118 131

Median 11 11 19 22 14 15 23 23 19 14 22 25 12 16 4 4 8 9 6 6 10 8 5 3 8 8 3 4 30 32 44 50 31 33 55 57 39 37 57 66 37 38 30 32 44 50

Min 3 1 7 6 5 3 5 10 6 3 9 12 5 8 0 0 1 1 1 0 1 2 1 0 2 2 0 1 8 6 19 19 15 16 28 32 23 15 29 33 12 22 8 6 19 19

RI PT

Rein left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right left right

SC

Rider position Sitting Sitting Sitting Sitting Posting Posting Sitting Sitting Light seat Light seat Sitting Sitting Light seat Light seat Sitting Sitting Sitting Sitting Posting Posting Sitting Sitting Light seat Light seat Sitting Sitting Light seat Light seat Sitting Sitting Sitting Sitting Posting Posting Sitting Sitting Light seat Light seat Sitting Sitting Light seat Light seat Sitting Sitting Sitting Sitting

TE D

Low (N)

Gait Walk Walk Trot Trot Trot Trot Left lead canter Left lead canter Left lead canter Left lead canter Right lead canter Right lead canter Right lead canter Right lead canter Walk Walk Trot Trot Trot Trot Left lead canter Left lead canter Left lead canter Left lead canter Right lead canter Right lead canter Right lead canter Right lead canter Walk Walk Trot Trot Trot Trot Left lead canter Left lead canter Left lead canter Left lead canter Right lead canter Right lead canter Right lead canter Right lead canter Walk Walk Trot Trot

M AN U

Rein tension statistic Median (N)

Max 47 71 77 83 28 36 47 57 48 54 59 68 40 58 22 30 47 45 18 17 28 37 12 14 33 37 16 17 76 138 127 154 63 101 99 133 97 114 130 119 80 179 76 138 127 154

ACCEPTED MANUSCRIPT

31 33 55 57 39 37 57 66 37 38 137 129 215 243 157 164 265 264 203 149 269 309 171 179

15 16 28 32 23 15 29 33 12 22 39 28 104 82 72 58 84 130 88 58 123 152 51 100

RI PT

127 129 61 64 26 29 62 65 29 31 155 161 118 131 127 129 61 64 26 29 62 65 29 31

SC

left right left right left right left right left right left right left right left right left right left right left right left right

M AN U

Posting Posting Sitting Sitting Light seat Light seat Sitting Sitting Light seat Light seat Sitting Sitting Sitting Sitting Posting Posting Sitting Sitting Light seat Light seat Sitting Sitting Light seat Light seat

TE D EP AC C

Area under curve (Ns)

Trot Trot Left lead canter Left lead canter Left lead canter Left lead canter Right lead canter Right lead canter Right lead canter Right lead canter Walk Walk Trot Trot Trot Trot Left lead canter Left lead canter Left lead canter Left lead canter Right lead canter Right lead canter Right lead canter Right lead canter

63 101 99 133 97 114 130 119 80 179 470 730 777 859 317 457 559 674 478 522 592 665 352 742

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT