Nature’s “responsible” response to disease

With the emergence of the COVID-19 pandemic, there have been global calls for the implementation of “social distancing” to control transmission. Throughout the world, some have resisted this requirement with the unfounded argument that it is unnecessary or ineffective. Social distancing, however, is a natural consequence of disease across animals, both human and nonhuman. Stockmaier et al. reviewed responses to disease across animal taxa and reveal how these responses naturally limit disease transmission. Understanding such natural responses and their impacts on pathogenic transmission provides epidemiological insight into our own responses to pandemic challenges.

Science, this issue p. eabc8881

Structured Abstract


Contagious pathogens can trigger diverse changes in host social behaviors, rewiring their social networks and profoundly influencing the extent and pace of pathogen spread. Although “social distancing” is now an all too familiar strategy to manage COVID-19, nonhuman animals also exhibit a suite of pathogen-induced changes in social interactions, either as precautionary measures by healthy hosts or as physiological consequences of infection in sick individuals. These diverse changes in the social behaviors of both healthy and infected hosts in response to pathogens are widespread across taxa, but we still have much to learn about their underlying mechanisms and epidemiological and evolutionary consequences. Studies of social distancing behaviors in nonhuman animals have the potential to provide important and unique insights into ecological and evolutionary processes relevant to human public health, including pathogen transmission dynamics and virulence evolution.


We synthesize the literature on pathogen-induced changes in sociality in nonhuman animals and in humans. These include active and passive changes in pathogen-exposed and -unexposed group members occurring both before and after individuals develop an active infection. Behavioral changes that reduce social interactions—and thus pathogen spread—include changes driven by infectious hosts, such as sickness behaviors and active self-isolation, as well as changes driven by healthy hosts, including active avoidance or exclusion of infectious individuals and proactive social distancing in the face of pathogenic threats. Although species have evolved behavioral social distancing because it reduces infection risk, these behaviors also incur costs by limiting access to the many benefits of group living, such as protection against predators and cooperative food finding. Thus, many species appear to have evolved the ability to alter the expression of these behaviors in ways that maximize benefits and minimize costs. The most susceptible individuals of some species show the strongest avoidance of sick conspecifics, and social distancing behaviors are sometimes foregone in interactions with close relatives. Pathogen-induced changes in sociality also apply important selection pressures on pathogens. Because social distancing reduces transmission and thus fitness, pathogens may evolve lower levels of virulence, presymptomatic transmission, or the ability to disguise cues that enable hosts to recognize their presence. Finally, pathogen infection can also increase social interactions when healthy individuals lend aid to pathogen-contaminated or sick conspecifics. Helping sick individuals is a major part of human and eusocial insect societies but is less commonly observed in other, nonhuman animals. Whether pathogens can evolve to elicit helping behavior in hosts, thus augmenting their own transmission, remains unknown.


The structure and dynamics of social contact networks fundamentally determine the fate of disease outbreaks, that is, how fast and far they spread and who will be infected. In the race to combat the COVID-19 pandemic, numerous studies have begun to address the public health utility of unprecedented social distancing efforts. Nonhuman animal systems, particularly those with social structures similar to those of humans, present unique opportunities to inform relevant public health questions such as the effectiveness, variability, and required duration of social distancing measures. Further, the experimental tractability of nonhuman animal systems allows study of the coevolutionary dynamics generated by social distancing behaviors, which themselves have public health implications. Selection for or against social distancing behaviors has the potential to create a conflict of interest and could incentivize selfish behaviors that are not in the best interest of everyone.

Social distancing in humans and nonhuman animals.

(A) Pathogen-exposed forager ants self-isolate and their nestmates increase social distance to each other (image: Timothée Brütsch). (B) People social distance during COVID-19 (image: Forest Simon). (C) Sick vampire bats reduce grooming non-close kin (image: Gerald Carter). (D and E) Under certain conditions, Trinidadian guppies avoid parasitized individuals (D), (image: Sean Earnshaw, University of St. Andrews) and house finches avoid sick conspecifics (E) (image: Jeremy Stanley).


Spread of contagious pathogens critically depends on the number and types of contacts between infectious and susceptible hosts. Changes in social behavior by susceptible, exposed, or sick individuals thus have far-reaching downstream consequences for infectious disease spread. Although “social distancing” is now an all too familiar strategy for managing COVID-19, nonhuman animals also exhibit pathogen-induced changes in social interactions. Here, we synthesize the effects of infectious pathogens on social interactions in animals (including humans), review what is known about underlying mechanisms, and consider implications for evolution and epidemiology.

A crucial issue for understanding the spread of infectious diseases is determining how animal social networks change through time as uninfected, exposed, or sick individuals alter their behavior in response to contagious pathogens (15). Compartmental epidemiological models (e.g., susceptible-infected-recovered models) often make simplifying assumptions that hosts participate equally in contacts and contacts remain constant through time (6). However, contact rates clearly vary among individuals and over time, and network-based epidemiological models demonstrate that contact variation substantially affects disease dynamics (710). One key source of contact rate heterogeneity lies in behavioral responses to pathogens by infected, contaminated, or susceptible individuals. Although some parasites famously manipulate the behavior of their hosts to facilitate transmission to new hosts [reviewed in (11)], behavioral responses to infection in social animals are more often host driven (12). Here, we review the diverse suite of host-mediated behavioral responses to pathogens, which include sickness behaviors (immune-mediated lethargy and social disinterest) by infected hosts and avoidance or exclusion of potentially infectious conspecifics by susceptible individuals, both of which likely suppress population-level pathogen spread. Conversely, infected individuals may receive care from uninfected group members, facilitating transmission. Such social responses to infection are commonly observed in nonhuman animals; therefore, considering the evolutionary and epidemiological implications of social distancing in nature could shed important light on our understanding of human outbreaks.

Pathogens change social cues, signals, and behaviors

Pathogens induce a wide array of changes in the behaviors of individuals that they successfully colonize and their uninfected group members. These changes arise at distinct stages across systems, beginning as early as initial host exposure to the pathogen (i.e., “contamination”) or as late as symptomatic stages of disease (Fig. 1). Some externally transmitted pathogenic fungi in social insects elicit host behavioral changes as early as 15 min after exposure, when individuals are already potentially infectious but not yet infected, which requires fungal spores to pierce the cuticle (13). For example, termites exposed to entomopathogenic fungal spores produce immediate vibratory alarm signals that trigger avoidance or hygienic responses in nestmates (13, 14); the same pathogen induces self-removal and care responses in ants within hours of exposure (3, 1517), suggesting that social insects detect cues associated with the pathogen itself on the surface of the cuticle.

Fig. 1 Behavioral changes in response to pathogens (yellow) and their potential cues (blue) can occur upon initial exposure, during the presymptomatic incubation period (teal), or during the symptomatic period (orange).

The degree to which behavioral changes overlap a pathogen’s infectious period (red) will determine their effectiveness at preventing spread. In ants, chemosensory recognition can occur immediately after exposure, triggering self-isolation and proactive social distancing. In other systems (e.g., Trinidadian guppies and ectoparasitic worms), infectiousness and avoidance behavior are aligned with clinical signs or, for parasites that do not cause obvious clinical signs, with changes in chemical cues (i.e., mandrills and protozoal parasites). Behavioral changes can also occur later in infection (i.e., humans and influenza infections). Semitransparent arrows indicate variability and uncertainty in timing across systems.

In other systems, changes in behavior are triggered by modifications in social cues and signals caused by infection itself or by challenges of the immune system with pathogenic compounds either during the incubation period while the host is not yet infectious or during the symptomatic disease phase (Fig. 1). For example, virus-infected or immune-challenged mice produce specific olfactory cues (18), feces of protozoa-infected mandrills have a distinct smell (19), immune-challenged humans have more aversive body odor (20), and fungus-infected ant pupae produce chemical cues that trigger hygienic behaviors in adult ants, including destruction and disinfection of the cocoon (21). Visual cues can also be altered by infection; for example, Trinidadian guppies can avoid conspecifics using visual cues of ectoparasitic worm infection (22, 23), and humans can identify immune-challenged individuals by examining facial photos (24). Infection or immune stimulation can also affect auditory cues, as is the case for vampire bats challenged with immunogenic lipopolysaccharide (LPS), which reduce contact calling rates (25), and LPS-challenged men, who experience audible breathing changes (26).

In addition, innate immune responses to pathogens typically stimulate physiological (e.g., fever) and behavioral changes in the infected host, including lethargy and reduced social interactions, particularly early in infection (2730). These “sickness behaviors” occur widely across host taxa and in response to diverse pathogens (2731). Because the predominant physiological mediators of sickness behaviors are the proinflammatory cytokines that link the immune, endocrine, and nervous systems (28, 29, 32), the consequences of these behavioral changes for social interactions can be experimentally explored by injecting hosts with immunogenic substances such as LPS or cytokines (2, 28, 33). Given how common they are across taxa, sickness behaviors could also serve as a relatively universal cue for recognizing infected conspecifics. Detecting sickness behavior may be easier when the observer is familiar with the baseline behavior of the sick individual. Therefore, such recognition mechanisms could be more common in species that live in close-knit groups. However, the sensory and neural mechanisms responsible for the recognition of such indirect cues are still poorly known.

How individuals detect, recognize, and respond to disease-related cues, especially chemical cues, has received much attention (e.g., 18, 19), and changes in appearance, smell, vocalizations, or behavior are known to induce “social distancing” (i.e., reductions in potentially transmission-causing contacts) in both animal and human societies. However, natural selection can also lead to seemingly altruistic behaviors such as helping infected conspecifics, which may instead increase disease transmission. Here, we focus on six pathogen-induced physiological or behavioral changes in hosts that cause changes to social interactions in groups and can be driven by conspecifics or the potentially infectious individual (Fig. 2).

Fig. 2 Effects of pathogen exposure on social behaviors can be driven by susceptible or pathogen-affected individuals and increase (help) or decrease contact (see orange squares).

For individual-driven effects, we distinguish direct effects and indirect effects (gray area). Animals are highlighted based on available studies in respective systems in birds, nonhuman mammals (mouse symbol), humans, insects and other invertebrates (mostly eusocial insects; ant symbol), reptiles and amphibians (lizard symbol), and fish.

Passive self-isolation by potentially infectious individuals

Passive self-isolation is a component of sickness behavior (2729) that occurs when a sick individual directly or indirectly reduces contact with others while remaining within the group. It can occur directly when infected animals lose motivation to engage in physical social behaviors such as grooming or food sharing (33, 34), a phenomenon called social “disinterest.” For instance, immune-challenged vampire bats reduce grooming of certain conspecifics (33), virus-infected bees share less food with nestmates (34), and humans challenged with bacterial endotoxin self-report feelings of social disconnectedness that may reduce contacts (35).

Passive self-isolation can also occur indirectly due to physiological responses to infection such as lethargy, which is challenging to tease apart from direct effects without measuring the motivation of the test subject. Passive isolation can happen, for instance, when sickness-induced lethargy reduces individuals’ social investments in the biological marketplace, such as allogrooming or provisioning of food (36), which could reduce reciprocal services from, and contacts with, group members. Additionally immune challenge and the resulting lethargy can reduce social vocalizations, which as an incidental side effect may make group members less inclined to interact with the sick individual (25). Lethargy can also alter patterns of movement and dispersal, which determine contact with other individuals (2, 5). Thus, reduced movement could restrict the spread of directly transmitted pathogens between clusters of individuals. However, such passive isolation likely does not evolve as an adaptation specifically for this purpose.

Active self-isolation by potentially infectious individuals

Potentially infectious human and nonhuman animals sometimes actively remove themselves from others, thereby preventing susceptible individuals from interacting with them. This differs from passive self-isolation, in which susceptible group members can maintain interactions with lethargic infected individuals. For instance, although immune-challenged bats perform less grooming, they remain part of the group and still receive food donations from conspecifics (33). By contrast, fungus-exposed ants spend more time outside the nest, and thus actively self-isolate, limiting encounters with susceptible nestmates [(3, 37, 38); Fig. 3]. Self-isolation is a seemingly altruistic act hypothesized to evolve through kin selection, as evidenced by its widespread occurrence in eusocial insects, in which high within-colony relatedness favored the evolution of numerous collective disease defenses called “social immunity” (3, 3739). Active self-isolation appears to be a general response to apparent detection of impending death, not only from pathogens but also from CO2 poisoning and toxins (37, 40). However, the cues and mechanisms underlying the initiation of self-isolation remain unknown.

Fig. 3 Network changes and epidemiological consequences of pathogen exposure in ants (case study).

(A) Common garden ant (Lasius niger) queen and workers marked with fiducial markers used for automatic detection of social interactions (image: Timothée Brütsch). (B) Social interaction networks before and after exposure of some workers (gray dotted squares) to infectious fungal spores. Circles represent nonexposed individuals, and circle colors represent the predicted intensity of exposure to the pathogen based on epidemiological simulations run on each network [data from (3)].

Despite anecdotal observations, such as a tuberculosis-infected badger leaving its group to die alone (41), systematic investigations of active self-isolation in animals outside of eusocial insects are lacking. By contrast, infected humans are known to actively self-isolate, as evidenced by historical outbreaks (42). However, whether such self-isolation is driven by personal initiative or governmental policy directives is often difficult to disentangle.

Avoidance of potentially infectious conspecifics

In animals affected by contagious pathogens, selection should favor susceptible individuals who can detect and subsequently avoid potentially infectious conspecifics (43). Indeed, avoidance of exposed or infected conspecifics occurs in diverse nonhuman animals including lobsters (44), Trinidadian guppies (22, 23), mandrills (19, 45), and termites (14). Humans can avoid conspecifics based on facial cues or chemosensory cues (20, 24), and awareness of infectious threats can exaggerate avoidance responses even without proximity to infected people [i.e., gathering information online (46)]. Avoiding infectious conspecifics requires (i) cues that differentiate potentially infectious individuals from healthy group members, (ii) sensory systems to detect those cues, and (iii) neurological pathways that translate cues into behavioral changes. Pathogen exposure and infection alter a suite of sensory cues that need not be pathogen specific [generally “disgust eliciting” (4749)] and could therefore result in rapid manifestation of avoidance behaviors in animal populations (50).

Exclusion of potentially infectious individuals (enforced isolation)

Exclusion of infectious conspecifics, by aggression or other means, represents another direct way in which pathogens lead to social distancing. We distinguish exclusion from self-isolation in that exclusion is enforced by uninfected individuals. Aggressive exclusion of infected individuals is mostly documented in eusocial insects (39, 51), for example, virus-infected honeybees being forcibly dragged out of the nest (51). Much like avoidance, active exclusion requires an ability to recognize infected individuals. In Lasius neglectus ants, destructive removal of infected broods is driven by changes in cuticular chemical composition that allow uninfected workers to detect otherwise asymptomatic fungal-infected broods (21). Enforced exclusion has not been experimentally demonstrated in mammals, although observational evidence exists (52), and enforced quarantine has occurred throughout human history and remains an important public health measure against pathogens such as Ebola and severe acute respiratory syndrome (SARS) (53, 54).

Increases in social contact through caregiving

Helping is one of the main aspects of human health care and can involve family, friends, or health care workers (55). However, such caregiving incurs increased infection risks for caregivers (56). The extent of caregiving in nonhuman mammals is still unclear, so far relying largely on opportunistic field observations (55). The clearest evidence for caregiving behaviors outside of humans comes from antifungal grooming in eusocial insects: Ants and termites routinely physically remove or chemically deactivate infectious fungal spores on contaminated nestmates, thereby decreasing the risk of infection for their nestmates but also increasing their own risk of low-level infection [(15, 16, 57); Fig. 3]. Recognizing infected or exposed conspecifics is a precondition of caregiving. Such recognition could occur through detecting infection cues (see above), the pathogen itself in the case of external contamination, or active solicitation of help such as the vibratory alarm behavior that termites use to elicit care (13).

Proactive social distancing among susceptible or asymptomatic individuals to slow spread

In the ongoing COVID-19 pandemic, most countries have implemented generalized social distancing, requiring asymptomatic and uninfected individuals to minimize all contacts. This drastic measure has proven effective in reducing transmission rates in affected communities (58, 59). A similar strategy is used by colonies of black garden ants: Upon entry of fungus-contaminated nestmates, nurses and foragers increase their social distance from one another, reducing intergroup contact rates (3). This early colony-wide reaction likely reduces the risk of an epidemic by limiting inadvertent transmission from asymptomatic carriers (Fig. 3).

Epidemiological consequences for directly transmitted pathogens

The structure and dynamics of social contact networks fundamentally determine the fate of contagious pathogen outbreaks, including how fast and far they spread and who becomes infected (710). Contact rates vary among individuals based on social structure, sex, and age, among other things, and shape individual and community level risks of transmission (8). Studies of human viruses such as influenza shed light on how individual-level behaviors, such as social withdrawal during infection, could inform public health responses (5). In the race to combat COVID-19, numerous studies have examined the public health utility of unprecedented large-scale social distancing [e.g., (58, 59)]. By studying pathogen-induced social network changes in nonhuman animals, we may learn about the efficacy of naturally evolved social distancing rules that could inform the management of contagious pathogens in humans.

Passive and active self-isolation, avoidance, exclusion, and group-wide social distancing can profoundly affect the spread of contagious pathogens by reducing the degree of contact between susceptible and sick individuals and thus altering network-level contact heterogeneity [(13, 5, 60, 61); Fig. 4]. For instance, network centrality of wild vampire bats is reduced when their immune system is challenged but this effect diminishes over time (61). Similarly, immune-challenged mice reduce connectivity to their group because of lethargy (2), and ant social networks undergo deep restructuring to prevent colony-wide spread of an infectious fungal threat [(3); Fig. 3]. Unfortunately, most research on sickness behaviors has been done in the laboratory setting, such as with mouse models, which are often in dyads to identify physiological mechanisms, and this mechanistic focus prevents inquiry into epidemiological effects in larger populations and networks. Such larger-scale research is increasingly possible because of technological advances such as next-generation proximity loggers and automated tracking of radiofrequency identification (RFID) tags or quick response (QR) code labels, which provide high-resolution data on network structure and track how individuals and group properties change over time (62). A key future research goal is to understand how the social effects of infections alter both the topology and overall transmission properties of contact networks.

Fig. 4 Social distancing mechanisms affect the number (connecting lines) and strengths (width) of network connections for susceptible (green circles) and exposed and/or infected (orange circles) individuals.

Mechanisms are ordered based on transmission risk from caregiving (high risk) to complete removal and/or exclusion (low risk). Representative taxa are shown for each mechanism.

Epidemiological studies of passive self-isolation in humans are mainly modeling based (63) or come from surveys of social contacts in the presence or absence of infection. For example, influenza-induced sickness behaviors reduce the number of social contacts, and hence the virus’s reproduction number, to about one-fourth relative to expectations without sickness behavior (5). However, the effectiveness of passive self-isolation in suppressing transmission will depend on the extent to which behavioral changes align with the infectious period of a given pathogen (Fig. 1). Further, sickness behaviors and their effect on social interactions are themselves confounded by other factors such as social stress, sex, and kin relationships [(12); Fig. 5]. In humans, sociocultural factors can affect the expression of sickness behavior. For instance, there are often economic or social motivations for persistent work attendance when sick, a phenomenon known as “presenteeism” (64). There is also preliminary evidence that personality traits or cultural norms such as stoicism and familism affect sickness behavior differently based on demographic characteristics (65). Any epidemiological benefits of isolation and sickness behavior cannot accrue in sociocultural systems that stigmatize rest, recuperation, and isolation or do not provide individuals with the means to safely engage in these behaviors. The additional level of complexity contributed by environmental, biological, and cultural variation in the expression of isolation and sickness behaviors should be incorporated in future models of pathogen-induced behavioral changes and transmission (Fig. 5).

Fig. 5 Individuals within a species can vary social distancing behaviors based on immune status, kinship, and social or work pressure.

Icons show species for which these patterns have been shown to date.

Active self-isolation prevents conspecifics from interacting with infected individuals, whereas passive self-isolation may not have the same effect. Therefore, active self-isolation, particularly when such behavior occurs early in the infectious period, should decrease transmission-causing contacts more effectively (Fig. 4), as shown recently for ant foragers that self-isolate when exposed to a fungus, greatly reducing their contacts with other colony members [(3); Fig. 3]. The epidemiological effects of enforced exclusion should be similar because it prevents all subsequent interactions with conspecifics. For instance, in L. neglectus ants, removal, destruction, and disinfection of infected broods was shown to reduce transmission by 95% (21). Measures that isolate infectious individuals are more effective when asymptomatic transmission is rare and lose efficiency as asymptomatic transmission increases (66). In the latter case, active isolation must be supplemented with other proactive measures such as quarantine of contacts due to contact tracing or generalized social distancing (66, 67).

Avoidance behaviors by uninfected individuals reduce pathogen spread but require cues that may not align well with a pathogen’s infectious period (Fig. 1). Further, there can be substantial interindividual variance in avoidance based on traits such as an individual’s immune susceptibility or kinship to the sick conspecific [(12); Fig. 5]. Theory suggests that risk-based evaluation of infectious conspecifics could have important epidemiological consequences and determine whether pathogens persist or disappear (60). Environmental cues such as unsanitary conditions might also modulate avoidance behaviors (68) and could be incorporated into epidemiological models.

Caregiving inherently increases contact between helpers and infectious individuals but may accelerate recovery of sick individuals, reducing infectious period length. Their combined impact on pathogen transmission will depend on the nature of the caregiving behavior (e.g., directly removing pathogens such as antifungal grooming in ants versus mitigating harm to sick individuals) and how carefully caregivers mitigate their own risk. Health care workers are among the most affected in recent pandemics, comprising 18.6% of Middle East respiratory syndrome (MERS) cases, 21% of SARS cases (56), and >95% of positive PCR tests in a study comparing healthcare and non-healthcare workers at the onset of the COVID-19 pandemic. (69). Because of their risky occupations, health care workers can become “superspreaders” (9), connecting patients, their families, and friends, and contributing disproportionately to overall spread than the average person. Some health care workers continue to work while symptomatic, despite acknowledging that this places patients at risk, due to structural concerns about staffing and cultural norms that support presenteeism (70). Thus, targeted infection control procedures for health care workers are imperative, as are policies that discourage presenteeism such as sufficient paid sick leave. Because evidence of helping behavior is rare outside of humans and eusocial insects, we know little about its epidemiological effects in nonhuman animals. Social insects, however, highlight the complex balance between costs and benefits of care behavior; for example, ants with high disease susceptibility preferentially use safer care behaviors such as antimicrobial spraying over riskier behaviors such as grooming (17). In humans, lower susceptibility (i.e., no preexisting conditions) or targeted vaccination might affect decision making about when and how intensively to care for the sick and that care’s impact on pathogen spread.

Evolutionary consequences for pathogens

Despite social distancing strategies, contagious pathogens persist in human and animal populations. Thus, social distancing behaviors and the capabilities of pathogens to counteract these behavioral defenses could result in evolutionary arms races and thus shape the evolutionary trajectories of both hosts and pathogens (12, 71).

Pathogens are predicted to evolve a virulence level (i.e., damage to host) that optimizes their overall growth rate within the population. Theory suggests that this growth rate for contagious pathogens reflects a balance between within-host replication to generate infectious propagules (e.g., pathogen shedding), which typically underlies virulence, against the need to get those propagules to new hosts before the infectious individual dies or recovers (72, 73). There is very little research on how pathogens evolve to optimize the trade-off between sufficient shedding by hosts while reducing host symptoms that induce social distancing by infectious or susceptible hosts. Pathogens may evolve counteradaptations that minimize host isolation, whether by mitigating symptoms (e.g., lethargy) that affect behavior of infectious hosts or altering sickness cues available for detection. Inhibition of proinflammatory cytokines that drive passive self-isolation, including lethargy, might be a target for such counteradaptations by pathogens. Inhibition of cytokine responses is well described in bacterial and viral pathogens (74). Similarly, up-regulation of anti-inflammatory cytokines could theoretically affect sickness behaviors and social contact rates (75).

All forms of social distancing, whether driven by infectious or susceptible hosts, should generally select for less virulent pathogens with milder symptoms or asymptomatic infectious periods (72, 73), especially in pathogens for which transmission is weakly reliant on virulence (75). Further study is needed to determine whether avoidance behaviors favor presymptomatic infectious periods as pathogen counterstrategies. Conversely, helping behavior may increase transmission opportunities, potentially favoring increased pathogen virulence because high virulence no longer limits, and may even facilitate, transmission opportunities. An intriguing prediction is that pathogens might evolve to elicit helping behaviors (e.g., inducing signals of distress) to attract susceptible caregivers. Such pathogen manipulation occurs for other contact behaviors such as augmented aggression [reviewed in (11)] or increased acceptance of noncolony members in honeybees (34).

Evolutionary significance for the host

Social distancing can have substantial fitness costs for infected individuals, which may experience loss of social status, increased exposure to predators, decreased foraging efficiency, and reduced social support (27, 29, 30, 76). Susceptible individuals can also incur costs such as reduced mating opportunities when they avoid or exclude infectious conspecifics (48), particularly if there are false-positive signals. Proactive general social distancing may compromise other collective functions such as food sharing or information flow (3). This raises questions about the evolutionary origin and persistence of pathogen-induced social distancing in humans and nonhuman animals.

Social distancing by susceptible individuals (i.e., avoidance, exclusion of infectious individuals, and proactive distancing) should be favored whenever the benefits of avoiding infection outweigh the costs of distancing, which include indirect effects of disrupting the social group. These mechanisms should therefore mostly evolve in loose social groups, in which the costs of forgoing social interactions are small (77, 78), or in the face of virulent pathogens, in which the costs of contracting infection are high (79). Consistent with these predictions, highly social animals appear less likely to avoid sick peers, and low-virulence diseases such as sarcoptic mange in gray wolves do not elicit exclusion (50, 76). Further, because the costs of contracting infection can even vary among individuals within a species, avoidance behaviors should be variable such that highly susceptible individuals show stronger avoidance responses, as occurs in Trinidadian guppies [(22); Fig. 5]. In addition to variable costs, because the benefits of social interactions vary according to individuals’ social roles and position, avoidance behaviors should depend on both individual risk (50) and social context. For instance, in humans, population-level differences in disgust perception and sensitivity (80) may be linked with differences in pathogen threat (81). Social status and financial resources clearly affect an individual’s ability to absorb costs of social distancing, and in humans, some costs of distancing may be lowered through virtual interactions (50). Further work is needed to clarify the connections among individual social status, role, and (in humans) attitudes, practices, and behavioral changes.

The evolution of social distancing enacted by potentially infectious individuals themselves is a more complex question because leaving the group incurs substantially higher costs for isolated individuals (who forgo the benefits from all kin or group members) than for remaining group members (who only experience a small decrease in group size). This asymmetry in costs may lead to conflicts of interest between infectious and susceptible group members, in which concealing an infection may be beneficial to sick individuals if it allows them to maintain the benefits of sociality (i.e., presenteeism in humans). This is supported by studies showing that social context alters the expression of sickness behaviors (82); for example, immune-challenged zebra finches express stronger behavioral sickness symptoms when housed alone than when housed in a group environment (83). Other cues (e.g., olfactory or visual signs of infections) may be less plastic, harder to conceal, and potentially constitute more honest information for conspecifics.

Whether sickness behaviors are expressed as an inevitable side effect of infection or as an active, adaptive host response has been highly debated (84) given the difficulty in disentangling the behavioral and inflammatory components. Sickness behaviors are generally hypothesized to improve recovery by redirecting energy to costly immune responses (27). Direct tests of the adaptive benefits of sickness behavior are rare [but see (85, 86)]. However, multiple studies have found that sickness behavior and physiological responses to infection, such as fever, are not always correlated and can arise independently of one other (30, 87). This led to the hypothesis that passive social distancing mediated by sickness behavior, as well as active self-isolation, may confer additional indirect benefits to infectious individuals beyond the beneficial effects on recovery (30).

The most obvious social benefit of self-isolation is kin protection, because social distancing reduces the risk of transmitting pathogens to related group members, thereby increasing the indirect fitness of infected individuals (88). Kin selection should therefore favor the evolution of self-isolation within highly related groups, as likely occurred in many eusocial insects (3, 12, 37, 38), which are characterized by unparalleled levels of relatedness among group members (89). In social insects, active self-isolation cannot be a mere side effect of infection because it often occurs after exposure but before the onset of infection (3, 38) or even in the absence of an infectious organism as a response to other causes of mortality such as poisoning (37, 40); instead, it appears to be a seemingly altruistic act that contributes to the colony’s cooperative disease defenses (39). Self-isolation in humans could have the opposite effect of self-isolation in other animals because it might decrease contact with unrelated individuals outside the home but increase contact with family members, thus putting kin at higher risk than non-kin.

Kin selection theory also predicts that caregiving should evolve among relatives because increased kin survival may outweigh the risks associated with caring. This is supported by multiple studies of nonhuman animals: Mandrills do not avoid grooming parasitized offspring and half-siblings (45), and antifungal grooming is omnipresent in eusocial insects, greatly increasing the survival of exposed workers (15, 16, 39, 57). Similarly, humans are more likely to receive aid from relatives than strangers across a range of conditions [reviewed in (90)]. However, helping behavior can also evolve in groups with low relatedness provided the benefits of aiding a diseased group member outweigh potential costs to helpers. This may occur in close-knit groups strongly reliant on cooperation for survival or host-pathogen systems with low transmission risk or intermediate pathogen virulence (12, 76, 77).

An alternative explanation for the evolution of caregiving is that it confers direct benefits to helpers. For example, in social insects, caring individuals gain protection against secondary infection with the same pathogen through a temporary boost of their immunity (15, 57); similarly, in humans, being close enough to recognize an individual’s ailment might prime the caregiver’s immune system (91). Other benefits could accrue through reciprocity (i.e., delayed benefits) or reputation enhancement and subsequent reputation-dependent benefits from third parties (90).

Conclusions and future directions

Social distancing behaviors have been studied extensively in both humans and nonhuman animals. Whereas these behaviors (especially sickness behaviors) are often studied using immunostimulants, far less research has been done with pathogens that have naturally coevolved with their hosts. This is an important next step because the considerations outlined above suggest that the strength and nature of distancing behaviors may be a key element of host-parasite coevolution (12, 71), which may favor changes in virulence, presymptomatic or asymptomatic periods, and pathogen-induced cues. We should use naturally coevolved systems to examine how effectively sick individuals are isolated; the physiological, sensory, and neurological basis of any isolation; and the epidemiological effects of isolation. How do individuals sense their own (or others’) illness or pathogen exposure? When during the infection do cues arise and are some of them present before obvious signs are noticeable? To what extent is the timing of cues driven by host versus pathogen-mediated mechanisms? How does perception of cues influence decisions to change social dynamics and group structure? Understanding these mechanisms and their consequences is crucial for then predicting how broadly hosts can use them in the face of diverse pathogens and how and when pathogens may coevolve to combat these mechanisms. Although we largely discuss pathogen exposure and infection interchangeably here, the fact that some species can respond to mere pathogen presence whereas other behaviors are expressed only when individuals become visibly sick raises important questions about the extent to which the cues used for social distancing correlate with infectiousness (i.e., pathogen shedding). Specifically, what are the epidemiological effects of early versus late pathogen-induced social distancing, and at what stage do the benefits to the host of preventing ongoing transmission outweigh the costs of distancing?

Studies in eusocial insect societies have been especially productive because these animals practice seemingly altruistic behaviors such as active self-isolation and caregiving, which decrease the risk of outbreaks through the colony (3, 3739). These social networks share many characteristics with those of human societies and have evolved properties to prevent pathogen transmission (3, 4, 39). As a result, their social distancing strategies may prove key to investigating the epidemiological effects of such behaviors and thus their potential public health utility. There are, however, important differences in interpreting how social network structures evolve in response to pathogenic threats. In eusocial insects, the behavioral repertoire known as “social immunity” most likely represents group-level adaptive behaviors that evolve in response to high relatedness in the group and result in collective properties (39). By comparison, pathogen-induced changes in social networks of other animals including humans (2, 5, 61) often do not have the same properties, such as high relatedness levels, and can create conflicts of interest that incentivize selfish behaviors.

Public health measures experienced during past and current pandemics have raised awareness for social distancing, and epidemiological studies are actively evaluating their effectiveness and required duration. Humans are by no means alone in using social distancing to mitigate risk of infection (92). The widespread occurrence of pathogen-induced changes to social behaviors across animals in diverse taxa represents a valuable opportunity to investigate underlying mechanisms, epidemiological consequences such as effectiveness and required duration, and host-parasite coevolution. Nonhuman animals’ social distancing strategies may be experimentally tractable, enabling manipulative experiments or multigeneration observations that are impossible with humans. These systems represent a valuable guide to understanding how contagious pathogens spread through social networks, how networks change in response to pathogens, and how these bidirectional feedbacks alter pathogen dynamics and evolution.

Acknowledgments: We thank two anonymous reviewers for helpful comments and Carley Beyer for graphical assistance with figures. Figures were created with Funding: N.S. was supported by a European Research Council (ERC Starting Grant DISEASE no. 802628). Competing interests: The authors declare no competing interests.