Fig 1.
Classification of the number, type and intensity of training sessions.
Data are numbers of training sessions stratified by training intensity, month of the year or track surface. A total of 3568 training sessions were included in this dataset from a single trainer in Australia. Data were available throughout the year. Training intensity (soft/med/hard canter; soft/med/hard gallop) was derived from calculating sextiles of the fastest furlong (200m interval) for the overall dataset. Statistics were generated by analysing proportions (percentage of group total) by chi-square (Genstat v20, VSNi, Rothampsted, UK).
Fig 2.
Speed, stride and influence of multiple training sessions on different track surfaces by sex of work rider.
a-f: Values are predicted mean ± S.E.M. for continuous data recorded by ‘Equimetre’ in a cohort of racehorses in Australia (n = 130 different racehorses, n = 1,754 different training sessions). Data were available throughout the year. Training intensity (soft/med/hard canter; soft/med/hard gallop) was derived from calculating sextiles of the fastest furlong (200m interval) for the overall dataset. The statistical model generated predicted means (± S.E.M.) with the pre-specified interaction, training intensity × rider sex (or training session, e,f) fitted last after inclusion of rider registration status and track surface as fixed effects, HorseID and rider name as random effects, since both horses and riders completed multiple sessions. All data analyses were conducted using Genstat v20 (VSNi, UK) and graphs produced using Graphpad Prism v9.0 (La Jolla, USA).
Table 1.
Descriptive data for the complete cohort of racehorses training at a single racing yard in Australia.
Fig 3.
Racehorse heart rate on different track surfaces by sex of work rider.
a,b,c,e,f: Values are predicted mean ± S.E.M. for continuous data recorded by ‘Equimetre’ in a cohort of racehorses in Australia (n = 130 different racehorses, n = 1,754 different training sessions). Data were available throughout the year. Training intensity (soft/med/hard canter; soft/med/hard gallop) was derived from calculating sextiles of the fastest furlong (200m interval) for the overall dataset. The statistical model generated predicted means (± S.E.M.) with the pre-specified interaction, training intensity × rider sex fitted last after inclusion of rider registration status and track surface as fixed effects. HorseID and rider name were included as random effects, since both horses and riders completed multiple sessions. d) describes calculation of delta HR and HR area-under-the-response-curve. All data analyses were conducted using Genstat v20 (VSNi, UK) and graphs produced using Graphpad Prism v9.0 (La Jolla, USA).
Fig 4.
Recovery of racehorse heart rate on different track surfaces by sex of work rider.
a,b,c,d: Values are predicted mean ± S.E.M. for continuous data recorded by ‘Equimetre’ in a cohort of racehorses in Australia (n = 130 different racehorses, n = 1,754 different training sessions). Data were available throughout the year. Training intensity (soft/med/hard canter; soft/med/hard gallop) was derived from calculating sextiles of the fastest furlong (200m interval) for the overall dataset. The statistical model generated predicted means (± S.E.M.) with the pre-specified interaction, training intensity × rider sex fitted last after inclusion of rider registration status and track surface as fixed effects. HorseID and rider name were included as random effects, since both horses and riders completed multiple sessions. HR area-under-the-response-curve (AUC) calculated as described in methods. All data analyses were conducted using Genstat v20 (VSNi, UK) and graphs produced using Graphpad Prism v9.0 (La Jolla, USA).