A week into my stay at California’s Sagehen Creek, I was lounging among the grass tufts next to a Polyergus breviceps ant nest with Faerthen Felix, the station’s assistant manager. I was trying to determine exactly how their raids transpire, and we had shown up early to catch the moment when the action began. But for most of the day not a single Amazon had shown her face. When the first slavemaker appeared at 4 P.M., she showed a supreme indifference to the slaves that were industriously collecting bits of my jelly sandwich. By 4:45, a hundred Amazons were milling about. While the foraging slaves went far afield, the Amazons stayed within 2 to 3 meters of their nest entrance.
That had been the habit of every colony I’d observed: no Amazon ants most of the day, then a slow buildup of milling workers near the nest late in the afternoon. Faerthen and I tried to figure out what all the wandering Amazons were doing. In time, the raid would depart, and they would join it, but until then they occupied their time examining every cranny near the nest. Were they foraging? Faerthen asked. No: finding meals is slaves’ work. The milling ants never picked up a thing, nor was there any indication that they laid trails or took notice of one another. Was their exploration part of colony defense? Perhaps. I had heard that when raiding workers encounter another Amazon colony, they attempt to destroy it and carry off its brood as a source of slaves of their own kind.1 So I dropped an alien Polyergus in front of the milling workers to see if they would respond—and was not surprised, given the threat she represented to their colony, that they immediately punctured her to death.
This Amazon worker is positioning her daggerlike mandibles to pierce the head of an ant from an alien nest. Her own head has already been punctured behind the eye.
But as this was the extent of their safeguarding activities, there had to be a better explanation for the slavemakers’ desultory bustle. I would find that Amazon colonies raid almost simultaneously at the end of each day, with virtually every wanderer joining in. By the time the raiders from one colony have gone far enough to encounter a competing nest, the raiders from that second colony have probably departed along their own path. Protection against other Amazon colonies must depend chiefly on the surfeit of workers that never leave the nest.
If not foraging or supplying defense, what do the milling Amazons accomplish? Watching their incessant scrambling within this staging area, I proposed to Faerthen that they might be energizing themselves for the upcoming battle. That would make them a bit like human troops performing drills—sharing movements and chants, a practice known to deepen identification with the regiment.2 Wolves and wild dogs engage in similar rallies, which assure that everyone is “awake, alert, and ready” before a hunt.3 In any case, after this seeming display of bravado, which typically ended between 5 and 6 P.M., the raid started fast and moved quickly.
The first indication of an incipient raid is a surge of Amazon workers in one direction, with the milling ants in that quadrant joining an outpouring from the nest. The exodus bears some similarity to the way a raid explodes from an army ant bivouac, except that with Amazon ants, in the few minutes after most of the milling ants join the raid, the number of ants lagging behind declines to zero. A stream of workers a few meters long moves away from the nest like a swift snake, albeit one that is straight as a stick.
For a distance of up to 140 meters, the procession glides at speeds amounting to nearly 200 meters an hour—or a foot every few seconds. That’s ten times faster than an army ant raid. Though I’ve become pretty good at stalking animals, on two afternoons I staked out a colony and missed the whole raid: both times I looked away from the nest for a moment, and all the ants were gone. Even the ants’ orange color and the open landscape didn’t help me find the raid once it slipped out of sight.
What triggers the ants to leave the nest, and how are they able to travel so quickly and efficiently? Most of the details of the subjugation of Formica by Polyergus breviceps have been gathered by Howard Topoff. Working in Arizona, Howard discovered that, unlike raiding army ants, Amazon ants target a single location and are brought there by a leader. Earlier in the day, in fact while her nestmates are ambling haphazardly around their nest, this motivated individual has been scouting for victims, and when the raid begins, she is in charge of the group.4
Researchers studying other populations and species of Polyergus, however, have denied the existence of scouts.5 Watching the Amazons at Sagehen Creek, I understood their doubts. I had been unable to pin down the leader of the raid or to discover her earlier in the day, while she was seeking a target colony. But animal behavior can vary. Even the species of ant enslaved by the Amazon ant differs between locations. Perhaps the California and Arizona Amazons had different scouting behaviors as well. I took it as a personal challenge to confirm the existence of scouts at Sagehen Creek.
And I failed. Hour after hour, day after day, I could not identify a single scout.
Not only was I unable to pick out any leader within a raid, but I was puzzled by certain things about the raids themselves. I could scrape away the soil in front of the column and the ants would continue, which told me that, like foraging army or marauder ants, the Amazons were not tracking a scent. But if the foremost individuals weren’t following an obvious leader, they also didn’t replace each other the way those at the front line of an army or marauder ant raid do, by advancing incrementally before retreating. Instead, all the Amazon workers kept pace with each other while moving forward, with only occasional meanders and a few abrupt shifts in course. They had to be acting on some kind of guidance. That would explain why Amazon raids advance more quickly than army ant raids, despite the fact that the army ant workers are much faster runners.
More than anything else, the Amazon ant raids brought to mind the termite-hungry gangs of Pachycondyla I’d seen while studying driver ants in Nigeria. Instead of taking victims en route, as army ants do, the Pachycondyla raids target a clump of termites, make their kills, and then retreat. Compared to the Amazon raids, though, the Pachycondyla groups were slow and even more tightly compact—one or two ants wide—making it easy to pick out the group leader who had scouted out the termites earlier in the day.6
It seemed, then, that my best hope was to spot the Amazon trailblazer before the raid started, as she departed the nest to scout for Formica. But what chance did I have to spot one ant, when a whole raid could so easily escape my notice? I decided the problem was manpower.
So a year later, I hired two eagle-eyed assistants and returned to Sagehen Creek. Starting at 4 P.M. each day, we walked in wide circles around three different Amazon nests, beyond the reach of the milling ants. I figured any Amazon traveling that far from home, that early in the day, would have to be a scout. Dizzy from circumambulating, we finally located a loner moving in a beeline away from one of the nests—a scout for sure! We watched her progress closely, beguiled by the possibility of untangling a mystery. She was unswerving during her outbound run, undistracted by any Formica worker or nest she came across. When I dropped a Formica pupa in front of her, she didn’t take notice, even though capturing brood was the whole point of the exercise. In the course of a week, we spied several more scouts, and the story was always the same. Only after at least 25 meters did each one cease her epic run and look around, as if a switch had been flipped in her head.7 For the first time, she sensed and responded to the presence of free-living Formica.
Clearly, she couldn’t do her job if she were distracted by single foragers, because they were everywhere, and their occurrence would not necessarily indicate a colony nearby. It was only the presence of a nest that caused the scout to head home and recruit a raid party. I didn’t see it, but Howard tells me a scout may stick her head into an entrance to confirm that she has found an active colony. I imagine that Formica guarding the nest entrance will preemptively ensnare a scout if she is detected during her act of espionage.
In Arizona, Howard found that the scouts are few, typically the same two or three individuals every day, each going a different direction: if one didn’t succeed, another might. That seemed to be true at Sagehen Creek as well. On some afternoons, no raid departed, presumably because no scout had located a Formica nest. After waiting for a raid to begin its mad dash, I was often dismayed to see the workers continue to mill about the nest until the light declined and they trickled back to their abode.
Raiding army and marauder ants gamble on finding prey without sending out any scouts at all, a strategy that pays off because their strength in numbers increases the odds that they will catch any prey they do find. Amazons gamble in another way, by putting their options in the hands of a few. Why not engage more scouts?
Howard may have found one answer. When he removed scouts as they left a nest (returning them each night after the raiding period was over), within a few days their number increased to as many as thirty. Apparently there is a supply of ants that can take on the reconnaissance role as necessary. While a bigger party of these scouts would find the day’s target colony more quickly and dependably and perhaps closer to home, the fact that scouts have blinkers on with respect to Formica until they have traveled a certain distance suggests it is to the slavemakers’ advantage to avoid overharvesting the most easily available sources of slaves. Two or three scouts must be successful enough at finding the remote colonies that drawing from the reserve of scouts in the nest is a rare contingency.
Preparations for the Amazon raid begin with a couple of intrepid scouts venturing into the wide beyond, while dozens of warriors rally near the nest entrance and thousands more wait inside. When a scout returns bearing news of a Formica nest, she mobilizes this widely spread workforce. As she passes one of her milling nestmates, she touches it vigorously. That worker in turn excites others. A chain reaction ensues.
That process begun, the scout enters her nest, where she continues to assemble the troops. Though she may contact just a few individuals in the minutes before she heads out again, by then hundreds have been alerted to follow her. (Some afternoons more than one scout locates a Formica nest, and multiple raids begin in conflicting directions. All but one of them usually quickly aborts. A reasonable hypothesis is that the outbound ants prefer the most “enthusiastic” scout—the one that conveys the strongest signals.) The raid advances, led by the former scout, who now serves as guide. The other ants diligently follow the scent she releases, very likely reinforcing her pheromone with some of their own to help keep the ensemble in formation.
I never did learn how to pick out the leader within an Amazon raid. There were always several identical ants spilling across a raid front as wide as my hand. How could I identify her? One time, I used a forceps to snatch a worker who was slightly out in front of the others and put her in a jar. The raid continued, unaffected. I took the next-foremost worker, then the next. Six ants later, the raiding workers suddenly broke formation and spread over the ground. It seems I’d finally captured the scout. (Clearly she hadn’t been in the lead at first; her entourage must swarm loosely around her. But presumably she never lags far behind this vanguard.) I examined my captive with a magnifier. She was more darkly colored than her sisters. This is usually a sign of an older individual—in this case, I assumed, a relatively experienced worker who had taken on this essential leadership role, making her by far the hardest-working ant in her family.
I treated my captive with respect, dropping her gently among her followers. Having ceased to receive her guidance, they were now acting the way the workers normally do when a raid reaches its destination, spreading out to search for a nest to attack. With her retinue dispersed, the leader wasn’t able to wrangle them any farther. Soon they all gave up and went home.
How does the guide find her way back to the Formica nest she located earlier, raiding squadron in tow? She did not leave herself any clues to her prior route—no pheromone breadcrumbs to follow. Howard discovered that she steers by the sun, using solar navigation not only on her initial hunting excursion but also on her return to mobilize an army, and again on her second outbound run, when she leads that army to the Formica booty. With no other information at her disposal, she takes a slightly different route each time.
The leader doesn’t need to see the sun itself. She can use polarized light, the scattering of sunlight through the sky. This allows the ants to determine their direction even if the sun is hidden behind a tree or a cloud. But trees and clouds hide polarization cues as well, so a patch of sky must be visible somewhere. I illustrated this with an experiment Howard suggested: I taped a piece of waxed paper between two sticks and held it taut above the foremost raiding ants. Immediately the workers scattered. Although the paper had let through plenty of light, the polarization signals were obscured, so the guide ant could no longer shepherd her nestmates. After some fruitless snooping around, the workers returned to their residence.
Amazon ants at the advancing end of a slave raid rush long distances at Sagehen Creek, California, led by a single worker.
Rain, like clouds, may force a raid to be canceled. In another experiment suggested by Howard, I was able to get a raid to turn back in a panic by showering the ants with a pinch of water. Fortunately, most afternoons during the raiding season are warm and sunny, the kind of weather the Amazon ants prefer.
Once the procession is under way, the guide simply halts when she reaches the site where she found the Formica nest, presumably ceasing to release her pheromone signal and thereby allowing the battalion to spread over the nearby terrain. At this juncture, her troops seem to switch from “follow the leader” mode to “search for nest” mode. (The change is remarkable and abrupt, mirroring the change in the scout’s behavior from “run ahead single-mindedly” to “search for nest” on her initial, outbound trek.) Insofar as searching for slaves is the closest the slavemaker workers come to foraging, foraging has begun.
As I watched the Amazons rooting around for a passage into the besieged colony, a simple explanation emerged for the wandering they did at their own nest prior to a raid. They were acting no differently than they were now. Perhaps the milling ants had been looking for Formica from the moment they emerged from their nest. The leader’s pheromones (I hypothesized) could have a hypnotic effect on her nestmates, diverting the overly eager raiding workers from their ferreting behavior long enough to take them to a location more sensible for their search.
A worker leading her companions with only the help of the sun is unlikely to arrive at the precise location of the Formica nest entrance she found earlier in the day. No matter: colonies have multiple entry holes. Brought to the outskirts of the targeted colony, her comrades need only spread over a few square meters to find a way in. Each raider seems to emit a pheromone as soon as she detects an entrance, for reinforcements come fast—with luck, before the Formica can erect a blockade.
Whether the raid is a success or a failure, the return trip is a cinch. The pack is no longer dependent on its leader. The pheromones the workers released to keep themselves together on their outbound journey persist long enough for them to follow home.8 The retreating ants make use of polarized light as well, though. Taking on a Pied Piper role, Howard has managed to delude the ants into turning around and returning the brood to the nest they just raided by shading the advancing column and using a mirror to redirect the sun.
HOW DID SLAVEMAKING BEGIN?
In The Origin of Species, Darwin proposed that the ancestors of slavemakers may have eaten the brood of other ants, much as army ants do today. He suggested that when the booty wasn’t consumed fast enough, some transformed into adult workers that imprinted on their captors and ipso facto became slaves. At first the nascent slavemakers would have relied on their slaves most heavily for jobs that were costly or dangerous, such as foraging and defense. But with their slaves constantly performing all duties out of habit, the slavemakers would have gradually lost their other domestic skills as well, culminating in the modern Amazon ants that will starve to death if there are no slaves to hand them food. Over many generations, the addition to the supply of laborers would have become so valuable, even essential, that predatory raids metamorphosed into slave raids. Darwin saw the raids in effect as a circuitous kind of food search: what was once a foraging enterprise became a quest for individuals that would do the foraging—that is, a fresh batch of slaves.
Today, the most widely accepted conjecture about the origin of slavery in ants holds that the slavemaker ancestors were competitive species that took brood as part of their war booty and then ate it, even though ant brood wasn’t an everyday part of their diet, and some of these occasionally survived to become slaves.9 This idea has gained favor because the groups of ants that most often evolve slavery have been species that fight over territories or other resources rather than the specialist cannibals of other ants. Despite their predatory raids, for example, no army ant species has become a slavemaker.10 Remember the acorn-dwelling Temnothorax that are subject to enslavement by Protomognathus americanus? In territorial battles between their colonies, Temnothorax eat any immatures they seize.11 But on rare occasions, and seemingly by mistake, they will rear one to adulthood. Such an individual will serve as an accidental slave—unless her confused nestmates kill her first. Indeed, the evolution of slavery must have included a suppression of the worker impulse to kill foreigners, thereby allowing would-be slaves to coexist with them.
The honeypot ants of the American Southwest are a territorial species with an aptitude for slavery. They show fascinating ritualized behaviors designed to avoid lethal conflict, engaging in mock fights called tournaments in which ants from different colonies circle one another on tiptoe to try to make themselves look larger. In this so-called stilt walking, they will sometimes climb on pebbles to stand taller than their neighbors, a ploy that anthropologists call tactical deception, which is associated in primates with keen intelligence.12 For species with a modest labor force, ritualized conflict is a reasonable strategy to avoid casualties, though if one side determines that their workers are larger and outnumber the others, they raid the weaker nest, gorge on its brood, and drag back its honey-filled replete workers as slaves.
Slavery evolved in both ant and human societies out of the need for a compliant and manageable labor force to drive their extensive systems of collective production. Ants are concentrated on producing the brood that will be the source of the next generation, and the labor force geared to this task is normally made up of the ant colony’s own workers. But why work? If we consider the slavemaker colony as a superorganism, it’s as if a creature obtained its arms not by growing them but by grafting on fully formed limbs that it pulled off someone else and can replace whenever it needs to, something like the horticulturalist who grafts a branch of one apple tree onto the trunk of another. Apart from the effort required to make this graft—or steal the slaves—in the first place, the outcome is cost free and pays off handsomely as long as each slave brings in more food than the meat she would have provided had the slavemakers eaten her instead of letting her live. By this measure, a slave probably nets the colony a profit in a few days, with an hour-long raid yielding many thousands of hours of services by the ants enslaved.
The graft has the most value when the initial cost to the colony of procuring it is least. Capturing pupae is the key. Although nursing the larvae is labor intensive, pupae—ants at the nonfeeding stage between larva and adult—yield a new slave crop quickly and with almost no effort. In fact, the slaves raise only pupae to become new slaves; the few eggs and larvae sacked by their masters become their food.13 These snacks reduce the number of mouths the colony has to feed. There’s no explanation for why the slaves make this decision about what to eat. But the slaves even consume the pupae if they are hungry enough, as always regurgitating a portion of each meal to the Amazon workers.
When honeypot ants engage in ritualized combat over food, the colony that draws fewer big workers retreats, often without injury. By standing on a pebble, the worker at right is “cheating” to appear larger than she is, driving off a larger opponent near Portal, Arizona.
These observations are a clue that sustenance and slave raiding are linked: after all, raids bring in the slaves that will in turn bring in the food (or, at the whim of other slaves, end up as food themselves). Amazons raid less often when there is a food glut, which makes sense, given that a well-fed colony probably already has an adequate supply of enslaved foragers.14 Future research should resolve whether raids are prompted by hungry slavemakers and slaves. If they are, eating more of the booty may be a means of sustaining a colony during the time between procuring brood and turning it into foraging slaves.
Slavemakers are commonplace in the temperate zones, even in suburban backyards. But their absence in tropical climes is an enigma. There are theories, of course. One has to do with numbers of ants. Temperate-zone ants frequently subject to enslavement, such as Formica and Temnothorax, tend to be superabundant and yet easy to attack. Despite the large number of ants in the tropics, few tropical species are as plentiful as Formica and Temnothorax, and the ones that are, like the weaver and marauder ants, are extraordinarily well defended.
I once traveled a few weeks on the Paria Peninsula of Venezuela and in the Arima Valley of Trinidad with Robin Stuart, an expert on the North American slavemaker ants that reside in acorns. We spent our time watching little brown ants called Nesomyrmex come and go from nests in the hollow twigs of roadside shrubs—while trying to ignore the roaring logging trucks at our backs. These weren’t attractive locations, but Robin and I had decided Nesomyrmex were promising candidates for enslavement, and it would be a coup for us to find ant slavery in the tropics.
Nesomyrmex interested us not only because they are common and innocuous but also because they are cousins to the temperate Temnothorax, or acorn ant. By occupying a number of acorns, a single acorn ant colony is fragmented into isolated housing units (polydomy), each of which may have one or more egg-laying queens (polygyny). Both these attributes may increase the openness of the colony to invasion by outsiders, including slave raiders. Robin and I found that Nesomyrmex colonies are likewise characterized by polydomy and polygyny, and yet we found no social parasites in their nests.15
Perhaps it is not the differences in colony organization but rather the seasonality of the temperate zones that is conducive to ant slavery. The production of brood in temperate-zone ants has an annual cycle, and slave raids of the Amazon ant take place during the few summer weeks when the free-living Formica nests contain workers in the coveted pupae stage of development. Earlier in the year, while the Formica colonies are still rearing their larvae, the Amazons are full- time stay-at-home loafers, their slaves tending the Amazon queen’s brood that will mature into a fresh cadre of warriors by the time they are needed, in the raiding season. Seasonal production of brood is less pronounced in the tropics. It’s also been proposed that seasonally cool temperatures might dull the ability of ants in the temperate regions to recognize the alien queens when they infiltrate their colonies, a necessary step in the evolution of many slavery species.16
I believe there is another possibility. Slavemakers show a rare forbearance in giving up the immediate gain of eating food (stolen brood) for the potentially greater long-term benefit of having slaves.17 Such delayed gratification can be especially advantageous in temperate environments due to the hardships arising from predictable changes with season, as well as more extreme and unexpected cold or warm spells.18 In essence, slavemakers such as the Amazon ant have chosen to hoard not meat but slaves, whose efforts help tide them over in lean times. Hoarding in the tropics is less often a life-and-death matter, because animals are more likely to procure a steady-enough food supply that they can eat meals when they find them or shortly thereafter.19 For Amazons, the payoff is a comfortable lifestyle in which workers avoid work outside of the raiding season, leaving their slaves to toil, come rain or shine.