Introduction

Captivity can serve as an important tool for education or population recovery of endangered species; however, captive environments drive changes in natural behavior, and it is important to understand and manage these changes1. Documented cases of natural behaviors altered in captivity range from foraging patterns2,3,4 to breeding and reproduction5,6 and social behavior7. Across species, these changes in social dynamics have been shown to impact levels of cooperation and the use of social information, leading to increased aggression, aberrant behaviors, morbidity, and mortality7,8,9,10,11,12.

One important impact of captivity on sociality is more disparate relationships in dominance8,13,14. Dominance, described as an asymmetrical relationship or distribution of behavior between two individuals15, is characterized by increased possession of a resource16,17. Furthermore, dominance status in a hierarchy is enforced by top-down intraspecific aggression between group members18,19. In captive carnivores, African wild dogs display a higher frequency of challenge for alpha than wild counterparts, resulting in increased levels of aggression between higher ranking individuals20, and dominant wolves display levels of intraspecific aggression towards pack members four times higher in captivity than in the wild19.

Unlike other carnivore species, wild prides of African lions are egalitarian, with highly symmetrical relationships between females17. Although feeding is considered the ‘most common context for aggressive competition in lions’17,21, wild female lions respect an “ownership” rule, and rarely supplant members of their same age-sex class from food, displaying no discernible dominance (increased possession/consumption of resource) or feeding hierarchy (top-down aggressive control). Lions possess formidable weaponry (teeth and claws) that ordinarily inflict costs that outweigh the advantage of gaining access to a single food source17. However, feeding habits of female lions in captivity have suggested the contrary, and dominant-subordinate relationships regarding food resources have been heavily observed by professionals, but are yet to be reported in the literature (professionals in the field: Hildegard Pirker, Dr Christine Steyrer, Kevin Richardson, Dr Jessica Burkhart). Hence, in the artificial confines of captivity, dominant-subordinate relationships and associated aggression arising in a naturally “egalitarian” species could potentially pose welfare and management issues17. Therefore, we formally tested whether unnatural dominance relationships are present in captive female lions, and, if so, investigated a potential ameliorative solution.

Administration of oxytocin, a neuropeptide highly implicated in social cognition22,23, has been shown to flatten dominance hierarchies in rodents and primates24,25,26. Intranasal oxytocin increases behavioral synchrony and efficiency in communicating dominance status, and reduces differences in related social behaviors between dominant and subordinate male macaques24, and, interestingly, oxytocin increases the social status of subordinate female macaques over males by increasing their (dominant) threatening and vocalization behaviors25. Previously, we have found that intranasal administration of oxytocin decreases vigilance and increases affiliative behavior in African lions in a non-feeding context27,28, however, the impacts of intranasal oxytocin have shown sex- and context-dependent effects (Ma et al. 2018). In the current study, we sought to further investigate the effects of oxytocin in the specific context of competition during feeding in female African lions to see whether intranasal administration of oxytocin could, in fact, restore symmetry to female lion relationships, reinstating the “ownership” rule. We hypothesized that: A) When presented with a food resource, a dominant-subordinate relationship between captive pairs of female African lions would present as consistent asymmetries in possession, and that B) Oxytocin would ameliorate the difference in pairwise dominance relationships by closing the gap between differences in resource possession, and reduce the subordinate lion’s tendency to behave submissively, consistent with the “ownership” rule.

Methods

Subjects

All trials were performed at Lionsrock Sanctuary FOURPAWS, in South Africa between February and October 2023. Procedures were approved by the Institutional Animal Care and Use Committee of the University of Minnesota, and in accordance with ARRIVE guidelines. All subjects (n = 20 females, 10 pairs) were captive born, healthy female adults in prime condition (age 4–18), housed in greater than 1 hectare, species appropriate enclosures at their current location for greater than one year. Participants were selected by sanctuary staff based on health and pair availability. All 20 female lions were included equally across trials and analyses. No studies were performed on females who were lactating, pregnant, or in estrus, and no animals in this study are used for breeding purposes. Pairs were selected for the study that had been housed amicably together (outside of a feeding context) since early adolescence or birth (direct levels of relatedness are unclear due to the nature of rescues), and all individuals were similar in age and condition within each pair. Pairs of females came from mixed group compositions and were separated into pairs for the study only on the days trials took place. Pairs from groups containing multiple females were selected randomly based on the animals that the handler separated together from the first baseline trial, and these were maintained throughout the trial period. All animals in the study were pre-trained to take meat blocks (approximately 0.1 kg per block) from the researcher at the gate (in the same manner they are accustomed to being fed meat blocks by the sanctuary staff) prior to study; any animals that showed hesitancy to approach individually were excluded from the study, and all animals participated spontaneously. Two baseline trials were performed to habituate the lions to the researcher and the experimental protocol. The lions are fed twice a week, and all trials were performed on non-feeding days, with no enrichment given prior to trial to ensure the animals were motivated by the resource and not satiated by feeding earlier in the day.

Trial types

Dominance can be characterized as monopolization over a resource17. In this study, we use food in the form of blocks of raw beef as a resource (meat was provided by the sanctuary and was within feeding guidelines for each animal). Two baseline trials were performed to habituate the lions to the researcher and the study procedure. The lions are fed twice a week, and all trials were performed on non-feeding days, with no enrichment given prior to trial to ensure the animals were interested in the resource and not full from feeding or distracted. Asymmetry in resource possession and changes in aggression were measured.

Each trial day consisted of two tests (neutral and non-neutral). First, the neutral test determined whether a dominant individual presented within the pair of females (determined by possession of resource). This was directly followed by the second test, the non-neutral test, to determine if dominant individuals would respect the ownership rule or if they would overtake the resource from the subordinate. The set of tests was administered consecutively on a trial day, and repeated on a total of four days (two days with oxytocin treatment and two days with saline control, in alternating order), with 10–14 days between (differences based on weather conditions). The possessor of the resource was identified as the animal that ultimately consumed the 0.1 kg piece of meat.

Trial protocol

Treatment (10 IU Oxytocin) or vehicle (saline, as control) was administered intranasally 45 min prior to trial commencement (presentation of resource), and both individuals in the pair received the same treatment at the same time. Oxytocin or saline was administered in alternating order across the four treatment days, with half of the pairs receiving oxytocin first, and the other half receiving saline first in order to control for order bias.

To commence the trial, the female pair was brought to the gate being lured by meat (Fig. 1). For the neutral test, the resource (meat) was placed on a stick against the fence at an equal distance between the two females, giving them equal opportunity to approach and possess the “resource.” The meat block was placed on the stick to make it difficult to pull off and consume (possess), therefore giving each lion time to supplant the first individual that tried to take “ownership” by biting the meat with approximately 30 s between each round to allow for a reset. For each treatment for each pair, meat was presented a total of 10 times occurring on two separate days (except in rare cases when one female was reluctant to return, usually after receiving excessive aggression, then only repeated 6 times as a humane endpoint). A lioness was considered “dominant” if she monopolized (consumed) the resource at least 70% of the time [note: in all but one pair the ratio was 90–100%]. Neutral tests were directly followed by the non-neutral tests in which an additional set of 5 rounds of meat presentation was administered per treatment day (total of 10 rounds). In non-neutral tests, the meat was held closer to the subordinate, as determined by the neutral test, giving the subordinate a chance to reach the meat first and claim “ownership.” It was then observed whether the dominant female allowed the subordinate to maintain possession, or if she overtook the resource regardless of the “ownership” rule normally observed in female lions in the wild.

Fig. 1
figure 1

Dominance in female African lions as determined by possession of the food resource. A high value resource (meat) is held at the fence allowing two females to compete for the resource. A) one female behaves submissively while the other dominates the resource. B) the possessor of the resource defends herself, behaving less submissively than in (A) to maintain possession of the resource.

Full size image

Pairs were selected randomly to receive oxytocin or saline in alternating order for a total of 4 trial days. The dose was 10 international units (IUs) of oxytocin administered intranasally via a DeVilbiss atomizer. Intranasal administration was achieved by placing the tip of the atomizer 1 cm into the subject’s nostril27. One spritz from the atomizer equals one international unit. Intranasally administered oxytocin bypasses the blood–brain barrier, peaking at 45 min and lasting ≈ 4 h29,30. Trials therefore commenced 45 min post oxytocin administration and lasted approximately 15 min. Treatments were prepared by research assistant, Dubois, so that the researcher, Burkhart, was blinded to the treatment in order to control for biasing behavior, interpretation, or interaction during trials. All trials were scored live by two researchers and also recorded using a Gopro Hero 5 and Samsung s20 5G + for later reference.

Behavior was measured during each round to determine dominance as follows: who possessed/consumed the resource, as well as aggressive vocalizations and physical actions (activation of weaponry: teeth or claws). The occurrence of vocal and physical aggression was scored as either 0 (absent) or 1 (present). Aggressive encounters were then scored and categorized by severity, 1 = brief non-contact, 2 = brief contact, 3 = prolonged contact. This number was considered the “aggression level”. A total aggression score was calculated by summing the 0/1 vocal score + the 0/1 physical score + the 1–3 level of aggression. The total aggression score could therefore range between 0–5. Also measured was how many body lengths apart the individuals were while the resource was consumed.

Statistical analysis

Differences in resource possession

Successful possession of the food resource by dominant vs subordinate lions was our primary indicator for the presence of dominance relationships within pairs. We therefore compared the ratio of within-pair resource possession across trials. Binomial generalized linear mixed models (GLMMS) were employed using “glmer” (R-package lme4) to model the likelihood of dominant and subordinate individuals to successfully possess the food resource. This “possession” likelihood was modeled as a function of treatment (saline or oxytocin), dominance (dominant or subordinate individual), and the interaction between treatment and dominance. We considered neutral vs non-neutral trials as separate tests and ran these data in separate models. Pair identity was added as a random effect to our mixed models to account for repeated measures of lion pairs across trials.

Differences in aggressive behavior

We explored differences in aggressive behavior and intensity across trials using 4 aggression metrics, as changes in aggression for both dominant and subordinate individuals was a likely mechanism for changes in pair-wise resource possession. We first modeled the presence/absence of vocal aggression (1) and physical aggression (2) using binomial GLM models, and then modeled physical aggression intensity level (3) and total aggression count (4) using poisson GLM models. Data from neutral vs non-neutral trials were again modeled separately, as was dominance status for easier interpretation of effect sizes. All aggression metrics were modeled as a function of treatment (saline or oxytocin), with pair identity used as a random effect to allow for pair and individual differences in displayed aggression.

Results

Asymmetrical relationship pairings in female African lions were first determined, then tested, in a sanctuary setting after administration of oxytocin or vehicle. Two types of trials were designed; first, a neutral trial with equal opportunity to possess the meat resource, in order to establish dominance, and second, a non-neutral trial where the subordinate animal had first right of possession to test oxytocin’s effect on the “ownership” rule.

Resource possession in neutral trials

In neutral trials (designed to test dominance over equal opportunity possession), we observed during control (saline) treatments that across all pairs, one consistent individual (hereafter named the “dominant”) possessed (consumed) the resource on average 94% of the time (range of 70% – 100%) (Fig. 2). Results of binomial models show that dominant individuals were significantly more likely to possess the resource across neutral trials (Table 1; log odds: −5.3 ± 0.59, p < 0.0001).

Fig. 2
figure 2

Asymmetrical resources possession, demonstrating dominant-subordinate relationships, decreases with oxytocin (n = 10 pairs, 380 rounds). Data show mean within-pair proportion that dominant or subordinate individual possessed the resource under neutral (N) and non-neutral (NN) treatments. Error bars show standard error. In all treatments, the subordinate individual was less likely to possess the resource than the dominant (Binomial GLMM β: −5.3 ± 0.59, p < 0.0001 for neutral, β: −1.98 ± 0.3, p < 0.0001 for non-neutral). However, in non-neutral trials testing the “ownership” rule, the subordinate displayed a significant increase in maintaining possession after oxytocin (β: 1.2 ± 0.4, p = 0.01), decreasing asymmetry in pairwise dominance relationships.

Full size image
Table 1 Binomial GLMM models for resource possession. Models were run separately for neutral vs non-neutral trials. All models included pair # as a random effect: Possession ~ Treatment * Dominance + (1|Pair). Models used reference categories for each coefficient, and contrast levels are listed in parentheses for each variable. Significance indicated by (✧ > 0.10; 0.05 > *; 0.001 > **).
Full size table

Post oxytocin, dominant animals possessed the resource on average 89% of the time (range of 40% – 100%). Subordinate lions were 4.32 times more likely to maintain possession of the resource post-oxytocin compared to control, but this difference was only marginally significant (Table 1; log odds: 1.46 ± 0.75, p = 0.052).

Resource possession in non-neutral trials

In non-neutral trials (designed to bias possession toward the subordinate in order to test the “ownership” rule), during control treatments dominant individuals still possessed the resource on average 74% of trials (range of 30% – 100%), and dominant individuals were again significantly more likely to possess the resource (Fig. 2; Table 1; log odds: −1.98 ± 0.3, p < 0.0001). However, post-oxytocin, dominant individuals possessed the resource on average 61% of the time (range of 10% – 100%), and subordinate lions were 3.05 times more likely to maintain possession of the resource compared to saline trials, significantly increasing their ownership (Fig. 2; Table 1; log odds: 1.2 ± 0.4, p = 0.01).

Differences in aggressive behaviors across trials

Chi-sq tests were used to test for differences in body-length proximity score (BL score) of lion pairs during trials. No differences were found in BL score in either neutral (chi-sq test: df = 3, p = 0.42), or non-neutral trials (chi-sq test: df = 3, p = 0.17).

We found no significant differences in subordinate or dominant lion aggression across neutral trials (Table 2; Fig. 3). However, in non-neutral trials dominant lions were less likely to display vocal aggression post-oxytocin (Fig. 3a; log odds: −1.0 ± 0.37, p < 0.01), while subordinate individuals were more likely to display physical aggression (Fig. 3b; log odds: 1.1 ± 0.47, p = 0.02), displayed greater aggression intensity (Fig. 3c; odds: 0.36 ± 0.14, p = 0.01), and increased overall aggression count (Fig. 3d; odds: 0.93 ± 0.29, p = 0.001).

Table 2 GLMM models for Effects of Oxytocin on Aggression Intensity. Models were run separately for neutral vs non-neutral trials and for submissive vs dominant individuals. We used binomial models for presence of vocal and physical aggression (estimates reported as log odds), and Poisson models for aggression level and total aggression count data (estimates reported as odds). All models included pair # as a random effect: Aggression ~ Treatment + (1|Pair). Models used oxytocin as the Treatment reference category, and indicate the effect of oxytocin. NA values indicate model was unable to converge. Significance indicated by (✧ > 0.10; 0.05 > *; 0.001 > **).
Full size table
Fig. 3
figure 3

Subordinate female lions increase aggression in order to maintain possession of resources after oxytocin administration during non-neutral ownership trials. a-d) Changes in displayed aggression across trials. a-b) mean presence of vocal or physical aggression; c) mean level of physical aggression, and d) mean total aggression level. Error bars show standard error. e–f) effect sizes from Binomial and Poisson GLMM models indicating significant differences in both dominant and subordinate aggressive behavior post oxytocin (* p < 0.05, ** p < 0.005), subordinates increased occurrence of physical aggression, mean aggression level, and total aggression, while dominants decreased vocal aggression thus indicating context-dependent effects of oxytocin administration.

Full size image

Discussion

Female lions in the wild are unique in their symmetrical, egalitarian behavior. However, here we observed that, in captivity, there are strong dominant-subordinate relationships between female African lions, as indicated by asymmetry in possession of a high-value resource. During neutral trials, where both members of a pair were offered equal opportunity for possession, a “dominant” individual possessed the resource an average of 94% of the time.

Interestingly, however, we found that these asymmetrical relationships decrease with intranasal administration of oxytocin. Post oxytocin, subordinates maintain possession of the resource significantly more often during non-neutral trials, which were designed to bias possession towards the subordinate in order to test the “ownership” rule observed in wild lions. Post oxytocin treatment, subordinate individuals also behaved less submissively by significantly increasing aggressive defense (level of physical aggression, total aggression level, and mean total aggression) in order to maintain the resource, while dominant individuals decreased in vocal aggression—their most common form of aggression, highlighting the potential for oxytocin to elicit differing context-specific effects.

Mech (1999) proposed that differences in social-dominance exist because free-roaming familial groups are able to disperse, where unrelated pairings or groups are forced to stay together for years in captivity. Captive environments hardly come close to mimicking lions’ natural home range or lifestyle5,31, and both size and quality of the physical captive environment play a crucial role in lion behavior. Less natural environments lead to increased abnormal behaviors5,11, and the inability to behave naturally, such as regulating social dynamics during carcass feeding, negatively impacts levels of aggression and relationship dynamics in captive lions21. Specifics of group dynamics, such as group size19 and sex ratios32 can also have a significant effect on the level of affiliative or aggressive behaviors displayed among captive group members. An array of factors likely leads to the unnatural formation of dominance relationships in female lions as we have observed here. While some species, such as primates, have been shown to employ protective strategies including tension-reduction and conflict-avoidance models to mitigate levels of aggression generated by the constraints of captivity7,33, the details of these models have not been well studied in captive carnivores.

The formation of unnatural asymmetrical dominant-subordinate relationships in captive female lions could, indeed, be a product of tension-reduction or conflict avoidance models brought on by captivity. We see here that oxytocin can be used as a potential tool to intervene. One possible explanation for this may be oxytocin’s ability to attenuate stress-induced social avoidance caused by dominance34, and regulate fear responses, promoting defense behaviors35,36. Specifically, oxytocin has been shown to attenuate response to conditioned fear37,38, which could be beneficial for managing continual forced subordination in a closed (captive) environment. However, the current study only analyzed relationships between pairs of individuals, which could change in relation to the remaining group members. Group level dynamics and the specific functions for why these social dominance relationships have developed in captivity should be further investigated before the implications of altering them with oxytocin can be fully understood.

Differences in group dynamics are further emphasized by the idea that specific personality traits drive behavioral variance and stress levels in captive big cats39,40. The difference in personality traits and affiliated stress levels likely impact the expression of dominance in captivity, as stress levels are directly linked to dominant-subordinate relationships34,41. Research has shown that the effects of oxytocin may differ among individuals with altered baseline states42, and here we show that oxytocin has contrasting behavioral impacts on dominant vs subordinate behavior. Importantly, these findings further our understanding of oxytocin’s specific effects and potential applications considering specific aspects of pairwise relationships and in varied contexts, helping to clarify the perception of oxytocin having a negative or “dark” side43, and emphasizing the importance of carefully considering differences in personality and related behavioral traits, as well as social-relationship dynamics, before administering oxytocin.

Furthermore, the social salience hypothesis emphasizes oxytocin’s role in enhancing attention to cooperative vs. competitive environments44. Oxytocin can both facilitate prosocial behavior in the presence of a positive social stimulus45 or increase protective responses towards adverse stimuli46. While oxytocin is mostly known for its ability to promote prosocial behavior, it has been shown to increase self-serving and defense-motivated behaviors when protecting against self-vulnerability in a competitive context4447,48;, which is in line with our current findings. While our previous work emphasized the cooperative aspects of the salience hypothesis by highlighting the importance and effectiveness of providing positive stimulation during socializations of African lions28, in the current study, we tested the competitive aspects by creating an environment of aggressive competition over a high-value resource, in order to obtain a more complete perspective of oxytocin’s potential to manage carnivore behavior across contexts.

In our previous work27, we also found that oxytocin reduced vigilance (number of roars) in response to out-group threats (roars) and increased in-group affiliative and tolerance behavior (proximity to each other and not play object). However, we did not find that oxytocin administration had an impact on observed social behaviors in response to a high-value food item. Those trials, which differed from the current study in several key parameters (group size, sexes, space allocation, controlled approach to resource), were not designed to elicit sufficient changes in ownership to allow us to identify the effects observed here. We consider the current results complementary to that prior work. Furthermore, we have previously found that repeated administration of oxytocin in pairs and groups of lions in non-feeding contexts lead to lasting prosocial effects28. While the current study revealed that, after a single dose, oxytocin administration had immediate, distinct impacts on dominant and subordinate counterparts during competition over a resource, dominance relationships were thereafter restored (as evident by following trials), and therefore, the long-term effects of continued administration of oxytocin on relationships within a competitive feeding context should be investigated further.

Aspects driving social differentiation in group-living species are multifactorial15,49, and captivity adds an additional array of challenges with a multitude of issues which potentially contribute to the formation of hierarchical societies21,31,50,51,52. Further investigation is required to understand all the ways in which behavior is manipulated by an environment, and how oxytocin may directly impact these behaviors. Nonetheless, this study implicates the use of oxytocin as a recovery tool for the formation of unnatural behaviors in captivity, by ameliorating asymmetrical relationships and restoring the “ownership” rule in female African lions.