3 GRADES IN THE EVOLUTION AND CLASSIFICATION OF INSECTS [Presidential Address to the Australian Entomological Society, Melbourne, January 15, 19671 By I. M. MACKERRAS* Abstract Grades of organization, of the kind recognized in vertebrates by de Beer and J. S. Huxley, are a conspicuous feature of the insects also. Four can be distinguished by well-defined anagenetic gaps: aptery ote; palaeopterous and exopte gote; neopterous and exoptery- gote; neopterous and enffopterygote.As grades can be dezed more clearly than ph logeny at higher taxonomic levels and the reverse is true at lower levels, it is suggestelthat a classification can be made to reflect both without confusing the user. Insects have been remarkably successful animals. They probably differentiated from other lowly terrestrial arthropods about 350 to 400 million years ago, as small, soft-bodied hexapods that lived in moist, sheltered situations and crawled or ran among the primitive vegetation that was then be ning to develop. Some of them began to fly about 300 million ears ago. Therear ter they radiated widely, evolving aripassu with the developing rand flora, until today about three-fourths of all the own species of animals are insects, while the numbers of individuals defy com- Lputation. Their success in exploiting a wide range of environments, from high latitudes to the equator, from rain forest to desert, from mountains to shore, even including a limited invasion of the sea, and in adapting themselves to varied ways of living-saprophagous, phytophagous, carnivorous, parasitic, social-has been one of the outstanding events in the whole course of animal evolution. It can be brought into focus sharply enough for our purpose by recalling that, whereas the Australian mammalogist has 22 families and about 250 native species to think about, the Australian entomologist must cope with some 700 families and 54,000 known species, including nearly 4,000 species of the family Curculionidae alone. This superabundance of material has naturally brought many problems in its train, not the least of which is that it becomes diflticult to see the wood for the trees. It may be appropriate, therefore, that the first presidential address to this Society should be an attempt, however imperfect, to look at the evolutionary and taxo- nomic picture as a whole. It is, in part, a sequel to more general examination of taxonomic principles (Mackerras 1964), into which I was led by my first contact with numerical taxonomy.

SCE~CMETHOD AND TAXONOMY As this discussion is concerned with taxa and their evolutionary relationship to one another, we must be clear at the outset about the taxonomic principles upon which the classification we use is founded. In the first place, it is essential to remember that taxonomy can be regarded as a scientific discipline in its own right only if those who practise it conform to the basic rules that are held, by common consent, to govern any kind of scientific en- deavour: (l), that the scientist should describe and interpret nature objectively (i.e., without bias) as it is, and not merely chop it up into tidy little watertight com- partments to suit his convenience-though he may do so as a preliminary to closer examination; and (2), that the collection and sorting of facts (“descriptive science”) is not an end in itself, but a stepping stone to theoretical knowledge. It might, indeed, be salutary, in the li t of some retrogressive modern trends, to define facts (information)as the materia of science, science as the use that is made of the facts to construct and test hypotheses.?kh Second1 , the basic premise of evolutionary animal taxonomy is that animals have evolved by descent with modification, diversity being the result of speciation, and the purpose of the classification is to express that relationship. In its classical form, every taxon is a statement of the hypothesis that its members are more nearly related to each other by propinquity of descent than to members of any other taxon, so the classification is called “phylogenetic” or “Darwinian”. As every episode of *Division of Entomology, C.S.I.R.O.,Canberra, A.C.T. ?It follows that empirical or Adansonian taxonomy, as expressed in a “phenetic” or “numerical” classification, is not a scientific discipline. It is purely a system of filing information, so it need not concern us here. J. Ausr. enr. SOC.,1967, 6:3-1 I. 4 I. M. MACKERRAS speciation is a point of origin of diversity of unpredictable extent, monophyly may be defined, strictly, as the evolution of a taxon from a single ancestral species. It follows, too, that the relative times at which splitting occurred are the fundamental criteria in constructing phylogenetic hierarchies (Hennig’s principle of “sister groups”-see his 1965 review). Thirdly, there are two com onents in diversity: the multiplication of phyletic lines, and the development of pf: enetic divergence between them. The two are not necessarily co-ordinate, and a strictly phylogenetic classification is concerned only with the first. Much confusion has resulted from failure to appreciate this principle. Finally, it must be emphasized a ain that taxa are statements of hypothesis. Like all scientific hypotheses, they are %ased on a critical analysis and judgement of the evidence available at the time, and they are tested by analysis of new evidence as it becomes available. Again like other hypotheses, they are modified when necessary, and abandoned when found untenable. The evidence that is used, both for formulating and testing, is primarily anatomical (including ultrastructure), and it can be analysed in a variety of wa s. No modern taxonomist would depend on one set of structures or one method or analysis when he can use more. The essential point is that the classification is dynamic: it unfolds a little more with every increase in our knowledge, and becomes progressively more illuminating as time goes on. No one can ex ect that finality will ever be reached (Mackerras 1964). With this gackground we may pass to a consideration of the complications that arise from the existence of grades of organization in animals.

THE NATUREOF GRADES Julian Huxley (1957, 1958) has pointed out that there are three main processes of evolution, “leading respectively to biological improvement, to diversification, and to persistence”. He adopted the terms anagenesis, cladogenesis, and stasigenesis for these processes, and he designated the steps in anagenetic advance as grades and monophyletic lines as clades. In this way he crystallized ideas which had been developed by Darwin and T. H. Huxley a century earlier. Phylogenetic classifications are composed of clades, and have been considered in the preceding section. We need not concern ourselves with persistence, except to note that it is treated as positive when the taxon survives, negative when it declines. Failure to advance, however, does not necessarily mean extinction. Unspecialized animals may survive, either because they can evade competition, or because they are very competent at something (Manton 1958), and it is a common observation that animals that are extremely primitive in most respects also have compensating specializations in a few. That biolo ‘cal improvement is the consequence of natural selection is the core of Darwinian #eory, and Huxley has defined it as including “detailed ada tation to a restricted niche, specialization for a particular way of life, increased eli ciency of a given structure or function, greater differentiation of functions, improvement of structural and physiological plan, and higher general organization”. He noted, too, that Darwin had recognized that diversification is a special form of general biological improvement. Clearly, ana enesis OCCUTS at all taxonomic levels and may be of all degrees of magnitude. If t%at were always so, grades would be merely arbitrary divisions in a continuum, and would have little special si cance. But it is not always so. Observations have shown that particular structur modifications have, from time to time, been associated with the appearance ofP extensive new adaptive radiations. Events of this kind have long been known in the vertebrates, and we shall be examining their occurrence in the insects. They are clearly reco nizable, at least at higher taxonomic levels, and it is to them that the term gra%- e may be usefully restricted. Thus, though anagenesis is continuous, grades are dis- continuous. De Beer (1954) showed that the transition from one grade to the next took place by what he called mosaic evolution. Archaeopteryx, for example, had a mixture of characters, primitive ones shared by reptiles and birds, some reptilian characters that are lost in birds, and some truly avian characters that reptiles do not possess. Also, as perhaps we might expect with mosaic evolution, the transition often seems to depend, not on a change in any one structure or function, but on what amounts

6 I. M. MACKERRAS and agility would have come into operation, leading to the development of longer, stronger legs carried on a larger, more powerful thorax. The mechanical benefits of the hexapod gait could be fully realized; functional division of the trunk into a thoracic tagma specialized for locomotion and an abdomen specialized for tro hic and reproductive functions would be facilitated; and raising the anterior end of! the substrate would have facilitated evolution of the varied types of hypognathous mouth-parts that occur in the insects. Even more important, from the point of view of subsequent evolution, were the provision of a point of balance in the anterior third of the body and the results of selection pressures for perception of, and reaction to, more distant as well as nearby objects in the environment. These two features, hexapody and emancipation, thus combined to provide the foundation on which the rest of insect evolution was built, so the Apterygota may be regarded as the base grade of the class. The problem becomes more complicated when we consider evolution within the Apterygota. The group is homogeneous as a whole, and there is little doubt that it was monophyletic; but it became divided very early (though the geological record does not yet show it) into a more primitive division with monocondylar mandibles which includes the Archaeo atha and the Palaeozoic Monura (Sharov 1966), and a generally more specialize group with dicondylar mandibles which includes the Thysanura. All Pterygota (exceptr possibly some Ephemeroptera) have dicondylar mandibles, and are usually presumed to have evolved from the thysanuran stem. This hypothesis is supported by some additional anatomical evidence, but it cannot yet be taken as fully established. Nevertheless, Hennig (1962) accepts it. He con- sequently rejects the Apterygota as a sister group of the Pterygota, makes the primary subdivision of the insects (“Ectognatha”) into Archaeognatha and Dicon- dylia, and divides the Dicondylia into Zygentoma (= Thysanura) and Pterygota. There is an evident clash in the kind of evolutionary information presented in the two classifications and we will have to consider which is to be preferred. The Evolution of Flight The development of wings was the outstanding event in the evolution of the insects, because it provided a means to escape from enemies, facilitated dispersal, and so opened up new environmental niches for the pterygotes to exploit. Some- thing like the aptery ote sha e, with its muscular, relatively rigid thoracic tagma and potentially aero(BE ynamic alance, was a prerequisite for true flight, and to that extent the apterygotes may be said to have been re-adapted. But they were tied to the ground, not only by lacking wings, but by lac&I * g spiracular regulatory mechan- isms to control repiratory water loss and in being dependent on indirect transfer of sperm for fertilization. Both these deficiencies (and some others) had to be over- come before full advantage could be taken of an ability to fly. As with the birds (de Beer 1954), mosaic evolution was involved, and it was successfully achieved even in the most primitive of the surviving pterygotes. It is widely held that wings developed from paranotal lobes on the thoracic terga, those on the prothorax taking little part in the process for aerodynamic reasons. We are not concerned here with the selection pressures that transformed paranotal processes into wings (see Wigglesworth et al. 1963; Flower 1964); but we are very much concerned with whether flight may have evolved more than once and with discerning the earliest type of wing articulation. These are difficult problems, and it may simplify the discussion if we begin with the preliminary hypotheses that wings evolved rn only one ancestral stock and that the palaeopterous articulation was the primitive one. The strongest evidence for the monophyletic origin of wings is that the venational fields and the general arrangement of veins within them can be homolo- gized throu out the Pte gota. This proposition can be tested in two ways. Firstly, is it true? Ti?ere is certain7 y no inherent objection to it in the venational plans of all the surviving orders, except one: the Odonata are undoubtedly odd. Not only has the tracheation gone offthe rails, which may not matter very much; but venational with other orders can be established only by deleting the whole of the homoloFMP an CuA sections of the venation, as supported by Tillyard (1926), or b assuming that the Odonata were derived from an extremely ancient stock in whici GRADES IN THE EVOLUTION AND CLASSIFICATION OF INSECTS 7 the cubito-anal section was still rudimentary (Riek, personal communication). Of the Palaeozoic orders now extinct, most conformed to the general plan, the Megani- soptera (= Protodonata) and possibly Megasecoptera to the odonate plan. It is necessary to stress here that parallel or convergent evolution is remarkably frequent in the Arthropoda, from the most ancient divisions of the phylum (Tiegs and Manton 1958) to the most specialized orders of insects (e.g. Hinton 1955, and many other papers), so it should not be forgotten that functional requirements demand a corrugated arrangement of supporting girders and struts of varying strength that could not differ very reatly from the general venational plan of the insects. This is not a test of the Tt ypothesis: it simply enjoins extra caution in accepting similarity of structure as firm evidence of homology, and it applies with as much force to the dicondylar mandibles of the Thysanura and Pterygota as to the venational problems we are now considering. Secondly-and this is obviously important in view of the foregoing caution- what independent lines of evidence can be found that bear upon the problem? We may approach this question by looking at two orders, Ephemeroptera and Odonata, that were already clearly foreshadowed before the end of the Carboniferous. The Ephemeroptera are palaeopterous, with a fully developed venation; they have an appendix dorsalis, mate in the air, and the clasping organs of the male are ninth-segment coxites and styles (telopodites). Their origin can be traced with con- siderable confidence through part of the Palaeodictyoptera to apterygote ancestors not so very different from those that survive today. The question is whether those ancestors had monocondylar or dicondylar mandibles, and it is raised because the larvae of some recent mafles are said to have monocondylar mandibles (mouth- parts are atrophied in adults). Mr. Riek has dissected larvae of a few Australian groups, and their mandibles appeared to us to be dicondylar. However, the point is so important that it should be checked by examining a wider range of primitive types and making reconstructions from serial sections. The Odonata are also palaeopterous, but they have a reduced or incompletely developed venation, no appendix dorsalis, no ninth-segment appendages in the male, and have evolved a completely novel method of sperm transfer in flight. They shared a proximate common ancestry with the huge Meganisoptera, and both poups were highly specialized for active, predatory, aerial life, even m the Palaeozoic. The mandibles of Odonata are unequivocally dicondylar in both larvae and adults, and the structure of the ovipositor indicates that the apterygote stock from which they were ultimately derived had eighth- and ninth-segment appendages in the females ; but it is difiicult to imagine the selection pressures that would have produced such a remarkably divergent and complicated mating mechanism if the ancestral males had had as potential1 efficient claspers as the ancestors of the Ephemeroptera. There are thus j: ee niggling doubts raised about the ancestors of these two order+venation, mandibular articulation, and mechanism of mating-and these need to be resolved before we can accept the first hypothesis with complete con- fidence. Turning to the problem of the wing articulation, there are three possible solutions, and each has been supported by different workers. 1. The theory that the palaeopterous type was the most ancient was developed independently by Martynov and Crampton more than forty years ago, and there is a good deal of evidence to support it. In the first place, the Palaeoptera were highly diversified by the end of the Carboniferous, whereas the Neoptera (except for Blattodea) appear to have been limited in the Carboniferous but radiated greatly in the Permian. This suggests that the Palaeoptera were robabl the older. Secondly, the most primitive known Neoptera can be derived Pdrom p aeopterous ancestors on characters other than the wing articulation, but the reverse involves some rather serious evolutionary improbabilities. Thirdly, the transition from a fixed to a flapping aerofoil involves less modification of structure when it is made on a simple rectilinear basis than when it also involves an independent ability to retract the wing. One would expect evolution to occur in stages, rather than all at once. Fourthly, the relatively greater part played in fQht by the direct muscles in Palae- optera and lower Neoptera, as compared with lugher Neoptera, suggests that their primary action was to flap the wings and that their other functions are secondary. 8 I. M. MACKERRAS This is not to suggest that there was complete rigidity. Some degree of controlled antero-posterior movement and an ability to twist the win s articularly by raising or lowering the costal margin, were essential for controKyF o speed, balance, and manoeuvrability from the beginning; but these could be achieved with any articu- lation that was not too rigid, as they have been, most efficiently, in the Odonata. 2. The Blattodea are also extremely ancient, their way of life is intermediate between terrestrial and aerial, and they have been thought by some to be more or less direct descendants of the ancestral pterygotes, which would therefore have been neopterous (see Martynova 1961, Sharov 1966). Against this is their specialized wing venation, even in the oldest known fossils, the probably secondary nature of their terrestrial mating (the aerial mating of the Ephemeroptera, with male below, seems likely to have been the most primitive method of true copulation in the insects), and the associated reduction and modification of the clasping organs. The most likely hypothesis, on morphological grounds, is that the blattoid-orthopteroid complex was derived from alaeopterous palaeodictyopteran ancestors. That is to say, they are more likely to L aerial insects returning to the ground than terrestrial ones becoming ada ted to the air. 3. Sharov (192) 6) has rejected both the earlier hypotheses, and concluded that the original pterygotes (Archoptera) had swept-back wings like the wing-sheaths of some exopterygote larvae. He derives the Palaeoptera from one section of the group and the Neoptera from another. The group itself is based on only a few, very ancient, and not very convincing fossils, includin the Devonian Eopterum; and the theory is open to the objection that, although e cient fixed-wing flight (including gliding) can be obtained with swept-back wings, iiefficient flapping flight, at least in its simpler forms, requires that the long axis of the aerofoil should be nearly at right- angles to the air flow. There would have to be a mechanism from the beginning to rotate the wing forward as well as to fla it, and if that had happened it is difficult to see why palaeoptery should have evoP ved at all. The Palaeoptera, then, are a possibly diphyletic assemblage representing, as nearly as we can judge, the most ancient of the effectively flying insects. On the other hand, they do re resent quite clearly, especially when we look at their Palaeozoic radiation, a nota 1le anagenetic advance on the A terygota, and they have achieved this by at least a modicum of mosaic evolution. #hey may therefore be regarded as a grade, whether or not they ultimately prove also to be a clade. The Exopterygote Neoptera There are two main questions to consider here. First, are the Neoptera monophyletic? It is at least doubtful for two sets of reasons. On the one hand, the Diaphanopterodea, and possibly one or two other Palaeozoic groups, seem likely to have been independent neopterous developments from different Palaeozoic ancestors (Riek, rsonal communication). On the other hand, the surviving Neoptera may be divi ed into three great clades, the blattoid- ortho teroid orders, the hemipteroid orders,d" and the endopterygote orders, and it is dlffcult to bring these three clades back to a sin e point of common ancestry. The hemipteroids might possibly have been derivetl' from primitive blattoids, but the evidence from wing architecture and venation and from enital structure and aerial matin point rather strongly to an inde ndent origin of the endopterygotes. As always wt en dealing with ancient events, t e evidence is not conclusive, but one is left with the impression that selection pressuresr favouring neoptery must have been quite strong. Under the circumstances, it seems curious that Hennig accepts the Neoptera as a clade, though he rejects the Palaeoptera on evidence that is little, if any, stronger. Second, are we justii3ed in treating the exopterygote Neoptera as a grade above the Palaeoptera? The selective advantages of neoptery are evident. An insect that can fold its wings can move freely in confined spaces, and so can work better in vegetation or on the ground than one that cannot. Moreover, the folded wings provide protection for the soft sides of the body, and division of function between fore and hind wings-an important development-becomes possible. The relative success of the exopterygote Neoptera is evident from the table; and it may also be significant that the Palaeoptera declined as the Neoptera rose to prominence to- GRADES IN THE EVOLUTION AND CLASSIFICATION OF INSECTS 9 wards the end of the Palaeozoic, that the only ones to survive have been those that either confine their feeding to the larval stages (Ephemeroptera) or hawk for prey upon the wing (Odonata), and that the early stages of both these groups are very good at exploiting special (aquatic) niches. The anagenetic gap here is not as great as that between apterygotes and pterygotes or between exopterygote and endopterygote Neoptera next to be con- sidered, and it is not so clearly supported by evidence of mosaic evolution. Never- theless, it is quite well defined, it helps to clarify our understanding of the stages in evolution, and it is useful for those reasons to accept it as marking a grade.

The Endopterygote Neoptera There is general agreement that holometaboly, as expressed in endopterygote development, represents a marked biological improvement on hemimetaboly, and it is worth examining how this has come about. The only constant (or almost constant) difference between exopterygotes and endopterygotes is that the wing rudiments are free to develop rogressively on the surface of the body of exopterygote larvae, whereas they are wit tdrawn into pockets invaginated beneath the surface in endopterygote larvae. The advantages are of the same kind as neoptery in adults: the endopterygote larvae can begin to exploit confined habitats that are not so accessible to exopterygote larvae. Degrees of metamorphosis are not involved at this stage, and it is worth remembenng that dragonfly larvae differ as much from the adults in habitat and structure, and under- go nearly as much metamorphosis, as do those of Megaloptera. The foundation is laid, but the potential cannot be realized until something else has happened. This was the evolution of the pupal stage, and here I would follow Hinton (1963), whose theory involves fewer assumptions than the others that have been proposed. He points out that there is insacient room in the larval thorax for the wings to reach a size at which they can complete their growth after the final larval- adult apolysis (see below). A last larval exopterygote stage is necessary to bridge the gap, and it is this that becomes the pupa. Now this larval stage, being exoptery- gote, had lost the advantages of its earlier instars, and, lacking functional wings, it also lacked the advantages of the adult. It was suspended, like Mahornet’s coffin, between two ways of life, and so it seems inevitable that it would become secluded into a resting form in which the change from larval to adult organization could take place. Thus was provided the opportunity for that progressive dissociation between larval and adult structure and way of life that has reached such remarkable levels in the higher endopterygotes. To one who was brought up on vertebrate neurology, the most extraordinary of the many extraordinary events in metamorphosis is the way in which a nervous system adapted to larval needs is broken down and replaced by a nervous s stem adapted to adult needs. It is this sort of thing that provides the key to the bior ogical success of the endoptexy otes. But that is not the whole story. None of t%ese changes that so ’ustly excite our wonder could have occurred, if it had not been for the existence od pharate phases of development that could vary in duration according to ada tive needs. A digres- sion is necessary to explain what is meant by this, because, a! though the facts and their interpretation have been known for more than thirty years (Snodgrass 1935, p. 64), a surprising number of people still seem to think that the change from one stage of growth to the next (“transformation”) takes place suddenly. It is not like that at all. Arthropod epicuticle is inextensible, and arthropods grow in stages that are separated from one another by renewal of the cuticle. The stages are facts of nature, there to be discerned, not merely arbitrary divisions in a continuous process. A stage of growth is completed when the body of the arthropod ills the space that is available within its cuticle. Preparatory changes begin in the epidems, the cells retract from the old cuticle, undergo more or less extensive reorganization that determines the form of the new instar, and a new cuticle is laid down on their external borders. The freeing of the cells from their previous restraint at apolysis, as the retraction has been termed (Jenkin and Hinton 1966), clearly marks the end of one stage of growth and morphogenesis and the beginning of the next. 10 I. M. MACKERRAS During the first, more vulnerable part of its develo ment, the new instar re- mains enclosed in, and protected by, the old cuticle, to wfL ‘ch it remains connected in a variety of ways, and throu which it can still perform (see Hinton 1958). This is the p9 urute, or cloaked, phase of and it ends at ecdysis when the arthropod is ready to face the times it never is ready, and the whole of the stage is pharate, as with the pupae of cyclorrhaphous flies and some other insects. The crucial importance of the pharate phase, from our present point of view, is that change takes time, and it provides the time. When the chan e is slight, the pharate phase is usually short; when the change is great, as between arva and pupa and between pupa and adult in the hi er endopterygotes, the pharate7 phase is prolon d by as much as is needed. Wi ttout this adaptwe flexibility in its duration the en ropterygotes could not have evolved in the way that th We thus have in the endopterygotes a beautiful example of x ehave* mosaic evolution to which de Beer drew attention m the birds and mammals. But there the com- parison ends. If the suspicion that the endopterygotes evolved from Palaeodicty- optera separately from the exopterygote Neoptera proves to be correct, then they took two major anagenetic steps so nearly together in time that they cannot be separated on present knowledge. It would simplify the story considerably if we could point with confidence to an exopterygote neopterous ancestor of the endo- pterygotes. However, we can still a-t the endopterygotes as representing both a clade and a grade, and we can also reject the Hemimetabola and Holometabola as formin a satisfactory primary taxonomic division of the winged insects. This, at least, cf ears away some of the obscurity. CONCLUSION Anagenesis and cladogenesis have combined to produce the vast array of s that we have to classify, and a complete evolutionary classification animalshould 3Yerefore reflect both. But we are limited by ractical considerations to a two-dimensional classification, and no consistent two-&mensional arrangement can reflect more than one of them.. We do not want multiple general classifications of the animal kingdom, if we can do without them, so the essential question is whether it is really necessary to be consistent at all levels. We should not avoid this question, as Simpson (1961) has done, very neatly and effectively, simply by defining mono- phyly as “derivation of a taxon . . . from one immediately ancestral taxon of the same or lower rank,” although this is often as far as it is possible to go in practice. Nor should we ignore it, as Hennig (1965, and earlier papers) has done, when the resulting structure is so rigid that arts of it conceal more than they reveal, a criticism that may justly be made oP the tabular statement in his 1962 aper. It is our duty to descnbe and interpret nature as it is, and I can see no logicBp reason for insisting on consisten (or any other abstract rinciple) if it obscures an aspect of nature that we are en7 eavouring to understandl There are three sets of circumstances that help to simplify the problem. The most signiiicant is that, the further one has to go back in time to find the ori a major group of animals, the clearer do grades become and the more of? scure Of becomes the phylogeny. The converse is equally true. In the relatively younger, lower ranking taxa, phylogeny can often be deduced with considerable confidence, whereas anagenesis can be seen but dimly. The second consideration is that this situation has, in fact, been incorporated, consciously or not, in widely accepted classifications, which at least some evolutionary taxonomists seem to find more illuminating and intellectually satisfyin than a rigid1 phylogenetic classification. Conservatism is no more a criterion oB scientific vali&ty than is convenience, but the opinions of serious students of evolution should at least be examined. The third consideration is that, although taxonomy is a scientific discipline, the classification in which its results are ex ressed is a tool, and it is im ortant to know for what purposes the tool is to be usecf There are four main uses o a classification: as a summary of the organic composition of the taxa and the evolutionaryP relation-

*This is true of phyletic diversity and phenetic divergence in general ( . 4), but that is a more difficult probltm, and the arguments that follow do not necessarily appg to it. GRADES IN THE EVOLUTION AND CLASSIFICATION OF INSECTS 11 ships between them; as a guide to the closeness of relationship between taxa that are being compared by morphologists, physiologists, ecologists, ethologists, and others; as a basis for evolutronary zoogeography; and as a filing system for in- formation. A mixed classification serves all these needs, because the general student of evolution, who cannot neglect grades, is usually concerned with higher taxa, zoogeographers and others who depend on hylogenetic relationships are pre- dominantly concerned with levels at which t lose relationships are most clearly expressed, and the nature of the system does not matter for information storage so long as it is mechanically efficient. If, then, there may be virtue in a mixed classification, the question arises: at what level should the influence of anagenesis fade and cladogenesis become strictly reflected? It seems impossible to be precise, because the level is like1 to vary from group to group, and some ancient groups (such as nematodes, per r;aps) may not show evident anagenetic discontinuities at all. I do not think that it matters, and I would also a ree with Simpson that it is not necessary to incorporate a terminology to distinguisi between anagenetic and more strictly phylogenetic taxa. To anyone who wishes to make scientifk use of a classification the framework can serve only as aide-mkmoire. He must know the background in considerable detail as well, and therein will he find where the change of emphasis occurs.

REFERENCEs DE BEER,G. R. (1954).-Archneopteryx and evolution. Advmf Sci., Lond. 11: 160-170. FLOWER,J. W. (1964).-On the origin of fli t in insects. J. Insect. Physiof. 10: 81-88. HENNIG,W. (1962).-Ver&nderungen am pt ylogenetischen System der Insekten seit 1953. Tagungs- berichte No. 53, pp. 29-42. Bericht iiber die 9 Wanderversammlung Deutscher Entomologen, 6-8 Juni. 1961, in Berlin. (Deutsche Akademie der Landwirtschaftwissenschaften zu Berlin.) HENNIG,W. (1965).-Phylogenetic systematics. A. Rev. Enr. 10: 97-116. HINTON,H. E. (1946).-Concealed phases in the metamorphosis of insects. Nature, Lond. 157: 552-553. HINTON,H. E. (1955).4n the structure, function, and distribution of the prolegs of the Panorpoidea, with a criticism of the Berlese-Imms theory. Trans. R. ent. SOC.Lond. 106: 455-545. HINTON,H. E. (1958).--Concealed phases in the metamorphosis of insects. Sci. Prog.. Lond. 46: 260-275. HINTON,H. E. (1963).-The origin and function of the pupal stage. Proc. R. enf. SOC.Land. (A), 38: 77-85. HUXLEY,J. S. (1957).-The three types of evolutionary process. Nature, bnd. 180: 454-455. HUXLEY,J. S. (195 ).-Evolutionary processes and taxonomy with special reference to grades. Uppsalu Univ. A8,,,r., 1958: 21-38. JENKIN,P. M., and HINTON,H. E. (1966).-Apolysis in arthropod moulting cycles. Nature, Land. 211: 871. MACKBRRAS,I. M. (1964).-The classification of animals. Proc. Linn. SOC.N.S.W. 88: 324-335. MANTON,S. M. (1958).-Habits of life and evolution of body design in Arthropoda. J. Linn. SOC.(Zool.) 44 (Bot.) 56: 58-72. MANTON,S. M. (1964).-Mandibular mechanisms and the evolution of arthropods. Phil. Trans. R. SOC. (B) 247: 1-183. MARTYNOVA,0. (1961).-Palaeoentomology. A. Rev. Enr., 6: 285-294. SHAROV,A. G.(1966).-"Basic Arthropodan Stock with Special Reference to Insects,'' 271 pp. (Pergamon Press: Oxford). SIMPSON,G. G. (1961).-"Principles of Animal Taxonomy," 247 pp. (Columbia University Press: New York). SW~M, R. E. (1935).-"Principles of Insect Mo hology," 667 pp. (McGraw-Hill: New York). Tnwa, 0. W.,and MANTON,S. M. (1958).-The evxution of the Arthropoda. Biol. Rev. 33: 255-337. ~YW,R. J. (1926).-"The Insects of Australia and New Zealand," 560 pp. (Angus & Robertson: Sydney). WWOL[EFWORTR,V. B.,ef al. (1963).-Discussion of the origin of flight in insects. Proc. R. enr. SOC.Lond. (c]28: 23-32.