The Ants Chapter 11

CHAPTER 11. THE ORGANIZATION OF SPECIES COMMUNITIES

Species belonging to the same community are said to be organized when the species interact in such a way as to make certain subsets more frequent than expected from chance alone. In Elton's terms (1933), the community has “limited membership.” At the outset, only some of the species in the surrounding pool have the dispersal ability to reach the habitat, so that admission is restricted. And of those that colonize, some fit in less well than others and pass into extinction sooner, making expulsion from membership discriminatory. Organization is based on the latter process of differential extinction. But the addition and loss of species is not the only way that organization can occur. It is also possible for species to affect one another's abundance, spatial distribution, and behavior.

By definition the more numerous the connections among the species, and the more complex the hierarchies they create, the more organized the community (Diamond et al., 1986). The most obvious type of species interaction that might prove effective is competition. But other such processes occur, including predation, parasitism, commensalism, and mutualism. In addition, it is possible for species to alter the physical environment in a way that allows other species to colonize it more readily. The key point of this chapter is that all of these phenomena have been demonstrated at one time or another in the ants.

Interspecific competition
In Chapter 10 we saw that competition is the hallmark of ant ecology. Many, perhaps most ant species employ aggressive techniques up to and including organized warfare. What are the consequences of this fierce competition for community organization? For nearly forty years myrmecologists have recognized the existence of dominance hierarchies among species, in which some species regularly displace others at territorial sites. These hierarchies are often linear through three or more links. Documentation has been provided in a series of detailed studies by Kaczmarek (1953), Marikovsky (1962b), Yasuno (1963, 1965b), Brian (1965b, 1983), Dlussky (1965), Wilson (1971), Zakharov (1972, 1977), Pisarski (1973, 1982), Reznikova (1975), Czechowski (1979), Jutsum (1979), Levings and Traniello (1981), Pisarski and Vepsäläinen (1981), Vepsäläinen and Pisarski (1982), Rosengren and Pamilo (1983), Fellers (1987), Savolainen and Vepsäläinen (1988), Ward (1987), and others.

Vepsäläinen and Pisarski have devised a useful three-tiered classification from general winners at the top to general losers at the bottom. On the Tvärminne Archipelago, a cluster of small islands off the Finnish Baltic coast, this hierarchy is as follows:

I. The lowest level comprises species that defend only their nests; examples include Formica fusca and the three Finnish species of Leptothorax.

II. At the intermediate level, species defend their nests and food finds; in other words they maintain spatiotemporal territories. Examples include Tetramorium caespitum, Camponotus herculeanus, and Camponotus ligniperda, as well as smaller colonies of Lasius niger.

III. At the top level species successfully defend their nests and all of their foraging areas as absolute territories. Examples include Formica exsecta, Formica sanguinea, Formica truncorum, and members of the Formica rufa group of species, as well as large colonies of Lasius niger. They are the “large-scale conquerors” recognized by Rosengren and Pamilo (1983) for Fennoscandia generally. The most prevalent of all species on the mainland is Formica aquilonia.

Some of the population and behavioral characteristics of

Tab 11-1 representative species are given in Table 11-1. Intermediate (level II) species are often extirpated entirely from the extensive domains of colonies belonging to level III. On the other hand colonies of level I species, which fight with alien workers only in the immediate vicinity of their own nests, are able to survive better in the presence of the dominants. The compatibility they enjoy is nevertheless far from perfect, because Formica polyctena workers periodically invade nests of the low-ranking Formica fusca and Lasius flavus and carry out adults and brood as prey. As a result of this pressure, few colonies of these two species survive within tens of meters of the Formica polyctena nests.

Since the dominant species patrol such large foraging areas, local faunas can be profoundly influenced by relatively few of their colonies. On the island of Joskär, for example, a single colony of Formica polyctena containing over a million workers dominated 20,000 square meters of terrain (the size of two football fields). Entire communities can be closed against some species. Formica rufibarbis is absent from the Tvärminne Archipelago, even though it is abundant on the adjacent mainland. The reason appears to be that its favored habitat, open dry pine forests, is preempted by colonies of the more aggressive Formica sanguinea.

The species hierarchies offer an unusual opportunity to analyze the role of aggressive behavior and interference competition in the assembly of communities. As Vepsäläinen and his co-workers have demonstrated, it is possible to translate upward from experimental studies of the encounters of individual ants on baits to local faunal lists. It is also possible to make qualitative predictions of relative abundance. The lowest ranking, level I species, for example, appear to be the same as the “opportunistic” species recognized by Wilson (1971) at food baits. Their workers are adept at finding new food but abandon it when confronted by more aggressive species. The exemplar in the Tvärminne Archipelago is Formica fusca. The workers of this species utilize a wide range of food items, which they locate quickly, and they shift to less desirable food when put under pressure from higher ranking species.

As one proceeds from small to large islands in various archipelagos around the world, the number of ant species increases roughly as the fifth to third root of the island area (Wilson, 1961; MacArthur and Wilson, 1967). It is also true that the growing assemblages often form nested sets. That is, on extremely small islands only species A-C might be present, on somewhat larger islands A-F, on still larger islands A-H, and so on. The sequence is seldom if ever of the form A-C, D-I, K-X, and so forth (Wilson and Hunt, 1967). This regularity, which is far from perfect and evidently subject to considerable stochasticity, reflects the fact that larger islands provide a greater array of microhabitats in which specialized ant species can settle. But it is also likely to be shaped in part by the kind of rank-ordering and preemption illustrated in the Finnish studies. The nested-set phenomenon is unusually clear-cut in very small mangrove islands in the Florida Keys (Cole, 1983a). As depicted in Figure 11-1, the species of arboreal ants build up in a regular manner with an increase in volume of the foliage. Each species has a minimum island volume it requires for indefinite survival, which for some species at least is that amount of plant surface needed to protect the colony from excessive stress due to wind and wave action. The nests are especially vulnerable, because the mangrove trees are rooted in tidal mud. Two species, Crematogaster ashmeadi and Xenomyrmex floridanus, are dominant. When a colony of either one is established first, it precludes the invasion by a colony of the other species. When one of these two species is present on islands with volumes less than 5 cubic meters, it precludes an invasion by the remaining, subordinate species. Workers of the two dominant species are consistently aggressive toward workers of all other species, while those of the subordinate species (Camponotus sp., Pseudomyrmex elongatus, and Zacryptocerus varians) almost invariably run from enemies. When Cole removed the dominants from islands less than 5 cubic meters in volume, the subordinates were able to invade the trees and persist indefinitely.

Dominance hierarchies among ant species are a worldwide phenomenon. In the arid and semi-arid regions of Australia, the overwhelming alpha species are members of the dolichoderine genus Iridomyrmex (Greenslade, 1979; Greenslade and Halliday, 1983; Andersen, 1986a,b). As described by Greenslade in his study of the fauna of South Australia, Iridomyrmex are successful competitors and generally they dominate associations of ant species. The more abundant, diurnal Iridomyrmex species compete with each other and since they tend to be mutually exclusive their colonies often form a patchwork on the ground. The distribution of dominant Iridomyrmex then sets up patterns in space and time with which other ants must conform. At high temperatures, above-ground activity by the Iridomyrmex ceases and they are replaced by Melophorus. . . Some Camponotus species are nocturnal, foraging at night when many dominant Iridomyrmex are inactive. Others are much larger than the Iridomyrmex with which they coexist and reduce interaction by the size difference. Others again nest in no-man's land between colonies of Iridomyrmex. The behaviour of Camponotus is normally evasive when they encounter other ants and they seem to have succeeded as subordinate competitors. Other common genera are Rhytidoponera, Monomorium and Pheidole. As a rule the species are not closely adapted to the rest of the ant community. Instead they tend to be rather unspecialized, catholic feeders, flexible in their time of foraging and able to occupy a wide variety of habitats. Together with some species of Iridomyrmex and Paratrechina they often seem to be opportunists, exploiting areas or resources that are not intensively used by other ants.

An example of the way in which the subordinate species fit into the Iridomyrmex community is provided by a formicine of the genus Melophorus that lives in unoccupied chambers of nests of Iridomyrmex purpureus, emerging onto the surface and foraging in the middle of the day in hot weather when the meat ants are inactive. This general adaptation to extreme heat, paralleling the thermophily of Cataglyphis and Ocymyrmex in Africa, has evidently been a key factor in the evolution of Melophorus into a large number of species that fill many nesting and feeding niches. The genus includes general predators, predators on termites and other ants, and seed harvesters. One species has a minor worker with a conspicuously flattened head and alitrunk that allows it to forage under the peeling bark of Eucalyptus trees.

The dominant species form the core of the local community. They affect the composition and abundance not only of other ant species but even of plants and other arthropods. Their pervasive influence is clearly marked in the canopies of tropical forests of West Africa, where they spread out into mutually exclusive territories that cover from one to several trees each. The resulting “ant mosaic” was originally described by Leston (1973a) as follows:

The humid tropics fauna include certain dominant ants. Where they occur these are more numerous than other ants and to the exclusion of other dominants, i.e., those ants which elsewhere are more abundant. Dominants are usually non-nomadic, arboreal, multi-nested, saccharophilic and predatory, practicing mutualism with Homoptera: they have a potential for rapid population growth. Optimally dominants are spaced out in a three-dimensional mosaic with few lacunae but forest degradation leads to a two-dimensional structure. Three mosaic patterns are recognized: (1) the forest mosaic, occurring too but simplified in the crowns of cocoa, coffee, coconut and oil palm, etc.; (2) the forest understorey pattern; (3) the Crematogaster striatula pattern, peculiar to West Africa. . . Each dominant is the centre of a positive association of other insects, spiders and non-dominant ants, and of a negative association. The positive associates are not merely those insects with which the dominant is in mutualistic symbiosis. In the Old World tropics many tree crop pests have patchy distributions through their negative association with some dominants, their predators, and positive association with others. “Flitting” insects are more affected then “flying” or endophytic species: insects can be arranged in an “ant impact” hierarchy. Manipulation of the mosaic leads to changes in the overall population levels of many pests. Mosaics occur throughout the Old World tropics and probably in the New World too: it is likely that Azteca and its allies are dominant, to some extent limiting leaf-cutting Attini.

The details of the African mosaics were further elucidated by Room (1971), Leston (1973b,c), Majer (1976a-c), and Jackson (1984). Room studied over 100 species of ants in cocoa farms in Ghana and plotted their pairwise occurrences. The five dominant species and their associates are shown in Figure 11-2, while the negative associations are given in Figure 11-3. The reason why some species in the African canopy repulse one another is clear enough, but the basis of the positive association between many pairs is unknown. A strikingly similar mosaic pattern occurs in tropical Australia, with Oecophylla smaragdina and a species of Crematogaster playing key roles (Figure 11-4). Majer established the dominance of Oecophylla longinoda, Crematogaster striatula, Crematogaster depressa, and Tetramorium aculeatum in that order. He proved competitive displacement by removing colonies of some of the dominants and watching the competitor species move in, along with their favored ant and other arthropod associates.

Changing and removing dominants
If dominant ant species are really organizing agents, then changing from one dominant species to another should have marked effects on the remainder of the ant community. We have seen that this is indeed the case in passing from one site to another in the African canopy. The ants and other insects associated with Oecophylla longinoda are very different from those associated with the species of Crematogaster, creating a spatial mosaic across the tree tops. Fox and Fox (1982) described a changeover through time of the dominant ants in regenerating heath at Hawks Nest, Australia. There was a shift in only two years from a community dominated by three species of Iridomyrmex to one dominated by two other species of Iridomyrmex and Tapinoma minutum. Not only were the dominant species replaced (all dolichoderines), but also the entire communities of ants associated with them. The Foxes observed a change in species diversity and density of individual ants across the changeover point. Similarly, major shifts in the local ant fauna were observed by Greenslade (1971) in changeovers of the dominant species Anoplolepis longipes and Oecophylla smaragdina in coconut plantations of the Solomon Islands. When Anoplolepis longipes flourished, species diversity fell sharply; and when it declined, diversity increased.

In communities lacking such dominant species, or “large-scale conquerors” of level III importance, the hierarchical arrays of remaining species are much less predictable (Dobrzanski and Dobrzanska, 1975). Some of this change is due in various localities to the harshness of the local environment and an increase in diversity of microhabitats (Gallé, 1975; Boomsma and van Loon, 1975), but the relaxation of competition and predatory pressure almost certainly also plays a role (Savolainen and Vepsäläinen, 1988). In the grasslands of southern England, the three principal species are Tetramorium caespitum, the overall dominant, Lasius alienus, and Lasius niger. Formica fusca is a marginal species, able to live within the territories of the Tetramorium but avoiding contact by building a single small entrance to the nests and foraging more widely in places less frequented by the Tetramorium. Farther north, in the vicinity of Strathclyde, Tetramorium caespitum is absent and Formica fusca is replaced by the related Formica lemani. Under these circumstances Formica lemani is a dominant of sorts, taking the favored nesting sites and preying on two species of Myrmica (Brian, 1983).

Community structure is clearly affected by climate, the diversity of nest sites, the diversity of food, and interspecific competition. But the relative importance of these factors, despite excellent and strongly suggestive studies conducted on several continents, remains to be definitively measured. To this end, comparative analyses of a quantitative nature of different communities are a challenging task for students of ant ecology.

The dominance-impoverishment rule
We have noticed a worldwide tendency in the relation between behavior and species diversity, as follows: the fewer the ant species in a local community, the more likely the community is to be dominated behaviorally by one or a few species with large, aggressive colonies that maintain absolute territories. The relation holds in the relatively species-poor canopies of Africa (Room, 1971; Leston, 1973a), Australia (Hölldobler, 1983), and the Solomon Islands (Greenslade, 1971); the boreal faunas of northern Europe (Vepsäläinen and Pisarski, 1982; Rosengren and Pamilo, 1983); the mangrove islets of the Florida Keys (Cole, 1983a); the small islands of the West Indies (Levins et al., 1973); valley riparian woodland in California (Ward, 1987); and the arid and semiarid as opposed to mesic habitats of Australia (Greenslade, 1976; Greenslade and Halliday, 1983; Andersen, 1986a,b).

What is cause and what is effect in this rule? Are some ant faunas impoverished because of the suppressing effect of the “large scale conquerors,” or have the large scale conquerors originated in environments with impoverished ant faunas? At first it might seem that the first alternative is more likely to be true, because the dominant species have been repeatedly shown to reduce species diversity and abundance within their territories. When the populations of colonies of such species are dense, the effect can be widespread. Indeed, Carroll (1979) has postulated that the low biomass of stem-dwelling ants he found in Liberia in comparison with those in Costa Rica is due to the greater prevalence of dominant ant species (such as Oecophylla) in Liberia.

However, initial intuition can be wrong in this case. We have concluded that the opposite is true, that impoverished faunas promote dominant species rather than the other way around. Our reasoning is as follows. If the appearance of dominant species promotes impoverished faunas (the first alternative), the loss should occur only in alpha diversity, in other words the number of species found in the particular sites where dominant species are present, but not in beta diversity, the number of species occurring across many localities with and without dominants. Put another way, the faunas of whole regions and habitats in which dominants prevail should be rich even though the local, individual sites where they occur are poor. But this is not the case. The regions where the dominants occur are generally ones where the faunas as a whole are small, such as boreal Europe and small tropical islands. Furthermore, arboreal ant faunas in tropical forests are generally much less diverse than the terrestrial ant faunas that lie just below them, and they are also the ones dominated by species with large aggressive colonies and absolute territories. This appears particularly to be true in West African forests, which were hard hit during the Pleistocene dry periods (Moreau, 1966; Carroll, 1979). This difference exists even though the leaves and branches of the arboreal zone are of the same geologic age as the litter and humus of the terrestrial zone.

We therefore suggest that the primary causal chain is from impoverishment to dominance rather than from dominance to impoverishment. Some habitats have relatively few species because they are physically harsh, are restricted to a limited area, or are geologically young, or some combination of these features. As a consequence there is a relatively limited number of specialists available to preempt the narrower niches of nest sites and food items. The opportunity exists for a few other, generalist species to expand ecologically and to occupy a wide range of nest sites and food items. They will tend to evolve a large colony size and behavioral mechanisms, such as absolute territories, that vouchsafe control of the larger niches into which they have moved.

The conditions for coexistence of species
Interference between colonies belonging to the same species has the important effect of increasing the numbers of competing species that can coexist. This result is predicted in the graphical models devised by Gause and Witt (1935) on the basis of the Lotka-Volterra competition equations. In words, the Gause-Witt theory states that, if two species interfere with one another to any extent, one will always replace the other unless the following condition is met: the population densities of the two species must be self-limiting in such a way that they will stop increasing before the other species becomes extinct. The most familiar way in which such an equilibrial coexistence can come into being is if the two species occupy sufficiently different niches. Then one will tend to reach a limit in the part of the habitat to which it is specialized and reaches maximal densities before it is able to crowd out the second species in the part of the habitat optimal for the second species. This special case has become so familiar in ecological writing as to be frequently referred to as the “Gause hypothesis,” “Gause's law,” and so forth, but the Gause-Witt model embraces other potential mechanisms as well. Consider, for example, the possibility that the population density of each species is under the control of a parasite specialized for feeding on it. This, too, could easily lead to stable coexistence of the two prey species--as well as of the two parasitic species. It follows that any density-dependent control peculiar to a species will contribute to the stable coexistence of competing species.

The greater the difference existing between species in their respective niches, comprising nest sites, time of foraging, food, and so forth, the more likely each species will be independent in its population controls. In this light the work of Pontin (1960, 1961, 1963) takes on a particular significance. Pontin made careful studies of the ecology of two related species of formicine ants, Lasius flavus and Lasius niger, with special reference to the ways in which each affects the survival and reproduction of the other. In calcareous grassland near Wytham, England, the two species are dominant species, and their colonies are intermingled at saturated densities. In order to measure the consequences of interaction, Pontin first placed newly mated Lasius queens in tubes with openings large enough to admit workers but too small to permit the escape of the queens, and seeded them within the territories of mature colonies. He found that the queens were attacked and destroyed preferentially by workers of their own species. Studies were then made of the relation between productivity of new queens and the distance between the nest of the colony and the nearest nests belonging to both species. In a related experiment, colonies of Lasius flavus were transplanted to new positions in a circle around nests of Lasius niger in order to increase the competitive pressure on them. The results showed conclusively that queen productivity is reduced more by intraspecific than by interspecific interference. Therefore, through both the depression of the production rate of new queens and their destruction following the nuptial flights, each of the two species controls its own population densities to a greater extent than those of its competitor. This effect fulfills, at least in principle, the essential condition of the Gause-Witt equilibrium. The behavioral basis of the effect can only by guessed. Perhaps the reason why workers attack alien queens of their own species preferentially is that, as Brian (1956a) has claimed to be the case in Myrmica, they tend to be repulsed at a distance by the odors of both queens and workers of alien species. Such a response, which serves primarily as an adaptation to avoid injurious conflict, could not be extended to members of the same species without interfering with normal communication within the colonies.

Interference between mature colonies seems to be reduced by innate ecological differences between the two species. Lasius niger is a versatile ant that nests in rotting stumps, beneath stones, or in the open soil (often in mounds), and forages both below and above ground and up onto low vegetation. Lasius flavus is a primarily subterranean species that builds mounds in the open soil. Where the two species live together in the Wytham grassland, niger inserts itself in suitable nest sites between the flavus mounds. By competition for space and food (and limited predation on flavus, which is not reciprocated) niger depresses the queen production of flavus. Symmetrically, flavus takes away space and food from niger; and it also interferes with niger by using stones as props for the mounds of excavated earth, thus covering them and denying them to the niger for use as nest covers. But the degree of interference between the two species seems to be superimposed on a larger degree of interference among colonies belonging to the same species. The latter, intraspecific interference is not only enough to stabilize the populations of colonies, but it is also sufficiently greater than interspecific interference to permit the permanent coexistence of the two species. A similar relationship appears to occur between two species of honeypot ants, Myrmecocystus depilis and Myrmecocystus mimicus, which have much the same ecological requirements and frequently coexist in the same habitats (Hölldobler, 1981a).

Once the nest site of an individual ant colony has been selected, and if the area of its foraging ground is fixed by the colonies in residence around it, the productivity of the colony will probably depend on the food yield of the territory. It follows that in truly territorial species the production rate of new queens in mature colonies will increase as a function of the size of the territory. The relation has been verified in the two analyses in which the proposition has been tested: that of Pontin (1961) on Lasius and that of Brian et al. (1966) on Myrmica.

Actually, the factors controlling the density of populations of social insects are probably what ecologists refer to as intercompensatory. This means that, in a given environment at a given time, one factor is usually limiting, and, if it were removed, the population would increase until a second factor became limiting, and so on. If nest sites became unlimited in a food-rich area, the colony density would increase until food became scarce. If food were then presented in unlimited amounts, the populations would presumably increase until territorial behavior (to be sure, centered around smaller territories) stabilized the population at a new, still higher level. This simple sequence is based on only a few empirical observations, mostly those of Brian, and it is speculative when applied to social insects as a whole. It is altogether probable that other sequences and other factors are at work. It is even likely that different populations belonging to the same species equilibrate under different schedules of controls.

Against this theoretical background it is possible to make more sense of the diverse interaction phenomena of ant species, and to gain some understanding of the mechanisms that organize species communities. In briefest form, species use an amazingly wide range of procedures either to push back competitors, fit among them unobtrusively, or escape into marginal environments. Those documented to date will now be examined.

Niche differentiation. The reduction of interference by foraging at different times of the day is well known (Baroni Urbani and Aktaç, 1981; Hölldobler, 1981a; Klotz, 1984). But even when this occurs a dynamic tension between the species can “fine-tune” the adjustment. In Australia Iridomyrmex purpureus and Camponotus consobrinus utilize the same food sources, and they often nest in close association (Greaves and Hughes, 1974). As Hölldobler (1986a) found, the Iridomyrmex forage mostly during the day and the Camponotus mostly at night, with the two species replacing each other at particular homopteran aggregations and other persistent food sources (Figure 11-5). Where either of the species occurs alone, its foraging period is usually longer by one to two hours. And where the Iridomyrmex and Camponotus occur together, they shorten each other's foraging period by direct interference. In the morning Iridomyrmex purpureus workers gather around the nest exits of Camponotus consobrinus and close them with pebbles and clumps of soil. At dusk the situation is reversed: the Camponotus gather to prevent the Iridomyrmex from leaving the nest (Figure 11-6).

An equally striking displacement of diel schedules has been discovered in several other ant species. In Costa Rica, Swain (1977) observed that workers of the dolichoderine Monacis bispinosa ceased tending scale insects at night when a large yellow formicine of the genus Camponotus appeared. In order to recreate the displacement experimentally, he lured Monacis workers to a sugar bait in a new nearby site. At first, the Monacis workers continued feeding after dusk, the normal activity pattern of the species in other localities. But as soon as Camponotus workers found the bait, the Monacis retreated. Those who hesitated were attacked and killed, and the diel displacement was quickly established. In the Siberian steppes, colonies of Formica subpilosa are usually most active during the mid-afternoon. However, in the presence of Formica pratensis they shift the peak of their activities to the evening. Stebaev and Reznikova (1972) were able to induce the change by moving nests of Formica pratensis to the vicinity of those of Formica subpilosa. In forests of Ghana the arboreal myrmicine Tetramorium aculeatum spaces out its foraging time and becomes more nocturnal in the immediate presence of Crematogaster clariventris and Oecophylla longinoda (Leston, 1973c; Majer, 1976a).

Comparable cases of displacement leading to dynamic equilibria exist in the differentiation of space. On Hicacos Island near Puerto Rico, Levins et al. (1973) observed that when food baits were shaded, they were dominated by Pheidole megacephala workers. When the same baits were flooded by sunlight, the Pheidole retreated and the baits were taken over by Brachymyrmex heeri. When the baits were again shaded, the Brachymyrmex left and the Pheidole returned. The turnover occurred within a half hour and was accompanied by sporadic fighting.

Density specialization. Davidson (1977b) suggested that ant species might divide the environment by specializing on different densities of seeds and other food items. Those that forage in large numbers along trunk trails are likely to enjoy an advantage when food is denser. They are especially effective in rich patches where cooperating groups operate most efficiently. At lower food densities the energetic cost of foraging in this manner might exceed the energetic yield, allowing solitary foragers to take over. Davidson provided some empirical evidence instantiating this model of niche division. Among granivorous ants of the southwestern American desert, trunk-trail foragers such as Solenopsis xyloni and Veromessor pergandei concentrate on high-density patches of seeds. In contrast, individual foragers like Aphaenogaster cockerelli and Pheidole desertorum harvest seeds that are mostly dispersed and hence require independent discovery. They also spend more time searching than trunk-trail foragers. A mixed strategy was used by Pogonomyrmex rugosus, which use trunk trails during peak seed abundance and revert more to individual foraging when seeds are scarcer. In experiments, trunk-trail foragers were more selective in their choice of food when food was abundant than were individual foragers. In addition, Hölldobler et al. (1978) demonstrated that Aphaenogaster cockerelli is less efficient in recruiting nestmates to food patches than is Pogonomyrmex rugosus.

The interesting consequence of density specialization, to the extent it occurs in nature, is that it permits the coexistence of species that are ecologically identical in all respects except in the distribution of their food items. It is also a form of niche division that can profoundly affect the modes of communication and foraging techniques of ant species.

The concept of density specialization can be fitted at least loosely to the phenomenon of dominance hierarchies among species. In some environments, the trunk-trail, high-density specialists would be the same as the highest ranking competitors in the dominance orders. Such species not only utilize rich food sources but also monopolize them by means of territorial aggression (Hölldobler, 1976a). To return briefly to the Tvärminne Archipelago, Formica polyctena and Formica lugubris are approximately equal dominants. Their colonies attain very large size, and the workers monopolize aphid clusters and other rich food sites as part of their absolute territories. Their large mound nests represent considerable energetic investments and are seldom abandoned. The two species are separated by a slight difference in habitat preference: Formica polyctena favors older pine forest with undergrowth and Formica lugubris newly grown pine forest. At the next competitive level is Formica truncorum, the colonies of which have smaller populations and occupy less food-rich and stable habitats, thus partially escaping competition from Formica polyctena and Formica lugubris. They also build smaller, less “expensive” nests. Close behind in the hierarchy is Formica exsecta, which is similar in most respects to Formica truncorum but has a smaller worker size. Below both of these species is Camponotus ligniperda, which defends only the food sources it happens to find. It nests in sites outside the territories of the dominant Formica, such as open, rocky pine forest. At the bottom of the hierarchy are Formica fusca and the species of Leptothorax and Myrmica, characterized by much smaller colonies and marginal nest sites. The workers search individually, and most frequently in areas with low food density where none of the dominant Formica occur (Pisarski and Vepsäläinen, 1981; Vepsäläinen and Pisarski, 1982). In southern California, a similar relation between rank, nest site, and foraging pattern has been noted in the dominance of Formica haemorrhoidalis over Camponotus laevigatus (MacKay and MacKay, 1982).

Size differences and worker polymorphism. Size differences have often been implicated in the reduction of competition among closely related animal species. Ants offer some persuasive examples. In a study of three Pogonomyrmex species at Portal, Arizona, Hansen (1978) found a close correlation between the size of the workers and the size of the seeds they collected. The mean wet weight of the foraging workers was 5.8 mg in Pheidole desertorum, 8.9 mg in Pogonomyrmex maricopa, and 15.1 mg in Pogonomyrmex rugosus, while the mean weight of seeds collected by them was 1.0, 2.2, and 2.8 mg respectively, with pairwise differences all significant at the 99 percent confidence level or higher. Size differences are a striking feature of the western North American species of Pogonomyrmex generally, as we have illustrated in Figure 11-7. It is a notable fact that the only species occurring in eastern North America, the Florida harvester Pogonomyrmex badius, is strongly polymorphic, with the continuous size variation of workers from a single mature colony spanning most of that covered by many species in western North America (Figure 11-8). It seems possible that Pogonomyrmex badius workers also collect a wider range of seeds than those of individual species in the west. In other words, in the absence of competition Pogonomyrmex badius might have evolved into a generalist. However, the full diets of Pogonomyrmex badius and other Pogonomyrmex species remain to be described, and James Traniello (personal communication) has found no correlation of forage load weight and worker size in Pogonomyrmex badius.

A strong case for the evolutionary relation of polymorphism to competition has been made for the harvester Veromessor pergandei by Davidson (1977a,b, 1978, and as a co-author in Brown et al., 1979b). This “supreme specialist” is the most abundant ant in the least productive deserts of the American west. Its huge colonies use a mixed strategy of trunk-trail and individual foraging to exploit seeds that are either clumped in patches or independently scattered. At different localities the degree of size variation within single Veromessor pergandei colonies is inversely correlated with the presence or absence of competing granivorous ant species. At the eastern edge of the species range, the moderately productive Sonoran Desert of south-central Arizona, Veromessor pergandei coexists with up to four potential competitors that are either larger or smaller but approximately monomorphic. Here the Messor workers display less within-colony polymorphism, and those with intermediate size predominate. In desert habitats with still greater resource productivity and a correspondingly higher diversity of ant species, Veromessor pergandei is replaced by species whose colonies possess workers of nearly uniform body sizes and forage on narrower ranges of seed densities.

It is still an open question as to whether the correlation of body size and seed size is always realized in polymorphic harvester ants. Davidson (1978) reported such correlations in Veromessor pergandei at four separate locations, but not at a fifth site. On the other hand Rissing and Pollock (1984) working in different habitats were unable to detect size matching in this species. The phenomenon was demonstrated in the fire ant Solenopsis invicta (Wilson, 1978). Rissing and Pollock (1984), however, point out that in this case the prey size taken by small workers (head width less than 0.8 mm) is much smaller than that taken by large workers. This is most likely due to physiological limitation on the carrying ability of small individuals. Veromessor pergandei workers “are substantially larger, with most workers measuring 1.0 - 2.0 mm in head width. Size-matching is absent in Veromessor pergandei and in Solenopsis invicta workers of the same size range as Veromessor pergandei (Wilson, 1978). An analysis of the effect of burden on velocity of Veromessor pergandei foragers (Rissing, 1982) indicates these individuals are substantially larger than necessary to carry seeds commonly found in their habitat. This suggests that physiological limitations on harvestable item size does not occur in Veromessor pergandei. Rissing and Pollock conclude: “To the extent that size-matching does not occur in Veromessor pergandei, worker size variability must be a weak to nonexistent factor in determining diet breadth or foraging efficiency in this species.”

There is one other aspect which might in part account for the sometimes conflicting results. Rissing (1987) recently found in Veromessor pergandei a “distinct annual cycle in mean worker body size that replicates across colonies and habitats; this cycle occurs through alteration of the worker size distribution.” Rissing argues that the worker size variance over the year is a mechanism to maintain a constant, large worker force, in which smaller workers appear in the foraging force following periods of reduced seed availability, reduced favorable times to forage, and alate production during winter months. Since Veromessor pergandei exhibits interspecific territoriality, a large and constant worker force appears of selective advantage. If it is too costly to make large workers, the colony chooses to make smaller workers rather than reduce the size of the worker force.

Trophallactic appeasement
One mechanism by which ant species might mutually adjust to one another is through some form of appeasement whereby dominance is recognized and halted short of fatal aggression. Interactions between competing colonies is often thought to consist entirely of threat, fighting, and avoidance. These are indeed the prevailing responses observed under natural conditions. But it also appears that some ant species regurgitate liquid food to adversaries during hostile encounters, and that they benefit to some extent by stopping or delaying physical attacks. Kutter (1963b, 1964) found that when he placed colonies of Lasius fuliginosus and various species of Formica (exsecta, pratensis, rufa, truncorum) in containers close together and connected them by wooden bridges, intense fighting broke out as expected. But eventually the surviving workers grew more friendly, engaging in mutual grooming and feeding. The Lasius workers nevertheless remained hostile to the Formica queens and hunted them down, so that in time the former “alliance colony” turned into a pure Lasius colony. Similarly, workers of Pheidole dentata and Pheidole morrisii placed with fire ants (Solenopsis geminata and Solenopsis invicta) close together in plastic cells or larger laboratory nests appeased the more aggressive fire ants by regurgitating food to them. The same proved true of the two fire ant species when placed together, as well as various combinations of Formica. These encounters consistently lowered the frequency of overtly aggressive acts (Bhatkar and Kloft, 1977; Bhatkar, 1979a,b, 1983; Kloft, 1987).

Is trophallactic appeasement, to use Bhatkar's phrase (1979b), just an artifact of laboratory confrontations? Evidence that it occurs under more natural conditions was obtained by Bhatkar (1983), who observed the behavior of ant species attracted to sugar-water baits in a lightly wooded area of northern Florida. He found that whenever Pheidole dentata, Pheidole morrisii, or Solenopsis geminata workers arrived at the baits before those of Solenopsis invicta, the invicta scouts challenged them by raising the gaster, secreting a droplet of venom, vibrating the abdomen (probably to dispense the venom as an aerosal, as described by Obin and Vander Meer, 1985), and advancing slowly with mandibles open. The food-laden workers of Pheidole and Solenopsis geminata often responded by opening their mandibles, regurgitating a droplet of the sugar water, and offering it to the attacking Solenopsis invicta. The invicta workers became less aggressive, holding their own mandibles open to receive the food, or stopping the interaction entirely to groom themselves. Similar exchanges were observed among different species visiting the same aphid associations to collect honeydew.

Thus trophallactic appeasement occurs in nature, but we have too little information to assess its extent among ants generally or its role in community organization. The exchange appears to be specialized as an adjustment to aggressive, dominant species and is most likely to occur around long-lasting food sources that attract large numbers of foragers.

Enemy specification
The major predators of ant species are frequently other ant species, including many specialized to prey on ants (see Chapters 15 and 16). It is equally true that the most serious interspecific competitors in ant communities are dominant territorial species. It is therefore not surprising to find that at least some ant species have evolved defensive maneuvers directed in a precise way to identify and confound their most dangerous adversaries.

This phenomenon of “enemy specification” was first discovered in Pheidole dentata, a small myrmicine abundant in woodland over most of the southern United States (Wilson, 1976b). The native fire ant Solenopsis geminata occurs in many of the same habitats and to some extent utilizes the same nest sites as Pheidole dentata. It forms large and aggressive colonies that are strongly territorial. The same is true of the red imported fire ant Solenopsis invicta, which has spread throughout much of the southern United States during the past forty years. Fire ant scouts recruit masses of workers to food sites. They also treat Pheidole dentata as food and can destroy a colony within an hour. The Pheidole can avoid this fate by intercepting the scouts before the Solenopsis are able to mount an invasion. The Pheidole minor workers respond to the presence of only one or two Solenopsis. Within seconds, some of them start to run swiftly back and forth to the nest, dragging the tips of their abdomens over the ground. The trail pheromone thus deposited attracts both minor and major workers from the nest in the direction of the invaders. The majors are especially excited by the combination of the fire ant odor adhering to the bodies of the returning minor worker scouts and to movement while they are being contacted. They do not lay odor trails of their own. Upon arriving at the battle scene they become even more excited, rushing about and snapping at the fire ants with their powerful mandibles and chopping them to pieces (Figure 11-9). The recruited minor workers also join in the fighting, but they are less persistent and remain in the area for much shorter periods of time. As a result the majors increase in proportion, and for all but the most transient invasions they eventually outnumber the minors, despite the fact that they constitute only 8-20 percent of the worker population in most nests. The majors remain in the battle area for an hour or more after the last Solenopsis has been dispatched, restlessly patrolling back and forth.

In this manner the Pheidole are able to “blind” colonies of fire ants by destroying the scouts of these dangerous enemies. But they do not react to other potential adversaries in the same way. Ants of a wide variety of species in other ant genera tested by Wilson were neutral or required a large number of workers to induce the response. In a more recent study, Carlin and Johnston (1984) demonstrated that Pheidole dentata colonies can be sensitized to react more swiftly with a defense recruitment of majors when encountering other ant genera, such as Tetramorium, Crematogaster, and Atta, provided they have previously been repeatedly exposed to these genera. However, even then they never exhibit the same immediate and trigger-like response as when exposed to Solenopsis. Similar reactions to Solenopsis have been reported in Pheidole militicida, a seed-harvesting species of the southwestern United States, and Pheidole morrisii, an omnivorous species of eastern U.S. woodlands (Feener, 1986, 1987a).

Enemy specification has also been discovered in African and Asian weaver ants of the genus Oecophylla (Hölldobler, 1979, 1983). At the Shimba Hills in Kenya, only a few ant species coexist with Oecophylla longinoda on the same tree. Some of these, including a large species of Polyrhachis, are occasionally hunted and attacked by the weaver ants. However, most of the time the Oecophylla are relatively indifferent, and the ants are able to avoid capture by quick, skillful movements. Other species, such as a common Camponotus of the region, are never found on the same trees as the Oecophylla. When workers are placed on an Oecophylla territory, the weaver ants react by recruiting masses of defending workers. The response is as strong as that to other colonies of Oecophylla and is of the kind that sometimes leads to major warfare. Oecophylla smaragdina reacts with equal violence to Pheidole megacephala, Podomyrma sp., and Iridomyrmex sp. some of its principal competitors in Melanesia and Australia. On the other hand Hölldobler (1983) observed in Queensland, Australia, that where Oecophylla smaragdina workers ventured onto a tree occupied by Podomyrma laevifrons, they were immediately attacked by the Podomyrma workers. In fact, where only 20 Oecophylla workers were released on a Podomyrma tree, Podomyrma reacted with an effective defensive recruitment. Alarming scouts summoned nestmates by laying trails with poison gland secretions. Using their massive mandibles, the Podomyrma quickly dismembered the Oecophylla workers and carried the body parts into their nests. These results suggest that Oecophylla, and possibly Camponotus, Podomyrma and other dominants in the arboreal ant mosaic, recruit large defensive forces only when confronted with the most dangerous enemies. According to Vanderplank (1960), a limited number of ant species are effective predators of Oecophylla, including Pheidole megacephala, Pheidole punctulata, Anoplolepis longinoda, and perhaps some species of Crematogaster.

Enemy specification of one form or another appears to be a very wide occurrence in the ants. Species of Camponotus and Novomessor, for example, rapidly evacuate their nests when army ants of the genus Neivamyrmex approach (LaMon and Topoff, 1981; McDonald and Topoff, 1986). The response so far as known is triggered exclusively by army ants, and it is evidently based on the recognition of chemical cues specific to Neivamyrmex. The approach of Neivamyrmex also causes Pheidole desertorum and Pheidole hyatti to evacuate their nests (Droual and Topoff, 1981; Droual, 1983, 1984).

Enemy specification may also prove to be a key process in the organization of communities of ant species. On the one hand it provides the means whereby a vulnerable species can live alongside a dominant one, in the way Pheidole dentata manages to coexist with the Solenopsis fire ants. On the other hand, it can lead to the opposite effect by excluding one species from within the territories of other species, thereby sharpening the patterns of the ant mosaics. A case of this second effect is provided by the species of Oecophylla and their co-dominants.

Character displacement
There is yet another consequence of competitive interaction on which ecological analysis can profitably focus. We have seen evidence that in order for species to coexist, it is necessary that each of them be sufficiently different to reach their equilibrial densities before eliminating their competitors, and the usual way this occurs is through differences in critical dimensions of the “niche,” namely, those parameters of habitat, nest site, diet, foraging periodicity, and other factors capable of limiting populations. Now when the ranges of two species first meet, it may be that the species are already so different that competition is negligible, and the ranges come to overlap with no difficulty. But if interference is considerable, it will be of adaptive value for the species to diverge ecologically in the zone of overlap. In the case of social insects such “character displacement” (Brown and Wilson, 1956; Futuyma, 1986; Grant, 1986) will have a dividend measurable in increased colony survival or queen production or both. In view of the kind of competition revealed by Pontin's analysis it is appropriate that one of the first and best-documented examples of ecologically based character displacement occurs in the genus Lasius (Wilson, 1955a). Lasius flavus, one of the protagonists in Pontin's study described earlier, occurs throughout the temperate portions of Europe, Asia, and North America. Over most of this range it occupies a wide array of principal habitats, including grassland and both deciduous and coniferous woodland with varying degrees of shade and drainage. In the forested portion of eastern North America, however, it encounters a very closely related species, Lasius nearcticus. There it is more restricted in habitat choice, being limited chiefly to open woodland and fields, while nearcticus occupies the darker, moister woodland habitats. Where they occur together, the two species can be distinguished by no fewer than five morphological characters, including differences in eye size, color, maxillary palp development, antennal length, and head shape. At least the first three of these reflect a greater adaptation on the part of flavus to a less subterranean existence. From the Great Plains of North America, where nearcticus is left behind, westward to the Pacific coast and beyond across Asia and Europe, flavus displays both a wider ecological range and a greater variation in each of the morphological characters sufficient to incorporate the nearcticus as well as the eastern flavus traits. In short, where flavus overlaps the range of nearcticus, it is displaced to more open habitats, and its populations display morphological characteristics correlated with specialization to these habitats.

A case of character displacement in actual progress has been recorded in fire ants by Wilson and Brown (1958a). The South American species Solenopsis invicta was introduced into the port of Mobile, Alabama, around 1918, and in the 1940s began a rapid expansion that was to extend its range over most of the southern United States by 1970. The species builds up very dense populations in open habitats, but is largely absent from woodland. Solenopsis xyloni, a closely related native species which also favors open environments, was mostly eliminated from its old range in the southern United States in only twenty years. Solenopsis geminata, on the other hand, has been only partially displaced by invicta; whereas previously geminata occurred in both open and woodland environments, now it is limited mostly to woodland where the invicta do not penetrate. There has been a concurrent morphological change in the geminata population inside the invicta range. Previously the open environment was occupied chiefly by a reddish color form and the woodland by a dark brown color form; with the advent of invicta, the reddish color form has been mostly eliminated.

The evolutionary phenomenon of character displacement may or may not be of general occurrence in social insects. It can usually be detected only by extensive studies of geographic variation and so far has been recorded only in several ant genera, including Odontomachus, Rhytidoponera, Pristomyrmex, Solenopsis, and two species groups of Lasius (see also Taylor, 1965b). It deserves closer study because of the likelihood that interspecific competition sets constraints not only on the distribution, ecology, and morphology of particular social insect species, but on their social characteristics as well. A species restricted to pieces of dead wood or some other cramping nest site will, by interspecific competition, tend to evolve a smaller mature colony size and lower its production rate of sexual forms. If the reduction is great enough, it may abandon odor trails as a form of communication.

The taxon cycle. There is another dimension to the displacement phenomenon, revealed in biogeographic analyses of entire faunistic regions. Wilson (1959a, 1961) showed that ant species invade and subsequently evolve to endemicity within New Guinea and other parts of Melanesia through a “taxon cycle” of range expansion and contraction. Most of the Melanesian fauna has been derived ultimately from Asiatic stocks entering by way of New Guinea; some invading species are able to spread beyond this great island to Queensland, to the Solomon Islands, and to other parts of outer Melanesia. A smaller part of the fauna has been derived from old Australian stocks that have entered by way of New Guinea or New Caledonia. Faunal flow from New Guinea through outer Melanesia has been largely unidirectional, with an ever diminishing number of species groups found outward from the Bismarcks to the Fiji Islands (see Figure 11-10).

From the totality of these distribution patterns, Wilson inferred the cyclical pattern of expansion depicted in Figure 11-11. Following the invasion of Melanesia (Stage I, primary), the pioneer populations may then diverge to species level (Stage II) and further diversify. Eventually the source populations outside Melanesia contract, leaving the species group as a whole peripheral and Melanesian-centered (Stage III). Endemic Melanesian species occasionally enter upon a secondary phase of expansion (Stage I, secondary) but are rarely if ever able to push beyond Australia or the Philippines.

The expanding, Stage I species are characterized on New Guinea by their greater concentration in “marginal” habitats, which have the lowest species diversity. These habitats include open lowland forest, savanna, and seashore. They are evidently the most favorable beachheads for invasion, as well as launching areas for further range expansion. Stage-I species are also characterized by their individual occurrence in a greater range of major habitats. These species also make up a significantly higher proportion of the faunas of the archipelagos of central Melanesia, including the Bismarck Archipelago, Solomon Islands, and the New Hebrides. Stage II and Stage III species, in contrast, are concentrated in the “central,” high diversity habitats of the lowland and montane rain forests.

It appears that ant species invade New Guinea by way of the marginal habitats. Evolutionary opportunity is nevertheless limited in the marginal habitats, and there is strong selection pressure favoring re-entry into the inner forest habitats. In general, Stages II and III, leading to the origin of the great bulk of the Melanesian fauna and its most distinctive elements, are played out primarily in the inner rain forest. Stage I species are characterized on the average by larger colony size and the use of trunk trails (Wilson, 1961). These, it will be recalled, are also traits of the dominant species of the boreal forests and other species-poor habitats in other parts of the world. Passage through the taxon cycle appears to entail changes in biological traits consistent with the dominance-impoverishment principle described earlier: in other words, species with large colonies dominating parts of their foraging range do best in species-poor environments.

Ecological expansion and contraction
Entry into a smaller fauna is often accompanied by ecological release. On New Guinea some of the most widespread of the Indo-Australian ant species, notably Rhytidoponera araneoides, Odontomachus simillimus, Pheidole oceanica, Pheidole sexspinosa, Pheidole umbonata, Iridomyrmex cordatus, and Oecophylla smaragdina, are mostly or entirely limited to species-poor “marginal” habitats, such as grassland and gallery forest. But in the Solomon Islands, which has a smaller native fauna, these same species also penetrate the rain forests, where they are among the most abundant species. In Vanuatu (New Hebrides), which has a truly impoverished ant fauna, the species of Odontomachus and Pheidole just listed almost wholly dominate the rain forests as well as the marginal habitats. Ecological release in the opposite direction, from central to marginal habitats, has also occurred. In Queensland and New Guinea, Turneria is a genus of rare species mostly confined to rain forests. It is also the only genus of the subfamily Dolichoderinae to have reached the northern islands of Vanuatu. On Espiritu Santo, for example, two species of Turneria are among the most abundant arboreal insects in both marginal habitats and virgin rain forest.

The degree of compression or release in new environments varies among species and is difficult to predict in advance. A case in point is the marked difference in behavior between two of the thirteen ant species that have succeeded in colonizing the Dry Tortugas, the outermost of the Florida Keys. In the presence of such a sparse fauna, Paratrechina longicornis has undergone extreme expansion. In most other parts of its range it nests primarily under and in sheltering objects on the ground in open environments. On the Dry Tortugas it is an overwhelmingly abundant ant and has taken over nest sites that are normally occupied by other species in the rest of southern Florida: tree-boles, usually occupied by species of Camponotus and Crematogaster, which are absent from the Dry Tortugas; and open soil, normally occupied by the crater nests of Conomyrma and Iridomyrmex, which genera are also absent from the Dry Tortugas. In striking contrast is the behavior of Pseudomyrmex elongatus. This ant is one of ten species that commonly nest in hollow twigs of red mangrove in southern Florida. It tends to occupy the thinnest twigs near the top of the canopy and is only moderately abundant. Pseudomyrmex elongatus is also the only member of the arboreal assemblage that has colonized the Dry Tortugas, where it has a red mangrove swamp on Bush Key virtually all to itself. Yet it is still limited primarily to thinner twigs in the canopy and, unlike Paratrechina longicornis, has not increased perceptibly in abundance (MacArthur and Wilson, 1967).

Species packing and equilibrium
It is generally thought the numbers of species inhabiting islands and other more or less closed habitats are at equilibrium, with the rate of species extinction approaching the rate of immigration of new species. The result is the more or less regular area-species curve, in which according to taxonomic group and location the number of species on individual islands increases as the fifth to third root of the land area. As a rule of thumb, the number of species doubles with every tenfold increase in area. This relation has been documented in ant faunas of different parts of the world by Wilson (1961), Baroni Urbani (1971b), Goldstein (1975), Pisarski et al. (1983), Vepsäläinen and Pisarski (1982), Ranta et al. (1983), and Boomsma et al. (1987). However, there exist universal qualities of biology at the species level which dictate that equilibria must always be quasi-equilibria. If we observe a well established fauna for short periods of time, we will note that the species extinction rate at least approximately equals the species immigration rate; but if we continue to watch it for a long period of time we will probably note a steady and very slow drift in the average species number, most likely in an upward direction. Wilson (1969) postulated four stages of quasi-equilibria:

1. Non-interactive equilibrium. This approximate balance is reached in certain cases prior to sufficiently high population densities, which make competitive exclusion and other forms of species interaction major factors in species survivorship. The condition and the following stage were documented by Simberloff and Wilson (1969) in ants and other arthropods colonizing small islands in the Florida Keys.

2. Interactive species equilibrium. When the populations of individual species become dense enough to make competitive exclusion and other forms of species interaction major factors in species survivorship, or at least significantly more important factors than when densities are very low, the equilibrium can be said to be interactive. There is no way to predict from theory the direction or velocity of change in attaining this stage. In the Florida Keys it occurred within several months and entailed changes of less than 20 percent in species numbers.

3. Assortative species equilibrium. After interactive equilibria were reached on the islands observed by Simberloff and Wilson (1969) in the Florida Keys, the composition of arthropod species continued to change rapidly. Thus new combinations of species were being generated. Inevitably, combinations of longer-lived species must accumulate by this process on island systems in general. Such species persist longer either because they are better adapted to the peculiar conditions of the local environment, or else because they are able to coexist longer with the particular set of species among which they find themselves. As a rule, we would expect such assortative equilibria to consist of a greater number of species than their antecedent interactive equilibria.

4. Evolutionary species equilibrium. If the community persists for a sufficiently long period of time, its member species can be expected to adapt genetically to local environmental conditions and to each other. The result should be the lowering of the extinction rate and consequently an increase in species diversity.

The existence of evolutionary species equilibria implies a tighter packing of species and a greater compatibility based on such mechanisms as enemy specification, species dominance hierarchies, and niche division leading to separate density-dependent controls. Although the process cannot yet be treated in the context of quantitative theory, some idea of the amount of change in species numbers can be gained by comparing newly assembled biotas with older ones containing at least some endemic species. Wilson and Taylor (1967a) took this approach in their analysis of the Polynesian ant faunas. There are no native ant species in the far-flung archipelagos east of Tonga. The islands are inhabited instead by “tramp” species carried inadvertently by man from many different parts of the world. Most were introduced by shipping during the past 200 years. The total number of such species known to occur in the Pacific area is 38. Remarkably, this synthetic fauna behaves as though it is equilibrial on the separate islands, because it follows a fairly regular area-species curve. In particular, the number of species found on individual islands increases approximately as the cube root of the area. Islands the size of Upolu, in the Samoan group, contain 15 to 25 tramp species. The islands of extreme western Polynesia also possess old, partly endemic ant faunas. Altogether, 43 undoubted native species occur in this area, of which 35 are found on Upolu. Thus the pools of tramp and native species are similar in size (38 versus 43 species), but the equilibrial number of tramp species on Upolu (22) and islands of similar size is much lower than the equilibrial number of native species occurring on Upolu (35). Wilson and Taylor concluded that the transition from the early interactive or assortative equilibria, roughly duplicated at the present time by the synthetic tramp faunas, to the evolutionary equilibria of the native faunas resulted in an increase in the number of species of 1.5 to 2 times. And since the area-species relation is a simple power function, the same result should apply to other Polynesian islands, as depicted in Figure 11-12.

The influence of parasites and predators
Students of ant ecology have recently begun to document a role of parasitism and predation in the structure of species communities. In essence, differential mortality caused by parasites and predators can put some ant species at a disadvantage with reference to others, altering not only their abundance but also their foraging behavior as the workers attempt to meet the threat.

Feener (1981) found phorid flies in Texas of the genus Apocephalus that attack soldiers of Pheidole dentata when they come out of the nests to defend food baits against Solenopsis texana. In an attempt to lay their eggs, the flies stampede the Pheidole majors, which retreat back to the nests leaving the minor workers to confront the Solenopsis. Neither the Pheidole minor workers nor the Solenopsis are parasitized by the flies. Under these circumstances the balance is tipped in favor of the Solenopsis, which dominate the baits. When the flies are absent, in the early spring and late fall, the Pheidole soldiers are able to operate in full strength, and the Pheidole largely replace the Solenopsis at the food baits. It is not difficult to see how the presence of just one effective parasite or predator can contribute to the coexistence of two otherwise incompatible ant species.

Such complex interactions among hosts, parasites, and competitors may be more common than previously realized. Feener (1981) has observed parasitic phorid flies in all warm temperate woodlands in which he has searched. Workers of Camponotus pennsylvanicus, one of the most abundant ants of the southern and eastern United States, were seen to abandon food baits on the approach of the phorid Apocephalus pergandei. These baits were then quickly occupied by other ants, including Crematogaster punctulata, Pheidole dentata, and Solenopsis geminata. Ants may go so far as to change their foraging times in response to phorids. Pheidole titanis is a specialized predator on termites in the southwestern United States and western Mexico. During the dry season it conducts its raids during the daytime. In the wet season, under pressure from Apocephalus flies that attack both soldiers and minor workers, Pheidole titanis raids only at night, when the flies are inactive. As a result fewer termites are harvested during the wet season, a differential that presumably increases opportunities for other ant species feeding on termites (Feener, personal communication).

Vertebrate predators can similarly exercise mediating effects in community organization. In the Mohave Desert the horned lizard Phrynosoma platyrhinos preys on the individually foraging harvester ant Pogonomyrmex californicus ten to a hundred times more frequently than on the trunk-trail harvesters Pogonomyrmex rugosus and Veromessor pergandei. The trunk-trail foragers are able to mob the lizards in large numbers, causing the predators to retreat. Pogonomyrmex californicus appears to use other devices to avoid the lizards, including small, inconspicuous nests, the storage of refuse in underground chambers, and the cessation of aboveground foraging when lizards appear (Rissing, 1981b).

The future of community studies
Although still in an early stage, studies in the behavioral ecology of ants have revealed a startling variety of social mechanisms that appear to adjust species to one another and hence organize local communities. We are persuaded that what is known today is only a fraction of the processes that actually exist. We also believe that the bottom-to-top approach is the best way to understand communities. In other words, it is best to start with the identification of the processes in individual species and proceed to a simulated synthesis of the community of species within the context of the entire local fauna and flora. Ants, with their great abundance and easily observed social behavior, are superb organisms for the study of community ecology.