contact. In Singapore I watched marauders steer hostile weaver ants up the tree in which this canopy species was nesting, and then the marauders gathered by the hundreds for a meal: they tore off the tree’s bark, rotating bits of it between their mouthparts and forelimbs while sucking out the sap. This food ordinarily draws the marauder ants only in times of scarcity, and indeed at the time there had been no rain for a week.
From springtails and seeds to frogs and large fruit, marauders harvest a cornucopia. They are reminiscent of humans, who apply the dictum “because it was there” not only to climbing mountains but also to adding tasty morsels to our diets. Marauders and people are exceptions to the general rule that in the tropics, where so many different organisms live together, most species, like the springtail-hunting trapjaw ants, become specialists in a narrow niche to survive the intense rivalry for resources.23 Marauder ants, in contrast, by interfering with all contenders for each meal and taking prey where others fail, exceed expectations by being geniuses at the competition game.
TRACKING FOOD FROM A TRUNK TRAIL
In Singapore’s Botanic Gardens one day, I placed a meter-wide plywood board in front of a raid. The ants crossed it in swarm formation, which confirmed my suspicion that their raids don’t depend on workers finding food or retracing old routes. Even so, I knew the ants were no fools—their raids slowed in areas with little to offer, the number of workers in them declining as the ants drained back to the nest until, if the dearth continued, the whole army would retreat. I decided to find out how the plenitude or distribution of booty changed an army’s strength and direction.
The marauder ant’s vegetarian proclivities made the job easy: it’s more difficult to manipulate caterpillars and crickets than to move fruit and seeds.24 Loaded with supplies from the grocery store on Orchard Road, I headed back to the Botanic Gardens and spread canary seed in a line extending from a trunk trail. It didn’t take long for the marauder workers to leave their highway and flow along this line. They tracked the seeds precisely, continuing outward in a column even after they had passed the last seeds. I had launched my own raid!
Did the distribution of food affect how the raid progressed? I poured a seed pyramid ahead of a swarm. The ants continued forward for several minutes after contacting this jackpot and then drained back to the food, where they rapidly built up in numbers. The swarm raid now over, the excess arriving ants radiated from the seed pile in a network of branching column raids spread over several square meters (a process called recruitment overrun, described in chapter 2). I had seen marauder ants generate similar trail networks under trees dropping fruit, which they thus track down quickly. While column raids are ineffective for catching fast prey, these bifurcating formations shine when it comes to fanning a foraging populace out over large areas. Each time one of the weak raids in a network encounters a bonanza, any number of workers can be summoned within minutes from the trunk trail to seize and consume it.25
What if the enticements are less concentrated? My next approach was to scatter a few seeds in a meter-wide swath off to one side of a swarm raid that was crossing a field with little in the way of food. The raid turned and followed my swath its entire 15-meter length, even though I laid few seeds—one every 20 square centimeters or so, which would put three of them in an area the size of my palm. Somehow, raids track subtle changes in food density, even though the workers coming upon each seed are ignorant of the food distribution as a whole.
How does that happen? While the ants follow exploratory trails at the raid front, they are more attracted to any recruitment trails they come across, which lead to food. When there are more seeds on one side of a raid, ants must be drawn to them by the buildup of recruitment pheromones left by the successful foragers from that direction. New arrivals tend to follow the strengthened routes leading to the food-rich region, causing the raid to turn and track the seeds without any of the ants comprehending what is happening—a fine example of what artificial intelligence experts call collective or swarm intelligence, in which the raid viewed as a whole deals effectively with problems by adapting to changes in the environment. A.I. experts would describe the raid as “robust.” Indeed, from computers to the natural world, scientists have found that seemingly thoughtful processes often emerge spontaneously from the integrated actions of simple-minded agents, like ants, with no need for leaders or any kind of management or centralized control.26
I went back to Orchard Road, depleting the grocery shelves of bird seed to continue my experiments. What mattered to the marauders seemed to be the relative abundance of food: when a raid was bringing in lots of other victuals, I needed more seeds to alter its course. The raids turned out to be smartly responsive to food in a variety of ways, branching or shifting in direction, width, and strength on the fly. Even though the absence of scouts made the raid blind to meals at a distance, the aggregate response of the workers to food at hand apparently enabled the raid as a whole to follow the food distribution in bountiful regions.
It’s a subject of endless fascination for scientists that each ant can only proceed locally on the limited information at hand, and yet their societies manage to act globally. Darwin was right when he wrote that for all ants do with their modest endowments, “the brain of an ant is one of the most marvellous atoms of matter in the world, perhaps more marvellous than the brain of man.”27 But the true power of the mind of an ant emerges at a superorganismic level, when those brains join to produce colony-level actions to accomplish a goal. Lewis Thomas, the author who first introduced me to the superorganism idea in my youth, described an ant society as “an intelligence, a kind of live computer, with crawling bits for its wits.”28
HOMEWARD BOUND
One afternoon it occurred to me that I could use the marauder ants’ ability to track seeds to unravel a mystery. Every trail has two directions. How do workers select the correct way home?29
In most situations, the ants have no problem choosing a direction. Because workers ordinarily find food at the raid front—the end of the trail—every returning ant has but one way to go. Along the route, though, are junctions with other trails. Some of these don’t present a problem: trails split at sharp angles, so nest-bound ants will make the right choice if they take the route that lies closest to straight ahead.30 Still, in the labyrinth of trails between raid and nest, I saw many situations in which the ants could have made directional mistakes but rarely did. Why?
I realized that by pouring seeds in an arc, connecting one point on a trunk trail to another point farther along the same trail, I could give the ants a choice of two equally good directions back to the nest. I watched in anticipation as the troops rushed from the trunk trail to track the line of seeds along each end of the arc. Every ant who picked up a seed from the advancing front of either column then turned around and carried it directly back to the trunk trail. When the advancing armies met, the ants now had the option of completing the full loop, and they often did so if they had’t picked up a seed. From their point of view, they were simply continuing as they had been going, away from the nest. A worker that picked up a seed after passing the site where the troops met would not turn around but rather would continue onward—a choice that, in any “normal” situation (not a loop), would have led her away from the nest.
The result was that all the seeds flowed away from where the armies converged. I called the trail segment within a centimeter or two to either side of this point the transition area because ants acquiring seeds in that stretch weren’t consistent in their choice of direction. The transition area was usually near the middle of the arc, but I could change its location by laying down the seeds earlier at one side of the loop, causing the ants who found that end of the loop to travel farther than they did on the other side before the armies merged.
At first, I guessed that the ants had marked the trail with some kind of “arrow,” as invisible to our eyes as the pheromone trail itself, which told their colleagues, “Go this way!” But that hypothesis crumbled when I waited until the seeds were nearly gone and
