Philosophical Transactions of the Royal Society B: Biological Sciences

Climate change is increasing wildfire risk globally and peatlands have become increasingly vulnerable to fire in recent decades. We combine social research methods with analysis of historical ecological (palaeoecological) records to understand links between fire, climate, vegetation and land use. Stakeholders in the Peak District (UK) highlight the need for scientific research and local knowledge to be more frequently embedded into policy. Analyses of historical ecological datasets reveal regime shifts in moorland vegetation following periods of fire activity and managed burning. This type of disturbance can lead to dominance of grasses, which may have a negative impact on peatland carbon balance under warmer climatic conditions. Recent fires are contributing to the loss of Sphagnum moss and greater dominance of heath and grass. Restoring peatlands, through re-establishing native woodland and improving peat bog hydrological conditions, alongside careful planning around controlled burning, are key measures to enhance resilience to future fire events.

Drumming--rhythmic, percussive sound production using body parts or external objects--is rare among non-human animals, with confirmed tool-assisted cases previously limited to primates and Palm Cockatoos. Here, we report the first documented instance of spontaneous, tool-assisted drumming in a Galah (Eolophus roseicapilla). A captive, male Galah produced rhythmic tapping by striking a coconut shell against a metal bowl. Across 14 recorded sessions, the bird displayed consistent temporal structure characterised by two stable tapping rates (approximately 0.8 s and 0.2 s inter-onset intervals) arranged into recurring phrases. This pattern indicates a simple hierarchical rhythmic organisation with a 4:1 ratio between metrical levels. The birds behaviour emerged without training, apparent reinforcement, or known exposure to conspecific or human drumming models, suggesting an intrinsic capacity for rhythmic tool use. Although the function of the behaviour remains unclear--play, nutrient extraction, or communicative signalling--these observations extend known rhythmic and tool-using abilities within cockatoos and raise new evolutionary questions. Our findings highlight the potential for rhythmically structured, instrumental behaviour to arise in a broader range of avian taxa than previously recognised, motivating further comparative and experimental work on the cognitive and biomechanical foundations of drumming in parrots.

A populations social structure, often represented as a social network, shapes fundamental biological processes including the spread of disease, information, and behaviour. The Friendship Paradox is a network phenomenon whereby the average individual has fewer friends than their friends do. This effect can be quantified as relationship disparity (the difference between an individuals connectedness and those they are connected to) which captures the local social environment. Previous work has shown that such relationship disparity can be exploited in effective outbreak monitoring, targeted health interventions and optimized contact tracing. Yet, how its magnitude varies across real-world social networks remains poorly understood. Here, we analyse relationship disparity across 391 empirical animal social networks to test how intrinsic network properties and biological attributes predict its extent. We find that smaller and sparser networks exhibit stronger relationship disparity, and that mammalian and avian social systems generally showed stronger relationship disparity than reptilian systems. After controlling for variation in individual sociability, mammalian and reptilian social networks displayed weaker relationship disparity than expected based on network structure alone. Together, these findings demonstrate that both network structure and biological attributes shape relationship disparity in natural social systems, providing a foundation for predicting how higher-order network architecture influences social processes such as contagion. Significance StatementIn natural populations, social connections are unevenly distributed, often resulting in a small subset of individuals that are highly connected while many are relatively peripheral. The Friendship Paradox is a measure of relationship disparity between individuals and their local social environment. Understanding how features of the social network and biological system are associated with relationship disparity can contribute to understanding what shapes social behaviour. Relationship disparity may not just be an emergent network property but could reflect a higher level of social structuring, and therefore shape processes that depend on social contacts. Our findings demonstrate the value of comparative network analysis for generating insights into fundamental principles structuring real-world societies.

Interpersonal neural synchrony (INS) in mother-child dyads is often interpreted as a neural marker of relational quality and sensitive caregiving, yet findings on its predictors remain heterogeneous. One possible source of this variability is the diversity of interactional paradigms used in hyperscanning research. This study examined how maternal personality, child temperament, and affective states relate to INS across interaction contexts varying in social interactivity. Thirty-three mother-child dyads (n = 20 female children) participated in a functional near-infrared spectroscopy hyperscanning experiment involving passive video co-exposure, a structured cooperative task, and free interaction. Fronto-temporal activity was recorded simultaneously, and INS was computed using wavelet transform coherence. Above-chance levels of INS emerged in inter-brain region combinations primarily involving the mothers left inferior frontal gyrus (IFG) and the childs right IFG (adjusted ps < 0.030, Cohens d range = 0.14-0.31). Maternal neuroticism was the only significant predictor of INS, with higher levels associated with increased synchrony during passive video co-exposure (adjusted p = 0.012) and free interaction (adjusted p = 0.021), but not during the structured game. These findings indicate that maternal dispositional traits shape INS in a context-dependent manner. Notably, the positive association between neuroticism and INS suggests that heightened neural synchrony may reflect over-attunement in more anxious caregivers, rather than optimal coordination. Excessive synchrony may therefore index tightly coupled, over-monitoring interaction dynamics, consistent with models of affiliative vigilance in anxious parenting. Overall, INS may follow a non-linear pattern in which moderate levels are most adaptive, highlighting its flexible, dynamic, and context-sensitive nature.

Pretend play is a hallmark behavior in childhood where children create nonliteral meanings. Empirical data supporting the role of social cognition and the decoupling from literality are still scarce during early development. We explored here how the comprehension of pretense affects the visual exploratory behavior of toddlers (n = 44) and adults (n = 65) when they were exposed to short video clips in which an actress performed either real actions (e.g., eating jelly) or pretend actions (e.g., pretending to eat with imaginary food), while varying the complexity of those actions. We analyzed participants exploration of the face in the videos as exploitation of social information. We showed that all observers paid more attention to the face in pretend scenarios than in real ones, measured as longer total looking time in adults and more fixations and revisits to the face in both age groups. We also found more gaze shifts (a measure of information sampling) between the face and the moving hand in the pretend videos in both age groups, mainly at the initial stages of the actions. Additionally, analyses of the scanpaths structure using gaze entropy showed less order in the exploration of pretend videos in both age groups, suggesting that pretense involved greater uncertainty and increased information seeking. The less structured trajectories were observed again mainly in complex pretend scenarios. Taken together, our gaze results indicate that from its developmental origins, the comprehension of pretense relies on social processes linked with information seeking and exploration. Significance StatementDevelopmental theories have long debated whether pretend games are born in conjunction with social capacities in the second year or become integrated later in life. Our study shows that, much like adults, toddlers visually explore pretend scenes gathering more social information and in a less structured manner compared to real-world scenarios, suggesting that the emerging capacity to play with the meaning of things is linked with that of thinking of other minds early in life.

Cannibalism is among the strongest and most widespread food taboos in human societies, yet archaeological, ethnographic, and historical evidence indicates that it has repeatedly emerged across diverse human populations. This coexistence of recurrent practice and persistent prohibition raises a fundamental question: when does cannibalism become adaptive, and why is it typically suppressed? We address this problem using a formal model that treats cannibalism as a potential food source subject to energetic benefits and multiple sources of cost. Nutritional gains are modelled using a saturating function of caloric intake, while costs arise from acquisition, digestion, and infection. Infection costs are represented as a stochastic process whose mean increases with the length of the trophic transmission chain, capturing the risks associated with repeated within-species consumption. Analysing the expected energetic balance across levels of food availability and cannibalism order reveals narrow ecological conditions in which cannibalism yields a positive expected balance and broader conditions in which it is strongly disfavoured. The model provides a framework for interpreting archaeological and ethnographic findings by specifying boundary conditions and identifying the most probable ecological scenarios under which different forms of cannibalism are expected to occur. The results predict that cannibalism is most likely under extreme resource scarcity, when acquisition costs are low and infection risks are constrained, while sustained or high-order cannibalism rapidly becomes unviable due to escalating infection costs. Overall, the findings suggest that cannibalism is best understood as a conditional trade-off rather than a behavioural anomaly, with cultural taboos functioning as adaptive responses to nonlinear epidemiological risks.

The growing field of comparative thanatology aims to shed light on how and why the understanding of death evolved. Observations across different nonhuman primate species have reported care-taking behaviour of dead infants, but also cannibalism. Various hypotheses have been put forward to explain these infant-directed behaviours, ranging from responses to infantile cues to an understanding of death. To aid comparative analyses and test some of these hypotheses, we report behaviours directed at dead infants in a wild population of Guinea baboons (Papio papio) living in the Niokolo-Koba National Park, Senegal. During 12 years of field observations (2014-2025), 67 infants died before reaching 1 year of age. In 4 cases, we could not establish when the infants had died because the field station was closed due to the COVID-19 pandemic. In 22 of the remaining 63 cases, mothers, but occasionally also other group members, carried, protected, groomed, and dragged dead infants. In 6 cases, cannibalism occurred. None of the mothers expressed any signs of emotional distress in response to infant death. We suggest that a concept of death in Guinea need not be invoked to explain the observed behaviours. Instead, selection appears to have favoured post-mortem caretaking behaviours to avoid abandoning an infant that might temporarily be unresponsive. The lack of infant responses to maternal behaviour and the disintegration of the corpse may drive the transition from perceiving the infants as an object that evokes caretaking to one that resembles food, which ultimately facilitates occurrences of cannibalism.

Infant corpse carrying is widely observed in chimpanzees and bonobos, yet its underlying mechanisms remain debated. Analysing 83 published cases using Bayesian mixed-effects models, we show that ICC duration varies with infant age at death, cause of death, and site-level interbirth intervals, with longer carrying following disease-related deaths, older infant age, and slower life histories. These results suggest that variation in infant corpse carrying duration is parsimoniously accounted for by the persistence of maternal behavioural systems being modulated by carrying risk, dyadic bond strength, and life-history context rather than by mothers recognising death as an irreversible biological state. Given the close evolutionary relationship of Pan and Homo, this implies that the complex cognitive frameworks required to recognise deaths finality likely emerged in the hominin lineage after divergence from the Pan-Homo last common ancestor.

Global populations are ageing at an unprecedented rate. For many diseases, age is a strong indicator of susceptibility, morbidity, or mortality. Principles of evolutionary biology can be leveraged to understand how pathogens may optimally exploit new populations, and the impact of this on the global burden of infectious-disease-induced mortality. We parameterised an age-specific R0 model with 2017 epidemiological data on Measles, Tuberculosis, Meningitis, and Ebola, and age-specific demographic estimates for 2017 and 2050, for the seven Global Burden of Disease super-regions. We explored the theoretical trade-offs between pathogen virulence & transmission, and virulence & host recovery, parameterising trade-off parameters using Latin Hypercube Sampling. Population ageing between 2017-2050 saw an increase in virulence induced mortality in four settings: 1) Ebola in sub-Saharan Africa, 2) Measles in central/eastern Europe & central Asia region, 3) Measles in North Africa & the Middle East and 4) Tuberculosis in the central/eastern Europe & central Asia region. The decrease in infection duration due to an increase of elderly people drives pathogen virulence down for diseases in the remaining settings. Understanding the mechanisms that shape pathogen dynamics and leveraging this to predict future challenges is key to endemic disease management in a rapidly changing world. Author SummaryKey aspects of disease transmission including susceptibility to infection, the severity of infection, and the probability of dying from that infection, are affected by host age. Global populations are rapidly ageing, so that the mean age of most populations is generally higher than it used to be and is set to continue on this trajectory. This suggests that the dynamics of infectious diseases are also likely to change, although infectious disease dynamics tend to be non-linear as these key parameters interact. We have developed a dynamic modelling framework to explore how changes in population age structure might impact the optimal level of pathogen virulence in a population. We have chosen four infectious diseases as case studies, that differentially impact certain age classes to illustrate these dynamics. We have parameterised this framework with open access data for each of the seven Global Burden of Disease super-regions and show that population ageing can increase virulence for several diseases in differing global regions, whilst increased background rates of mortality can drive virulence down in others.

A diversity of intelligences arises from the constraints under which animals evolve. However, characterizing how constraints shape intelligence is challenging because it requires relating the restrictions on cognitive mechanisms to those that affect their evolution. We demonstrate the potentially complex interaction between constraints by considering the case study of working memory. Here, information-processing capability is limited by the storage capacity available to hold representations, and the degree of control over those representations. We present an evolutionary model that is mechanistically detailed enough to capture the interactions between capacity and control. This allows us to make quantitative predictions about the distinct patterns of information processing that might be observed across animals. Further, our models cognitive detail allows us to fit recall performance on the retro-cue task, illustrating how model predictions can be tested by comparing humans and rhesus monkeys (Macaca mulatta). We find that capacity and control are synergistic and amplify each others effects. However, evolution prioritizes investment in capacity because it is required for control to be effective. The strength of synergy varies due to interactions between these cognitive components depending on task complexity, cue reliability, and the availability of metabolic energy. Consequently, our model predicts diversity in investment in capacity and control across animals, and identifies a small number of regimes into which lineages could evolve. We discuss how the computational structure of tasks exerts selection on cognitive designs. Significance StatementTheories about the cognitive processing required for intelligence have been developed largely independently from analyses of evolutionary pressures. This produces an impediment to understanding the diversity of intelligences across animals. We present a mathematical model that bridges these two theoretical domains, focusing on working memory. Our analysis reveals a rich interaction between the capacity to store information and the ability to control that process. We predict that animal intelligences fall into a few major cognitive regimes: initially capacity is prioritized, then capacity and control increase together due to synergy, and finally capacity returns to the fore as adding further control yields diminishing returns. We demonstrate our model makes predictions that can be compared against empirical data via cross-species experiments.

Humans have a remarkable ability to act efficiently and accurately in familiar situations while remaining flexible in novel circumstances. Nonparametric contextual inference has been proposed as a computational principle that can model how agents achieve flexible yet stable behaviour in dynamic and possibly unknown environments. However, it remains an open question how humans learn, deploy and reuse stable contextual task representations so efficiently. To address this question, we propose the nonparametric Bayesian Contextual Control (NP-BCC) model, which integrates nonparametric contextual learning with two well-established cognitive mechanisms: repetition-based automatisation and schema-like prior knowledge. These two mechanisms are assumed to support behavioural stability and facilitate novel task acquisition. Simulations in dynamic multi-armed bandit tasks of increasing difficulty illustrate how the NP-BCC can acquire and reuse contextual task representations, with the proposed mechanisms operating in the intended, functionally meaningful manner. Specifically, we show via simulations that automatisation not only enhances task performance but also stabilizes contextual inference and structure learning, while structured prior knowledge accelerates the acquisition of novel contexts. We discuss the implications of our findings for computational accounts of adaptive behaviour and contextual learning, and outline directions for future empirical work, including investigations of context-dependent behavioural dysregulation relevant to conditions such as substance use disorders. Author summaryPeople are very good at repeating well-learned actions in familiar situations, but they can also quickly adjust their behaviour when circumstances change. How the brain balances stability and flexibility is still not fully understood. There is growing evidence that the brain organizes experience into different "contexts", which are mental representations of encountered situations. Computational models based on this idea can in principle reproduce flexible behaviour, but they often become unstable in complex environments. To improve stability, we borrow two simple strategies from everyday human behaviour. First, people tend to repeat actions that have worked well before. Second, when facing something new, they often reuse strategies from similar past situations. Using simulations, we show that combining these strategies with context-based learning produces more reliable behaviour in the model. Prior experience helps the model understand new situations more quickly, while repeated actions help stabilise behaviour once a situation becomes familiar. Taken together, our findings show how such mechanisms can give rise to both flexible and stable behaviour in the model.

Movement is costly, and animals are under strong selective pressure to move efficiently, yet, in patchy, dynamic landscapes, decision-making is inherently uncertain. We quantify the energetic savings achieved by using up-to-date information presented within social cues for reducing movement costs. We use an agent-based model, founded on realistic aeronautical rules and parametrised on the Andean condor (Vultur gryphus), to study movement in patchy landscapes. By explicitly considering altitude, flight results in a sequence of soaring and gliding in the 3D space. We investigate how the cost of movement to an overall goal varies when birds use social information from others that are either fixed in space or moving collectively to the common goal, and under different risk-taking speed strategies, from slow and cautious to fast and risky. The value of social information is operationalised as energetic savings in units of basal metabolic rate. Under low predictability, agents with intermediate risk and high social-information use exhibit lowest movement costs, with up to 41% energy savings over asocial movement. By extending classical aeronautical theory to social and variable environments we demonstrate the adaptive value of social information for efficient movement in patchy, unpredictable landscapes.

The emergence of tolerance between distinct social groups is a central question in social evolution, with relevance to both nonhuman primates and early human societies. Ecological factors such as resource heterogeneity and social threats like bachelor male presence have each been proposed as drivers of intergroup proximity, but their combined effects remain unclear. We developed a spatially explicit agent-based model to examine how resource patchiness and bachelor threat jointly shape aggregation dynamics and intergroup tolerance among mixed-sex groups. In the model, groups forage across heterogeneous landscapes and expand their ranges with increasing patchiness. Bachelor males roam independently and pose localized threats, prompting groups to aggregate probabilistically according to threat intensity. Aggregation decisions follow a sigmoid response, and familiarity accumulates through repeated overlap or joint aggregation. Intergroup tolerance thus arises from encounter histories rather than being preprogrammed. Simulations show that resource heterogeneity promotes tolerance by increasing overlap and encounter frequency, while bachelor threat induces aggregation as a protective response. Tolerance can also emerge without consistent aggregation, provided that ecological conditions repeatedly bring groups together. When both heterogeneity and threat are high, aggregation and familiarity peak, indicating a synergistic effect. These outcomes are robust across a wide parameter space and do not require explicit coordination or cooperative intent. Our findings highlight how simple behavioral rules embedded in ecological and social contexts can yield complex intergroup outcomes, offering a general framework for understanding the evolution of intergroup tolerance in primates and humans.

Orca, wolves, chimpanzees and humans share a similarly impressive capacity for group hunting, where individuals coordinate behaviour together to capture prey. Studying hunting behaviours has important implications for understanding how behaviour in group contexts may be indicative of cognitive decline. Despite growing interest in brain circuits for prey capture, the brain regions involved in tracking prey during a hunt and the behaviours in group hunt linked to success remain unclear. Here we combined functional near infrared spectroscopy (fNIRS) and a virtual minecraft world to examine behaviour, brain dynamics and brain synchrony involved in group hunting behaviour. We focused on the prefrontal cortex (PFC) due to its known role in planning and social coordination and recorded from pairs of individuals as they either cooperated to hunt another person (prey) or simply followed another person. Hunters were more successful if they managed to keep a smaller distance to the prey and moved at speeds that were more synchronised with their co-predator. At high-range frequencies for fNIRS (0.1-0.2Hz), we found greater brain-to-brain synchrony in lateral and medial (frontopolar) PFC regions during hunting compared with chance levels. Together, these findings provide insights into what behaviours and brain dynamics associated with successful group hunting.

Collective decision making is a fundamental aspect of group behavior in both animals and humans, and often involves reaching a consensus on the best of n options, using empirical evidence. Although many parallels have been drawn between human and animal collective decisions, collective human behavior is rarely studied in the type of embodied scenarios that animals are often faced with. In this study, we placed human groups in a virtual setup similar to nest site selection in social animals, in which they explored a shared environment and reached a consensus based on their observations of empirical features. In groups of up to 10, participants had to reach consensus on the empirically largest of four candidate sites without verbal communication, instead using movement-based interactions in a custom-developed 3D virtual environment for online multi-participant experiments. The results showed that the speed and accuracy of consensus was importantly modulated by perceptual difficulty and information availability, but that no speed-accuracy trade-off was present. Participants attempted to reach consensus on the empirically largest site by flexibly adapting their use of social information to perceptual difficulty, their spatial position, and the time already spent supporting some option. When a minority of informed individuals were present, these individuals exercised greater independence and influenced the group to faster and more accurate consensus. These results extend previous findings on social decision making strategies in humans to nonverbal scenarios akin to those of social insects.

Protogynous sex change, where individuals first function as females and later as males, is a key life-history strategy among polygynous reef fishes. In haremic systems, sex change is typically socially regulated, with dominants suppressing subordinates sex change through aggression. Females within a harem form a size-based hierarchy that can remain stable in most species through the threat of eviction. We studied a different situation in the cleaner wrasse Labroides dimidiatus, where larger females have incomplete control, as they spend most of their time alone at their own cleaning territory. We tracked over 400 individuals for 12 months, recording growth, behavior, social organization, and sex change. We confirmed earlier reports that both sexes direct aggression primarily at those ranked immediately below them. However, we observed 30 cases where smaller females outgrew larger ones, revealing hierarchy instability. Of 42 sex change events, 43% occurred in presence of the male, and half of these early sex changers were not the largest female, but individuals overlooked by the male. Fast growth relative to harem-mates and harem switching increased the likelihood of sex change. Local population densities also influenced growth and sex change, with individuals in high-density demes growing faster and changing sex at larger sizes. Our findings reveal flexible sex change dynamics in a system with incomplete social dominance. Such incomplete control and observations that becoming male confers both higher reproductive success and survival highlight the need to expand game-theoretical and life-history frameworks to encompass such strategic flexibility. Lay summaryDominant cleaner wrasse cannot fully control subordinates as individuals occupy distinct core areas. Tracking 400 fish for a year, we found that smaller females could outgrow initially larger ones, and early sex change despite a larger male. Fast growth and harem switching increased the chances of becoming male. Population density also shaped these strategies. Our findings reveal flexible sex change dynamics in a system where becoming male confers both higher reproductive success and survival.

1.Many group-living mammals and birds groom the fur (or preen the feathers) of their close associates, and this social grooming (or social preening) seems to build and maintain affiliative relationships. Female common vampire bats (Desmodus rotundus) lick each other in ways that appear to be responsive to cues of need, which suggests that this social grooming could be a low-cost form of helping in addition to being a social signal. If social grooming is a form of helping, then vampire bats should preferentially groom others in locations that are difficult to self-groom. We show that social grooming (n = 1586 events) did indeed occur most often on parts of the recipients body where self-grooming (n = 1515 events) was least likely, often in locations where the recipient could not lick itself, like the back of the head. The finding that vampire bats preferentially groom each other in hard-to-reach locations provides further support for the hypothesis that social grooming is a low-cost form of help in vampire bats.

Finding resources for the colony is one of the most difficult and risky tasks for a social insect worker. A worker on a foraging trip can face a number of challenges, including interference from other individuals, her own errors, and environmental disturbances. Collectively, colonies may use a variety of strategies to minimize the impact of such perturbations on the foraging process. Here, we investigated how individual Solenopsis xyloni ant workers react to perturbation of an established pheromone trail. We trained foragers from colonies in the field to either a low or high concentration sucrose solution in a feeder on a T-maze setup, then replaced a section of floor covering, removing a section of the pheromone trail previously laid. We found that while ants made correct choices on the T-maze when the trail was intact, their choices did not differ from chance when the trail was absent, indicating strong reliance on a pheromone trail (and not, for example, memory) to return to the resource. Moreover, when the trail was absent, we found that a majority of ants abandoned the resource, and that even the ants that were able to reach the resource did not repair the perturbed trail. However, with a high-quality resource, more ants persisted in attempting to reach it (instead of abandoning). We interpret these responses in the framework of robustness mechanisms discussed in systems biology. Our study thus links individual and collective responses to perturbations, and provides an empirical example of how information use interacts with system robustness. Statements and declarationsThe authors have no competing interests to declare that are relevant to the content of this article.

O_LISeasonal patterns of species appearances constitute a major component of diversity variation. Theory attributes this phenological structuring of communities to the alignment of life cycles to suitable moments and to constraints of seasonality on development, yet the specific mechanisms operating across taxa remain largely unresolved. In insects, body size and colour are key functional traits that contribute to driving spatial community assembly through their link to thermoregulatory performance and development. C_LIO_LIHere we analyse variation in mean body size and colour lightness of 483 butterfly assemblages across Great Britain and throughout the season to test whether trait alignment with seasonal environment and developmental constraints may shape the phenological structuring of communities. C_LIO_LIBoth body size and body colour varied more along season than across space, emphasizing the importance of phenology on diversity variation. Body size was larger early and late in the season, i.e. under conditions of low temperature and solar radiation. This pattern contrasted with the spatial trends found and was driven by species overwintering as adults, which we interpret as being likely due to energetic constraints. Body colour, conversely, was darker early and late in the season, mirroring the spatial pattern found, and suggesting a thermoregulatory alignment with seasonal conditions. Furthermore, covariation between body size and colour suggests a thermoregulatory interaction between both traits. C_LIO_LIOur findings suggest that life-cycle constraints and seasonal thermoregulatory alignment contribute to shaping the phenological structure of insect communities. C_LI

Collective behavior in animal societies can buffer individual costs and confer resilience to environmental challenges. However, the mechanisms by which groups sustain function when members are compromised remain poorly understood. In the presented study, we investigate how social context shapes collective fanning, a thermoregulatory behavior critical for colony function, in Western honeybees (Apis mellifera). Using oxytetracycline (OTC), a known physiologically disruptive antibiotic to honeybees, to selectively impair certain group members, we tested our hypothesis that the presence of untreated bees would rescue the fanning response in mixed-composition groups. We show that groups containing untreated individuals fan at levels comparable to fully untreated groups, despite the presence of OTC-impaired bees. This preservation of collective thermoregulatory function was correlated with both treated and untreated individuals in mixed groups shifting their interaction dynamics and social network positions. These findings reveal a decentralized mechanism of collective resilience, whereby behavioral compensation by individuals sustains group-level thermoregulation under partial disruption. Our results provide a framework for understanding how social insect colonies maintain function in the face of individual-level perturbations, with broader implications for predicting the limits of collective resilience in animal societies experiencing increasing environmental pressures.