Journal of Animal (1987), 56, 1-9

CAUSES OF ECOLOGICAL SUCCESS: THE CASE OF THE . THE SIXTH TANSLEY LECTURE* BY EDWARD 0. WILSON

Museum of ComparativeZoology, Harvard University, Cambridge,Massachusetts 02138, U.S.A.

INTRODUCTION The key remaining questions of are more ecological than genetic in nature. In spite of the continuing great vitality of genetics in general, the known mechanisms of heredity appear to be sufficient to explain the observed phenomena of evolution. This is not to say that all processes of genetic change have been discovered. It may even prove true that extragenetic constraints on embryonic development, such as fundamental physical limits on cell size and configuration, play a role in evolution. What I am suggesting instead is that nothing empirically known at the present time about the nature and rate of evolution even hints at the existence of undiscovered hereditary processes. Consequently, what we know for sure that we do not know is largely ecological in nature. And ecologically-based problems of basic importance abound. One of the more obvious is why there are a certain number of species on Earth and not some other, in particular why 10 million (if that, say, proves to be the number) and not a thousand or a hundred million. Another unanswered question is why certain groups of organisms speciate profusely and spread over the world while otherwise similar ones remain undiversified and static. And that leads ineluctably to a third, related question, the meaning of ecological success. The subject has received remarkably little attention, perhaps because it is difficult to express with any precision and may appear to be more semantic than scientific. To take an example on which all can hopefully agree, we intuitively think of the ants as a successful group. Can success be usefully defined in such organisms? And if so, which if any evolutionary strategies consistently confer it?

WHAT IS ECOLOGICAL SUCCESS? The principal force impelling evolution, natural selection, is simply all the events that cause differential survival and reproduction. It seems intuitively right and indeed can be defended on a deeper, ontological level that persistence of a phyletic line through geological time is the key measure of ecological success. For a species to have a higher probability of existence at any randomly chosen point in geological time has more meaning than for it to have a lower probability of existence, because the result is a greater potential to create new lines, influence other species: in short, to change the Earth. If this definition appears to emphasize , the qualification is correct and justified. A great deal of macroevolution has occurred through the differentialextinction of species.

* Based on the biennial Tansley Lecture of the British Ecological Society, given at Edinburgh on 18 December, 1985.

1:

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions 4~~~~~~~~~~~~~~~~~~N

.....EDWAD.0 ILSO

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions 2 Tansley lecture to the B.E.S. The half-life of species is less than 10 million years in most taxonomic groups, and almost all species that ever lived have become extinct (Raup 1986). The effects on biotic composition and evolution made by the small fraction of surviving lines have been very great. If evolution is conceivably predictable even to a limited degree, it is likely to be made so in reality by picking out the successful living groups and characterizing their biological traits. Persistence of phyletic lines may in any case seem at first a spurious measure of 'success,' since all species existing today have a direct ancestry that goes all the way back to the origin of life. But what is meant is not persistence of the line relative to all life but rather persistence relative to other phyletic lines or 'clades' (Raup & Valentine 1983) that originated at about the same geological time and occupied the same major habitat. A sharply focused example is the following: if an adaptive radiation from one ancestral species in the Oligocene Epoch produced ten species, and only one survived as an evolving clade to the present time (perhaps speciating secondarily along the way), the one was more successful than the remaining nine. On the basis of the criterion of phyletic longevity the ants are relatively very successful. They have survived as a taxonomic family, the Formicidae, since the lowermost part of the upper Cretaceous, or about 100 million years B.P. All of the Cretaceous fossils belonging to the superfamily Formicoidea, in other words ants and their immediate ancestors, can be referred to the single formicid genus Sphecomyrma.The Sphecomyrmi- nae originate from localities across a wide portion of old Laurasia, from present-day Kazakhstan, north-central Siberia, and the Magadan region of easternmost Siberia to Alberta, Canada, and New Jersey, U.S.A. (Wilson 1985a, 1987a). Eomyrmex gucheng- ziensis, a species which I interpret to be intermediate between Sphecomyrma and the contemporary primitive subfamily Ponerinae, has been recorded from the early Eocene Fushan deposits of Manchuria (Hong et al. 1974). By mid-Eocene times, at least three modern subfamilies, the Myrmicinae, Dolichoderinae and Formicinae, had appeared, as represented by recently discovered fossils in the Arkansas amber (Wilson 1985a). By comparison with other animal groups, the Formicidae are older than nineteen of the twenty living mammalian orders (that is, all but the Marsupalia). Of the twenty mammalian families that apparently originated during the Cretaceous Period, only one, the Didelphidae or opossums, survived to the present time and hence are as old as the ants (Clemens et al. 1979). The Formicidae are also older than all living families of birds and all but three of the twenty-eight living orders (Phoenicopteriformes, Gruiformes, Charadrii- formes). Three of the bird orders originating in the Cretaceous, also those just listed, survive today, while the remaining two (Hesperornithiformes, Ichthyornithiformes) were confined to that geological period. These comparisons are impressive but of course wholly dependent on the similarity of orders and families as these categories are subjectively judged by taxonomists working independently on the insects and vertebrates respectively. At closer range, the ants are not exceptional among the families of the . As shown in Table 1, no fewer thari twenty-five of the thirty-six families known from Cretaceous fossils, including the Formicidae, are still represented by living species. One remarkable group, the primitive sawflies of the family Xyelidae, have survived since the Triassic Period. Turning to the next lower taxonomic level, only one of the five genera (Iridomyrmex) thus far recorded from the Cretaceous and Eocene deposits is extant. However, no fewer than twenty-four genera, or 56% of the forty-three total representedin the early Oligocene Baltic amber ant fossils still survive, including such currently

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions E. 0. WILSON 3 TABLE1. Hymenopteranfamilies in the geologicalrecord, by periods(based on Carpenter1986) Triassic Jurassic Cretaceous Tertiary Extinct Recent Extinct Recent Extinct Recent Extinct Recent No. families 0 1 14 8 11 25 4 52

0/) Families 0 100 64 36 31 69 7 93 abundant and widespread taxa as Ponera, Tetraponera,Aphaenogaster, Monomorium, Iridomyrmex, Formica, Lasius, and Camponotus (Wheeler 1914). At least one species, Lasius schiefferdeckeri,is so close to living species of the L. niger group that it can be distinguished only by minor overlapping traits in antennal and mandibular form (Wilson 1955). This modern aspect is even more evident in the Dominican amber, which is apparently early Miocene in age. Here no fewer than thirty-four genera, or 92% of the total thirty-seven, still survive. Furthermore, the great majority of species thus far analysed have been placed in modern species groups, and in a few instances are difficult to separate at the species level (Wilson 1985b). Modern taxonomic studies of most other hymenopteran families in the Oligocene and Miocene deposits are still too incomplete to allow quantitative comparisons with the ants at the generic and specific levels. However, enough data are available on the Bethylidae and Braconidae of the Baltic amber to indicate percentages of generic survival comparable to that in the ants (see review by Larsson 1978). In summary, the ants are notable as a whole for their persistence as a clearly defined taxonomic family and hence of a clade that presumably originated as a single ancestral species during the Cretaceous Period. Genera and species groups, thought to represent radiations from post-Sphecomyrma lines, are also relatively ancient. The ants substan- tially exceed warm-blooded vertebrates in longevity at all taxonomic levels, but not, so far as the limited fossil data suggest, the families and genera of solitary Hymenoptera. Sheer longevity depends upon at least four general qualities at the population level, which will now be considered in turn. (1) Number of species generated through time. An ancestral line that penetrates many adaptive zones is more successful than one that survives as a single species, because through unintended effect it has 'balanced its investments' and ceteris paribus will probably persist longer into the future. The described world ant fauna consists of approximately 300 genera and 8800 species. The number of still unrecognized genera is evidently small. Judging from the rate of discovery during the past 50 years and the alacrity with which uncontestable novelties at the generic level are publicized, I would venture to guess that fewer than 100 still remain unrecognized. In contrast, the number of undescribed species is evidently very large. An extreme example is provided by Pheidole, the most speciose of all ant genera. In addition to the 524 currently unchallenged names in the New World alone, of which at least half probably represent valid species, I estimate that several hundred undescribed species are present in collections. It is quite possible that 20 000 or more species of ants constituting 350 genera exist in the world. Furthermore, a high diversity has characterizedthe Formicidae throughout most of the Tertiary Period. The fossils most thoroughly studied to date, those in the Baltic amber, belong to forty-three genera and ninety-two species (Wheeler 1914). The next best known

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions 4 Tansley lecture to the B.E.S. deposit, the Dominican amber, contains thirty-seven genera and a large but still undetermined number of species that I would roughly estimate (directly from available material) to be between fifty and eighty. The Dominican amber fauna is especially instructive because it lived on an island (Hispaniola, which may have been united with the remainder of the Greater Antilles) and therefore can be compared directly with the modern circumscribed fauna of Hispaniola. In fact, the known modern fauna comprises thirty-seven genera and eighty-eight species, a remarkablecorrespondence to the diversity existing on the island in Miocene times (Wilson 1985b). (2) Whetheror not one or more derivativelines uniquelyoccupy a major adaptivezone. If this level of adaptive radiation has been attained, the phyletic assemblage as a whole is less likely to suffer extinction through competitive exclusion. The ants have accomplished the feat several times, creating new unique adaptive types. The fungus-growing ants (tribe Attini) are among the few animals of any kind that utilize symbiotic fungi or yeasts as food, sharing the trait with a few taxa (in the Anobiidae, Brenthidae, Curculionidae, Lymexylidae, and Scolytidae) as well as the macrotermitine (Wilson 1971; Batra 1979). One apparently monophyletic group within the Attini, the twenty-four species of Acromyrmex and fifteen species of Atta (leafcutter ants), are unique within the animal kingdom in their ability to convert large quantities of fresh vegetation into a fungus substrate (Weber 1972). They accomplish this feat through an assembly-line division of labour in which the vegetation fragments are passed from one group of workers to slightly smaller workers and, thus, on down through four steps for serial processing into a highly modified substrate (Wilson 1978). The leafcutter ants are extremely abundant throughout the New World tropics, where they are more effective herbivores than any other animal group of comparable species diversity, as well as among the most destructive and versatile insect pests (Cherrett 1982). Ants of the tribe Cerapachyini (subfamily Ponerinae) and the famous army and driver ants, which according to Gotwald (1979) comprise three independently derived tribes in the Dorylinae and Ecitoninae, are unique in their use of group foraging to overwhelm large and formidable prey, including colonies of termites and other species of ants (Schneirla 1971). A wholly independent derivation of group foraging to capture insects has recently been discovered in the myrmicine genus Pheidologeton by Moffett (1984). Ecitonine army ants of the genus Neivamyrmex have existed at least since Miocene times in the New World, as evidenced by a recent find in the Dominican amber (Wilson 1985b). Many ant species keep aphids, scale insects, pseudococcids, and other homopterous insects as 'cattle' in their nests, moving them from one site to another in response to danger or food shortage. Some, including species of Acanthomyops, Acropyga, and Melissotarsus, depend largely or entirely on the symbionts in their care (see, for example, Delage-Darchen 1972). Perhaps the most remarkable species of all is the Malaysian Diabolus cuspidatus,which Maschwitz & Hanel (1985) have recently shown to be a true herding species. These large ants move their pseudococcids from one set of arboreal nesting and feeding sites to another in a continuously shifting pattern. (3) The population density that can be sustainedfor long periods of time. Mathematical models of population density support the intuitive principle that the larger the population and the less the numerical fluctuations in time, the longer the waiting time to extinction (Frankel & Soule 1981). In addition, the population is less likely to lose fitness through inbreeding depression. A rough principle used in conservation biology is the '500-50' rule: when the effective breeding size of the population falls below about 500 individuals the

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions E. 0. WILSON 5 rate at which alleles are lost through random fixation rapidly increases; and when it falls much below 50 individuals, the percentage of homozygotes rises sharply (see, for example, Senner 1980). Many ant species are fabled for their high population densities, huge nuptial swarms, and hence large effective breeding sizes (Wilson 1971; Brian 1983). Ant faunas as a whole are extremely dense in comparison with those of other animals, including other insects. In rain forest near Manaus, Brazil, they make up about one-fifth of the total animal biomass (Fittkau & Klinge 1973). This proportion is probably close to that prevailing in most other tropical and warm-temperatehabitats, from deserts to forests of every kind with the exception of montane cloud forests. In the Amazonian forests of Brazil and Peru, ants are by far the most abundant insect group, exceeding , lepidopterans, and other comparably ranked taxa considered individually (Erwin 1983 & personal communica- tion). From one tree in the Tambopata Reserve of Peru that was fogged with an insecticide and thoroughly censused for ants, I identified forty-three species belonging to twenty-six genera: a diversity approximately equal to that of the entire ant fauna of the British Isles (Wilson 1987b). In addition, the effective breeding sizes of ant populations may be generally more stable than those of most other, non-social insect taxa, including the solitary Hymenoptera. The reason is that the basic breeding unit is the colony, or more precisely the single queen or relatively few such individuals contained therein. As documented in the pioneering work of Pickles (1940), a drastic reduction can occur in the number of individual workers, reducing the average population per colony, without changing the number of colonies or queens very much. When conditions improve, the colonies serve as nuclei in the rapid restoration of the overall populations of workers-again, without greatly altering the number of colonies, queens, and hence the effective population size. However, field studies are too few to establish whether ant populations are in fact generally more stable than those of non-social hymenopterans and other insects occupying the same habitats. Ants weigh heavily on most terrestrial ecosystems and have profound effects on the biology of other animal groups. To take a less familiar example, it has been my impression that ants reduce the abundance of ground-dwelling carabid beetles and spiders, especially when these arthropods are specialized for existence in the soil and rotting vegetation. In mountain zones high enough to be mostly free of ants, such as the summit of Mt Mitchell in North Carolina and the Sarawaget Mountains of Papua New Guinea above 2500 m, carabids and spiders appear to increase markedly in abundance. Darlington (1971), a premier collector of the Carabidae worldwide, reached the same conclusion with reference to these insects. He considered the effect to be more pronounced in the tropics than in the cold temperate zones and greatest of all for wingless ground-dwelling species, at least in New Guinea. (4) Breadth of geographic range. The wider the geographic range, the less likely is it that the entire clade will be reduced to extinction prior to microevolutionary adjustments and renewal of population growth in one or more of the species. The Formicidae are one of the most widespread of all groups of organisms of comparable taxonomic rank. They range from the Arctic circle to the southernmost reaches of Tasmania, Tierra del Fuego, and southern Africa. The only places free of native species are Antarctica, Iceland, Greenland, Polynesia east of Tonga, and a few of the most remote islands in the Atlantic and Indian Oceans. Four genera (Camponotus, Crematogaster, Hypoponera, and Pheidole) extend over most of this range separately.

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions 6 Tansley lecture to the B.E.S. SPECIFIC CAUSES OF SUCCESS

The general population-level qualities contributing to phyletic longevity are usefully distinguished from the individual and colony-level traits that confer these qualities on particular species. The question of interest is: What unusual or unique biological traits have enabled the ants to remain abundant and relatively unchallenged morphologically for over 50 million years? The answer appears to be that the ants were the first group of predatory eusocial insects to both live and forage primarily in the soil and rotting vegetation on the ground. After entering this major adaptive zone no later than the Upper Cretaceous, they appear largely to have preempted its occupation by other candidate groups among the insects. Eusocial behavior of any kind is a rare achievement among insects. From the evidence of living groups, it has originated about twelve times within the Hymenoptera and once in the protoblattoid line that gave rise to the termites (Wilson 1971; Michener 1974). By definition is the combination of brood care, overlap of generations, and, most importantly, the division of colony members into reproductive and non-reproductive castes. Ants are far advanced in all of these traits. Tightly organized colonies of the kind made possible by eusociality enjoy several key advantages over solitary individuals. Under most circumstances groups of workers can better forage for food and defend the nest, because they are able to switch from individual to group response and back again, swiftly and according to need. Individual ant workers for the most part can perform as competently as individual solitary : except, of course, in the case of reproduction. When a food object or nest intruder is too large for one worker to handle, nestmates can be assembled by alarm or recruitment signals. Of equal importance, the execution of multiple-step tasks is accomplished in a series-parallel sequence instead of a parallel-series sequence (Oster & Wilson 1978). That is, individual ants can specialize on particular steps, moving from one object (such as a larva to be fed) to another (a second larva to be fed). They do not need to carry each task to completion from start to finish as, for example, first check the larva, then collect the food, then feed the same larva. Hence, if each link in the chain has many workers in attendance, a series directed at any particular object (a given hungry larva) is less likely to fail. Moreover, ants specializing on particular labour categories typically constitute a caste specialized by either age, body form, or both. There has been some documentation of the superiority in performance and net energetic yield of various castes for their modal tasks, but careful experimental studies are still relatively few in number (e.g. Porter & Tschinkel 1985; Wilson 1985c). That much being noted, what makes ants unusual even within the select company of eusocial insects is the fact that the workers are all wingless and typically well adapted for nesting and foraging in the soil and rotting vegetation. Ants are the only eusocial predators occupying this major niche. The termites live in the same places and also have wingless workers, but they feed almost exclusively on dead vegetation. The ants have a number of special adaptations fitting them for their special way of life. One of the most striking is the elongation of the mandibles into working tools. The primitive formicid mandible is a blade whose inner border is lined with a row of sharp teeth used for gripping and cutting. This basic shape is found in most species among both the primitive and advanced subfamilies, but it has been altered in a few to resemble sickles and other shapes, which serve either for the capture of unusual prey or as fighting instruments in the defence of the colony.

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions E. 0. WILSON 7 A second important innovation is the metapleural gland, a pair of cell clusters that open into chambers located at the extreme rear corners of the major middle portion of the body. The gland produces phenylacetic acid and other antiseptic substances. It may well have played a role in the successful colonization of the moist, micro-organism-ridden environment in which the great majority of ant species occur (Maschwitz, Koob & Schildknecht 1970). The metapleural gland is the closest we have to a single diagnostic character separating the Formicidae from all other aculeate Hymenoptera. However, it has been secondarily lost in a few phyletic lines, chiefly those specializing in arboreal and hence drier, cleaner environments (H6lldobler & Engel-Siegel 1984).

HOW TO AVOID CIRCULAR REASONING Broad evolutionary arguments of the kind I have just advanced are particularly susceptible to circular or ex post facto reasoning about cause and effect. In its fatal form the circularity goes as follows: group x is successful; group x has traits n1,n2, n3. . . ; hence traits n1, n2, n3 ... made group x successful and they are to be regarded as generally success-promoting. But this conclusion might be entirely wrong. A closer look could reveal that group y also possesses n1, n2, n3 ... and is on the verge of extinction after a relatively short geological existence. Hence, in order either to strengthen or disconfirm the hypothesis about the causes of success (n1, n2, n3, .. .) it is necessary to compare the attributes of success among many clades that either possess or lack the listed traits. In the case of ants and other social insects this procedure leads to a provisional support of the hypothesis. Termites are the only other group of insects that remotely approach the ants in their combination of advanced eusociality, wingless worker caste, and residence in the soil and decaying vegetation. Also, like the ants they are dramatically successful by the respective criteria of long geological tenure (Lower Cretaceous origin at the latest), high species diversity, penetration of unusual and unique adaptive zones, large populations, and wide geographic range. It is a remarkable fact that, so far as is known, no major insect clade possessing an advanced eusocial organization is anything less than very old and very successful. With reference to the five criteria, the eusocial wasps and bees rank below the ants and termites but are still far above the great majority of solitary phyletic groups. In the rain forest near Manaus, for example, these four eusocial groups constitute an astonishing 75% of the entire insect biomass (Fittkau & Klinge 1973). Although such comprehensive studies have not been made elsewhere, my own subjective impression is that the four groups will prove to make up most of the insect biomass in the majority of biomes around the world. It appears that with reference to sheer preponderance, insect life became primarily social no later than early Tertiary times, about 50 million years ago. Finally, no principal phyletic group that attained advanced eusociality, with large colonies and well-marked castes, is ever known to have failed. For example, no 'para-formicid' or 'para-isopteran' or anything like them hangs on in rareness and geographic isolation in a peripheral part of the world's land mass; none has ever been encountered in the fossil record.

CONCLUSIONS If we accept length of life through geological time as the ultimate criterion of ecological success for a species or its descendants, at least four very general causes can be deduced

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions 8 Tansley lecture to the B.E.S. that can act singly or in combination. They are the number of the species in the monophyletic group, the occupation of unusual adaptive zones, the magnitude and numerical fluctuations in time of the effective population size, and the breadth of total geographical range. The ants (Formicidae) rank extremely high in all four of these qualities. Species longevity and its general causes are attributes that emerge at the population level. They are usefully distinguished from the underlying specific causes, which are particular traits of individual organisms (or colonies) that vary from one organism to the next. In the case of the ants the causes appear to be characters associated with advanced eusocial organization, predatory behaviour by groups, and life spent in the soil and rotting vegetation on the ground. Such analyses concerning specific causes are always endangered by circular reasoning; the successful group has certain traits, therefore the traits caused the success. This difficulty can be diminished by seeing whether the postulated traits are associated with phyletic longevity and its population-level causes in other, parallel lines. The termites, which are closest to the ants in geological age, social organization, concurrent operations, and habitat, are also highly successful. So in fact are all of the other advanced eusocial insects. The ants, termites, social bees, and social wasps, for example, constitute about 75% of the insect biomass of the Amazonian rain forest. No known advanced eusocial group ever evolved that was not highly successful by the population-level criteria just cited.

ACKNOWLEDGMENTS

I am grateful to William L. Brown, Bert Holldobler, and Mark W. Moffett for a critical reading of the manuscript, and to Frank M. Carpenter for a valuable summary of the fossil history of the Hymenoptera based on his monumental treatise on insect paleontology. I am also indebted to A. W. Crompton and R. A. Paynter, Jr, for information and advice on the Cretaceous and birds. My recent research has been supported by National Science Foundation Grants BSR-81-19350 and BSR-84- 21062.

REFERENCES

Batra, L. R. (Ed.) (1979). Insect-Fungus Symbiosis: Nutrition, Mutualism, and Commensalism. Wiley, New York. Brian, M. V. (1983). Social Insects. Ecology and BehaviouralBiology. Chapman & Hall, London. Carpenter, F. M. (1986). Hexapoda. Classes Collembola, Protura, Diplura, Insecta. Part R, Arthropoda, 4, Treatise on Invertebrate Paleontology (Ed. by R. A. Robison), (in press). The Geological Society of America, Inc. and The University of Kansas, Lawrence, Kansas. Cherrett,J. M. (1982). The economic importance of leaf-cutting ants. The Biology of Social Insects (Ed. by M. D. Breed, C. D. Michener & H. E. Evans), pp. 114-118. Westview Press, Boulder, CO. Clemens,W. A., Lillegraven,J. A., Lindsay, E. H. & Simpson, G. G. (1979). Where, when, and what-a survey of known Mesozoic distribution. Mesozoic Mammals: The First Two-thirdsof Mammalian History (Ed. by J. A. Lillegraven, Z. Kielan-Daworowska, & W. A. Clemens), pp. 7-58. University of California Press, Berkeley. Darlington, P. J. (1971). The carabid beetles of New Guinea. Part IV. General considerations; analysis and history of fauna; taxonomic supplement. Bulletin of the Museum of ComparativeZoology, Harvard, 142, 129-337. Delage-Darchen, B. (1972). Une fourmi de C6te-d'Ivoire: Melissotarsus titubansDel., n. sp. Insectes Sociaux, 19, 213-226.

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions E. 0. WILSON 9 Erwin, T. L. (1983). Beetles and other insects of tropical forest canopies at Manaus, Brazil, sampled by insecticidal fogging. TropicalRain Forest: Ecology and Management (Ed. by W. L. Sutton, T. C. Whitmore, & A. C. Chadwick), pp. 59-75. Blackwell Scientific Publications, Oxford. Fittkau, E. J. & Klinge, H. (1973). On biomass and trophic structure of the central Amazonian rain forest ecosystem. Biotropica, 5, 2-14. Frankel, 0. H. & Soule, M. E. (1981). Conservationand Evolution. Cambridge University Press, Cambridge. Gotwald, W. (1979). Phylogenetic implications of army ant zoogeography (Hymenoptera: Formicidae). Annals of the Entomological Society of America, 72, 462-467. Holldobler, B. & Engel-Siegel, H. (1984). On the metapleural gland of ants. Psyche (Cambridge), 91, 201-224. Hong, Y.-C., Yang, T.-C., Wang, S.-T., Wang, S.-E., Li, Y.-K., Sun, M.-R., Sun, H.-C. & Tu, N.-C. (1974). Stratigraphyand paleontology of Fushun coal-field, Liaoning Province. Acta Geologica Sinica, 2,113-149. Larsson, S. G. (1978). Baltic amber: A palaeobiological study. Entomonograph,1, 1-192. Maschwitz, U. & Hanel, H. (1985). The migrating herdsman Dolichoderus (Diabolus) cuspidatus:an ant with a novel mode. Behavioral Ecology and , 17, 171-184. Maschwitz, U., Koob, K., & Schildknecht, H. (1970). Ein Beitrag zur Funktion der Metathoracaldriise der Ameisen. Journal of Insect , 16, 387-404. Michener, C. D. (1974). The Social Behavior of the Bees. A Comparative Study. Belknap Press of Harvard University Press, Cambridge, MA. Moffett, M. (1984). Swarm raiding in a myrmicine ant. Naturwissenschaften,71, 588-589. Oster, G. F. & Wilson, E. 0. (1978). Caste and Ecology in the Social Insects. Princeton University Press, Princeton, NJ. Pickles, W. (1940). Fluctuations in the populations, weights and biomasses of ants at Thornhill, Yorkshire, from 1935 to 1939. Transactionsof the Royal Entomological Society of London, 90, 467-485. Porter, S. D. & Tschinkel, W. R. (1985). Fire ant polymorphism: The ergonomics of brood production. Behavioral Ecology and Sociobiology, 16, 323-336. Raup, D. M. (1986). Biological extinction in Earth history. Science, 231, 1528-1533. Raup, D. M. & Valentine, J. W. (1983). Multiple origins of life. Proceedings of the National Academy of Sciences, 80, 2981-2984. Schneirla, T. C. (1971). Army Ants: A Study in Social Organization(Ed. by H. R. Topoff). W. H. Freeman, San Francisco. Senner, J. W. (1980). Inbreeding depression and the survival of zoo populations. ConservationBiology: An Evolutionary-Ecological Perspective (Ed. by M. E. Soule & B. A. Wilcox), pp. 209-224. Sinauer, Sunderland, MA. Weber, N. A. (1972). GardeningAnts: The Attines (Memoirs of the American Philosophical Society, No. 92). American Philosophical Society, Philadelphia. Wheeler, W. M. (1914). The ants of the Baltic amber. Physikalisch-okonomischenGesellschaft zu Konigsberg,55, 1-142. Wilson, E. 0. (1955). A monographic revision of the ant genus Lasius. Bulletin of the Museum of Comparative , Harvard, 113, 1-205. Wilson, E. 0. (1971). The Insect Societies. Belknap Press of Harvard University Press, Cambridge, MA. Wilson, E. 0. (1978). Caste and division of labor in leaf-cutter ants (Hymenoptera: Formicidae), I: The overall pattern in A. sexdens. Behavioral Ecology and Sociobiology, 7, 143-156. Wilson, E. 0. (1985a). Ants from the Cretaceous and Eocene amber of North America. Psyche (Cambridge), 92, 205-216. Wilson, E. 0. (1985b). Invasion and extinction in the West Indian ant fauna: Evidence from the Dominican amber. Science, 229, 265-267. Wilson, E. 0. (1985c). The sociogenesis of insect colonies. Science, 228, 1489-1495. Wilson, E. 0. (1987a). The earliest known ants: An analysis of the Cretaceous fossils. Paleobiology (in press). Wilson, E. 0. (1987b). The arboreal ant fauna of Peruvian Amazon forests: a first assessment. Biotropica (in press).

(Received 27 May 1986)

This content downloaded from 131.215.225.9 on Thu, 30 Jan 2014 22:59:41 PM All use subject to JSTOR Terms and Conditions