Relationship Between Physical Activity, Blood Cortisol Levels and Innate Immune Response in Common Bottlenose Dolphins (Tursiops truncatus) in a Controlled Environment

TL;DR

This study explores how physical activity affects cortisol levels and immune function in captive bottlenose dolphins, revealing a strong link between stress hormones and immune response.

Contribution

The study provides new insights into the relationship between cortisol, physical activity, and innate immune function in cetaceans.

Findings

Physical activity significantly increased cortisol levels by 122% in dolphins.

Phagocytic activity in granulocytes and monocytes decreased during periods of physical activity.

A negative correlation was found between cortisol levels and immune cell function.

Abstract

Cortisol has been extensively studied in captive cetaceans, but its relationship to immune function remains poorly understood. An apparently asymptomatic individual exposed to certain exogenous or endogenous factors could experience an increase in blood cortisol levels, which could alter the function of immune system cells and influence its well-being. To better understand these mechanisms and their implications for animal welfare, this article will evaluate the relationship between blood cortisol levels and their influence on the function of the innate immune system in captive bottlenose dolphins.

This study investigates the effects of physical activity on serum cortisol levels and phagocytic capacity of the innate immune system in eight common bottlenose dolphins under human care. Analysis of 8 pairs (16 samples) revealed a significant increase in cortisol during periods of physical activity (mean increase of 1.27 µg/dL, 122% elevation), accompanied by decreased phagocytosis in granulocytes (92% reduction) and monocytes (52% reduction). Statistical analyses demonstrated consistent negative correlations between cortisol levels and phagocytic function, suggesting that physical activity influences hypothalamic–pituitary–adrenal axis activation and, consequently, innate immune system function. The male showed an attenuated cortisol response, but maintained monocyte sensitivity. These findings highlight the complex interplay between the neuroendocrine cortisol response and immune function in cetaceans, for the management and evaluation of animals under human care. A multi-method statistical framework incorporating Bayesian analysis, bootstrapping, and traditional approaches ensured robust inference despite limited sample size.

Figures

Tables

Table 5. Regression model summary.

Predictor	Granulocytes (1)	Monocytes (2)
Physical Activity	27.229	−20.166 *
Cortisol	5.517	−5.775
Monocytes	2.457 ***	—
Granulocytes	—	0.273 ***
Constant	−55.014	40.667 ***
R2	0.860	0.933
Adjusted R2	0.808	0.908

Full text

1. Introduction

Cortisol is a glucocorticoid synthesized from cholesterol, secreted by the adrenal cortex and released into the blood [1]. In blood plasma, most cortisol (65%) binds with high intensity to corticosteroid-binding globulin (transcortin), 30% binds to albumins, while 3–5% remains in a metabolically active form (free) [1,2,3].

Cortisol secretion follows a circadian rhythm, with highest levels occurring in the early morning hours (30–50 min after waking) [1,2]. Additionally, certain factors such as physical activity episodes, exercise, food intake, or individual stress can generate an increase in blood cortisol levels, peaking 10–30 min after stimulus interaction [1,2].

In general, this neuroendocrine response can affect immune system (IS) functioning. When an individual perceives any of the described factors, a physiological response is triggered in the hypothalamus (central nervous system), which converts nerve impulses into hormonal signals, releasing cortisol into the blood, that subsequently acts on primary and secondary lymphoid organs, T and B lymphocytes, neutrophils, monocytes, and macrophages, triggering alterations in the IS [4,5,6].

The immune system communicates and modulates through cell contact or low-molecular-weight proteins called interleukins (ILs) or cytokines. Studies in mammals have observed that corticoids can inhibit immune cell migration and proliferation, as well as numerous immune responses, such as increased chemotaxis and IL production, especially proinflammatory ones like IL-8 [7,8]. In addition, studies with free-ranging cetaceans observed that when individuals are subjected to stressful stimuli such as capture and subsequent release, proinflammatory IL-8 increases [9], potentially causing IS suppression. This provides evidence of the relationship between prolonged or constant stressor exposure, blood cortisol levels, and IS functionality.

Although analysis of cetaceans’ cortisol levels and their relationship to stress in controlled environment under human care has been addressed in several previous studies [10,11], immune system functioning and its correlation with blood cortisol levels remains a poorly explored research line. Understanding general IS functioning and how certain factors influence participating cells could be relevant, as previous studies with free-ranging dolphins observed that individuals typically do not present observable clinical disease manifestations until very advanced stages, thus showing no signs of weakness and avoiding predator attraction [12]. In captivity, although predators are absent, an apparently asymptomatic individual exposed to certain exogenous or endogenous factors could experience increased blood cortisol levels, potentially altering IS cell functioning [12].

The present study employs complementary statistical approaches to explore associations between the factor “physical activity”, cortisol dynamics, and the functionality of certain cells of the innate IS in bottlenose dolphins under human care, aiming to better understand these mechanisms and their implications in bottlenose dolphins in controlled environments.

2. Materials and Methods

2.1. Study Subjects

The study involved eight bottlenose dolphins (Tursiops truncatus), comprising one male and seven females, maintained under controlled conditions at Zoo Aquarium Madrid, Spain. During the study, the females were not provided with any contraceptive method, as this could influence our results. It should also be noted that although some specimens were not born at the Madrid Zoo Aquarium, it was previously verified that all had spent most of their lives under human care (Table 1).

The dolphinarium facilities include three interconnected pools: two indoor pools without public access, connected to an outdoor pool with public access. All eight dolphins could move freely among the three pools throughout the day, except during physical activities performed during training and educational interactions, when they remained in the main pool.

The amount of physical activity performed by all study subjects during public educational sessions was 15 min twice daily. Additionally, in the mornings before the zoo opened to visitors, caretakers conducted enrichment games and physical activities.

This demographic composition provides ecological validity, reflecting typical population structures under controlled conditions, while allowing for the examination of possible sex-specific responses, although it should be noted that the limited male representation requires a cautious interpretation of sex-based differences.

2.2. Sample Collection and Stress Minimization Protocol

Since cortisol levels are known to be higher in the morning; to assess how cortisol increases influence IS function, blood samples were collected at two different times of day:

• Time 1 (WPA—Without Physical Activity): Early morning (approx. 10 AM), before first meal and without physical activity.
• Time 2 (DPA—During Physical Activity): Late morning (approx. 12 PM), after physical training and before educational encounter activity.

Importantly, before initiating research, all individuals underwent physical and medical evaluations to assess health status and rule out any pathology that could influence results. Additionally, samples were always collected without prior feeding, as previous studies demonstrated that fish contain high cortisol concentrations that could alter results [13].

Sample collection was always carried out within the pool, without the use of any lifting platform, and was performed voluntarily by all study subjects. All individuals had been previously trained for blood sample collection. Keepers used hand signals to indicate to the individuals that they should turn slightly within the pool so that blood could be drawn from the caudal fin. During sampling, if an animal did not approach or cooperate after a couple of attempts, it was left alone without further prompting. Because it was not a forced process, the stress component was minimized, and consequently, its potential effect on blood cortisol levels was reduced.

As a result, sample collection took longer, as some days obtaining samples from all individuals was not possible, since they were never forced, restricted, restrained, or isolated for collection. For all analyses, 1 mL of blood was drawn from each individual and separated into two BD Vacutainer® Heparin Tubes (0.5 mL in each). A total of 96 blood samples were obtained for cortisol level analysis and 96 samples for innate immune response assessment, distributed as shown in Table 2:

2.3. Blood Cortisol Analysis

At the Biomedical Research Unit (UIB) of the Alfonso X el Sabio University in Madrid, Spain, serum cortisol concentrations were evaluated. For this purpose, once the blood was drawn, each tube was centrifuged at 1300× g for 10 min to separate the serum, and this serum was subsequently stored at −20 °C. The Demeditec Cortisol ELISA kit, a competitive enzyme immunoassay for the quantitative measurement of cortisol, was used to measure cortisol levels in the 96 samples.

Multi-well plates were coated with anti-cortisol antibody. The unknown cortisol amount in our sample competes with a cortisol–horseradish peroxidase (HRP) conjugate for binding. The bound peroxidase conjugate amount is inversely proportional to cortisol concentration in the sample. Therefore, after substrate addition, the developed color intensity is inversely proportional to sample cortisol concentration.

2.4. Innate Immune Response Assessment

The evaluation of the innate immune response was performed at the University of the Balearic Islands (UIB) and the Jiménez Díaz Foundation (Madrid, Spain). To do this, the blood samples, once obtained, were transferred from the Zoo to the UIB or the Jiménez Díaz Foundation, kept at 4 °C (maximum 40 min), to later analyze the phagocytic capacity of granulocytes and monocytes using the IngoFlowEx®, (EXBIO Praha, Prague, Czech Republic) kit. This kit evaluates the ingestion of Escherichia coli (E. coli) bacteria by these immune cells, as this bacterium can infect cetaceans, including dolphins [12,14,15]. The IngoFlowEx® kit was selected because it has been previously used as a diagnostic tool to evaluate phagocytosis in marine mammals [12].

Kit-provided bacteria were labeled with fluorescein (FITC) for subsequent flow cytometry evaluation. Phagocytic capacity was determined by the percentage of FITC-positive cells, analyzing a minimum of 20,000 events (cells) per dolphin and sample. Cell populations were identified based on size and granularity: the largest and most complex were classified as granulocytes (neutrophils), and the largest with intermediate complexity as monocytes. This approach represents a validated methodology previously used in cetaceans where species-specific antibodies are limited [12,16,17]. Analyses were performed on the CytoFLEX cytometer (Beckman Coulter, Brea, CA, USA) using CytExpert 2.4 software (Beckman Coulter) at both Fundación Jiménez Díaz and Alfonso X El Sabio University (Madrid, Spain).

2.5. Statistical Analysis

Given the small sample size ( $[eqn]$ dolphins, 16 paired observations), we employed a comprehensive multi-method statistical framework to explore associations and address potential limitations of individual approaches. It should be noted that the limited sample size in this study is context-specific, reflecting ethical and logistical constraints of working with protected marine mammals, rather than inherent to the species itself. This integrated strategy allows for the assessment of result consistency across multiple analytical perspectives.

2.5.1. Normality Testing and Non-Parametric Analysis

Data were initially assessed for normality using the Shapiro–Wilk test, as many statistical tests assume normal distribution and deviations from this assumption can lead to incorrect conclusions. Due to ambiguous results in some variables, particularly non-normal distribution of granulocyte data at Time 2, we selected non-parametric tests appropriate for paired data with limited sample sizes. Specifically, we used the Wilcoxon signed-rank test for paired data to compare conditions with and without physical activity.

Preliminary Shapiro–Wilk testing revealed that granulocyte data during DPA did not follow a normal distribution ( $[eqn]$ ), while other variables showed acceptable normality (cortisol WPA: $[eqn]$ , DPA: $[eqn]$ ; granulocytes WPA: $[eqn]$ ; monocytes WPA: $[eqn]$ , DPA: $[eqn]$ ). These results justified the selection of non-parametric approaches for all analyses.

2.5.2. Bootstrapping for Confidence Intervals

To address uncertainty inherent in small sample sizes, we implemented bootstrap resampling (1000 iterations) to generate empirically derived confidence intervals for all primary metrics. This approach makes no distributional assumptions and provides empirical interval estimates even when classical parametric assumptions are violated. Bootstrap confidence intervals offer particular value when working with rare or endangered species where sample sizes are necessarily constrained by ethical and practical considerations.

2.5.3. Bayesian Analysis

We complemented frequentist approaches with Bayesian analysis to provide probability-based inference particularly suited to limited-sample contexts [18]. Bayesian methods allow incorporation of prior information from the literature on dolphin cortisol physiology while quantifying uncertainty through posterior distributions. We specified weakly informative priors and conducted sensitivity analyses to assess the consistency of conclusions across different prior specifications. Bayesian posterior probabilities provide more intuitive interpretation than p-values, directly quantifying the probability that physical activity affects physiological parameters.

2.5.4. Correlation and Regression Analysis

We explored correlations between variables to identify potential relationships and interactions. Additionally, we estimated multiple regression models to determine significant predictors and provide more comprehensive understanding of how different factors influence outcomes.

This comprehensive approach addresses limitations of individual statistical methods and allows for the assessment of result consistency across multiple analytical frameworks. However, it should be noted that applying several analytical methods to the same limited dataset does not generate independent lines of evidence, and therefore, the findings should be interpreted as exploratory associations rather than definitive causal relationships.

3. Results

3.1. Serum Cortisol Analysis and Distributional Patterns

Serum cortisol levels (µg/dL) were analyzed in each sample, and observed values fell within ranges previously described by St. Aubin [19] (0.4–3.6 µg/dL), confirming the physiological relevance of our measurements.

Overall, all individuals showed increased cortisol levels after training and before the educational encounter (Time 2, DPA), with values consistently higher than those observed during the WPA condition.

Analyses revealed profound physiological alterations between conditions. Cortisol exhibited a 122% increase during DPA (mean increase of 1.27 µg/dL), rising from a mean of 1.04 ± 0.12 µg/dL without physical activity to 2.31 ± 0.61 µg/dL with physical activity. This increase represents a physiologically substantial and consistent change across all studied individuals.

Cortisol distribution shifted upward during DPA, while phagocytosis distributions shifted downward (Figure 1).

Bootstrap analysis (1000 resamples) yielded 95% confidence intervals that did not overlap between conditions (WPA: 0.82–1.26 µg/dL; DPA: 1.70–2.92 µg/dL). Bayesian analysis yielded a posterior probability >0.99 for higher cortisol during physical activity, with posterior mean difference of 1.28 µg/dL (95% credible interval: 0.61–1.94 µg/dL).

Consistent directional changes were observed across all individuals, with notable inter-individual variability in response magnitude (Figure 2).

Notably, in the case of the male dolphin (Individual 5), the increase in serum cortisol during DPA was not as significant as in females (WPA: 1.1 µg/dL; DPA: 1.5 µg/dL)

3.2. Individual Response Consistency

All subjects showed an increase in cortisol and a decrease in phagocytosis during the physical activity period (DPA) (Figure 3). This directional consistency was observed in all individuals, though the magnitude of change varied among subjects.

3.3. Innate Immune Response Analysis

Flow cytometry analysis of FITC-labeled E. coli phagocytosis by granulocytes and monocytes under WPA and DPA conditions revealed notable changes reflecting cortisol influence on the immune system.

3.3.1. Female Response

Females showed significant decrease in granulocyte phagocytosis during DPA (3.44%) compared to WPA (42.86%), representing a 92% reduction. The flow cytometry histograms (Figure 4 left and right) from one of the females illustrate this drastic suppression of phagocytic activity.

Similarly, decreased monocyte phagocytosis was observed during DPA (2.89%) compared to WPA (19.67%), although this 52% reduction was not as pronounced as that observed in granulocytes. The following are flow cytometry data (Figure 5 left and right) from one of the females that demonstrated this monocyte suppression pattern.

These findings are consistent with the negative correlation observed between physical activity and phagocytic capacity.

Bootstrap confidence intervals for female granulocyte phagocytosis showed no overlap between conditions (WPA: 35.2–50.5%; DPA: 1.8–5.1%), strongly supporting the physiological significance of this decrease. Bayesian posterior probability that physical activity reduces granulocyte phagocytosis exceeded 0.99.

3.3.2. Male Response

Unlike females, no significant differences were observed in the male dolphin regarding FITC-labeled E. coli phagocytosis by granulocytes between WPA (31.78%) and DPA (30.56%) conditions. Flow cytometry histograms (Figure 6 left and right) show maintained granulocyte function across both conditions in the male subject.

However, decreased monocyte phagocytic activity was observed during DPA (2.48%) compared to WPA (8.84%). Flow cytometry analysis (Figure 7 left and right) demonstrates this selective monocyte suppression in the male dolphin.

In this individual, monocyte phagocytosis decreased during DPA while granulocyte phagocytosis remained stable, showing a differential response pattern between these cell types.

3.4. Wilcoxon Test

Wilcoxon test results showed significant differences ( $[eqn]$ ) for all variables between Time 1 (WPA) and Time 2 (DPA) conditions (Table 3).

All variables showed significant differences between conditions, with large effect sizes ( $[eqn]$ ).

3.5. Correlations Between Variables

The following correlations were observed between variables, as detailed in Table 4. These correlations allow better understanding of relationships between physiological and immune responses during “physical activity” periods.

Serum cortisol levels showed a strong positive correlation with the physical activity condition ( $[eqn]$ ). Cortisol levels showed negative correlations with granulocyte ( $[eqn]$ ) and monocyte ( $[eqn]$ ) phagocytosis. Granulocyte and monocyte phagocytosis were strongly positively correlated with each other ( $[eqn]$ ), and both showed negative correlations with the physical activity condition ( $[eqn]$ and $[eqn]$ , respectively).

Bootstrap correlation confidence intervals confirmed the stability of these relationships, and Bayesian correlation analysis provided posterior probabilities >0.95 for all major correlations, indicating high confidence in these associations.

3.6. Multiple Regression Models

Multiple regression models were fitted to provide more comprehensive understanding of how different factors influence outcomes (Table 5):

• The model for granulocytes explained 86% of variance ( $[eqn]$ = 0.860), with monocytes as a significant predictor ( $[eqn]$ ).
• The model for monocytes had the best fit, explaining 93.3% of variance ( $[eqn]$ = 0.933), with physical activity ( $[eqn]$ ) and granulocytes ( $[eqn]$ ) as significant predictors.

Table 5: Regression model summary.

<table><thead><tr><th align="center" rowspan="1" colspan="1">Predictor</th><th align="center" rowspan="1" colspan="1">Granulocytes (1)</th><th align="center" rowspan="1" colspan="1">Monocytes (2)</th></tr></thead><tbody><tr><td align="center" rowspan="1" colspan="1">Physical Activity</td><td align="center" rowspan="1" colspan="1">27.229</td><td align="center" rowspan="1" colspan="1">−20.166 *</td></tr><tr><td align="center" rowspan="1" colspan="1">Cortisol</td><td align="center" rowspan="1" colspan="1">5.517</td><td align="center" rowspan="1" colspan="1">−5.775</td></tr><tr><td align="center" rowspan="1" colspan="1">Monocytes</td><td align="center" rowspan="1" colspan="1">2.457 ***</td><td align="center" rowspan="1" colspan="1">—</td></tr><tr><td align="center" rowspan="1" colspan="1">Granulocytes</td><td align="center" rowspan="1" colspan="1">—</td><td align="center" rowspan="1" colspan="1">0.273 ***</td></tr><tr><td align="center" rowspan="1" colspan="1">Constant</td><td align="center" rowspan="1" colspan="1">−55.014</td><td align="center" rowspan="1" colspan="1">40.667 ***</td></tr><tr><td align="center" rowspan="1" colspan="1"> <inline-formula> <math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mrow><msup><mi>R</mi><mn>2</mn></msup></mrow></math> </inline-formula> </td><td align="center" rowspan="1" colspan="1">0.860</td><td align="center" rowspan="1" colspan="1">0.933</td></tr><tr><td align="center" rowspan="1" colspan="1">Adjusted <inline-formula><math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mrow><msup><mi>R</mi><mn>2</mn></msup></mrow></math></inline-formula></td><td align="center" rowspan="1" colspan="1">0.808</td><td align="center" rowspan="1" colspan="1">0.908</td></tr></tbody></table>

Bayesian regression models yielded consistent parameter estimates with credible intervals that excluded zero for significant predictors.

However, given the limited sample size and high correlation among predictors, these regression results should be interpreted as exploratory, as the high $[eqn]$ values may partially reflect the data structure rather than true explanatory power.

Data analysis provides robust evidence of physical activity’s impact on subjects’ physiological responses. The use of non-parametric tests, correlation analysis, multiple regression models, bootstrap resampling, and Bayesian inference ensured comprehensive and nuanced data understanding.

4. Discussion

This analysis explored associations between physical activity and physiological changes in bottlenose dolphins under human care. The application of multiple statistical approaches allowed for the assessment of result consistency.

Common bottlenose dolphins (T. truncatus), like other marine mammals, face exogenous factors both in the wild and in controlled environments that can cause alterations in individual behavior, group interaction, and physiological homeostasis [19,20,21]. Regarding blood cortisol levels, previous studies have already observed that multiple factors such as sex, age, season, time of day, noise, and temperature variations can alter these levels [10,22,23,24,25,26]. In our study, we have observed that not only does the onset of physical activity in T. truncatus significantly influence blood cortisol levels, but also that, as a consequence, the function of cells of the innate immune system is altered through complex neuroendocrine-immune interactions.

To assess the reliability of the blood cortisol levels obtained, they were compared with those intervals described in previous research (0.4–3.6 µg/dL) [20,22,27]. We did not use a control subject in our study because this would have required separating the control subject from the other participants, depriving them of exercise and the company of the group. This isolation could have affected their well-being, triggered stress in the study subject, and potentially lead to alterations in cortisol levels, resulting in unreliable results.

It is well established that multiple factors influence individual cortisol variation, including sex, age, season, time of day, and exogenous variables such as acoustic environment and thermal fluctuations in the enclosure’s water. In our controlled comparison between conditions differentiated primarily by physical activity presence (DPA) or absence (WPA), we observed a consistent elevation of this glucocorticoid in all individuals, with particularly pronounced responses in females during DPA phases. This pattern strongly suggests that physical activity represents a significant modulator of cortisol dynamics in T. truncatus, producing quantifiable increases in circulating levels that could reflect at least some physiological activation by the individuals.

A critical consideration is whether these cortisol elevations represent transient, exercise-induced spikes or whether these levels persist throughout the day. If cortisol levels remain elevated, the possibility of chronic stress and the potential influence of other variables on blood cortisol levels should be assessed. Therefore, future research should consider more extensive monitoring protocols, including post-physical activity and pre-night sampling, to distinguish between transient physiological responses and persistent stress states.

The multi-method statistical framework proved useful for small-sample contexts. Bootstrap confidence intervals provided empirical estimates without distributional assumptions, while Bayesian posterior probabilities (>0.99 for cortisol increase) offered intuitive interpretation. The consistency of results across frequentist, bootstrap, and Bayesian approaches suggests that observed cortisol changes reflect genuine physiological patterns. Furthermore, replicating these analyses in other centers with similar social structures would further improve the generalizability of the findings. The convergence of evidence across frequentist, bootstrap, and Bayesian approaches substantially strengthens our confidence that observed cortisol changes represent evidence of physical activity’s impact on the subjects.

On the other hand, this pattern found in the male could have a physiological explanation in the well-characterized interaction between hypothalamic–pituitary–adrenal (HPA) and hypothalamic–pituitary–gonadal (HPG) axes. In the classical stress response pathway, sympathetic–adrenal–medullary (SAM) axis activation is initiated when sympathetic preganglionic neurons issue catecholamines released in response to physical activity or stress, exert immunomodulatory functions. This cascade increases plasma energy substrates while suppressing certain reproductive hormones, including testosterone [28,29,30].

Notably, the absence of male–male competition in our study population suggests that even minor cortisol elevations causing slight testosterone reductions likely do not compromise individual condition. However, singular male representation prevents definitive conclusions about sex-specific effects, necessitating future studies with balanced sexual representation.

Our investigation of innate immune function focused on granulocyte and monocyte/macrophage phagocytic capacity for a novel exploration of cetacean physiology in controlled environments. Previous marine mammal immunology has emphasized extreme stressors such as transport and habitat translocation [16,17,31], while routine management factors like physical activity remain poorly explored. We selected E. coli as a relevant phagocytic target given its established pathogenicity in both terrestrial and marine mammals [12,14,32,33,34].

Observed immune suppression patterns reveal sophisticated physiological regulation. Female subjects exhibited coordinated decreases in both granulocyte and monocyte phagocytosis during DPA, alongside strong negative correlations between cortisol and phagocytosis.

This aligns with established mammalian stress immunology where glucocorticoids and catecholamines released during responses to physiological activity or stress exert immunomodulatory functions [5]. In additon, bootstrap confidence intervals for female granulocyte phagocytosis showed no overlap between conditions, and Bayesian posterior probabilities (>0.99) supported the observed activity-associated suppression. The consistency across analytical methods provide additional confidence in these findings despite the limited sample size. Therefore, our findings suggest physical activity triggers cortisol-mediated immune regulation, potentially increasing transient vulnerability to environmental pathogens [5,35].

The male subject showed distinct immune modulation, with maintained granulocyte function but suppressed monocyte phagocytosis during DPA. On the one hand, this dissociation suggests that each cell type has a specific sensitivity to neuroendocrine signals, with monocytes potentially being the most sensitive indicators of blood cortisol variations. The greater sensitivity of monocyte phagocytic activity observed in this study may reflect fundamental biological differences in glucocorticoid responsiveness between innate immune cell types [36,37]. In this regard, previous studies have described the existence of two types of glucocorticoid receptors (GRs): alpha and beta. Humans neutrophils have a higher concentration of beta glucocorticoid receptors compared to the alpha isoform [38]. The beta isoform acts as a glucocorticoid inhibitor [39], which makes neutrophils less sensitive to the presence of cortisol than other peripheral blood mononuclear cells (PBMCs) such as monocytes [40]. Considering that males have a much lower increase in cortisol than females, and that glucocorticoid receptor isoforms are likely to be present in different proportions, these factors could explain the differences in phagocytosis response between males and females. Glucocorticoids preferentially suppress monocyte and macrophage functions, including cytokine production and phagocytosis, while neutrophil activity may remain relatively preserved during acute endocrine activation [41,42]. In addition, recent evidence further supports the role of monocytes as highly plastic endocrine-immune sensors, responding dynamically to systemic hormonal fluctuations [43,44].

Similar to our results, previous cetaceans research suggests that stress-related immune modulation may preferentially affect regulatory innate immune functions, consistent with the pattern observed here [12,17].

It should be noted that the monocytes possess multifaceted immune functions beyond phagocytosis, including inflammation mediation and immune signaling through molecules like TNF, IL-6, and IL-15. They also modulate other immune components including interferon- $[eqn]$ [44], which itself decreases E. coli phagocytic capacity in macrophages [45].

Our methodology detected phagocytically active monocytes (macrophages), but not necessarily all functional subsets of monocytes, which leaves open the possibility that in the case of the male, physical activity influences other functions of monocytes without altering phagocytic parameters.

Likewise, this male bottlenose dolphin’s immune profile could also reflect testosterone-mediated immunomodulation. Previous research has shown that testosterone can influence circulating immune cell populations, particularly monocytes, granulocytes, and platelets [43], and modulates monocyte inflammatory responses by increasing the production of TNF, IL-6, and IL-15 [44,46,47], without necessarily increasing phagocytic capacity. It is worth noting that, as with other findings from this study, these results are not entirely conclusive due to the limited sample size. Therefore, it is uncertain whether the observed differences are simply due to individual variability or sex-specific traits. On the other hand, although the sample size is insufficient to establish definitive physiological reference values for the species, the data may still provide valuable preliminary information for potential future studies of bottlenose dolphins in controlled environments.

Regarding methodological considerations, the voluntary sampling protocol after extensive behavioral training minimized potential stress artifacts, while the multi-analytical framework addressed inherent limitations of individual statistical approaches. Large effect sizes (Cohen’s d = 1.31–2.42; Wilcoxon $[eqn]$ 0.9) were observed despite a modest sample size.

The integration of bootstrap resampling and Bayesian analysis proved useful for addressing small-sample uncertainty. Bootstrap methods provided empirically derived confidence intervals without distributional assumptions, while Bayesian posterior probabilities offered intuitive interpretation. The consistency of findings across frequentist, bootstrap, and Bayesian approaches suggests that observed effects reflect genuine biological patterns rather than statistical artifacts.

As previously mentioned, various exogenous factors, such as variations in temperature, barometric pressure, and water pH, can influence adrenocortical activity [10,24,25,26]. In our study, we observed that another factor, “physical activity”, causes increases in blood cortisol levels and, consequently, alters the functionality of certain immune system cells. It should be noted that the “physical activity” condition in this study represents a binary operational classification (presence vs. absence) encompassing training and pre-educational sessions, rather than a quantified physiological dose. Consequently, the present results should not be interpreted as intensity-specific effects but as responses associated with routine activity periods. In addition, acute cortisol peaks represent normal physiological responses, sustained elevation due to inadequate adaptation could potentially compromise the health and survival of individuals [10,24,25,26].

Likewise, cortisol elevations observed here should not be interpreted necessarily as indicators of distress, as similar increases are well documented during adaptive exercise-induced arousal [48,49]. Likewise, future research should incorporate more frequent physiological monitoring throughout diurnal cycles, including cortisol and phagocytosis assessment before and after exercise and during evening periods when cortisol typically reaches its nadir. In addition, without concurrent markers such as lactate or cytokine profiles, the welfare implications of these endocrine changes must be interpreted cautiously. Therefore, analysis of additional stress biomarkers such as lactate and complete cytokine profiling would provide deeper physiological insight, although logistical and ethical challenges of frequent sampling require careful consideration.

The findings could help in understanding the complex interaction between physiological stress and immune function. Understanding these relationships is fundamental to assessing how specific factors can influence individual physiological responses and, consequently, immune system function. Furthermore, as previously mentioned, other unmeasured variables may be influencing cortisol levels and phagocytosis parameters in our subjects. Therefore, it is crucial to replicate this study in groups at other centers with similar social and facility structures.

Finally, although the current findings are not entirely conclusive to obtain physiological values for the population due to the small sample size, they could reflect relevant trends at the species level.

5. Conclusions

This study describes significant effects of physical activity on cortisol dynamics and innate immune function in bottlenose dolphins under human care. These findings highlight the importance of integrating physiological parameters into welfare assessments. Monitoring cortisol and immune function may contribute to improved management strategies in controlled settings. Rather than defining fixed intensity thresholds, we recommend incorporating longitudinal cortisol and immune monitoring into routine management, particularly during periods of repeated or clustered physical activity, with special attention to female individuals. Likewise, it would be important to replicate the study in other centers, preferably those with social structures similar to ours, in order to make a comparison between males.

Key implications include the following:

• Importance of physiological and immunological monitoring: The evaluation of cortisol and immune function could be incorporated into the routine management of captive individuals, thus allowing for more thorough monitoring of the animal’s well-being.
• Activity Scheduling: Consideration of immune consequences when planning physical activity regimens.
• Sex Differences: Potential endocrine-immune interactions requiring further investigation.
• Methodological Framework: Multi-method statistical approaches enhance inference in limited-sample research.

Based on our findings, we propose the following practical recommendations for facilities housing bottlenose dolphins: (1) Consider scheduling intensive physical activities with adequate recovery intervals, particularly for female individuals who showed more pronounced cortisol responses; (2) implement periodic monitoring of cortisol levels alongside routine health assessments, ideally including morning baseline samples (before physical activity) and post-activity samples to establish individual response profiles; (3) when feasible, incorporate phagocytosis assays into health monitoring protocols as complementary indicators of immune status; (4) future studies should aim to quantify physical activity intensity (duration, energy expenditure) to establish dose–response relationships that could inform specific activity thresholds for optimal welfare management.

Although limited sample size ( $[eqn]$ , one male) limits generalization and prevents definitive conclusions about sexual differences, our findings indicate biologically significant trends with important implications for captive management. The consistency of results from traditional non-parametric tests, bootstrap resampling, and Bayesian analysis provides additional confidence in these observations despite sample constraints.

Future research should expand sample size and include multiple facilities with similar social structures to determine response consistency across populations, enabling more robust statistical analyses and refined understanding of how factors like physical activity influence T. truncatus physiology and immune function. Collaborative multi-institutional studies would be particularly valuable for accumulating sufficient sample sizes while maintaining the ethical imperative of minimizing invasive procedures.