The Ants Chapter 10

CHAPTER 10. FORAGING STRATEGIES, TERRITORY, AND POPULATION REGULATION

Ant colonies exploit the environment by wholly social means. They deploy aging workers into the field as expendable probes to blanket the terrain around the nest in a near continuous and instantaneous manner. In many species scouts that encounter a large food supply recruit nestmates in numbers appropriate to the size and richness of the discovery. When they meet enemies, they can recruit defenders or else withdraw, closing the nest entrance and lasting out the danger with the aid of food that has been conveniently stored in the crops and fat bodies of the younger workers.

In the formal framework of a theoretical ecologist an ant colony resembles a plant as much as it does a solitary animal. It is sessile, in other words rooted to a nest site. Its growth is indeterminate, so that under stress it can revert to the size and caste composition of younger colonies. As a consequence the size and age of colonies are only weakly correlated. As Rayner and Franks (1987) have pointed out, close parallels also exist between ant colonies and mycelia of fungi. “Both are collectives of genetically related or identical semi-autonomous units, consisting respectively of discrete multicellular individuals and hyphae.” The patterns created by the growing hyphae are remarkably similar in geometric design to the foraging columns and swarms of army ants and trunk trail systems of wood ants. In both cases the probes of genetically different cultures avoid one another, creating an interdigitating pattern that thoroughly exploits the available resources.

To continue the comparison, many ant colonies adjust to environmental change like plants more by flexibility of response than by movement from place to place. They attain this adaptability by a broad range of social behaviors, including adjustable alarm and recruitment, differential clustering, alterations of nest architecture, and shifts in foraging pattern.

Foraging theory
Foraging theory can be said to have begun, at least in mathematical metaphor, with Emlen (1966) and MacArthur and Pianka (1966). It has been extended greatly by Schoener (1971), Pyke et al. (1977), Orians and Pearson (1979), Stephens and Krebs (1986) and others. The special case of social insects has been theoretically treated by Oster and Wilson (1978) and advanced by many investigators through excellent field and laboratory studies.

Foraging theory is composed mostly of optimality models. The marginal value utilized, or “currency,” is usually energy. The modelers visualize the organism as seeking to maximize its net energetic yield. Evolution by natural selection, they reason, has modified behavior in one or the other of four basic ways so as to increase this currency: (1) choice of food items (optimum diet); (2) choice of food patch (optimum patch choice); (3) allocation of time invested in different patches (optimum time budget); and (4) regulation of the pattern and speed of movement. In the case of ants and other animals with a nest or other permanent retreat, the foraging patterns are analyzed in terms of sallies from a central location. This special case is the subject of “central-place foraging theory,” about which we will say more later.

Optimal foraging patterns are generally thought to be constrained by two forces, one external to the organism and the other internal. The external constraint is mortality due to accidents, disorientation, and attacks by predators and competitors. Many field studies have shown that worker populations of ant colonies are subject to terrible losses through predation and battles with competing colonies. It is also clear that the foraging patterns of individual species are profoundly influenced by these pressures. The internal constraints, on the other hand, are particular physiological constraints and the limited sensory and psychological capacities of organisms. Animals, and especially small-brained creatures like ants, can rarely if ever maximize their net energy yield by calculating exact costs, benefits, and mortality risks for each new situation in turn (of course, even human beings are hard put to accomplish that much). Instead, they are more likely to employ “rules of thumb,” which are quick decisions triggered by relatively simple stimuli or stimulus configurations. The rules of thumb work adequately for individual worker ants most of the time, and they may result in considerable precision when added up across multiple workers during the process of mass communication. Rules of thumb used by fire ants that we introduced earlier to illustrate colony homeostasis (Chapter 9) include the following: “continue hunting for a certain foodstuff if the present foraging load is accepted by nestmates” (it will probably satisfy the needs of the colony); “follow a trail if enough pheromone is present” (it will probably lead to food); and “feed the queen more if final-instar larvae are present” (this means that the colony developmental pipeline is in good working order).

In adapting foraging theory to the social insects, Oster and Wilson (1978) treated risk of worker mortality not as a constraint but as an energy cost that must be born to varying degrees by the colony. When a worker is killed in action, the game is far from over; the colony has merely lost a packet of energy that had been invested earlier through the processes of egg laying, larva rearing, and pupa care. The ultimate pay-off in colony-level selection is the summed production of reproductives over a colony generation. In general, ant colonies appear to act so as to build as large a population of workers as possible during the ergonomic phase, then “cash in” with a partial or full conversion to reproductives during the breeding season. Consequently, a reasonable measure of fitness for a colony in its ergonomic phase of growth is energy, with the growth rate of the adult biomass expressed as calories/unit-of-time and the full colony biomass achieved at the end of the ergonomic phase expressed in net caloric accumulation. The analytic advantage of employing energy as the basic currency is that it can be folded one way or the other into all colony processes. The key particulars are as follows:

• The biomass and biomass growth of the colony or any portion of it can be translated to energy units.

• Foraging success, brood care, and other nurturent activities can be translated into energy gained.

• Protective activities, including nest defense and construction, can be converted into calorie equivalents saved because of the reduction in mortality and impaired colony functions that such activity avoids.

• Metabolism and mortality can be converted directly into caloric loss.

It is of course an oversimplification to measure ergonomic success entirely by the energy equivalents of biomass. Reproductive success depends not just on colony size at the end of the ergonomic phase but also on the proportion of immature and mature forms as measured by the relative frequency of castes at this critical time. It also depends on the forms and kinds of “capital” at the disposal of the colony, particularly the amount of stored resources and the structure and location of the nest. Nevertheless, energy equivalence is probably the overriding determinant of reproductive success. The optimal caste frequency distribution and nest structure at the onset of the reproductive phase are unlikely to differ from those built up during the ergonomic phase, while few species store large amounts of food materials near the end of the ergonomic phase. In addition to such an audit of the total life span, analyses of particular components of foraging appear to be energy-efficient in the case of ants. As we shall see shortly, the time spent selecting food items and the degree of selectivity are consistent with the models of central place foraging theory, which are based on the postulate of net energy maximization. Also, in the case of leafcutting ants, net energy yield appears to be the key function among all those that might have influenced the evolutionary choice of leafcutting ant medias as foraging specialists over minors and majors (Wilson, 1980b).

Foraging behavior has two principal interlocking components: the search and retrieval of food items, often accompanied by the recruitment of nestmates, and the avoidance or defeat of enemies. It is feasible in theory at least to assess the relative importance of the two components by equating them in terms of energetic cost and gain.

The temperature-humidity envelope
Every ant species operates within ranges of temperature and humidity that can be depicted as a two-dimensional space; in other words, every species has a temperature-humidity envelope. Also, the tolerance of an isolated foraging worker is very different from that of an entire colony. The forager is more swiftly affected by the preexisting ambient temperature and humidity of the microenvironment, whereas the colony can control the microenvironment by shifting into deeper reaches of the nest or by the retention of metabolic heat and moisture through clustering. Thus no nuptial flights of the fire ant Solenopsis invicta occur when morning temperatures of the soil (from the surface to 10 cm deep) are below 18°C, and colony founding by newly inseminated queens occurs only if the soil temperatures at 5 to 7 cm below the surface are equal to or greater than 24°C. In addition, young colonies survive winter poorly even in southern Mississippi (Rhoades and Davies, 1967; Markin et al., 1973, 1974). As a consequence the imported fire ant appears already to have reached its geographic range limit in the northern part of the coastal states. To the west, the colonies are likely to be able to spread across desert areas only by colonizing stream beds, irrigated agricultural fields, and urban areas. Foraging fire ant workers, on the other hand, have even less tolerance for extreme cold, heat, and low relative humidity, and they are unable to survive in ambient conditions in which colonies as a whole do well (Francke and Cokendolpher, 1986). The main point to consider is that if fire ants were solitary insects, the geographic distribution of the species would be far more restricted than it is in fact.

A large portion of the available data on temperature tolerances by foraging workers is summarized in Table 10-1. Not surprisingly, a rough correlation exists between the tolerances and the environment, with desert species at the upper end and some (but not all) species from cold temperate forests at the lower end. The extreme thermophiles of the world fauna belong to the myrmicine genera Ocymyrmex of southern Africa and Cataglyphis of North Africa and Eurasia. The two are biogeographic vicariants of each other. In particular, both are specialized in similar manner for diurnal foraging on extremely hot desert terrain. The workers hunt singly for the corpses of insects and other arthropods that have succumbed from the heat (Harkness and Wehner, 1977; Marsh, 1985a,b; Wehner, 1987). Their behavior and even outward physical appearance are remarkably similar. Ocymyrmex barbiger forages in the sun at surface temperatures up to 67°C, which must be close to the record for insects generally. Cataglyphis fortis is a specialist of the extremely hot, dry, and food-impoverished terrain of the Saharan salt pans. At the opposite end, the cryophilic (cold-loving) species Camponotus vicinus and Prenolepis imparis start foraging at just above freezing and cease when the temperature reaches about 20°C. Similarly, the Australian species Nothomyrmecia macrops forages exclusively after dusk, and the workers seem to be more active at times when temperatures are low (5-10°C). Hölldobler and Taylor (1983) suggested that low temperatures hamper the escape of potential prey encountered in the tree tops increasing the hunting success of the Nothomyrmecia foragers (Plate 12).

Most diurnal desert ants pass through two periods in the day when their preferred temperatures occur, in the morning and late afternoon. As a consequence they have a bimodal distribution of foraging activity, with a period of decline around midday. In mesic temperature habitats, such as northern coniferous woodland, the reverse pattern occurs among diurnal species, with foraging peaking in the middle of the day. These patterns are probably determined by external environmental cues. Hunt (1974) was able to shift the desert bimodal pattern of the Chilean dolichoderine Dorymyrmex antarcticus to a mesic unimodal pattern merely by shading the nest at midday.

While the temperature tolerances of ant species are correlated with climate and major habitat, the relation is only loose. Added variance comes from two sources, microhabitat specialization and competition avoidance. Forest species, such as Aphaenogaster rudis and Paratrechina melanderi, have lower tolerance than other species, such as Monomorium minimum and Tetramorium caespitum, which are adapted to clearings in the same immediate area. It is also of advantage for ants to use unusual activity regimes to escape competitors. The thermophilic species of Cataglyphis and Ocymyrmex have the desert terrain virtually to themselves at midday. In the grassy meadows of Massachusetts, the little black ant Monomorium minimum recruits at higher temperatures than does its three closest competitors, Lasius neoniger, Myrmica americana, and Tetramorium caespitum. It combines this specialization with an effective form of venom dispersal to appropriate a substantial fraction of dead insects and other food finds (Adams and Traniello, 1981).

The subtle differences in microhabitat that characterize species are very well illustrated in Formica perpilosa and Trachymyrmex smithi, two ant species common in the Chihuahuan desert of southern New Mexico. As illustrated in Figure 10-1, Formica perpilosa is active in a much broader area around its nest than is T. smithi. The reason is that the [[Formica'' are more resistant to water loss, and they also tend to forage on mesquite and other forms of vegetation, where the relative humidity is higher (Schumacher and Whitford, 1974).

Quantitative studies of humidity preference by foraging ants of the kind conducted by Schumacher and Whitford are rare in the literature, but abundant anecdotal evidence exists to show that the higher the relative humidity, the greater the temperature tolerance. An increase in foraging activity when humidity rises at high temperatures has been observed in Pheidole militicida (Hölldobler and Möglich, 1980), Formica polyctena (Rosengren, 1977b), Prenolepis imparis (Talbot, 1943b, 1946), and a wide range of semi-desert ant genera in Australia (Briese and Macauley, 1980); and it is probably a general phenomenon in ants as a whole. Another source of differential foraging in the face of varying humidity is simple physiological resistance to desiccation, a phenomenon that has been observed in Tetramorium caespitum by Brian (1965b), Pogonomyrmex by Hansen (1978), and diurnal ant species in Australia by Briese and McCauley (1980).

On the other hand, rain halts most foraging in places where the drops pelt the ground and form small puddles and rivulets (Hodgson, 1955; Lewis et al., 1974b; Skinner, 1980a). Every collector is familiar with this phenomenon, to his frequent frustration and especially during tropical wet seasons. On the other hand, there are a few long-legged ants, including the species of the dolichoderine genus Leptomyrmex of Australia and New Guinea and the myrmicine “giraffe ants” Aphaenogaster phalangium of Central America, that use their unusual stature to navigate water films. As a result they are among the first insects to forage after the rain has ceased.

Daily cycles of activity
Each ant species has a distinctive daily foraging schedule. Some are active for only a few hours each day. To take an extreme example, there is a remarkable degree of precision in the changeover of ant species at dusk in the heath of southwestern Australia (Wilson, 1971). In midafternoon, the ground and the branches and leaves of the low bushes that dominate the vegetation contain hordes of workers, mostly brown, red, or black in color and with medium-sized compound eyes, belonging to ten or so species of Myrmecia, Rhytidoponera, Dacryon, Iridomyrmex, and other typically Australian genera. As dusk falls, first one species, then another, begins to pull back into their nests, while the nocturnal species--pale-colored, mostly large-eyed species in Colobostruma, Iridomyrmex, and Camponotus--make their appearance in a regularly staggered succession. So orderly is the changeover that approximately the same number of foraging workers remains on the bushes throughout.

Are such daily cycles based on circadian rhythms, or are they guided hour-by-hour by changes in temperature and other external stimuli? The answer appears to be that circadian rhythms affecting foraging behavior are widespread, but in many ant species they can be overridden--or at least frame-shifted--by colony hunger or certain environmental changes. McCluskey and Soong (1979), for example, established diel rhythms in four species that occur together in southern California. Pogonomyrmex californicus and Pogonomyrmex rugosus, which are harvesters, commence foraging at early midday and remain very active through the afternoon. Two other harvesters, Veromessor andrei and Veromessor pergandei, as well as Formica pilicornis and Myrmecocystus mimicus, which are general insectivores, have bimodal activity schedules, declining toward midday and picking up again from late afternoon through dusk. When colony fragments were placed in the laboratory at a constant temperature with alternating light and dark, they maintained their natural schedules to a significant degree. A similar result was obtained by Rosengren (1977b) with Finnish colonies of the wood ant Formica polyctena. Workers responded to a 12:12 light/dark cycle in the laboratory by a rise in activity toward the end of the dark period, as though anticipating daybreak. The ants could be entrained to new 12:12 cycles out of phase with that occurring in nature. When kept in complete darkness on a 12:12 warm-cool cycle, the workers grew most active in the middle of the warm period. Thus light alone is not crucial for the entrainment of diel cycles. In later work, Rosengren and Fortelius (1986a) extended this result to a variety of species in the exsecta and rufa groups of Formica. They observed a tendency of workers to increase activity before the artificial “dawn,” but there was also a great deal of variation in the overall activity pattern among colonies. Ambiguity is also the rule in the giant ponerine ant Paraponera clavata (Figure 10-2). McCluskey and Brown (1972) observed that workers on Barro Colorado Island emerged regularly at dusk and apparently foraged until dawn. When transferred to an artificial nest in the laboratory they displayed striking peaks of activity at dawn and dusk. But when placed in a dark room with alternating light and dark and constant temperature, they became diurnal rather than crepuscular and nocturnal. In Costa Rica over a period of two years, we observed considerable variation in the diel pattern of activity by Paraponera workers, although these giant ants appear to be primarily nocturnal. The same phenomenon was observed by Harrison and Breed (1987), who succeeded in training Paraponera workers to come to the sites of sugar baits at fixed times during both the day and night.

In fact, many ant species seem quite capable of shifting the time of peak foraging back and forth as an adjustment to vagaries of the environment. We have seen Atta cephalotes, a dominant leafcutter of the moist lowland forests in Central and South America, change from predominantly diurnal to nocturnal foraging over a period of a few days in the Brazilian Amazonian forest. Lewis et al. (1974a) found that in Trinidad the foraging columns emerge decisively and reach peak traffic within two hours. This event usually occurs at night and lasts for an average of 12 hours; when it unfolds during the day it persists for 7.5 hours on the average. Colonies in the same locality are often unsynchronized, and it seems likely that they are entrained in an independent manner by food discoveries rather than by some more pervasive cue from the physical environment.

This form of adaptive plasticity has in fact been demonstrated in the case of Messor galla and Messor regalis, two granivores inhabiting the savanna of northeastern Ivory Coast (Lévieux, 1979). During the dry season, when seeds are abundant, the colonies forage mostly at night and in tight columns. During the rainy season, or in periods of the dry season with poor seed crops, the ants shift to diurnal foraging and the columns are less organized. It is very common for desert ants to switch from nocturnal foraging in the summer to diurnal foraging in the winter. The phenomenon has been observed in species of Aphaenogaster (Whitford and Ettershank, 1975), Atta (Mintzer, 1979b), Veromessor (Tevis, 1958), Pheidole (Hölldobler and Möglich, 1980), and Pogonomyrmex (Hölldobler, 1976a). Nearly the reverse pattern is displayed by Aphaenogaster rudis and Paratrechina melanderi in the hardwood forests of Maryland: peak activity is reached around midnight during the summer months, shifting to more nearly uniform round-the-clock activity during the spring and summer and quiescence during the winter (Lynch et al., 1980). Authors who have reported these patterns generally consider them to be adaptations to promote thermoregulation. Where rodents can remain strictly nocturnal year-round, at least in deserts, ants are more sensitive to ambient temperature changes and must alter their daily activity accordingly (Davidson, 1977b; Brown et al., 1979a,b).

Differences in foraging rhythms among sympatric species of ants can serve in temporal partitioning of significant resources. Such activity differences might be based proximately on different humidity and temperature ranges tolerated by the species, yet be an ultimate, evolutionary result of interspecific competition. Although there exist only a few studies which directly address this question, some of the observations reported by Talbot (1946), Greenslade (1972), Bernstein (1979b), Swain (1977), Klotz (1984), and other strongly suggest temporal partitioning of resources. Klotz reports that in his study area in Kansas both Camponotus pennsylvanicus and Formica subsericea utilize the same aphid honeydew sources, but in a temporally displaced manner. Camponotus is more active at cool temperatures and predominantly nocturnal, while Formica is active at higher temperatures and is primarily diurnal. We made similar observations on Camponotus socius and Camponotus floridanus in central Florida. Where both species coexisted and used the same honeydew sources, socius was primarily diurnal and floridanus primarily nocturnal.

The approximately 8,800 known ant species use a dazzling variety of procedures for the discovery and retrieval of food. Several authors have attempted classifications of this important category of behavior, including Oster and Wilson (1978), Passera (1984), and Moffett (1987b). We will provide a synthesis here of what we regard as the most useful elements of these schemes. The resulting classification breaks all of the phenomena into three categories: hunting (3 kinds), retrieving (4 kinds), and defense (4 kinds). The elements in each can be combined to form 3 x 4 x 4 = 48 possible three-state foraging techniques. This arrangement provides an informal and convenient framework for the description of the behavior of individual species.

Hunting: (1) by solitary workers; or (2) by solitary workers directed to specific trophophoric fields by trunk trails; or (3) by groups of workers searching in concert, in the manner of army ants.

Retrieving: (1) by solitary workers who return home on their own; or (2) by individual workers who return home along persistent trunk trails; or (3) by individual workers recruited to the food sites by scouts; or (4) by groups of workers who gang-carry the food items.

Defense: defense by guard workers during hunting, or an absence of such defense; and either defense by guard workers during harvesting and retrieval of food, or the absence of such defense (altogether, 4 combinations are possible).

Although no complete accounting has been attempted across all of the ants, it is our impression that virtually all of the 48 possibilities in this broad space of foraging techniques is employed by one species or another. At one extreme are the completely solitary foragers, including scavengers and specialized predators and in all subfamilies other than the army ants of the subfamilies Dorylinae and Ecitoninae. The workers hunt singly and retrieve food items entirely on their own. At the other extreme are the army and driver ants of the genera Eciton and Dorylus, which hunt in groups, often gang-retrieve large prey items, and deploy specialized guard workers along the perimeters of the foraging columns.

The total foraging strategy of an ant species often comprises two or more three-state techniques, each chosen according to the quality and nature of the food of the moment. For example, when the solitary foragers of Pogonomyrmex encounter a seed or dead insect small enough to be carried singly, they carry it home alone. When the object is too large to be transported by one ant, or there is a cluster of small food items, the worker lays a recruitment trail with secretions from its poison gland. When the food is persistent at a particular site, as is the case for a continuing seedfall, the ants deposit trunk trails with secretions that come at least in part from the Dufour's gland (Hölldobler and Wilson, 1970; Hölldobler, 1971b; Hölldobler, 1976a). The minor workers of the Amazonian myrmicine Pheidole embolopyx also search in solitary fashion and single-load the small insects they encounter. When the food item is too large to move, the scouts recruit other minor and major workers from the nest. The majors spend most of their time carving up the insect and defending it from intruders, while the minors concentrate on drinking the hemolymph and carrying fragments back to the nest (Wilson and Hölldobler, 1985). Minor workers of very small Pheidole dentata colonies spend more time feeding away from the nest and transferring the food to nestmates within the nest, as well as less time recruiting, than do minor workers in larger colonies. Evidently the effort of a single worker can more nearly satisfy the requirements of the small colony (Burkhardt, 1983).

Recruitment thus permits a multiple strategy of foraging and permits the exploitation of a wider range of food items. When workers of Monomorium, Myrmica, Formica, and Lasius forage singly, prey size is positively correlated with worker size. But when cooperative foraging based on recruitment is used, the correspondence breaks down. The ants are able to handle items far larger than their individual size (Traniello, 1987a). Whereas the workers in Traniello's study varied in fresh weight from 0.1 to 6.8 mg, their arthropod prey varied from 1 mg to over 2 grams, a range of 2000 times.

Solitary foragers
The approximately 8,800 known ant species use a dazzling variety of procedures for the discovery and retrieval of food. Several authors have attempted classifications of this important category of behavior, including Oster and Wilson (1978), Passera (1984), and Moffett (1987b). We will provide a synthesis here of what we regard as the most useful elements of these schemes. The resulting classification breaks all of the phenomena into three categories: hunting (3 kinds), retrieving (4 kinds), and defense (4 kinds). The elements in each can be combined to form 3 x 4 x 4 = 48 possible three-state foraging techniques. This arrangement provides an informal and convenient framework for the description of the behavior of individual species.

Hunting: (1) by solitary workers; or (2) by solitary workers directed to specific trophophoric fields by trunk trails; or (3) by groups of workers searching in concert, in the manner of army ants.

Retrieving: (1) by solitary workers who return home on their own; or (2) by individual workers who return home along persistent trunk trails; or (3) by individual workers recruited to the food sites by scouts; or (4) by groups of workers who gang-carry the food items.

Defense: defense by guard workers during hunting, or an absence of such defense; and either defense by guard workers during harvesting and retrieval of food, or the absence of such defense (altogether, 4 combinations are possible).

Although no complete accounting has been attempted across all of the ants, it is our impression that virtually all of the 48 possibilities in this broad space of foraging techniques is employed by one species or another. At one extreme are the completely solitary foragers, including scavengers and specialized predators and in all subfamilies other than the army ants of the subfamilies Dorylinae and Ecitoninae. The workers hunt singly and retrieve food items entirely on their own. At the other extreme are the army and driver ants of the genera Eciton and Dorylus, which hunt in groups, often gang-retrieve large prey items, and deploy specialized guard workers along the perimeters of the foraging columns.

The total foraging strategy of an ant species often comprises two or more three-state techniques, each chosen according to the quality and nature of the food of the moment. For example, when the solitary foragers of Pogonomyrmex encounter a seed or dead insect small enough to be carried singly, they carry it home alone. When the object is too large to be transported by one ant, or there is a cluster of small food items, the worker lays a recruitment trail with secretions from its poison gland. When the food is persistent at a particular site, as is the case for a continuing seedfall, the ants deposit trunk trails with secretions that come at least in part from the Dufour's gland (Hölldobler and Wilson, 1970; Hölldobler, 1971b; Hölldobler, 1976a). The minor workers of the Amazonian myrmicine Pheidole embolopyx also search in solitary fashion and single-load the small insects they encounter. When the food item is too large to move, the scouts recruit other minor and major workers from the nest. The majors spend most of their time carving up the insect and defending it from intruders, while the minors concentrate on drinking the hemolymph and carrying fragments back to the nest (Wilson and Hölldobler, 1985). Minor workers of very small Pheidole dentata colonies spend more time feeding away from the nest and transferring the food to nestmates within the nest, as well as less time recruiting, than do minor workers in larger colonies. Evidently the effort of a single worker can more nearly satisfy the requirements of the small colony (Burkhardt, 1983).

Recruitment thus permits a multiple strategy of foraging and permits the exploitation of a wider range of food items. When workers of Monomorium, Myrmica, Formica, and Lasius forage singly, prey size is positively correlated with worker size. But when cooperative foraging based on recruitment is used, the correspondence breaks down. The ants are able to handle items far larger than their individual size (Traniello, 1987a). Whereas the workers in Traniello's study varied in fresh weight from 0.1 to 6.8 mg, their arthropod prey varied from 1 mg to over 2 grams, a range of 2000 times.

Foraging strategies
The extreme category of solitary hunting combined with solitary retrieval is illustrated by the desert scavengers of the genera Cataglyphis and Ocymyrmex (Harkness and Wehner, 1977; Wehner et al., 1983; Schmid-Hempel, 1984, 1987; Schmid-Hempel and Schmid-Hempel, 1984; Wehner, 1987). The total foraging pattern of Cataglyphis bicolor is described in the following striking manner by Rüdiger Wehner in his 1987 review:

When the daily foraging activity of a desert ant's colony is compressed in both space and time by assuming a bird's eye point of view and using a quick-motion camera, an impressive picture emerges: a constant stream of particles radiates out from a centre and spreads evenly over a roughly circular area. Within this field of flow each (individually labelled) particle follows its own radial path, lingers at some distance from the start, and returns to the centre. Some time later, the same particle reappears, follows a similar path, disappears again in the centre, reappears, and continues to do so until at the end of the day the whole stream of particles ceases to flow. Now assume that the time window is expanded from one to several days and the time machine speeded up. One would then become aware of particles disappearing from the scene forever after having performed a number of moves, and new particles taking their place. It is as though one were observing a gigantic slime mold in which cells continually moved in and out from an aggregation centre, but in which the inward and outward movements of the cells were largely desynchronized. Scaling up the scene by about five orders of magnitude and returning to real time, the central-place foraging behaviour of desert ants unfolds in front of one's eyes, and a number of questions about the spatial and temporal organization of this behaviour immediately spring to mind.

The workers of Cataglyphis bicolor (Figure 10-3) make about five to ten forays each day, and an average of 290 foragers of a colony conduct 1500-3000 such expeditions. The entire effort is made during daytime. The ants retreat entirely into the nest at night and often close the entrance with soil. As morning approaches they dig out the entrance again. At sunrise, a worker comes out slowly and stands about for a minute or so within a few centimeters of the entrance. As Harkness and Wehner describe it, the first foragers then begin their runs:

Then one or two more come out and will set off away from the nest in more or less a straight line in some direction, which varies all round the compass with different individuals. Each individual, however, keeps to a constant foraging direction, that is maintained at least over a period of weeks. The frequency of ants leaving is low at first but eventually rises to 100-200 per hour (according to the size of the nest) later in the morning. In half-an-hour or more the first ants begin to come back, usually carrying the corpse or part of the corpse of a dead insect. The arrival of an ant with a burden seems often to be followed by the exit of a group of half-a-dozen or more ants. . . . After an ant returns from a foray, commonly it goes out again in a matter of minutes. By midday there is a constant traffic in and out. Although the frequency of exits and entries stays high for several hours, in general the activity of an individual ant is restricted to a limited period of time that is the same every day and lasts 2-3 hours. Towards sunset the traffic in and out of the nest falls away.

The solitary foragers are guided by sun-compass orientation and the location of bushes and other landmarks, with the latter being remembered as successive frames of configuration (Wehner, 1982, 1987). The foragers tend to persist in only one or a very few directions for their lifetime, if for no other reason than that travel outside the nest is very dangerous and life is short. Most of the workers are soon picked off by spiders and robber-flies, in spite of their ability to run at great speed, up to a meter per second. In southern Tunisia they enjoy a half-life time of only 4.2 days and a life expectancy of 6.1 days (Schmid-Hempel and Schmid-Hempel, 1984). Yet the system is so efficient that the average forager retrieves a food weight during its lifetime 15 to 20 times greater than its own body weight (Wehner et al., 1983).

When occasionally changing direction, both Cataglyphis and Ocymyrmex appear to follow a simple rule of thumb: continue to forage in the direction of the preceding foraging trip whenever the trip proved successful. Otherwise abandon that direction and select a new one at random, but decrease the probability of doing so as the number of previously successful runs increases. Following these rules, Wehner (1987) designed a model that closely described the directional choices actually made by individual foragers. The result was in accord with the observation that colonies exhibiting low foraging efficiencies simultaneously develop low sector fidelities. The overall effect of this method of foraging is that each colony blankets the area around the nest entrance so evenly that maps of foraging activity do not provide a clue to the location of the nest entrance (see Figures 10-4 and 10-5).

As Schmid-Hempel (1984) has pointed out, the rule of thumb used by Cataglyphis and other desert ants creates a kind of polymorphism in the population of foragers. When some of the ants encounter aggregated items, so that they are repeatedly rewarded, they concentrate their efforts. Other workers enter terrain with more scattered resources and disperse their efforts to a corresponding degree. In the ensuing total pattern, the entire foraging domain of the colony is covered efficiently, with a temporary focus on the most productive sites.

The efficiency of deployment is enhanced by “noise” in the repetitive journeys. That is, unpredictable deviations regularly occur away from paths taken on previous sallies. Such errors increase the chance that the workers will strike food items missed in earlier efforts. The same explanation applies to the scatter of fire ants around odor trails, a variance that sometimes allows the workers to locate moving prey (Wilson, 1962b, 1971).

Jacques Pasteels, Jean-Louis Deneubourg, and their co-workers have taken the next step of reasoning and developed a sophisticated theory of error adjustment to optimize net energetic yield during recruitment (Pasteels et al., 1982; Deneubourg et al., 1983; Verhaeghe and Deneubourg, 1983; Champagne et al., 1984; Pasteels et al., 1987). Each species concentrates on food items with a certain pattern of dispersion and size. When the items are extremely clumped and large, say as in the case of a termite colony, the recruitment needs to be very precise, as it is in the case of termitophages Neoponera laevigata and Megaponera foetans. When food is less aggregated and more easily handled by single workers, the error in recruitment is typically larger. Such seems to be the case, for example, in the Messor and Pogonomyrmex harvester ants. Furthermore, the error level may be adjusted within the repertories of individual species in ways that accord with the Pasteels-Deneubourg models. Food items that are small or of poor quality elicit less recruitment pheromone and tactile signaling and more error in the response, leading to a wider exploration of the environment--as in the single-forager strategy of the desert ants (Wilson, 1962b; Hangartner, 1969b; Hölldobler, 1976a; Verhaeghe, 1982; Crawford and Rissing, 1983). The same is true of food sources that are distant from the nest (Wilson, 1962a; Ayre, 1969; Hölldobler, 1976a; Taylor, 1977).

Ortstreue and majoring
Workers belonging to the same colony display a striking degree of specialization in their foraging activity. A growing body of research has revealed the widespread occurrence of individual persistence in foraging site, the phenomenon called Ortstreue. We have seen how Cataglyphis workers return repeatedly to the same approximate area for as long as they are rewarded with food. Similar learning and particularization of behavior has been documented in the ponerine ants Diacamma rugosum, and Neoponera apicalis, both of which are solitary hunting ants (Uezu, 1977; Fresneau, 1985); leafcutter ants of the genus Atta (Lewis et al., 1974a); harvester ants of the genera Chelaner, Messor, Pheidole, and Pogonomyrmex (Hölldobler, 1976a; Onoyama, 1982; Onoyama and Abe, 1982; Hölldobler and Möglich, 1980; Davison, 1982); the carton-building “shining black ant” Lasius fuliginosus of Europe (Dobrzanska, 1966; Hennaut-Riche et al., 1979); the North American species Lasius neoniger (Traniello and Levings, 1986); and ants of the Formica rufa group (Økland, 1930; Jander, 1957; Rosengren, 1971, 1977a; Herbers, 1977; Rosengren and Fortelius, 1986b).

Experiments on Cataglyphis, Formica, and Pogonomyrmex suggest that the memories of location in site fidelity are primarily visual in nature. In extreme cases the learning persists for weeks or months. Rosengren and Fortelius were able to show that in Formica aquilonia at least, olfactory orientation is used to the same purpose in the dark. Specifically, the workers appear to follow scent marks used in home range marking. Thus the Formica evidently depend on a hierarchy of cues of the kind demonstrated in homing Pogonomyrmex workers by Hölldobler (1971b, 1976a). A cue in one sensory modality (such as vision) is used predominantly until it fails, then a cue in a second modality (such as olfaction) is employed, and so on. Pheidole militicida, which forages extensively during the night, also employs primarily olfactory cues in home range orientation and Ortstreue (Hölldobler and Möglich, 1980).

It is reasonable to expect that memory and fidelity to particular sites improve, through natural selection, as food sources become richer and more persistent. Such appears to have been the case in herds of honeydew-producing aphids and other homopterans (Pasteels et al., 1987; Goss et al., 1988). When mistakes are made, so that the ants arrive at the wrong tree, the result can be a substantial loss of time and energy. Just such an unproductive leakage has been documented in Formica lugubris by Sudd (1983), who found that a small percentage of workers wander from a species of pine supporting large populations of aphids to another species with few or no aphids.

Another way to divide the environment among foragers belonging to the same colony is by majoring, or specialization on different kinds of food. It is well known that individual honeybees and bumblebees persist in visiting one species of flower for days at a time, even when the favored blossoms are intermingled with those of other flower species (Heinrich, 1979; Seeley, 1985). A similar phenomenon occurs in ants, although it has been documented less extensively. In field experiments by Rissing (1981a), for example, workers of Veromessor pergandei and Pogonomyrmex rugosus continued to harvest one species of grass seed mixed with two other species, even when larger seeds in the medley were available and being collected by nestmates. The specialization sometimes lasted for several days but occasionally shifted rapidly. The choices made were not just a function of body size, with larger ants selecting larger seeds. This variable accounted for only 4 percent of the variance in seed size. It is not known how the initial selections are made and why preferences change, but the process is far from a random first choice: Messor and Pogonomyrmex workers examine as many as 60 seeds before carrying one home (Hölldobler, 1976a; Rissing, 1981a).

Central place foraging
When a forager is bound to a nest or sleeping site, it faces a different set of problems in harvesting energy than if it merely rested at intervals while conducting a search. How these problems can be solved so as to maximize the net energy yield is the subject of the special set of models that make up “central place foraging theory” (Orians and Pearson, 1979; Schoener, 1979; Stephens and Krebs, 1986). The basic reasoning must be modified to accommodate the social insects, which commit expendable “energy packets” in the form of nonreproductive workers (Oster and Wilson, 1978; Seeley, 1985). The number of workers and hence the energy costs can be fitted through mass communication to the spatial distribution of food items and thus the yield of energy moment by moment. Also, the workers can individually specialize on particular sectors of the terrain and on different kinds of food items, allowing a nearly simultaneous coverage of a wide area.

Most of the predictions made from the foraging models are intuitively clear and possess a double heuristic value in the study of ant ecology. First, they allow a test of the basic proposition that natural selection shapes behavior and, in the case of social insects, does so by acting at the level of colonies. Second, by framing research in terms of these models, the ecologist asks questions and searches for predicted phenomena that might otherwise be easily missed.

Let us now consider some of the predictions and see how they have fared so far in empirical studies. The key overall assumption is that the greater the energy expenditure to get to a patch of food items, the more choosy the animal should be in selecting the item out of the patch to carry home. In other words, the more energy you spend to get there, the larger the energy package you try to bring home. This general conception leads to a series of explicit predictions, as follows:

The greater the distance the ant travels from the nest to a food patch, the longer it should take to select the food item. This prediction was confirmed by Schmid-Hempel (1984) for workers of Cataglyphis bicolor, which spend more time searching around a productive site (where dead insects were found) when the site is farther from the nest. It was also confirmed by the finding of Rissing and Pollock (1984) that workers of Veromessor pergandei take longer to search through more distant patches of seeds once they have arrived on the site. It is not supported by data of Shepherd (1982) on the leafcutter Atta colombica, the workers of which fail to increase their search time on leaf baits when they have traveled farther to get there. However, the Atta data are ambiguous because the large variance in individual times obscured possible real differences in a one-way analysis of variance conducted by Shepherd. For ants generally, then, this test of the basic proposition of energy conservation appears to be supported by the available data, but more experimental studies with different species are needed to make the test convincing.

The greater the distance the ant travels to a food patch, the more selective the ant should be in choosing a food item. The expected relation was obtained by Davidson (1979), who found that the farther from the nest Pogonomyrmex rugosus foragers journeyed, the narrower the range of barley seeds they accepted in a patch, with smaller seeds being consistently more ignored. In similar experiments on Pogonomyrmex occidentalis, Pollock (1977) detected the same trend, but his data were not strong enough to be statistically significant. (Davidson points out that Taylor's tests were conducted over shorter distances than her own.) Rissing and Pollock (1984) discovered that while Veromessor pergandei workers generally take larger seeds from patches, the size of the seed is not correlated with the time spent selecting it and hence the time spent in traveling to the patch. Possible nutritive cues other than size could not be discounted. Overall, the available data appear to favor the energy model, but more studies are clearly needed.

The higher the temperature, the more selective the forager should be in food patches. To our knowledge no experiment has been performed to test this expected correlation. However, at varying temperatures Traniello et al. (1984) have examined the behavior of Formica schaufussi in the pursuit of scattered prey, as opposed to variable prey aggregated in patches. As the temperature rose, the ants became less selective, accepting smaller and less profitable prey. Ants, like other insects, increase their metabolism and hence pay higher energy costs at higher temperatures (Nielsen, 1986). We would therefore expect that if the ants encounter food items in groups, so that they are able to pick and choose, the ants will become more selective at higher temperatures. But if they encounter prey items singly, they will be inclined to be less selective, because “a bird in hand is worth two in the bush.” This is the result obtained by Traniello and his co-workers.