Eusociality has been a key driver in why ants have become some of the most successful insects in the world, inhabiting most terrestrial environments on the planet (1). The high degree of cooperation and the division of labor of ants benefit the colonies in the form of more effective brood care, foraging and protection of resources, which has allowed ant colonies to grow large, often comprising of hundreds of thousands or even millions of individuals (2). Sociality also comes with costs. Ants live in dense colonies with a low genetic diversity among frequently interacting nestmates, which creates favorable conditions for pathogens to spread (3, 4). Ants also forage and interact with a wide variety of organisms in their environment, which makes them vulnerable to exposure to pathogens (5). The combination of high risk of exposure as well as favorable conditions for transmission of pathogens makes ants particularly susceptible for outbreaks of infectious diseases (3).

To cope with the high pathogen pressure they face, a wide variety of both physiological and behavioral strategies have evolved in ants. Ants share an innate immunity with other insects as a first line of defense, but also have unique adaptations in the form of anti-microbial gland exudates (6–8). Ants initially try to avoid pathogens, but if exposed they can mitigate the effects of them through hygienic behaviors such as grooming of both self and nestmates (9) and contain pathogen spread by corpse processing and cemetery formation and management (10). The extreme level of sociality also means that the anti-pathogenic strategies of individual ants in a colony lead to a whole which is larger than the sum of its parts, providing protection against pathogens on a colony level as a form of social immunity (11, 12).

As a complementary strategy, ants could potentially use self-medication to stave off the threat of disease. Self-medication is the use of biologically active compounds by animals to fight off disease in a way which would be harmful or costly for uninfected individuals (13, 14). The self-medication response is plastic and can take place either before infection takes place (prophylactic) or after (therapeutic) (15), and can be directed either at self or kin (13, 16). To be qualified as true self-medication, the behavior needs to fulfill four criteria: the compound must be deliberately contacted (I), the compound must be harmful for the pathogen (II), the behavior must lead to a higher fitness for infected individuals (III), the use of the compound must be costly for uninfected individuals (IV) (13). However, some studies using vertebrates have adapted slightly different criteria, possibly due to difficulties in carrying out field studies compared to controlled laboratory studies using insects (14, 16, 17).

Evidence of self-medication behavior in ants and other social insects is still scarce (18–20) and to date the clearest evidence for self-medication has been provided in solitary insects (15, 21–24). So far, there is only one demonstrated case of therapeutic self-medication in ants (20), where workers infected with a highly lethal dose of a generalist entomopathogenic fungus consumed food laced with hydrogen peroxide (H₂O₂), a reactive oxygen species (ROS) which is part of their natural immune responses (25–27), to increase their chances of survival. There is also a case to be made that the collection and processing of resin into the nest material is a form of prophylactic self-medication in ants (8). However, the extent of self-medication responses and how they play out in the colony foraging dynamics is still unknown in ants.

Due to the eusociality of ants, individual self-medication responses are likely to engage colony level processes which regulate the collection and distribution of medicinal compounds in the colony. The division of labor in ant colony means that the foragers, who commonly make up around 10–20% of the workers in the colony (1, 28–30), provide nutrition for the entire nest, and tune their foraging choices according to the nutritional needs of their nestmates, influenced by feedback-loops of food acceptance rates by other nestmates (31). The food is then distributed according to which groups or individuals have a higher need for specific nutrients (32). Therefore, if nestmates are battling disease, the nutritional needs of the colony change, and foragers should respond by foraging on medicinal compounds regardless of their individual infection status to distribute medicine to infected nestmates as a form of kin medication (13).

In this study, we tested how different levels of infection by the generalist entomopathogenic fungus Beauveria bassiana affect the self-medication behavior of Formica fusca ants in the form of foraging choices of colonies, and whether they lead to enhanced survival against the pathogen. The use of different levels of pathogen exposure combined with the use of either fixed food diets or a diet choice allowed us to explore questions relating to the plasticity of the self-medication response. We predict that as the prevalence of disease in a colony increases, so does the strength of the self-medication response. By using colonies in which only part of the workers is infected, we could also investigate a potential kin medication component to the self-medication response. We also predict that both infected and infection free workers will forage on medicinal food, with the task of distributing it to the rest of the colony. We used a dose of pathogen which had a low effect on mortality, which is considered to be more naturally relevant (33). Studies using highly lethal doses have also been shown to negatively affect normal immunologic behaviors such as grooming (34).

We collected 47 wild nests of Formica fusca from around the Hanko peninsula in southern Finland (59°54'46.9“N, 23°15'53.8”E). Formica fusca is a weakly polygynous species in this area, with between 40 and 45% of the nests being monogyne, and queen numbers of the nests varying between 1 and 26 (35, 36). Some of the colonies we collected contained several queens, but as precise numbers of queens were not relevant to this study, we did not aim to excavate whole colonies, but taking subsamples and allowing the rest of the nest to survive.

From the nests we made a total of 133 experimental colonies of one queen and 100 workers each and placed them in separate plastic containers (8 × 15 × 10 cm). Whenever several experimental colonies were made of a single nest, they were used in the different setups. The walls and lids of each container was lined with a 20% (w/v) mixture of ethanol and talcum powder to keep the ants from escaping (37). A small portion of the lid was cut off to get a clear view of the foraging area of the containers. Each container had a 2 cm deep plaster cast base with a 1 cm deep circular indentation (3 cm Ø) as a nesting site with a ceramic tile (4 × 4 cm) on top for shelter.

Formica fusca ants have previously been shown to self-medicate by incorporating food containing ROS in the form of H₂O₂ into their diet in response to a highly lethal dose of B. bassiana (20). Here we show that a similar response is also induced with a much less lethal dose, and that the strength of self-medication response is modulated by the ratio of infected ants in the colony, as the colonies were choosing to forage increasingly on ROS food as disease prevalence increased in the colonies, as we predicted.

Adjusting foraging on ROS food according to pathogen prevalence implies plastic and deliberate use of the compound, which is in accordance with the first criterion of self-medication. Our experiment also clearly shows that another criterion is met: ROS food is harmful for healthy individuals. Simone-Finstrom and Spivak (18) have previously confirmed a further criterion, that H₂O₂ lowers the fitness of B. bassiana. The remaining criterion we cannot clearly prove in our experiment—i.e., the benefit of the dietary change as a response to the infection. We do see however, that when having access to a balanced diet of both medicinal and normal food, the level of infection does not affect the survival of ants in a significant way compared to healthy colonies, so there is some evidence for a benefit to changing the diet.

Two factors may have masked any benefits in survival caused by changes in foraging behavior: the low effect of the pathogen dose on ant mortality and the relatively short period of time observing the colonies. Observing mortality after seven days gives us only a view on the effects of the primary infection, but the threat of disease could still linger in the colonies due to the fungus killed corpses as dead ants were not removed from the colonies (40). Fungal growth on corpses is often visible 2 days after death (40), and ants can differentiate between fungus and freeze killed corpses (41), so it is likely that the self-medication response in the colony would remain upregulated even if the effects of the primary infection would decrease. Therefore, it is reasonable to expect, that the colony would continue to forage on medicinal food which due to its deleterious effects on survival could render short-term benefits low or unclear, but in a longer timeframe benefit the colonies in either worker survival or other measures of fitness. The issue of determining costs and benefits on more than just the short-term survival of individual ants has been raised before (46) and remains an important part in future studies on self-medication behavior.

An interesting result of the experiment was that also infection free colonies with a food choice diet were observed foraging on ROS food, even if this was rarer than in infected colonies. The reason for this behavior is speculative, but it could be possible that the ants are displaying a fixed prophylaxis behavior. de Roode and Lefèvre (47) argued, that if insects would face constant pathogen pressure, evolution would favor prophylaxis to become a fixed behavior. Although this experiment was not enough to provide strong enough evidence for it, Bos et al. (20) reported a similar observation of healthy colonies incorporating ROS food into their diet even if they were free of active infection. As ants are under constant threat of disease due to their behavior and the prevalence of pathogens in their nest and surroundings (3, 5, 48, 49), a fixed prophylaxis would be a feasible strategy for ants to stave off disease, and an interesting study into the extent of the self-medication behavior in ants.

Our results also suggest that obtaining ROS even when only part of a colony is exposed to pathogen infection might not only be an individual response, but instead a response on the colony level.

Foraging on H₂O₂ supplemented food in the 50% infected colonies was done by both uninfected and infected ants without showing any significant difference due to the health status of the foragers. As the foraging frequency was quite low overall, it is likely that all individual ants in the colonies were not individually foraging on the foods, but instead the foragers of the colony were distributing the food according to the needs of the colony. Buffin et al. (50) have previously shown how food is effectively distributed in colonies of F. fusca in a matter of a few hours. If the foragers are distributing medicinal food to infected nestmates as a form of therapeutics as well as uninfected nestmates as prophylaxis, then there could also be a kin medication aspect to the self-medication behavior of ant colonies, however this would have to be confirmed in a separate experiment.

The response of uninfected individuals in a colony to also forage on ROS food implies that there is simultaneous prophylactic as well as therapeutic self-medication within colonies. Thus, the self-medication response is active on both the individual as well as the colony level and opens up the possibility of active kin medication in colonies suffering from pathogen infection. This sort of kin medication also draws parallels with mass drug administration strategies (MDA) used by humans against diseases such as malaria (51, 52). In MDA, the entire population is treated against a prevalent disease regardless of whether they are actively infected or not, meaning that simultaneous therapeutic and prophylactic medication takes place, and it has been shown to be an efficient strategy to contain and eliminate the disease (51). Therefore, kin medication driven by foragers could also be a strategy also used by ants to combat pathogens as a part of their social immunity, which opens up an interesting avenue of self-medication research in ants as well as other social insects.

In conclusion, this experiment adds to the evidence of the ability of ants to self-medicate against disease. In our experiment, we found that the strength of the self-medication response is plastic and increases when a larger part of the colony is infected with a fungal pathogen. This experiment also takes into consideration the criticism of past experiments, where highly lethal doses of pathogens have been used, to provide a more natural level of exposure of pathogens. The fact that ants alter their foraging behavior in response to not only to prevalence of pathogens in their colonies, but also the quality of the medicinal food available and possibly the severity of the disease (20), means that ants have a highly complex understanding of their surroundings and changes in it caused by pathogens leads to intricate decisions on a colony level to respond to them.

The data is now available on Dryad: DOI https://doi.org/10.5061/dryad.0zpc86707.

JR developed the idea for the experiment, analyzed the data, and provided the first draft of the manuscript. JR, HH, and DF designed the experiment and carried it out. All authors worked on the revisions of the manuscript and approved the final version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.