Pick from the Past
Natural History, November 1985

The Case of the Vanishing Caterpillar

A butterfly larva’s fate depends on who finds it first—its ant friends or ant foes.




Map by Joe LeMonnier
wenty-four years ago, I was a graduate student researching ecology in the Sierra de Tuxtla, an isolated volcanic range that rises from the Gulf coastal plain of Mexico, one hundred miles southeast of Veracruz. One remote habitat in the Sierra especially intrigued me because biologists had never noticed it. This was a small pine community found only on several of the steep ridges radiating down the southeastern slope of Volcán Santa Marta, a dormant, 5,250-foot-high volcano. There I discovered a new species of metalmark butterfly (lepidopterist Harry Clench of the Carnegie Museum honored me in 1962 by naming it Ross’s metalmark, Anatole rossi).

During the first month of my work in the tropics, when I began unraveling the metalmark’s life history, some of the particulars came easily. I rapidly learned that the butterflies live in colonies and that females lay their eggs only on a particular species of croton, a ground-cover plant common in the region. I even discovered a few young larvae, or caterpillars. But then my luck ran out. Even though I found many severely damaged croton plants—they had apparently been eaten by mature caterpillars—I couldn’t find a single large caterpillar or a chrysalis, the developmental stage ii which the caterpillar metamorphoses into a butterfly. Where were the culprits?

I spent my days crawling about on hands and knees, often in rough terrain, searching for large caterpillars. Not finding any, I became very discouraged and resorted to hatching butterfly eggs and rearing the young larvae myself. My combination home and laboratory was a thatch-roofed, mud-floored hut, provided by my hosts, John and Royce Lind, American missionary-linguists who were studying the local Popoluca Indian culture.

Reared in plastic sandwich-box nurseries, the young green larvae grew into large, plump caterpillars, shedding their skins six times in the process. After the second molt, each caterpillar began to secrete droplets of clear fluid from two openings on its back, near the hind end.


Two Ross’s metalmark caterpillars, accompanied by wood ants, emerge from the base of a croton plant.
Near its front end a pair of tentaclelike organs were periodically protruded and withdrawn. In addition, a pair of hard, bladelike structures, which constantly vibrated, projected over the caterpillar’s head. At the time I hadn’t a clue as to the functions of these organs, but my in-house experiments at least proved that the tiny larvae did grow into larger versions. Why couldn’t I find any in the field?

One day, as I sat on a log in the midst of one of my metalmark colonies, nibbling a peanut butter and jelly sandwich (my lunch each day for several months) and pondering the mysterious absence of caterpillars, my reverie was disturbed by motion on a croton plant just three feet from where I sat. I dropped to the ground and drew closer. There, to my surprise, was a fully mature caterpillar, which I recognized from my “nursery” research.

Swarming over it were several large, reddish black ants. My protective instinct moved me to rescue the caterpillar from its predicament, but then I noticed that the ants were not attacking the caterpillar, which was feeding docilely; they were probing it, caressing it with their antennae. The caterpillar was protruding its tentacles in response, while the ants climbed aboard and drank droplets of the clear liquid that it released at its hind end.

The caterpillar then began crawling down the plant, and within moments, it was in an underground chamber at the shrub’s base—something I had never seen before. The ants followed, some hitching rides on the caterpillar’s back. Once they were all inside the chamber, the ants closed it with dirt pellets, and all vestiges of insect infestation disappeared. My adrenalin soared! I quickly checked some partly defoliated crotons nearby. Adjacent to each plant was a subterranean niche containing from one to three caterpillars—large caterpillars—and ants; some of the chambers even contained chrysalises. Here was the solution to the mystery, and it had come about so unexpectedly.


One of two honeydew-secreting glands projects from a caterpillar’s hind end. An ant stroking the glands with its antennae will be rewarded with a sweet droplet.
Although I’ve spent many years observing butterflies, caterpillars, and ants, I have never again seen these ants and caterpillars above ground during the day. They come out from their hiding places only at night. Had it not been for’ my chance encounter with what I like to think of as a gluttonous caterpillar sneaking in an extra meal, I probably would have given up the project of completing the metalmark butterfly’s life history.

Following this revelation, I spent most of my hours—now at night, as well as during the day—on the pine ridges observing the caterpillars and their ant attendants. At night I used a flashlight covered with red cellophane, since my lantern, like any white light that the insects could interpret as dawn, disturbed the insects’ activity. Even my camera’s electronic flash had to be used sparingly and on intermittent nights in order to maintain nearly normal field conditions.

My nighttime excursions into the forest with camera, tripod, flashlight, and writing paraphernalia, sometimes during torrential thunderstorms, aroused the curiosity of the Popoluca villagers, whose large repertoire of folklore regarding “evils” of the night discourages nocturnal ventures. Within a short time I became the talk of the community, and speculation about my activities reached my American hosts. I learned that the most common theory about my wanderings was that I was searching for lost Spanish treasure, which the Popolucas believe to be buried in their area. At first, this reaction elicited only a chuckle from me, but then the situation became serious. One day I noticed that someone had been digging in one of my colonies and had actually destroyed several of my study plants and their underground inhabitants. Since the Popolucas were unaware of the insects, they paid them no heed as they dug where I had been crouching and peering the previous night. They only wanted to see what treasure I had found.

Following this disaster, my hosts and I met with several of the village leaders. We assured them I was only a “nature lover,” interested in simple things like ants and grubs on plants, and that if their digging continued, my project would be ruined and I would have to leave the area disappointed. Thereafter, I was not disturbed, although I continued to elicit strange glances from many Popolucas, especially the village witch doctor, who continued to spy on me from the nearby shadows.



Protuberances over the caterpillar’s head, left, shield two short sword-shaped beaters. These specialized appendages produce faint clicking sounds that attract wood ants.

In the years since my discovery of Ross’s metalmark, I have worked out its life history, which is probably the most complex of any known butterfly species. The butterfly is relatively small, with checkered brown, orange, white, and black markings. Like most other metalmarks around the world, the adults spend most of their time flitting rapidly two to five feet above the ground or clinging to the tops and bottoms of leaves, where they bask in the sun with their wings outstretched. In clear weather the butterflies are particularly visible late in the afternoon, when they dart about erratically and chase each other for short distances in the last rays of the setting sun.

The adult insects feed almost exclusively on the nectar from the white flowers of Croton repens, a wild relative of the more showy crotons raised indoors by many Americans and Europeans. The first indication that the butterfly’s life cycle is unusual is that, while the croton plants are widely distributed, the butterflies congregate in small, isolated colonies. These colonies, which contain eight to fifty butterflies, are found only on certain sunny ridge crests and upper slopes and only in the vicinity of the few human settlements in the area—five small Popoluca Indian villages.

Within the butterfly colonies, the female insects lay their eggs on the undersurfaces of croton leaves. Hidden on the underside of the leaves, the very young, inconspicuous green caterpillars spend their days and nights nibbling the plants’ tissues. Eight to ten days after hatching, a larva develops the specialized structures that I had noticed in my nursery. These enable it to attract its ant associates. A pair of glands near the larva’s hind end produce a sugary secretion known as honeydew, which ants eat with the gusto of children devouring candy. A variety of insects-aphids, many scale insects, and to a lesser degree, certain butterfly and moth caterpillars-produce similar honeydew secretions. The tentaclelike organs near the front of the caterpillar secrete a gaseous chemical pheromone, or scent, which also attracts the ants. And the caterpillar’s vibrating, bladelike structures produce faint clicking sounds, inaudible to humans but within the range of insect perception.



An ant grasps a globule of honeydew in its jaws.

Not long after these specialized organs become active in a newly molted larva, the caterpillar attracts its associates, ants of the species Camponotus abdominalis, commonly referred to as wood ants or carpenter ants because they nest in old logs, fence posts, and even house beams. This species can be found throughout most of tropical Mexico and Central America. At night these ants scour the countryside (and villagers’ kitchens) looking for, above all else, sweet things to eat. Needless to say, the Anatole caterpillars with their honeydew glands fill the bill.

When it first encounters a caterpillar during a nocturnal foray, the ant uses its antennae to stroke the caterpillar in the vicinity of the honeydew glands. The caterpillar responds by secreting tiny drops of moisture, which the ant imbibes. Usually within a few minutes, the ant worker is joined by several others of its kind, all of which consume the seemingly endless supply of treats. The wood ants do the caterpillar no harm, and after some thirty minutes of feasting, depart in relays to dig a small, shallow trench around the base of the croton. This excavation takes several hours but usually is complete by dawn. The resultant circular excavation is one to two inches deep. The ants then swarm about the caterpillar, constantly stroking it with their antennae. The caterpillar is induced to crawl down and soon finds itself underground. The ants quickly seal the entrance to the hole with small pellets of dirt from their previous digging. In a sense, the ant ranchers have herded their “cow” into its pen and closed the gate so that the milking becomes a private affair.

At dusk, between 7:00 and 7:10, the pellets of dirt are removed from the roof of the pen and the ants emerge, temporarily leaving their captive behind. They run back and forth over the leaves of the plant in a manner that suggests a search for potential enemies. To test if this was the case, on several occasions I placed small predatory arthropods, such as bugs and spiders, on the plants. The ants immediately seized these victims with their powerful mandibles,


In a subterranean chamber, a caterpillar attaches itself to a croton stem, sheds its skin, and becomes a chrysalis, or pupa. Although it produces no honeydew, the chrysalis will be tended by the ants for nine days of the eleven-day pupa stage.
sprayed them from their abdomens with formic acid—an acrid liquid that smells something like vinegar—and carried them down the plant, depositing the virtually lifeless forms on the ground a few inches away. (Incidentally, the Popolucan name for these ants means “the sour-smelling ones.” The villagers’ language is rich with descriptive names for most common plants and animals; however, it doesn’t contain a name for the elusive metalmark caterpillar.)

After about ten minutes of running over the plant, the ants return to the subterranean enclosure and vigorously stroke the caterpillar with their antennae, apparently coaxing it to move. The caterpillar now ascends the stem, shepherded by its carpenter ants, and eventually settles on a tender leaf near the top of the plant to begin feeding.

The feeding continues intermittently throughout the night, with the ants constantly in attendance, crawling over the caterpillar, pausing occasionally to imbibe droplets of honeydew, and running up and down the croton plant in search of marauders. The caterpillar, meanwhile, frequently emits its pheromone and vibrates its bladelike beaters, causing the ants to become more active and attentive. Between 4:30 and 4:45, just before dawn, the ants herd their charge downward and into the underground chamber. The roof is replaced, and even before the first streaks of light cross the tropical sky, both cow and herders are secure within their corral until the next twilight.

The daily cycle is rarely altered or interrupted during the summer and fall months. Even torrential rains do not have any observable deterrent effect. The underground pens are always located on sloping terrain of heavy red clay that provides excellent drainage, so the pens are rarely flooded. Besides, the insects are on the croton plants during the night, when most of the rains occur.

The caterpillars require nearly two months to complete their development. Then, like all butterfly and moth larvae, they enter a resting and transformational stage called the pupa. Unlike moth larvae, butterfly caterpillars do not encase themselves in cocoons. Instead, they attach their abdomens to stationary objects and their skins to reveal a new form, the chrysalis. Then, one of the greatest marvels of the insect world occurs: the lowly crawler is transformed into a winged gem of the skies.

The metamorphosis of Ross’s metalmark takes eleven days. The process begins with the Anatole chrysalis in its chamber, attached to the subterranean portion of the croton stem or to its root. Although the organism does not produce honeydew in this stage, it does possess glandular openings corresponding in position to the pheromone-producing organs of the caterpillar, and I conclude that a similar chemical is released into the air. At any rate, the ants remain in the pen for nine days—until two days before the butterfly’s emergence. Then, as the adult form nears perfection, its pheromone glands atrophy. The ants depart, leaving the pen open. The butterfly later emerges from the old pupal skin, climbs out of the darkness it has inhabited during the daylight hours for nearly two months, and enters the bright, sunny world of the pine ridges. After an hour or so of drying and exercising its newly formed wings, the butterfly departs for a free, aerial existence, which it enjoys for as long as a month, apparently oblivious to its past ant associates.



Transformed, the pupa becomes a butterfly.

Many questions arose in the course of my observations. Do the ants remain with their charge throughout its development or do they sometimes return to their nest site? What would happen if the caterpillars were isolated from their ants? Is the association between ants and caterpillars of benefit to both species?

I experimented. First, I carefully opened a pen, and taking out the ants one by one, I dabbed each on its back with a drop of quick-drying paint. I repeated this procedure with four nearby pens, using a different color for each. Three days later I observed that all of the painted ants had either been replaced by unpainted ones or else had moved to different chambers. There evidently was a changing of the guard as the ants circulated to and from their central nest and over their territory. Eventually I learned that the ants’ nest may be up to fifty feet away, its inhabitants exploiting a variety of resources-saps, flower nectars, and other sweets. The caterpillars seem to be just one source of food they encounter. The ants’ response-secreting the unwieldy prize in a hole instead of attempting to carry it back to the nest—may not have been specifically evolved to cope with the Anatole caterpillar. Similar behavior has been observed in several species of ants. The herding behavior may be a more specialized feature of the ants that evolved in association with these caterpillars.

In another experiment I removed the attending ants from several crotons and placed net cages around the plants to keep them out. When thus isolated from their caretakers, the caterpillars did not return to their subterranean cells after feeding but, instead, remained on the leaves and fed intermittently throughout the day and night. Evidently the daily journeys up and down the plants were forced on the caterpillars by the ants.

Then I removed several caterpillars and placed them on croton plants that were outside the butterfly-caterpillar colonies. Left unattended, the caterpillars were invariably attacked by workers of Ectatoma tuberculatum, large reddish ants festooned with numerous hairs and spines. This species is a member of a widespread group of ants, the Ponerinae, known for their fierce stinging ability and voracious predatory habits. The Popoluca’s name for this species translates as “the robbers,” appropriate for sinister-looking creatures that are frequently seen carrying assorted insect prey to their tree-hole nests.



A newly emerged Ross’s metalmark butterfly dries its wings.

I was personally well acquainted with Ectatoma. Just a few days prior to my relocation experiments, I had been stung on the thumb as I leaned against an oak tree. My finger was numb for nearly twentyfour hours and swollen for another day. I could imagine the damage such a sting would inflict on a caterpillar. I have since learned that, when stung, a larva undergoes extensive tissue destruction within a few minutes.

The relationship between the carpenter ants and the metalmark caterpillars thus appeared to be one in which the larvae furnished the ants with tasty sweets and in return received protection from a vicious armed predator, another ant species. To confirm this, I carefully investigated the pine ridge crests and upper slopes, exclusive habitats of the metalmark butterfly. The Ectatoma ants infested these open areas and spent the daylight hours solitarily stalking potential prey. At night, however, they remained secluded in their nests. The Anatole caterpillars and their wood ant guardians were prime targets, but they were hidden beneath the ground during the day when the predators were on patrol. At night, with the plundering halted as the Ectatoma ants stayed in their own nests, the caterpillars could feed in complete safety, exposed on their plants. Therefore, the relationship between the carpenter ants and metalmark caterpillars is mutually beneficial. Biologists call such a symbiotic relationship, in which both parties benefit, mutualism.

There is one other chapter to this story. During the cooler months of winter and early spring, the ant-caterpillar activities are somewhat altered. Between November and April, nighttime temperatures may drop to 50°. Adult butterflies are no longer on the wing, having died as the cool, damp weather set in. The ants deepen the pens into five- to six-inch vertical tunnels, probably to increase insulation from the cold air. The butterfly’s larval stage is greatly lengthened as the caterpillars and their attending ants become sluggish. Frequently a caterpillar will crawl up a plant during the early night hours, take a few bites, and then be herded immediately back into the pen. At times, particularly during very cool spells, the caterpillars do not emerge at all for many consecutive nights, presumably able to subsist on stored nutrients because of a reduced metabolism.

As spring, with its warmer temperatures, begins in the Sierra de Tuxtla, so does the dry season. Much of the undergrowth in the various ecological communities rapidly withers and dries.


The cycle is renewed as two metalmarks mate. The female will lay the eggs on the underside of a croton leaf. If the resultant caterpillars are not found and hidden by wood ants, they are frequently destroyed by stinging ants and other predatory insects.
At this time, humans become an important factor in the perpetuation of the metalmark’s life cycle. The Popoluca men toss lighted matches alongside the numerous trails on the pine ridges adjacent to their villages. The pine needles and dried undergrowth make excellent tinder, and the flames spread rapidly along and down the ridges. Because the ground is only sparsely covered, there is no widespread inferno, only low ground fires that creep in irregular patterns. The goal of this annual activity is to clear clutter from the ground in order to encourage the growth of fresh grasses that will serve as fodder for mules, burros, and horses. This allows the work animals to be pastured only short distances from the Popoluca villages.

By mid- or late April, selected sections of the pine ridges—and practically all butterfly-caterpillar colonies—have been burned over. However, because the wood ants previously deepened the pens during the cool months, the fires do not harm the secluded insects. They emerge the night after the firing and find a new world, one devoid of food plants and littered with burned debris. The absence of food for these winter-season caterpillars now triggers the onset of their long-delayed pupal period. No more than twelve days after fire passes through a colony, all the caterpillars transform into chrysalises. In two more weeks—by late April or early May—a fresh generation of butterflies is flying about.

Soon mating occurs and the females begin to lay their eggs. They select only the small shoots of croton plants that have recently sprouted from the bare, scorched patches of ground near their original larval food plants. The newly emerged cr0tons look much healthier than others growing in unburned localities. They are shorter, brighter green, and have relatively smooth leaves. The butterflies instinctively avoid placing their eggs on larger, more spindly plants with hairy leaves. While rearing caterpillars in my field laboratory, I observed that the young larvae could not penetrate the velvety surface of these larger plants; they soon became emaciated and died, presumably from starvation.

The distinctive growth form of the crotons in the metalmark colonies may be a response to the increased sunlight that pours into the burned-over areas. In unburned, grassy areas, even new-grown crotons produce hairier leaves. This difference, so important to the young caterpillars, remains in effect throughout the year. Thus, both the spring burning carried out by the villagers and the selectivity of the egg-laying butterflies insure colony stability and are critical to the survival of the species.

With the successful reproduction of the “spring crop” of butterflies, the cycle closes. Ross’s metalmark seems to reproduce exclusively in an environment that is hostile to most other defenseless insects. It lives and breeds in a tiny habitat—pine ridges close to Popoluca Indian villages—that is burned regularly and systematically by the indigenous population and is the hunting territory of a potentially devastating ant predator. The butterfly thrives, where others cannot, because of its pact with a specific wood ant—a pact that, if somehow broken, would almost surely spell the butterfly’s immediate extinction.



Ross in Mexico in the early 1960s

When he began his graduate work in biology at Louisiana State University in 1962, Gary Noel Ross went on a six-month expedition to investigate butterfly populations in Mexico’s relatively unknown Sierra de Tuxtla. Little did he realize that he would discover a new species in the process. This butterfly, named Ross’s metalmark in his honor, became the subject of his master’s


Ross in the early third millennium
thesis and is the basis of his article in this issue. His doctoral dissertation dealt with all of the region’s butterflies. Ross has also done fieldwork elsewhere in North and South America, Africa, and Southeast Asia.
     [Update 2006:] At the time this article was published, Ross was studying the natural dyes and biological motifs used in the weavings of the Zapotec Indian culture of Oaxaca, Mexico. That research became the subject of “The Bug in the Rug” (March 1986). Among his other articles that have appeared in
Natural History is “Butterfly Wrangling in Louisiana” (May, 1995), which won the John Burroughs Nature Essay Award. A resident of Baton Rouge, Louisiana, where he formerly was professor of biology at Southern University, Ross is now a research associate at the McGuire Center for Lepidoptera and Biodiversity at the University of Florida, Gainesville.

Copyright © Natural History Magazine, Inc.

Return to Web Site Archive, Picks from the Past