Self-portrait, after stepping on a marauder ant trail near Malacca, Malaysia.

      Unless I bothered them further, all the patrollers would make their own way back to the trail within fifteen minutes. However, after stepping on marauder ant trails hundreds of times, mostly by painful accident, I noticed that on occasion a weak column would emerge from the bedlam and remain active much longer, advancing away from the trunk trail and branching here and there. Supplied with more ants pouring off the trunk trail, a minority of these columns would expand gradually into a wide, fan-shaped swarm raid that often reached 2 to 3 meters across—the largest I measured was 5 meters—and contained troops that pressed forward in concert. At this point, the ants no longer seemed to be looking for me but were again expanding out in a regimented hunt for food. So what began as a response to a footstep or possibly to a tree branch that had crashed onto a trail had transformed into mass foraging in epic proportions. Indeed, the more food the workers came upon in their journey, the more epic the raiding response seemed to become.

      This was different from army ant behavior. Marauders raid both in columns and in swarms, with an occasional column expanding into a swarm raid. Army ants typically raid either in columns or in swarms, but not both. Most army ant species are column raiders, whose raids stay in narrow columns from start to finish, whereas swarm-raiding army ants spill directly from their nests in a broad swath, and the raid continues as a swarm throughout.¹¹

      What triggers the marauder ants to launch a raid? Though there is no scout to shepherd a raiding party toward any particular meal, I noticed that when a patrolling worker fell by chance upon some morsel, nearby ants converged on the site immediately—suggesting that the worker who made the discovery had released a pheromone—and with their arrival a new pathway soon formed. This episode of conventional recruitment to something tasty could escalate into a raid when an excess of ants coming to the meal continued to advance in a column beyond it, a response known as recruitment overrun.

      Was food necessary to the process? Determined to get closer to the truth about how raids develop, I set up camp one weekend, coming as close as anyone has to roughing it in the Singapore Botanic Gardens. A tent would have called attention to myself, so I didn’t bring one. In any case, my intent was not to sleep, or even to move from one spot. All I needed was a camp chair and a stockpile of grub—Grainut cereal, fruit and cheese, and jerky. Stationing myself just far enough away from a 50-meter trunk trail so as not to disturb the action, I watched a 2.7-meter-long segment of the route for fifty hours straight. The midday sun left me roasted. During the second night, a storm waterlogged my notes. But it was worth it. Twice during that period I saw a raid start spontaneously, with a column of marauders streaming out from the trunk-trail throughway without food or provocation. If that had been typical for the entire trail, the colony would have been spawning a raid every forty-five minutes.

      I still needed to get a picture of how the raids related to each other. For a week, Paddy joined me at the Botanic Gardens to help me find out what the marauder ants were up to in the long term, in their choice of raiding locations. We mapped raids by marking each path with bamboo skewers emblazoned with neon-colored flags. Within days, the ground around the trunk trail resembled my back after my one session with a Singaporean acupuncturist.

      Most often, the raids crisscrossed the belts of land flanking the trunk trail. That is where the pattern became clear. Marauder ant raids moved readily both over virgin soil and across or along the course of prior raids, even ones from a few hours earlier. When a raid passed over an abandoned path, the foragers at the front seldom showed a change in conduct, neither avoiding it nor turning to follow it. On occasion, a raid seemed to retrace an old path a short distance; presumably, there’s a latticework of residual scents from an old raid that must dissipate with time. But in general, each raid went its own way.

      This differs from the activities of most ants, such as the seed-harvesting ants of the American Southwest. At first glance, the masses of harvester workers might be mistaken for an army ant raid as they pour out of their nest each day. Actually, though, they are less an advancing army than commuters caught in a traffic jam, reestablishing a trail to areas where they will then scatter to unearth seeds by foraging in the desert sand. Marauder raids resemble those of army ants in not being based on set courses; these mass foragers aren’t obliged to retrace their steps, and they easily cross unfamiliar terrain. The actions of the individual workers may be severely limited, but those of the raid as a whole are not. This is foraging in the pure sense, invoking the freedom to search unknown terrain, in this case moving as one.¹²

      MAKING SENSE OF ANT SCENTS

      A year and a half into my Asian sojourn, I made my way by train and bus from Singapore to the island of Penang, Malaysia, where I stayed for a month at a delightful research station on the beach. On several occasions, and with little warning, I was asked by the station manager to vacate my bungalow with its one small bed: a VIP from the American consulate required it for the weekend with his twenty-something “daughter.”

      Thus evicted, I would take the opportunity to travel to another rainforest site on the island that was thick with marauder ants. Curious about how the ants communicate to stay in tight formation, on these excursions I studied the marauder ant’s ability to produce chemical trails. With a field microscope, I dissected workers to extract two organs associated with their rudimentary sting: the Dufour’s gland and the poison gland, both known sources of pheromone signals in other ant species. I used a fine forceps to tease free the ant’s infinitesimal glands: thin, translucent sacs small enough to be an amoeba’s luncheon treat. I then smeared the contents on the ground near processions of marauders, to lead the ants where I wanted them to go.

      As I hoped they would, the workers dutifully followed the artificial runways. But their reactions indicated that the functions of the two glands differed. They followed the Dufour’s gland trails steadily and accurately and for a long time, which suggests that its secretion is critical in establishing trails, especially stable ones like a trunk trail. By contrast, their response to a crushed poison gland was to run like mad and sloppily follow the route for several seconds. Their brief excitement suggested the poison gland’s contents were reserved for inciting ants to capture prey or destroy an enemy.

      This wasn’t enough evidence to produce a set of sound scientific conclusions, but workers engaged in raids were too sensitive to my presence to allow for experiments on them. From watching the workers follow the scent trails I had drawn, I hypothesized that a marauder ant raid is prompted by two trail signals, as has been proposed for army ants as well. The majority of the routes within a raid must originate when workers at the front deposit exploratory trails: pheromones, likely derived from the Dufour’s gland, released as a forager moves on a new path.¹³ In addition, the workers in the vicinity of food lay recruitment trails that yield a massive response if the quarry—perhaps struggling prey—is attractive to many ants. Networks of columns materialize within the raid fan even when there is no food, however, suggesting that workers reinforce a selection of the trails from the front lines.¹⁴ This would lead eventually (I hypothesize) to the accretion of Dufour’s gland secretions into the base trail, and eventually, if that trail continues to be used and reinforced over time, into a trunk trail.

      Recruitment signals come and go as food is harvested. The ever-present exploratory trails are the glue that binds individuals into a foraging group, the closest parallel in ant societies to the adhesives that join the cells of our bodies. The front-line workers’ pheromonal scents keep the foragers immediately behind them close together and moving ahead as a unit, all the while leading the raid forward.

      For both the marauder ant and army ants, the varied attributes of the raids—the cohesive advance, the lack of a target other than the general land ahead, the absence of scouts—seem unrelated. But these features are manifested in a simple series of actions so circumspect and tentative that in humans they might be equated with separation anxiety. Unless they are diverted to kill prey en route, the ants are committed to a single goal: to follow fresh trails leading ultimately to unexplored terrain. Each ant stays near her sisters on routes that draw her inexorably out from the nest and onward, eventually bringing her to the raid front. There,