Oecophylla weaver ants swarm through the tropical rainforests of Africa, Asia, and Australia, but because life in the trees has so many advantages, the New World has its own hyperaggressive canopy-dwelling ants.
One morning in late spring 1990, I found myself slung by ropes a dozen meters above the jungle floor rummaging for beetles in clumps of litter on tree branches. I was in Peru, on assignment for National Geographic magazine, to document the rainforest canopy, one of my research specialties.1 Since finishing my thesis four years earlier, I had served as the curator of Harvard’s ant collection (where I’d first seen the marauder ant and decided to make it my quest) at the Museum of Comparative Zoology under my former adviser, Ed Wilson. But now I had gone freelance, planning to support my life and research with writing and photography.
That morning at sunrise I had been on the ground with Terry Erwin, a Smithsonian beetle expert who inventories the species diversity of canopy insects. To get close to our targets I had sent a fishing line into the tree with a slingshot and used it to pull a climbing rope over a branch; I then got into my climbing harness, clipped two ascenders to the harness and the rope, and wriggled skyward.
But as I rose, my support rope shifted; I abruptly fell several inches and began to spin in space. Plant bits shaken loose from the branch above whirled into my eyes and blinded me. My hands were full of cameras and entomology gear. To stabilize myself, I threw my legs around a branch high to one side.
Big mistake! Swinging through the air, my foot smashed a mass of canopy-rooted plants, or epiphytes, that concealed a well-defended ants’ nest. In an instant, workers covered my legs and then dropped like dive bombers onto the rest of my body. As they gashed my skin with their mandibles and sprayed formic acid into my wounds, I recognized not only the species, Camponotus femoratus, but also the fact that I’d found my first “ant garden”—albeit the hard way.
Regaining my balance while slapping at this vicious species of carpenter ant, I noticed the presence of a second ant on my skin—the smaller Crematogaster levior, a shy species of acrobat ant that does not bite. The ant garden is a result of their collaboration and represents an infrequent instance of harmony between ant species. Nestled in this mass of epiphytes, a confederation of these two ants had constructed a quarter-meter-wide treetop house of carton, papery sheets they produced by masticating plant matter and soil. The workers then collected seeds and embedded them in the carton. There the seeds grew into cacti, bromeliads, figs, orchids, philodendrons, and anthuriums, creating a bounteous garden.
The plants and ants depend on one another. The plant roots strengthen the carton, keeping it from disintegrating in rain and giving the ants a stable home.2 The ants, in turn, seem to be necessary for the plants’ survival, since these particular species of flora never occur on their own.3 (Though we can’t say yet if the seeds die if the ants don’t find them, or if the ants are so thorough at snapping them up that these plants have no opportunity to germinate elsewhere.) In any case, the ants were clearly protecting both nest and garden with zeal.
In this striking example of mutualism, the Camponotus and Crematogaster jointly created the nest and protected the epiphytes. They shared trails, helped each other find prey (though Camponotus can be a bit of a thief), and tended the same sap-sucking insect “cattle” as an additional source of fuel. The acrobat ants then drank the honeydew excreted by the smaller Homoptera, or aphid relatives, and reared them to a size suitable for milking by the carpenter ants, which played the more important role in finding and planting the seeds that developed into fresh garden plants.4
Nauseated by an overdose of ant toxins in my bloodstream, I pushed myself away from the ant garden to another tree trunk. The garden was truly elegant, I could see, though for the moment the ants on it seethed. As I recovered my balance, I pondered what it was about life in the canopy that fostered both mutualism and belligerence.
BIOLOGICAL SUCCESS
Success in nature is often described in terms of the number of species in a group. By this measure, the ant-garden ants I had stumbled upon belong to two of the three most successful genera of ants (the third is Pheidole, or big-headed ant), with hundreds of species each. But success isn’t always associated with a proliferation of species; the number of individuals and their effects on nature can matter more. Indeed, ants are the prime example.5
The tropical forest canopy, with its multiple levels of foliage and branches, can have ten times more habitable real estate than the ground, a much higher ratio than in the temperate zones. With all that elbow room, it may be no surprise that a study in the Amazon basin found eighty-two ant species in a single tree, almost twice the number of ants in the entire British Isles.6 Though that sounds like a lot, compared to other insects in tropical canopies, ants have an almost negligible diversity. As Terry Erwin points out, a single tree in Peru can contain thousands of species of beetles alone. Still, arboreal ants more than make up for their relatively few species with an astonishing bounty of individuals. Workers, in particular, often make up 20 to 40 percent of the organisms in trees, microbes aside. Measured by weight rather than numbers, all of the ant species in combination account for 10 to 50 percent of the mass of arthropods living in tropical trees. Ants also weigh more than all the vertebrates in the same area, from frogs and lizards to parrots, monkeys, and leopards. With so many ants and so little else, canopy ants sustain their populations through heavy reliance on plant matter, as we saw for weaver ants.7 The same is likely to be true for tropical ants living on and in the ground, where they also roam in overabundance.8
Pervading the tropics of three continents with just two species, weaver ants are a particularly good model of success without diversity. In this regard, weavers and humans have a similar history. Our ancestors adapted better than Neanderthals and earlier branches of our evolutionary tree, which stopped producing offshoots in the wake of Homo sapiens’ aggressive dominance of the Earth—with six billion members now and counting. Weaver ants seem to have followed a similar course, controlling the environment to such a degree that they are often able to push out or mow down the competition. Along with South America’s Camponotus femoratus, they are among the most militant ants on Earth, capable of eliminating all adversaries except the most fierce. This they accomplish by being numerically and behaviorally, and therefore ecologically, dominant, using their force of numbers and tactical skills to suppress or conquer territorial competitors and thereby control the environment.
Is numerousness essential to weaver ants’ success in fighting, or is it their belligerence that allows them to expand their population? The two conditions seem to go hand in hand, making it difficult to distinguish cause and effect. Although marauder and army ants at times use strength of numbers and battle skills to overpower the competition, the goal of most violence in these mass-foraging ants is the practical one of securing food supplies. In contrast, weaver ant societies, much like Peru’s ant-garden ants, fight other colonies to control the surfaces on which they live.
Weaver ants tearing apart a driver ant captured in Ghana.
This difference in goals has parallels in human groups. Most early hunter-gatherers moved often in pursuit of foods that offered immediate large payoffs, such as big game. After the Pleistocene, human population pressure caused these slowly replenishing foodstuffs to become depleted and eventually forced people to settle down in areas chosen for the availability of fast-breeding foods such as grain and small game, which required more time and labor but could be harvested sustainably. This shift in turn necessitated vigorous defense of these territories against would-be usurpers.9 In the insect version of this “broad-spectrum revolution” (as anthropologists refer to this shift in human diet), each densely packed ant garden or weaver settlement, with its foraging centered on a broad range of such quickly renewable resources as insect prey and nectaries, has come to approximate a warfare state. Among animals, all-out war against their fellows occurs only among the largest societies of humans and ants.10
TAILOR-MADE ANT ACCOMMODATIONS
Where do so many ants find homes in the trees? Many nest in hollowed twigs or galleries in bark, or the litter that accumulates among the roots of orchids and between the leaves of tree ferns.11 Cavities capacious enough to hold large colonies are rare, though, and the success of such colonies often depends on constructing nests, such as the weaver ant’s tents, with materials they find in the canopy or produce themselves. There are other ants that use silk, usually combined with leaves, to build their nests—mostly larval silk, though in one African species the adults synthesize silk of their own from a gland near their mouths, and an Australian ant steals its silk from spider webs.12 The ant-garden ants are among many species that use carton.
There are ants that nest terrestrially and forage in the trees, giving them a toehold in both environments. This is more common than the reverse, a canopy-nesting species that primarily forages on the ground. It’s a classic suburban commuter’s compromise between the best housing and the best income: the forest floor provides more roomy nesting opportunities, and the food and other resources found in the canopy make the transit worthwhile.
While the canopy species mentioned thus far live on any plant that offers a suitable nesting cranny, certain trees, epiphytes, and vines provide custom-fitted ant accommodations. Some of these so-called ant plants cater to a specific ant, providing food and board suited to no one else.13 Why? These residents are proficient at eating herbivores, and they kill anything that sets foot on their host. As a boy, I read how Cecropia trees house Azteca ants in spacious compartments in their trunk joints, feeding them pale, glistening “food bodies” more nutritionally balanced than nectar or honeydew, which exude from the base of each palmate leaf. On my first trip to the tropics as a college student I ran into one such tree—literally—and learned that Azteca don’t just pick on creatures their own size.
Weaver ants, though similarly aggressive, do not occupy specialized ant plants. They can live in any tree by creating their own nests and finding their own food (if the plant has nectaries, so much the better). Childhood experience guides the choice of plant homestead: workers and queens prefer to nest and forage on the tree species they grew up on as larvae and young adults, and, like humans, they become more set in their ways as they get older. Still, the ants show a special affinity for mango and citrus, a fact that encouraged the Chinese to use Oecophylla to control citrus pests as far back as A.D. 304.14 In parts of Africa and Asia, their use in biocontrol continues—though pity the laborers who climb those trees to pick the fruit!
Is the weaver ant’s presence good for the trees? To answer such a question, ecologists conduct a cost/benefit analysis. In some ways the benefits clearly outweigh the costs: weaver ants cull leaf-munching insects, and tree foliage lasts longer where the ants reside. One type of beetle, though fond of foliage, flies away from a tree the moment it senses weaver ant pheromones.15 Another benefit to trees might come from weaver ant hygiene, or rather the lack of it. Nutrition is a problem in tropical forests, where soils are thin—but a tree can absorb nutrients through its foliage as well as its roots.16 The workers in some ant-plant mutualisms use leaves and stems as toilets and trash chambers, thereby feeding their plant. The fecal droplets that weaver ants scatter over leaves to mark their territorial claims might serve as fertilizer, too.
Known as the “dinosaur ant” for her primitive appearance, this Nothomyrmecia macrops worker from Poochera, South Australia, is guarding scale insects that have exuded so much honeydew that they appear to be covered in sugar.
On the negative side, the leaves that the weaver ants incorporate into their nests may be lost prematurely to wear and tear, as the ants pull them from their ideal alignment for photosynthesizing.17 However, since only a tiny percentage of vegetation is tied up in nests, these costs to the tree should be low.
A bigger deficit item may be the ants’ nurturing of their homopteran cattle. In the United States, you can locate Formica propinqua ants by the dead cottonwood trees around their nests, which have been sucked dry by the aphids the ants raise.18 In addition, some sap-sucking insects carry infections, making them the plant version of the malaria mosquito. In most situations, though, the cost of Homoptera to trees is not so severe. Azteca, for example, raise sap-sucking insects in moderation on their Cecropia hosts to no evident ill effect. In fact, some trees may have evolved to be tasty to such insects because they attract protective ants, as an alternative to producing nectaries.19
Some of the weaver ant cattle reside not on the trees themselves, however, but on vines in their crowns, which have wide vessels ideal for feeding by Homoptera. Heavy infestations of “plant lice” raised by Oecophylla may in this case inhibit vines from shading a tree or weighing it down and breaking its crown, thus working to the tree’s advantage.20
Overall, weavers are thought to benefit most trees. Could the relation of the ants to choice tree species such as mango and citrus be a rudimentary mutualism, as with Azteca ants and their Cecropia trees, though less precise and obligatory? Researchers have noted of mango and citrus trees that the “odors of the plants may have evolved to attract ants for protection.”21 And the tree wouldn’t be the only one to profit from this arrangement: anything that increases its vigor should benefit the ant colony it houses, by yielding more durable homes and sweet and savory foods—honeydew and prey.22
SPECTACULAR DEFENSES
A few years after my trip to Peru, Dinah Davidson of the University of Utah offered to show me another dominant ant species and impressive adversary of the weaver ant in Brunei, a small, oil-rich country in northern Borneo. After touching down to an evening view of the Sultan Omar Ali Saifuddin Mosque, I arose the next morning for a forest river journey on a canopied boat. Kuala Belalong Field Studies Center was just as I remembered it, solidly built at the base of thickly wooded hills. Dinah, a compact woman with hair cut short for the field, took me up a steep path while pointing out weaver ant territories, which alternated with trees occupied by any of sixteen species of Camponotus carpenter ants belonging to what’s known as the cylindricus group.
The cylindricus ants have dramatic methods of defense. The major worker’s head, for example, is flattened into a disc, enabling her to serve as a living door to nests in hollow branches. She allows her nestmates inside only after they identify themselves by tapping the blockading disc with their antennae (a technique also seen in other ants). Dinah took me to the territory of one of the more unusual cylindricus species and told me to grab a minor worker that was climbing the trunk. I did, and the ant’s leg fell away in my hand, in much the way that a lizard will lose its tail.
Still other cylindricus species exhibit the most extreme behavior of all, employing the “suicide bomber” response to its enemies that I had come to Brunei to see. Wishing me luck, Dinah left me at the base of a tree occupied by one of these colonies and departed. I pulled out my camera, adjusted my flashes, dripped some honey next to the tree from a vial in my pocket, and waited. After an hour, weaver ants along with another species of carpenter ant located the bait and started arriving at the cylindricus-occupied tree. One of them started up the trunk, but then came down again. That one would live another day. Another climbed a bit higher and attempted to walk by a cylindricus minor worker. Just as I clicked the shutter there was a splash of yellow, and both ants were immobilized in a sticky, grotesque tableau.
That picture made my journey halfway around the world worthwhile. Photography is, for me, a tool for storytelling, and this ant’s story left my heart pounding. Approached by an adversary, the cylindricus had blown herself up, her body rupturing with a muscular convulsion that spewed forth a toxic, lemon-colored glue that pinned her foe to the ground, killing both of them straightaway.23
A Brazilian ant I’ve yet to see, Forelius pusillus, has an equally fatal approach to protecting the nest. Up to eight sacrificial individuals stay outside at night to seal the entrance with sand, kicking the final grains in place until no trace of the hole is visible. Walled off from their sisters, by dawn almost all are dead, for reasons unknown—perhaps the squad consists of the old or sick. The ants in the nest then clear the passage to begin the day’s foraging. That night, more victims seal the door.24 No one can say what prompts this preemptive defense, though dangerous army ants would be one safe bet.
A Camponotus cylindricus–group ”exploding ant” has ruptured her body to spew a sticky yellow glue, which has killed both her and the larger worker of another species of carpenter ant in Brunei, Borneo.
In northern Borneo, cylindricus often jointly control their canopy territories with certain Polyrhachis, which have their own self-destructive defense. The first time I saw a gleaming gold specimen of one big, attractive species of this genus, I couldn’t resist touching her—and immediately had the worker embedded in my finger and unable to remove herself, thanks to the fishhook-shaped spines on her back. Birds and lizards must learn to avoid these pincushions.
DOMINANCE AND SUBORDINATION
If these colonies are viewed as organisms, a worker’s death is of no more consequence than a man cutting his finger. The larger the colony, the less consequential the casualty. Extremist defenses, then, are a manifestation of a large labor force. Such extremism in handling risk is an example of how death without reproduction can be of service to queen and colony, and a reminder that anything humans concoct—even suicide missions and terrorism—probably has a parallel in nature.
Just as trained armies and impersonalized warfare came into being among people as populations exploded with the development of city-states, inexorable, large-scale offensive and defensive conflicts between rival ant nests usually involve the numerically dominant species, with their huge colonies. One likely reason is that the necessary communications are best orchestrated in large societies, whether they involve written languages in humans or pheromones in ants. Another reason is simple efficiency. Larger human settlements have higher per capita productivity, with fewer resources required to feed and house each individual.25 This pattern, if similar among ants—which remains controversial26—may enable large societies to more easily accrue the spare time, energy, and resources that can be invested in creative endeavors (by people) and armies (by both ants and people). As a result, not only are the ants of large societies more expendable individually, but the group as a whole may also be able to take more large-scale risks, given that losing 10 percent of an army will be more devastating for a society of ten than for one of a million.27
Yet another advantage of community size is that populous societies control large spaces, and large spaces have relatively small perimeters (a large circle, for example, has a smaller circumference relative to its area than a small one). Thus, the bigger a colony, the smaller the proportion of its population that needs to be employed in border surveillance, and the more troops it has free to commit to offensive battles.
Ecologically dominant species are usually too competitive to coexist. On occasion there is a détente between two of them based on different nutrient and housing needs, as with the antgarden ants, which eat different-sized prey, or the Polyrhachis species that share territories with the exploding ants. Polyrhachis nest on the ground and eat honeydew and insects, while cylindricus nest in trees and specialize in licking the microbial film growing on foliage.28 Even with a large labor force, it pays to be selective in targeting competitors. The Polyrhachis and exploding ants have largely put aside their differences in a coalition against the weaver ant. Weavers in turn show enemy specification, picking out intruders of other dominant species and excluding them from their territories as if they were competing weaver ant colonies. Weavers don’t win all their battles, and sometimes they choose to retreat. The exploding ants can keep them at bay, while the workers of the Asian Technomyrmex albipes, a species known for noxious chemical warfare, may force them to aggregate into fist-sized balls of thousands of brood and workers that rain out of the trees to the safety of the ground.29
Yet other, less quarrelsome canopy dwellers manage to survive in the territories of dominant ants by being overlooked or ignored. They creep out of view, run from trouble, or blend with the environment. Commonly referred to as nondominants, these ants may even compete for the same resources as the dominant or “extirpator” species. They may be opportunists, to use the entomological term, subject to attack by the dominant species but able to harvest food before the bullies drive them off. Others are insinuators, who rob meals from under the dominants’ noses by virtue of speed, stealth, or tiny size, in some cases even parasitically sneaking along the dominantants’ recruitment trails. In other cases, the insinuator is active at times of day when the dominant ant is not, or it may simply forage in a nonthreatening manner, for food the dominant species does not want or at sites that it cannot reach, such as the narrow furrows in bark that the speck-sized workers of the ant Carebara explore under the feet of weaver ants.
Most nondominant ants have societies of a few thousand or less, and often much less, as colonies of mere dozens thrive in any ready-made spaces they can find. The relative timidity of these small colonies parallels the behavior of small bands of human hunter-gatherers, which similarly lack basic infrastructure, with no entrenched dwelling places, territorial land, elaborate trail constructions, or stockpiled resources to protect. Full-bore warfare is unnecessarily risky for groups of this size: for human hunting bands, for example, most conflicts are small in scale and arise, as they do for many animals, over issues of power or reproduction.30 Given their mobility and lack of rivalry over land and resources, small groups are otherwise more likely to choose flight over fight—making nondominant ant species easy targets for domination.
Wherever weaver ants occur, they rule over the best sources of nectar and honeydew—those with the most amino acids and sugars. The subordinates remove whatever’s left over, at times sharing the inferior spoils among themselves equally or accessing them in a pecking order.31 Every once in a while one of the more tenacious of these subservient species has its moment of glory, taking over a swatch of the canopy when dominant species are absent.
The weavers’ control of the canopy is so extreme that in times of food scarcity, they will raid the nondominant species nesting within their territories to eat their larvae, in a sense using the contents of these colonies as if they were reserves of food.32 Subordinates could move away, but they may prefer to live with the enemy they know. They might also rely on weaver ants as a homeland security system, scaring off their usual competitors.
Canopy conditions encourage aggressive dominance by species like the weaver ant. For all their complexity, the living spaces in forest canopies are easier to control than areas of equivalent size on flat ground. Hill forts made it easier for people to fight approaching armies; ants similarly take advantage of the height and geometry of plants, which results in chokepoints that limit access to a territory, simplifying its defense. The borderlands that canopy ants guard are restricted; they consist mainly of vines, tree limbs, and the boundaries between tree trunks and the earth. If it were possible to squash flat all the trees occupied by a weaver ant colony, the surface area would be a pancake with a border hundreds of meters long, a frontier as imposing to patrol as the Mexican border is for the U.S. government. To control the equivalent area in the trees against even large armed forces, the ants need only employ a few expert fighters at a few bridgeheads.33 Their payoff in terms of arboreal land per military expenditure is thus vastly increased.
In 1993, I spent an afternoon at Guanacaste National Park, Costa Rica, watching the thin, wasp-like workers of Pseudomyrmex ants that resided in the hollow thorns of their ant plant, an acacia tree. The action was ongoing as ants killed or drove off caterpillars and other insects. But what especially irritated the workers was a vine touching one of the acacia twigs. They examined its looping tendril, then spent an hour pulling and shredding its tissues, at which point the vine fell away. Why so much attention to what for this predatory species must be inedible vegetable matter? The answer is simple: a vine can become an access point for invasion by neighboring colonies. This specialized form of clearing, which was so thorough that the ground around the base of their tree had been denuded of all other plants, served the acacia as well, if only coincidentally.34
A Pseudomyrmex worker in Guanacaste, Costa Rica, tearing the tendril of a vine that has touched her nesting acacia tree. If it were permitted to grow, the vine might overwhelm the tree.
Jorge, a Matsigenka Indian, standing in a spirit garden in Manu National Park, Peru. The undergrowth has been cleared away by Myrmelachista schumanni ants, whose aggressive attacks have warped the bases of the surviving trees.
It’s common for ants to clear the area around their ant plants, and those that do it the best are found in Peru’s forest glades, called spirit gardens by the local Matsigenka Indians, who believe that spirits clear the underbrush. I visited a half-acre spirit garden with an Indian named Jorge and Doug Yu, at the time a graduate student at Harvard studying the coevolution of ants and plants. Sweat bees, wasps, and killer bees, hungry for the salt in our perspiration, landed in such droves that we had to shout over the buzz and hold our arms out stiffly like Frankenstein’s monster to keep from being stung. The trunks of the few big trees were swollen like potbellies with malformed bark, which the Matsigenka ascribe to fires set by the spirits. Slicing a trunk with his knife, Jorge showed us the true cause of the deformity: tunnels eaten out by the minute Myrmelachista schumanni ant, which were killing the trees. Doug pointed out the ant plants nearby, small trees, easy to miss in the clearing, that shelter the brood and queens of this ant in their stems. The trees were doing well under the unobstructed, illuminated conditions provided by their ants. Sparing only these hosts, the workers will spray formic acid at any other plant, whether scrawny herb or mammoth tree; poisoned in this way, a sapling loses its leaves in five days, while the large trees barely get by or slowly expire. A spirit garden can contain three million ants and, judging from the slow rate of expansion of the garden space, last for eight centuries.35
The patchwork distribution of residents within the trees is called a mosaic. This distribution becomes most ecologically interesting when ants of different species live near—but not with—one another, separated typically by territorial friction, like that seen between weavers and the “exploding” ants.36 Ant mosaics are not universal. They are often indistinct in New World tropical forests, which can be more uniformly dominated by colonies of the same ant-garden ant species and some of the aggressive ants that control ant plants.37 And they are rare at extreme latitudes, where winters preclude occupation of the canopy year-round and most of the foraging in the treetops is done by ground-nesting ants.
Still, any environment where multiple ant species control exclusive territories can produce a mosaic that resembles a geopolitical map of Europe—even on the ground, despite the high defense costs. In America’s Southwest, the harvester ants Pogonomyrmex barbatus and P. rugosus control seed-rich swaths of desert flatlands in separate territories with the tenacity of gold miners guarding claims. Yet they ignore nondominant harvester ant species, which survive on the inferior seeds the dominant ants overlook. In the tropics, though, mosaics on terra firma tend to be hard to make out: most colonies there are small and packed together, with on average five nests per square meter.38 One reason may be the army ants, which are the dominant tropical ground-dwellers. Because of their concentrated, shifting raids, army ants don’t partake in mosaics. Rather, they move across the litter and soil like a hurricane, plowing through other colonies, giving less dominant ants an opportunity to move into the land they have cleared.39
What makes mosaics obvious in the canopy are the trees, whose crowns form discrete units without parallel on the ground. While different dominant ants at times control different portions of one crown, commonly each tree is a nation-state unto itself, owned by a colony. The territorial checkerboard of arboreal ant colonies is ever shifting, established by the ants’ history in much the way international borders define territorial stakes for us. Ranges expand and contract depending on conflicts with other colonies and the growth and death of trees. Competition may be somewhat reduced by the fact that dominants are often partial to particular plants, such as the weaver ant to citrus and the Azteca ants to Cecropia trees, or by the tendency of some ants to restrict themselves to certain layers of the canopy. Otherwise, though, the whole canopy is open for confrontations. Remove a dominant colony, and an adjacent colony of the same species may take over, an annexation simplified by the fact that its workers can exploit the previous tenant’s trails.40 Or a different dominant species might seize the space, starting a cascade of community shifts.41
These shifts occur because each dominant ant species alters its territory as an ecosystem engineer. This means that, as an outcome of their social skill set and incessant activity, the workers modify or create the environments in which they live and maintain that milieu thereafter, much as humans do in their societies.42 But each ant species is different and fosters the survival of different treetop residents, among them the insects it may tend for honeydew, while driving off its competitors and culling its prey.43 Many ants are similar to people who intentionally encourage the reproduction of some living things, like those we use for constructing our homes (such as trees), while unintentionally encouraging others, like the molds and vermin that consume our refuse. In such ways, the patchwork of ant species must profoundly enhance the diversity of a rainforest.
Terry Erwin estimates that a hectare of rainforest canopy contains a thousand trillion individuals belonging to a hundred thousand species, most of them invertebrates.44 Ant mosaics could be a factor in the forces that cause some of these species to evolve in the first place. On islands such as the Galápagos, populations evolve independently from others of their kind, diverging into separate species. Similarly, both tree crowns and ant territories can function much like islands for small, fast-breeding insects.45 But rather than being isolated like the Galápagos in an ocean of water, crowns and territories exist within a quiltlike sea of other crowns and territories, each combination of which is acceptable to a different degree to each canopy-dwelling insect.46 A beetle species, for example, may thrive where its food tree is occupied by a carpenter ant, but be killed where weaver ants live on the same tree species. Populations of that beetle will therefore come and go across the landscape as their ideal island habitats, defined by both ants and trees, change over the centuries. And so it must be for countless other canopy residents.