Workers of Proatta butteli seizing a wasp at the Singapore Botanic Gardens.

      In Malayan rainforests to the north of Singapore I would later find a second swarm-raiding Pheidologeton species called silenus, closely related to P. diversus but with raids twice as fast, matching the raid speeds of a slow army ant.⁴ Sluggish or not, the painstaking searches don’t hinder the hunting prowess of these Pheidologeton species, particularly in the diversus marauder ant, which usually takes food in abundance.

      I came to believe that there was simply no need for the mass of marauder ants to move along any faster. In fact, in the Botanic Gardens I came upon a different kind of ant nesting at the base of a withered tree that taught me that mass foraging might not require the group to move at all.⁵ Lumpy beasts with unimpressive jaws, Proatta look incapable of doing anyone harm. Yet that day I saw three workers grab a wasp that must have outweighed them by a factor of fifty; trying to escape, it nearly lifted them all from the ground in an attempt to take flight. Nearby comrades, attracted by the commotion, seized the quarry by the hind legs. Then more nestmates, perhaps drawn to the site from a distance by pheromones, helped to pull the wasp into their nest.

      Groups of the same Proatta ants also killed marauder ants that lagged behind on the base trail after a raid. What accounted for their success? While much of their effort is spent scavenging by themselves for all kinds of tidbits, the Proatta workers accumulate in such numbers within inches of their nest entrances that when an insect walks by, several ants are often close enough to pin it down. Proatta essentially stay in place and let prey come to them, using a group version of the ambush tactic employed by a human duck hunter hidden in a blind, or by solitary-living species such as the snapping turtle, which lies in wait for fish to pass by.⁶

      As we have seen, proximity is not essential for ants to act as a group. But the Proatta behavior demonstrates that, as in the packed raids of the marauder and army ants, a high density of participants increases the likelihood that encountered prey will be caught. When their workers are close together, some ant species are even able to avoid active foraging almost entirely. Ants squeeze together inside their nests, and a Mexican Leptogenys species takes advantage of this density by giving their living quarters a scent that attracts the pill bugs on which they feed. The ants jointly kill and feast on these little crustaceans without leaving home.⁷

      But by keeping on the move in a crowded mass, the marauder and army ants accelerate the odds of encountering dinner, compared to these sit-and-wait strategists. It is the difference between dragging a net through water and leaving it fixed in place. Both methods work, but a motile net almost certainly catches more fish during a given period.

      ANATOMY OF A RAID

      One morning as I watched the foremost workers in a marauder raid nose their way forward through the grass, Paddy leapt up behind me with an insect net. Suddenly the 12-centimeterlong praying mantis he was chasing landed with a shudder within my swarm. Paddy backed away with a mild expletive as the ants overcame the mantis. Some grabbed the wings by their edges and spread them out to their full green glory, while others took its head between their jaws until it cracked open like a nut seized by pliers. Soon the marauders were slicing and dicing the mantis with the cold efficiency of slaughterhouse employees.

      Army ants, and particularly swarm-raiding army ants, are exceptional for their ability to consistently trap difficult, even dangerous, prey. I now saw that this attribute applied to swarms of marauders as well. To uncover the secrets of the marauders’ predatory success, I began to study the moment-by-moment organization of their raids. By watching where the ants first advanced at the front and then doing a slow scan back to the base column, I discovered I could treat what I saw along the way as a chronological sequence. In practice, this wasn’t necessarily straightforward. While the workers might be fearless with prey, they are skittish when it comes to other interruptions. They will retreat from a simple breath of air. To interpret their behavior, therefore, I stood as far off as possible. At times I used binoculars, once confusing a group of birdwatchers with my concentration on what must have appeared to be barren earth.

      The ants in the narrow swarm behind the raid front seem to move randomly, going backward, forward, and sideways with respect to the front. There obviously must be a net movement ahead to account for the raid’s progress, but it’s hard to detect ants following one another in that direction. Few trails are evident, and the ants appear to be moving through a diffuse cloud of orientation signals. The swarm advances to new ground every few minutes, and the land it formerly occupied is taken over by the forward part of the fan as more ants begin to form columns by running along specific tracks. The fan is differentiated from the swarm by the fact that it has these columns, and where that demarcation is made depends on the observer’s ability to pick them out. Farther back in the raid, the columns become fewer and busier, with an increasing proportion of ants on identifiable routes. Ultimately, at the back of the raid, the ants funnel onto one path, the base column.

      Roughly speaking, each part of a raid has a different function, turning the ants collectively into a food-processing plant. Prey is located by the foragers at the front, subdued within the swarm, then torn up in the fan. From there, it is transported along the base column to the trunk trail and delivered to the nest, where most of it is ingested by the ants. Things aren’t always so clear-cut, of course; as when kids crisscross the same ground on an Easter egg hunt, it is possible for workers in a swarm to find something the lead ants have missed, or for those in the fan to contact prey that’s on the run from the other ants.

      Mass foraging permits workers to flush prey and act in concert to catch it, like sportsmen engaged in a fox hunt but with the scale of operations increased a thousandfold. The downside of the ant stratagem is that the colony has to pack its greatest resource—its labor force—into an entity compact enough to cross a small area, instead of spreading those workers far and wide on individual search missions, as would a solitary-foraging species. The result is that the same number of workers finds less, but catches more.⁸ How? The deployment of these ants maximizes the capture of quarry too large for solitary species, yielding an intake of food that compensates for the slow encounter rate. All ant colonies stash a reserve of workers in the nest, to draw from as needed. Marauder ant swarms are made up of such assistants, transplanted from the nest to the site where food has been discovered.⁹ Keeping a reliable labor supply close at hand means that a raid can quickly respond to changing conditions—an essential component of success. Prompt conscription to the battlefront through explosive recruitment minimizes the time between the moment when workers first find prey and the arrival of reinforcements to pounce on it. No matter how fierce or capable the quarry may be, with no opportunity to make a getaway it will generally be overpowered by the rapidly escalating force of its assailants.

      In his book on military theory, The Art of War, Sun Tzu recommended this stratagem in the sixth century B.C.: “Rapidity is the essence of war; take advantage of the enemy’s unreadiness, make your way by unexpected routes, and attack unguarded spots.”¹⁰ The marauder ant, in its raids, has mastered this strategy beautifully.

      HOW RAIDS BEGIN

      The only time marauder ants are motivated to leave their manicured avenues and raiding paths to strike out as independent individuals, rather than in a coordinated raid, is in the face of disaster. Whenever I trod on a trail and mangled a bunch of ants, both the workers I panicked and their dead and damaged comrades released pheromones causing widespread alarm. The agitated survivors, whom I call “patrollers,” rushed about in a frenzy, dispersing up to a third of a meter from their trunk trail. Each appeared to take her own path away from the trail rather than tracking those around her. The patrollers seemed to be in a frantic search for the source of the problem and would give my leg a serious chew