Student ID: N1343635
Evaluating Equine Compliance Time in Correlation with Stress: Insight Into Invasive and Non-Invasive measurements
Abstract
Since stress affects a horse’s behaviour, physiology, and wellbeing, accurate and ethical methods of evaluation are required. By examining the relationship between cortisol levels and thermographic eye temperature (IRT), the impact of a calming spray on compliance times, and the possibility of predicting compliance time using physiological and environmental variables, this study examined stress. Data on IRT, heart rate, cortisol levels, weight, and ambient temperature were gathered from 498 horses throughout the baseline and post-calming spray periods.
IRT was supported as a non-invasive stress indicator by a weak but statistically significant positive correlation found between it and cortisol levels (r=0.1286, p=0.0040). The calming spray reduced compliance times, with a median showing decrease from 35 to 30 seconds. Regression analysis identified heart rate (BPM) to be the only significant predictor of compliance time, while other variables showed to not have any effect on the prediction of the compliance time.
The results indicate the importance of using both invasive and non-invasive methods to conduct the research regarding stress evaluation. Non-invasive methods as IRT and heart rate are useful and statistically significant to assess stress. Invasive methods, such as cortisol blood samples are highly accurate and are recommended to be included in the future studies with the non-invasive methods to make the results more accurate. Model for prediction of compliance time should be improved by including behavioral factors and tactics to improve equine welfare as a priority for the future studies.
Key words: Calming spray, Compliance time, Cortisol, Thermographic Eye Temperature, Stress
Introduction
Stress is a critical factor that influences the behavior, physiology, and welfare of animals, including horses. In horse populations, stress can arise from a variety of stimuli, including new environments, transporting, training, and veterinary interventions (Squibb et al., 2018). Evaluating a horse’s stress level is essential to maintaining their well-being and improving their performance (Budzynska, 2014; Ijichi et al., 2023). Conventional approaches frequently depend on physiological indicators like heart rate and cortisol levels as well as behavioural observations (Aragona et al., 2024; Budzynska, 2014). Alternative indicators must be investigated, considering that these techniques may be intrusive or impacted by external factors.
A commonly used behavioral method assessing stress level in horses is measuring their compliance time, which refers to the duration an individual took to complete the task (Esch et al., 2019). Compliance time is useful in the settings where horses are exposed to novel or somewhat unpleasant stimuli (Esch et al., 2019, Squibb et al., 2018). Longer compliance time could be a sign of increased stress, given that anxiety or fear can lead to avoidance or nervousness (Henshall et al., 2022). Tasks that require horses to overcome unknown obstacles, for example, can bring out their stress reactions, offering information about both individual temperament and general welfare (Henshall et al., 2022; Ijichi et al., 2023).
To measure the stress levels in horses, both invasive and non-invasive methods are being used (Aragona et al.,, 2024). Although the accuracy and credibility of those methods are valuable, they come with serious disadvantages. Some of the procedures in an instance, blood sampling, is an invasive procedure that requires restraining the animal, which raises their stress levels and can cross the results (Aragona et al., 2024; Sikorska et al., 2023). Furthermore, the stress that comes with sampling raises ethical concerns, as well as the validity of the taken samples, since they are obtained in such conditions (Tannenbaum and Bennett, 2015).
Alternative methods that prioritise animal welfare have been popularised. One of those non-invasive techniques is thermographic eye temperature (IRT) (Johnson et al., 2011). This technique, by using infrared thermography, measures heat emission from the periorbital region, which is impacted by sympathetic nervous system activation (Aragona et al., 2024; Johnson et al., 2011). Increased blood flow to the eyes causes an increase in IRT during a stressful situation, making it a trustworthy and non-intrusive stress indicator. Furthermore, other non-invasive methods, such as heart rate or a behavioral assessment, can be conducted on ethical grounds where animals do not have to be physically restrained, minimising their stress levels (Aragona et al., 2024; Johnson et al., 2011; Redaelli et al., 2019).
It is understood that a combination of behavioural and physiological indicators is the most effective way to assess stress (Young et al,. 2012). This allows for a more thorough assessment of stress and prevents the false conclusions that could be drawn from using only one measure (Waren & Randle, 2017; Young et al,. 2012).
The horse’s overt behavioural reactions to test conditions or management techniques might not fully represent the importance of the physiological stress response (Jolivad et al.,2023 ; Squibb et al., 2018). Even in the absence of overt behavioural stress responses, horses may display notable, quick physiological reactions when confronted with challenging situations (Jolivald et al., 2023).
This study investigates the correlation between thermographic eye temperature (IRT) and cortisol levels. Based on the hypothesis that stress may cause changes in both variables, which may potentially indicate a correlation between them. Furthermore, the effect of a calming spray was evaluated under the assumption that applying the spray reduces stress, leading to a decrease in the compliance time. Lastly, the potential to predict compliance time was assessed, with the hypothesis that all variables, such as air temperature, cortisol, IRT, heart rate, and weight, collectively influence compliance time, making it measurable through multiple factors.
Methods
This study is aimed at investigating compliance time as a type of measurement of stress in horses. This experiment was conducted in two phases:
Baseline experiment
Where compliance time was conducted as horses navigated through a new obstacle (shower curtain). Data was collected on various variables, including sex, weight, heart rate (BPM), thermographic eye temperature (IRT), blood cortisol levels, and ambient air temperature.
Post calming spray experiment
The same group of horses was put through navigating the same obstacle after the area was sprayed with a substance designed to calm horses. Compliance time was measured again under those conditions, as well as all various variables.
Study Group
A total of 498 individuals participated in the study. It comprised of 249 males and females. Data was stored in two separate Excel files; one of them contained a baseline data, and the second one included post-calming spray data.
Statistical Analysis
All data analysis was performed using R 4.4.1 and RStudio, methods of calculating data were presented on Rpbus website.
Research questions
A total of 3 research questions have been asked regarding the experiment:
Is there a correlation between the variables IRT and Cortisol?
Does the application of the calming spray have an effect on compliance times, and if so, what is it?
Is it possible to predict compliance time?
Results
Correlation between Thermographic Eye Temperature and Cortisol
A Pearson product-moment correlation was used to analyse relationships between thermographic eye temperature (IRT) and blood cortisol levels. The dataset suggests that there is a weak but statistically significant positive correlation (r=0.1286, n=496, p=0.0040). The correlation coefficient demonstrates weak positive correlation between IRT and cortisol levels. While calculations indicate that this correlation is statistically significant, the strength of this relationship suggests that those two variables are minimally related. P-value in this test provides evidence that the observed correlation is candid and that it is unlikely to have happened by accident.
#Get summaries of both euine and eqsub data
> summary(equine)
>summary(eqsub)
#Create Pearson Correlation for the data
> pearson_corr <- cor(equine$irt, equine$cortisol, method = "pearson", use = "complete.obs")
> print("Pearson Rank Correlation Coefficient:")
[1] "Pearson Rank Correlation Coefficient:"
> print(pearson_corr)
[1] 0.1286011
> # Pearson correlation significance test
> pearson_test <- cor.test(equine$irt, equine$cortisol, method = "pearson")
> print("Pearson Correlation Test")
[1] "Pearson Correlation Test"
> print(pearson_test)
Pearson's product-moment correlation
data: equine$irt and equine$cortisol
t = 2.8881, df = 496, p-value = 0.004046
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
0.04119998 0.21404903
sample estimates:
cor
0.1286011 # Load ggplot library
> library(ggplot2)
#Create a scatterplot and linear model
>ggplot(equine, aes(x=irt, y=cortisol))+
+ geom_point(alpha=0.3, size=3)+
+ geom_smooth(method=lm , color="red", fill="#69b3a2", se=TRUE)+
+ labs(title = "Correlation Between IRT and Cortisol",
+ x = "Thermographic Eye Temperature (^0C)",
+ y = "Cortisol (mcg/dL)")+
+ theme_minimal()+
+ theme(plot.title = element_text(face="bold", hjust =0.5))
`geom_smooth()` using formula = 'y ~ x'
> #| label: load-packages
> #| include: false
Fig. 1. A scatterplot of a correlation between Thermographic Eye Temperature (IRT) and Cortisol
Effect of a calming spray on compliance time
To determine whether the application of calming spray has an effect on horses and the compliance times, a Saphiro-Wilk normality test was used. The results of the test indicated that the differences between comp (compliance times before use of the calming spray) and comp2 (compliance times after use of the calming spray) are not normally distributed (W=0.84738, p-value<2.2x10-16). Since the differences are not normally distributed, further calculation has been taken with a non-parametric test. The Wilcoxon signed-rank test was used in contrast to a paired t-test. The result of the Wilcoxon signed-rank test indicates that the application of the calming spray has a statistically significant effect on compliance times (V=124,251, p-value<2.2x10-16). The findings imply that the compliance times before and after using the relaxing spray differ significantly. To determine whether calming spray lowered the compliance time median of both variables, it was calculated. Lower median in comp2 confirms that there is an improvement in compliance time.
#Inspect the data, making sure that compliance time before and after exist in eqsub dataset.
>head(eqsub)
# A tibble: 6 × 4
ID sex comp comp2
<chr> <chr> <dbl> <dbl>
1 EQ001 Female 37.2 31.5
2 EQ002 Female 34.5 29.6
3 EQ003 Female 36.3 30.8
4 EQ004 Male 35.3 30.1
5 EQ005 Female 37.4 31.7
6 EQ006 Male 33.5 28.6#Calculate the differences between both compliance times. Used Saphiro-Wilk test to test normality of the differneces.
>differences <-eqsub$comp-eqsub$comp2
> shapiro.test(differences)
Shapiro-Wilk normality test
data: differences
W = 0.84738, p-value < 2.2e-16#The p- value is not greater than 0.05, a non-parametric Wilcoxon signed-rank test is needed.
#Run Wilcoxon signed-rank test
>wilcox.test(eqsub$comp, eqsub$comp2, paired = TRUE)
Wilcoxon signed rank test with continuity correction
data: eqsub$comp and eqsub$comp2
V = 124251, p-value < 2.2e-16
alternative hypothesis: true location shift is not equal to 0#Compare the median of both compliance times before and after use of the calming spray.
>median(eqsub$comp)
[1] 35.04061
> median(eqsub$comp2)
[1] 29.98684#Create a histogram of a compliance time before the use of a calming spray
>ggplot(eqsub, aes(x=comp))+
+ geom_histogram(aes(y=..density..), binwidth=0.5, fill="grey", color="black", alpha=0.7)+
+ geom_density(color="yellow", size=1, adjust=1)+
+ labs(title="Histogram of a compliance time before the use of a calming spray", x="Compliance time (seconds)", y="Frequency")+
+ theme_minimal()+
+ theme(pot.title=element_text(hjust=0.5))
Fig.2. Histogram of compliance time before using calming spray.
#Create a histogram of a compliance time after the usage of a calming spray
>ggplot(eqsub, aes(x=comp2))+
+ geom_histogram(aes(y=..density..), binwidth=0.5, fill="grey", color="black", alpha=0.7)+
+ geom_density(color="green", size=1, adjust=1)+
+ labs(title="Histogram of a compliance time after the usage of a calming spray", x="Compliance time (seconds)", y="Frequency")+
+ theme_minimal()+
+ theme(pot.title=element_text(hjust=0.5))
Fig. 3. A histogram of compliance time after the use of calming spray.
#Create a boxplot to compare compliance times before and after use of the calming spray
>boxplot(eqsub$comp, eqsub$comp2,
+ names = c("Before Calming Spray", "After Calming Spray"),
+ ylab = "Compliance Time (seconds)",
+ main = "Comparison of Compliance Times",
+ col = c("slategray1", "lightskyblue3"))
Fig. 4. A Boxplot comparing compliance time of before and after the usage of calming spray.
Predicting Compliance Time
A multiple regression linear model was used to determine if it is possible to predict compliance time. A dependent variable (compliance time) was used to assess a relationship with independent variables such as heart rate, weight, air temperature, cortisol levels, and thermographic eye temperature (IRT). The summary of the regressions linear model showed that only heart rate (BPM) was a significant variable (p<0.001) with a positive coefficient (0.5050.5050.505). This implies that, while all other factors are held equal, a 1 BPM increase in heart rate corresponds to a 0.505-unit increase in compliance time. Other variables, including IRT (p=0.335p), cortisol levels (p=0.748p), weight (p=0.694), and air temperature (p=0.405), were not significant predictors. Since only BPM was significant, a linear model was refitted as a fundamental predictor.
| Predictor | Coefficient | Std. Error | t-value | p-value |
|---|---|---|---|---|
| Intercept | -39.10 | 3.64 | -10.74 | <0.001*** |
| Weight | 0.0013 | 0.0034 | 0.394 | 0.694 |
| IRT | -0.0308 | 0.0319 | -0.964 | 0.335 |
| Air Temperature | -0.0226 | 0.021 | -0.833 | 0.405 |
| Cortisol | -0.0008 | 0.0036 | -0.321 | 0.748 |
| BPM | 0.505 | 0.0223 | 22.62 | <0.001*** |
Table 1. Table presenting regression coefficients, standard errors, and significance levels.
#Fit a regression Model
> model <- lm(comp ~ weight + irt + air + cortisol + BPM, data = equine) > > summary(model) Call: lm(formula = comp ~ weight + irt + air + cortisol + BPM, data = equine) Residuals: Min 1Q Median 3Q Max -2.88796 -0.65339 0.00537 0.58312 2.79258 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -3.910e+01 3.641e+00 -10.739 <2e-16 *** weight 1.325e-03 3.361e-03 0.394 0.694 irt -3.078e-02 3.191e-02 -0.964 0.335 air -2.256e-02 2.708e-02 -0.833 0.405 cortisol -8.361e-04 2.602e-03 -0.321 0.748 BPM 5.048e-01 2.228e-02 22.661 <2e-16 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 1.003 on 492 degrees of freedom Multiple R-squared: 0.5129, Adjusted R-squared: 0.5079 F-statistic: 103.6 on 5 and 492 DF, p-value: < 2.2e-16
#Create a scatterplot of Compliance Time vs Heart Rate
> ggplot(equine, aes(x=BPM, y=comp))+
+ geom_point(alpha=0.3, size=3)+
+ geom_smooth(method="lm", color="dodgerblue1")+
+ labs(title="Compliance Time in Contrast with Heart Rate", x="Heart Rate (BPM)", y="Compliance Time (seconds)")+
+ theme_minimal()+
+ theme(plot.title = element_text(face="bold", hjust=0.5))
Fig. 5. A scatterplot of compliance time in correlation with Heart Rate (BPM)
#Fit linear regression model
>model <- lm(comp ~ weight + irt + air + cortisol + BPM, data = equine)
#Generate predicted values
equine$predicted <- predict(model)
#Create a scatteerplot of Actual vs Predicted compliance time.
library(ggplot2)
> ggplot(equine, aes(x = comp, y = predicted)) +
+ geom_point(color = "gray36") +
+ geom_abline(intercept = 0, slope = 1, color = "red", linetype = "dashed") +
+ labs(title = "Actual vs Predicted Compliance Time",
+ x = "Actual Compliance Time (seconds)",
+ y = "Predicted Compliance Time (seconds)") +
+ theme_minimal()+
+
+ theme(plot.title=element_text(face="bold", hjust=0.5))
Fig. 6. A scatterplot of comparison of the actual and predicted compliance time.
Discussion
The compliance time required to complete training tasks has long been a widely used measure for assessing stress in horses (Esch et al., 2019). This study aimed to investigate stress by evaluating compliance times and the relationship they might have with various variables, both psychological and environmental factors. This includes thermographic eye temperature (IRT), blood cortisol levels, heart rate, weight, and ambient air temperature.
Correlation between Thermographic Eye Temperature and Cortisol
The analysis demonstrated a significantly positive correlation between IRT and cortisol (r=0.1286, p=0.0040) which suggests a weak but statistically significant relation. These findings align with previous studies that also identified IRT as a promising non-invasive tool for measuring stress in horses (Readelli et al., 2019; Valera et al., 2012).
Because of its accuracy and ability to directly detect the hypothalamic-pituitary-adrenal (HPA) axis response, invasive tests such as blood cortisol collection continue to be used most commonly for evaluating stress (Aragona et al., 2024). However, there are several disadvantages to these approaches. Animals are naturally stressed during blood collection, which could bias the results (Sikorska et al., 2023). Non-invasive techniques, like IRT, on the other hand, are less invasive but can be affected by outside influences like illumination and ambient temperature. For instance, environmental variability may impair the accuracy of IRT data, requiring thorough calibration and controlled circumstances (Aragona et al., 2024).
While IRT has the advantage of being non-invasive, it is less specific in comparison with cortisol. The weak correlation observed in this study demonstrates how crucial it is to combine both invasive and non-invasive methods in order to gain an accurate understanding of equine stress response.
Effect of a calming spray on compliance time
The application of the calming spray reduced compliance times, as demonstrated by the Wilcoxon signed-rank test. The effectiveness of the spray was demonstrated by the decrease in median compliance times, which went from 35 seconds prior to the spray to 30 seconds following it. Similar findings were reported by McDonnell, Miller & Vaala (2014), who observed compliance time and stress reaction during the conduction of 12 healthcare examinations and treatment procedures. Their study demonstrated that calming spray containing alpha-casozepine effectively reduced stress behaviors in horses.
The significant reduction in compliance time allows to accept the hypothesis that the calming spray has an effect on the decrease of stress in horses. This study provides useful advantages in real-world situations, especially in high-stress settings like sporting events, transit, or veterinary checks.
Calming sprays are non-invasive and align with ethical guidelines for improving equine welfare. From the ethical perspective, the use of calming spray aligns with the principles of 3Rs; Replacement, Reduction, Refinement (Tannenbaum and Bennett, 2015). Calming sprays are an ethical substitute for intrusive treatments, which, although helpful, may cause discomfort or violate ethical welfare norms, as they have improved stress removal methods without causing harm.
The effectiveness of calming sprays, however, could be affected by outside variables like a horse’s temperament, task habituation, and the level of the environmental stimuli. Because these factors were not specifically accounted for in this study, the results may have been subject to variation. Additionally, although compliance time is a valuable stand-in for stress, it is an indirect metric that does not fully reflect the range of physiological alterations or behaviours associated with stress.
Predicting Compliance Time
Horse compliance times were found to be influenced by both physiological and environmental aspects by the multiple linear regression analysis. The model explained a moderate proportion of the variance in compliance times (R2=0.513R^2 = 0.513R2=0.513, Adjusted R2=0.508R^2 = 0.508R2=0.508), indicating that while variables included in the linear model have an effect on stress, additional factors are likely to be present as well. Among all the variables, heart rate (BPM) was the only significant variable (p<2.2×10−16p < 2.2 \times 10^{-16}p<2.2×10−16).
The positive association between heart rate and compliance times under stressful conditions aligns with existing studies. Findings of Dufourni et al. (2024) demonstrate the importance of a heart rate as a reliable and non-invasive method for assessing stress in horses.
Additionally, other variables—weight, thermographic eye temperature, ambient air temperature, and cortisol levels—did not show significant relationships with compliance time. Although cortisol is a well-established stress indicator, its lack of relevance in this study may be due to measurement variability or the complex connection between short-term and long-term stress reactions. Similar to this, although IRT is thought to be a non-invasive technique for identifying stress-related physiological changes, its predictive value might have been affected by individual or environmental factors.
Those findings emphasise the presence of the various factors of stress responses in horses. While heart rate was a strong indicator, the limited power of the model (R2=0.513R^2 = 0.513R2=0.513) suggests that other factors that were not measured for this study might be taking place, such as the temperament of the individual, prior training, or genetic factors that may significantly have an effect on compliance times. Physiological measurements, when combined with behavioural observations, may offer a more thorough comprehension of stress reactions.
Evaluation if Invasive and Non-invasive methods
This study evaluated both invasive and non-invasive approaches for assessing stress levels in equines. Each of those demonstrated certain points as well as limitations for the research.
Invasive methods such as blood sampling for measuring cortisol levels are widely used methodologies, as they have high accuracy in their results. Cortisol is considered a biomarker that offers reliable insights into animal stress levels (Aragona et al., 2024). However, methods of obtaining the blood samples can themselves induce stress, potentially raising cortisol levels above baseline levels, which may bias results.
Alternatively, a study by Ijichi et al. (2019) suggests that salivary cortisol samples are as effective as blood samples for measuring stress. Unlike blood sampling, salivary sampling is a non-invasive method, making it a more ethical option that could be widely adopted for animal welfare purposes.
The findings of this study imply the importance of integrating invasive and non-invasive methods for a comprehensive assessment of stress. Non-invasive methods can be used as initial evaluation tools, providing useful and ethically acceptable choices for real-time stress monitoring. On the other hand, invasive techniques offer comprehensive physiological information that can support non-invasive results and take potential biases into consideration.
For instance, researchers can cross-verify data by combining cortisol readings and thermographic eye temperature, which reduces the disadvantages of each technique when used separately. This strategy not only improves the accuracy of stress evaluations but also complies with ethical standards, including the 3Rs: Replacement, Reduction, and Refinement (Tannenbaum & Bennett, 2015), which lessen the overall toll on animals while upholding scientific integrity.
Implications for animal welfare
The results of this study have important implications for equine welfare. The significant reduction of compliance time after the use of a calming spray implies the potential for non-invasive methods to reduce stress and improve task performance. Furthermore, the ability to predict compliance time by using heart rate measurements provides valuable insight into the strategies of the management of the individuals.
The integration of both invasive and non-invasive methods indicates an improved approach for the equine research with enhanced welfare. With the utilisation of both methods, future studies can obtain more accurate and reliable data.
Conclusion
This study answers key questions about physiological and behavioural responses, contributing to the expanding corpus of research on equine stress evaluation. It demonstrated the importance of compliance time as an indicator of stress in equines and the enhancing functions of invasive and non-invasive techniques for evaluating physiological reactions and behaviours associated with stress. Equine stress reactions are complex, as demonstrated by the significant correlation between IRT and cortisol, the impact of the calming spray on compliance times, and the capacity to predict compliance time using a variety of factors.
Future studies
Findings in this study highlight the importance of ethical considerations in animal research. Future research should also focus on developing various methods that combine invasive and non-invasive procedures, as well as on improving determining models by adding behavioural variables. Such tactics will increase the precision of stress assessment while encouraging more ethical methods of stress reduction that put the welfare of the animals first.
References
Aragona, F., Rizzo, M., Arfuso, F., Acri, G., Fazio, F., Piccione, G. and Giannetto, C. (2024). Eye Temperature Measured with Infrared Thermography to Assess Stress Responses to Road Transport in Horses. Animals, [online] 14(13), pp.1877–1877. doi:https://doi.org/10.3390/ani14131877.
Budzyńska, M. (2014). Stress Reactivity and Coping in Horse Adaptation to Environment. Journal of Equine Veterinary Science, 34(8), pp.935–941. doi:https://doi.org/10.1016/j.jevs.2014.05.010.
Dufourni, A., Buschmann, E., Vernemmen, I., Van Steenkiste, G., van Loon, G. and Decloedt, A. (2024). Effect of physiological and pharmacological stress on heart rate, blood pressure, and echocardiographic measurements in healthy Warmblood horses. Journal of Veterinary Internal Medicine, 38(1), pp.398–410. doi:https://doi.org/10.1111/jvim.16967.
Esch, L., Wöhr, C., Erhard, M. and Krüger, K. (2019). Horses’ (Equus Caballus) Laterality, Stress Hormones, and Task Related Behavior in Innovative Problem-Solving. Animals, 9(5), p.265. doi:https://doi.org/10.3390/ani9050265.
Henshall, C., Randle, H., Francis, N. and Freire, R. (2022). Habit Formation and the Effect of Repeated Stress Exposures on Cognitive Flexibility Learning in Horses. Animals, 12(20), p.2818. doi:https://doi.org/10.3390/ani12202818.
Ijichi, C., Green, S., Squibb, K., Carroll, A. and Bannister, I. (2019). Zylkéne to load? The effects of alpha-casozepine on compliance and coping in horses during loading. Journal of Veterinary Behavior, 30, pp.80–87. doi:https://doi.org/10.1016/j.jveb.2018.12.009.
Ijichi, C., Wilkinson, A., Riva, M.G., Sobrero, L. and Dalla Costa, E. (2023). Work it out: Investigating the effect of workload on discomfort and stress physiology of riding school horses. Applied Animal Behaviour Science, [online] 267, p.106054. doi:https://doi.org/10.1016/j.applanim.2023.106054.
Johnson, S.R., Rao, S., Hussey, S.B. and Morley, P.S. (2011). Thermographic Eye Temperature as an Index to Body Temperature in Ponies. Journal of Equine Veterinary Science, [online] 31(2), pp.63–66. doi:https://doi.org/10.1016/j.jevs.2010.12.004.
Jolivald, A., Ijichi, C., Hall, C. and Yarnell, K. (2023). The mane factor: Compliance is associated with increased hair cortisol in the horse. Applied Animal Behaviour Science, 258, p.105819. doi:https://doi.org/10.1016/j.applanim.2022.105819.
McDonnell, S.M., Miller, J. and Vaala, W. (2014). Modestly Improved Compliance and Apparent Comfort of Horses With Aversions to Mildly Aversive Routine Health Care Procedures Following Short-Term Alpha-Casozepine Supplementation. Journal of Equine Veterinary Science, 34(8), pp.1016–1020. doi:https://doi.org/10.1016/j.jevs.2014.06.003.
Redaelli, V., Luzi, F., Mazzola, S., Bariffi, G., Zappaterra, M., Nanni Costa, L. and Padalino, B. (2019). The Use of Infrared Thermography (IRT) as Stress Indicator in Horses Trained for Endurance: A Pilot Study. Animals, 9(3), p.84. doi:https://doi.org/10.3390/ani9030084.
Sikorska, U., Maśko, M., Ciesielska, A., Zdrojkowski, Ł. and Domino, M. (2023). Role of Cortisol in Horse’s Welfare and Health. Agriculture, 13(12), pp.2219–2219. doi:https://doi.org/10.3390/agriculture13122219.
Squibb, K., Griffin, K., Favier, R. and Ijichi, C. (2018). Poker Face: Discrepancies in behaviour and affective states in horses during stressful handling procedures. Applied Animal Behaviour Science, 202, pp.34–38. doi:https://doi.org/10.1016/j.applanim.2018.02.003.
Tannenbaum, J. and Bennett, B.T. (2015). Russell and Burch’s 3Rs Then and Now: The Need for Clarity in Definition and Purpose. J Am Assoc Lab Anim Sci, [online] 120-32. PMID: 25836957; PMCID: PMC4382615. Available at: https://pmc.ncbi.nlm.nih.gov/articles/PMC4382615/#abstract1.
Valera, M., Bartolomé, E., Sánchez, M.J., Molina, A., Cook, N. and Schaefer, A. (2012). Changes in Eye Temperature and Stress Assessment in Horses During Show Jumping Competitions. Journal of Equine Veterinary Science, 32(12), pp.827–830. doi:https://doi.org/10.1016/j.jevs.2012.03.005.
Waran, N. and Randle, H. (2017). What we can measure, we can manage: The importance of using robust welfare indicators in Equitation Science. Applied Animal Behaviour Science, 190, pp.74–81. doi:https://doi.org/10.1016/j.applanim.2017.02.016.
Young, T., Creighton, E., Smith, T. and Hosie, C. (2012). A novel scale of behavioural indicators of stress for use with domestic horses. Applied Animal Behaviour Science, 140(1-2), pp.33–43. doi:https://doi.org/10.1016/j.applanim.2012.05.008.