Insectes Sociaux

, Volume 58, Issue 3, pp 281–292

The largest animal association centered on one species: the army ant Eciton burchellii and its more than 300 associates


  • C. W. Rettenmeyer
    • Department of Ecology and Evolutionary BiologyUniversity of Connecticut
  • M. E. Rettenmeyer
    • Department of Ecology and Evolutionary BiologyUniversity of Connecticut
  • J. Joseph
    • Department of Ecology and Evolutionary BiologyUniversity of Connecticut
    • Department of Ecology and Evolutionary BiologyUniversity of Connecticut
REVIEW ARTICLE (C.W. Rettenmeyer memorial paper)

DOI: 10.1007/s00040-010-0128-8

Cite this article as:
Rettenmeyer, C.W., Rettenmeyer, M.E., Joseph, J. et al. Insect. Soc. (2011) 58: 281. doi:10.1007/s00040-010-0128-8


As possibly two of the last true naturalists, Carl Rettenmeyer and his wife Marian dedicated their lives to the study of army ants and their associates. Over the course of 55 years, the Rettenmeyers went on numerous field trips mainly to the Central American tropics and analyzed hundreds of self-collected samples and those sent by a multitude of other scientists, who were inspired by Carl’s enthusiasm. It comes as no surprise that Carl Rettenmeyer became the world’s leading expert on army ant associates. This paper, which the Rettenmeyers nearly completed before Carl’s death in 2009, gives the first comprehensive list of animals known to be found in the company of a single army ant species: Eciton burchellii. The 557 recorded associates range from birds to insects and mites and comprise the largest described animal association centering around one particular species. Although some of these associates may be opportunistic encounters, we are confident that approximately 300 of the recorded species depend on the ants, at least in part, for their existence. The extinction of E. burchellii from any habitat over its vast area of distribution is likely to cause the extinction of numerous associated animals at that site. This overview will hopefully inspire researchers throughout the world to follow in the Rettenmeyers’ footsteps and continue the investigation of army ants and their associates.


MyrmecophilesAnt guestsSymbiosisArthropodsTropical ecologyBiodiversity


Army ants are functionally defined by sharing a suite of life-history characteristics: they are carnivorous and raid for food in groups of hundreds to thousands of individuals, their queens are permanently wingless, and the whole colony emigrates periodically. Army ants can be found in the tropical and subtropical regions throughout the world. This paper, however, concentrates only on one neotropical army ant species. All New World army ants belong to the subfamily Ecitoninae (Hymenoptera: Formicidae), which includes about 150 species in five genera (Watkins, 1976). The majority of these species are primarily subterranean and rarely seen on the ground surface. Although a few species raid in columns above ground, only two species, Eciton burchellii (Westwood) and Labidus praedator (Fr. Smith), forage in large swarms above ground. Of these two species, only Eciton burchellii create temporary nests or bivouacs above ground in brush piles, on tree trunks or under or inside logs and hollow trees (Fig. 1). An average E. burchellii colony has about 500,000 workers (Franks, 1985) and follows a strict cycle of statary and nomadic phases (Rettenmeyer, 1963a; Schneirla, 1971). During the 20-day statary phase, the ant pupae and newly laid eggs develop. While statary, each daily raid, starting from the central bivouac, generally proceeds in a different direction, which reduces the chance that the ants raid the same area twice within a few days (Franks and Fletcher, 1983). At the end of the statary phase the eggs hatch and the pupae eclose, initiating the nomadic phase. During this period, when the thousands of larvae develop, the colony raids every day and follows the raid with an emigration almost every night to a new bivouac site. After 15 days the larvae pupate in synchrony, and the colony becomes statary once more. Colonies grow for approximately 3 years, after which the largest colonies reproduce through colony fission: The parental colony splits into two parts, one headed by a daughter-queen, the other by the original queen or another of her daughters (Franks and Fletcher, 1983).
Fig. 1

The bivouac nests of E. burchellii are made up of the ants themselves. By hooking their tarsi to one another, the ants create a biting and stinging mass in whose center the queen and brood are well protected. Photo: Stefanie Berghoff

The spectacular swarm raids of E. burchellii, with several thousand ants swarming over the ground surface and occasionally up into the forest canopy, are guaranteed to impress any visitor to tropical forests (see e.g. Bates, 1863). Within a single day’s raid, the ants can retrieve up to 30,000 prey items (Franks, 1982; Fig. 2). The fierce reputation of these ants, however, has been exaggerated even by biologists. The entomologist and later director of the U.S. National Zoo, William Mann (1934) wrote: “Even men flee as the mighty column writhes through the jungle, wiping out all insect and animal life in its path.” In Mann’s defense, we suspect that the editors of the National Geographic Magazine felt compelled to enhance his language. Although E. burchellii occasionally kills small vertebrates, it cannot cut skin or flesh off bones. Thus, vertebrates do not belong to the prey spectrum of this army ant, whose typical prey is larvae and pupae of other ants and wasps and many other adult arthropods (Schneirla, 1971). By attacking in large groups, E. burchellii can overwhelm arthropods that are many times its size. It seems like a paradox that such a predatory ant species has more animal species associated with it than any other species so far reported.
Fig. 2

Eciton burchellii workers carrying prey items back to the bivouac. Photo: Stefanie Berghoff

Wasmann (1887) described the first beetles collected from E. burchellii. Seven years later, he published the first catalog of arthropods reported from ant and termite colonies throughout the world (Wasmann, 1894), already listing 1,177 ant-guests. Since 1894, this is the first attempt to provide a comprehensive list of ant associates—in our case, we list all the species associated with the single army ant species E. burchellii. Our collections and the publications we have surveyed add up to an amazing number and diversity of E. burchellii associates. Because the ants’ raiding habits and swarm sizes are unique among neotropical army ants, we are confident that the ants have been correctly identified in hundreds of published records going back more than 100 years.

We will discuss the associates of E. burchellii in order of three main ecological groups: (1) associates found at the ants’ swarm raids, (2) associates found in the ants’ refuse deposits, and (3) inquilines found in bivouacs and columns of the army ants.

Studying Eciton burchellii and its associates

It has become customary to call ants the “host” of associated “guest” arthropods, independent of any knowledge about any biological connection that might exist between the species. Wasmann (1903) created a behavioral classification of ant guests, using the terms symphile, synechthran, and synoekete based on his imagining various behaviors between the ants and their guests. Wasmann never went to Central or South America and never saw living army ants or any of the arthropods living with them. His speculations distort the behavior of these arthropods, and we will not use his classification. Following this first classification, other authors have proposed different systems classifying animals living with army ants, based on the behavior, location, feeding preferences, degree of specialization, or on a combination of these (see Gotwald, 1995 for an overview). However, no system is completely satisfactory, which is in part due to the fact that we know so little about most animals found with army ants. We thus prefer and will use the term “associate” since it indicates that there is some connection between two species but does not imply benefit or harm to either.

It is likely that many associates benefit in some way by eating the food the ants collect or the waste they discard. Such associates can be referred to as commensals. Associates that live in the nest of the ants are called inquilines. There are a few ectoparasites on E. burchellii, yet we know of no diseases they may transmit. Similarly, no interior parasites of E. burchellii have yet been documented although a few phorid flies and possibly some diapriid wasps may be internal parasites of the ants. Associated with E. burchellii are also many phoretic species—mites and some insects that ride on the ants for transport. Phoretic species have often been misinterpreted as ectoparasites or as commensals being carried by the ants. If a beetle is under the head of an ant, it might be carried, but most often it is phoretic.

Eciton burchellii has a wide distribution and ranges from Mexico to southern Brazil, Paraguay, and Bolivia (Watkins, 1976). Throughout its range, E. burchellii is divided into five subspecies (Borgmeier, 1955), which differ in part in their geographical range, color, and habitat. The status of these subspecies is still disputed, although currently under investigation (E. Rodriguez, pers. comm.). We have observed four E. burchellii subspecies and found no striking differences in their basic behavior. For the scope of this paper, we will therefore treat all the subspecies of E. burchellii together.

We started our personal collection of army ants and their associates in 1952 and now have data from over 1,200 colonies. We supplemented these records with samples from 345 colonies sent to us by other collectors. With the help of several field assistants and two excellent ant sorters, we have examined the more than one million E. burchellii adults and immatures in our collection for associates. We collected flies over swarm raids by net. Insects in ant raid and emigration columns were collected using straight tube aspirators. In addition, we collected large ant samples from bivouacs, anesthetized with ether or carbon dioxide. After we removed the associates, the ants were returned to their colonies. Some ant samples were also preserved directly in alcohol and later sorted under a dissecting microscope. Associated arthropods were sent to experts for identification (where available). Voucher specimens are found in the collection of the identifying taxonomists and in the Carl W. and Marian E. Rettenmeyer Ant-Guest Collection at the University of Connecticut, the world’s largest collection of arthropods associated with neotropical army ants.

Associates of E. burchellii: results and discussion

Overall, we recorded 557 species associated with E. burchellii (Table 1), with hundreds of specimens still awaiting identification. Most associated arthropods are poorly known taxonomically and large groups of associates such as Collembola, mites and staphilinid beetles remain undescribed—mainly due to a shortage of taxonomists. While taxonomic information of arthropod associates is often unsatisfactory, data on their biology are, in general, missing completely. It is very rare that larval and adult stages of an associated arthropod are known. The entire life cycle for any of these species is unknown. It is likely that some of the recorded associates were chance encounters of the ants (especially some of the Lepidoptera). However, if only those associates are counted for which we have some indication that they depend on the ants for part of their living, approximately 300 species remain in this list. Details on all the recorded families and species such as scientific names, collection sites within the ant colony, and relevant references are given in supplementary Tables 1–9.
Table 1

The associates of Eciton burchellii. Detailed information on the associate groups is found in Supplementary Tables 1–9


Common name












Several hundred unidentified specimens in our collection







Association of several species unresolved







Several hundred unidentified specimens in our collection; 403 Ptiliid specimens part of a current revision







>500 unidentified specimens in our collection, some possible leaf litter species







Association of this single species is unclear







>850 unidentified specimens in our collection







Numerous unidentified specimens in our collection







List likely includes several opportunistic species












557 (ca. 300 species if an estimated number of opportunistic species is excluded)

BV bivouac, EC emigration column, LT light trap near bivouac, RC raid column, RD refuse deposit, SR swarm raid

aAs hardly any mite taxonomists are available in many cases, only the higher taxonomic groups can be identified, leading here to a higher number of families than genera

Associates found at swarm raids


Throughout Central and South America, over 200 species of birds have been observed foraging at the swarm raids of E. burchellii. Birds are primarily found with this army ant, but some birds are also found at the smaller swarms of L. praedator (Faria and Rodrigues, 2009; Kumar and O’Donnell, 2007; Swartz, 1997). Occasionally, birds may also follow other species of army ants (Willis and Oniki, 1978).

It is likely that almost any insectivorous bird in the vicinity of a swarm raid takes advantage of the food the ants indirectly provide by flushing thousands of arthropods and small vertebrates out of the leaf litter as the ant swarm advances. Many of these flushed prey animals are nocturnally active or would otherwise stay hidden in the leaf litter. By flushing them out, the ants provide an important service to the birds. At some swarms the birds are so abundant that they significantly reduce the ants’ success rate in capturing prey, indicating that the birds may act as cleptoparasites of the ants (Wrege et al., 2005). When we observed and videotaped swarm raids of E. burchellii, we noticed that the ants are much more selective than one would believe from the literature. For example, we found that millipedes and some species of cockroach, walking sticks, and katydids apparently have some kind of defense, protecting them from E. burchellii. If these species are preyed upon by the antbirds, the impact smaller bird aggregations have on the ants’ raiding success may become less pronounced. We strongly encourage all observers of antbirds to try to get species identifications of their prey. As insects picked up by antbirds at army ant swarms are often visually identified over some distance, we assume that there is a strong bias for identifying large prey that are more readily seen (but see Chesser, 1995).

Often people have assumed that the birds are feeding on the ants. However, a number of studies (see e.g. Willis and Oniki, 1978; Del Hoyo et al., 2003) as well as our own observations (Rettenmeyer, 2009) have conclusively demonstrated that the few army ants eaten are probably attached to the prey the birds eat.

Some birds are known to derive more than half of their food from army ant swarms and thus to depend on the ants for their survival (see Willis and Oniki (1978) for references to many early papers and Del Hoyo et al. (2003) for an extensive review of the literature for all species of antbirds). Obligatory antbirds are mainly found in the families Thamnophilidae (the “typical” antbirds, see e.g. Fig. 3), Formicariidae (ground antbirds), and Furnariidae (subfamily Dendrocolaptinae; woodcreepers). We list 29 species which we believe to be the most complete summary of obligatory antbirds (Suppl. Table 1). In addition to these obligate antbirds, many other bird species regularly attend E. burchellii swarms (see e.g. Chaves-Campos, 2003; Faria and Rodrigues, 2009; Kumar and O’Donnell, 2007; Roberts et al., 2000; Swartz, 1997, 2001). Although most of these species probably follow the ants opportunistically, the actual association may not be easily determined. Estimating the proportion of ant-derived food in the birds’ diets is a difficult task, and the birds’ behavior may vary between populations or over a bird’s life span (Swartz, 2001; Willis and Oniki, 1978). Moreover, the frequency of a bird species observed at ant swarms may not be an indicator of its dependency, but of its commonness in the area. Building on the assumption that obligate antbirds must keep track of several ant colonies in order to maintain a constant food supply, Swartz (2001) suggests using “bivouac checking behavior” to distinguish between obligate and opportunistic antbirds. At least four obligate antbirds were regularly observed flying (or walking) to an ant bivouac, observing it for some time and then flying off to, presumably, check on another colony (Swartz, 2001). Although the recording of such behavior will involve many hours of quietly observing the vicinity of army ant bivouacs, we encourage anyone studying antbirds to spend some time doing this. In this way, we may refine our knowledge about the dependency status of many antbirds.
Fig. 3

The Ocellated Antbird (Phaenostictus mcleannai, family Thamnophilidae) is regularly found at swarm raids of E. burchellii, from which the bird obtains the majority of its food. Photo: Adam Fuller

The means by which birds locate the army ant bivouacs is still unknown. The birds may remember the bivouac location from previous days. Another possibility is that the birds find the bivouacs by smell. Eciton burchellii has a strong and characteristic odor that is species specific and very different from the odor of other Eciton species. We know of no tests that have been done to see if the birds can smell the army ants and thus locate the ants’ swarms or bivouacs. In the light of recent discoveries, showing that the sense of smell in birds may be more important than generally believed (Steiger et al., 2008), it would be interesting to investigate how birds find the ants.


Zikán (1929) was the first to describe that hesperiids, or skippers, frequent the bird droppings at army ant swarm raids. Drummond and Boyce (1976) reported that other types of butterflies have the same habit. Ray and Andrews (1980) first used the term “antbutterflies”. Today, at least 239 species have been seen or collected at E. burchellii swarm raids (Suppl. Table 2).

Based on the observation that skippers, which are especially difficult to sample, are attracted to bird droppings at ant swarms, the Ahrenholz Technique was developed (Lamas et al., 1993). This technique consists of placing small pieces of white tissue paper or “bird dropping mimics”, wetted with saliva, on the forest floor. Such lures are known to attract butterflies which will drink from them. Many of the recorded butterflies may be opportunistic feeders at the lures or ant-associated bird droppings. However, a first study investigating the effect of the ants’ presence on butterfly recruitment to lures showed that some of the species accumulate significantly faster at lures if they are placed in the presence of E. burchellii (DeVries et al., 2009). This indicates that at least some hesperiids respond to cues that are not present in the lures alone and are thus true associates. Such additional cues attracting the butterflies may be (1) the sight of the birds, (2) the sound of the birds, (3) the sight of the (numerous) bird droppings, or (4) the odor of the ants. As far as we know, no experiments have been done to distinguish between these alternatives.


The Diapriidae is a diverse family of tiny wasps (2–3 mm). Species are mainly endoparasitoids of fly larvae and pupae, with some species known to attack ants and other insects (Masner and García, 2002). When we did our main collecting over swarms, we did not yet know that such tiny wasps were found with Ecitoninae. Once we became aware of them, we regularly observed these tiny wasps over swarm raids, but did not collect them, partly because they were too small to be caught by our insect nets. Today, 17 diapriid genera have been found with Eciton army ants (Suppl. Table 3). That this co-occurrence of diapriids and army ants is not only a coincidence but also an indication of a true association at least in some cases is suggested by the following observation: All neotropical genera of Diapriinae associated with army ants are conspicuously missing from the Antilles region (Lesser and Greater Antilles minus Trinidad), a region also devoid of army ants (Masner and García, 2002). We predict that the deliberate collection of these tiny wasps over E. burchellii swarm raids and from its colonies is likely to increase the number of known associated species.

We regularly observed and even filmed (Rettenmeyer, 2009) spider wasps of the family Pompilidae at swarm raids. Although we have not seen the “tarantula hawks”, the largest species of this family, the smaller pompilids are fairly common. The observed wasps could be as small as 1 cm. We propose that some species of these wasps may have evolved to find their spider hosts by following army ant swarm raids. Wasps from the families Scelionidae and Proctotrupidae, consisting of mainly parasitic species, are also present at swarm raids. The reason for their presence remains to be investigated.


The large and certainly charismatic birds following swarm raids have received much attention for over 100 years. However, what birders and other researches often fail to notice are the thousands of flies that regularly accompany a single swarm raid. The number and diversity of flies differ slightly between the swarm raids of E. burchellii and those of L. praedator (Rettenmeyer, 1961a). We assume that these differences are based on different raiding behavior and/or prey preferences of the two ant species.

Eciton burchellii swarm raids are generally accompanied by flies of the families Calliphoridae (blowflies), Conopidae (thick-headed flies), Sarcophagidae (flesh flies), and Tachinidae (tachinid flies) (Suppl. Table 4). Also present at swarm raids, yet less conspicuous due to their small size, are Phoridae (scuttle flies), which are very common in the ants’ refuse deposits and bivouacs.

There has been much speculation but few data to show what the flies are doing over the swarm raids. Flies of the genus Stylogaster (Conopidae; Fig. 4) are commonly seen to dart at cockroaches and other prey pursued by flies of the genus Calodexia (Tachinidae). So far, not a single Stylogaster species has been reared. However, examining 1,802 Calodexia and 531 Androeuryops (Tachinidae) flies, we found 17 eggs inserted in different body parts of these two species, one egg per fly (Rettenmeyer, 1961a). The elongate, elliptical eggs of Stylogaster are easy to see with a dissecting microscope and are species specific in their morphology and sculpturing (Rettenmeyer, 1961a). Harpoon-like, the eggs have a sharp point and backward pointing barbs that help to hold them in the host. Female Stylogaster have an exceedingly long abdomen that looks like it might serve as a blowgun. We have not been able to see if the eggs are shot out as from a blowgun or whether the fly must come in contact with its host. Both male and female Stylogaster are common over swarm raids. Here, these small and dark colored flies alternate between hovering and zipping around, which makes them exceedingly difficult to watch. From our own data and the few reports in the literature (see Rettenmeyer, 1961a), we assume that the most likely hosts for Stylogaster are cockroaches and possibly other arthropods with some species being possible parasites of Tachinidae. It should be pointed out that the range of Stylogaster extends far beyond the range of army ants, i.e. north as far as Massachusetts. In the United States, Stylogaster is usually collected as it feeds on flowers. This would give the opportunity to collect these flies and to investigate their host range and egg-laying behavior.
Fig. 4

Flies of the genus Stylogaster are abundant at E. burchellii swarm raids where they likely parasitize fleeing cockroaches and possibly other arthropods. Photo: Daniel Kronauer

There are at least 28 species of Sarcophagidae (flesh flies) associated with E. burchellii (Suppl. Table 4). We found four new genera and 18 new species during our research on Barro Colorado Island, Panama (Dodge, 1968), and we assume that collections from other areas will increase the number of new species. Most of the observed flies were definitely attracted to the swarm raids, for they were concentrated over the ants in numbers never encountered in general sweeping. Moreover, many of these flies have so far only been collected over this ant’s swarm raids.

We regularly observed flesh flies landing on small vertebrates killed by E. burchellii, which the ants abandoned due to their inability to tear the killed animal apart. We were able to rear Notochaeta amphibiae and Notochaeta bufonivora from frogs which had been killed by E. burchellii. The high species diversity among the sarcophagids we found at Barro Colorado Island (Dodge, 1968) may suggest that some of these flesh flies have evolved to follow swarm raids to find the dead vertebrates the ants kill but do not eat. We assume that the sarcophagids larviposit on a variety of animals, including large arthropods the ants may not be able to carry off.

Flies of the genus Calodexia (Tachinidae) are generally the most abundant and most conspicuous flies at E. burchellii swarm raids. Curran (1934) lists 23 Calodexia species from Central and South America. He states that the flies were collected over swarm raids of army ants, without giving the ant species. When collecting over E. burchellii swarm raids at Barro Colorado Island, we regularly sampled 12 of the 13 species reported by Curran (1934) for this site. We thus propose that most of the other flies listed by Curran are also associates of E. burchellii (Suppl. Table 4).

Calodexia are rarely found in the vicinity of the ant’s bivouacs but may appear within minutes once the ants start to raid. Often numbering in the hundreds, Calodexia can be found perched on the foliage and leaf litter in front of the advancing ants. The flies constantly shift their position, avoiding contact with the raiding ants. Calodexia found at the swarm front are almost all females (we collected 1,783 females but only five males on Barro Colorado Island; Rettenmeyer, 1961b). The flies dart at cockroaches and crickets fleeing the ants while ignoring most other arthropods. We observed and were able to film one instance where three flies attacked a large ctenid spider (Rettenmeyer, 2009). Unfortunately, the flies and the spider both got away. Collecting cockroaches and crickets the flies had darted at, we were able to observe the development and pupation (and in some cases eclosion) of more than 30 Calodexia larvae from at least four species (Rettenmeyer, 1961b; Fig. 5). Calodexia does not lay eggs but deposits active larvae that penetrate the host. The first instar larvae are species specific in their complex pattern of hooks that facilitate penetration of the host (Rettenmeyer, 1961b). Our data indicate that the different Calodexia species are all larviparous and internal parasites of Blattaria and Orthoptera. We estimate that of the cockroaches and crickets escaping the ants, 50–90% are subsequently parasitized by the flies accompanying the ants. In most cases, this parasitism is fatal to the host.
Fig. 5

a Flies of the genus Calodexia sit in the leaf litter at the ants’ swarm front and deposit live larvae on cockroaches and crickets fleeing from the ants. b This parasitized cricket from Trinidad died 1 day after collection, the fly larvae already visible behind its head. c After 3 days, the fly larvae had consumed most of the cricket and were clearly visible. d Four days after collecting the parasitized cricket at the swarm front, the flies pupated away from the cricket and eclosed another 10 days later. Unfortunately, these and other reared flies were lost before they could be identified. Photos: Stefanie Berghoff

The only other tachinid fly common over swarm raids, Androeuryops ecitonis (Beneway, 1961), is often missed due to its small size. We assume that the fly has a connection with E. burchellii since we have collected a total of 529 specimens over swarm raids and only two by sweeping where there were no ants. In stark contrast to Calodexia, 345 of the collected Androeuryops were males.

Associates found in refuse deposits

Refuse deposits, the garbage dumps of E. burchellii, are sometimes directly under a bivouac. More frequently, the refuse is deposited about 30 cm away from the bivouac edge. When bivouacs are inside logs, the refuse may be dropped off the end of the log’s opening. When bivouacs are high up in a tree, which is rare, the ants may carry their refuse more than 30 m to the ground.

As the ant larvae grow, they become less able to enter hard body parts such as legs which thus land with edible flesh intact on the refuse deposits. We recorded a maximum amount of refuse added in 1 day to the deposits as being about 100 ml. During the 3 weeks that the ant colony is statary, the refuse accumulates to a considerable pile. The smaller deposits of nomadic bivouacs are usually eaten or removed by scavengers within a few days after the departure of the ants. Refuse deposits have a distinct, somewhat fecal, odor which is different from the odor of the bivouac itself. It seems possible that rove beetles (Staphylinidae) and other insects are attracted to this odor.

Refuse deposits are teeming with life. A single refuse deposit can yield more than 1,000 arthropods, and we have more than 100,000 “refuse specimens” in our collection. Due to an unfortunate shortage of taxonomists, many of these arthropods remain unidentified. The main groups found in the refuse deposits are mites, springtails, staphylinid beetles, phorid flies, and a variety of insect larvae. Many of the occurring species are likely to be opportunists, taking advantage of the abundant food the ants supply. However, some arthropods seem to have a more intimate association with the army ants.

Mites are the most abundant arthropods both in number of species and individuals in all refuse samples. While many mites probably just happen upon the refuse deposits from the surrounding area, there is some evidence that mites living within the ant colony use the refuse deposits of the statary phase to feed and reproduce (Berghoff et al., 2009). Most mites are likely to feed on decaying booty or fungi, or to prey on other refuse inhabitants. There are large numbers of immature Uropodina mites in the refuse. However, none shows any obvious morphological modifications that would connect them with the Uropodina that are found in bivouacs. Unfortunately, the life cycle of none of the mites found with the ants or their refuse is known in any detail, rendering the matching of immature and adult stages exceedingly difficult. As the majority of mites remain unidentified in many cases, we can provide only the higher taxonomic groups found in the refuse deposits (Suppl. Table 5). The most common mites in the ants’ refuse are hypopi, the deutonymphs of Anoetidae, which are phoretic on the adult ants and many other adult insects running among the refuse (see Houck, 1991 for a general discussion of phoresy by mites).

Collembola (springtails) occur in high numbers of species and specimens in refuse deposits. Most belong to non-myrmecophilous genera also known from leaf litter. However, we recorded two species of the likely myrmecophilous Cyphoderus (Rettenmeyer, 1961b; Suppl. Tab. 6).

More Coleoptera than any other order of insects are found in the refuse deposits of E. burchellii. Most are staphylinid and histerid adults, and presumably their larvae, yet beetles from 12 families are regularly present in this habitat (Suppl. Table 9). We collected more than 7,000 staphylinids from E. burchellii refuse deposits at Barro Colorado Island alone, and thousands more could have been collected had there been any prospect of getting the specimens identified. Thus, most beetles remain undescribed, and no beetles have so far been reared from army ant refuse.

Muscoid fly larvae are common. We collected and reared Euryomma panamensis and E. rettenmeyeri from E. burchellii refuse deposits (Chillcott, 1958). These larvae were abundant in these refuse deposits, with one exceptional sample containing 439 Euryomma larvae in addition to 435 other dipterous larvae. We never found Euryomma larvae near dead mammals, rotting fruit or vegetation or excrement, indicating that these species may be restricted to the ants’ refuse deposits for breeding sites. In addition, we successfully reared 33 adult Neivamya sp. from a single refuse deposit. Phorid flies and their larvae are also very common in E. burchellii refuse deposits (Suppl. Table 4). We were able to rear Pulicimyia triangularis, two species of Megaselia and 78 Ecituncula species on dead E. burchellii ants or their refuse (Rettenmeyer and Akre, 1968). The majority of the flightless females hatched between days 15 and 22, so that eggs laid near the beginning of the statary phase of the ants can develop into adults before the colony emigrates (Rettenmeyer and Akre, 1968). The large-winged males eclosed on average a little after the females, an adaptation that would increase dispersal and cross-mating by forcing some males to fly to bivouacs of different colonies. Most phorids are scavengers that feed on booty refuse, but may also feed on booty and live brood within the bivouacs.

A wide variety of wasps can be found around refuse deposits of E. burchellii (Suppl. Table 3). Many of these wasps are known parasites of flies and other insects and their larvae. We suggest that some of these wasps recruit specifically to the refuse deposits of E. burchellii to find their hosts in abundance.

Associates found in bivouacs and ant columns

An inquiline is a species that lives in the colony or nest of another species. We consider any arthropods running in the raid or emigration columns and those found in bivouac samples of E. burchellii as inquilines.


Mites associated with E. burchellii are very diverse (Suppl. Table 5) and present in every colony. Investigating raiding worker ants in Panama, a median number of six mites per 150 investigated ants was found, resulting in an estimated average of 20,000 mites associated with each colony (Berghoff et al., 2009). Mites can be collected from bivouac samples of E. burchellii or by picking up single ants from raid or emigration columns, reducing the possibility of contamination with mites from soil or detritus to a minimum. We are therefore confident that the majority of collected mites are associates of the army ants.

Mites of the families Scutacaridae and Pygmephoridae are very abundant, riding below or between the coxae of worker ants (Berghoff et al., 2009; Rettenmeyer, 1962a). All collected mites were females and considered to be phoretic. Both families are also found in the refuse deposits. Decreasing numbers of phoretic mites during the ants’ statary phase indicate that the mites might dismount the ants to feed and reproduce in the refuse deposits during this time (Berghoff et al., 2009).

Of all the mites recorded, only a few have been shown to be ectoparasites of adult ants. One of these species is Rettenmeyerius carli (Ascidae) which attaches to a membrane at the base of the mandibles of the two largest worker castes (Elzinga, 1998). We observed the mite to swell as it ingests blood. The abundance of R. carli varies significantly between colonies, infesting most major workers of some colonies and apparently lacking in other colonies (Berghoff et al., 2009). Another parasite is Planodiscus, of which four species have been described from E. burchellii (Suppl. Table 5). All are found only on the underside of the hind or, rarely, on the middle tibia of the ants’ legs. When the mite first climbs up to its preferred position, the ant appears to be bothered. We assume that this is due to the mite inserting its mouthparts in the membrane between the ant’s femur and tibia. Once a mite is attached, the ants show no sign that they are aware of it, possibly due to the mites’ cuticular sculpturing and arrangement of setae, which closely match those of the ants’ legs (Kistner, 1979). The anterior of these mites is pointed with lateral dorsal groves that allow the ant to bend its leg (Elzinga, 1978, 1979).

Associated mites often show specific morphological adaptations for living with the army ants. Circocylliba, for example, have a highly convex dorsum which covers, and likely protects, the entire body and legs (Fig. 6). The mites’ concave ventral surface allows them to be closely appressed to their phoretic sites—the ant’s head, thorax, and gaster or to larval surfaces (Elzinga and Rettenmeyer, 1974). The two species of Larvamima, on the other hand, are mimics of E. burchellii larvae and follow the ants’ emigrations most likely by being carried just like the ant-larvae (Elzinga, 1993). Pinoglyphus has a peculiarly shaped adhering plate, yet is always detached from the ants in alcohol samples. In Trinidad, we observed one specimen attached with its sucker to the eye of a living worker ant. These are just a few examples of the diverse adaptations of mites associated with E. burchellii. The list is likely to grow if the hundreds of mites in our collection are identified.
Fig. 6

Mites are the most diverse group of E. burchellii associates and have developed numerous strategies to stay with the ants. Here, a Circocylliba mite sits tightly on the head of a worker ant. The inset shows an EM photograph of the mite’s ventral side. Photos: Eberhard Wurst


Most inquiline beetles belong to one of three families: Histeridae, Ptiliidae, and Staphylinidae. Beetles living within the bivouac often show morphological adaptations, such as having an ant-like form or matching the color of the E. burchellii subspecies they are found with (Kistner and Jacobson, 1990). The tear-drop-shaped body of most associated ptiliids effectively deflects an ant’s mandibles, enabling the beetles to ride out the ants’ occasional attacks without injury. Many of the staphylinid and histerid associates regularly groom the ants, possibly transferring the colony odor to themselves (Akre, 1968; Akre and Rettenmeyer, 1966). In laboratory nests, we observed inquiline beetles from all three families to feed on booty and sometimes on the ants’ larvae. Beetles found within the bivouac were generally able to follow fresh E. burchellii trails (Akre and Rettenmeyer, 1968; Akre and Torgerson, 1969) although many of these inquilines ride on the ants or their larvae during emigrations.

While histerid beetles mostly stay within the colony, some staphylinid beetles are regularly found in the ants’ raid columns (Suppl. Table 9). Once the researcher’s eye has adapted to differentiate between worker ants and the ant-like beetles, Ecitomorpha and Ecitophya specimens can be collected from most E. burchellii raid columns with some patience (Fig. 7). Some ant colonies apparently have high populations of these beetles, and we could collect up to 70 beetles within 1 h from some colonies (see also Reichensperger, 1925). Another staphylinid beetle regularly observed outside the ants’ bivouac is Tetradonia. Frequently, this most common beetle stay on the periphery of the colony, follows raid and emigration columns and is common in refuse deposits. Of all species we have observed, Tetradonia is the only regular predator of E. burchellii adults (Fig. 8). The beetles attack not only healthy but also injured E. burchellii workers and try to drag them away from the remaining ants; in laboratory nests the ants were dismembered, and the beetles fed on the ants’ body fluids (Akre and Rettenmeyer, 1966). Unlike many other inquiline beetles, which may still have the ability to fly but refuse to do so even when provoked, Tetradonia is often observed flying into and out of the ants’ refuse deposits.
Fig. 7

A beetle of the genus Ecitophya (Staphylinidae) running in an E. burchellii raid column (just below the major worker). The close resemblance of these beetles to their hosts makes their detection among the running ants a challenge. Photo: Daniel Kronauer
Fig. 8

Beetles of the genus Tetradonia (Staphylinidae) are the only regular predators of adult E. burchellii. Here, a beetle attacked a worker ant and tried to drag it away from the main ant column. Photo: Daniel Kronauer

Two Cephaloplectinae species (family Ptiliidae) are described from Eciton burchellii colonies. However, the subfamily is currently being revised for the first time in nearly 70 years (W.E. Hall, pers. comm.), and it would not come as a surprise if there are new associates and/or host records among the over 1,400 ptiliid beetles we collected from E. burchellii colonies in four countries (Suppl. Table 9). In some colonies, Cephaloplectinae can be observed in raid columns not too far away from the ants’ bivouac. Here, they often wait at the top of a small shelf in the ants’ trail, which forces returning ants carrying large booty to slow down. The beetles use this moment to grab hold of an ant or the carried prey and ride it back toward the bivouac. We assume that the beetles use their ride to feed on the booty. Other Cephaloplectinae are found in bivouac samples or riding on the ants and their larvae during emigrations (Suppl. Table 9). One pair of Cephaloplectinae was observed mating while riding on a callow E. burchellii male in an emigration column at the beginning of the nomadic phase (D. J. C. Kronauer, pers. comm.). Whether this mating site and time is the exception or the rule remains unknown—as is the place where the larvae develop.

The only beetle and in fact the only E. burchellii arthropod associate whose life cycle could be clarified in some detail is Vatesus clypeatus (Staphylinidae) (Akre and Torgerson, 1969): Adult V. clypeatus live within the bivouac and feed on booty and possibly the ants’ larvae. Their wing tips are broken off along a fold line, rendering them unable to fly. The eggs are probably laid at the beginning of the statary period, and larvae are found in the refuse deposits where they feed on booty or are predators of other refuse inhabitants. Vatesus larvae follow the ants’ emigrations during the first days of the nomadic phase. After a few days, the beetle larvae disappear from the ant columns and probably pupate in the leaf litter. The wings of the matured pupae are fully developed. Since the beetles can follow only fresh trails, the newly emerged adults are assumed to fly to a new host colony. We assume that this life cycle of V. clypeatus could also describe, with some adaptations, the life cycle of other E. burchellii associates. The validity of this prediction, however, remains to be shown.


Phoridae (scuttle flies), with at least 84 species and four unassigned males, is the most diverse family of regular E. burchellii associates (Suppl. Table 4). Almost half of the phorids, for which we have records of their collection sites, are found within the bivouacs or ant columns. Some of these phorids have sclerotized ovipositors and could be parasites of the army ants or other arthropods (R. H. L. Disney, pers. comm.). However, most of the myrmecophilous phorids are probably scavengers on booty and refuse (Borgmeier, 1928; Rettenmeyer and Akre, 1968). In Trinidad, Diocophora appretiata was observed at several occasions sitting on leaf litter along raid columns, watching the passing ants. The flies regularly darted at workers carrying prey (Disney and Berghoff, 2005), whether to snatch some food or to oviposit on the prey or the ants, we do not know. The majority of phorids regularly found with E. burchellii are females with vestigial wings. Males are less often seen but may be found flying into and out of the refuse (Rettenmeyer and Akre, 1968), where we assume many species reproduce during the ants’ statary phase. Phorids follow emigration columns, running in a typical zigzag darting manner. Their numbers can rise spectacularly toward the end of an emigration, when all but the last few ants have emigrated (Rettenmeyer and Akre, 1968 and numerous observations thereafter). The number of phorids per colony can vary from a few specimens to several thousands. Many species are still known only from the females, and we recommend any researchers interested in studying these flies to try to rear phorids (see Rettenmeyer and Akre, 1968 for a rearing method) to hopefully enable linking some more males to conspecific females.


Collembola were reported living in the bivouac of E. burchellii as early as 1919 (Beebe, 1919). More than 90 years later, it still remains to be shown what they are doing among the ants. In laboratory nests, we occasionally observed Collembola running among eggs and larvae, where they could possibly be feeding on substances on the surface of the larvae. From our observations, it seems doubtful whether the ants can catch and kill such small and active insects. The most common and likely myrmecophilous springtail we collected from E. burchellii colonies in Panama was a new species of Cyphoda. However, more than 500 springtails in our collection still await identification (Suppl. Table 6).


Two species of Thysanura (bristletails) are associated with E. burchellii (Suppl. Table 7). Rettenmeyer (1963b) summarizes the records and behavior of these species, from which we will point out the main traits: Trichatelura manni is the most common species found with E. burchellii and other army ant species (Fig. 9). Most T. manni were found in emigration columns or in nighttime raid columns. Both sexes as well as juveniles were present, with the smallest individuals mainly found on the first nomadic days. This suggests that T. manni may lay eggs during the statary period and possibly synchronizes its reproductive cycle with that of its host. This bristletail was observed to feed on fluids oozing from the ants’ booty and to scrape secretions or particles off larvae and adult ants. The single specimen of the second species found with E. burchellii, Grassiella rettenmeyeri, was running in a raid column just behind the advancing swarm front. The remaining nine specimens of this species were found with five different Ecitoninae species. In addition, three specimens were found in leaf litter away from any army ants. The lack of morphological adaptations, the wide host range, and the record of free-living specimens indicate that G. rettenmeyeri, unlike T. manni, is not closely associated with its host.
Fig. 9

The bristletail Trichatelura manni (Nicoletiidae) can generally be observed only at night when it runs in the host colonies’ raid or emigration columns. Photo: Daniel Kronauer


Millipedes are fairly common with the New World army ant genera Labidus,Nomamymex, and Neivamyrmex (Loomis, 1959; Rettenmeyer, 1962b). However, the single species found with E. burchellii, Prionodesmus fulgens, is a unique record and probably a detritus species not really associated with the ants (Suppl. Table 8).


The fauna associated with this one army ant species is incredibly diverse and abundant. However, the 557 associates reported here are likely only the tip of the iceberg. Thousands of specimens still remain to be taxonomically described and their association analyzed. The readily available methods of DNA sequencing may reveal new cryptic species, i.e. morphologically similar but distinct species, which have been classified as a single species (e.g. Bickford et al., 2007). In addition, the following consideration must be taken into account: Eciton burchellii has a geographical distribution from Mexico to southern Brazil. Within this vast area, there must be hundreds, perhaps thousands, of undiscovered associates of E. burchellii.

Throughout our analysis of our E. burchellii data and the literature, we noted records of animals associated with other army ant species. Reporting on these observations would go far beyond the scope of this paper. We can say, however, that the data and references indicate that Labidus praedator, which has much smaller and transitory swarm raids (Schneirla, 1971), has the second largest number of associates. Most associates of the remaining, and less well studied, Ecitoninae species probably still remain to be discovered.

Copies of the DVDs

“Associates of Eciton burchellii” and “Astonishing Army Ants” can be obtained by sending a check for $25 (per copy ordered) made out to the University of Connecticut Foundation (memo line—Carl and Marian Rettenmeyer Ant-Guest Endowment). Please mail the check to Business Services Supervisor, Ecology and Evolutionary Biology, U-3043, 75 N. Eagleville Road, Storrs, CT 06269, USA. Please indicate if you need European encoding (PAL).


We are grateful for the help of numerous field assistants and students aiding in the collection and sorting of army ants and their associates. Naming all of them would go beyond the scope of this paper. Our special thanks go to the following taxonomists for identifying and/or describing E. burchellii associates: Sidney Camras, Henry Disney, Ernst Ebermann, Richard Elzinga, David Kistner, Sandor Mahunka, Lubomir Masner, and Alexy Tishechkin. We would also like to thank Charlene and Adam Fuller for many hours spent in the field gathering information about E. burchellii and their associates and Adam for photographing and identifying antbirds. We thank Daniel Kronauer for comments on an earlier draft of this paper and for the permission to use some of his photographs as illustrations. Last but not least, we thank Phillip DeVries (Lepidoptera) and Henry Disney (Phoridae) for checking and adding to the respective tables. This work was supported by the British Ecological Society (SEPG 39/39), the Carl and Marian Rettenmeyer Ant-Guest Collection Endowment of the University of Connecticut, the Leverhulme Trust (F/00 182/AI), the Smithsonian Institution (SRA), the State Museum of Natural History Stuttgart, and the National Science Foundation. We thank the Asa Wright Nature Centre (Trinidad), the Estación Biologica de Monteverde (Costa Rica), the Organization for Tropical Studies (OTS) (Costa Rica), the Smithsonian Tropical Research Institute (STRI) (Panama), and the Summer Institute of Linguistics (Ecuador) for providing facilities and arranging collection and export permits. This work conforms to the legal requirements of the countries in which samples were collected.

Supplementary material

40_2010_128_MOESM1_ESM.pdf (160 kb)
Supplementary material (PDF 160 kb)

Copyright information

© International Union for the Study of Social Insects (IUSSI) 2010