It was January 2005, and I was in Africa again. I had already had many adventures in the sub-Saharan region. Years before, pursuing my love of frogs, I had hunted the goliath frog in Cameroon, hoping to set a new world’s record with a 3.3-kilogram specimen but settling for one that weighed a little less and was a meter long with its legs extended. In Gabon, I’d surveyed ants in the rainforest canopy, working on a canopy raft—a network of pontoons placed 40 meters high on the tree crowns. (My fellow researchers and I were so bothered by the dominant Crematogaster, or “acrobat” ants, that we nicknamed them “Crematobastards.”) In Ivory Coast and Senegal, I’d collected praying mantises, and in South Africa and Namibia I’d searched for spiders, flies, termites, and other curiosities. But this trip to Nigeria would be my first opportunity to investigate the similarities between the army ant and the marauder ant, my Pheidologeton diversus.
On the previous trips I’d stumbled upon Africa’s infamous Dorylus army ants and especially noted the swarm-raiding species commonly known as driver ants for their habit—shared with swarm-raiding army ants in the New World and the marauder ant in Asia—of herding their prey before them.1 I came to regret the stumbling. The Dorylus I encountered bit me fiercely when, distracted by the sight of a vervet monkey in the trees, I stepped on a packed file of the dark, shining workers. Their violent response reminded me of Ogden Nash’s character study in his poem “The Ant”:2
The ant has made himself illustrious
Through constant industry industrious.
So what?
Would you be calm and placid
If you were full of formic acid?
The pain of a driver ant bite has more to do with mechanical damage than poison, though. Each chomp forced me to stop and extract a worker from my skin or crush with a thumbnail those gnawing at my clothes. (Their heads stayed embedded in the fabric through several washings.) Bites on a fingertip were so agonizing that pulling the ant off wasn’t an option: when I gripped the offender between two fingers of the opposite hand, she would clamp down even more savagely on the delicate finger pad. In time I found a solution: I inserted the finger in my mouth and crushed the ant’s head between my teeth, which immediately disengaged her jaws. The ant was about the size of a Tic Tac breath mint and just as crisp.
Munching on the insect, I detected a hint of nuttiness and a trace of the pungent sourness of formic acid. Driver ants are not as oily as the plump marauder food-storage repletes. Nor are they as tangy as weaver ants, which have a mentholated lime flavor and are served in India as a condiment with curry dishes. Certainly they aren’t as desirable as honeypot ants, which Australian aborigines and southwestern American Indians find delectable.
Though they do not offer the candied delights of honeypots, driver ants are toothsome enough to chimpanzees to be one of their dietary mainstays, and that’s what had brought me to Nigeria. Caspar Schöning, who at that time was studying driver ants at Copenhagen University, had invited me to join him while he organized a research project on ant consumption with University College of London primatologist Volker Sommer. Joining us would be Darren Ellis, a student of Volker’s.
Just as different human cultures have developed different techniques and tools to kill prey, from spears to snares, the chimps across Africa have evolved distinctive traditions and methods to hunt ants.3 (Until recently, this kind of cultural diversity was thought to be unique to humans.) The chimps at Gashaka-Gumti National Park, where Caspar and I were going, use branches as harvesting tools, stripping off their leaves and inserting them into driver ant nests or, possibly, poking them at ant trails; the technique is known as ant dipping. Putting the stick in their mouths like a lollipop, they peel away the furious ants, which cling to each other in strings. Eating ants sounds painful, but driver ants don’t sting, and as I discovered, they don’t bite your tongue if you chew fast. The chimps’ use of sticks may have more to do with reaching deep into the nest in order to stir up as many ants as possible than with avoiding bites. In other places, chimps bravely reach into driver ant nests with their hands, which enables them to grab the delicious larvae and pupae.
But when driver ants are on the offensive, they’re fearsome hunters. Caspar, a perceptive, gentle German with the physique of a welterweight fighter and blond hair cut military-style, was an enthusiast of the ant’s prowess. The driver ant expert at Utica College, Bill Gotwald, had been told by a village chief in Ghana about a human baby that had been killed by the ants.4 I had heard that driver ants can even bring down a cow. Many African tribes believe that before a python feeds, it checks its surroundings for driver ants—which would flay the snake alive with their knife-blade mandibles if they found it too distended by a meal to escape.
As Caspar and I set out from Abuja for Gashaka-Gumti, on the border with Cameroon, we found ourselves in the midst of a desert storm. So much dust was being carried on Harmattan winds from the Sahara that our view on the two-day ride to the park was abysmal. The sun shone no more brightly than a full moon.
Jammed into a crowded jeep for the final step of our many-hour journey, we crossed a wide river just before reaching the field station inside the park. The water rushed up past the tires, but the driver plowed forward, knocking against rocks as the water continued to rise. A fellow passenger pointed to the spot where a student had gone for a swim the year before and been caught by a crocodile. His lifeless body was found under a rock, where the crocodile had stashed him for a later meal. Balanced precariously in the flatbed in the back, clinging to six other people, stacks of luggage, and tied bundles of squawking chickens, I concentrated on the happy fact that the air was clearer at this slightly higher elevation.
Once at the field station, a few low concrete buildings, Caspar and I dropped our baggage and headed up a footpath to start surveying the local ants. The undulating landscape was a mix of olive-green lowland forest and woodland savanna dominated by pale grasses. The earth was dust dry. Leaf litter crackled underfoot. On the way we passed handsome, colorfully dressed men and women of the Jibu and Chamba tribes. They gracefully balanced baskets of meat and fruit on their heads—trade goods that they would carry on the three-day walk to Cameroon. The air was split by the pea-yaow call of the putty-nosed guenon monkey.
Seeking moist ground, we left the path within an hour to search the taller gallery forests along a stream. Soon we were joined by chimpanzee expert Andy Fowler, a soft-spoken Englishman with a dependable wry smile, and Darren Ellis, a thin, bespectacled American, who began a dialogue with Caspar that continued nonstop for three weeks as they hammered out the protocols Darren would use for his master’s thesis. He was studying the stick tools made by chimps, the driver ant’s response to the tools, and the importance of these army ants to the chimp diet. This last topic required that Darren count the ant parts in any chimp dung he could find—a task that turned out to be even less pleasant than it sounded.
Andy led us to a driver ant nest he’d spotted two days earlier. From the base of a toppled-over kapok tree spilled pyramids of soil that extended into a streambed. We saw some old, weak, and wounded individuals withdrawing to the garbage heap to expire. Some were being captured by workers of a black-and-coffee-colored acrobat ant, who were waving their heart-shaped abdomens in excitement at the easy repast. Bagging vulnerable workers in the trash piles or the wounded left behind after an army ant raid is an industry for certain ant species.
This driver ant was Dorylus rubellus. As I watched their dead and dying, I thought of how my mentor, Edward O. Wilson, had figured out how ants recognize their deceased kin. In 1958, working at Harvard, Ed and two colleagues proffered ants squares of paper soaked with a series of foul chemicals associated with decay. One compound, oleic acid, yielded a full-blown necrophoric response, inciting workers to haul the paper to the trash pile. When the researchers daubed oleic acid on a live ant, her determined nestmates dragged her off as well. Until she licked herself clean, the unfortunate individual was repeatedly thrown back onto the midden with every attempt to enter the nest.5 Smell like the dead, and dead you must be.
Searching along the streambed, I found the driver ants’ trail, which was exposed for 3 meters before it climbed an embankment and disappeared into the forest. A few workers were carrying chunks of insects, probably collected in a raid deep in the forest, slung under their bodies in classic army ant style. What I saw next caused me to drop to my knees. The drivers looked and behaved so familiarly that it was easy for me to believe I was in Asia again watching a stream of marauder ants. Part of the reason was the posture of the guards, who stood or strolled near the trail in marauder ant fashion, high on their legs, with raised heads and open mandibles.
I soon spotted differences from the marauder, though. An inordinate number of medium and large driver ants, rather than the small workers, had taken guard roles. The traffic on the trail also seemed chaotic, seldom sorting itself into lanes. Contrary to some descriptions I had read, the ants ran side by side with their feet on the ground, not atop one another. Workers stepped on those next to them, however, and big ones strolled over small ones—common practice in polymorphic species, preventing traffic snarls behind workers that come to a sudden halt.
On portions of the route, the guards interlinked in a way I had never seen before: hooking their clawed feet together like some horror-film version of armored cheerleaders forming a pyramid, they welded a lattice over the traffic below them in a defensive shield bristling with jaws. Unlike marauder ants, driver ants have no eyes, and all their joint activities are truly examples of the blind leading the blind. Watching the guards link one leg to the next, I wondered how they sensed the hordes beneath them in order to correctly position their bodies. This shield tore whenever it was snagged by the food in transit below, which meant the basketwork of ants constantly had to be rebuilt.
Food also got snagged whenever the trail passed under leaves. Lacking the major-caste road crews of the marauders, the drivers could get past these obstructions only after relentless buffeting from the brute force passage of workers below, except at one moist site, where the ants had eschewed their bristling covers of live guards in favor of marauder-style earthen barricades.
I was lost in thoughts of road construction when two giant forest hogs, hirsute and high as my chest, ran into the streambed ahead of me. They gave me a look of evident horror, then dashed off noisily. An hour later, an impressively virile olive-colored baboon mock-charged, fangs bared, while his females walked behind him bleating nervously.
THE DRIVER ANT RAID
Back at the field station that night I saw my first driver ant raid. It was advancing at full force outside the dining room, where the savanna grasses had been chopped to create a lawn. While the primatologists remained on the porch drinking Star Beer, Caspar, Darren, and I got on our bellies to ant-watch. The raid extended from the nest of a Dorylus rubellus colony at the base of a tree 15 meters distant. That meant the ants were just getting started: driver ants surge ahead for 80 to 120 meters before retreating. Behind a front about 7 meters wide was a prodigious swarm 15 to 30 centimeters deep. Small workers rushed through to reach the front, where they slowed down to search for prey before retreating. Larger ants were mostly located in the whirlpool of activity farther back in the raid. Great numbers of them stood in guard posture all around the reticulating columns in the raid fan, while others were busy killing, dismembering, and carrying the kills.6
A swarm raid requires substantial ant power. The regiment has to be packed tightly because there’s no telling where the next kill will show up within the “net” of ants. Within this raid there were tens of thousands of ants, with two to five workers occupying each square centimeter of ground. They were so numerous that the sound of them rummaging in the litter or dropping from twigs was like the patter of rainfall. (Indeed, they did create a shower: driver ants scrounging for prey will climb up plants, but they save time by dropping to the ground instead of climbing back down.) In two hours, however, all this ruckus bagged the colony only two thimblefuls of invertebrate meat—ant-sized plant hoppers, centipedes, worms, and spiders.
Perhaps the raid was passing through an unfruitful stretch of land. Because driver ants and other army ants have been found to travel much farther during their raids than do marauder ants, I’d expected that they wouldn’t be as sensitive as marauders to local shifts in food abundance. Their unresponsiveness to food distribution would encourage army ants to continue through a barren part of the landscape until they finally found prey, or until their distance from their nest ensured diminishing returns. Such doggedness may be important when a search party is concentrated in a raid rather than spread widely, as it is in solitary-foraging ants, especially when food is scarce and scattered, as it is during Nigeria’s droughts.
Yet this seemed to be only part of the explanation, because there was prey around that the driver ants missed. The tens of thousands of ants we watched in those two hours took in less than several nearby packs of two hundred or so Pachycondyla ants that were on their evening excursions, recruited by scouts to catch termites. Grasshoppers, crickets, and Pachycondyla broke away from the trawling rubellus raid and survived, even when we increased the ants’ chances by tossing the escapees back into the swarm.
I thought about the slowness of marauder ant raids and the meticulous care the workers put into combing the raided area to extract food; generally, far more marauder ants than driver ants return laden with booty.7 Could that difference reflect the greater significance to the driver ant diet of tracking down, not lone invertebrate prey, but food that comes in widely separated bonanzas? This ant’s typical meal is found in prodigious stockpiles: brood plundered from ant nests. Indeed, mass foraging in army ants likely arose as a strategy to effectively surprise-attack other ant colonies. Army ants are thought to have begun to regularly kill large, nonsocial prey like spiders and centipedes only after their ant-plundering colonies had evolved to huge sizes and developed the capacity for wide swarm raids.8
The overkill population of tens of thousands within a raid, most doing nothing but walking around before leaving again, may represent a particularly huge reserve force, to be drawn on when the occasional megameal is encountered. If big ant nests are the mainstay of the driver ant diet, there’s a good chance that raids may go bust on some days.9 Yet swarm-raider army ants can prowl over 10,000 square meters in the course of a week. That’s equivalent to combing the length and the breadth of three football fields, an area that should contain plenty of ant colonies, large and small.
That night I sat under the brilliant Milky Way and made notes in my journal in front of the campfire. Finding windfall meals, I scribbled, requires that ants maintain the size of their raids even during periods when their take almost always ends up being small. What surprised me most about the driver ants was not the strength of the day’s raid but something else I noticed: throughout the raid, workers were constantly going home early and empty-handed, while equal numbers streamed out to replace them.
Theodore Schneirla, the dean of army ant research, concluded that army ants are inefficient. When a raid is at full steam, each worker’s transit between nest and raid can take an hour. Based on my records from that first raid and others I saw later on, I calculated that an aggregate of thirty hours’ work time was lost to the ant society every single second from all this walking to and fro by thousands of ant workers.
Why doesn’t each worker stay out in the raid until she has something to show for her efforts? Coming and going from the front lines, might workers spread among themselves the risks of the hunt? Or do workers tire out and plod home for food and rest as fresh troops stream from the nest to replace them? These notions were illogical. Rather than commute to the nest, workers could save an hour by doing their R&R within the raid itself. That’s where the food is, and indeed many workers within the raid do stand around. I have said they serve as guards, but they sometimes look more like office workers stealing a power nap at their desks in the middle of a grueling day.
Pausing in my writing to watch a burst of falling stars, I thought of another explanation: maybe the ants roaming the trails make contributions outside the raid, such as building arcades or guarding the route. However, there were times when all the ants ran between raid and nest without pause. I was confident that most ants returning from a raid hiked all the way home, accomplishing nothing along the way but cardio exercise.
Feeling the weight of dinner in my belly, I wondered if the ants heading back to the nest were transporting hidden booty.10 In the marauder ant, nest-bound workers often have abdomens bloated from drinking the syrup of overripe fruit.11 But this kind of “tanking up” would not account for the huge numbers of driver ants going home holding no visible reward in their jaws. Where would the food needed to fill so many bellies come from? Not fruit, ordinarily: though driver ants do eat certain native forest fruits, in general vegetable matter is a minor part of their diet. With so few ants hauling prey corpses, the only way the homebound ants could have full bellies would be if most of the prey were being consumed on the spot. Though raiding workers do lap tasty juices off worm and larva prey, I saw no evidence of such activity on a large scale. The ants haul most flesh back to the nest to eat, ingesting during the raid only those foods they cannot cut into pieces.
Then again, remembering how the basket-laden people walking from Nigeria to Cameroon paused to chat with their returning friends—perhaps to compare notes on the value of the goods they had sold—I wondered whether the ants were sharing information. Through signals I could not recognize, returning ants might inform the nest of a raid’s success in acquiring food or finding a new nest site. Fresh troops would then depart for the raid with updated knowledge of the colony’s current requirements. This feedback might draw more ants into a raid, or lead to its retreat or to the start of a migration. If information does flow between raid and nest, that might help explain evidence for day-to-day differences in the duration and distance of raids.12
For societies as large as an army ant’s, this hypothesis seemed reasonable. As I saw for the springtail-catching trapjaw ants, the care with which workers in small colonies conduct their business can reflect their limited operating budgets. A large ant colony almost invariably shows more frenzied activity and a faster tempo than a small one. Both ant and human societies can be more productive per capita as their size increases, despite all the mad rushing about: in large cities, people interact with numerous others, exchanging and creating ideas at a high rate.13 There’s a payoff for all their “type A” behavior. Workers in large ant colonies likewise glean information from the crowds around them. In the seed-harvester ant of the southwestern United States, for example, unemployed workers perceive how many of their compatriots are devoted to different tasks by the scent each passerby has picked up from her environment, which reveals the job she is performing—one of several known instances where ants show a capacity to accumulate evidence before making a decision. The workers then adjust their efforts accordingly, shifting, say, from nest maintenance to foraging when foragers are in short supply.14 Some individuals’ rough assessment of the labor situation may be mistaken, but a large colony can afford errors, and the “foraging for work” method enables ant societies to redistribute labor effectively without the need for a supervisor.
Among ants, who are acting without a leader, each individual responds based on the small amount of information available to her. But by gathering all those bits together, the superorganism as a whole behaves sensibly. The raid’s features emerge from the collective decisions of the incompletely informed masses, each ant contributing so infinitesimally as to be essentially irrelevant to the outcome. Indeed, the organization of a swarm raid has been accurately re-created by a computer just by modeling the ants in terms of a single, simple set of behavior rules.15
Under many circumstances this wisdom-of-the-crowd is characteristic of humans, too—a valuable feature of human democracies. The average of a large number of decisions, even by individuals who are poorly informed, often turns out to be surprisingly smart and accurate. The U.S. military, for example, located a lost submarine with scant information by averaging the guesses of a variety of experts as to its fate, even though no individual guess was close to correct. This has been put forward as a reason to avoid, in the way ants do, an overdependence on a few leaders or “experts,” whose judgment can be less reliable than that of a crowd.16
One result of the greater information flow in large societies is that larger colonies are more homeostatic than small ones, which is to say that they are more stable in their internal interactions and their relation with the physical environment, in much the way the health of a human body is maintained by a flow of information through our tissues generated by our endocrine glands and nervous system.17 Large ant societies tend to have a more dependable influx of food, for example, and their nests have internal temperatures better regulated to suit the varied conditions required by developing brood—as in the sun-exposed nest mounds of temperate ants.18 As I found out for the marauder ant, raids also appear tightly regulated and appropriately responsive to their environment.
While there may be something to this hypothesis for many aspects of colonial life, keeping such a vast labor pool in constant motion makes no sense as a way to run a business—or an ant raid. Perhaps the structure and momentum of a raid are somehow sustained by the manner in which the ants cycle between the nest and the raid front as an unbroken part of the superorganism. It may be that the incessant long hauls between raid and nest are a by-product of this dynamic, with which, in much the way humans have retained their (now useless) wisdom teeth, the ants have been unable to dispense.
RAIDING A NEST
Day after day, Caspar and I explored the terrain for driver ants, but at night the convenience of the dining room colony was irresistible. I’d eat yam tubers and ground cassava on the porch with my companions while gazing at the tree with the driver ant nest. With the monkeys asleep, the primatologists’ fieldwork was over for the day. Ours, however, had just begun. Each night after feeding and watering ourselves, Team Ant would scoot over to check out rubellus in action.
On my second night at the field station, the colony raided away from the dining room, into an expanse of savanna extending along the station’s border. This raid contained far more big workers, about the size of a marauder ant major but without their boxy heads, than the one the night before. I had no idea why. Could the ants adjust their work crews according to some labor need I couldn’t perceive?
Stepping into the 2-meter-tall grass, I worried about not seeing the bloodthirsty ants before they saw me. (Army ant studies invariably suffer from a variation of the Heisenberg Uncertainty Principle: an observer may influence a raid simply by agitating the subjects with the slightest disturbance. If only myrmecologists could concoct an antigravity device that allowed them to hover over a colony without touching anything, not to mention an oxygen tank to keep them from breathing on the ants. I’d order the deluxe unit, with a personal air conditioner, because the ants don’t appreciate sweat dripping onto them, either.) Fortunately, I soon found that I could watch these ants without their noticing because there were workers swiftly ascending grass stalks directly in my line of sight. With this early-warning system, I was able to stop walking and survey the ground before the more slowly progressing ants in the dense thicket there began to ambush my feet.
Then I noticed that the rubellus were driving carpenter ants, each one clutching a larva or a pupa, up to the tips of the stalks, where they froze. Because of the paltry space on a blade, as few as two or three driver ants were spread out behind the carpenter ants on a meter’s length of stem. It looked as if each driver ant was pursuing the carpenter ants on her own, beyond the raid front, perhaps (given the ants’ blindness) by using vibrations or chemical signals. As I scribbled in my notebook, a driver ant tackled the carpenter ant at the tip of one stalk. Both fell in a tangle into the maelstrom below, where the carpenter ant’s pupa was snatched away and her head torn off before she was buried in driver ants.
Tracking an individual driver ant in the welter of a swarm is well-nigh impossible, which made the action on the blades of grass a great opportunity for me. I watched intently as one carpenter ant after another met the same fate. (As one friend pointed out, it’s cruelly ironic that the most popular product for killing ants is called Raid. Talk about living by the sword, dying by the sword.)
In a desperate gambit, a carpenter ant retreats up a grass stalk with a pupa as driver ants raid the ground below in Gashaka, Nigeria.
Rooting in the earth, I located the carpenters’ nest entrance. Driver ants were emerging with carpenter brood, as survivors scurried to temporary safe havens with their charges. The larvae looted from the nest numbered in the hundreds. It wouldn’t matter how many ants escaped death, though. They were doomed unless the queen survived as well. The colony could be resurrected only with her continued supply of eggs.
Based on the number of driver ants pouring out of the nest, however, I was sure that she had been killed and her colony vanquished. Then I recalled my dad, who had worked on an ambulance in his youth, complaining that people made up to be dead in the movies were often shown with wounds that, though horrific, wouldn’t be fatal. The same was likely true of attacked ant colonies. In tropical American forests with high densities of army ants, every spot on the ground is raided on average at least once a day, so all the colonies have to be able to survive repeated attack.19 Despite the devastation, the fact is that army ants largely cull rather than eliminate their prey. This culling may make the army ant’s role in the ant community similar to that of a grazing mammal that crops grass just enough so that it grows back.
Do army ants practice sustainable harvesting, then? Not likely. I’ve never seen an army ant worker hold back on making a kill. After all, army ants don’t control territories, so it’s likely another army ant colony would reap the benefits of an earlier group’s restraint. I imagine their assaults simply cease when they reach a point of diminishing returns.
Whether the carpenter ant colony was defunct or wounded, the local population of ants would have plummeted, giving other ant species a chance to colonize the area during the following weeks. Over this same time, the driver ants’ other prey—the crickets, spiders, scorpions, nightcrawlers, snails, and varied kin whose numbers had been diminished through death or exodus—would crawl, hop, or slither back, repopulating the ransacked ground.
TRACKING PATCHES AND PREY
Before my departure to Nigeria, I had spent days rummaging through the Biosciences Library at the University of California, Berkeley, reading all I could about army ants, particularly African driver ants. I found that because of the time it takes a prey population to recover, it has been widely assumed that army ants avoid overhunting a site, perhaps warned off by the chemical trace of past raids. This idea was corroborated by the behavior of the most-studied army ant, the New World Eciton burchellii. This species has a highly mobile life organized around a huge, synchronized brood produced in a predictable pattern. Colonies migrate almost every day for a couple of weeks while feeding their hungry larvae, then settle in one spot for another two or three weeks until the clutch of tens of thousands of adult ants emerges. During this stationary period, the ants minimize the overlap between their forays to some extent by raiding in a starburst pattern around the nest, with each day’s raid separated from the tracks of previous days like the spokes of a Ferris wheel.20
Most army ants don’t lead such regimented lives. Their queens produce brood continuously, or at least in a less regular way, and their migrations and raids are less predictable and probably, in the case of the migrations, less frequent. Nevertheless, I expected that most army ants would avoid recently raided areas. I was therefore surprised, on my third night at the field station, to see a raid going on in the area near our dining room where I had documented my first raid by this colony only two nights before. As they had that last time, they dined on watermelon we had discarded, bulldozing their way into the fruit, an unusual treat for them. Eciton burchellii has been observed to raid night after night under electric lights that attract clouds of insects, demonstrating that this species can be motivated to deviate from its starburst raid pattern. Perhaps under certain circumstances it doesn’t pay to give up on a dependable food source.
Patch is the scientific term used to connote a local food supply that takes a long time for an animal to consume, and patchiness describes an uneven distribution of resources.21 Some patches are bonanzas, like a watermelon rind in one lump.22 Once the recruits arrive at such a large object, they need search but little, if at all, to harvest it completely. Other patches consist of a cluster of scattered pieces; once that type of patch is located, the ants must seek out the constituent items, such as the brood scattered within a raided ant nest. The two kinds of patches aren’t mutually exclusive. Our watermelon castoffs each represented a bonanza, but all of the pieces, scattered across the lawn, represented a patch, too. Some patches are restocked, as with the excess fruits that fall continually from a tree or are tossed daily from a dining room—these are patches that keep on giving. Ants finding a long-lasting food supply can save themselves a lot of foraging effort by returning to the site as long as visits pay off.
Shuttling individuals to resources in appropriate ways is a matter of having an appropriate transit system. For marauder ants, the grand pattern of food acquisition depends on the colony’s durable trunk trails, which are rare in swarm-raiding army ants. My time in Asia had shown me that when most of the land within range of a marauder ant colony is barren with the exception of a verdant field rich in seeds and bugs, the colony’s trunk trail will point to that field like an arrow.23 In addition to accessing patches of widely scattered food like this over weeks, branches of a trunk trail can lead to concentrated bonanzas on which the ants may feed uninterrupted for days. Because the food present in another colony’s jurisdiction is available only at the cost of war, the most stable routes also lead away from the trunk trails of other colonies. This configuration allows competing colonies to avoid each other even when their nests are close together.24
That third, moonless night in Gashaka, I set my alarm at two-hour intervals and groggily checked the dining room colony repeatedly to confirm that the ants spent until 5 A.M., well after the raid itself had ended, eating the watermelon down to its hard green casing. As a spotted eagle owl hooted in a nearby tree, I watched ants ply the same trails without pause until they devoured the feast, keeping the former raid route open, at least temporarily.
A driver ant emigration is being staged beneath this protective envelope of aggressive workers in Gabon.
Apparently, driver ants can gorge themselves at one place in much the way marauder ants do, albeit for a shorter interval. A colony Caspar studied in Kenya spent three days carving away at a colobus monkey corpse until it was reduced to fur and bone.
Driver ants also reuse abandoned trails that extend into patches rich with scattered food. I saw an example of this at the start of that third evening, when the first ants returned to the watermelon-strewn area near our kitchen. The initial sign of the impending raid was a trickle of workers crossing the lawn in a narrow file. I was puzzled, because this species should raid only in swarms, not columns. Then I realized that it was not a raid at all—instead the workers were retracing the path they had used for their raid two days earlier. The feeble vanguard turned out to be a harbinger of the swarm that arrived as a separate wave an hour later.25 Just as the chemical traces of an old route could (in theory, at least) repel future raids if the original swarm had left little to eat behind, another old route may attract raids to a locale that remains desirable over an extended period—it remains as a persistent track, or trunk trail.
While trunk trails are invaluable for accessing food patches, they have their uses no matter how food is distributed. It’s basic geometry. Marauder ant raids, for example, extend 20 meters at most from their starting point. If every raid departed from the nest, the colony would be able to hunt in a circle with only a 40-meter diameter, a region that could quickly be exhausted of food. But the trunk trails of most colonies are 30 to 60 meters long, vastly expanding the reach of their raids. In a sense, these thoroughfares are part of the marauder ant’s nest: sturdy and safe and, even when no foraging is going on, holding a reservoir of ants that comes in handy when a raid is launched.
Army ants also employ trunk trails to access distant regions, and they can do so at breakneck speed. An Eciton burchellii colony in its stationary phase may develop several routes radiating in different directions. Rather than departing the nest in their usual swarm, the ants run in a line along one of their abandoned trails, retracing the trunk trail’s path for 50 to 100 meters. Only then does foraging begin in earnest, as ants pour off the trail in raid formation to explore promising new terrain.
It isn’t known yet how army ant workers mark their more productive paths to make them attractive for future reuse as trunk trails. In general, detecting an old route, and whether to follow or avoid it, should be easy for army ants, whose capacity to pick up scent is legendary. Some New World species will follow nearly anything, even a thin streak of water laid on the ground. This versatility suggests that their skills could even extend beyond recognizing the trails of their own species to tracking the trails of their prey. Workers at the front lines of a raid likely take whatever cues they find to lead them to a meal, scents left by other ants included. Caspar and I were focused on swarm raiders, but around the world most army ants mount column raids, which illustrate this ability well. Narrow columns don’t have the ant power to take down big prey, which will escape them easily. Instead, most column-raiding army ants depend on finding ant nests, with their hoards of brood. Even a weak raid can stage an effective attack on an ant nest if the workers, recruited from the column network, are quickly able to accumulate in numbers.
There is a logistical problem, however: the nest entrance can be difficult for a raid column to find. Picture the column as an elongating line, and the nest entrance as a point on the ground; the odds of them intersecting aren’t good. But if the column raiders avail themselves of their scent-tracking skills, their raid need only cross the trail of the other ant species. That’s an easier proposition, because everywhere in a rainforest there is a tapestry of the pheromonal guidance signals deposited by all kinds of ants.
I have devised an experiment to determine whether army ants are tracking prey by following their scent. When I find a column raid in progress, I scrape away the ground surface ahead of their front line. If the ants are still in search mode, they will continue across the upturned ground without hesitation. But if they are in pursuit of a prey species of ant, my action will have removed the pheromone signals they are tracking and disrupt their advance. Twice on my visits to Barro Colorado Island, a research station operated by the Smithsonian in Panama, I stopped the column raids of an Eciton hamatum colony cold—until they picked up the scent again and continued on their way.26
Whether reused on occasion or not, army ant trunk trails are generally less ubiquitous, obvious, and persistent than those of the marauder ant. Army ants focus on discovering virgin hunting grounds through migrations and shifting raids, because their diets are dominated by ant colonies and large invertebrates that are slow to replenish. Because marauder ants regularly eat fruit, seeds, and small and large prey, they can generally gather food even in the frequently reraided areas near their nests. There may be limits to the utility of their trunk trails, though: typically after a few weeks, a colony abandons one highway and starts another. I haven’t been able to prove it, but it may be they need a change of venue once they have depleted a region.
Conversely, the predilection among army ants for meals of large invertebrates and social insects may explain why their raids travel ten times faster and several times farther than those of marauder ants. With their great breadth of diet, marauders can afford to be slow and methodical in their searches, whereas a sizable army ant swarm must do considerable reconnoitering to take in enough of their widely separated social insect prey.
I took off for an hour to look for ants along the path to Cameroon, unable to stop thinking about how an aggregate of army ants, using the interweave of their columns and scents, explores the world much as an organism does. My mind turned to the driver ants’ blitz of the carpenter ant colony. Two young men strolled into view, baskets filled with dried meat balanced on their heads. Judging from their stares and whispers, I must have been an odd sight indeed.
After years of living on foreign soil, I have developed a method to deal with this reaction, based on the assumption that, since I am the center of attention anyway, I might as well make things interesting. Whether faced by a crowd at a bus stop in Nepal, by kids leaving school in a mountain town in Bolivia, or by weary foot travelers in Nigeria, I walk back and forth like a professor in a class, carrying on out loud about whatever is on my mind.
“The swarm raid rolled over the carpenter ant nest as if it were nothing,” I said to the two astonished villagers. “Just a tiny minority of the driver ants were involved in the kills. What does that tell us?” I turned dramatically to the young men, who were grinning at my animated speech. They gave my hand a shake and continued on their way, chattering.
My brief soliloquy led me to the obvious conclusion. The driver ant raiding enterprises were superorganism-level organs geared for much larger confrontations than those I had witnessed so far. Two days later, this suspicion was confirmed.