Proc. Nat. Acad. Sci. USA Vol. 68, No. 6, pp. 1254-1258, June 1971

Interconnected Patterns of Biogeography and Evolution (carabid beetles/South Pacific/New Guinea/dispersal/group selection) PHILIP J. DARLINGTON, JR. Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts 02138 Communicated April 12, 1971

ABSTRACT Analysis of the fauna ofthe carabid beetles that are represented in the lowland fauna (because the of New Guinea reveals both a broad dispersal pattern and a local turnover pattern that together fit into a world-wide details of their origins are more obscure for the mountain- pattern of successive dispersals and replacements that run living stocks). The arrows in Fig. 2 show directions and from large to small areas and from more to less favorable distances of dispersals, and the widths of the arrows are climates. This pattern coincides broadly with a world-wide proportional to the numbers of separate stocks that have pattern of species numbers. Evolution by group selection, apparently made the dispersals, the actual numbers being proceeding most rapidly and effectively where species are most numerous, connects the patterns and can supply given on the arrows. It will be seen that dispersal has been the force that gives direction to the dispersal pattern. predominantly from the Oriental area toward Australia, and Directional change at any level of complexity involves that the disparity increases with distance. Among stocks that movement that results in the formation of diverse groups have reached only to New Guinea, the ratio of Oriental to of units (which are themselves groups of smaller units) and differential survival. This process-generalized group Australian stocks is only about 3:2 (37:25); among those selection-has been continuous from chemical evolution that have reached Cape York in one direction and the Moluc- on the earth's surface, through the origin of life, and into cas in the other, more than 2:1 (9:4); and among those that successive interacting levels of organic evolution. A corol- have reached the main part of Australia or the Oriental area lary is that evolution should make situations favorable to respectively, 5:1 (35:7). And, among stocks that have made itself, by group selection, and has probably done so in (for example) tropical rain forest, where new information still longer movements (not distinguished in Fig. 2), the about group evolution may be sought. disparity is still greater; many Carabidae (including some groups not now represented on New Guinea) seem to have Completion of taxonomic work on the carabid beetles (Family dispersed from the mainland of Asia to Australia, while only Carabidae, "predaceous ground beetles") of New Guinea (1) one primarily Australian carabid stock (the genus Gnatha- permits analysis of the New Guinean carabid fauna as a phanus) seems to have reached the mainland of Asia. whole. The results will be reported in detail elsewhere (2). A world-wide dispersal pattern, formed by successive The present paper summarizes new evidence of fundamental dispersals of dominant groups from large to small areas and geographic patterns of dispersal and evolution, suggests from more to less favorable climates, with some counter- connections among the patterns, and considers their con- dispersals, but with the latter fewer and shorter than the sequences. main dispersals, is indicated by the distributions and fossil FACTS AND PATTERNS records of vertebrates (4). Wilson (5) has shown that the dispersals of ants in the Oriental-Australian area fit this Broad dispersal patterns pattern, and the dispersals of New Guinean Carabidae The New Guinean fauna of Carabidae proper (excluding tiger (Fig. 2) fit it too. So far as I know, these ants and beetles beetles, subfamily Cicindelinae) consists of 667 known species, supply the first detailed evidence that major groups of of which 434 constitute the lowland fauna, defined as including invertebrates do fit the vertebrate dispersal pattern. Note all species found below an altitude of 500 m. Most of the that the existence of the pattern implies a force, presumably species are small (mean length usually 12 mm or less, rarely generated by evolution, that is centered in large and favorable over 20 mm) and all are winged or derived from ancestors areas, and gives direction to dispersals. that were or may have been winged when they reached New Guinea. These characteristics (which facilitate dispersal) Dispersal in relation to time; turnover patterns and also the geographic relationships of the insects indicate The numbers of dispersals of New Guinean Carabidae are that New Guinean Carabidae are primarily an accumulation plotted against time in Fig. 3. Only time is considered in this of actively dispersing forms. The numbers, directions, and case, not direction or distance, and all carabid stocks now times of their dispersals are all significant. represented on New Guinea are included. In making the The directions and distances of dispersals apparently made correlation with time, it is assumed that nonendemic species by existing New Guinean carabid stocks within the area reached New Guinea most recently; that endemic species are mapped in Fig. 1 are summarized in Fig. 2. [Criteria for older on the island, and that their ancestors arrived during a determining directions of dispersal are discussed elsewhere longer span of time (necessary for evolution of species); and (3).] In this connection, only those stocks of New Guinean that endemic genera are still older, and that their ancestors Carabidae have been counted that do yield reasonable evi- arrived during a still longer time span. It is assumed also that dences of direction of dispersal (some do not), and only those the endemic species and endemic genera vary in age, and that 1254 Downloaded by guest on September 30, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Patterns of Biogeography and Evolution 1255

ORIENTAL I I i AUSTRALIA AREA MOLUCCAS NEW GUINEA: CAPE YORK BELOW C.Y. I

37

I

35 K2111I11111<9 25 .7 FIG. 1. Map of areas distinguished in Fig. 2. FIG. 2. Dispersals of New Guinean Carabidae. the more recent outnumber the older stocks in each category; the endemic species and endemic genera are accordingly di- vided into subgroups that are arbitrarily fitted to an orderly world-wide pattern of species numbers, which also runs from curve in Fig. 3. The numbers of separate stocks, each repre- large to small areas and from more to less favorable climates. senting one dispersal, have been counted or (in the older The coincidence suggests that the patterns may be connected. groups) closely estimated. Fig. 3 is thus based on a combina- And evolution with group selection offers a connection. tion of assumptions, arbitrary procedures, and actual counts. Group selection of demes and species However, the essential fact seems beyond question: many dispersals are represented in the most recent part of the Group selection is defined (by description) below. It has been New Guinean carabid fauna, very few in the oldest part. emphasized by Wynne-Edwards (8) and others to account for This correlation of amount of dispersal with time has two the evolution of group phenomena (e.g., peck orders and possible explanations. One is that New Guinea was formerly "altruistic" behaviors), and de-emphasized by Williams (9), more isolated than now and more difficult for Carabidae to who undertakes "a defense of the thesis that group-related reach. The other is that turnovers have occurred in the phenomena do not, in fact, exist", and who argues that group New Guinean fauna, with continual extinctions of older behaviors can evolve by gene substitutions that arise by groups and replacements by more recently dispersing ones. selection of individuals. In favor of the second explanation is the fact that the whole The concept of group selection is best approached in the existing New Guinean carabid fauna seems more recent in its same way that the concept of individual selection was ap- origins than the faunas of New Caledonia and New Zealand, proached historically. Understanding of the role of selection as if old stocks have been replaced on New Guinea. And in evolution began with the realization (by Darwin and Wal- turnovers on New Guinea would be consistent with the lace, after reading Malthus) that more individuals come into turnovers (studied chiefly among ants and birds) that have existence than can survive, that fitness varies, that fitter occurred on smaller islands east of New Guinea and elsewhere individuals must tend to survive and reproduce while less-fit (6). do not, and that a consequence-evolution by individual selection-can be inferred. This is still all that can be said The world-wide pattern of species numbers about individual selection in most cases: it is still usually Species of plants and animals are most numerous-the largest inferred, very rarely demonstrated. Similarly, it can be ob- numbers occur together, e.g., in 1 km- in the largest and served or realized that more groups (demes, species, etc.) most favorable (and most stable) areas, which are in the come into existence than can survive, that fitness of groups continental tropics; fewer occur in smaller areas and less varies and that fitter groups must tend to survive and "re- favorable climates, and very few occur on small isolated produce" (e.g., by multiplications of species) while less-fit islands and in very cold or otherwise inhospitable places. This is one of the fundamental patterns of biogeography (7). ' NON-E ANCESTORS OF ANCESTORS OF THEORY SPEC. ENDEMIC SPECIES ENDEMIC GENERA IN NON-ENDEMIC The patterns correlated GENERA The carabid fauna of New Guinea reflects and connects two 137 138 16 dispersal patterns: a world-wide pattern in which the sum of many dispersals has a net direction, from large to small areas C,, MORE 0 RECENT and from more to less favorable climates; and an insular U, turnover pattern, formed by continual extinctions and new dispersals. The two patterns seem in fact to be the center and m111 OLDER a: margin of a single combined pattern of evolution of successive w groups of plants and animals in large, favorable areas, and of MORE dispersals outward from these areas, with extinctions and RECENT OLDER z replacements occurring throughout the pattern but most conspicuous at the margin. This combined pattern of evolu- TIME tion, dispersal, and replacement coincides broadly with the FIG. 3. Dispersal of New Guinean Carabidae in relation to time. Downloaded by guest on September 30, 2021 _ 1256 Zoology: Darlington Proc. Nat. Acad. Sci. USA 68 (1971)

DISPERSAL FIG. 4. Dendrograms of evolution of: A, four independent species; B, one widely distributed species of four interbreeding FIG. 5. Diagram of components of dispersal and evolution in demes; C, four species with repeated group selection and re- group selection. A, main evolution center; B, secondary evolution radiation. center; C, small marginal area. Each horizontal line represents dispersal of one or more species derived by radiation from one ancestor; each vertical line, one step up a "ladder of successive and that a group become extinct, consequence-evolution by favorable adaptations" (see text). Extinctions are not indicated. selection-can be inferred. Strictly speaking, this is a diagram of phylad selection in one Group selection need not involve competition among phylogenetic line; in actual cases, selection involves complex sibling groups; local groups may become extinct for many interactions of many such lines, related and unrelated. reasons other than competition among siblings, and other (related or unrelated) groups may reoccupy the vacated areas evolution is not likely to be effective in producing generally later. And group selection must usually occur through selection adapted plants and animals. Large panmictic populations may of individuals; groups do not usually become extinct instan- evolve more effectively, but their mass retards their rate of taneously but by death or failure to reproduce individuals evolution. But if a large population is divided into many during many generations. Nevertheless, some groups do localized demes that evolve partly independently, but inter- become extinct and are replaced by other groups which, on the breed occasionally (Fig. 4B), it can store much recombination average, are presumably more fit. In this sense, group selec- variability, can adapt rapidly to both temporary and perma- tion does occur. nent changes in the environment, and by "trial and error" can enter new adaptive zones even when the environment is How much group selection? essentially stable, and can rapidly exploit new major ecologic The numbers of groups that can exist in limited areas are opportunities (13). In short, a large population divided into limited; if demes and species multiply enormously, as they occasionally interbreeding demes is in a position to evolve sometimes do, enormous numbers must become extinct, as both rapidly and effectively. Darwin (10) noted. The fossil record shows enormous numbers If group selection too occurs among the interbreeding of replacements at and above the species level, although it demes, it should remove some less-fit demes and thus accel- cannot show them among demes that are too-little differen- erate adaptation And, if isolating mechanisms evolve so that tiated to be distinguished as fossils. Many species of plants interbreeding among demes ceases, group selection will con- and animals live on unstable "checkerboards" (discontinuous, tinue at the species level, and should continue to accelerate shifting habitats) and are thus divided into isolated demes, adaptation by what may be called the "multiple-climber among which continual extinctions and replacements occur. effect". Adaptive evolution is like a climb up a ladder of This process is conspicuous among species associated with successive favorable adaptations. If each species climbs the temporary ponds and among crop pests that continually ladder alone (Fig. 4A), each will ascend relatively slowly, repopulate isolated fields, but the process probably goes on having to wait at each step for the appearance of needed among many other plants and animals: e.g., continual re- mutations and recombinations in the single population. But dispersal is apparently essential to many carabid beetles, if (say) four species climb the same ladder as an (involuntary) among many of which wings are retained apparently only team, with continual extinctions, reradiations, and group because they enable populations to exist on shifting ecologic selection (Fig. 4C), the climb will be more rapid, since at each checkerboards (11). In short, group replacements (which step the team will have four times as good a chance to originate imply group selection) can be observed in some cases, and needed innovations, and the first member of the team to make some situations suggest that they are inherent in the mode of a significant step can then radiate and start four new climbers existence of many plants and animals. Lewontin (12) reviews at the new level. The advantages of this process should be actual examples of interdeme selection, but concludes that it cumulative: the more species in the team, the more rapid the is less important than individual selection. This is surely true, climb; and the more steps, the greater the statistical advan- but group selection is not an alternative to, but a supplement tage of the larger team. to, individual selection. The patterns connected Consequences of group selection Group selection, as here simply described, varying in rate and One possible consequence of group selection is the evolution of effectiveness with species numbers, offers a credible, functional altruistic and other group phenomena. connection between the world-wide patterns of species Another possible consequence is that group selection may numbers and of dispersal, and may supply the "force" needed increase the rate and effectiveness of adaptive evolution. to give direction to the dispersal pattern (see ref. 14 for Isolated small populations may evolve rapidly, but their further discussion). Downloaded by guest on September 30, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Patterns of Biogeography and Evolution 1257

The geographic pattern formed by group selection is on the earth's surface before the origin of life, selection favored diagrammed in Fig. 5, and combines a dispersal component stability of individual molecules. But when some molecules (represented by horizontal lines) and an evolution component began to reproduce themselves, the selective advantage (represented by vertical replacements) that vary in different shifted from individual stability to reproductive capability. places. Within a main theater of evolution (A), where group Groups of self-reproducing molecules survived longer than selection is occurring among many species and contributing to individual molecules did. And evolution by selection occurred rapid and effective species evolution, the dispersal component both within and among the groups. Complex molecules may consist only of local movements made by species evolved more rapidly and effectively in large groups than in spreading into adjacent areas vacated by other species. But small ones-the multiple climber effect-and large groups at the margin of the pattern (C), on islands too small to main- were therefore selected as wholes, so that evolution did make tain independent populations for long periods, the dispersal or improve situations in its own favor. Group selection may component alone may be present, and no appreciable evolu- also have reduced the "life spans" of individual molecules; too tion of species may be occurring (although the island fauna as much molecular stability would have impeded the evolution a whole, as a group of interacting species, may evolve as some of new kinds of molecules, and groups in which individual species drop out and others replace them). At secondary molecules were too stable would have been selected against. evolution centers (B), of which New Guinea is one so far as If so, this was the beginning of what is called "altruism" at Carabidae are concerned, the dispersal component is strong, higher levels of organic complexity. but there is an evolution component too, revealed by the exis- Similarly if, of two competing species that are at first tence of locally evolved endemic species and genera. equally fit, one is panmictic and the other is divided into many Extensions and corollaries occasionally interbreeding demes, the second species should evolve more rapidly and effectively, and should eventually Up to this point, I have been concerned with patterns actually replace the first one. Evolution should thus improve the observed in the course of taxonomic and biogeographic work, situation in its own favor, by group selection at the species and with inferences made from the observed patterns. The level, and in doing so should favor any characteristics of theoretical results, however, can be extended and have individuals or of populations that divide populations into corollaries. separate, but occasionally interbreeding, demes. Character- Fig. 5A can be extended to represent directional change at istics that might contribute to this result include reproductive any level, down to (and below) the level of simple chemical and dispersal rates sufficient to ensure some, but not too reactions, and up to the level of whole faunas (and of solar much, exchange of individuals among demes, and also extreme systems and universes). At each level, movement results in niche specialization, which should favor existence in widely formation of diverse groups of units (that are themselves interrupted, thinly occupied ecologic checkerboards. This groups of smaller units) and differential survival occurs, i.e., situation should in turn favor existence of large numbers of some groups survive longer than others, although none is species occupying slightly different niches in the same areas, immortal. This can be expressed as a pseudo-mathematical and this in turn should favor group selection and the multiple equation: climber effect among many species. This description, based on Mo ->GF+DS= DC, group-selection theory, seems to fit situations existing in the Mo being movement, GF group formation, DS differential tropical rain forests. survival, and DC directional change. At the level of chemical CONCLUSIONS change, movements of atoms and molecules form diverse combinations, some of which survive longer than others. At Group selection, proceeding most rapidly and effectively the most obvious level of organic evolution, movements of where species are most numerous, provides a credible, func- genes in populations form diverse combinations (groups of tional connection between world-wide patterns of dispersal genes, which correspond to "individual" plants and animals), and of species numbers, and can provide the "force" that, some of which survive longer than others. At the chemical gives direction to the dispersal pattern. Group selection, level, surviving groups are "more stable"; at the level of reduced to simplest terms, is a process of movement, group organic evolution, they are "fitter" and are "selected"; but formation, differential survival, and directional change that. the essential process is the same and has presumably been occurs at all levels of inorganic and organic complexity; continuous from chemical evolution on the earth's surface, organic evolution is the sum of group-selection processes at through the origin of life, and into organic evolution. In fact, many interacting levels. A corollary is that evolution makes organic evolution is the sum of group-selection processes pro- situations favorable to itself, by group selection, and that it ceeding and interacting at many levels. The essential simplicity does so at all levels, from groups of prebiotic molecules to and uniformity of group selection is masked by constraints complex groups of species. New information about evolution imposed on dispersal at different levels, especially by the con- at the highest levels should be looked for in places of maximum straints on dispersal of genes in populations imposed by specie numbers and diversity, especially (on land) in un- self-replication of living material, but relaxed by bisexual disturbed tropical rain forest, where evolution probably has reproduction. in fact made situations especially favorable to itself. The principal corollary of this hypothesis is that evolution I thank Prof. Edward 0. Wilson for reading this manuscript should make situations favorable to itself, by means of group and for comments and references that have substantially im- selection: group selection at any level will favor groups in proved the paper. which evolution is occurring rapidly and effectively at lower 1. Darlington, P. J., Jr., Bull. Mus. Comp. Zool., 107, 87-252 levels. For example (I shall describe this case as if it happened, (1952); 126, 321-564 (1962); 137, 1-253 (1968). although it is hypothetical), when molecules began to evolve 2. Darlington, P. J., Jr., Bull. Mus. Comp. Zool, in press. Downloaded by guest on September 30, 2021 1258 Zoology: Darlington Proc. Nat. Acad. Sci. USA 68 (1971)

3. Darlington, P. J., Jr., Zoogeography (John Wiley and Sons, 8. Wynne-Edwards, V. C., Animal Dispersion in Relation to New York, 1957), Introduction. Social Behaviour (Oliver and Boyd, London, 1962). 4. Darlington, P. J., Jr., Quart. Rev. Biol., 23, 105-123 (1948); 9. Williams, G. C., Adaptation and Natural Selection (Princeton Zoogeography (John Wiley and Sons, New York, 1957), Univ. Press, 1966). chap. 9; Evolution, 13, 488-510 (1959). 5. Wilson, E. O., Evolution, 13, 122-144 (1959); Amer. Natur., 10. Darwin, C., On the Origin of Species (1859; facsimile of 1st 95, 169-193 (1961); in The Genetics of Colonizing Species, ed. ed., Harvard Univ. Press, 1964), p. 109. H. G. Baker and G. L. Stebbins (Academic Press, New 11. Darlington, P. J., Jr., Ecol. Monogr., 13, 37-61 (1943). York, 1965), pp. 7-24. 12. Lewontin, R. C., in Annu. Rev. Ecology Systematics, ed. 6. Mayr, E., Science, 150, 1587-1588 (1965); MacArthur, R. R. F. Johnston (Annual Reviews, Palo Alto, 1970), Vol. 1, H., and E. 0. Wilson, The Theory of Island Biogeography pp. 1-18. (Princeton Univ. Press, 1967). 13. Wright, S., Proc. Amer. Phil. Soc., 93, 471-478 (1949); 7. Darlington, P. J., Jr., Biogeography of the Southern End of numerous other refs. the World (Harvard Univ. Press, 1965), chap. 2, 5. 14. Darlington, P. J., Jr., Evolution, 13, 488-510 (1959). Downloaded by guest on September 30, 2021