OECOPHYLLA LONGINODA , AN ANT PREDATOR OF ANOMMA DRIVER ANTS ( HYMENOPTERA FORMICIDAE )

Ants o the subgenus Anomma, one ot 6 subgenera composing the genus Dorylus, are commonly reerred to as "driver ants." The driver ant receives its name appa.rently because it "drives everything before it capable o muscular movement, so ormidable is it rom its numbers and bite ." (Savage, 847). O the Dorylus subgenera, Anomma is most conspicuous. Its species are aggressive, primarily epigaeic oragers (i.e. surface adapted as opposed to subterranean) with colo.nies consisting ot up to 22 million individuals (Raignier and van Boven, 955). Although the erocity ot the Anomma driver ants has been fictionalized and tales o their behavior are incorporated into Arican olklo.re, their pugnacity has been repeatedly documented (Loveridge, 922 Raignier and van Boven, 9.55 Savage, 847, 849; Wheeler 9o, 922). The driver ants have also been described as clearly dominant animals (Weber, 943). Without a doubt, they appea.r as efficient predato,rs, having evolved a system o group predation o significant adaptive advantage in tropical environments. However, these carnivores are themselves no.t without predators. Bequaert (922) reports Anomma species rom the stomachs ot 4 species o.x Arican toa.ds (genus Bufo) and 3 species, ot Arican rogs (in the genera Rana, Kassina, Hemisus). He also reports that driver ants are commonly eaten by Arican skinks o the genus Mabuya and by pangolins (scaly anteaters) o the genus Manis. Chapin (I932) indicates that driver ants are eaten by several species ox Arican birds, including the Guinea owl, Phasidus niger Cassin. My own observations in Ghana place the domestic chicken among Anomma predators, or driver ants requently orage in village reuse heaps where chickens are also regular visitors. These chickens walk along the columns and clusters o. worker ants, picking up and swallowing individual workers while carefully avoiding any other contact with the ants. Insects and other arthropods have not been observed as important Anomma predators. While flies o the genera

Bengalia (Calliphoridae) and Stylogaster (Conopidae) are recorded as A nomma predators (Cohic,948), the term predator is applicable only in the case of Bengalia.Bengalia flies fly about and land near Anomma columns and eventually dart toward individual workers, stealing their prey or brood (Bequaert, I922).Stylogaster, on the other hand, is parasitic on insects, particularly cockroaches, that flee before Anomma foraging swarms, and on tachinid and muscid flies that are attracted to such sw.arms(Smith, 967).Stylogaster is also abundant over the swarms o.f New World dorylines, and para- sitizes cockroache's and tachinid flies (Rettenmeyer, 96 ).The eggs of these conopids are inserted into the host cuticle, and the larvae are internal parasites.
During May and June o 97, while collecting driver ants in Ghana., I observed on 5 separate occasions the ant Oecophylla long- inoda (Latreille) (subfamily F0rmicinae) successfully attacking .4nommaworkers.It becam e apparent that O. longinoda is a com- mon predator of driver ants and that it may be one of the very t:ew insect predators of the subgenus.
Two other species of ants were seen to.attack A nomma, but each case involved unusual circumstances.In the first case, workers of the ponerine ant Paltothyreus tarsatus (Fa.bricius) carried off the larvae and pupae of D. (Anomma) nigricans Illiger that had been discarded while the nest of the .4nommacolony 'was being excavated.
In the second case, a small species of Crematogaster dragged away Anomma workers that had been injured by an automobile that passed over an Anomma column.(Note: driver ants move in columns, usually along well marked trails, during f.oraging and during emi- gration or nest relocation).Cohic (948) also records an encounter between Crematogaster and Anonma.
O. longinoda, commonly re'erred to as the red tree ant, is con- centrated in several areas in the Ethiopian Region and is considered to be an efficient preda.torwith a painful bite (Vanderplank, 96o).This ant is dimorphic, with its maxima caste performing the o.raging tasks (Weber, 949).Altho.ughO. longinoda previously has been recorded as a predator of Anomma by Co.hie (948), Ledoux (95o), Vanderplank (96o), Way (x954), and Weber (x949), the extent of this predation was, in most cases, not indicated.Note: the major weapons in predatory attack for both Oecophylla and Anomma are the mandibles.O. longinoda, a ormicine ant, is sting- less, while A nomma workers possess what appears to be a function- less sting.
Of the 5 observations of Oecophylla predato,ry behavior, 2 oc- 35o Psyche [December curred in the coastal scrub and grassland region at Lego.n and 3 in the moist semi-deciduous orest region (: at Kade .and at Taro), and each involved predation on the species Anomma nigricans.Al- though all African populatio,ns o, OecoDhylla are considered to be one species, t'he observations at Kade involved the commonly occur- ring dark orm.

DISCUSSION
O. longinoda attacks both emigrating and oraging ,columns o d nomma, but attacks the columns at a limited number o selected points.In some cases the A nomma trails over which the columns move are marked by soil particle walls, while in other cases, the columns march along completely exposed trail.sor trails covered in part by grasses and organic debris.Some trails are subterranean or covered with soil particle ceilings built by the d nomna workers.In the observations reported here, the A nomma trails either had soil particle valls or were pal"tially covered by organic debris (primarily leaves).
Single Anomma workers are removed from the. column by individ- ual oraging O. longinoda workers.The O. longinoda workers are either positioned outside of the soil particle walls or on leaves and twigs adjacent to or overhanging the A nomma column.An O. longinoda worker quickly thrusts its head into the column, grasps an A nomma worker in its mandibles and pulls it rom the moving col- umn.In one case (Tafo) the O. longinod.aworkers removed A nom- ma workers rom a diffuse column expansion.The Anomma workers are rarely alarmed by the removal of a single sister worker, although in one instance, the O. [onginoda attack was so widespread that the ztnomma column was subsequently disrupted.O. longinoda workers are obviously afforded some protection by the trail wall or by the space between the column and their position on the surrounding vegetation (Fig. I).d nomma workers are generally excitable and easily disturbed, and thus it is surprising that they are usually so little disturbed during .anOecol)hylla attack.Way (I954) notes that O. longinoda grabbed A. nigricans workers t:rom the column, but adds that the A nomma workers had "strayed rom the main stream."While it is true that the A nomma workers selected by O. longinoda o.ragers are peripherally located in the co.1- umn, they are not usually "strays."Way (I954) also repo.rts that A nomma soldiers were never taken, but I observed on several occa- sions soldiers being attacked and immobilized.Although most A nomma workers attacked were not o the soldier caste, this may Psyche [December be merely a reflection o the polymorphic proportions ot the colony population.
While the removal o Anomma workers rom the column is per- ormed by individual O. longinoda workers, the immobilization o prey is the result o group action.O. longinoda workers surround the initial oraging worker and its Anomma captive and bite and grasp the A nomma worker extremities, and at the same time begin pulling (Fig .).Additional O. longinoda workers are recruited to this task, perhaps by the secretion o alarm substances.The mandibles, legs, antennae and parts o the alitrunk, petiole, and gaster are grasped (Fig. I).Anywhere rom 5 to 20 O. longinoda workers arrange themselves around the Anomma worker.In all cases, regardless o Anomma worker size, ca.ptured workers were incapable o mounting an adequate defense and were unable to bite their captors (i.e. they are unable to reach the O. longinoda worker because oi: their position in its mandibles).
The Anomma worker is immobilized through prolonged stretch- ing, and this method o. prey immobilization is co,mmon in Oecophyl- la (Gressitt, 956; Ledoux, 1950; Vanderplank, 960 Way, 1954;   Weber, 949).The stretching o prey, spread-eagle ashion, may be common to predaceous ants whose mandibles, perhaps in combina- tion with their stings, are not efficacious, at least at the individual level, in quickly killing or immobilizing prey.For instance, New World doryline swarm raiders, such as Eciton burchelli (Westwood) and Labidus coecus (Latreille) stretch their large prey until it is torn to pieces (Schneirla,97).In this case, not only does the stretching serve to kill the prey but to.subdivide it as well, thus i:acilitating the transpo.rto the prey back to.the bivouac.The O. longinoda mandibles do not appear to pierce the integument o the Anomma workers, although Way (954.)notes that the s,ott cuticle o limb and abdominal joints ot prey is sometimes breached.He turther reports that the O. longinoda workers sometimes curl their gasters dorsally and spray "poison" on these wounds.
Ater the Anomma worker is immobilized, it is usually carried by one or two Oecophylla workers to their arboreal nest.I the Anomma worker is not completely immobilized, more than 2 workers may cooperate in carrying it to the nest.In one ob,servation, the A nomtna worker was pulled at and stretched or 20 minutes, re- suiting in ahnost complete immobilization.Four O. longinoda workers then proceeded to carry the worker up a tree trunk, grasping the A nomma mandible, an antenna and each the metathoracic legs.The Anomma worker occasionally moved its legs., catching its tarsal .claws on lichens and mosses growing on the bark.Eventually 2 workers managed to carry it up into the tree.Way (I954) timed the stretching o a honeybee by O. longinoda workers and ound it also lasted 20 minutes.
The predatory attack o O. longinoda on Anomma driver ants is thus divisible into 3 phases: (I) the attack phase, in which an individual Anomma worker is seized and removed rom the column; (2) the immobilization phase, in which additional O. longinoda workers are recruited to stretch the A nomma worker until it is no longer able to move; and (3) the transfer phase, in which the im- mobilized Anomma worker is carried to the Oecophylla nest.
Although in all observations the Oecophlla workers carried off many A nomma workers, the number may not be signitcant with re- spect to the total number o A nomna workers comprising a colo.ny.Way (954) notes, in his observations o Oecophlla attacks on A. nigricans, that in one case, in a period o an hour, 348 Anomma were carried into one tree and 252 into another.
The importance o driver ants in the diet o O. longinoda can- not be calculated at this point.It is known that insect prey and sugars, are essential or the survival and reproduction o an O. long- inoda colony (Vanderplank, 96o).The role o Anomma as a oo.d source may depend, to a large extent, on chance movements o, Anom- ma columns into Oecophylla colony territories.Obviously the prob- ability o such an occurrence depends, in part, on the colony densities o prey and predator in a given area.The efficiency o oraging behavior in A nomma may in itse'l increase the probability o chance encounters with Oecophylla.Numerous oraging columns, or in- stance, may simultaneously work out rom an Anomma nest, may requently change direction, and may start anew e.ach day, thus en- abling the colony, in its search or nutrient sources, to probe the environment in many directions.
Are the roles o predator and prey in the Oecophylla-Anomma in- teraction ever rever,sed?Evidence or such a reversal might be ound in the prey materials carried by A nomma workers back to the are less likely to do so than they are to forage in lower vegetation.Cohic (948) reports that while he observed D. (dnomma) nigri- cans foraging in trees to heights of about o meters, the nests of Crematogaster and Oecohylla were spared or avoided.SUMMARY I. O. longinoda is an important predator o nomma driver ants and may be the single most important insect predator o A nom- ma in areas where the two are sympatric.
2. In attacking Anomma, an O. longinoda wo,rker reaches into a moving Anomma column (foraging or emigrating) and s.eizes an individual Anomma worker in its mandibles., pulling it quickly xCrom the column.This is termed the attack phase o the preda- tory interaction.
3. After the Anomma worker is remo,ved from the column, the ini- tial O. longinoda forager is joined by sister workers, which to- gether immobilize the prey through prolonged stretching.This is termed the immobilization phase.
4. When the Anomma worker is suciently immobilized, it is car- ried back to the Oecophylla nest by or more workers.This is termed the transfer phase.
5. Individual Anomma workers are unable to successfully defend themselves, and little or no alarm is generated in the column when the worker is removed.
6.All Anomma workers, regardless of size, including soldiers, are subject to O. longinoda attack.
7. Immobilization of prey through prolonged stretching may be employed by predaceous ants whose ma.ndibles and/or stings are ineffective in subduing prey at the individual level.
8. O. longinoda is an effective predator of Anomma beca.use,(a) it usually attacks the d nomma column at a limited number of points and generally avoids disrupting d nomma movement; (b) it seizes and removes Anomma workers from the column quickly, thus avoiding a widespread alarm response; (c) it at- tacks from defensively advantageous positions outside of the col- umn, reducing its vulnerability to .4nommaalarm response.9.In areas densely populated by O. longinoda (where the probabil- ity of chance encounter is high), an Anomma colony may lose several hundreds, theoretically thousands, of workers per day to Oecophylla predation.
nest.I made extensive surveys o prey taken rom Anomma .oragingcolumns in areas where O. longinoda is also o.und, and in only one sample was there such evidence.It contained the gaster o an adult O. longinoda wo.rker.It appears that O. longinoda is not oten attacked by A nomma, though it i.s not likely that Oecophylla could effectively resist an A nomma oraging swarm.However, it is probable that the arboreal nests o O. longinoda are simply out o reach o A nomma oragers.My own observations indicate that while Anomma does forage in trees at heights greater than o feet, they