Early on my first evening at Gashaka-Gumti, after the long day’s search for driver ants, I collapsed on the hard earth outside my room at the field station and contemplated the parrots flying overhead. But then I became aware of movements in the grass, and I turned my head to witness a remarkable sight: a row of handsome, 2-centimeter-long cylindrical Pachycondyla analis workers, right next to my face.
Scientists studying Pachycondyla have determined that raids of species like the one at Gashaka don’t proceed like those of driver ants and other army ants. Rather, they are led by an individual that has scouted a promising target population of termite prey and recruited a couple hundred workers to harvest it. I was witnessing that now. Traveling in a compact squad 2 meters long and two to four ants across, the workers in front of me marched at a steady pace of 1 meter per minute, following the leader to dinner. There were no stragglers. Compared to the swarm raids of the army ants, this raid seemed leisurely and orderly in the extreme—another example of how ants in smaller societies move slowly and with care.
Eventually the ants entered the brush, where I couldn’t follow. Circling the field station, I saw more columns sallying forth, one from each of several nests that were apparently operating on a tight and synchronous schedule. Thirty minutes to an hour later, the ants reemerged from the brush and headed home in identical formations. Only now, each one held several termites, stunned by the toxins from the ant’s stinger, bundled between her jaws.
Two nights later, a feeble column from the Dorylus rubellus driver ant colony near the kitchen, retracing the route taken by the raid of two days before, passed next to one of the Pachycondyla nests. Some of the rubellus ants stopped in the open-jaw guard position, preventing the much larger Pachycondyla from departing on their raid. Every minute, one or two of the besieged Pachys (as ant experts call them) stuck her head out of the entrance to the nest and jabbed at the tormentors. Occasionally a Pachy succeeded in grabbing a driver ant and pulling it below, where I could just make out the workers tearing it apart.
Every fifteen minutes or so, there was a surge of activity in the driver ant column, and rubellus workers poured down the Pachy hole—twenty, fifty, a hundred of them, a veritable blitzkrieg that must have prevented the Pachys from implementing a coherent defense. I envisioned the savagery below, the feckless Pachys massacred, their brood consumed. Eventually, though, the Pachys forced the driver ant horde to retreat. With a flashlight I made out, faintly, what I took to be the survivors in their bunkers, nursing their wounds.
One hour later, the column was succeeded by the full-bore swarm raid of driver ants, which swept through, swamping everything. The disoriented scramble that ensued reminded me that, compared to Pachycondyla’s methodical predation on termites, army ant raids seem based less on finesse than on brute force. I’m convinced driver ants have little success with prey smaller than themselves not because such prey isn’t worth the effort—since so few ants take food most of the time, this argument hardly seems viable—but because the sightless rampagers are individually clumsy.1 This again is what we expect with ants with a large colony size: faster movements and more inefficiency.
Even so, there were signs of order. As I watched the raiding ants struggle with a scarab beetle, others walked over or alongside them, delivering food to the nest. The frenzied workers on the beetle never seemed to confuse the movements of their prey with those of the dead insects being convoyed past them. Meanwhile, their food-hefting sisters managed to stick with their job and ignore the fighting, even though the ants embroiled with the beetle must have been discharging a powerful alarm pheromone. I couldn’t imagine how they kept it all straight.
I turned my attention back to the entrance where the driver ants had poured into the Pachy nest. By now the swarm had departed, leaving the hole deserted, in ominous stillness. There was not one Pachy to be seen.
But in the morning, I saw the Pachys marching out once again, with no dead of their species in evidence. Remarkably, they had survived the rubellus attack. In fact, the Pachys must have been picking off the driver ants one by one and (I imagine) eating them underground the whole time the multitudes were passing overhead. Instead of facing their demise, the Pachys had beaten the odds with a classic maneuver: by taking advantage of the choke point at the nest entrance, they had greatly decreased their adversary’s access and combat power. They had been in control of the situation the whole time, transforming the driver ant raid into the ultimate stay-at-home feast.
How could they take such a pounding? I found out by dropping a Pachy onto a driver ant trail. In one fell swoop, she was buried from view by a mass of driver ants. That was the sort of brutality I expected from an army ant! Convinced she was done for, I returned twenty minutes later to find driver ants still laying siege. I extracted the hapless Pachy with a pair of forceps and, shaking off all but two of her attackers, put her down for a look. She lay motionless yet looked intact. When I picked off the last two ants, she roused herself and ran off. She had been playing dead.
I surmised that the Pachys were too well armored to be killed—the driver ants’ mandibles slip right off their exoskeletons. Because army ant raids pass by within ten or twenty minutes, a victim need only stay immobile and wait for her assailants to give up. On several subsequent occasions, I saw a Pachy escape after driver ants had restrained her.
Fighting back is rarely an effective way to survive an army ant raid. No matter how many of its workers die while catching prey, the raid never appears to retreat; the attackers just keep piling on—an advantage to having a humongous army. Some beetles and millipedes avoid death by exuding noxious chemicals; driver ants respond by burying them with soil and abandoning them with no harm done. Spiders and praying mantises avoid capture by New World army ants by freezing in position; unable to detect prey except by their escape response, the ants leave them unscathed.
Many ant species have evolved other defenses to give their colonies a chance to survive—it helps, for example, to be built like a tank, like the Pachys. Other species climb grass stalks in a gambit to carry themselves out of reach, or barricade themselves in their nest chambers. The workers of New World Stenamma alas mold a tiny sphere of clay that they then use as a door for their nest entrance. A worker closes this portal upon detecting predators—especially army ants. This is reminiscent of a defense used by the ancient Cappadocians, who lived in what is now Turkey and carved stone discs that they would roll across the entryways to their underground dwelling places when an attack was imminent. Stenamma take it a step further, constructing false entrances to blind-ended tunnels that lead their foes astray.2
TERRORIZING TERMITES
The next night the mood at Gashaka was depressed. Caspar and I had been at the station only four days, but several of the scientists had been counting the months. The malaise that can descend on people isolated for too long in the field had worsened that day, when one researcher had a recurrence of malaria and retreated to bed. Meanwhile, everyone had been on the lookout for chimp feces for Darren, and the samples were piling up; Darren had spent the miserably hot day trying to sieve ant parts from a single turd. He had a few driver ant heads to show for it, but at this rate, his thesis would require several unappetizing years.3
Anxious to escape the conversational doldrums at the dinner table, I checked on the nearby driver ant colony. I was surprised to find meter after meter of workers carrying hundreds of ghostly bodies along their route, which I reckoned were brood being transported in a migration. Then I noticed that they were heading in the wrong direction, from the savanna toward the nest. A look through my macro lens revealed that the cargo was Macrotermes, termites known for their castlelike nests of clay.
Compared to the small incursions that Pachys make on termite galleries, this reflected a battle royal going on somewhere in the dark. Here at last was a show of the voraciousness for which army ants are celebrated. My spirit soared with the primal recognition of “nature, red in tooth and claw,” as Alfred, Lord Tennyson, described it. Was this the boom in the raiding “boom and bust” I’d been looking for?
Neither Caspar nor I had read about driver ants conducting an attack of this sort.4 These swarm raiders were thought not to invade termite mounds, on the theory that they were unable to penetrate within. Termite capture was believed to be the sole expertise of other kinds of African army ants with more subterranean habits. (The Mofu people of mountainous northern Cameroon hold these belowground termite hunters in great respect, calling them the “prince of the insects.” The villagers collect the workers in a calabash gourd and pour them out in their houses, then lay a trail of ochre on the ground; this is meant to guide the ants to the most termite-infested sites in their homes.)5
Termites, like ants, have a caste system that can include small workers and soldiers. Looking closely at the trail, I watched the ants hauling the termite workers, pale blobs about their own size. Once in a while, the corpse of a soldier termite would go by; it was also the size of a worker but had an elongated orange head and needle jaws. At rare intervals came the headless body or the rust-red, bodiless head of a second, larger kind of soldier. That night I roused myself every couple of hours for an update on the progress of the ants, checking out the action on the trail by flashlight beam. By morning, the booty of the ants included the rotund carcasses of the developing reproductives, the cockroach-shaped kings and queens (termites are in fact social roaches).6 This evidence told me that the ants had breeched the colony’s defenses and were now invading the nest proper.
Where was this rampaging taking place? Caspar followed the driver ant trail to where it disappeared underground near an eroded Macrotermes mound a couple of meters across. These termites have huge colonies and raise a type of fungus to help them break down their woody food. We dug a meter down into the nest, exposing dozens of chambers containing the termites’ soft gray fungus gardens, each about the size of a softball. Crawling in and out of pits and holes in each garden were both adult termites and the younger, more delicate nymphs. As we dug, Pachy ants, homing in on a good thing, raided the exposed gardens and soon had stacks of termites in their jaws. There were no driver ants to be seen: either the troops were attacking from a different direction or we had the wrong termite colony.
All day and through the next night, the driver ants continued to drag the termites to their nest. But where were they coming from? Scanning the surrounding landscape, Caspar and I eventually found two more columns of the same driver ant colony in the savanna. Judging by the differing proportions of the small worker, soldier, and queen termite castes being moved along each of the three highways, it seemed they represented three separate attacks, on either different termite colonies or different battlefronts within the same termite nest. In the end we estimated the ants hauled away at least half a million termites, large and small—several flaccid kilograms in all.
Driver ant workers investigating a termite fungus garden presented to them in an experiment in Gashaka, Nigeria. The same colony had attacked a large Macrotermes termite nest at this site.
An intriguing story, but how incomplete. In science, we learn by bits and pieces, leaving others to unravel further details. I could only guess at the scene that had unfolded somewhere beneath our feet. During the underground portion of their forays, the driver ants must have clashed with termite soldiers ganged along narrow access routes to their castle in much the way Pachy ant workers had shielded their nest entryway. But in this case the driver ants broke through and launched into wholesale looting of the corridors beyond, transforming a steady raid advance into a focused attack that continued for hours.
Since I hadn’t witnessed the original killing spree, I decided to reconstruct it. I took a fungus garden from the termite nest that Caspar and I had dug up and deposited it next to a file of driver ants. There was no response. Apparently, the ants did not perceive the garden as a source of prey. Their assault began only after I had crumbled the fragile material to expose the termite workers inside. Then the driver ants infiltrated every crevice and pulled out dozens of the buttery-soft termites.
During their second night of gathering termite booty, I could hardly get close to the trail, which had become completely walled over by a bristling envelope of guards. By knocking the guards away, I succeeded in getting a view of the ants below, many of which were now carrying termite corpses away from the ant nest. They were transporting their own brood as well, slung under their bodies in the same way they carried food. The raid trail had become a migration route, and it was clear why. Caspar and I held our noses against the stench arising from the nest: under our feet, the termites had begun to rot. The driver ants were abandoning ship, taking with them any salvageable meat and leaving the garbage-thieving acrobat ants to scavenge the decomposing bodies left behind.
Army ants, including driver ants, often migrate along a prior raiding route, while at the same time conducting a raid along another. Near the migration’s midway point, the queen makes her run, shielded by a retinue of workers. To see this happen requires round-the-clock diligence. I took advantage of this opportunity and lay down on my side at what I thought was a safe distance from the trail, with my headlight duct-taped to a nearby tree branch to cast a steady beam of light. Unfortunately, my days had been long ones. I recall noticing that among the migrating Dorylus rubellus ants were tiny workers I had never seen at any other time; presumably, they served as nurses. But then I fell into a dream about being a dwarf ant—only to be awakened soon after by the sound of my own scream, and the pain of rubellus ants embedding themselves in my arm.
Generally, the exodus of migrating driver ants continues round the clock for two to four days. The number of ants participating can reach into the millions. But this seemed to be a small colony, in the hundreds of thousands, and its migration was over by the second morning. Caspar and I located its new nest, 67 meters from the old site, and we pried up a rock for a view of the massive company beneath. The workers were piling termites in a larder 15 centimeters wide in a preexisting cavern.
This was the first time a driver ant colony had been found to stash its food. If their raids have been crafted by evolution to take advantage of the rare windfall, the ants should be masters at stockpiling an excess catch in this way.7 However, whereas the Pachy ants sting their prey to keep it incapacitated but alive, the driver ant kills it, and corpses rot fast in tropical heat. Driver ants and other army ants also lack repletes and don’t take the seeds that marauder ants can horde.8 Napoleon observed that an army travels on its stomach—anticipating the idea of a superorganism by seeing the ensemble as an individual—and the same is true of an ant army: unable to keep a fresh larder, this colony was forced by its stomach to stay on the move.
PREDATION VERSUS DEFENSE
Two weeks later, my arms were blotchy from bites. I had stared at ants so long, I saw their flowing columns even when I closed my eyes. I felt like an obsessed FBI profiler investigating the habits of a serial killer. By now, the daily activities of rubellus ants had fallen into a predictable pattern. Raids began early in the evening and continued into the morning hours. After a raid ended, the flow of returning workers on the trail could proceed for hours, even on into a second night.
From my first day at Gashaka watching the open-jawed guards, I’d documented how protective the driver ants were of the commerce on their trails. The rubellus reacted even more hysterically to my presence than did the marauder ants. When I so much as breathed on them, the food-transporting ants retreated, while the other workers scurried off to patrol up to 30 centimeters from the trail. I used my mask, constructed earlier from a disintegrating T-shirt, to keep from creating a ruckus.
I had thought that standing guard over a procession and patrolling near it after a commotion must be part of the colony’s defense, not part of its foraging behavior. And to some degree this must be true. Except for a few long-term trunk trails, all the trails of driver ants and other army ants are created during raids that have recently cleared the surroundings of food. For this reason, the numbers of guards or patrollers are out of proportion to the likelihood that those ants will find a meal; ipso facto, they will more likely serve to protect a column than locate prey.
This is unambiguously true for the majors of the New World Eciton army ants, which have fishhook-shaped mandibles suited only for suicidal defense against vertebrates—their jaws have to be pulled out of the skin with tweezers.9 Except for these specialist saber-bearers, which never catch prey, there’s no evidence that any army ants distinguish enemies from meals—a driver ant’s actions don’t differ whether she bites an entomologist or an aphid.
But no matter what the ant species, the line between defense and foraging can be blurry, because any concentration of ant workers has the capacity to serve as a snare for food. As an example, an insect might flee from the army ant raid front into the raid fan, where ants stationed as guards along the network of trails can participate in dispatching it. Dorylus rubellus on trails far from the raid reacted to grasshoppers, crickets, foreign ants, and striped mice in the same way they did to my clumsy presence: by patrolling and attempting to seize them. They caught two of the crickets and a carpenter ant, and in another case a grub that had caused no disturbance, and cut them up and carried them off as food.
The same thing happens during marauder ant patrols. In fact, for both driver and marauder ants, the workers on patrol appear to take on the movement patterns of those within a raid, absent an advancing front. The hypothesis I developed in Nigeria was that the only thing that stops a raid from developing after a disturbance is that the patrollers are soon drawn back to the overwhelming scent of the thoroughfare from which they came. But food can break the workers from this attraction: I scattered tiny bits of meat in front of the driver ants on patrol, and this was enough to set off a small raid from the side of the trail, as prey near trails often does with marauder ants.
The reaction of rubellus to disruptions along the trail is mild compared to their reaction to threats to the nest. Driver ants have a unique response to such disturbances, perhaps because, unlike in the barricaded constructions of big ant societies (weaver ants and leafcutter ants, for example), the populace often can be viewed from outside. One afternoon, Caspar and I tried the chimpanzee luncheon technique of jamming a stick in the midst of the ants visible within the wide nest hole of one colony. Workers poured out of the hole and began patrolling thickly within a meter of it. Others ran along the stick and cascaded in strings from the end. Within an hour the ants had closed the gap with a 25-centimeter-wide plug of their menacing bodies.
For this colony, our meddling resulted in an eviction. When I stopped by the next day, the ants were busily abandoning their nest, tracing a dense migratory route through the forest that shimmered with the gaping mandibles of the jet-black soldiers.10 I sat down at a safe distance and took out my notebook. At one moment a driver ant colony can be rushing headlong into battle with a termite army a million strong; at the next it might be fleeing from a chimpanzee with a stick or the breath of a man on its trail. Advance or retreat, eat or be eaten—these are choices even army ants have to make.
Workers of the driver ant Dorylus nigricans in Ghana transporting huge numbers of their colony’s pupae during emigration to a new nest site.
The resemblance between patrolling near a trail and swarming in raids set me thinking about how easy it is to make assumptions about the function of behavior, which can lead to misinterpretation. This seems to have been the case for the South American ant Allomerus decemarticulatus. Its colonies occupy shrubs that have hollow pouches at the base of their leaves, making for multiple living quarters. The workers also build shelters along the plant stems, using fine hairs spliced off the plant and bound together with fungi and feces. These thatched roofs, it is claimed, serve as traps.11 Reportedly, the workers reach through the gridwork of openings in the thatch to jointly ambush prey of a size and vigor normally caught only by army ants, pinning and dismantling grasshoppers against the platform as if it were a torture rack.
This notion of a “trap” implies that a grasshopper, for example, would avoid the ants if they were not hidden. This seems unlikely; I doubt grasshoppers could notice the minute ants of this species, particularly in mid-leap, let alone change course to avoid them.
During a research trip to Tiputini, Ecuador, I put the trap idea to the test. I hung a mosquito net over a plant with a thriving Allomerus colony, added a hundred grasshoppers and katydids, and sat inside for the next five mornings—an unusual case of using a mosquito net to keep insects in instead of out. Even after the grasshoppers settled down, they were indiscriminate in their movements, hopping from where ants hid under the structures to where ants strolled in full view to where there were no ants at all. When they landed among the ants, even on the structures, they got away unhurt. Certainly if the structures served as traps, they were inefficient ones.12
Colonies of Allomerus decemarticulatus build defensive covers over their trails. The workers are emerging from the circular entryways in these covers to catch an intruding Pheidole (“big-headed”) ant.
The Allomerus constructions run continuously from one nest pouch to the next on different branches of their shrub and contain a highway of workers commuting from nest to nest. Other plant-dwelling ants erect similar covers over their trails, even with similar holes through which they come and go to forage. Such arcades probably serve primarily to protect the enclosed traffic against enemies (out of sight being out of mind), as do the trail covers of soil built by marauder ants and many driver ants.
Indeed, the Allomerus workers at my study site didn’t wait in ambush hour after hour at each “foxhole,” as would be expected if the structures were sit-and-wait traps; when conditions were calm, most of the holes were usually empty. But that wasn’t true at times of danger to the ants on the passageway within. After a day of pulling grasshoppers from my hair, I noticed interlopers of another ant, a species of Pheidole, or big-headed ant, climbing the plant to pin down a wounded grasshopper missed by the Allomerus. Upon the arrival of the Pheidole ants, the Allomerus workers began to guard each of the several dozen entrances to their arcade nearest the commotion caused by the intruders. These guards, aided by nestmates roaming the arcade surface, also caught and killed one Pheidole and carried it off.
Ants of many kinds will on occasion catch and kill enemies and prey along their trails, especially when workers are densely packed; it’s a matter of overwhelming the quarry, as army ants do, through staggering numbers, a tactic that can succeed even for a timid species if their legions are great enough. In this way the organization of a superorganism can be more responsive than the tissues in a body: trail-bound workers can shift seamlessly in their behavior from transport to protection to predation. It’s as if one’s liver could change function when the heart is incapacitated, and pump blood.