Battles between ants (Hymenoptera: Formicidae): a review

Introduction

Ants (Hymenoptera: Formicidae, Latreille 1809) are ubiquitous. There are native ant species living on every continent on Earth except Antarctica (Moffett 2010; Peter Convey, British Antarctic Survey, personal communication). They are found in urban areas, remote areas, and everywhere in between. Ants occur in several ecological niches and indeed, they act as keystone species in various ecosystems. In their interactions with others, ants may act as prey, herbivores, predators, scavengers, seed dispersers, and farmers and may engage in mutualistic relationships (Schultheiss et al. 2022). Ants first appeared on Earth over 140 million years ago, and the current number of unique species has been estimated at 11,800 (Moreau et al. 2006) and 15,700 (Schultheiss et al. 2022). A recent study conservatively estimates an abundance of 20 quadrillion ants (Schultheiss et al. 2022).

Given their large numbers and extensive terrain coverage, ants must compete with other ants for resources. We will discuss examples of conflict among ants, which takes a myriad of forms and can range from ritualized encounters, in which ants posture to demonstrate their size and/or numbers, to lethal battles. Physical fighting typically involves the use of mandibles for biting, dragging, or holding and the release of venom, which most often takes the form of stinging. Ants may engage in combat with conspecific ants from other colonies or with members of other species. In rare cases, even ants from the same colony may fight. Differences in levels of aggression have been noted among species but, even within species, context-dependent responses lead to a diversity of aggressive behaviors and strategies.

Due to the often large number of combatants engaged and the scope of the battles, ant warfare has been compared to human warfare. In both, individuals may gain advantage by using various tactics, and certain strategies may benefit the group collectively. For example, the ability to recruit larger forces can lead to victory, and using more expendable individuals can increase overall colony efficiency. A notable difference between human and ant warfare is that there is no leader or central decision maker for ants; collective strategy arises out of individual behaviors and the interactions between individuals. These battles often capture the public’s imagination, appearing in games, television, and other media.

The purpose of this manuscript is to comprehensively collect, for the first time in a single review, studies in the steadily growing field of ant combat. We examine under what circumstances ants are likely to engage, with whom, and using what strategies. We also analyze models of ant combat and describe several case studies with various species of interest. For modeling and mathematical analysis, we provide a different perspective on Lanchester’s laws than a recent review (Clifton 2020) and provide a more focused assessment of ants than another review on group conflicts (Green et al. 2020).

Factors That Affect Levels of Aggression

Early anecdotal reports established that ants of different species do not always fight when they encounter one another (Dobrzanski and Dobrzanska 1975). The authors concluded that the probability of fighting was dependent on several variables including levels of aggression in the individual, the behavioral state of the colony, and the relative numbers of potential combatants. There is also marked variation in ants based on their species, that of their opponent, and colony size. Cuticular hydrocarbons are thought to act as a major component of colony member recognition [and supercolony; see Brandt et al. (2009)] and likely interspecific species recognition, and should thus be expected to play an essential role in any biological pathways involved in determining levels of aggression.

We discuss below what may affect the propensity to engage in aggressive behavior including factors such as the species encountered, whether the individual encountered is a neighbor or not, proximity to the nest, individual variation including size of the combatants, and resource distribution. These studies involved the use of comparison groups to provide information on the factors most likely to elicit aggression. However, we note that it is difficult to compare across multiple studies because there is no standardized protocol for experimental design. Ants may be presented to other ants dead or alive, in arenas of varying kinds or in situ, free or tethered. Aggression may be measured by degree of aggressive response (e.g., with ratings of 0, 1, 2) or by tallying individual behaviors (e.g., mandibles open, biting, holding, etc.). Thus, specific comparative studies would likely be needed to fully understand differences between the ranges of responses between groups of ants.

Combat Among Colony Members

Typically, when ants engage in battle, it is with ants from a different colony or even a different species. It is unusual for combat to occur between ants from the same colony. An exception is found in the genus Cardiocondyla (Hymenoptera: Formicidae: Myrmicinae). Unlike most ants, Cardiocondyla species produce wingless males, which do not engage in nuptial flights with queens (Stuart et al. 1987). These species have multiple queens in the colony, providing an opportunity for wingless males to mate without having to leave the nest. Wingless males have noticeably stronger mandibles than winged dispersing males and are thus equipped for aggressive competition among the males in the nest.

Cardiocondyla wroughtonii produce both winged males who do not fight and wingless males who typically fight to the death such that there is only one per colony. These males often engage in one-on-one combat until one male grips the other for a period that may last for hours. Unlike these protracted contests among adults, the resident wingless males make quick work of young eclosing competitors (Stuart et al. 1987). Cardiocondyla venustula (Hymenoptera: Formicidae: Myrmicinae) ants show an intermediate level of aggression among males vying for mating opportunities (Frohschammer and Heinze 2009, Jacobs and Heinze 2017). Older males find and defend territories within the nest which allows greater access to females. Those Cardiocondyla species that do not show male–male aggression have a mating strategy in which many sexually receptive females occur for short periods in large nests, thus resulting in limited opportunities for the males to compete for mating territories (Frohschammer and Heinze 2009, Jacobs and Heinze 2017).

In addition to the morphological and behavioral differences observed for fighting, differences exist in the pattern of sperm production. Winged males produce sperm early and store them in advance of their relatively short-lived nuptial flights, whereas wingless males produce sperm throughout their lifetime enabling them to mate with multiple queens within the nest (Heinze and Holldobler 1993). This, in turn, increases the likelihood of physical competition among the wingless males over a longer period of time. When greater numbers of queens are present, the colony produces more wingless fighter males (Cremer and Heinze 2002).

Females may also compete with other females for the purpose of reproduction. In Harpegnathos saltator (Hymenoptera: Formicidae: Ponerinae), workers of the same colony will engage in duels after the loss of the queen. During these duels, the workers strike each other with their antennae and although most workers engage for only a short period of time, some continue fighting for months and become gamergates (“pseudoqueens”) with noted changes in molecular, physiological, and behavioral measures (Opachaloemphan et al. 2021). Additionally, brain plasticity was demonstrated. Reproductive workers show reduced brain volumes, but brain volume was regained when the ants were reverted to a non-reproductive state (Penick et al. 2021). In contrast, research on harvester ants, Pogonomyrmex californicus (Hymenoptera: Formicidae: Myrmicinae), suggests that a lack of fighting among cooperatively nesting queens may be a way to enhance lasting colony survival and growth relative to single-queen colonies (Ostwald et al. 2021).

Recruitment

An increase in interactions with ants from an opposing colony often leads to an escalation of fighting (Hoover et al. 2016). A positive feedback loop is in play as increased interactions lead to increased recruitment, referring to the ability of a colony to bring more workers to a particular site. This, in turn, increases the probability of interacting with another ant. Two forms of recruitment have been described in this context [with several examples (Holldobler and Wilson 1977, Adams 1994)]. Short-term recruitment occurs after interactions with potential intruders and results in local clustering in the region of encounter. This recruitment involves rapid movements not seen when recruiting for the purpose of foraging. The ants will drag the middle (not rectal) abdomen presumably to lay scents from the sternal gland (Holldobler and Wilson 1977) or pygidial gland (Adams 1994). Long-range recruitment is accomplished by some of the ants who move over a larger area or lay a trail back to the nest. They release pheromones from the rectal gland and then prime responding nestmates with a tactile signal (Holldobler and Wilson 1977, Adams 1990). Recruitment success is tied to combat success. Larger groups of A. trigona won the majority of battles and were more likely to have increased recruiting rates (Adams 1990). When smaller colonies did win, it was correlated with a more rapid recruitment rate allowing for quicker growth in numbers.

Holldobler and Lumsden (1980) explored the effects of territorial structure and behavioral recruitment responses. When neighboring ants were placed within the territory of Oecophylla longinoda (Hymenoptera: Formicidae: Formicinae) African weaver ants, recruitment increased dramatically among the resident ants. Weaver ants demonstrated a more vigorous response to specific types of intruders, particularly predator ants or direct competitors for resources close to their nests where a stronger defense can be mounted (Hölldobler and Lumsden 1980). Wilson described minor workers of Pheidole dentata (Hymenoptera: Formicidae: Myrmicinae) recruiting both minor and major workers from the nest after encountering intruders (Wilson 1975, 1976). At the site of combat, the major workers (“soldiers”) outnumber the minors (although the reverse is true for the colony at large) and do most of the job of dispatching opponents. A full response is induced when even a single Solenopsis worker is detected, likely since a fire ant invasion poses a serious threat to the Pheidole colony. When the opponent numbers grow, fighting occurs closer to the nest where majors form a tight defense. In the face of defeat, the Pheidole will scatter carrying eggs, larvae, and pupae (Wilson 1975).

Combat Methods

When in a fight, ants can use two basic types of attacks. Nearly all ants can make effective use of their mandibles (Fig. 1A), which can be used for cutting or crushing opponents. Ants will tend to attack enemy weak points, such as eyes, antennae, legs, and the junctions between the head, thorax, and abdomen. The second weapon is venom, which is often dispensed via an abdominal stinger and is a potent weapon in the arsenal of many ant species. This can result in an evolutionary arms race between stronger venom and mechanisms to detoxify it. For instance, the tawny crazy ant, Nylanderia fulva (Hymenoptera: Formicidae: Formicinae), secretes a detoxifying substance and spreads it on their cuticle (LeBrun et al. 2014). This substance is useful at reducing the effectiveness of S. invicta fire ant venom, which could explain the displacement of invasive fire ants by invasive tawny crazy ants. Some ants can use venom in a ranged attack, allowing them to inflict damage on an enemy from a distance without immediate risk of injury to themselves. The African myrmicine ant, Crematogaster striatula (Hymenoptera: Formicidae: Myrmicinae), can use this method to effectively attack termite nests, and other species of ants were observed to retreat when confronted with this tactical advantage (Rifflet et al. 2011). A more sacrificial tactic is used by Colobopsis ants (Hymenoptera: Formicidae: Formicinae), which explode in combat, covering their foes with toxic substances that can inhibit or kill enemies (Laciny et al. 2018).

One simple way for ants to improve their combat power is better armor, usually thicker carapaces. For example, Megaponera (formerly Pachycondyla) analis ants (Hymenoptera: Formicidae: Formicinae) may be able to survive attacks by Dorylus rubellus driver ants (Hymenoptera: Formicidae: Dorylinae) by having a sufficiently thick carapace, rendering the driver ant mandible attacks ineffective (Moffett 2010). More unusual armor is also possible, as seen in a recent study on the leafcutter ant, Acromyrmex echinatior (Hymenoptera: Formicidae: Myrmicinae), which found that the ants benefit from the development of biomineral armor composed of magnesium calcite (Li et al. 2020). Such armor substantially improved survival during combat. Similarly, improvement of weapons such as mandibles with metals can effectively increase their hardness and sharpness. In the context of cutting leaves, this was seen in Atta cephalotes (Hymenoptera: Formicidae: Myrmicinae) (Schofield et al. 2021), though likely such improvements would also increase combat efficiency. Improvement of mandibles would also require fewer metals than increasing armor across the whole carapace. Another way to improve mandibles would be to specialize them for fighting against a particular opponent. This has not been documented thus far, but leafcutters have been observed to produce individuals with variable mandible morphology to enable increased access to different food sources (Püffel et al. 2021), suggesting the possibility that some of this variation may play a role in combat.

In many species of ants, castes of larger, powerful majors (often termed “soldiers” when they specialize in combat) can provide a significant advantage in battle, particularly when using tactics that combine their strengths with minor workers (Fig. 1B). In Pheidole pallidula (Hymenoptera: Formicidae: Myrmicinae), for instance, minors will work to immobilize enemy ants, and then majors (“soldiers”) will arrive to efficiently deliver the killing blow (Detrain and Pasteels 1992). These valuable colony members, which represent a greater investment for the colony due to their size, are often only mobilized to battle when enemy majors are present or for large-scale battles.

Another way to preserve colony strength is to attempt to rescue wounded ants, allowing them to later heal and continue to contribute to the colony. M. analis ants were observed to carry back injured nestmates after termite raids and perform wound treatment to avoid infection (Frank et al. 2018). This significantly increased the chance that their nestmates survived and thus reduced the cost to the colony of engaging in battle. Such behavioral adaptations may reduce the threshold required for colonies to undertake potentially resource-intensive behavior, such as raiding colonies of other social insects.

The need to minimize resource losses during battles can also affect the strategies of ant societies in other ways. Older workers usually undertake the riskier tasks, allowing younger workers to make important contributions in or near the nest, thus extending the average lifespan of workers and maximizing overall productivity (Whitehouse and Jaffe 1996). One study found that fire ants go even further. In staged encounters with ants from another colony, younger workers feigned death, which increased their chance of surviving the encounter (Cassill et al. 2008). In contrast, older workers did not show this behavior and fought the enemy ants. Another possible contributing explanation is that younger workers have a softer exoskeleton (Cassill et al. 2008), which reduces their fighting ability and may prevent them from effectively contributing to a battle.

The most obvious strategy in a battle is to bring superior numbers to the site of the confrontation (or possible confrontation). In some cases, larger numbers may provide an overwhelming advantage if these numbers can be brought to bear (see section on Lanchester’s laws below). The advantage of higher numbers of workers can be seen in several contexts. For example, prey being taken to nests by larger groups of Formica schaufussi (Hymenoptera: Formicidae: Formicinae) was stolen at much lower rates (Traniello and Beshers 1991). The effect of number of combatants in individual encounters was investigated in F. xerophila ants (Tanner 2006). In this study, groups of varying size were placed together, and then a varying number of these were placed in artificial arenas with a food source. F. integroides soldiers, which are somewhat larger than F. xerophila, were also placed in the arena. In general, F. xerophila were substantially more aggressive if initially placed in a large group, even when engaged in subsequent one-versus-one encounters. In such encounters, F. xerophila did not prevail in any fights and were occasionally killed, but in five-versus-five encounters, both species suffered similar average casualty ratios. Observations indicated that F. xerophila used “team-fighting” tactics, ganging up against a single enemy, to substantially increase their combat power (Fig. 1C).

On a larger scale, McGlynn (1999) examined the sizes of invasive ants species, concluding that individuals were usually smaller than native ants from the same genus. Examining the data more closely, this size difference was far more common among ants that fought for control of resources. This suggests that investing colony resources in superior numbers rather than larger individual ants could be a successful strategy to win battles between similar ants. Indeed, analyzing at colony size rather than individual size may be a better proxy for combat power, and this “super-organism size” could account for the invasive species’ competitive advantage. Colony size was not assessed in this study, but some reports have noted particularly large invasive colonies, as exemplified by Argentine ants where different colonies can form “supercolonies” that do not fight or compete with each other (Human and Gordon 1996, Holway 1999). Indeed, supercolonies appear to be over-represented among invasive ant species allowing for large group size with decreased intraspecific aggression (Holway et al. 2003, Buczkowski et al. 2023).

Numerical superiority can be negated in siege type warfare, where the fighting space is often limited in underground passages. This can allow a numerically disadvantaged colony to prevail or at least minimize their losses compared to an open field conflict when more combat power can be brought to bear (see Lanchester’s laws section below). Space is often even more limited at ant nest entrances (Baroni Urbani 1998, Fischer et al. 2015), so a common context of battles involves one side trying to deny the other entry. Special techniques can be employed in this situation, such as some species of Cephalotes (Hymenoptera: Formicidae: Myrmicinae) and Carebara (Hymenoptera: Formicidae: Myrmicinae) ants with flat “plug” heads that block the entrance from invaders (Baroni Urbani 1998, Fischer et al. 2015). In longer-term competition, a unique offensive technique is employed by Dorymyrmex (formerly Conomyrma) bicolor (Hymenoptera: Formicidae: Dolichoderinae) against various Myrmecocystus species (Hymenoptera: Formicidae: Formicinae) (Möglich and Alpert 1979). This involves dropping small stones into nest entrances, thus temporarily blocking them and substantially inhibiting the colony from carrying out normal operations outside the nest. Siege warfare can occur not only in underground nests, but also in trees (Riginos et al. 2015) where trunks and branches can restrict the number of invading ants able to attack simultaneously, thus allowing defenders to similarly mitigate a numerical disadvantage.

Lanchester’s Laws

With ants generally following a set of rules for their individual behaviors, mathematical modeling of ant wars is practical and can demonstrate well-known principles of combat. The best studied examples involve Lanchester’s laws (Lanchester 1916) (Table 2), which have previously been reviewed in the context of social animal warfare (Borges 2002, Clifton 2020, Green et al. 2020). Such laws state that combat power is continuously applied by a certain number of individuals with potentially different combat ability. In the “linear” form of Lanchester’s laws, the number of individuals participating in combat is limited by a certain frontage (Fig. 2A), referring to the width of space over which combat takes place. In this case, the total power of an army is proportional to the combat power of units that can fit at the front (which can be different between species, so the linear law does not necessarily imply one-on-one battles) multiplied by the total number of units (Lanchester 1916). This kind of situation could exist on a battlefield with lots of restricted terrain, such as inside a colony. It could also prevail in open terrain if there is a solid line between the two combatants that cannot easily be crossed by either side to bring more forces to bear on the opponent. In a second “square” form, each individual unit may engage in combat simultaneously, unlimited by frontage of a line or any other restriction (Fig. 2B) (Lanchester 1916). In this case, the total power of an army is proportional to the combat power of individual units multiplied by the number of individual units squared. Thus, if army A is outnumbered 2:1, its units must inflict damage at four times the rate of individuals in army B for both armies to have equal combat power. This was suggested as an explanation for the success of invasive ants, which are generally smaller than similar native species, thus potentially allowing them to field superior numbers of combatants with the same resources (McGlynn 1999).

Lanchester’s laws were first applied to ants by Franks and Partridge (1993), who suggested that ants would try to maximize their relative combat power against an enemy colony. Ants with greater numbers would use battle strategies that maximized the number of units that could simultaneously engage the enemy (resulting in a battle following Lanchester’s square law). Ants with inferior numbers would attempt to limit the battlefield in a manner allowing each side to bring forth only a fraction of their total force, fighting a series of one-on-one combats (and thus, follow Lanchester’s linear model). Army ant colonies, with their high numbers, would be expected to follow the former strategy. Indeed, they have been observed to attack from all possible directions, maximizing the width of the front lines and potentially allowing them to bring greater numbers to the fight (Franks and Partridge 1993). Furthermore, despite the presence of larger, more powerful workers in army ant colonies, the ants participating in these attacks are overwhelmingly medium-sized workers, allowing a greater total number of ants to occupy the front line (Fig. 2C). Similarly, Whitehouse and Jaffe (1996) found that significantly more soldier class Atta laevigata (Hymenoptera: Formicidae: Myrmicinae) were recruited to simulated vertebrate disturbances than to disturbances by other ants, where they were not preferentially recruited. They further observed that in induced battles between Atta colonies, media ants participated in the bulk of the fighting (though this class was similarly responsible for most foraging and thus, initiation of combat). Interestingly, large numbers of minima ants were recruited to the battle site, though these mostly did not participate in the battle, which is not necessarily consistent with predictions from Lanchester’s square law. Slave-making ants, by contrast, usually have significantly inferior numbers compared to their opponents, and thus, they prefer to fight one-on-one battles where their individual superiority becomes more significant (Franks and Partridge 1993). This is accomplished by the use of signaling substances, which disorder the enemy colony, preventing them from using team-fighting tactics to their fullest potential.

The effect of group size was examined in conflicts between natural populations of acacia ants (Palmer 2004). The study involved four similar species, Crematogaster mimosa, Crematogaster sjostedti, Crematogaster nigriceps (all three Hymenoptera: Formicidae: Myrmicinae), and Tetraponera penzigi (Hymenoptera: Formicidae: Pseudomyrmecinae) in a natural setting. Estimated colony size was found to be the most important predictor of success in these battles, but the death count on both sides was approximately equal, consistent with observations that most fights were one-on-one. These findings held true when the number of combatants was varied by preventing some individuals from reaching battle sites. This indicates support for Lanchester’s linear law, which is consistent with the confined terrain in which these battles took place that limited the size of the front line and thus the total combat power that could be brought to bear at one time.

Several studies attempted to demonstrate Lanchester’s law by direct experimental manipulation. McGlynn (2000) used a natural environmental setting in Costa Rica. Bait was placed in enclosures with large or small openings, to allow entry of variable numbers of ants, mimicking the situation described in Lanchester’s square or linear laws. Instances where small, medium, or large ants ended up fully controlling the bait food resources were recorded, and it was found that large ants controlled the enclosures with small openings more frequently than smaller ants, in support of Lanchester’s linear law. However, the complicated experimental situation makes the conclusions of this study less firm, with foraging success and dynamic modulation of aggression potentially having a larger effect than potential combat, particularly given the small total numbers involved.

Another experimental study by Plowes and Adams (2005) attempted to test the application of Lanchester’s laws to fire ant battles. Size-controlled ants from different colonies were mixed in isolated containers, and the casualties were counted at the end of the resulting 24-h battle. In this study, the linear law was found to be a better match to the data than the square law, indicating that the groups of ants were not able to bring their full potential combat power to a confrontation due to the battlefield’s geometry. However, individual fighting quality was also estimated simultaneously (and varied widely, even between similarly sized ants), which may have confounded the analysis, since several combinations of fighting quality and geometry parameter (which could be linear, square, or the space in between) could produce similar outcomes. The authors did attempt to control for this by mixing ants at several ratios for each group. It is also unclear how the artificial “staged fight” setups of this and similar studies represent actual confrontations in natural settings when ants may use recruitment to more readily bring greater numbers to bear in a fight.

Batchelor and Briffa (2010) investigated the applications of Lanchester’s laws to wood ant (Hymenoptera: Formicidae: Formicinae) battles measuring the individual body size, number, and energy levels (as measured by sugar concentration in individuals) of each side in staged combats. They found that a group containing individuals with greater individual body mass was more likely to inflict the first casualty on the enemy, while group size and energy levels did not significantly affect this. However, this outcome measure may not be suitable for investigating the long-term outcome of battles, particularly in natural conditions, since larger size may be dominant in artificial encounters where ants are less able to use higher numbers (or bring them to bear quickly), putting a premium on individual fighting strength. A follow-up study by the same authors also examined staged conflicts in wood ants, obtaining several pictures of each fight for analysis (Batchelor and Briffa 2011). Larger groups here were actually able to bring greater numbers to bear in a fight, but they also had a larger fraction of unengaged individuals than smaller groups. Combat performance, as measured by enemy casualties, was improved with increasing individual body mass and, to a lesser extent, higher group size, potentially indicating weak support for Lanchester’s square law, or at least some level of group-size effect beyond linear.

A recent study investigated termite battles at multiple ratios in petri dishes, also tracking the progress of the battle’s course as it occurred, rather than relying only on final casualty estimates (Clifton et al. 2022). Though not ants, termites tend to move and fight like ants in similar social colonies, making the lessons from this study potentially broadly applicable to considerations for ants. Based on analysis of these staged termite battles, Lanchester’s square law provided a better match for the results than the linear law, but the stronger army still had somewhat less combat power than predicted by the square law. Additionally, closer investigation revealed that substantial portions of both armies, even if equally sized, remained unengaged in the battle. This was counterbalanced by the fact that two or more termites ganging up on a single enemy are able to inflict much higher damage than termites involved in one-versus-one confrontations. Evidence from other sources supports this notion (le Moli and Parmigiani 1981, Powell and Clark 2004, Tanner 2006, Clifton et al. 2022), making it a potentially important consideration for modeling of ant battles.

One-on-one battles of the meat ant I. purpureus showed no relationship between the outcome of the fight and the size of the worker (Han et al. 2023b). This may have been due to small differences in worker size, but more likely, the statistical power of the study simply was not enough to detect size-level differences. This was partially because only a small fraction of worker conflicts had a clear winner. In competitions over food between colonies of the same species, the larger colony usually prevailed, though in some cases, there was still no clear winner, with both colonies retaining access to the food. This potentially implies that both colonies could cease fighting if the cost would likely be high compared to the benefit of accessing a single food source. Average worker size in these colony-level conflicts also had no effect.

Most recently, various combinations of army size and individual unit strength were assessed both computationally and experimentally. Using a computer game, the authors found that an army with strong units tended to prevail over larger armies more frequently in complex battlefields, defined as having more patches of terrain that would prevent movement of individuals (Lymbery et al. 2023). Based on Lanchester’s laws, this was the expected result. The restricted terrain limited the number of units that could be engaged more severely, pushing the result more closely to the linear law, where army size gave less advantage. The authors also experimentally tested combat between the larger meat ant I. purpureus and the smaller Argentine ant L. humile. Though the variance in outcome was high, their results were similar to the simulations, with restricted terrain improving the outcome for the stronger meat ants. This study generally supports the notion that restricted terrain leads to outcomes predicted by Lanchester’s laws, with restricted terrain limiting the number of individuals that can engage simultaneously and thus favoring armies with strong individuals. However, the quantitative details of both the simulations and experiments do not support the use of the square law in open terrain, giving an outcome close to that mathematically predicted by Lanchester’s linear law. The authors note that this was partially due to nonbattle deaths, but other factors could also have influenced it, such as the units using a somewhat linear formation in the open engagements (preventing all units for the side with greater numbers from engaging until they move around the sides of the line), or simply there not being enough physical space for all units to engage, even if the units from the side with smaller numbers are completely isolated from each other.

Overall, these studies suggest that Lanchester’s laws provide a good starting point for analysis of large-scale ant battles, but they may fall short for accurately approximating and understanding outcomes (Table 2). Perhaps most importantly, further studies should attempt to avoid artificially staged encounters, or at least present them in semi-natural situations, such as induced battles between natural or laboratory colonies, rather than simply mixing various combinations of ants together. Though allowing more natural behavior to be observed, this could be difficult in practice, since an ant colony at a severe disadvantage (such as one dependent on Lanchester’s square law) would be substantially less likely to engage in a full-scale battle unless their colony was under attack, and even then, they would need to be able to mount a coherent defense quickly enough for the subsequent battle to be easy to interpret mathematically. In general, there would be large evolutionary pressure to avoid the consequences of being highly outnumbered in Lanchester’s square law, with retreat and other measures such as colony abandonment that could still carry negative consequences often being preferable to losing a large force without being able to inflict significant damage in return.

There is also much room for improvement in the models themselves. Adams and Mesterton-Gibbons (2003) considered that Lanchester’s law in the pure linear or square forms was unlikely to capture the true dynamics of combat in social groups, and they developed modifications to address certain perceived issues. Particularly, the square law was thought to overestimate the advantage accrued from possessing numerical superiority, with the actual value falling between the linear and square laws. The recent study of termite combat in a laboratory environment by Clifton et al. (2022) supports this notion. Adams and Mesterton-Gibbons (2003) further argue that individuals with higher fighting power are not only able to inflict more damage, but also are able to suffer more damage before being incapacitated. This can potentially be modeled with the current form because “fighting quality” is essentially a constant representing the attacking power of one type versus the defense strength of the other type. However, when attempting comparisons between three or more species, this awkward use of ratios breaks down. Thus, going even further, we propose that future work could also incorporate not just fighting quality, but a separate and independently varying “toughness” quality, allowing models to better simulate interspecific confrontations, or those between different castes. This would allow representations of situations, for example, where large ants with thick cuticles are highly resistant to damage, but still not able to contribute much more attacking power than smaller ants. However, it should be noted that for some ant species, attacking power and toughness may still not be constant when considering all possible opponents, with the possibility, for example, of “paper-rock-scissors” advantages between three types that could not be explained by fixed model parameters.

For situations with linear or close to linear geometry, the width (the space needed by an individual to engage in combat) should be considered, which could vary substantially between species and castes of different sizes. In such cases, smaller ants are already thought to be better front-line fighters due to their greater numbers, but the increased advantage of ganging up on an enemy may further amplify this consideration. However, it should be noted that for species with castes of substantially different size, maximum combat power could potentially be brought to bear if different castes are able to co-occupy the front line without substantial interference due to their different heights. Finally, maximum fidelity could potentially be obtained by quantifying the specific proportions of ants that are unengaged, engaged in one-on-one battles, engaged against multiple opponents, and engaged together with comrades against a single opponent, with variable damage output in each of these situations. These could be potentially complex functions of total numbers on each side as well as battlefield geometry. Such considerations could replace Lanchester’s laws at the cost of greater difficulty in obtaining these measurements.

Other Mathematical Models of Combat

While most mathematical modeling of ant combat has been based around Lanchester’s laws, other frameworks may be more appropriate in some instances (Green et al. 2020). The earliest instance of such work examined the evolution of giant colony size in Dorylus army ants (Hymenoptera: Formicidae: Dorylinae) compared to Eciton army ants (Hymenoptera: Formicidae: Dorylinae) with a more modest colony size (Boswell et al. 2001). In their model of colony success, large colony size was particularly important for fighting power with intraspecific colonies, which was far more common in Dorylus due to the greater density of these ants. It also accounted for the propensity of Dorylus colonies to divide asymmetrically, preserving combat power in the main colony, compared to Eciton colonies, which undergo more symmetric colony fission.

A model for fights between Lasius paralienus and invasive Lasius neglectus (Hymenoptera: Formicidae: Formicinae) was developed based on a framework for modeling spatially explicit chemical reactions and repurposed to represent ants in battle (Santarlasci et al. 2014). It was then compared to staged fights in arenas. Of note, since this study was agent-based, it could better predict how individuals meet and interact compared to simpler Lanchester-based models. The study drew attention to the high level of variance in outcomes compared to more commonly used deterministic predictions.

Tetramorium immigrans pavement ants (Hymenoptera: Formicidae: Myrmicinae) establish territories based on many extended battles with low mortality between neighboring colonies. One study built several models of recruitment to battle sites inspired by field observations of such ants (Adler et al. 2018). Larger colonies were found to be able to hold more territory per worker, possibly due to the ability to recruit superior numbers to a battle site. However, smaller colonies could still hold nearby territories due to reduced travel time from the nest to the battle site, which provides an advantage in recruitment (Fig. 2D).

Another study also modeled abstract ant battles with individual-based simulations (Adams and Plowes 2019). Based on several rules involving recruitment and search patterns, three types of battles were predicted: those in which the fighting group occupy an oval area, those with elongated shapes when ants search along trails, and linear battles when fighting ants physically block those from either side from passing, forming a line of increasing width as more ants start fighting. The fighting area usually moves toward the colony that is at a local disadvantage, quickly in battles along trails and slowly for linear battles. Smaller colonies were usually at a disadvantage, but they could make up for this by recruiting more efficiently. In most cases, battles were resolved quickly when one side obtained an early advantage, but some struggles were more drawn out with higher casualties.

Mathematical modeling of ant fights has also generally been narrowly focused on considerations of fully engaged individuals. However, more could potentially be learned by taking cues from well-understood aspects of human conflict at the strategic level. These could include the objective of combat as well as logistical aspects of conflict. For example, an ant colony with inferior numbers and individual fighting quality could potentially still win an engagement if it is able to bring its full numbers to bear more quickly than its opponent, thus defeating the enemy in detail by always substantially outnumbering them at the point of conflict. An isolated battle for a distant food source would also produce substantially different reinforcement patterns than a battle around a nest, or a moving battle between colonies.

Other Studies

We here review several other research studies on ant battles that do not fall solidly into one of our previous topic categories. These are sorted by interesting focal species and together represent the remainder of the current literature in the field.

Army ants often specialize in honing in on the odors of specific prey species (Manubay and Powell 2020), representing a major danger to even large colonies of other types of ants. Indeed, some of the largest ant battles occur when leafcutter ants engage army ants. In one observation, Nomamyrmex esenbeckii ants (Hymenoptera: Formicidae: Dorylinae) attacked an A. cephalotes nest (Swartz 1998). This occurred by a single stream of army ants on the move, and it was quickly met by a force of leafcutter majors, barricading some nest entrances and forming a defensive ring. These were individually removed by army ants, which then penetrated the nest and apparently continued fighting underground, removing several larvae from the nest. A few days later after eliminating the leafcutter ant colony, the army ants exited the nest. Several thousand dead ants on both sides of the battle were observed. Another study reported similar observations, noting also that both sides of this conflict used team-fighting tactics to bring down opponents (Powell and Clark 2004). In these cases, a major would be the main combatant, with smaller ants providing assistance by immobilizing opponent legs. Looking at several raids, Powell and Clark (2004) also reported that in many cases, larger leafcutter colonies could successfully repel the army ant attack, usually when they could form a quick response to the invasion while it was still in early stages and had not penetrated deep into the colony. Complete elimination of the leafcutter ant colony was uncommon, even for small colonies. Subterranean raids were also observed against the fungus-growing Trachymyrmex arizonensis ants (Hymenoptera: Formicidae: Myrmicinae) by Neivamyrmex rugulosus army ants (Hymenoptera: Formicidae: Dorylinae) (LaPolla et al. 2002). In these cases, the T. arizonensis evacuated some of their brood when the army ants approached. Pheidole big-headed ants, on the other hand, employ strategies led by their super-major workers against various species of Neivamyrmex army ants (Huang 2010). Initially, major workers, with assistance from minors, fight outside the nest, possibly to delay an approaching army ant column. Then, recruited super-majors attempt to block the nest entrance with their characteristic large heads. In one event, big-headed ant workers of all types counterattacked the army ants after the army ants had retreated from the nest entrance. An attack on the army ant trail disrupted potential army ant reinforcements, eventually leading to abandonment of the raid. Not all ants fight defensively against army ants, however. Red tree ants (O. longinoda) were observed to remove individual medium-sized Anomma driver ants (Hymenoptera: Formicidae: Dorylinae) from their columns, killing and consuming them in groups (Gotwald 1972).

With their large colony size and specialized food source, mature leafcutter ant colonies rarely fight against different species, save for colony defense against army ants or large predators. However, one report described Paraponera clavate (Hymenoptera: Formicidae: Paraponerinae), which nest in or near trees that are also their foraging site, defending these trees against the leafcutter A. cephalotes (Wetterer 1994). The P. clavate attacked a leafcutter foraging column, first carefully removing and killing individuals from its outer defense and then proceeding to launch a massed attack against the foraging column itself, which was successful in driving off the leaf-cutter column. Battles between leafcutter ants are more common, with each establishing defended territories. Survival and growth of small colonies tended to be severely curtailed when closer to large colonies (Jofre et al. 2022). However, it was unclear what proportion of this effect was from indirect resource competition, as opposed to direct aggression.

Wars among the wood ant Formica polyctena (Hymenoptera: Formicidae: Formicinae) have been the subject of several studies due to high intraspecific aggression (Moli and Parmigiani 1982). One extensive investigation tracked several colonies over a large dune area in the Netherlands (Mabelis 1978, 1984). In spring, many battles lasting up to several days took place between these wood ant colonies, especially during warmer weather. Dead enemy ants were consumed, and larger colonies gained more territory during these encounters. During summer and fall, less aggression was observed, likely due to smaller territory size during summer and lower protein requirements in the fall compared to the spring when new colonies are produced. A follow-up study indicated that other wood ants represented 86% of one colony’s prey intake during spring (Driessen et al. 1984). Of note, one colony was observed to battle two others simultaneously. If being in such a strategically disadvantageous position is common, future studies could potentially investigate how colonies respond to such situations. Another species of wood ant, Formica aquilonia (Hymenoptera: Formicidae: Formicinae), is usually more peaceful, though aggression appears to substantially increase in logged areas (Sorvari and Hakkarainen 2004), possibly due to the need to secure more limited resources in these damaged, less productive habitats.

The Argentine ant, L. humile, is highly invasive on multiple continents, displacing many native species in part due to their superior combat capacity (Angulo et al. 2024). In a Northern California study, Argentine ants often consumed new winged queens of rivals, and they won the majority of battles with native ant species, eventually displacing them from a portion of the study area (Human and Gordon 1996). Though not individually superior to native ants, their colonies usually had a substantial advantage from superior numbers (Holway 1999) and initiated more aggressive encounters (Human and Gordon 1999, Zee and Holway 2006). This large colony size can be partially explained by the formation of supercolonies, separate but sufficiently genetically close Argentine ant colonies that do not engage in conflict with each other (Thomas et al. 2006). A recent study examined interactions between Tapinoma nigerrimum ants (Hymenoptera: Formicidae: Dolichoderinae) and Argentine ants in the Mediterranean (Leonetti et al. 2019). Based on their local dominance, observed competitive exclusion with Argentine ants, and ability to be invasive in other environments, T. nigerrimumare was thought to potentially have the capacity to slow Argentine ant invasions. However, it was found that T. nigerrimum were substantially less aggressive than Argentine ants, which attacked to control food resources and invaded T. nigerrimum colonies. Invasive Argentine ants even appear capable of displacing other invasive ant species, such as big-headed ants in Bermuda (Wetterer 2017) and fire ants in the United States (LeBrun et al. 2007).

Conclusions

Overall, ant battles remain understudied, perhaps because this topic lies at an intersection of the fields of ecology, evolution, animal behavior, mathematical modeling, and conservation biology. Yet, the tractable nature of these systems combined with their intrinsically high interest levels among both scientists and the general public means that at least some studies have been undertaken regularly since at least the 1970s (see reference list), with 7 publications in the 1970s, 8 in the 80s, and now reaching over 30 per decade. These initially were observational in nature, but increasingly have made use of models and experimental manipulation, combining different techniques for a more rigorous analysis. This research has revealed many interesting components of ant interactions, ranging from the individual-based decision-making characteristics of colonies to battle strategy and tactics, and even attempts to understand the outcomes of battles from a quantitative and statistical perspective.

Aside from intrinsic interest in these topics, research on ant fighting can potentially provide useful knowledge to other fields, such as similar competition among termites (Tuma et al. 2020, Clifton et al. 2022) or even battles among larger animals such as groups of mammals that are less tractable to study and manipulate experimentally compared to ant colonies (Wilson et al. 2002, Green et al. 2022). In such battles, individual leaders could potentially change outcomes compared to conflict in insect societies, particularly in decisions of whether or not to retreat in various circumstances. Another area where this research can be useful is for conservation purposes. By understanding the competitive advantage of invasive ant species over their native counterparts, strategies to counteract these invasions could perhaps be more easily devised or optimized. In regions where local ants offer more competition, for example, less intensive control measures may be needed, and even gene drive-based methods would enjoy more success (Liu et al. 2023), which may be particularly important for haplodiploid species (including all ants) where suppressive power is lower (Liu and Champer 2022). Examining the decisions of individual ants in complex combat environments could prove useful for studying human conflicts from an historical perspective. In ancient warfare, humans would often be involved in melee combat, and the losing army was usually the first where soldiers disregarded commanders’ orders and fled the battlefield. This would often be done with information from only a limited segment of a battlefield, more akin to how ants would make decisions on when to retreat from a battle. Studies on ant warfare may even have implications for weapons development in matters such as conflicts between semi-autonomous drones. For example, when would a “bottom–up” approach that ants or drones might use run into deficiencies that could be improved with a minimum of “top-down” coordination? Conversely, analysis of human combat strategies to avoid some disadvantages of being outnumbered (and the attendant consequences of Lanchester’s square law) may shed light on ant behavior in similar situations. Perhaps ant nests may have some defense features similar to castles or other fortifications in species at high risk of colony raids.

Looking toward future research in this area, there are many species that have not been examined at all in the context of ant fights. These are likely to employ many interesting tactics, weapons, and other innovations that have yet to be discovered, perhaps particularly so in the context of evolutionary arms races. Another major area of future work could be increasing the fidelity of quantitative models, both for how a colony allocates its resources, choosing when and where to fight, and also for the battles themselves. Such advanced models could still be derived from Lanchester’s laws. With studies of conflict in ants still in their infancy, the many pathways for future research promise to be richly rewarding.

Funding

J.C. was supported by laboratory startup funds from Peking University and the Center for Life Sciences, the SLS-Qidong Innovation Fund, and the NSFC Overseas Youth Fund.

Author Contributions

Jackson Champer (Conceptualization [Equal], Investigation [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), and Debra Schlenoff (Conceptualization [Equal], Investigation [Equal], Writing—original draft [Equal], Writing—review & editing [Equal])