I remember, some years ago, visiting E.O. Wilson at Harvard. For some reason, he had to leave the room for a few minutes, and during his absence I enjoyed watching a colony of weaver ants with which—I almost wrote with whom—he shared his office. At first sight their movements appeared incoherent and disorganized, but, as I watched, it became clear that they were moving a large prey item toward their nest. This impression of apparent incoherence and actual directedness is often experienced by those who, like me, are unfamiliar with ants. It is similar to the impression one receives looking down a microscope at the cells of a developing embryo. It would also be the perception of a visiting Martian observing the people in a large railway station during the rush hour. This analogy between developing organism, insect society, and human city is not new. It is very much in the minds of the authors of The Ants, as shown for example by their frequent use of the term “superorganism” for an ant colony. But it is an analogy we do not quite know what to do with: How should we make use of it, in our imagination and our analysis?

Let us first be clear that the analogy is a real one. For example, in each case the units of which the whole is made show a division of labor, whether into nerve cells, muscle cells, and fibroblasts, or into reproductives, workers, and soldiers, or into bricklayers, carpenters, and professors. In each case, there must be communication between the units to ensure integration. Perhaps most important, there is a common feature of the evolution of the three systems. In each case, the units were once noncolonial: multicellular organisms are descended from single-celled ones; the ancestors of the ants were solitary wasp-like insects; and we are descended, in the sufficiently distant past, from mammals in which no social bond existed beyond that between mother and child. So to understand the social wholes, we must understand how they evolved from solitary and independent units.

In view of these parallels, the first point, triumphantly made by Bert Hölldobler and E.O. Wilson, is that the proper study of ants is ants. It is not just that ants are ecologically important. It is estimated, for example, that one third of the animal biomass of the Amazonian rain forest is composed of ants and termites, and other ecological habitats may not be very different. Equally important, ants display an astonishing range of structure and behavior. Some of the most bizarre examples concern the modification of individual worker ants to perform tasks that in human society would be performed by inanimate tools. In honey-pot ants, liquid food is stored in grotesquely swollen workers, each the size of a pea, that hang from the ceiling of the nest chamber. In weaver ants, the nest is made by weaving leaves together with silk. Silk is produced by the grub-like larvae, but in some cases the weaving is performed by an adult worker, which holds a grub in its jaws and wields it as a man might wield a ball of string.

My own favorite ants are the leafcutters, which I met on my only protracted visit to a tropical forest, in Panama. Walking in the forest, one would occasionally encounter a great column of these ants, each bearing aloft a cut portion of leaf, or, on one psychedelic occasion, pale pink flower petals. Ants cannot digest plant fiber, nor, unlike their rivals the termites, do they harbor in their guts a population of microbes that digest the fiber for them. So what use are the leaves? Darwin’s great contemporary Henry Walter Bates suggested that the leaves were used to thatch the nest as a protection against rain, but it turns out that the leaves are chewed up by a caste of small workers, and serve as a compost on which grows a special fungus, whose threads have inflated tips on which the ants feed. The ants are farming a fungus which is specially modified in a way that makes it serve better as a food for ants.

On almost every page of The Ants I found new and fascinating facts about natural history. But what general sense can one make of the analogies between organism, colony, and human society? There are those who would argue that the analogies are potentially dangerous, and that the use, in reference to ants, of such terms as queen, caste, agriculture, war, or slave-making is misleading. Sometimes it certainly is: the function of the queen in an ant colony is quite different from the functions of either Elizabeth I or Elizabeth II. But if biologists invented pseudoclassical technical terms to replace the obvious colloquial ones, they would rightly be criticized as elitist and obscurantist. They would also fail: I still remember my own failure to introduce the term “kleptogamy” to replace the potentially offensive but universal term “sneaky-fucking” to describe a behavior widespread among male animals.


More important, if we ignored the analogies we would rob ourselves of a valuable source of insight. Let me give two, perhaps trivial, examples. In the green tree ant of Australia, aging workers leave the main nest for smaller nests near the edge of the colony territory, from which they emerge to attack invaders from other colonies. I think the authors can be excused for calling these nests “barrack nests,” and even for remarking that “a principal difference between human beings and ants is that whereas we send our young men to war, they send their old ladies.” To take a more striking difference that emerges when one compares the three systems, there appears to be nothing in either a developing organism or in an insect colony that has the regulation function of government in human society. I do not know why. A moral one might draw is that human societies would work better without governments. Perhaps a more plausible explanation is that human societies need governments because their members are genetically different from one another, whereas the cells of an individual body are genetically identical and the members of an insect colony are genetically similar.

My own guess is that neither of these responses contains the whole truth, and that an important part of the explanation lies in the ways the units communicate: by language in human society, mainly by chemical signals in an ant colony, and mainly by cell surface patterns in developing organisms. The essential distinction here is that the semantics are genetically specified in the latter two cases, but not in the first. None of these responses may be correct, but the question is interesting, and it would not be asked if we did not draw the analogies in the first place.

Analogies are fine, but we must use them properly. They cannot be normative: it would be idiotic to argue that, because some ant colonies make war, and take slaves, we should encourage warfare and reintroduce slavery. Nor can an analogy prove anything: the fact that ant colonies depend for their stability on the genetic relatedness of their members does not prove that human societies do so. In science, analogies can do two things. Used in the vague verbal form that I adopted in the last paragraph, they can suggest questions and possible answers. If the analogy is so precise that the same mathematical description can be used of two systems, then it may be possible for workers in one field to borrow mathematics from another, as biologists have borrowed the marginal value theorem and game theory from economists.

Hölldobler and Wilson are not primarily concerned with the analogies I have been discussing, although they cannot fail to be aware of them. Insofar as they have a general point to make, it is that the study of ants is important for biology as a whole, and they are clearly correct. Perhaps the two major theoretical problems that have relevance for all biologists are, first, the means whereby the actions of individual ants are integrated in the short run to give rise to coherent colony behavior, and second, the evolutionary origins and maintenance of cooperative behavior.

The interpretation of ant behavior is achieved largely by chemical signals. A number of chemical substances—pheromones—are produced by specific glands, and individual ants respond to these pheromones in genetically specified ways. Such signals are the main cause of caste differentiation in a colony, which may contain workers of different sizes (for example, the large leafcutter workers that cut and carry the leaves, and the small workers that farm the fungi underground), as well as a specialized soldier caste. Individuals can be induced by specific signals to develop into particular types. Much work has been done on two related questions. Are the numbers in the different castes such as to maximize the economic efficiency of the colony? If the relative numbers of different caste members are altered experimentally, can the colony restore the initial numbers, and if so, how? Other aspects of behavior that are regulated by chemical signals include the marking of trails while foraging, the recruitment of soldiers to sites of attack, and the capacity of colony members to recognize intruders. These chemical signals function in the short-term regulation of the colony, a process that has obvious similarities to the physiological regulation of individual organisms. As the authors point out, colony regulation is in some ways easier to study, because it is easier experimentally to alter the composition of a colony than that of an organism.


The second major theoretical problem concerns the evolutionary origin and maintenance of social behavior. Three distinct mechanisms—mutualism, parental manipulation, and kin selection—have been proposed. The idea behind mutualism is simple: if two or more individuals each benefit by cooperation in terms of “Darwinian fitness”—the expected number of offspring—then there is little difficulty in explaining the evolution of cooperation in Darwinian terms. To take a simple example, a single male lion cannot hold a pride of females, and so is unlikely to produce offspring, whereas two or more males acting together can hold a pride; hence it pays males to cooperate even if they are unrelated. A similar process has probably been important in the origin of many cases of cooperation, but clearly it cannot explain the characteristic feature of ant societies that there are sterile workers. To be an explanation, mutualism requires that everyone benefits by producing offspring, and a sterile worker certainly does not.

Parental manipulation is also a simple idea. If a female can sterilize one or more of her daughters, and by so doing can leave a greater total number of descendants (because her sterile daughters help her to raise their siblings), then this “dominating” behavior of the female will be favored by selection. Hölldobler and Wilson do not rule out the possibility that parental manipulation has played a role in the origin of ant societies, but they think that there are features of ant biology that cannot easily be explained in this way. Instead, they favor the process of kin selection. Here, the essential idea is that if a gene, A, causes an individual carrying it to act (or to be more likely to act) in such a way as to increase the survival and fertility of the relatives of that individual, then the A gene may increase in frequency in the population, even if the individual itself is less likely to produce offspring. The reason for the increase, of course, is that the relatives may also carry the gene A. The idea was first formulated in mathematical form, and applied to social insects, by the Oxford zoologist W.D. Hamilton. Its application to ants, bees, and wasps is complicated by the peculiar genetic system of these insects, in which males develop from unfertilized eggs, and hence have only a single set of chromosomes. It is a curious consequence of this fact that a female has more genes in common with her sisters than with her daughters. It might therefore pay her to stay home and look after her sisters, rather than leave home and look after her daughters.

I vividly remember reading Hamilton’s 1964 paper, in which this point was made, and feeling, as T.H. Huxley is said to have felt when he first read Darwin’s idea of natural selection, “how stupid not to have thought of that,” although I was slightly comforted by the fact that neither R.A. Fisher nor J.B.S. Haldane had thought of it either. Subsequently, people have pointed out that this idea has snags, of which the most damaging was the point made by the Australian geneticist R.H. Crozier that, even though a female has more genes in common with her sisters than with her daughters, she has fewer genes in common with her brothers. So unless she could distinguish her brothers from her sisters when they were still larvae, and raise only the latter, the peculiar genetic system would not bias her toward socially cooperative behavior. There have been a number of twists in the argument since then, and the subject has proved surprisingly hard to think clearly about. The authors review this difficult story, and conclude that kin selection has played a crucial role in the evolution of insect societies.

In the authors’ own words, this book was written “to celebrate a personal muse.” They care passionately about ants, and are able to communicate that passion. Although encyclopedic, the book is beautifully written, and has some of the most stunning animal paintings and photographs I have seen. Once a generation, our knowledge of these important and curious animals needs to be brought together. Hölldobler and Wilson have done it for our generation. Somewhere, someone will lay hands on this book, and will be inspired by it to do for the next generation what they have done for this one.

This Issue

September 27, 1990