The Ants Chapter 3

CHAPTER 3. THE COLONY LIFE CYCLE

The ant colony is an almost exclusively female society with the males remaining in the nest only until the time of their first, and invariably fatal nuptial flight. Also, the entire activity of the colony can be said to be pivoted on the welfare of the queen. It is, to paraphrase Samuel Butler's remark about the chicken and egg, the procedure by which a queen is used to make more queens. Seen in yet another way, the colony life cycle can be fruitfully analyzed as an orchestration of energy investments, in which workers are multiplied until such time as it is profitable to convert part of the net yield into new queens and males. In some extreme species this maturation point comes with the accumulation of only a few tens of workers, which are organized by means of the simplest caste and communication systems. For example, the average size of a colony of the fungus-grower Apterostigma dentigerum producing queens and males is 35 (Forsythe, 1981). The rare Central American myrmicine Basiceros manni reaches maturity at 50 workers (Wilson and Hölldobler, 1986). In others it is not attained until the worker population reaches tens of thousands and develops complex caste and communication systems. The extreme examples are the army ants, whose colonies do not divide until the worker populations exceed hundreds of thousands or even a million (Raignier and Van Boven, 1955; Rettenmeyer, 1963a).

The life cycle of a particular species can be viewed as the story of how the maturation point is attained with maximum combined speed and freedom from risk. Only by studying it as one strategy out of a great many possible can we expect to understand more deeply the way that a given species has adapted by social means to the particular environments in which it lives.

Stages of colony growth
Like the life cycle of the individual ant itself, the life cycle of an ant colony can be conveniently divided into three parts (Oster and Wilson, 1978). The founding stage begins with the nuptial flight. The virgin queen departs from the nest in which she was reared, leaving behind her mother, who is the queen of the colony, and her sisters, who are either sterile workers or virgin reproductives like herself. She meets one or more males and is inseminated. The males soon die without returning home, while the queen finds a suitable nest site in the soil or plant material and constructs a first nest cell. Here she rears the first brood of workers, drawing on her own tissue reserves to produce eggs and feed the growing larvae. Soon after reaching the adult stage, the workers take over the task of foraging, nest enlargement, and brood care, so that the queen is now permitted to confine herself to egg-laying. Over the coming weeks and months the population of workers grows, the average size of the workers increases, and new physical castes are sometimes added. The colony is now in the ergonomic stage: its activities are exclusively "ergonomic" in the sense that they are concerned with work devoted to colony growth, rather than with colony-level reproduction or dispersal (Figure Fig 3-1). After a period that ranges according to species from a single warm season to five or more years, the colony begins to produce new queens and males (reproductive stage). The sexual forms go forth to start new colonies, and the new colony life cycle has begun. As depicted in Figure 3-2, colonies of all known ant species are perennial. Like flowering plants, they issue a crop of seeds, then return to an interval of purely vegetative (i.e., worker) growth.

Substantial variation has been elaborated out of this elementary theme, especially with reference to details in the mode of colony founding and the number of egg-laying queens that coexist during the several stages of the life cycle. Figure 3-3 presents a classification of the variations and the relevant terminology. Monogyny refers simply to the possession by a colony of a single queen, as opposed to polygyny, which is the possession of multiple queens. The founding of a colony by a single queen is referred to as haplometrosis; when multiple queens start a colony the condition is called pleometrosis. The term metrosis refers generally to this biological variable. Monogyny can be primary, meaning that the single queen is also the foundress; or it can be secondary, meaning that multiple queens start a colony pleometrotically but only one survives. In a symmetric fashion, polygyny can be primary, in which multiple queens persist from a pleometrotic association, or secondary, in which the colony is started by a single queen and supernumerary queens are added later by adoption or fusion with other colonies. (The patterns of queen numbers will be discussed in greater detail in Chapter 6.)

Next, the mode of colony founding is subject to complicated variation among species. It can be accomplished by swarming, a process also called budding, hesmosis, or sociotomy, in which two or more forces of workers separate in the company of queens. We prefer to divide swarming into two types: the more common budding, in which a group of workers departs from the main nest with one or more queens and starts a new nesting unit; and fission of the kind used by army ants, in which queenright portions of the colony separate from each other and go their own way. Colony founding in ants is frequently claustral, meaning that the queen seals herself off in a chamber and rears the first brood in isolation. This is in fact the prevailing mode of independent colony formation in ants. However, the queens of such very primitive forms as Amblyopone and Myrmecia, as well as some more advanced ponerine genera, still forage outside their cells for food, the condition known as partially claustral colony founding (Wheeler, 1933b; Haskins and Haskins, 1950a,b, 1951). The same behavior has been observed in the myrmicines Acromyrmex, Manica, and some species of Pogonomyrmex, and the formicine Cataglyphis (Cordero, 1963; Le Masne and Bonavita, 1967; Fridman and Avital, 1983; Hölldobler, unpublished observations).

Nuptial flights and mating
The vast majority of virgin queens dies within hours after leaving the mother nest. Most are destroyed by predators (Figure 3-4) and hostile workers of alien nests, with the others being variously drowned, overheated, and desiccated. In species with large nest populations, such as the leafcutter ants (Atta) and fire ants (Solenopsis), it is not uncommon for one colony to release hundreds or thousands of the young winged queens in less than an hour. If the surrounding area is dominated by stable, mature colonies, only one or two of the queens might become the progenetrices of new colonies. Most of the rest will die before they can construct a first shelter--or even before they can find a mate. In an unusual study of its kind, Whitcomb et al. (1973) have produced a catalog of the many kinds of predators that decimate young queens of the red imported fire ant Solenopsis invicta. The few individuals that navigate all the dangers must also avoid breeding with males of other species, thus producing inviable or sterile offspring.

It follows that the brief interval between leaving the home nest and settling into a new, incipient nest is a period of intense natural selection among queens, a dangerous odyssey that must be precisely timed and executed in order to succeed. We should expect to find an array of physiological and behavioral mechanisms that enable the young queens simultaneously to avoid enemies, get to the right habitat on time in order to build a secure nest, and mate with males of the same species. Field studies have shown that such specialized traits exist in abundance.

As also expected from the evolutionary argument, mating patterns vary greatly from one species to the next. However, most of the patterns thus far studied fall into one or the other of two broad classes, or "syndromes" (Hölldobler and Bartz, 1985). In the first, the female-calling syndrome, the females, which are often wingless and sometimes just fertile workers, do not travel far from the nest. Standing on the ground or low vegetation, they release sex pheromones to "call" the winged males to them (Figure 3-5). This pattern is displayed by Amblyopone and Rhytidoponera, which are members of the phylogenetically primitive subfamily Ponerinae (Haskins, 1978); presumably also by the very primitive Nothomyrmecia macrops (Hölldobler and Taylor, 1983); at least one pseudomyrmecine, the Neotropical acacia ant Pseudomyrmex ferrugineus (Janzen, 1967); and the socially parasitic species of the myrmicine genera Doronomyrmex, Formicoxenus, Harpagoxenus, and Leptothorax (Buschinger, 1968a,b, 1971a,b, 1975b; see Figure 3-6). In general, the colonies of female-calling species are typically small at maturity, with 20 to 1,000 workers, and produce relatively few reproductives. So far as known the females mate only once. An unusual variation on this pattern is followed by the Florida harvester ant Pogonomyrmex badius. Females gather on the surface of their home nest and are inseminated by males; afterward they fly off to start new colonies. Van Pelt (1953) thought that the males came from the same nest as the females with whom they copulate, but S. D. Porter (personal communication) observed that they usually fly for about a quarter-hour first before settling on a nest different from their own. Porter observed one case in which a male mated with two females after alighting.

The second combination of traits during mating is the male-aggregation syndrome. Males from many colonies gather at specific mating sites, usually prominent features of the landscape such as sunflecked clearings, forest borders, hilltops, the crowns of trees, and even the tops of tall buildings. Sometimes, as in some species of Lasius and Solenopsis, the males cruise in large numbers at characteristic heights above the ground. The females fly into the swarms, often from long distances, in order to mate (see Figures 3-7 through 3-9 and Plate 2), and afterward they typically disperse widely before shedding their wings and excavating a nest. The winged queens and males of the fire ant Solenopsis invicta, for example, fly up to heights of 250 meters or more; 99 percent then descend to the ground within a 2-kilometer radius of their origin, while a very few travel as far as 10 kilometers. The ability of a single mature colony to disseminate fertile queens in many directions over long distances is one of the reasons the fire ant is so difficult to eradicate (Markin et al., 1971). Male-aggregation species typically differ from those utilizing female calling in two other key respects: the mature colonies are large, containing from several thousand to over a million workers and producing hundreds to thousands of reproductive adults yearly, and multiple insemination is common. An unusual reversal of the usual swarming procedure was recently discovered in some Pheidole species of the southwestern United States: the winged queens gather in aerial swarms, where they maintain a more or less uniform distance from each other while attracting males with pheromones. The males fly into the female swarms and mate with individual females (Hölldobler, unpublished). Swarms of variable composition, some predominantly male and others predominantly female (occasionally exclusively female), have been reported by Eberhard (1978) in the coccid-tending formicine Acropyga paramaribensis of northern South America.

Ant species can be classified another way into two broad types. When the males alight on the surface of the mating site, either in response to female calling or in swarms to compete directly with one another, they are often typically large and robust in form and possess well-developed mandibles. In contrast, males that gather in aerial swarms are usually (but not invariably) smaller relative to the queen than are males of the first type. Also, their mandibles are reduced in size and dentition, sometimes consisting of nothing more than vestigial lobate or strap-shaped organs. An example of this type is the small myrmicine Pheidole sitarches of the southwestern United States. Up to 50 males form circular swarms that hover from a few centimeters to two meters above the surface of woodland clearings. The virgin queens fly in slow, even circles through the aggregations until mounted in midair by a male, whereupon the pair cease flying and spiral to the ground together to complete the copulation (Wilson, 1957b).

The swarms of some ant species are among the more dramatic spectacles of the insect world. W. W. Froggatt (in Wheeler, 1916c) describes the flight of the giant Australian bulldog ant Myrmecia sanguinea as follows:

"On January 30th, after some very hot, stormy weather, while I was at Chevy Chase, near Armidale, N.S.W., I crossed the paddock and climbed to the top of Mt. Roul, an isolated, flat-topped, basaltic hill, which rises about 300 feet above the surrounding open, cleared country. The summit, about half an acre in extent, is covered with low "black-thorn" bushes (Bursaria spinifera).  I saw no signs of bull-dog ant nests till I reached the summit.  Then I was enveloped in a regular cloud of the great winged ants.  They were out in thousands and thousands, resting on the rocks and grass.  The air was full of them, but they were chiefly flying in great numbers about the bushes where the males were copulating with the females.  As soon as a male (and there were hundreds of males to every female) captured a female on a bush, other males surrounded the couple till there was a struggling mass of ants forming a ball as large as one's fist. Then something seemed to give way, the ball would fall to the ground and the ants would scatter. As many as half a dozen of these balls would keep forming on every little bush and this went on throughout the morning. I was a bit frightened at first but the ants took no notice of me, as the males were all so eager in their endeavors to seize the females."

Donisthorpe (1915) tells of the mass flights of the abundant Myrmica rubra from the distinctively British viewpoint of an earlier observer:

"Farren-White in 1876 observed a swarm of ants near Stonehouse rising and falling over a small beech tree. The effect of those in the air--gyrating and meeting each other in their course, as seen against the deep blue sky--reminded him of the little dodder, with its tiny clustered blossoms and its network of ramifying scarlet threads, over the gorse or heather at Bournemouth.  He noticed the swarm about thirty paces off, and it began to assume the appearance of curling smoke; at forty paces he could quite imagine the tree to be on fire.  At fifty paces the smoke had nearly vanished into thin air."

A still different mating pattern was described in the Australian formicine species Notoncus ectatommoides by W. L. Brown (1955a):

"In a cropped lawn at Montville, numerous small holes appeared, each opened by workers and accompanied by a minute pile of dark earthen particles. From these holes, males began to issue almost immediately in numbers, until within a few minutes there had accumulated on the surface a surprisingly large number of this sex and also a few workers.  The males travelled aimlessly over the sward in low, flitting flight from one blade of grass to another, never rising more than a foot or so from the ground.  Movement seemed to take place at random in all directions.  Suddenly, however, the males of one area all rushed simultaneously to a single focal point, which proved to be a winged female emerging from a small hole.  In a few seconds, the female was surrounded by a dense swarm of males in the form of a ball, which at times must have exceeded 2 cm in diameter.  This ball moved in a half-tumbling, half-dragging motion over and among the densely packed grass blades, and held together for perhaps 20 seconds, after which the female escaped, flying straight upward. She appeared not to be encumbered by a male, and no males were seen to follow her for more than a foot above the ground; she flew steadily, and soon passed out of sight.

Meanwhile, the lawn had become dotted with similar balls of frenzied males, each surrounding a female in a fashion similar to the first. Obviously, many more males than females were involved in this particular flight. On each occasion, the female left the ball after 20-30 seconds and flew straight upward."

In a similar fashion males and females of Formica obscuripes conduct nuptial swarms on the ground. Talbot (1972) observed them flying to "swarming grounds" near their nests which were maintained throughout the nuptial flight season and perhaps even from year to year. The males fly back and forth above the ground searching for females which "stand on grasses, forbs or bushes," and apparently signal their presence to the males by pheromones.

No encompassing theory exists to explain the extreme variation in the patterns of mating behavior so far observed. However, a close examination of individual species reveals details that clearly contribute to the greater success of the sexual castes. For example, flying queens of the formicine Lasius neoniger stay strictly within open fields, the exclusive habitat of the earthbound colonies. Fewer than one percent make the mistake of venturing into adjacent woodland, a habitat dominated by the otherwise closely similar Lasius alienus. In one experimental study (Wilson and Hunt, 1966), newly inseminated and flightless queens were labeled with radioactive material for easy tracking and displaced to woodland sites. They attempted to crawl out but were unable to do so. In other words the Lasius queens depend on controlled flight patterns to survive.

Like orientation, the timing of the flights is important for successful mating and colony foundation. Flights conducted as part of the female-calling syndrome do not appear to be well synchronized at the level of either the colony or the population of colonies. The search by airborne males for solitary calling females in fact resembles that of many solitary wasps (Buschinger, 1975; Haskins, 1978). In contrast, flights leading to male aggregation are tightly synchronized within the colony as well as among colonies of the same species.

The manner in which this coordination is achieved is typified by Pogonomyrmex harvester ants of the southwestern United States. The process has been described by Hölldobler (1976b). Just prior to take-off, males and females move restlessly in and out of the sandy crater nests or gather in clusters around the entrance, as shown in Figure 3-7. This preflight activity is especially pronounced in Pogonomyrmex maricopa, a morning flyer, the queens and males of which evidently need more time to warm up before taking wing. As the time of departure approaches, the reproductives run back and forth in mounting intensity. Now, in a frenzy, they climb up and down on grass leaves or small bushes around the nest. At this point many more workers pour out of the nest, running excitedly around the nest and attacking any moving object encountered (including the careless myrmecologist). When the first reproductives try to take flight, the workers at first delay many of them by pulling or carrying them back to the nest. However, once the flight is in full progress, workers cease to interfere. Although the timing of the take-off overlaps considerably between the two sexes, the males generally fly from the nest first. Once aloft both sexes appear at first to drift with the wind, but after a few seconds they take a course upwind or across the wind. Soon afterward they arrive at the swarm sites, centered on conspicuous landmarks such as tree crowns and the tops of hills or (in the case of Pogonomyrmex rugosus) merely flat local areas in the desert.

A similar marching order is observed by the carpenter ant Camponotus herculeanus, which nests in the trunks of both living and dead trees in the boreal forests of Eurasia and North America. Males leave before the queens, although the periods broadly overlap. The early departure of the winged forms is inhibited by the workers, who drag or carry many back to the nest entrance (Figure 3-10). However, when the males do succeed in taking flight, they discharge a pheromone from their mandibular gland. The concentration of this substance is highest at the peak of male activity--the gland emission can now be smelled readily by humans--enough to trigger the mass take-off of the females (Figure 3-11). Blum (1981b) reports methyl 6-methylsalicylate and mellein as two of the three components of the secretion. This pleasantly aromatic combination is shared by most other species of Camponotus, but considerable differentiation nevertheless is achieved by the addition of other substances, such as octanoic acid and methyl anthranilate, according to species (see also Lloyd et al., 1984). A similar function may be accomplished by vibrational signals rather than pheromones in Pogonomyrmex harvester ants. Both males and virgin queens stridulate just before and during take-off, running the sharp posterior rim of their postpetiole over the actively moving, striated file on the first gastric tergite (Markl et al., 1977).

Many entomologists, including especially Kannowski (1959a, 1963) and Weber (1972), have observed that each ant species, at least those displaying the male-aggregation syndrome, swarms at a precise time in the 24-hour diel cycle; and the time differs among species. Under controlled laboratory conditions, McCluskey (1958, 1965, 1967, 1974) and McCluskey and Soong (1979) demonstrated in fact that the rhythms of males are generally if not universally circadian and endogenous. Once set in a laboratory regime of 12 hours light alternating with 12 hours dark, the rhythms persist for up to a week in total darkness. They are also quite precise. McCluskey found that males of the harvester ant Messor (= Veromessor) andrei increase in movement during the last hour of darkness, then peak during the first hour of light. Throughout the remainder of the 24-hour cycle they are quiescent, usually stirring themselves only to groom, solicit food from the workers, or walk sluggishly about the nest. Males of the Argentine ant Iridomyrmex humilis, in contrast, are most active at the very end of the light period. Similarly distinctive rhythms, each spanning only one or two hours, have been documented by McCluskey and his co-workers across a wide diversity of species from four subfamilies (Ponerinae, Myrmicinae, Dolichoderinae, and Formicinae), including some that are wholly nocturnal.

Queens of at least two species, Pogonomyrmex californicus and Mesor (= Veromessor) pergandei, also display circadian rhythms, and these are more or less synchronous with those of the males (McCluskey, 1967; McCluskey and Carter, 1969). In the case of P. californicus at least, the rhythm persists even after the female has flown and lost her wings. But it ceases when she is mated.

In summary, the time of day in which flights occur is programmed by a species-specific diel rhythm. But what determines the particular day on which the flights occur? Several studies, including that by Boomsma and Leusink (1981), have shown that weather conditions play a major role in the timing of nuptial flights. One of the commonest triggering stimuli is rain, especially in species that occur in dry habitats such as deserts, grassland, and forest clearings. A typical species in this respect is Lasius neoniger, one of the most abundant ants in abandoned fields and other open environments in eastern North America. This small formicine emerges in immense swarms in the late afternoon in the second half of August or early September. The flights almost always occur within 24 hours after moderate or heavy rainfall on warm, humid days with little wind. For an hour or so the air seems filled with the winged ants. They rise from the ground like snowfall in reverse. After mating, the queens find themselves on moistened soil that is easier to excavate. They are also protected from desiccation due to overheating (Wilson, 1955a). A very similar pattern is followed by the North American leafcutter ant Atta texana, except that the flights occur well before dawn, between 0300 and 0415 hours (Moser, 1967a).

Because there are relatively few "best days" in which the young queens can be successfully launched, species belonging to the same genus are likely to swarm at the same time and location. In one respect this is a favorable result, since an apparent function of mass emergence and swarming in cicadas, termites, and other insects is the reduction of mortality by overloading predators (Wilson, 1975b). But in another respect it can be detrimental. In the tumult of the swarms, with males struggling to copulate with each female encountered, there is a strong likelihood of interspecific hybridization resulting in either sterility or the production of less viable hybrids. Applying the standard argument from natural selection theory, this circumstance favors the evolution of premating isolating mechanisms. The conventional explanation does seem compatible with a great deal of evidence. Species belonging to genera as phylogenetically diverse as Myrmecia, Pheidole, Solenopsis, and Lasius have been observed to conduct their nuptial flights within the major habitats occupied by the colonies, thus automatically avoiding sexual contact with closely related species limited to other major habitats. How widespread and efficient this isolating mechanism is among ants in general has not been determined. But it cannot be the sole device in deserts, savannas, and tropical moist forests, where large numbers of congeneric species nest closely together. To take an extreme case, in many forest localities in the Amazon Basin, thirty or more species of Pheidole can be found within a single plot of a few square kilometers. Another intrinsic isolating mechanism is differentiation in the preferred mating site within the major habitat. Among the sympatric species of Pogonomyrmex of Arizona, Pogonomyrmex desertorum and Pogonomyrmex maricopa congregate on bushes and trees, while Pogonomyrmex barbatus and Pogonomyrmex rugosus gather at different sites on the ground. In addition, males mark the sites with secretions from their mandibular glands, and apparently the females and other males are attracted by volatile pheromones contained in the material (Hölldobler, 1976b). It is possible (but not yet experimentally verified) that the pheromones are species-specific and serve as an additional isolating device.

Many congeneric species are further separated by the timing of their mating flight, either the season of the year or the hour of the day. In Figures 3-12 and 3-13 we have presented two sets of data from army ants that suggest just such a mutually repulsing spread of flight times across the seasons and the daily cycle respectively. The males of army ants, on which the data were based, fly for an unknown distance before entering the columns or bivouacs of alien colonies belonging to the same species. If the receptiveness of the workers is synchronized by the same circadian rhythm, even the hours of flight can serve as an effective barrier to "mistakes" and interspecific hybridization. Such staggering in the diel flight schedule appears to be common among ants. In Michigan, for example, Myrmica emeryana flies between 0600 and 0800 hours, Myrmica americana between 1230 and 1630 hours, and Myrmica fracticornis between 1800 and 1930 hours (Kannowski, 1959a). Similarly, in Arizona Pogonomyrmex maricopa flies between 1000 and 1130 hours, Pogonomyrmex barbatus between 1530 and 1700 hours, and Pogonomyrmex rugosus between 1630 and 1800 hours. As morning flyers, the Pogonomyrmex maricopa queens appear to be at some disadvantage. The heat of midday prevents them from beginning nest excavation for three or four hours, during which time they are subject to higher predation than the other species (Hölldobler, 1976b). Some of the most closely related European species of Leptothorax swarm at different times of the day; others come into contact, and occasionally hybridize (Plateaux, 1978, 1987).

Another potential advantage of synchronous nuptial swarming is the increase in the numbers of colonies participating and hence the degree of outbreeding. The sparse data on allozyme variation in ants collected so far indicates that outbreeding is indeed nearly total (Craig and Crozier, 1979; Pamilo and Varvio-Aho, 1979; Pearson, 1983; Ward, 1983a). Hence mating is either effectively at random, as demonstrated in experimental choice tests with Pogonomyrmex californicus by Mintzer (1982a), or disassortative, that is, directed away from nestmates.

The glandular sources of sex pheromones produced by female ants have been identified only for a few species. The reproductive females of Rhytidoponera metallica call males with a sex attractant from the pygidial gland, an intersegmental structure between the VIth and VIIth abdominal tergites (Hölldobler and Haskins, 1977). Although some of the contents of this gland have been chemically identified (Meinwald et al., 1983), the specific behavior-releasing components have not yet been established experimentally. In several myrmicine species glands associated with the sting apparatus have been pinpointed as the sources of female sex pheromones. Virgin queens release a male attracting pheromone from the poison gland in the myrmicines Xenomyrmex floridanus (Hölldobler, 1971); Harpagoxenus sublaevis (Buschinger, 1972a); Doronomyrmex kutteri and Doronomyrmex pacis (Buschinger, 1975b; see Figure 3-6); and Formicoxenus nitidulus (Buschinger, 1976a,b).

Buschinger (1972b) was also able to demonstrate that males of Doronomyrmex kutteri and D. pacis react to the other species' female sex pheromones, and that hybridization is possible in laboratory experiments. In the field, however, both species, which occur sympatrically, appear to be sexually isolated by different diel rhythms in mating activity. In general, specificity in sexual communication is consistent with phylogenetic relationships among the leptothoracines. The Canadian slavemaker Harpagoxenus canadensis shows the same mating behavior as the European H. sublaevis, and males of both species respond to the other species' female sex pheromones. Very similar sexual behavior and responses to sex pheromones have been described in several other social parasites of the "subgenus Mychothorax" of Leptothorax, whose hosts, like those of H. canadensis and H. sublaevis, also belong to the "subgenus Mychothorax." The same is true of at least some non-parasitic members of the subgenus. In fact, there appears to be no pheromone specificity among the Leptothorax species. In contrast, Protomognathus americanus males do not respond to H. canadensis or H. sublaevis pheromones. This anomaly suggests that P. americanus may be more closely related to its host of the "subgenus Leptothorax" than to the other genus of Harpagoxenus or the "subgenus Mychothorax" (Buschinger, 1975b, 1981; Buschinger and Alloway, 1979).

Poison gland secretions of Pogonomyrmex females also elicit attraction in males (Hölldobler, 1976b). In Monomorium pharaonis, on the other hand, the female sex pheromone is derived from the Dufour's gland and the bursa pouches (Hölldobler and Wüst, 1973).

Male ants are richly endowed with exocrine glands (Hölldobler and Engel-Siegel, 1982), but little is known about their function. One important fact, noted earlier, is that Camponotus herculeanus males discharge mandibular gland contents when departing from the nest that stimulate the virgin reproductive females to launch as well the nuptial flight. A variety of compounds of the mandibular gland secretions of several Camponotus species have been identified (for review see Blum, 1981b), but it is not yet clear which substance or combination of compounds elicits the behavior. Similarly, the males of Lasius neoniger discharge their mandibular gland contents sometime during the nuptial flight (Law et al., 1965), but the precise timing and function remain unknown.

Males of Pogonomyrmex discharge mandibular gland secretions when arriving at the mating sites. The collectively discharged pheromone appears to attract the virgin females to the lek (Hölldobler, 1976b). It is possible that in other species where males have well-developed mandibular glands and distinct blends of compounds, the secretions also function in promoting aggregation and competition. Examples include Lasius' and Acanthomyops (Law et al., 1965), Camponotus (Brand et al., 1973b,c; review by Blum, 1981), Calomyrmex (Brown and Moore, 1979), Myrmecocystus (review by Blum, 1981b), Tetramorium caespitum (Pasteels et al., 1980), and Polyrhachis doddi'' (Bellas and Hölldobler, 1985).

A hypothesis concerning a possible novel role of male pheromones in sexual selection in army ants has recently been proposed by Franks and Hölldobler (1987). A detailed morphological examination of the reproductives has shown a close resemblance of conspecific males and females. Males are remarkably queen-like. They are large and robust, and their long, cylindrical abdomens are partially filled with an impressive battery of exocrine glands similar in form and location to those of females. Because queens are flightless and never leave their colony, males must fly between colonies and run the gauntlet of the workers before they approach the queen. For this reason, the workers can choose which males will be admitted and which virgin queens will be inseminated by the males. Army ant workers might therefore be involved in a unique form of sexual selection in which they choose both the matriarch and patriarch of new colonies. If this interpretation is correct, males resemble queens not because they are deceitful mimics; instead, under the influence of sexual selection they have come to use the same channels of communication to demonstrate their potential fitness to the workers as those used by queens.

Worker involvement in sexual selection might not be restricted to the army ants. Wheeler (1910a) noted that males of Leptogenys elongata are also accepted into alien colonies to mate with the wingless ergatoid females, and Maschwitz and Mühlenberg (1975) observed that males run along permanent foraging trails of Leptogenys ocellifera, apparently in an attempt to find access to ergatoid females. It may therefore be significant that Hölldobler and Engel-Siegel (1982) discovered very large exocrine sternal glands in Leptogenys males. Some other ponerines have ergatoid queens and therefore are not likely to engage in ordinary nuptial swarms, including species of Diacamma, Dinoponera, Megaponera, and Ophthalmopone. Longhurst and Howse (1979a) observed that males of Megaponera foetens enter the nests of alien colonies, after utilizing recruitment pheromone trails laid by workers to guide them to the nest. No information is available, however, on the exocrine glandular system of Megaponera or for that matter most other ponerine genera. Males of Ophthalmopone berthoudi also enter strange nests after dispersal flights, but so far as known do not follow odor trails--O. berthoudi workers in fact forage in an exclusively solitary manner and hence are less likely to lay recruitment trails of any kind (Peeters and Crewe, 1986a, 1987).

Male ants compete for females in a rigorous fashion, whether they are orienting to calling females in the primitive manner, flying in aerial swarms, or massing on the surface of the ground and vegetation. The competitive nature of mating is vividly illustrated by Pogonomyrmex rugosus (see Plate 2). The males gather in what can properly be called leks of the vertebrate kind. That is, the males occupy the same site year after year, use pheromones to attract other reproductives of the same species, and then compete with one another for access to the females. In the desert near Portal, Arizona, Hölldobler (1976b) was able to locate only one such site in an area of approximately 120,000 m2. The mating arena covered 4800 m2 of completely flat land unmarked by any distinctive physical features. The winged reproductives approached the arena upwind, which may suggest the presence of an olfactory cue. The first individuals to arrive (at around 1630 hours) were males, which alighted and began to race about in a frenzied manner. Soon afterward the first females alighted. They were immediately surrounded by three to ten males, as shown in Figures 3-8 and 3-9. At the height of the activity thousands of such mating clusters carpeted the ground, in densities as high as 50 per square meter. The queens actively terminated mating after several copulations, and stridulated when prevented from leaving by other suitors. This stridulatory vibration evidently served as a "female liberation signal" that communicated the female's non-receptivity to approaching males and induced them to cease pursuit (Markl et al., 1977). The females then climbed onto grass leaves to launch their flights or else flew directly from the ground. Some landed a short distance away, but others traveled at least 100 meters and possibly much farther. Each then shed her wings and began to excavate a nest chamber in the soil (Figure 3-14).

The general activity at the Pogonomyrmex rugosus mating site lasted about two hours, ending completely by 1900 hours as darkness approached. The males then withdrew into shelters around the mating site, such as crevices beneath grass clumps or little cavities in the soil. There they remained clustered overnight and through the following day until 1500-1600 hours, when they resumed activity. As on the previous day, new males flew in to the site to swell the population, and shortly afterwards females began to arrive. This cycle was repeated on three more consecutive days.

The ant leks differ from those of the sage grouse, hammerheaded bats, and Hawaiian Drosophila (see for example, Bradbury, 1985) in one important respect. Ant males are constrained in a way that vertebrates and fruit flies are not: each male ecloses from the pupa into full maturity with all of the sperm that he will ever possess. Dissections of males from phylogenetically divergent genera such as Nothomyrmecia, Camponotus, Lasius, Myrmica, and Pogonomyrmex reveal that the males' testes have degenerated and all of the sperm have migrated to the expanded vas deferens (Hölldobler, 1966, and unpublished data). When the male mates, it discharges most or all of the sperm together with the secretions of the mucus gland; he is thus incapable of additional inseminations (see Figure 3-15). As a result, reproductive success in male ants does not increase with repeated copulations, as it does with other kinds of insects whose males continuously replenish their sperm supply. Furthermore, it does not appear, from the few cases known, that males have enough sperm to inseminate more than one female. In cases where the queen is destined to produce very large numbers of offspring, one male is not even able to supply all of her needs. In Atta sexdens, for example, each newly eclosed male has between 40 and 80 million spermatozoans, while each newly mated female contains between 200 and 310 million spermatozoans in her spermathecae (Kerr, 1962). Fire ant queens (Solenopsis invicta) receive a supply of about 7 million sperm initially, which they gradually parcel out over a period of almost 7 years until the supply is exhausted (Tschinkel and Porter, 1988). Male ants are thus under strong pressure in natural selection to husband their sperm carefully.

One obvious question concerning the ultimate reproductive success of males is whether it is better for a male to invest all of his sperm in a single female or else to copulate with several females. As Hölldobler and Bartz (1985) pointed out, it is important to note that in ants, unlike other nonsocial species, a male's sperm does not all go towards effective reproduction. This is because in order for an ant colony to begin to produce any reproductive forms, it first must produce many workers. In most advanced ant societies workers are rarely reproductive, and because workers are females derived from fertilized eggs, a substantial portion of a male's sperm is used in colony growth and maintenance rather than in direct production of new queens. The consequent trade-off for a male is obvious. If he inseminates only a single female, and if she mates with no other male, then the male is certain to father any reproductives that she eventually produces. However, mortality of colony-founding females is extremely high. Hence a male that inseminates only a single female puts all of his sperm in one fragile basket. If he were to inseminate several females, on the other hand, he increases the chance that his sperm will end up in a successful foundress of a colony. In this case, however, he might decrease the chance that his sperm is used by the queen to make alates. On the other hand, if males do inseminate several females, there may be selection favoring males whose sperm mixes with other males' sperm in the females' spermathecae. Mixing sperm increases the chance that each male will have at least some offspring among the new crop of alate queens. From allozyme variation studies in multiple mating ant species, it does in fact appear that workers in colonies are fathered by several males (Pamilo, 1982b,c; Pearson, 1983; Ward, 1983a).

In species with large mature colonies, whose females must mate with several males in order to acquire sufficient sperm, males seldom attempt to monopolize females (Cole, 1983b). In many other kinds of insects, and other organisms as well, sperm competition is an important selective force, and males are often favored to ensure that no other male copulates with his mate (Parker, 1970). In multiple mating ant species, however, a male that prevents his mate from mating again may well prevent her from acquiring enough sperm to generate a mature colony, that is, one large enough to produce reproductives. Males in these species therefore should be selected to mate with females that are already mated. The active vying for position in waiting lines behind copulating pairs in Pogonomyrmex species indicates at least that males do not discriminate against previously mated females, but it does not prove the optimum multiple mating hypothesis.

To summarize, male ants are faced with two limiting resources: a restricted number of females available for mating, and a finite supply of sperm that suffices for only one or at most several matings. An expected consequence in evolution is the fierce competition of the kind observed in the Pogonomyrmex leks. Davidson (1982) observed that Pogonomyrmex barbatus and Pogonomyrmex desertorum males indiscriminately seize females and attempt to mate, while the females actively resist copulation. As a result, large males are disproportionately successful at gaining access to mates. In addition, large females mate even more disproportionately with large males. And still further, the average size of males produced by an individual colony depends on the total number of reproductives reared in a given season, which in turn is a function of the size and vigor of the colony. In short, the bigger the colony, the more likely its individual males are to succeed in the mating arenas. Why hasn't this selection pressure created ever larger males in evolution? Davidson offers two reasons: larger males mean fewer males per colony, an obvious trade-off in colony fitness, and very large males (as opposed to merely large ones) have been observed to lose some of their advantage to those slightly smaller. The result is the existence of an optimum male size in Pogonomyrmex.

A confounding bit of data reported by Davidson (1982) is that not only do larger females tend to mate with larger males, but smaller females tend to mate with smaller males. If sexual selection is operating such that females choose larger males, why do small females not also choose to mate with larger males? The answer to this question may be that males are selected to be choosy as well. As we have pointed out male ants have only so much sperm at their disposal, and they cannot afford to be profligate. Selection may favor males who compete for larger females because there is a better chance that large females will survive to produce a mature colony. A result of the competition would be that the smaller, less competitive males must settle for the smaller, less desirable females.

The evolution of male biology has been subjected to few rigorous studies, and most questions concerning trends and optimality in its evolution remain unanswered. We are in a somewhat better position with reference to both data and theory on the number of female matings. As documented in Table 3-1, which includes most or all of the information available, some fraction of the queens of fully three-quarters of all species copulate with more than one male. It is also true, as revealed by allozyme marker studies (Pamilo, 1982b,c; Pearson, 1983; Ward, 1983a), that the sperm from different fathers contribute randomly to fertilization. Cole (1983b) established that multiple matings (polyandry) occurs more frequently in species with large colony size. He concluded, as West-Eberhard (1975) and a few other previous writers had suggested earlier on more intuitive grounds, that polyandry was therefore likely to be a response to the need on the part of queens in large colonies for more spermatozoans than one male can provide. In a study of 25 species in 5 subfamilies, Tschinkel (1987a) added stronger evidence from the number of sperm acquired by queens. In comparisons across species, the number of sperm increases very rapidly with the number of ovarioles. It ranged in Tschinkel's sample from a few tens of thousands in Ponera and Hypoponera, which form small, slow-growing colonies, to 400 million in the leafcutter Atta texana, which attains populations of over a million workers at a time. At another level, the number of sperm stored per ovariole (as opposed to per queen) increased from 2,000 for queens with only six ovarioles to about 30,000 for queens with about 200 ovarioles.

Not satisfied with the intuitively simplest explanation, however, Crozier and Page (1985) went on to employ the method of multiple competing hypotheses to test the adaptiveness of polyandry. They constructed no less than eight such explanations (some admittedly very improbable) to account for the trend documented by Cole. The explanation of limited male contribution favored by Cole and earlier authors was downgraded, because "males of species with big females are generally larger than those with small females, so there is no absolute bar to male size (within reason!)." This does not seem to be a very strong counterargument. Aerial swarmers can potentially benefit from smaller size, which confers greater agility during the approach to incoming queens. Also, as we have noted with reference to mating in Pogonomyrmex, there is a trade-off between male size and male numbers, still poorly analyzed, that might contribute to the preferred production by colonies of smaller males.

Hence it is prudent to keep alive the limited-sperm hypothesis of polyandry. Crozier and Page, after discarding most of the other competing explanations, hold on to three (not counting the limited-sperm hypothesis, which we favor) as both inherently plausible and compatible with the correlation between colony size and polyandry. The first is that caste determination might be genetic, and if so polyandry would allow fuller expression of the caste system in each colony. It follows that species with more complex caste differentiation (a trait associated with large colonies) should be more polyandrous than species with simpler caste systems. As Crozier and Page note, there is no evidence for genetic caste determination in ants to the present time, although recent evidence suggests some kind of genetic predisposition toward various forms of labor specialization in honeybees (Calderone and Page, 1988; Frumhoff and Baker, 1988; Robinson and Page, 1988). All of the many substantial studies to date have implicated a single genotype with multiple developmental pathways controlled by nutritive and other environmental factors (see Chapter 8). The second surviving explanation in the Crozier-Page analysis is that polyandry maximizes the production of divergent worker genotypes, quite apart from caste phenotypes, and hence the range of environmental conditions that a colony can tolerate. Broad-niche species, most often those possessing large colonies, should be more polyandrous than species with narrow niches. This broad relationship has not yet been tested empirically. The third favored hypothesis is that multiple matings reduce the chances of disaster due to the production of diploid males. Males of Hymenoptera, it will be recalled, ordinarily come from unfertilized eggs and are determined as males simply by being haploid, that is, having only one set of sex-determining genes. When one or a very few loci are involved in the process, and recessive male-determining alleles exist, it is also possible to get males from fertilized eggs, the so-called diploid males. The queen of an ant colony can ordinarily control male egg production precisely by opening or closing her spermathecal valve "at will," thus determining whether an egg in the vaginal passage is fertilized. But she has no control whatever over the production of diploid male eggs, because the effort to produce females will still result in a fixed percentage of males by Mendelian chance alone. This circumstance does not matter much if the strategy of the colony is to produce males during early stages of colony growth (beyond the very earliest, fragile stage of colony founding), a not uncommon event in species with a small mature colony size. But it can add a substantial energetic burden on species whose strategy is to hold off production of drones until the colony is large. By mixing sperm from multiple males, the variance of such a load is reduced. In other words, more colonies are likely to have some diploid males, but on the average they are less likely to produce large numbers of diploid males.

Reasoning in another mode, Woyciechowski and Lomnicki (1987) proposed that multiple matings prevent workers from producing male offspring. According to their model of kin selection, workers are at an advantage if they produce sons and care for nephews in the presence of a mother queen who mated only once, but they should avoid personal reproduction and care for brothers in the presence of a mother queen who mated several times. The existing data on queen mating patterns and worker reproduction are not adequate to test the hypothesis.

More recently, Sherman et al. (1988) have argued that the role of polyandry is to increase genetic variation within colonies, thereby reducing the likelihood that parasites or pathogens can decimate the worker force by overcoming all of its physiological and behavioral defenses at once. A balanced portfolio of investments in genetic variation, in other words, is more likely to produce the highest long-term probability of survival and successful growth. This argument is logical, but in our opinion does not accord with the remarkable correlations that exist between polyandry, sperm count, and colony size. This latter relation favors the limited-sperm hypothesis but no other.

In any case, the reproductive behavior of ants is a still poorly explored domain with rich possibilities for general evolutionary biology. More studies are needed along all fronts, including the comparative natural history of nuptial flights, the detailed analysis of individual males and females during mating, genetic studies of sex determination, and more sophisticated models of reproductive competition at the individual and colony levels.

The founding stage
As soon as the queens are inseminated, they shed their membranous wings by raking the middle and hind legs forward and snapping the wings free at the basal dehiscent sutures. Over the coming weeks the alary muscles and fat bodies are metabolized and converted into eggs, as well as food to rear the first batch of larvae. The latter nutrient materials are packaged as either trophic eggs (eggs that cannot develop but are used exclusively as food), specialized salivary secretions, or both. The basic process was first described in the classic study of the formicine ant Lasius niger by Charles Janet (1907). More recently, it has been found that the esophagus of the queen expands into a "thoracic crop" in which the converted tissues are temporarily held in liquid form. In Pharaoh's ant (Monomorium pharaonis), the esophagus diameter widens from 7-10 micrometers to 265 micrometers. The thoracic crop has been demonstrated in five genera of Myrmicinae and Formicinae so far (Petersen-Braun and Buschinger, 1975).

In the case of Solenopsis invicta at least, the conversion process is mediated by the corpora allata. Allatectomized queens fail to cast their wings or undergo wing muscle histolysis, while treatment of these operated individuals with juvenile hormone causes both processes to proceed (Barker, 1979).

The external trigger for wing shedding and histolysis is not exclusively insemination, as might be guessed. When virgin queens of some species are taken from the presence of other queens, they drop their wings after about a day and begin to behave more in the manner of inseminated nest queens. In the case of Solenopsis invicta in particular, the crucial signal is a relatively nonvolatile pheromone produced by queens and conveyed to the virgin alates (Fletcher and Blum, 1981; Fletcher et al., 1983).

The conversion of body tissues into food for the larvae was a vital evolutionary advance in the Formicidae. Wheeler (1933b) suggested that partially claustral colony founding (where the queen still leaves the nest to obtain some of the food) is the primitive state and fully claustral colony founding was derived from it. This inference is based on sound logic: if the ants did originate from predatory vespoid wasps, as the anatomical evidence suggests, they or their immediate ancestors were likely to have passed through a stage in which foundresses still captured insect prey and transported them to preexisting nests. In other words, the earliest ants are likely to have been partially claustral.

It is further true that species of the relatively primitive subfamily Ponerinae display finely graded steps leading from the partial to the fully claustral mode, from dependence on outside foraging to more or less complete freedom from it. The queens of Pachycondyla (= Bothroponera) soror, for example, are in an exactly intermediate stage. They forage outside the nest, but their wing muscles are still reduced and metabolized in the manner of the higher ants (Haskins, 1941). The queens of Odontomachus haematodus are capable of rearing their first brood at least partially with their own oral secretions, but, in one experiment performed by Haskins and Haskins (1950b), the larvae still failed to reach maturity. Finally, these authors found that the unusually bulky queens of Pachycondyla (= Brachyponera) lutea are able to rear the first brood all the way through solely with their own secretions, even though they continue to forage outside the nest when given the opportunity. From Pachycondyla lutea it is but a short step to the condition typifying most myrmicines, formicines, and other "higher" ants, in which complete claustral colony founding is the mode.

At least one substantial advantage of claustral colony founding is obvious and may in fact have played a role in its general adoption among the phylogenetically more advanced ants. Social insect workers suffer their highest mortality during foraging trips (Porter and Jorgensen, 1981; Schmid-Hempel, 1984), and it is probable that the same is true of founding queens forced to leave their nests to search for food.

The ergonomic (exponential) stage
The first workers produced by the queen are typically "nanitics" or "minims," that is, miniature forms somewhat smaller than the smallest workers encountered in older colonies of the same species (Figure 3-16). They are characteristically timid in behavior but otherwise perform the same repertory of tasks as do workers in older colonies. In the fire ant Solenopsis invicta at least, they differ from other worker castes in venom composition, specifically in the relative proportions of piperidine alkaloids (Vander Meer, 1986b). In the case of dimorphic species, the first-brood nanitics possess the basic anatomical structure of the minor caste. Major workers usually do not appear until later, and even then are initially smaller in average size. Minims are a general, perhaps universal phenomenon in ants, occurring not only in the "higher" subfamilies and genera, but also in the primitive Australian genus Myrmecia (see review in Wood and Tschinkel, 1981).

Ergonomic models designed to calculate the net energetic yield and hence growth rates of colonies support the intuition that the small size and timidity of the first workers represent prudent features built into the investment strategy of colonies as a whole (Oster and Wilson, 1978). A newly founded colony should strive to maximize the number of workers and their initial survival rate at the expense of everything else. The intuited reasons are as follows. With the queen's internal resources exhausted, there is a minimum number of workers needed to accomplish an adequate performance in each of the vital tasks--a certain number to enlarge the nest, a certain number to nurse the second brood, a certain number to forage, and so forth. There should also exist an optimum number, above this minimum, since adult mortality is probable before the second brood reaches adulthood. The optimum number of nanitics can be defined as that above which the survival probability of the queen can no longer be significantly increased and in fact is likely to be decreased. Because the biomass of adult workers that can be produced by the founding queen is very limited, it is efficient for the queen to divide it into many small workers, as documented by Porter and Tschinkel (1986) in the fire ant Solenopsis invicta. But this advantage is easily reversed, because to raise a great many such individuals would necessitate the production of excessively small nanitics unable to exploit the food items and nest sites for which the species is anatomically and behaviorally adapted. As a result, there should be an optimum number of nanitic workers, determined by the balance between the advantages of a larger initial worker force and the disadvantages of a smaller body size (see Figure 3-17). This result has been confirmed experimentally in Solenopsis invicta by Porter and Tschinkel (1986). They found that nanitics are less efficient on an individual basis than ordinary minor workers in rearing brood, but more efficient as a group than a group of minor workers of equal combined weight--because they are more numerous. On the other hand, they are energetically more expensive to maintain and thus are superior only for the brief period of colony founding.

Moreover, the small size of the incipient colony seems to dictate that its members be relatively timid in behavior. Suppose that an encounter with a single enemy such as a group of foragers from an alien colony results in the loss of five workers. For a mature colony containing thousands of members, this sacrifice is not only tolerable but desirable, if it clears enemy scouts from the territory on which the population depends for food. But for an incipient colony of only ten workers, the loss could be fatal. Furthermore, the potential gain from expelling territorial intruders is expected to be less, because the incipient colony is still living on a fraction of the available food supply yielded by the surrounding terrain.

The nanitics appear to owe their miniature size at least partially to the meager nutrients supplied them by the founding queen in their larval stage. Pheromonal or other programmed stimuli from the queen may also be important. When Wood and Tschinkel (1981) introduced newly inseminated Solenopsis invicta queens into groups of workers with differing numbers, the workers in the first brood increased in average size according to the number of attending workers in the adoptive group. Yet none of these sets of offspring were as small in average size as the nanitics of normal incipient colonies.

When the first brood of workers reaches the adult stage, the new colony undergoes a radical transformation. If the queen has been conducting the ordinary chores of the colony, she now stops in order to devote herself exclusively to egg-laying. The workers take over all of the remaining tasks, including the feeding of the queen herself. For a few worker generations, the average number of which varies among species, no new reproductive forms are reared. Also, with the exception of raiding species such as Myrmecocystus mimicus and Solenopsis invicta, few if any reproductive individuals or alien workers are adopted from the outside. Thus the colony is a semi-closed system devoted to its own exponential growth (Brian, 1957b, 1965b, 1983; Wilson, 1971; Oster and Wilson, 1978). The colony can in fact be viewed in this middle, ergonomic stage as a growth machine: its hypothesized "purpose" is to proliferate workers as quickly and safely as possible. The growth function is implemented chiefly by division of labor--the right number of foragers to harvest the surrounding terrain, the right number of nurses to stoke larval growth, and a sufficient but not excessive number of defenders and auxiliaries standing by for emergencies. The focus of the colony is not yet reproduction or dispersal. New nest sites are sought only when the old ones become environmentally untenable or too small to hold the expanding colony.

During the ergonomic stage competition within the colony is at a minimum (see our analysis of competition in Chapter 6). However, the beginning of the stage, or more precisely the transition to this stage from the preceding, founding stage, is sometimes accompanied by hostile interactions. In Lasius flavus, Messor pergandei, Myrmecocystus mimicus, and Solenopsis invicta, pleometrotic laboratory groups revert to monogyny when the first brood or mature workers appear. The Lasius queens fight with one another and then break apart into single-queen units (Waloff, 1957). Those of Myrmecocystus form dominance hierarchies, with the supernumerary individuals eventually being driven out by the workers (Bartz and Hölldobler, 1982). When multiple Solenopsis invicta queens are introduced to queenless workers, the latter usually execute all but one, both in the laboratory and under natural conditions (Wilson, 1966; Fletcher and Blum, 1983; Tschinkel and Howard, 1983). In the carpenter ants Camponotus herculeanus and Camponotus ligniperda, large colonies often contain several queens, but these individuals are intolerant of one another and maintain territories within the diffuse nests, a condition referred to by Hölldobler (1962) as oligogyny. The same phenomenon occurs in the Australian meat ant Iridomyrmex purpureus: queens that cooperated amicably during nest founding become antagonistic after the first workers appear, and in the end permanently separate within the nest (Hölldobler and Carlin, 1985).

If the colony survives the precarious period during which the first and second worker broods are being reared, it is likely to enjoy an interval of sustained exponential growth. But this growth, like that in all populations, can be expected to slow with time and eventually to come to a halt. Most data on the course of colony growth in social insects generally suggest curves that are roughly sigmoidal (hence "logistic") in form (Brian, 1965b, 1983). In a recent, thorough study of the fire ant Solenopsis invicta, Tschinkel (1988a) was able to show that colonies under natural conditions grow logistically, attaining the maximum worker population of about 220,000 in 4 to 6 years. This is the expected result, but the underlying density-dependent controls are more complex than those determining typical logistic growth in nonsocial insects. The theory of colony growth, based on the concepts of economies of scale and evolutionary optimization, has been developed in some detail by Oster and Wilson (1978). We will return to the subject repeatedly in later chapters as part of our analysis of caste, division of labor, foraging strategies, and defense.

Reproductive stage
If a monogynous colony were to maintain its worker population at zero population growth, it would be unable to reproduce, since total investment means by definition that no production of virgin queens and males is possible. Consequently, at some point short of its maximum possible size, the colony should devote part of its production to the creation of virgin queens and males. The timing of the conversion varies among species according to the special adaptations the species have otherwise made to their environments (Wilson, 1971; Oster and Wilson, 1978). Within species, the production of reproductives increases as a function of colony size. Among species of the genus Myrmica at least, the average worker stature (related to colony size) also has a particularly strong influence on queen production but not male production (Elmes and Wardlaw, 1982). It is a common event, as documented in Myrmica rubra by Brian (1957a,b), for males to appear in the nest prior to the females--and sometimes at erratic intervals, possibly as a consequence of workers laying eggs in competition with the queen. Brian has termed the interval of early male production the "adolescent" period of colony growth, coming between the "juvenile" (= ergonomic) period during which the worker population expands and the "mature" period during which new, virgin queens are produced. In truth, very little information has been published on the timing of these key events in the colony life cycle, so it is impossible to generalize about the sequence in which males and queens appear. In many, perhaps even most ant species, some colonies rear only males in a given season, others rear only queens, while still others rear a mixture (Nonacs, 1986a,b). Whether individual colonies change their strategy from one year to the next is not known. Table 3-2 gives most of the available data on colony size in various species of ants. A few additional data have been compiled by Baroni Urbani. These raw numbers tell us nothing directly about the growth rates or factors limiting mature colony size. Also, the numbers often represent underestimates of the typical mature colony size, because colonies of all ages were censused and colonies in many wild populations tend to be young. Nevertheless, the data do permit some inferences when comparisons are made between major groups:

1. No apparent correlation is yet evident between the size of the mature colony and the longevity of the colony, at least as measured in monogynous species by the life span of the queen (see also Table 3-3). However, the data are so few, especially of colony life spans, that it would be premature to draw any firm conclusion. There is a clear need for more longevity studies of ants of all castes because of the relevance of such data to population dynamics.

2. There is no clear relation between climate and colony size. If anything, temperate species tend to have somewhat larger colonies on the average. This is due to a special ecological effect connected with constraints in the nest site of many tropical species, as follows.

3. There is a strong relation between preferred nest site and mature colony size. Among the ants of New Guinea rain forests, for example, species that nest in rotting logs and other pieces of decaying wood on the ground (almost all Ponerinae and the majority of Myrmicinae) form smaller colonies than those living in less restricted nest sites, such as the open soil of the forest floor (Acidomyrmex, Pheidologeton, Leptomyrmex, Pseudolasius, Acropyga, most Paratrechina), open air at the ground surface (Aenictus), and various parts of the tree canopy (most Crematogaster, Iridomyrmex, Camponotus, Polyrhachis, and Oecophylla).

4. The most elaborate caste and communication systems occur in species with large, perennial colonies, for example the legionary ants (Dorylus, Eciton), leafcutter ants (Acromyrmex, Atta), and the marauding ants of the genus Pheidologeton.

5. The great variation in colony size among species belonging to the same taxonomic group (for example, the Myrmicinae) attests to the capacity of this population trait to evolve with relative speed. Small alterations in the physiological parameters of individual ants such as mean worker life span and thresholds in queen determination of larvae can bring about major differences in mature colony size. Given this potency to adapt colony size to local environmental conditions at the species level, we should feel encouraged about the possibility of inferring which conditions have been critical in evolution.

Brood care and larval reciprocation
The workers of all ant species thus far investigated lavish care on all of the immature stages, from egg to larva to pupa. This is true even of Nothomyrmecia and Amblyopone, both of which are very primitive anatomically and have the simplest social organizations known in the ants (Figure 3-18). In the case of the North American species A. pallipes, Traniello (1982) recorded the following behaviors by the workers on behalf of the brood: lick and carry all stages, place larva on prey, assist larval molt by licking ecdysial skin free, bank mature larva with soil to facilitate cocoon spinning, assist the removal of the meconium at the commencement of pupation, and remove empty cocoon after eclosion. All of these behaviors have also been observed in Nothomyrmecia (Hölldobler and Taylor, unpublished; Figure3-19).

The workers of most species belonging to the phylogenetically more advanced subfamilies Myrmicinae, Aneuretinae, Dolichoderinae, and Formicinae engage in "trophallaxis" (Wheeler, 1918a). That is, they regurgitate liquid food to other members of the colony, including the larvae. In some species, such as Formica sanguinea or Solenopsis invicta, the larvae elicit the response by rocking their heads back and forth, flexing their mandibles, and "swallowing" rapidly (Hölldobler, 1967a; O'Neal and Markin, 1973). In the phylogenetically more primitive subfamilies Ponerinae, Nothomyrmeciinae, and Myrmeciinae the existence of oral trophallaxis is less certain. Reports have been published of occasional trophallaxis among workers of Nothomyrmecia macrops (Taylor, 1978), and those of some species of the Ponerinae and Myrmeciinae (Le Masne, 1953; Haskins and Whelden, 1954; Freeland, 1958). However, these observations have to be confirmed by experimental evidence (Hölldobler, 1985).

Myrmecologists have long believed that special nutrients needed for the rearing of reproductive females are produced in one or the other of the exocrine glands in the head or thorax of the worker. The two most frequently mentioned by past authors have been the labial and postpharyngeal glands (see Figure 7-2). Early experimental investigations on the two glands in Formica polyctena were contradictory. Gösswald and Kloft (1960a,b), using radioactive phosphorus as a tracer, concluded that the labial gland is the principal source of nutrition. In contrast Naarmann (1963), using a similar technique, assigned the nutritive role to the postpharyngeal gland.

The discrepancy was resolved by more detailed studies of the labial, propharyngeal, and postpharyngeal glands in Formica polyctena by Paulsen (1969). It is first worth mentioning that the propharyngeal gland of ants is homologous to the pharyngeal gland in honeybees (Emmert, 1969). In bees it is the source of important nutritional substances used in the nursing of larvae (Rembold, 1964; Weaver, 1966). Paulsen found no indication, however, that the propharyngeal gland serves the same function in Formica polyctena. Instead, as suggested earlier for Camponotus herculeanus by Ayre (1963a), it is evidently a source of digestive enzymes. The postpharyngeal gland, as then noted by Paulsen, is found only in ants, and it occurs in all adult castes, including the queens, workers, and males. The organ is glove-shaped, with two symmetrical halves terminating in a varying number of finger-shaped projections. Its lumen is filled with a complex mixture of lipids. When Paulsen injected triolein labeled with radioactive carbon into the hemolymph of "storage workers," which have a large abdominal fat body, radioactivity accumulated in the postpharyngeal gland but not in the propharyngeal gland or labial gland. Some of the labeled material was subsequently transported to other workers as well as queens and workers. In addition, Paulsen found that larvae of Formica polyctena collected in the field had droplets in their gut microscopically identical to those in the postpharyngeal gland lumen of the nursing workers.

These and subsequent investigations suggest that the postpharyngeal gland is the source not only of basic nutrients but also of "profertile substances" (Gösswald and Bier, 1953) used by workers in the determination of reproductive females. The fundamentally nutrient role of the gland is supported further by the following findings:

(1) There exists a strong positive correlation between the secretion of liophilic substances in the postpharyngeal gland and the storage of lipids in the fat bodies of Formica workers.

(2) Formica workers store lipids in their fat body during the autumn and metabolize them following the winter dormancy (Kirchner, 1964). There is a parallel increase in the volume of the cell nuclei in the glandular epithelium of the postpharyngeal gland in the autumn and a decrease in size in the late spring and summer. The nuclear volume is believed to be an index of secretory activity (Bausenwein, 1960).

(3) A similar association has been found in freshly eclosed workers, which have larger fat bodies and postpharyngeal gland nuclei than those of older workers (Bausenwein, 1960).

The caste-determining connection involving profertile substances was made as follows: Bier (1958a) found that only recently eclosed workers or young workers that served as storage workers during winter dormancy have the ability to rear reproductive females. Data provided later by Paulsen (1969) strongly suggest that this capacity is due to the high secretory activity of the postpharyngeal gland in these workers.

How general is the nutritive role of the postpharyngeal gland in ants? After demonstrating that dyes fed to Messor workers are transferred directly from the pharynx into the lumen of the postpharyngeal gland, Barbier and Delage (1967) postulated that the postpharyngeal gland is an organ for digestion and nutrient resorption. However, in later studies one of the authors (Delage-Darchen, 1976) demonstrated that materials from the postpharyngeal lumen serve as nutrients and are fed to the larvae. Markin (1970) provided circumstantial evidence that the postpharyngeal gland of the Argentine ant Iridomyrmex humilis, a dolichoderine, produces important nutrients fed primarily to the queens and small larvae.

To summarize, a substantial amount of evidence points to a nutritive function of the postpharyngeal gland across at least three of the major ant subfamilies, the Myrmicinae, Dolichoderinae, and Formicinae. In the case of Formica in particular, both Naarmann (1963) and Paulsen (1969) concluded that the contents of the gland are first transferred to the crop, where they are mixed with other foods, and then regurgitated to larvae and adult nestmates.

This is not, however, the whole story. Recent work on the biochemistry of the postpharyngeal gland in fire ants (Solenopsis) indicates that the organ serves additional, still unknown functions in at least some species. Attygalle et al. (1985) found that the postpharyngeal gland of Solenopsis geminata is hypertrophied, forming four huge lobes that completely fill the upper part of the head cavity. The glandular secretion is an oil containing a complex mixture of hydrocarbons, of which the major constituents are heneicosane, tricosane, and tricosene (linear C21 and C23 hydrocarbons). Interestingly, these substances differ in proportions from one worker to the next, for reasons that have not been explained. Thompson et al. (1981) showed that the postpharyngeal gland of Solenopsis invicta queens also contains hydrocarbons as major components (63 percent), accompanied by triglycerides (32 percent) and some free fatty acids. The hydrocarbons are all methyl-branched and include 13-methylheptacosane, 3-methylheptacosane, 13,15-dimethylheptacosane, and 3,9-dimethylheptacosane. When these queen-derived substances were presented to a Solenopsis invicta colony, workers clustered around them. On the other hand, Attygalle et al. failed to obtain a similar response in Solenopsis invicta with the contents of the worker postpharyngeal gland.

The labial gland, unlike the postpharyngeal gland, is able to discharge its contents directly through an opening on the labium (see Figure 7-2). Data from Ayre (1963a) on Camponotus and from Delage (1968) on Messor indicate that the labial gland serves primarily as a source of digestive enzymes, and amylase in particular. This sugar-generating role of the organ was confirmed in Formica polyctena by Paulsen (1969), who found that the labial gland secretions are rich in glucose (0.46M). This is not all. When Paulsen injected radioactivity labeled glucose or fructose into the hemolymph of workers, he detected radioactivity 24 hours later in the labial gland but not the postpharyngeal gland. Paulsen also found that the labial gland secretions, which are mostly carbohydrates obtained from resorbed food or mobilized reserves from the fat body, are fed to queens, workers, and larvae. The labial glands are active in older workers and their secretory activity does not display the same strong annual cycle characterizing the postpharyngeal gland.

Stomodeal trophallaxis by larvae has been observed in nearly all of the major ant subfamilies (Wheeler, 1928; Le Masne, 1953; Haskins and Whelden, 1954; Maschwitz, 1966). The single exception is the Dolichoderinae, a negative generalization seemingly reinforced by the fact that Athias-Henriot (1947) found no evidence of secretory tissue in the labial glands of the dolichoderine species Tapinoma nigerrimum. The workers also actively seek the stomodeal fluid, since they lick the head region more frequently than the remainder of the body and occasionally employ the same kind of antennal stroking used to offer food to larvae (Stäger, 1923; Le Masne, 1953). The larval secretions of at least one species, Leptothorax curvispinosus, are avidly consumed by queens who "graze" from one larva to another (Wilson, 1974a), while final instar larvae of the Australian harvesting ant Monomorium (= Chelaner) rothsteini convert seeds given to them into secretions that are fed back to the workers (Davison, 1982). The origin of these liquids is not known, but the most likely source is the paired salivary glands, the only well-developed exocrine glands that open into the mouth. Wheeler (1918a) pointed out that these glands are hypertrophied in the myrmicine Paedalgus termitolestes, and he suggested that they serve to produce secretions attractive to the workers. Wheeler also speculated that the unusual thoracic abdominal appendages of larvae of the pseudomyrmecine Tetraponera (= Pachysima), which he called "exudatoria," produce liquids that attract and bind the affection of workers (see Figure 3-20). Similar structures are found in certain species of Crematogaster, including Crematogaster rivai (Menozzi, 1930).

The larvae of the migratory ant Leptanilla japonica have a specialized duct organ on each side of the third abdominal segment. Masuko (1987; and personal communication) demonstrated that adult ants imbibe larval hemolymph directly through this organ. The queen in particular seems to feed exclusively on larval hemolymph. In addition, Masuko (1986) recently found that larval hemolymph provides an important source of nutrients for the queens of Amblyopone silvestrii. In this case, however, no special organ is involved. The queen punctures the larval skin and imbibes hemolymph from the bleeding wound.

Wheeler's mutual attraction hypothesis was put forth as evidence favoring the Roubaud-Wheeler Theory of the origin of insect sociality through trophallaxis. As beguiling as this idea may be, it has not yet been subjected to critical experimental investigation. Maschwitz (1966) found that the stomodeal contents of Tetramorium larvae have much higher concentrations of amino acids than their hemolymph. He concluded that the stomodeal fraction "very probably originates from the salivary secretion," and thus the secretion must have nutritive value. Sorenson et al. (1983) obtained the same result in larvae of the fire ant Solenopsis invicta; the concentration of amino acids, to be precise, is two to three times greater in the oral secretions of the larvae than in their hemolymph. On the other hand, the total protein concentration in the oral secretions is only 60-70 percent that in the hemolymph. The material consists substantially of amylases and proteinases, which are used (in the final instar) partly outside the body to digest solid food placed below the head by the nurse workers (Petralia et al., 1980). At the present time, almost no information at all exists on the nature and function of the abdominal exudatoria or other possible glandular tissue located elsewhere on the bodies of larvae, with the single exception of the abdominal glands of Leptanilla just mentioned.

Ant larvae periodically produce yet another kind of liquid from the anal region. The workers often solicit this material, which appears likely to contain waste substances and to originate from the Malpighian tubules, by stroking the tip of the larva's abdomen with their antennae (Le Masne, 1953; Ohly-Wüst, 1977). In the case of Solenopsis invicta, O'Neal and Markin (1973) have distinguished two forms of the anal liquid. One is milky and sought by the workers, which gently squeeze the posterior sternites up to five times in a row in an attempt to obtain it. Produced by the late second instar as well as third and fourth (final) instars, the milky liquid excites and attracts nearby workers, and it appears to serve as a supplemental food for them. The second kind of liquid is clear and less viscous and never solicited. Whenever a droplet is extruded, a passing worker picks it up in half-opened mandibles, carries it to the edge of the nest, and deposits it. This material, unlike the milky liquid, appears to consist wholly of true excreta.

The possibility that larvae serve as specialized digestive castes has received substantial support from a series of studies on ants belonging to the subfamily Myrmicinae. In a pioneering analysis of the European harvester Messor capitatus, Delage (1968) showed that the stomodeal secretions of larvae (presumably originating in the salivary gland) contain lipases and proteases, but no carbohydrases. In contrast, workers produce carbohydrases in their cephalic glands (including the labial gland, which extends into the thorax), but no proteases. Delage therefore proposed that the larvae provided proteases to the workers while receiving carbohydrases in return. Messor feeds heavily on seeds, and such a mutualistic exchange between adults and larvae might contribute to a more efficient digestion of the starch and other carbohydrates.