Beyond cortisol: evaluating serotonin, brain-derived neurotrophic factor, and oxytocin as indicators of equine welfare across three training regimens.

This study tested whether serotonin, BDNF, and oxytocin—alongside cortisol—change before-to-after work in different types of sport and leisure horses, and whether such multi-marker patterns could reflect discipline-specific stress and recovery. Serotonin changed the most and in opposite directions by discipline (down after show jumping, up on race day in racehorses), while oxytocin and cortisol were largely unchanged; the authors conclude that multi-marker profiling shows promise for describing neuroendocrine dynamics but cannot, by itself, diagnose welfare.

What question did the study ask?

• Can a panel of circulating biomarkers—cortisol, oxytocin, serotonin (5-HT), and brain-derived neurotrophic factor (BDNF)—capture discipline- and context-specific stress–recovery responses in horses?
• Do these markers change consistently before versus after work/competition across three real-world regimens: show jumping, flat racing over a season, and leisure riding?
• Do sex and breed modify these responses?

Why these biomarkers?

• Cortisol: classic hypothalamic–pituitary–adrenal (HPA) stress hormone; acute rises to challenge but influenced by timing, circadian rhythm, and habituation.
• Serotonin (5-HT): neuromodulator linked to mood, arousal, coping; in blood, largely stored in platelets and can reflect physiological activation with exercise or stress.
• BDNF: neurotrophin involved in neuroplasticity and exercise adaptation; in serum, heavily influenced by platelet release, potentially responsive to workload and training.
• Oxytocin: peptide associated with social bonding, anti-stress effects, and autonomic regulation; peripheral measurements are challenging but may reflect affiliative context or recovery.
• Rationale: no single marker indexes “welfare,” but complementary signals across systems might better characterize stress–recovery balance.

How was the study done?

• Design: Paired serum sampling immediately before and after standardized work contexts within three cohorts:
  • Show jumping horses: n=17, sampled pre- and post-competition.
  • Racehorses: n=43 enrolled (n=41 analyzed), sampled pre/post at season start (T1), mid-season (T2), and race day (R).
  • Leisure horses: n=14, sampled at rest before and after the riding season.
• Assays: Commercial ELISAs for cortisol, oxytocin, 5-HT, and BDNF in serum.
• Outcome metric: Post/pre ratio of each marker modeled on the natural log scale [ln(post/pre)] to stabilize variance and allow symmetric interpretation of increases/decreases. Ratios were reported back on the original scale for clarity.
• Statistical models:
  • Show jumping and leisure: intercept-only linear models estimating the mean ln(post/pre) across horses.
  • Racehorses: linear mixed-effects models with horse as a random intercept (to handle repeated measures), and fixed effects for session (T1, T2, R), sex, and breed; tested for session-specific changes and sex differences.

Key results by discipline and marker

• Show jumping (pre vs post competition):
  • Serotonin (5-HT): decreased; post/pre ratio 0.634 (95% CI 0.490–0.820; p=0.002), about a 36% drop overall.
  • Sex- and breed-adjusted signals:
    • Mares: stronger 5-HT decrease; ratio 0.470 (≈53% decrease; p=0.026).
    • Geldings: BDNF increased; ratio 1.727 (≈73% increase; p=0.021).
  • Cortisol and oxytocin: no significant paired change reported in the abstract.
• Racehorses (seasonal repeated measures):
  • Serotonin (5-HT):
    • Race day (R): increased in mares (ratio 1.948; 95% CI 1.258–3.015; p=0.003) and stallions (ratio 2.207; p<0.001), roughly 95–121% rises.
    • Season start (T1) and mid-season (T2): no deviation from 1.0 (no change) overall.
  • BDNF and oxytocin: remained near 1.0 across T1, T2, and R (all p≥0.507), indicating minimal pre/post change under this sampling schedule.
  • Cortisol: no session-specific effects; stallions showed lower post/pre ratios at T2 and R that trended but did not reach significance (p=0.059 and 0.064).
• Leisure horses (before vs after riding season at rest):
  • No significant changes for any marker (all p≥0.166), suggesting stability under low-intensity, routine conditions and rest sampling.

Interpreting the discipline- and sex-specific patterns

• Serotonin as the most responsive marker:
  • Opposite directions by discipline suggest different physiological or psychological demands:
    • Show jumping competition: decreased 5-HT may reflect acute consumption/redistribution, altered platelet dynamics, or stress-related serotonergic changes in a high-focus, intermittent power sport.
    • Race day: increased 5-HT may reflect arousal/anticipation, sympathetic activation, or platelet release during sustained high-intensity exercise.
  • Strong sex effects on race day (mares and stallions) imply hormonal or behavioral modulation of serotonergic responses under maximal arousal/effort contexts.
• BDNF nuance:
  • Selective increase in show jumping geldings suggests context- and sex-dependent release, possibly tied to acute neuromuscular load or platelet-mediated responses.
  • Lack of change in racehorses may reflect different exercise modality, timing of sampling relative to peak BDNF release, or chronic training adaptations blunting acute serum changes.
• Oxytocin stability:
  • Near-1.0 ratios across sessions indicate little pre/post shift in peripheral oxytocin under these conditions or limited assay sensitivity to transient changes.
• Cortisol flatness:
  • Absence of clear session-specific increases suggests that the pre/post sampling windows may have missed peak HPA responses, or that habituation and anticipation offset net changes.
  • Trends toward lower post/pre in stallions on T2 and R could reflect individual or sex-related recovery dynamics but were not statistically conclusive.

Statistical meaning of the reported ratios

• A post/pre ratio below 1.0 indicates a decrease; above 1.0 indicates an increase. For example, 0.634 ≈ 36% decrease; 1.727 ≈ 73% increase.
• Modeling ln(post/pre) provides:
  • Symmetry for increases and decreases (e.g., +50% and −33% have equal magnitude on the log scale).
  • Compatibility with mixed-effects frameworks to account for repeated measures within horses (race cohort).
• Confidence intervals around ratios describe precision; intervals excluding 1.0 support a true change under the model.

What this does and does not say about welfare

• Findings:
  • Multi-marker profiling detects discipline- and sex-specific neuroendocrine dynamics in real-world equine sport contexts.
  • Serotonin shows the clearest and most context-sensitive signal; cortisol and oxytocin were largely uninformative under the current sampling framework.
• Limits:
  • Physiological change does not equate to “good” or “poor” welfare without behavioral, performance, and health context.
  • Opposite 5-HT directions across disciplines underscore that interpretation is task-specific and cannot be reduced to a single threshold.

Strengths of the study

• Real-world sampling across distinct disciplines (show jumping, racing, leisure) with paired pre/post design.
• Inclusion of sex and breed effects and use of mixed-effects modeling for repeated racehorse measures.
• Comparative multi-marker approach rather than reliance on cortisol alone.

Key limitations and caveats

• Timing and kinetics:
  • Single pre/post samples may miss rapid peaks/troughs (especially for cortisol and oxytocin), and do not map full recovery curves.
• Assay and matrix considerations:
  • Serum 5-HT and BDNF are strongly influenced by platelet release during clotting; changes may reflect platelet dynamics as much as central neuromodulation.
  • Peripheral oxytocin ELISAs can face specificity issues; peripheral levels may not mirror central activity.
• Context control:
  • Potential variability in exact sampling intervals, circadian timing, diet, handling, and environmental stimuli could affect hormone levels.
• Scope:
  • Leisure horses sampled at rest before/after season may not capture responses to actual riding sessions.
  • No concurrent behavioral or autonomic (e.g., heart-rate variability) measures to triangulate welfare states.
• Power:
  • Modest sample sizes in subgroups (e.g., show jumping by sex) limit precision and generalizability of sex-specific estimates.

Implications for practice and monitoring

• Use panels, not single markers:
  • Combine biomarkers with behavior, performance metrics, and injury/veterinary data for a richer welfare picture.
• Context-specific baselining:
  • Interpret changes relative to the horse’s own baseline and discipline; expect opposite 5-HT directions in different sports.
• Sampling strategies:
  • Consider serial sampling (e.g., immediate post, 15–30 min, 1–2 h) to capture kinetics, especially for cortisol and oxytocin.
  • Standardize time-of-day and pre-sampling routines to reduce variability.
• Analytical choices:
  • Account for platelet counts or use plasma protocols when interpreting 5-HT and BDNF; validate oxytocin assays for specificity.

Future research directions

• Integrate biomarker panels with behavioral ethograms, heart-rate variability, lactate, and GPS workload to link physiology with welfare-relevant outcomes.
• Map time courses with repeated post-exercise sampling to resolve peak timing and recovery for each marker.
• Investigate mechanisms behind sex differences, including reproductive status and hormonal milieu.
• Examine training adaptation vs acute response by tracking within-horse changes across mesocycles.
• Evaluate management factors (transport, stabling, social contact) on oxytocin and serotonergic responses.

Bottom-line takeaways

• Serotonin is the most sensitive and discipline-dependent blood marker here: down after show jumping competition, up on race day in racehorses (especially mares and stallions).
• BDNF rose only in show jumping geldings; oxytocin and cortisol were largely unchanged under the study’s sampling schedule.
• Multi-marker profiling can describe neuroendocrine dynamics but is not a stand-alone welfare index; interpretation must be discipline-, sex-, and context-aware and paired with behavioral and performance data.