The Ants Chapter 18

AntWiki: The Ants --- Online

CHAPTER 18. THE HARVESTING ANTS

Introduction

Harvesting ants are species that regularly use seeds as part of their diet. They constitute a broad assemblage representing many different evolutionary lines within the subfamilies Ponerinae, Myrmicinae, and Formicinae. They are distinguished from the even broader assemblage of ant species that collect the myrmecochores, which are nutritive appendages fitted like caps or sheaths on the seeds. The foragers discard seeds as soon as they have detached the myrmecochores, which act may occur anywhere from the plant to the nest. Ants are therefore major dispersers of myrmecochorous plants, as we showed in Chapter 14. Harvesting ants, in contrast, feed on the seeds themselves. Yet their effect on the plants they visit is not wholly negative. The “mistakes” they make, that is the seeds they lose along the way or discard by accident at the nest, also disperse plants and compensate at least in part for the damage caused by seed predation. It is entirely possible that seed dispersal by harvesting ants preceded myrmecochore-aided seed dispersal during the coevolution of ants and plants (Rissing, 1986).

The harvesting of seeds by ants in deserts and grassland is bound to impress human beings who live by the same activity. The storing of seeds by ants in underground granaries has equal appeal. From the Book of Solomon to the writings of the ancient Greeks and Romans, ants were established very early in Western culture as the symbols of industriousness and prudence. The metaphorical view of these insects was shaped largely by the abundant Old World harvesters of the myrmicine genus Messor. One of us (Wilson, 1984a: 6) paid a biologist's tribute to these ants and the tradition they engendered in the following way:

Once on a tour of Old Jerusalem standing near the elevated site of Solomon's Throne, I looked down across the Jericho Road to the dark olive trees of Gethsemane and wondered which native Palestinian plants and animals might still be found in the shade underneath. Thinking of “Go to the ant, thou sluggard; consider her ways,” I knelt on the cobblestones to watch harvester ants carry seeds down holes to their subterranean granaries, the same food-gathering activity that had impressed the Old Testament writer, and possibly the same species at the very same place. As I walked with my host back past the Temple Mount toward the Muslim Quarter, I made inner calculations of the number of ant species found within the city walls. There was a perfect logic to such eccentricity: the million-year history of Jerusalem is at least as compelling as its past three thousand years.

History of the study of harvesting ants

The history of our understanding of these insects has consisted of two active periods with a conspicuous intervening hiatus. Ancient writers were well aware of harvesting by ants because they lived in the Mediterranean Region and Middle East, where the phenomenon is prominent. The dominant species they encountered, as Wheeler (1910a) has pointed out, were undoubtedly Messor barbarus, which occurs throughout the Mediterranean littoral of Europe, Asia, and Africa southward to the Cape of Good Hope; Messor structor, which is absent in Africa but ranges all the way from southern Europe to Java; and Messor arenarius, which is abundant in the deserts of North Africa and the Middle East. These middle-sized, conspicuous ants are often serious grain pests, and it is to them that the writings of Solomon, Hesiod, Aesop, Plutarch, Horace, Virgil, Ovid, and Pliny almost certainly allude. The earliest work on ants published in the modern era, Jeremia Wilde's De Formica (1615), merely repeats the accounts of these authors, in Latin. When Gould (1747), Latreille (1802), Huber (1810), and other early entomologists of the modern era began to study ants in the field, they saw no evidence of harvesting and consequently doubted or even denied the classical reports. This turn of events was due entirely to the fact that they lived in temperate Europe, where harvesting by ants is rare or absent. When Europeans began to report from warmer, drier climates, however, the phenomenon was quickly validated. At Poona, India, Sykes (1835) observed Pheidole providens bring rain-soaked grass seeds out of the nest and place them on the grass to dry. Jerdon (1854) confirmed the phenomenon in Pheidole providens, Pheidole diffusa, and Solenopsis geminata, and he saw the workers of these species collect seeds from different species of plants and store them in the nests. Moggridge (1873), during a sojourn in southern France, worked out the procedure of seed harvesting by Messor barbarus and [[Messor structor in some detail. He found that the ants harvest the seeds of at least 18 families, and he confirmed reports of Plutarch and other ancient authors that the workers bite off the radicle to prevent germination, then store the deactivated seeds in granary chambers in the nests. He observed seed-drying by the same method as that described by Sykes. In a remarkably modernistic twist, Moggeridge also established that harvesters play an important role in dispersing plants by accidentally abandoning viable seeds in the nest vicinity or failing to deactivate them before they sprout. All of these key observations have been repeated many times by later observers, including Forel, André, Emery, Lameere, and, nowadays, an entire generation of younger researchers working on the ecology of harvesters.

Simultaneously, several early American entomologists addressed the subject of ant harvesting, including Buckley (1861), Lincecum (1862, 1866), McCook in his classic The Natural History of the Agricultural Ant of Texas (1879), and Wheeler in his important synthesis of the subject in Ants: Their Structure, Development and Behavior (1910a). Each made closely similar observations on harvesting ants in the deserts of the southwestern United States. The favored subject was the ubiquitous members of the genus Pogonomyrmex, but Aphaenogaster (= Novomessor), Messor (= Veromessor), and Pheidole were also included. One famous misconception due to Lincecum was that the Texas harvester Pogonomyrmex molefaciens deliberately sows the seeds of grasses of the genus Aristida around the periphery of its mound or crater nests and cultivates the crop in addition to collecting and storing the seeds in its granivores. Wheeler wrote,

“This notion, which even the Texan schoolboy has come to regard as a joke, has been widely cited, largely because Darwin stood sponsor for its publication in the Journal of the Linnean Society . . . Four years of nearly continuous observations of molefaciens and its nests enable me to suggest the probable source of Lincecum's misconception. If the nests of this ant can be studied during the cool winter months--and this is the only time to study them leisurely, as the cold subdues the fiery stings of their inhabitants--the seeds, which the ants have garnered in many of their chambers will often be found to have sprouted. Sometimes, in fact, the chambers are literally stuffed with dense wads of seedling grasses and other plants. On sunny days the ants may often be seen removing these seeds when they have sprouted too far to be fit for food and carrying them to the refuse heap, which is always at the periphery of the crater or cleared earthen disk. Here the seeds, thus rejected as inedible, often take root and in the spring form an arc on a complete circle of growing plants around the nest.”

This interpretation of the “crop” as an adventitious growth of harvested seeds is consistent with the observations of Moggeridge on Messor. It also fits studies of the relationships of Pogonomyrmex and other harvesters established by more recent researchers.

After the first fruitful period in the natural history of harvesters, extending roughly between 1860 and 1910, there was a lull in the study of these ants. An intense revival began in the 1970s when a new generation of ecologists recognized the convenience of Messor, Monomorium (= Chelaner), Pogonomyrmex and other harvesters for field and experimental studies in foraging and competition. This new work, which we reviewed in Chapters 10 and 11, has grown into an important chapter of modern general ecology. We will now review more general aspects of the natural history and environmental importance of harvesting ants.

The distribution of harvesting

The known harvesting ants are listed in Table 18-1. It can be seen that the life habit is disproportionately concentrated in the Myrmicinae. Within that subfamily a great many genera phylogenetically remote from one another are represented, including such physically disparate forms as Messor, Oxyopomyrmex, Meranoplus, and Pheidole. The degree of commitment of harvesters to a seed diet varies across species to the greatest imaginable extent, from occasional and optional in the African Atopomyrmex mocquerysi to total or nearly so in Monomorium whitei of Australia and Veromessor pergandei of North America. Much of this range of variation occurs among species belonging to single genera, including Rhytidoponera, Monomorium, and Pheidole. The workers vary equally in temperament. At one extreme is Pogonomyrmex, viciously combative and with the most toxic venom of any known insect poison, at least with respect to mammals (Schmidt and Blum, 1978). Pogonomyrmex workers are so aggressive that they fight members of alien conspecific colonies 80 percent of the time when they encounter them while foraging, and fatalities are commonplace (Hölldobler, 1976a; De Vita, 1979; see Figure 10-15). At the opposite extreme are Goniomma and Oxyopomyrmex, whose workers are so timid and few in number that nests are difficult to find (Felix Santschi in Forel, 1904).

Harvesting ants are dominant elements in the deserts and drier grasslands in warm temperate and tropical regions around the world, especially in North America, Australia, the Sahara, and South Africa (Wehner, 1987). Seed-harvesting species compose more than half of all ant colonies in some Australian localities (Briese and McCauley, 1981). In the Namib Desert they make up more than 95 percent of the total forager biomass.

An extreme granivore

Veromessor pergandei, which has been studied intensively by Diane Davidson, Steven Rissing, and others over the past twenty years, is one of the most specialized granivores found among ants anywhere in the world. It flourishes in the deserts of southwestern Arizona, southern California, and Baja California. In Death Valley, one of the driest and hottest places in North America, the Veromessor pergandei are the most abundant ants; they have a biomass approximately equal to that of the total rodent population in the same area (Went et al., 1972). The toughness of the species in the face of harsh conditions is legendary among entomologists. In the Coachella Valley of California, colonies survived even after twelve successive years of severe drought (Tevis, 1958). The key to this success is the tendency of the ants to store large quantities of seeds underground, and their evident ability to subsist entirely on this food without the supplementation of arthropod prey or nectar.

The population of a mature colony is very large, ranging into the tens of thousands. A full census has not been taken, because no one has succeeded in excavating a complete nest. Tevis apparently has come the closest. He was able to follow one gallery for 4 meters before losing track of the nest system. Of their own effort Wheeler and Rissing (1975a) wrote, “In Deep Canyon and Death Valley we tried slicing off the top or digging in from the side toward the center of the nest; even with the enthusiastic assistance of several students, we were never able to dig quickly enough to find any large concentrations of workers.” They succeeded in making a partial cast of a nest by pouring a casting resin into an entrance hole; the upper structure of the nest they revealed is shown in Figure 18-1. To accomplish more would require the planning and energy of an archaeological dig. The typical mature Veromessor pergandei nest, from what can be seen of it, has two or three active entrance holes 2 by 4 cm across. Each entrance is surrounded by a crater of sand and fine gravel. The material nearest the entrance is held together by a yellowish cement. This material also contains some substances that the workers recognize as belonging to their nest. A neat pile of chaff comprising the husks of seeds forms a semicircle on the northern perimeter. Just below each entrance is a large chamber, or “vestibule,” partly filled with chaff and a few seeds. Further down are seed-storage chambers and finally rooms containing mixtures of larvae and pupae.

Veromessor pergandei workers harvest seeds from a wide spectrum of plants. Among those recorded to date are 14 genera of plants in the Coachella Valley (Tevis, 1958) and 24 genera comprising 29 species in Death Valley (Rissing and Wheeler, 1976). However, the ants are far from indiscriminate. They probe through piles of seeds before selecting one to take home, and they consistently choose larger than average grass seeds of each species offered (Rissing, 1981a). Species of Chorizanthe, Franseria, and Lygodesmia are less favored, perhaps due to their exceptionally hard seed coats. Workers also tend to “major,” that is, to persist in harvesting one species of seed even when more desirable seeds are present in the same pile. In the nest the workers break the endosperm into fragments which are then placed directly on the larvae. Went et al. (1972) suggested that the larvae metabolize the seed materials and regurgitate carbohydrate-rich secretions back to the workers, in the manner later described in the Australian granivore Monomorium whitei by Davison (1982).

Veromessor pergandei workers observe two foraging periods in each day (Rissing and Wheeler, 1976). One begins in the early morning before there is any indication of daylight to the human observer and ends during the heat of late morning. The second period starts in the afternoon after the temperature has fallen from the midday high and continues until dusk, or even after dark on warm evenings. The workers emerge quickly in columns that follow preexisting trunk trails. On one occasion recorded by Wheeler and Rissing, a column containing about 17,000 ants extended 40 meters from the nest.

The success of Veromessor pergandei in the harshest American deserts is not due entirely to its heavy reliance on stored seeds, but in addition to three other traits that impart flexibility in the foraging strategies. First, the workers rely increasingly on individual searching when seed supplies are short, and on columns when seeds are encountered in patches. Second, the ants rotate their columns around the nest entrance in a way that brings them repeatedly to new patches and increases their yield over long stretches of time. Third, when the supply of desirable seeds is low, the workers turn to less desirable seeds and non-seed plant materials such as flower parts, leaves, and stems (Rissing and Wheeler, 1976). Davidson (1978) ascribed the considerable size variability of Veromessor pergandei workers to an adaptive polymorphism in which small workers tend to collect small seeds and large workers favor large seeds. If true this property of the worker caste would allow the ants to harvest seeds from multiple plant species more quickly and with greater energetic efficiency. However, in a later study, Rissing (1987) found that worker size accounts for less than 4 percent of the variance in the size of harvested seeds. He attributed the worker polymorphism to an annual cycle in the food made available to developing larvae. During the winter “triple crunch” (reduced seed availability, shorter foraging time, new queens and males being produced), worker-destined larvae evidently receive less food and end up smaller in size. If this interpretation is correct, the worker polymorphism of Veromessor pergandei is more parsimoniously interpreted as a nonadaptive epiphenomenon rather than an additional mechanism enhancing the flexibility of foraging.

Seed selection

All species of harvesting ants studied so far accept a wide array of plant seeds under natural conditions. The natural history of Veromessor pergandei shows why such latitude favors species most dependent on seeds for their livelihood. In the physically demanding and irregular environments in which harvesting ants live, few plant species can be depended upon to produce a profitable crop of seeds during any given year. This interpretation of the ecological significance of broad seed choice was nicely supported in a field study by Hahn and Maschwitz (1985) on Messor rufitarsis, an unusual harvester that occurs widely through central Europe. At the northern limit of its range, in the German state of Hessen, Messor rufitarsis exists in scattered populations occupying open habitats in which exceptional numbers of plant species grow. From May to October, two or more of these species produce significant quantities of seeds, and the ants pass from one set to the next across the growing season, rather like a person traversing stepping stones across a pond.

In spite of the advantages of broad dietary choice, harvester ants do discriminate among seeds to some degree. All other things being equal, there is a general tendency to gather seeds of the kind that are most abundant (Davidson et al., 1980). Among species of Pheidole and Pogonomyrmex in the southwestern United States, a strong correlation exists between the size of the worker caste and the size of the seeds they prefer (Hölldobler, 1976a; Hansen, 1978; Chew and De Vita, 1980). In semi-arid Australia, Monomorium rothsteini collects a higher proportion of small seeds than does its competitor Monomorium whitei (Davison, 1982). In the savannas of northern Ivory Coast, Messor galla forages at night in the dry season while collecting seeds from the short and medium grasses Monocymbium seresiiforme and Pennisetum hordeoides, whereas Messor regalis is primarily nocturnal and harvests from the tall grass Andropogon gayanus. Preference for the seeds of some plant species over those of others have also been reported in Monomorium (= Chelaner) (Davison, 1982), Pheidole (Mott and McKeon, 1977), Pogonomyrmex (Nickle and Neal, 1972; Whitford, 1978a), and Solenopsis (Risch and Carroll, 1986). In several cases the investigators observed a shift to other, less desirable seeds when those usually favored became less available. The chemical basis of the selectivity is unknown, although Buckley (1982b) noted the likely wide occurrence of repellents such as tannin as (imposed on top of nutritive attractants), while Ashton (1979) reported the existence of sweet substances attractive to ants in the seeds of Eucalyptus regnans in Australia. On the other hand, seeds may be protected by purely physical traits. Those of Datura discolor are among the largest and most energy-rich of any ephemeral species of the southwestern American deserts, but they are evidently protected from harvester ants by their thick, heavily sculptured coats (O'Dowd and Hay, 1980).

Foraging patterns

The commonest foraging strategy of harvesting ant species is a mix of individual foraging and column retrieval, adjusted according to need from one day to the next. This pattern is conspicuously displayed by the species of Messor and Pogonomyrmex (Figure 18-2). Scouts go forth to explore the terrain. They are guided variously by visual landmarks, by their compass direction relative to the sun, and by odor marks deposited in the vicinity of the nest exits (Hölldobler, 1971b). If a worker finds a solitary seed, it carries it back to the nest. If the ant encounters a patch of seeds, for example a seedfall beneath a grass clump, it carries one seed homeward while depositing an odor trail from the tip of its abdomen. Nestmates travel out to the seed patch along the trail, and while returning with burdens of their own they often add to the trail pheromone. In time, if the seedfall persists, the chemical deposits accumulate in sufficient strength to constitute a trunk trail along which large numbers of the ants travel back and forth. Even without reinforcement the trunk trail can remain active for days or weeks (see Figure 18-3). In Pogonomyrmex badius and other Pogonomyrmex species, the original recruitment substance is emitted from the poison gland and the longer-lasting orientation substance from the Dufour's gland (Hölldobler and Wilson, 1970; Hölldobler, 1976a; see Figure 18-4). In Messor rufitarsis, belonging to a wholly different stock of myrmicine ants, both pheromones come from the Dufour's gland (Hahn and Maschwitz, 1985). As food grows richer and more clumped in distribution, colonies shift more from individual orientation to foraging along trunk trails. Even in the best of times a few scouts wander away from the main routes. As a result they occasionally discover new seed patches and together provide the colony with the flexibility of response needed to exploit all of the surrounding environment efficiently. Overall, the density of foragers falls away steeply and exponentially with distance from the nest exit. In one census of Pogonomyrmex californicus made by De Vita (1979), the modal density was 1.6 meters from the exit, while few if any foragers ranged beyond 13 meters.

Harvester ants using foraging columns shift them to avoid competition at food patches with colonies of the same and closely related species. The typical foraging pattern of these insects is an overdispersion of nest sites: the sites are spread out more uniformly than probably from pure chance alone, and the trunk trails of different colonies never overlap. Sometimes the trails come so close that their final branches interdigitate, but they still do not cross or terminate at the same seed patch (see Figure 10-14). This pattern has been documented in Pogonomyrmex (Hölldobler, 1976a) and the European Messor (Lévieux, 1979), and it is implied in descriptions of foraging in the American Veromessor pergandei by Rissing and Wheeler (1976).

The commonest diel pattern of foraging in hot, dry climates is bimodal, with activity peaking in the cool of the morning and then again in the afternoon or early evening. But this trait is evidently very dependent on temperature or humidity. With the onset of the rainy (and cooler) season of Colombia, Pogonomyrmex mayri changes its pattern from bimodal to unimodal, and the peak shifts to midday (Kugler, 1984). During the rainy season in the Ivory Coast savannas, Messor galla and Messor regalis change from mainly nocturnal foraging with a secondary peak in the morning to principally diurnal foraging (Lévieux, 1979). Harvesters living in cooler, moister environments, such as the Japanese population of Messor aciculatus, reach their peak around the middle of the day (Onoyama, 1982).

A striking feature of harvester foraging is Ortstreue or “site tenacity,” the persistent return of individual foragers to the same restricted area trip after trip and even day after day. When the colony harvests seeds along multiple trunk trails, workers showing Ortstreue behavior choose one trail over others repeatedly. The phenomenon has been documented in Pogonomyrmex (Hölldobler, 1976a), Pheidole (Hölldobler and Möglich, 1980), Messor (Onoyama, 1982; Onoyama and Abe, 1982), and Monomorium (= Chelaner) (Davison, 1982). A closely related behavior is majoring, in which individual workers persistently choose one kind of seed out of two or more available in the same foraging area. This kind of behavior is known to occur at least in Monomorium (= Chelaner) (Davison, 1982) and Messor (= Veromessor) (Rissing, 1981a). Both types of individual specialization, Ortstreue and majoring, clearly have the potential of increasing both individual efficiency and colony-wide efficiency, since they by-pass the time-consuming procedures of exploration and prey choice. In order for this enhancement to occur, however, it is necessary for the species to add differentiation of worker choice among harvesting sites and seed type, as well as the capacity to shift rapidly to new sites and food items when old seedfalls are depleted. All of these properties have been documented in Messor and Pogonomyrmex, and they probably occur widely in other harvesting ants.

Effects of harvesters on vegetation

There is general agreement among students of ant ecology that harvesters strongly alter the abundance and local distribution of flowering plants, especially in deserts, grasslands, and other xeric habitats where the ants are most abundant. They tip the balance in competition among some plant species and promote equilibria in others. They also rearrange the local distributions of the surviving species.

Under many circumstances, seed predation by ants reduces seed density and the subsequent vegetative mass of the plants. When ants were removed from experimental plots in Arizona by Brown et al. (1979b), annual plants were 50 percent denser after two seasons than in nearby control plots with their ant populations still intact. In Eucalyptus woodland in southeastern Australia, seedling densities of Eucalyptus baxteri increased 15-fold after the ants were eliminated (Andersen, 1987).

On the other hand, ants often aid the exploited species by dispersing seeds more widely. Pogonomyrmex rugosus and Veromessor pergandei collect seeds of Plantago insularis and Schismus arabicus in Arizona deserts. Many survive long enough to take root in the refuse piles around the ant nests. There the growing plants are at least five times denser on the average than in nearby sites away from the nests (Rissing, 1986). Thus the plants and the harvesters can be said to exist in a state of mutualism. The plants feed the ants a certain fraction of their seeds in return for which the ants transport another fraction to sites (the nest perimeter) that are relatively rich in nutrients and free of competitors.

Having established the countervailing forces of predation and dispersal as a general phenomenon in ant-plant interactions, ecologists have begun to define a far more complicated array of second-order effects. In the dry tropical forest of northwestern Costa Rica, species of Pheidole and Atta are secondary dispersers of fig seeds (Ficus hondurensis). The seeds are first scattered by birds, coatis, monkeys, and lizards as these animals feed on the fruit. The ants then rearrange the “seed shadow” from this primary dispersal by picking the seeds from the feces of the vertebrates and uneaten fruit fragments left behind. Some of the seeds transported by the ants germinate (Roberts and Heithaus, 1986). Thus there may be a sort of mutualism among plants, vertebrates, and ants, although it is to be doubted that the ants coevolved to respond to seeds made available by the vertebrates.

At the opposite end of effectiveness from serving as weak secondary dispersers as just described, ants can exert powerful effects on competition and extinction. They can even serve as “keystone species,” meaning that they affect plant community composition to an extent disproportionate to their numbers or biomass. An important example is the fire ant Solenopsis geminata in the annual cropping systems of the wet tropics in Mexico and Central America (Risch and Carroll, 1986). The seed abundance and plant biomass of weeds, especially grasses, is lowest in sites where the ants are present. In plots of corn and squash studied in Mexico, the ants reduced the number of arthropod individuals ten-fold and the number of their species three-fold. Such manifold effects have only begun to be explored. Risch and Carroll (1986) examined the impact of the fire ant on four pairwise combinations of weedy grasses and found that the ants generally preferred the seeds of one over the other. The ultimate effects on plant biomass proved remarkably diverse. For one of the plant pairs, the ants reversed the usual course of competition by differentially preying on the seeds of the dominant plant, allowing it to be excluded in the end by the subordinate. In two other cases, the ants preferred seeds of the usually subordinate species, causing it to disappear more quickly. In the fourth combination, the ants created a stable equilibrium by holding down the dominant just enough to allow the subordinate to survive.

Although the full cascade of effects of harvester ants on vegetation is clearly of major importance, it is still very poorly understood. Speaking of the Australian flora, Andersen and Ashton (1985) summarized the problems--and our general ignorance--of the ant-plant interactions in the following way:

Although it is clear that ants can potentially destroy large numbers of seeds, there are many issues to resolve before their actual effect on seedling recruitment becomes known. For example, many sclerophyllous plants (such as species of Leptospermum, Melaleuca and Kunzea) have tiny seeds which might escape predation by falling amongst litter or by rapidly becoming incorporated into the soil. Second, seeds may avoid predation by falling during periods of low ant activity (e.g. winter) or by falling into areas where the activity of seed-eating ants is low. Investigations into patchiness of seed removal by ants indicate that even at sites where overall rates of removal are high, there are many places where removal is consistently low. Third, since most seed-eating ants are omnivorous, fluctuations in the availability of alternative food sources, such as insect prey, might have an important influence on removal rates. Fourth, stochastic events such as extensive seedfall in summer followed by an extended period of unseasonably cool and wet weather, might enable seeds to avoid predation, since rain reduces ant activity and promotes germination. Fifth, it has been assumed that all seeds removed by ants are ultimately destroyed, which obviously requires verification. Sixth, the importance of seed losses to ants depends on patterns of seedling recruitment; for example if there is high density-dependent seedling mortality, then seed predation may not be important. Finally, fire plays a key role in the reproductive biology of many sclerophyllous plants, and massive, fire-induced seedfall, resulting in predator satiation, might play an important role in successful seedling recruitment. It is also possible that ants are satiated in the absence of fire by plants which release their annual seed crop in a massive seasonal pulse.

To these complexities must be added competition of the harvester ants with other kinds of animals that exploit the seed crop. Exclusion experiments performed by J. H. Brown, Diane W. Davidson, and their co-workers in the deserts of southern Arizona have documented these higher-order phenomena. At the Silver Bell alluvial plain 60 kilometers northwest of Tucson, these investigators removed ants from one series of fenced-in 0.1-hectare plots and rodents from another. Within a short time the number of ant colonies rose 71 percent in the rodent-free plots, while the rodents in turn increased 20 percent in numbers of individuals (and 29 percent in biomass) in the absence of ants. These effects were evidently due to the greater number of seeds made available when one or the other of the two taxa was removed (Brown et al., 1979a). However, two years after the beginning of the experiment, the situation changed unexpectedly. The ants began to decline in the plots from which the rodents had been removed. The cause turned out to reside in the plants rather than the animals. With the rodents gone, the large-seeded plants normally favored by rodents began to increase, replacing the small-seeded plants favored by ants and thus reducing their food supply (Davidson et al., 1984). The same authors conducted similar experiments 250 kilometers to the east, on the Cave Creek alluvial plain near Portal, Arizona. In this area ants competed with rodents only for seeds produced during the less productive winter peak. When rodents were eliminated, Pheidole xerophila increased in numbers but Pogonomyrmex desertorum declined. Rodents did not change appreciably when relieved of ants. The reasons for the differences in the interactions between the ant species in each locality and, more generally, between all of the organisms across the two Arizona localities appear to entail (1) seasonality in the production of seed resources and in their use by the two taxa; (2) specialization by ants and rodents on different density distributions of seeds; (3) “diffuse compensation” or compensation spread over many species populations; and (4) indirect interaction pathways, mediated through competing resource classes.

Finally, the higher-order phenomena of the kind disclosed in the Arizona experiments are likely to change in kaleidoscopic fashion in passing from one biome or continent to another. In Australia, for example, many species of ants and birds but few mammals specialize on seeds, whereas all three are prominent in North America. Whether or not the Australian harvester ant fauna is correspondingly more diverse and abundant, as Brown et al. (1979a) suppose, cannot be ascertained with existing field data. In South America, despite the absence of specialized seed-eating rodents, granivorous desert ants and birds are not conspicuously more diverse or abundant than in North America (Mares and Rosenzweig, 1978). A point of comparison is the prominent harvester ant genus Pogonomyrmex, represented by 22 species in the arid environments of North America (Cole, 1968) and 22 different species in comparably arid environments of the southern half of South America (Kempf, 1972b). The South American species are characterized by much smaller colonies and evidently mount less dense populations.


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

The Ants - Table of Contents