The Ants Chapter 17

CHAPTER 17. THE FUNGUS GROWERS

Introduction
Members of the myrmicine tribe Attini share with macrotermitine termites and certain wood-boring beetles the sophisticated habit of culturing and eating fungi. The Attini are a morphologically distinctive group limited to the New World, and most of the 12 genera and 190 species occur in the tropical portions of Mexico and Central and South America. Besides their unique behavior and the many peculiar behavioral and physiological changes associated with it, the Attini are distinguished from other ants by a distinctive combination of anatomical traits, including the shape of the antennal segments; a less-than-absolute tendency toward hard, spinose, or tuberculate bodies; and a proportionately large, casement-like first gastral segment.

It is conceivable that fungus growing originated only once in a single ancestral attine living in South America during that continent's long period of geological isolation from late Mesozoic times to approximately four million years ago. Exactly when the event occurred is open to conjecture, but it was almost certainly prior to the Miocene Epoch. Extinct but modern-looking species of Trachymyrmex (Baroni Urbani, 1980) and Cyphomyrmex (Wilson, 1985h) have been found in the Dominican amber, which is believed to be either late Oligocene or early Miocene in age.

In Africa, southern Asia, and other parts of the Old World tropics, the Attini are replaced by fungus-growing termites (Macrotermitinae), which in their turn do not occur in the New World. No one can be sure whether this complementary global pattern is due to a mutual preemption involving competitive exclusion of one group by another or whether it is simply one more accidental outcome reflecting the extreme rarity of the evolutionary origin of fungus gardening. The latter is more likely to be the case, meaning that if attines were to be introduced today into the range of the macrotermitines, or vice versa, the two kinds of insects could coexist with little interference. This is possible because attines utilize insect excrement and fresh plant material for the most part, while the macrotermitines use dead plant material. Also, fungus-growing ants forage above ground, often even in trees, while fungus-growing termites are primarily subterranean.

The Attini are an enormously successful group where they exist. One species, Trachymyrmex septentrionalis, ranges north to the Pine Barrens of New Jersey, while in the opposite direction several species of Acromyrmex penetrate to the cold temperate deserts of central Argentina. In the vast subtropical and tropical zones in between, attines are among the dominant ants. Many of the species gather pieces of fresh leaves and flowers to nourish the fungus gardens, and Atta and Acromyrmex rely on this source exclusively. Since they attack most kinds of vegetation, including crop plants, they are serious economic pests. The species of Atta in particular are among the scourges of tropical agriculture. They are familiar to local inhabitants as the wiwi in Nicaragua and British Honduras, the bibijagua in Cuba, the hormiga arriera in Mexico, the bachac in Trinidad, the bachaco in Venezuela, the saúva in Brazil, the cushi in Guyana, the coqui in Peru, and the leaf-cutting or parasol ant in most English-speaking countries, the last name alluding to the fact that an Atta worker holding a leaf fragment over its head gives the impression of carrying a parasol. The problems of agriculture in Atta country have been humorously epitomized in the following anecdote by V. Wolfgang von Hagen, in connection with his attempt to grow a vegetable garden in Belize:

My Indian servants, dusky, kinky-haired Miskito men, lamented all this work. It was useless, quoth a toothless elder, to plant anything but bananas or manioc, as the Wiwis were sure to cut off all the leaves. Without the slightest encouragement the Miskito Indians would launch forth on the tales of the ravages of the Wiwi Laca, but unswayed by the illustrations, like Pangloss I could only remark that all this was very well but let us cultivate our garden. In two weeks the carrots, the cabbages, the turnips were doing well. The carrots had unfurled their fernlike tops, the cabbage grew as if by magic. From our small palm-thatched house my wife and I cast admiring eyes over our jungle garden. Our minds called forth dishes of steaming vegetables to replace dehydrated greens and the inevitable beans and yucca. Even the toothless Miskito elder came by and admitted that white man's energy had overcome the lethargy of the Indian. Then the catastrophe fell upon us. We arose one morning and found our garden defoliated: every cabbage leaf was stripped, the naked stem was the only thing above the ground. Of the carrots nothing was seen. In the center of the garden, rising a foot in height, was a conical peak of earth, and about it were dry bits of earth, freshly excavated. Into a hole in the mound, ants, moving in quickened step, were carrying bits of our cabbage, tops of the carrots, the beans--in fact our entire garden was going down that hole. I could see the grinning face of the toothless Miskito Indian. The Wiwis had come.

In fact, the leaf-cutting ants of the genera Atta and Acromyrmex are the leading agricultural pests of Central and South America. They were preadapted for this role by their ability to use many plant species with the aid of their symbiotic fungi, which serve as a sort of ancillary digestive system. They also build up high population densities, such as 5 colonies per hectare in Atta vollenweideri and 28 per hectare in Atta capiguara, with each colony containing a million or more workers (Fowler et al., 1986a,b).

Leafcutters are the dominant herbivores of the Neotropics, consuming far more vegetation than any other group of animals of comparable taxonomic diversity, including mammals, homopterans, and lepidopterans. The amount of vegetation cut from tropical forests by Atta alone has been calculated on the basis of twelve studies to lie between 12 and 17 percent of leaf production (Cherrett, 1986). Grass-cutting species of Atta, which are distinguished from other members of the genus by their short, massive mandibles, are equally voracious. Each colony of Atta capiguara uses about 30 to 150 kilograms of dry matter each year; the figure of Atta vollenweideri is 90 to 250 kilograms per year. Atta capiguara reduces the commercial carrying capacity of pastureland, measured by the number of head of sustainable cattle, by as much as 10 percent (Fowler et al., 1986a).

Because of the catholicity of their diets, or rather the diets of their fungus, leafcutters have an extraordinarily diverse impact on agriculture. It includes the direct distribution of most kinds of crops, loss of land surface to the large nests (from 30 to 600 square meters per nest, when soil erosion is included), accidents caused to animals and agricultural machinery, and highway and other right-of-way damage from the excavation of the huge nests. Because of the variation in damage from one country to the next, the total loss due to the ants is impossible to calculate, but it is probably in the billions of dollars. Yet research on leafcutters remains relatively neglected. According to Cherrett (1986), by the early 1980s only 1250 articles had been written on Atta and Acromyrmex, as opposed to 10,000 on locusts.

Leafcutting ants have been important to the economy of Latin America throughout historical times. The early Portuguese colonists, who dubbed Brazil the kingdom of the ants, left behind such testaments to the saúva as the following: “If there is not much wine in this land it is because of ants which strip the leaves and fruit” (1587), “In a word, it is the worst scourge that farmers have” (1788), and “Either Brazil kills the saúva or the saúva will kill Brazil” (1822) (cited by Mariconi, 1970). Deep within their huge nests, able to multiply themselves many times each year, the leafcutters are nearly invulnerable to anything but massive poisoning.

Because so many species thrive in cleared land and secondary forests, leafcutters as a whole have benefitted by the advent of European civilization. The ubiquitous Atta cephalotes, for example, is specialized to live in forest gaps, and as a consequence it is able to invade subsistence farms and plantations from Mexico to Brazil (Cherrett and Peregrine, 1976). Prior to 1954 Atta capiguara around São Paulo was limited to a small savanna south of the city and had little or no economic impact in the area. When nearby forests were cleared for conversion to coffee plantations and then pastureland, the species spread rapidly and reached pest proportions. After Acromyrmex octospinosus was accidentally introduced into the West Indian island of Guadaloupe, shortly before 1954, it spread rapidly to become an important agricultural pest (Therrien et al., 1986). If any leafcutter ants, and especially Atta, were to be established in sub-Saharan Africa or some other part of the Old World tropics, the result might be an ecological catastrophe. The terrestrial ecosystems of these continents are unprepared for a herbivore with the resiliency and proficiency of these highly organized insects.

Yet in spite of the problems leafcutters cause, it would be a mistake to think of them as the uncompromising enemy of humankind. During millions of years of coevolution with their natural environment, they have become an integral part of the ecosystems of the New World tropics and warm temperate zones. They supplant to a large extent the populations of herbivorous mammals, which are relatively sparse through most of the New World tropics. They prune the vegetation, stimulate new plant growth, break down vegetable material rapidly, and turn and enrich the soil. In the tropical moist forests, Atta are major deep excavators of soil and stimulators of root growth (Haines, 1978). If leafcutters were to be extirpated, a profound readjustment of the structure of forests and grasslands would result, including the extinction of at least a few species of plants and animals. Such considerations have led Harold G. Fowler and his co-workers in Brazil (personal communication) to call for the protection of Atta robusta, a local forest-dwelling species in São Paulo State now endangered by rapid deforestation within its range.

Leafcutting ants are among the most advanced of all the social insects. They arose in the New World, most likely in South America when that continent was isolated from the remainder of the Western Hemisphere. During the past ten thousand years, a mere eyeblink in geological time, these insects have encountered the most advanced product of mammalian evolution from the Old World, Homo sapiens. Certain difficulties have arisen from this contact, with the great bulk of the losses occurring on the human side. In order to redress the balance, we need to learn a great deal more about the biology of our adversaries, paying particular attention to the weak points that undoubtedly occur in their complicated social systems. But the goal should be intelligent management of their populations and never their complete eradication. Our advantage--and responsibility--lies in the fact that we can think about these matters and they cannot.

Fungus culturing
What happens to the vegetation after the Atta workers have carried it down their holes is a fascinating story that has been worked out through many decades of research. Bates, in his book The Naturalist on the River Amazon (1863), suggested that the ants use the leaves “to thatch the domes which cover the entrances to their subterranean dwellings, thereby protecting from the deluging rains the young broods in the nests beneath.” Other early observers believed that the leaves are eaten or used to maintain a constant nest temperature by heat of fermentation. Thomas Belt was the first to surmise the far stranger truth. In The Naturalist in Nicaragua (1874) he described the garden chambers deep within the Atta nests as being “always about three parts filled with a speckled brown, flocculent, spongy-looking mass of a light and loosely connected substance. Throughout these masses were numerous ants belonging to the smallest division of the workers, and which do not engage in leaf-carrying. Along with them were pupae and larvae, not gathered together, but dispersed, apparently irregularly, throughout the flocculent mass. This mass, which I have called the ant-food, proved, on examination, to be composed of minutely subdivided pieces of leaves, withered to a brown colour, and overgrown and lightly connected together by a minute white fungus that ramified in every direction throughout it. . . That they do not eat the leaves themselves I convinced myself; for I found near the tenanted chambers deserted ones filled with the refuse particles of leaves that had been exhausted as manure for the fungus, and were now left, and served as food for larvae of Staphylinidae and other beetles.”

It was left to Alfred Möller (1893) to observe for the first time the actual eating of the fungi. He found that the tips of the hyphae produce peculiar spherical or ellipsoidal swellings (Figure 17-1) which are plucked and eaten. Möller called these objects “heads of Kohlrabi” because of their fancied resemblance to the vegetable. Later Wheeler relabeled them gongylidia (singular: gongylidium), and this name has stuck. A group of gongylidia, to complete the terminology, is sometimes referred to as a staphyla (plural: staphylae), while a piece of the peculiar morel-like fungus of Cyphomyrmex rimosus is called a bromatium. The gongylidial clusters of Atta and Acromyrmex, averaging about half a millimeter in diameter, were later observed to be eaten both by adult workers and larvae. The structures are rich in glycogen, in a form readily assimilated by the ants (Quinlan and Cherrett, 1979; Febvay and Kermarrec, 1983). Kermarrec et al. (1986) have described the gongylidium of the Acromyrmex octospinosus fungus as “a goat-skin bottle which has a thick wall covered with mucilage and is filled with a finely granulated mictoplasm that maintains its turgidity.” It is a tank “filled with glygogen, hydrolases, and viral particles.”  About 56 percent of the dry weight of the mycelium as a whole, or interconnected mass of hyphae, of the Atta colombica fungus is available in the form of soluble nutrients, which include 27% carbohydrates, 4.7% free amino acids, 13% protein-bound amino acids, and 0.2% ergosterol and other lipids. The carbohydrates include trehalose, mannitol, arabinitol, and glucose but no detectable polysaccharides (Martin et al., 1969a). Why the Acromyrmex fungus has abundant glycogen while the Atta fungus lacks it, if this reported difference actually exists, is not known.

Yet in spite of the problems leafcutters cause, it would be a mistake to think of them as the uncompromising enemy of humankind. During millions of years of coevolution with their natural environment, they have become an integral part of the ecosystems of the New World tropics and warm temperate zones. They supplant to a large extent the populations of herbivorous mammals, which are relatively sparse through most of the New World tropics. They prune the vegetation, stimulate new plant growth, break down vegetable material rapidly, and turn and enrich the soil. In the tropical moist forests, Atta are major deep excavators of soil and stimulators of root growth (Haines, 1978). If leafcutters were to be extirpated, a profound readjustment of the structure of forests and grasslands would result, including the extinction of at least a few species of plants and animals. Such considerations have led Harold G. Fowler and his co-workers in Brazil (personal communication) to call for the protection of Atta robusta, a local forest-dwelling species in São Paulo State now endangered by rapid deforestation within its range.

Leafcutting ants are among the most advanced of all the social insects. They arose in the New World, most likely in South America when that continent was isolated from the remainder of the Western Hemisphere. During the past ten thousand years, a mere eyeblink in geological time, these insects have encountered the most advanced product of mammalian evolution from the Old World, Homo sapiens. Certain difficulties have arisen from this contact, with the great bulk of the losses occurring on the human side. In order to redress the balance, we need to learn a great deal more about the biology of our adversaries, paying particular attention to the weak points that undoubtedly occur in their complicated social systems. But the goal should be intelligent management of their populations and never their complete eradication. Our advantage--and responsibility--lies in the fact that we can think about these matters and they cannot.

As fresh leaves and other plant cuttings are brought into the nest, they are subjected to a process of degradation before being inserted into the garden substratum. First the ants lick and cut them into pieces 1-2 mm in diameter. Then they chew the fragments along the edges until the pieces become wet and pulpy, sometimes adding a droplet of clear anal liquid to the surface. Then, using side to side movements of the fore tarsi, they carefully insert the fragments into the substratum. Finally, the ants pluck tufts of mycelia from other parts of the garden and plant them on newly formed portions of the substratum. A newly inserted single leaf section 1 mm in diameter receives up to ten such tufts in five minutes. The transplanted mycelia grow rapidly, as much as 13 µ in length per hour. Within 24 hours they cover most of the substratal surface.

Michael Martin and his co-workers discovered that Atta workers contribute digestive enzymes in the fecal droplets they deposit on the fungus, including a chitinase, an a-amylase, and three proteinases (Martin, 1970; Martin et al., 1973). Subsequently, Boyd and Martin (1975) showed that the proteinases originate in the fungus and pass unaltered through the digestive tract of the ants back to the fungus. The ants avoid digesting fungal enzymes by the simple expediency of not secreting any digestive enzymes on their own. Acromyrmex octospinosus also lacks proteinases, but these smaller leafcutters produce their own chitinases in the labial glands (Febvay and Kermarrec, 1986). The metabolic capabilities of the attine ants and their symbiotic fungi have yet to be worked out in detail, but it is at least clear that the ants have lost some key enzymes. They depend heavily on their symbionts for many of their nutrients, while the fungi in turn depend on the ants for care and the recycling of some of the enzymes.

Still another chapter in leafcutter biology began with the revelation, by Barrer and Cherrett, that Atta and Acromyrmex workers feed directly on plant sap. As much as a third of radioactivity in experimentally labeled leaves is absorbed directly by the ants as a result (Barrer and Cherrett, 1972; Littledyke and Cherrett, 1976). It might seem possible at first that the ants merely contribute the liquid to the fungus as added nutrient rather than assimilate it themselves. However, the sap must be crucial to the workers because, as Quinlan and Cherrett (1979) found, only 5 percent of their energy requirements are met by ingestion of juice of the fungal staphylae. In contrast, the larvae are able to subsist and grow entirely on the staphylae. These findings suggest that adult workers use only the juice of the staphylae, while the larvae use every part of the staphylae. The queen, to complete the story, is known to obtain at least a substantial part of her food from trophic eggs laid by workers and fed to her at frequent intervals.

In summary, the main properties of the leafcutter-fungus symbiosis can be stated as follows. Adult ants are fundamentally nectar feeders, predators, and scavengers. Their entire digestive system, from their peculiar infrabuccal and proventricular filters to the delicate midgut and limited spectrum of digestive enzymes, is geared to this dietary commitment. They are ill-suited to be herbivores. The fungus, in exchange for protection and cultivation, digests the cellulose and other plant products normally inaccessible to leafcutters and shares part of the assimilable metabolic products with them. In the case of the “lower” attines, which do not cut leaves but use insect remains and excrement, the fungus converts the chitin and other products otherwise less available to ants.

Curiously, the single outstanding problem of attine biology all along, the identity and biological qualities of the symbiotic fungus, remains wrapped in mystery. The principal difficulty has been the reluctance--indeed, the near inability--of the fungus to form sporophores, the elaborate fruiting structures required for taxonomic diagnosis. Evidently the ants do not permit the fungi to form the mushrooms or other spore-bearing bodies under natural conditions. Instead the ants feed exclusively on the special gongylidial tips of the elementary mycelial clusters, a preference that appears to have resulted in the loss of the ability on the part of the fungus to produce sporophores. Reciprocally, the fungi utilize the ants for transport and do not have to depend on windborne spores to transfer themselves from nest to nest. Although Moeller did not clarify this problem in Atta, he was lucky enough to discover sporophores growing from abandoned Acromyrmex nests on four separate occasions. These proved to be agaracine mushrooms, wine-red in color, which Moeller formally named Rozites gongylophora. Mycologists have since confirmed their placement in the basidiomycete family Agaricaceae, but transferred the species gongylophora to the genus Leucocoprinus (Heim, 1957; Kermarrec et al., 1986). Subsequent attempts by entomologists to locate sporophores in abandoned attine nests and to culture them in the laboratory from the gardens of various attine genera have rarely succeeded. The most notable single advance was Weber's (1957b) use of a medium of sterile oats to rear sporophores of an apparent Leucocoprinus (= Lepiota) from mycelia originating from a Cyphomyrmex costatus garden. If future mycologists ever succeed in isolating a plant hormone or nutrient combination that enhances sporophore formation in fungi, dramatic further progress can be expected in this field.

The current evidence overall seems to support Roger Heim's opinion that the symbiotic fungus cultivated by all the attine ants is Leucocoprinus gongylophorus. The identification of this species or at least a set of closely similar forms placed variously in the basidiomycete genera Leucocoprinus, Lepiota, and Rozites has been confirmed by the rearing of sporophores from the garden mycelia in the attine genera Atta, Cyphomyrmex, and Myrmicocrypta. These ants represent almost the entire phylogenetic spread of the tribe Attini. According to Heim, it is unlikely that different attines picked up various leucocoprines here and there in the course of their evolution. In the absence of opposing strong evidence, this parsimonious hypothesis seems preferable to that of Weber (1979), who preferred to place the fungi of the lower attines (such as Cyphomyrmex and Myrmicocrypta) in a separate genus, Lepiota. Lehmann (1975, 1976) offered a third, truly radical opinion, that the attine fungus is not a basidiomycete at all but an ascomycete in the genus Aspergillus close to the symbiont of the fungus-growing beetles and Old World fungus-growing termites. This conclusion is probably too parsimonious. It is based on several tenuous morphological comparisons, including the supposedly primitive ascomycete appearance of the gongylidial swellings. It also flies in the face of contrary evidence based on sporophores cultured from attine mycelia. Weber (1979) has in fact identified an unusual Aspergillus in abnormal gardens of Atta and Acromyrmex, but it was strongly avoided by the ants and appeared to be a contaminant of unusually wet gardens. The only clear exception to the strict conformity of attines to Leucocoprinus or a closely related group of leucocoprine genera is the cultivation of a yeast by Cyphomyrmex rimosus (Wheeler, 1907b; Weber, 1979).

This last example may prove to be the tip of an iceberg in the study of the secondary microflora of the attine gardens. While it is true that the ant cultures are dominated by a single fungus species, microorganisms also exist and might even participate in the symbiosis. Recent research to this end, reviewed by Kermarrec et al. (1986), has revealed that both yeasts and bacteria are in fact present in substantial numbers. The metabolic activity of these microorganisms remains uncertain, however. It is possible that bacteria assist in the lysis of cellulose into products that are more readily utilized by the fungi. At least six species of Bacillus have been identified inside nests of Atta laevigata, and the ants appear to inoculate fresh vegetation with these microorganisms during preparation of the substrate. On the other hand, it is equally possible that the bacteria parasitize the ant-fungus symbiosis, draining away some of the energy that would otherwise flow directly through the fungus to the ants. The parasitism hypothesis gains credence from a finding by Kermarrec and his co-workers that the symbiotic fungus of Atta and Acromyrmex secrete substances antagonistic to bacteria and other kinds of fungi.

Upon reflection it is impressive how nearly pure the ants keep the fungal growth in their nest chambers. They build this monoculture by a variety of techniques: the plucking out of alien fungi, the frequent inoculation of the Leucocoprinus mycelia onto fresh substrate, the manuring of the substrate with enzymes and nutrients to which the Leucocoprinus are especially adapted, the production of antibiotics to depress competing fungi and microorganisms, and the production of growth hormones. These last two methods entail an instinctive form of chemical engineering on the part of the ants. Maschwitz et al. (1970) and Schildknecht and Koob (1970) identified phenylacetic acid, D-3-hydroxydecanoic acid (“myrmicacin”), and indoleacetic acid in the secretions of the metapleural glands of Atta sexdens workers. They suggested that these compounds play different roles in the purification of the symbiotic fungus culture: phenylacetic acid suppresses bacterial growth, D-3-hydroxydecanoic acid inhibits the germination of spores of alien fungi, and indoleacetic acid, a plant hormone, stimulates mycelial growth. As pointed out by Weber (1982), this interpretation can be confirmed only by a demonstration that the components of the metapleural gland are actually present in the fungus gardens at bacteriostatic and fungistatic levels (Weber, 1982).

The life cycle of leafcutter ants
Leafcutting ants comprise 24 known species of Acromyrmex (Table 17-1) and 15 of Atta (Table 17-2). Because the Atta workers are so large and spectacular in their behavior, many entomologists have set out to study their life cycle and biology. They include Moeller, the pioneer in the subject, Forel, Goeldi, Huber, von Ihering, and Wheeler, all of whose publications are exhaustively reviewed in the classic 1907 study of the North American Attini by Wheeler. More recent researchers have included Autuori, Bitancourt, Bonetto, Borgmeier, Eidmann, Fowler, Geijskes, Gonçalves, Jacoby, Kerr, Moser, Stahel, Weber, and others; their work is carefully reviewed in Weber (1972, 1982) and the symposium volume Fire ants and leaf-cutting ants edited by Lofgren and Vander Meer (1986).

All of the Atta species appear to have basically the same colony life cycle. The nuptial flights of some species, such as the infamous sexdens of South America, take place in the afternoon, while texana of the southern United States and a few others hold their flights at night (Autuori, 1956; Moser, 1967a). Because the ponderous females work their way high into the air before the males approach them, actual matings have not been observed. Nevertheless, Kerr (1962), by counting sperm from the spermathecae of four newly mated sexdens queens with the aid of a hemocytometer, was able to show that each individual is inseminated by at least three to eight males. The actual estimated numbers of sperm varied among the queens he examined from 206 to 320 million, seemingly more than enough to last an individual the ten or more years speculated to be the normal life span of an Atta queen.

During the nuptial flight and immediately afterward, as the queens attempt to start new colonies, mortality is extremely high. Out of 13,300 Atta capiguara founding colonies in Brazil, only 12 were alive three months later (Fowler et al., 1986b). From a start of 3,558 incipient Atta sexdens rubropilosa colonies, only 90 or 2.5 percent were alive after three months (Autuori, 1950). The survivorship of rubropilosa during the same time interval was 6.6 percent (Jacoby, 1944), while figures of 10 percent were obtained for Atta cephalotes and zero percent for Atta capiguara in Central America and Brazil respectively (Fowler et al., 1986b).

In 1898 von Ihering made the important discovery of how the fungus is transferred from nest to nest. Prior to departing on the nuptial flight the Atta sexdens queen packs a small wad of mycelia into her infrabuccal chamber,a cavity located (in all ants including Atta) beneath the opening of the esophagus just to the rear of the base of the labium. Following the nuptial flight, which in Brazil may occur anytime from the end of October to the middle of December, the queen casts off her wings and quickly excavates a little nest in the soil. When finished, the nest consists of a narrow entrance gallery, 12-15 mm in diameter, which descends 20-30 cm to a single room 6 cm long and somewhat less in height. Onto the floor of this room, according to Jakob Huber (1905) and Autuori (1956), the queen now spits out the mycelial wad. By the third day fresh mycelia have begun to grow rapidly in all directions, and the queen has laid the first 3 to 6 eggs. In the beginning the eggs and little fungus garden are kept apart, but by the end of the second week, when more than 20 eggs are present and the fungal mass is ten times its original size, the two are brought together. At the end of the first month the brood, now consisting of eggs, larvae, and possibly pupae as well, is embedded in the center of a mat of proliferating fungi. The first adult workers emerge sometimes between 40 and 60 days. During all this time the queen cultivates the little fungus garden herself. At intervals of an hour or so she tears out a small fragment of the garden, bends her abdomen forward between her legs, touches the fragment to the tip of the abdomen, and deposits a clear yellowish or brownish droplet of fecal liquid onto it (Figure 17-2). Then she carefully places the mycelial fragment back into the garden. Although the Atta sexdens queen does not sacrifice her own eggs as a culture medium, she does consume 90 percent of the eggs herself, and, when the larvae first hatch, they are fed with eggs thrust directly into their mouths. The queen apparently never consumes any of the growing fungus during the rearing of the first brood. Instead, she subsists entirely on her own catabolizing fat body and wing muscles. Soon after the first workers appear, they begin to feed themselves on the gongylidia. They also manure the fungal garden with their fecal emissions and feed their sister larvae with eggs laid by the mother queen. The eggs given to the larvae are larger than those permitted to hatch; a histological study by Bazire-Benazet (1957) has shown that they are in fact “omelets” formed in the oviducts by the fusion of two or more distinct but ill-formed eggs. After about a week the new workers dig their way up through the clogged entrance canal and start foraging on the ground in the immediate vicinity of the nest. Bits of leaves are brought in, chewed into pulp, and kneaded into the fungus garden. About this time the queen ceases attending both brood and garden. She turns into a virtual egg-laying machine, in which state she remains for the rest of her life. Now for the first time the workers begin to collect gongylidia from the fungal mass and to feed them directly to the larvae.

The growth of the colony is at first very slow. During the second and third years it accelerates quickly and then tapers off as the colony starts to produce winged males and queens. Using data provided by Autuori, Bitancourt (1941) demonstrated that the growth of an Atta sexdens colony, if measured as the increasing number of nest entrances, closely fits the classic formula of logistic growth. This means that the rate of growth can be expressed as an elementary function of the population size times the difference between the population size at the given moment and the size finally reached by the colony. The essential qualities of colony growth, as they are understood at the present time, are illustrated in Figure 17-3.

The ultimate size reached by the Atta nests is enormous. Autuori's nest contained slightly over 1,000 entrance holes at the end of the third year. Another three-year-old nest excavated by Autuori (cited in Weber, 1966) contained 1,027 chambers, of which 390 were occupied by fungus gardens and ants. Although in only its first year of production of sexual forms, the colony had generated no fewer than 38,481 males and 5,339 virgin queens. Still another Atta sexdens nest, 77 months old, contained 1,920 chambers of which 248 were occupied by fungus gardens and ants. The loose soil that had been brought out and piled on the ground by the ants during the excavation of their nest was shoveled off and measured. It occupied 22.72 m3 and weighed approximately 40,000 kg. Autuori also estimated that during the short life of the colony the workers had gathered no less than 5,892 kg of leaves to cultivate their fungus gardens! The garden substrate of an average Atta vollenweideri nest, according to Jonkman (1980b), is built up of 182 million pieces of grass.

As these numbers suggest, the populations of old colonies of Atta are metropolitan in size. In publications over many years summarized by Fowler et al. (1986b), the numbers of workers in single colonies have been estimated as one to 2.5 million in Atta colombica, 3.5 million in Atta laevigata, 5 to 8 million in Atta sexdens rubropilosa, and 4 to 7 million in [[Atta vollenweideri''.

The nests of mature colonies are also structures of extraordinary expanse and complexity, as documented in the studies of Eidmann (1935), Stahel and Geijskes (1939), Jacoby (1937, 1944), Moser (1963), and Jonkman (1980b). In well-drained soil the deepest galleries usually penetrate to more than 3 m below the surface, and in some cases they descend to more than 6 m. Their excavation for scientific study requires teams of laborers or, as in Moser's work on Atta texana in Louisiana, the use of a bulldozer. Stahel and Geijskes systematically observed the movement of small puffs of smoke released over various of the nest entrances; they were thereby able to demonstrate the existence of a primitive ventilation system in the intact nests. Air, it was found, tends to pour into those nest openings located near the nest perimeter and to pass up out of the openings located closer to the nest center. The intake and exhaust openings are about equally numerous. A third kind of opening, through which no movement of air can be detected, is even more common. There is a simple enough explanation for this pattern. Air is heated by metabolism more rapidly in the central zone of the nest, where the fungus gardens and ants are concentrated, and it therefore tends to rise through the central galleries. The movement in turn draws air from the remaining galleries, which are located in the peripheral zones. The “neutral” nest openings probably lead to blocked galleries or gallery systems with relatively few ants and fungus gardens. Thus the construction of large numbers of nest entrances in the Atta nests--a feature shared with only a few other kinds of ants--appears to be an adaptation to facilitate ventilation through the exceptionally large biomasses of the leafcutter colonies (see Figure 17-4).

Host selection
The first impression obtained in the field and laboratory is that Atta and Acromyrmex are indiscriminate in their choice of vegetation to serve as the fungal substrate. Under various circumstances many species accept fresh leaves, flowers, fruits, tubers, and stems of plants, as well as the endosperm of seeds. Colonies can be kept in laboratories indefinitely with a mix of processed cereals, such as rolled oats, and fresh leaves. Given a reasonable choice, however, the ants are moderately to strongly selective. Some are specialists on grasses. Examples include Atta capiguara, Atta vollenweideri, and Acromyrmex landolti. Others, including Atta mexicana and Acromyrmex rugosus, are specialists on dicots, while a few others, including Atta laevigata and Acromyrmex lobicornis, take both grasses and dicots (Fowler et al., 1986a). Of dicot harvesters and generalist species in Paraguay Schade (1973) wrote, “I have seen large orange trees, full of semi-ripened fruit, completely denuded in one night. The fruit, which the ants did not touch, soon fell to the ground after having been sunburned for lack of shade. The ants seem to prefer some cultivated plants more than others: citrus trees, roses, violets, medlar trees, onions, carrots, strawberries, alfalfa, and peanuts. Somewhat less popular are avocado trees, mandioca (Manihot utilissima), peach trees, guava trees, mulberry leaves and fruits, privet leaves, flamboyant trees (Delonix regia), and many other trees, shrubs and plants of agricultural or ornamental value. Corn and beans are sampled somewhat less frequently. Members of the Compositae, Solanaceae, and Euphorbiaceae, especially the tallowtree (Sapium spp.), are frequently attacked. The castor-oil plant, Ricinus communis, a euphorb, is not touched.” In the Florencia Norte Forest of Costa Rica, Blanton and Ewel (1985) found that the dominant leafcutter, Atta cephalotes, attacked only 17 of 332 available plant species. They cut proportionately more woody than herbaceous species, more introduced species than natives, and a higher proportion of species with below-average water content. Comparable data concerning the selection of host plants was obtained for Atta cephalotes in Guyana by Cherrett (1968) and Atta colombica in Costa Rica by Rockwood (1976). Within the favored plant species, leafcutters prefer freshly spouted shoots, leaves, and flowers. When preferred species decline in abundance, the colonies switch to less favored ones. For example, Acromyrmex versicolor foragers in the deserts of Arizona utilize freshly growing stems and leaves of dicots when they are available in the wet season but switch almost exclusively to grasses during the dry season (Gamboa, 1975). A similar seasonality in the choice of dicots is evident in detailed studies of Atta texana by Waller (1986).

To reach the plants of choice the workers follow trunk trails that commonly stretch for more than a hundred meters from the nest. The record, cited for a colony of Atta cephalotes by Lewis et al. (1974a), is 250 meters. Successful scouts of Atta recruit nestmates with a powerful pheromone, 4-methylpyrrole-2-carboxylate, which they expel from the poison gland through the sting. Other, still unidentified substances provide long-term orientation along the odor trails (see Chapter 7). Many of the Atta species clear broad highways along which their dense columns can travel unhindered. These “attian ways” are among the most conspicuous sights in the New World tropics.

What is the basis of host selection by the ants? In studies of the feeding behavior of Atta and Acromyrmex, Cherrett and Seaforth (1968) detected a wide range of plant phagostimulants, consisting principally of unidentified sapids and lipids. Such substances are unlikely to provide the sole basis of discrimination, however, because they occur widely and are generally efficacious as food stuffs. Equally important, Howard (1987) has shown that Atta cephalotes is little influenced by energy content, moisture, or amounts of nitrogen. The more likely basis of selectivity is the occurrence of greater concentrations of repellent substances in some plant species over others. Hubbell and Wiemer (1983) and Howard and Wiemer (1986) have begun the important task of screening and identifying these compounds in the plants rejected by leafcutters. Their technique is to allow workers to forage through a random checkerboard array of rye flakes treated with extracts and synthetic compounds. Virtually all of the repellent substances discovered by this means have turned out to be terpenoids, including a great diversity of mono-, sesqui-, di-, and triterpenoids. A key question is: are the ants avoiding these substances because they are toxic to the foragers and substrate processors who drink the sap, or because they are poisonous to the symbiotic fungus? The latter effect may prove to be crucial, because many terpenoids have strong fungicidal activity. In a recent study, Howard et al. (1988) provided the first experimental evidence that three out of four terpenoid substances tested exhibit deleterious effects either on adult Atta cephalotes workers or their fungus. This study also indicated some correlation between deterrent ability and toxicity. A potential goal of future research is to establish the mode of action and relative effectiveness of these substances and their distribution through space and time in the thousands of plant species with which leafcutters regularly interact.

Another question of broad ecological interest is whether the leafcutter ants husband their resources by directing their attacks so as not to kill off too many plants close to home. Foragers have often been observed to shift their attentions from one tree to another without denuding any one of them. Columns often travel past intact food plants close to the nest to attack others far away. These facts led Cherrett (1968), to speculate that the ants sacrifice energetic efficiency in order to protect the host plants and thereby to gain a longer sustained yield. However, this intriguing and perfectly logical idea has been cast into some doubt by the recent findings on repellent substances. The worker ants may simply be “shopping” among plants in order to locate those with the fewest toxic vegetation. The trunk trails, as Shepherd (1982) pointed out, tend to be laid to the temporarily most productive sites and to be changed around from time to time in a way that provides a high yield throughout the lifespan of the colony.

The origin of the Attini
The phylogenetic origin of the Attini remains a source of bafflement in spite of a century of speculation on the subject. One authoritative opinion was offered by Emery (1895), who, on morphological evidence, placed the Attini near Ochetomyrmex and Wasmannia. These taxa, together with the aberrant genus Blepharidatta, comprise the tribe Ochetomyrmecini (Brown, 1953b). The ochetomyrmecines are exclusively Neotropical, which is at least consistent with the hypothesis of some kind of evolutionary link to the Attini. The overall morphological resemblance between the two tribes is not at all close, however, and in fact the Attini stand well apart from almost all other ants in their morphology. Forel (1902) offered the contrary opinion that the Attini stemmed from the Dacetini, which in the old, broad sense included the tribes Basicerotini and Stegomyrmecini. However, the larvae differ in morphology, and the most primitive known dacetines are the genera Daceton and Orectognathus, which forage aboveground and on vegetation in a way that distinguishes them from the primitive attines, which are soil-dwelling.

A much more likely candidate for an ancestral or cognate taxon among living ants is Proatta butteli of tropical Asia (Figure 17-5). The adults closely resemble some of the small attines, especially Mycocepurus, in their overall body form and the distinctive spines and tubercles that cover most of the head and body surfaces. When Forel (1912) originally described the genus, he erected a new tribe, the Proattini, to receive it. Emery (1922), in the authoritative Genera Insectorum, transferred Proatta to the Attini, reducing it to the rank of subtribe and implicitly recognizing its close relationship to the fungus growers. But most myrmecologists did not accept this placement. They believed that Proatta is not a true attine, attributing its outward shared traits with the fungus-growers to convergent evolution. A whole new twist was then added when Wheeler and Wheeler (1985a) were able to study Proatta larvae, which had been collected for the first time by Mark Moffett in Singapore and Malaysia. These authors concluded that “the larva of Proatta is definitely attine. We have a prejudice against attaching a small monotypic genus found locally in the Oriental Realm to a large widespread tribe in the Neotropical Realm; hence we had hoped that the larva would be either strongly attine or strongly non-attine. It is neither, but it is as good an attine as Myrmicocrypta. It lacks the coarse pinules on the mandibles, which is an attine character, but so does Apterostigma, which is otherwise like the higher attines.” Moffett (1986d) found that Proatta butteli workers hunt small prey and scavenge for arthropod corpses. They also capture arthropods larger than themselves by a combination of rapid recruitment and group retrieval. However, most importantly, they neither grow nor feed on fungi. Perhaps it is a disappointment to learn that Proatta is not a fungus grower, but, as George C. and Jeanette Wheeler remarked, “is it really necessary that the ancestral attine already have that habit?”

The peculiar position of Proatta leads us to the question of the evolutionary beginnings of fungus gardening. There are three competing hypotheses. The first, due to von Ihering (1898), proposes that attines originated from harvesting ants with slovenly habits: “We know quite a number of ants, like the species of Pheidole, Pogonomyrmex and furthermore species of Aphaenogaster and even of Lasius, which carry in grains and seeds to be stored as food. Such grain carried in while still unripe, would necessarily mould and the ants feeding upon it would eat portions of the fungus. In doing this they might easily come to prefer the fungi to the seeds. If Atta lundi still garners grass seeds and in even greater than the natural proportion to the grass blades, this can only be regarded as a custom which has survived from a previous cultural stage.” In opposition, Forel (1902) suggested that the ancestral attines lived in rotting wood and gradually acquired the habit of eating the fungi they chanced to find growing on insect excrement left behind by woodboring insects. A slight variant of this idea was offered by Weber (1956), who believed that the ants might have begun feeding on fungi which grew from their own feces. The third hypothesis, due to Garling (1979), is that the attine fungus arose from the fungi that live in mycorrhizal symbiosis with plant roots. She noted that most ectomycorrhizal fungi belong to the same group as the ant fungus, the Agaricales. The ant symbiosis could have arisen by repeated encounters that must have occurred between soil-dwelling ants and the ectomycorrhizal fungi living on the roots around their nests.

Something like von Ihering's slovenly-ant hypothesis finds support in one of the findings by Moffett on Proatta butteli. The colonies he studied accumulated substantial amounts of prey remains and other inedible refuse within their nest chambers, and this material formed substrate on which wild fungus grows profusely. If the ancestor of the Attini had a similar tendency to keep refuse in the nest--an unusual but not unique habit in ants generally--fungus gardening might have arisen when the ants began to feed on the mycelia taking root there. As shown in Table 17-3, most of the genera of small attines, which are thought to be relatively primitive among living species, use insect remains and other detritus as fungus substrates. But some also use insect excrement, a circumstance consistent with Forel's hypothesis.

The genera are listed in an order that reflects the idea, held by most students of the Attini since the time of Emery and Forel, that Cyphomyrmex is primitive, Atta is advanced, and the remaining genera occupy positions of varying degrees of intermediacy. Of course such a vertical array is bound to be an oversimplification, because the evolution of the Attini, like that of almost all other large animal groups whose histories are better known from the fossil record, almost certainly unfolded in a more complex, dendritic pattern. But the principal evolutionary trends do seem clear enough when considered separately, and they are at least loosely interconsistent. There is a gradual increase in body size and, in a few of the largest species, the appearance of well-marked worker polymorphism. The body develops certain unusual anatomical features such as tuberculation of the body surface, unusual hair structure, and cordate head shape. The mature colony size increases from small (that is, a few tens or hundreds of individuals) through medium (hundreds or thousands) to large (tens of thousands to millions), with a corresponding growth in the size and complexity of the nest structure.

Now if these trends do reflect a true evolutionary history, it is reasonable to suppose that feeding behavior also evolved in roughly the same direction, namely from Cyphomyrmex to Atta and the other, “higher” attine genera. And if that much is granted, we can regard the use of nest refuse, including discarded arthropod remains, and insect feces as the culturing medium to be the primitive trait and the use of fresh vegetation to be the derived trait. It is therefore likely that a closer examination of the biology of Cyphomyrmex, along with that of the other presumably primitive attines and perhaps also of Proatta, will offer new light on the origin of the Attini and the fungus-culturing habit.

Whatever the beginnings of fungus growing, the event of greatest importance in the history of the Attini was the efficient utilization of all forms of fresh vegetation by the true leafcutter genera Acromyrmex and Atta. The achievement is unique within the entire Animal Kingdom. A close examination of the processing of the vegetation by Atta cephalotes and Atta sexdens reveals at least some of the reasons that fungus culturing of this particular kind is so rare (Wilson 1980a,b, 1983a,b). First of all, the ants must be relatively large. Workers of the two Atta species with head widths below 1.4 mm have difficulty cutting even the softest leaves and petals (the energetically most efficient size is 2.2 mm). Second, the workers have to be polymorphic. Individuals larger than about 1.2 mm are evidently unable to care for the minute fungi within the nest. Consequently, the cultivation of the fungus entails a remarkable assembly-line operation, as follows. The medias (head width mode 2.2 mm) cut and retrieve the vegetation, smaller medias (1.6 mm) slice it into smaller pieces, still smaller ones (1.4 mm) degrade the pieces into small lumps, and then successively smaller minor workers (1.2 mm to 0.8 mm) place the lumps in the substrate, implant strands of fungus on fresh substrate, and care for the fungus as it proliferates (see Figures 8-28 to 8-30). In addition, the ants use special procedures such as the recycling of chitinases and proteinases from the fungi (Boyd and Martin, 1975). Overall, Acromyrmex and Atta have traveled a long path in evolution by mastering the technique of gardening and then shifting to a substitute of fresh vegetation. Because they depend on a fungus to accomplish much of their initial digestion and by this means are able to by-pass the formidable array of terpenoids, alkaloids, and other defensive chemicals that deter most insect herbivores, the leafcutters have been able to exploit a very wide range of food plants, including most of the crop species grown in tropical regions.

Ant-fungus symbioses outside the Attini
A unique mutualistic symbiosis occurs between the European formicine ant Lasius fuliginosus, the “shining black ant” of some English-language literature or “glänzend schwarze Holzameise” in some German writings, and the ascomycete fungus Cladosporium myrmecophilum. The fungus grows exclusively in the walls of the Lasius carton nests, reinforcing them structurally. The ants exhibit specialized behaviors apparently directed at the cultivation of the fungus, and they transport inocula from one nest to another.

Colonies of Lasius fuliginosus transform large cavities in the soil and tree trunks by filling them with carton nests, whose internal structure is partitioned and resembles a sponge (see Figure 17-6). Maschwitz and Hölldobler (1970) found that the carton consists of particles of wood, dry vegetable material, and soil glued together with sugary secretions collected by the ants from aphids and other homopteran insects. The fungal mycelium grows through the walls of the carton and reinforces them in the same way that steel mesh or rods reinforce the walls of buildings.

The ants build their distinctive nest with a remarkable division of labor. Four castes based on age are employed. Workers of the first group, which are evidently older, collect the solid particles and carry them into the nest cavity, where they deposit them on the edge of the carton structure. The second group, also older, simultaneously collect homopteran honeydew and carry the liquid into the nest in their crops. Inside the nest they regurgitate it to nestmates who spend most of their time in nest construction and brood care. This third group of workers collect the solid particles from the edge of the carton and carry them to the construction site. There they regurgitate the sugary material onto the particles. They place the soaked particles onto the edge of the carton wall. During this process the workers often knead the fresh material with their mandibles while continuously touching the carton with their antennae and forelegs. The construction workers usually gather at specific sites where they line up along the edge of the wall in rows four to five centimeters long. Other ants, evidently constituting a fourth labor group, remove old particles from the lower surface of the carton wall and plant them on the upper, growing edge. Through this action they appear to be transferring fungal mycelia onto the new sugar bed, but the behavior could be a fortuitous outcome of the more ordinary distribution of nest materials--a common behavior of ants generally. The ants also crop the mycelium continuously. When the colony is separated from the carton, the fungal “lawn” sprouts out into a furry mass. According to Lagerheim (1900), the carton fungus in Central Europe, Cladosporium myrmecophilum, is known only from the nests of Lasius fuliginosus. The symbiosis appears to be truly mutualistic. However, unlike the attines, the Lasius do not consume the fungus as food. A possible second case of fungus cultivation outside the Attini occurs in the harvester ant Veromessor pergandei. Went et al. (1972) speculated that the ants feed on fungi growing on refuse within the nests. Their evidence, however, is circumstantial and tenuous. They noted that a very small percentage of items brought into the nests by foragers are arthropod exoskeletons and fragments of insect excrement. This material does not appear in the refuse piles surrounding the outside nest craters. A possibility exists then, certainly worthy of further study, that the material is being used as a fungal substrate.

'''Hölldobler, B. and Wilson, E. O. 1990. The Ants. Cambridge, Mass. Harvard University Press. Text used with permission of the authors.'''

The Ants - Table of Contents