By Sicien Chen

National Institute of Zoology, Academia Sinica, China

Transactions of the Royal Entomological Society of London 1946. 97:381-404


Notwithstanding the general interest that has been shown in the study of insect transformation, our knowledge of the evolution of larval insects still remains obscure. For the old authorities have followed Brauer (1869) in regarding the diversity of larval forms as being the result of secondary and adaptive modifications of a primitive type; and the modern students, by accepting the theory of Berlese (1913), have recognised that the so-called larvae are but precociously emerged embryos, and that their diversity of types is primarily a consequence of variations in time of emergence. On the one hand, the conception of the adaptive nature of larval insects abases their significance in phylogeny and makes the tracing of their whole evolutionary history unreliable. On the other hand, the ontogenetic sequence of larval types, as illustrated by Berlese, is not conformable to the evolutionary sequence, for the protopod larva is in no way a primitive type. Hence the theories of both Brauer and Berlese, while explaining the significance of the diversity of larval forms in insects, leave the evolutionary side of the problem unsolved.

It is the object of the present paper to deal with this side of the problem of insect transformation that has been so long neglected. The subject will be discussed under its two principal aspects, that is: (1) the larval sequence and (2) the larval evolution properly speaking. By larval sequence we mean the ordered succession of larval types. It is well known that similar types of larvae are often found in very different groups of insects, and that these types usually follow a similar sequence or parallel modifications in the most widely separated orders. But the sequence of larval types is not necessarily phylogenetic, whereas in tracing the larval evolution properly speaking, phylogeny is always involved. In the first case, interest lies in the consideration of the steps by which the various larval types have acquired their characteristic features in the course of evolution from the most primitive to the most recent. In the second case, it is the origin of insect larvae, of the various orders of Holometabola and their phylogenetic relationships that are to be considered. But in addition it seems desirable to give, first some remarks on the monophyletic origin of the Holometabola, and secondly some account of the different types of nymphs and their evolutionary sequence.


In dealing with the problem of larval evolution, the question naturally arises as to whether insect larvae are monophyletic or polyphyletic in origin. In other words, whether the Holometabola have originated from a common ancestor, or whether they have been evolved from several ancestral stocks, arising independently from the Hemimetabola. This old question of insect phylogeny is still unsettled. The monophyletic conception was held by Brauer, Packard (1898) and their followers, since they always claimed the derivation of all existing larval forms from a primitive type. This was chiefly questioned by the palaeontologists, notably Handlirsch (1908), who had traced the holometabolous insects back to several origins in the Hemimetabola. But as we shall see, the evidence obtained from the study of fossil wings is not quite conclusive, and it would be equally possible to deduce the alternative view from the same data. On the other hand, the conception of monophyletic origin of Holometabola, so far as the study of larval forms is concerned, may be justified by the following facts: (1) the development of the wings and other organs of the imago from ingrowing buds is a feature common to all orders of Holometabola. This similarity in the fundamental process of growth indicates similarity of origin. (2) The various types of insect larvae may be successively traced back to a primitive type. (3) The tracing of larval evolution in this sense may help us to explain the phylogenetic relations of the holometabolic orders, and this fact, in its turn, helps to demonstrate the possibility of the single origin of insect larvae. While the first point is clear and needs no remark, the remainder are to be elucidated in the following sections--indeed, the present communication is an exposition of these facts. For the time being, it is only necessary to insist on this monophyletic conception of larval origin in order that it may serve as a basis for further discussions.


Insect larvae are usually grouped into types either according to their general form (Brauer, Packard), or depending upon their ontogenetic sequence (Berlese). For the purpose of the present study, an attempt is made to characterise the types in a way conformable to their evolutionary sequence, that is, each type is taken to represent a definite advance that has been made in the course of evolution from the most primitive to the most recent; consequently, all the types of side lines are not dealt with.

Following the monophyletic conception of larval origin, the first requisite of such a classification is to distinguish a primitive type versus those derived from it. A comparative study of insect larvae has shown that, among the various groups, the more generalised types usually have better developed appendages, and, as the types become progressively specialised, the appendages are invariably more and more reduced. It seems therefore that the primitive type of larvae would be a type in which all the typical appendages are exhibited in a well-developed condition and from which the other types have been derived through successive reductions. While the majority of larval types appear to follow this rule, there are, however, some other types the origin of which is undoubtedly different. These latter types are found exclusively among the primary larvae of endoparasitic Hymenoptera; judging from their state of organisation, they are actually in very early stages of morphological development and cannot be considered to have been specialised from other more generalised types. On the other hand, they cannot be regarded as primitive forms, since the parasitic Hymenoptera are by no means primitive Holometabola. The only plausible account of their origin is, therefore, to regard them as representing early stages of ontogenetic development which are ordinarily passed within the egg but which have been, in these instances, brought into the larval state through precocious birth. Hence we may propose to arrange the larval types under three categories, namely, the primitive or genera lised type, the specialised types, and the premature types. The primitive type is the type from which all the other types have originated; the specialised types are those which have been derived through modification of pre-existing types, principally by reduction of appendages; and the premature types are those which have owed their existence through precocious emergence from the egg.


    The true primitive larva will probably never be discovered, but a living representative is found in the larva of the neuropterous genus Corydalis Latreille. This larva, being primitive by its systematic position, exhibits at the same time many generalised features from which the specialised types may be successively traced. It represents, in the sense of Berlese, a typical polypod type; but in the sense of Brauer, a true campodeiform larva, for it resembles Campodea Westwood not only in general form but also in structure by being a polypod. We shall designate this primitive type by the term campodeoid-polypod.

    Primitive type of larva: campodeoid-polypod larva of Corydalis.

    The larva (above) is of large size, massive and robust at its final stage, with a well-developed head, three thoracic and ten distinct abdominal segments. Ten pairs of spiracles are present, including two thoracic and eight abdominal. The antennae are prominent, 4-jointed, and the ocelli are six in number. The mouthparts resemble those of a Carabid larva, the mandibles are powerful and more or less sharply toothed; the maxillae exhibit the typical parts, with the palpi 4-jointed, the galea palpiform, but the lacinia small, sometimes not well differentiated; the labium consists of a mentum, a broad bi-lobed ligula and 3-jointed palpi. The thoracic legs are well developed, with all the typical parts and paired claws. The abdomen is provided with nine pairs of appendages; each segment except the ninth carries a pair. The appendages consist each of a limb-base or coxite, bearing a large ventral tubercle and a lateral process. The ventral tubercle is called by Snodgrass the vesicle (vs), and the lateral process the stylus (sty). The vesicle is entirely absent on the eighth pair; it is provided with tufts of respiratory filaments on the first seven pairs, but on the last pair, instead of filaments, it is terminated by a pair of claws.

    Among the characters which distinguish the present larva as a primitive type, the most important are the abdominal appendages, hence their nature requires consideration. The researches of Heymons (1896) and Snodgrass (1931) have already established the fact that they are rudimentary limbs, but with regard to the homologv of their parts, no definite conclusions have yet been reached. Snodgrass considers the limb-base to consist of coxal and subcoxal regions of a typical appendage, the two parts being either distinct or united; the stylus as a rudimentary telopodite; the vesicles as endite lobes of coxae, which in some cases have become normally invaginated. Nevertheless, after admirably discussing the subject, Snodgrass prefers to regard these interpretations as tentative. But if we confine our attention to the mere comparison of living types, and study carefully his work on insect morphology, we can scarcely avoid the conclusion that the abdominal appendicular processes of many insects are respectively homologous structures, that is, between the vesicles of Apterygota, the plantar lobes of caterpillars and sawfly larvae, and the gill tubercles (or vesicles) of Corydalis on the one hand; and the styli of Thysanura, the gills of Ephemerid larvae, and the lateral processes (or styli) of Corydalis and Sialis Latreille on the other.

    The possession by the Corydalis larva of abdominal appendages which are homologous with the vesicles and styli of Thysanura affords definite evidence of its primitiveness. This larva may, therefore, be taken to represent the most generalised type of insect larvae from which the other types may have been derived through reductive specialisation. Of course the larva, as a living representative of primitive type, should be very different from the true primitive larva, for it has certainly been much modified in the course of evolution. It would be, however, fruitless at present to trace the actual characteristics of the true primitive larva. But with regard to the abdomen, we may believe, from the known facts of insect morphology, that in the ancestral larva, the typical segmentation, with twelve segments including the telson, might be distinct; the appendages, consisting of the typical parts, might be present on the segments 1-10; the vesicles were probably devoid of respiratory filaments, etc.


    The specialised types occur universally in all holometabolous orders. They are extremely diversified in form and structure; but the fact that they have all been derived from the primitive type through successive steps of reductive specialisation has much facilitated the task of their classification. Eight types, representing the principal steps in their specialization, are thus recognised and described below.

    1. The Sialoid-polypod type. The essential feature of this type is, theoretically, the possession of a pair of styli on each of the abdominal segments 1-10, but actually, those of the posterior segments are usually no longer retained. The type seems, therefore, to be derived from the primitive or campodeoid-polypod type by the suppression of vesicles, as such a process also occurs on the eighth abdominal segment in Corydalis. It is found only in certain aquatic larvae, of which the Neuropterons Sialis and Sisyra Burmeister, the Gyrinid Dineutes Aube, the Dytiscid Coptotomus Say, and probably the Hydrophilid Hydrophilus caraboides L. (vide Boving and Craighead, 1931), are examples. In Sialis, Sisyra and Hydrophilus caraboides, the styli are evidently present on the seven anterior segments, and in Coptotomus, on the six anterior segments; in both cases, each segment carries a pair. But in the Gyrinid, the full number of 10 pairs are present; they are distributed on the nine anterior segments, the two terminal pairs being both apparently carried on the ninth segment.

    2. The Eruciform-polypod type. Examples of this type are afforded by the larvae of the Lepidoptera, the sawflies, and the scorpionflies. Their essential features are the cylindrical body-shape and the presence of abdominal prolegs. The latter vary much in number, but are primarily of ten pairs, occurring on the segments 1-10. This number is actually found in the sawflies of the family Xyelidae (Yuasa, 1922), and in the embryonic stage of many Lepidoptera. According to Snodgrass (1931), the typical proleg of a caterpillar consists of three parts, namely, the limb-base or subcoxa, the coxa and the planta; he has brought forward evidence to show that the structure of the planta is almost identical to the vesicle of Corydalis larvae. It is therefore the styli of the campodeoid-polypod larvae which are not represented in this type.

    3. The Carabcoid-oligopod type. The present type is represented by what is ordinarily known as the campodeiform larvae. Among the typical examples are the larvae of most Neuroptera and of many Coleoptera. They resemble those of the campodeoid-polypod and sialoid-polypod types in the flattened-shaped body, the 5-segmented thoracic legs and other characteristics, but have assumed an oligopod condition by the loss of abdominal appendages.

    4. The Staphylinoid-oligopod type. The typical representatives of this type are the ordinary campodeiform larvae of the Coleoptera Polyphaga. They can be distinguished from the caraboid-oligopod larvae by their 4-segmented thoracic legs, the tarsal segment being not differentiated. The type is evidently derived from the preceding, since in certain genera of Staphylinidae (Philonthus Stephens, Bledius Samouelle), the tarsal segment has still preserved its separated condition, and in Aleochara bilineata, it becomes indistinct only from the second instar onwards. The absence of a distinct tarsal segment marks, therefore, an advanced stage in specialization.

    5. The Eruciform-oligopod type. The eruciform-oligopod type is represented by the larvae of several orders, comprising certain sawflies, many Trichoptera, the later stage larvae of the Staphylinid genus Aleochara Gravenhorst, one neuropterous family Mantispidae, and the ordinary eruciform larvae of Coleoptera Polyphaga. Although universally characterised by the cylindrical body and absence of abdominal appendages, they are very variable in other respects, and appear to have originated in several ways. In the sawflies, the type is most probably a derivative of the eruciform-polypod type through reduction of abdominal appendages, as the various phases of reduction, from the Xyelidae and Tenthredinidae to the Pamphilidae, Cephidaae and Siricidae are evident. In the Trichoptera, Mantispidae and Aleochara, it seems to have been derived from the caraboid-oligopod type by assuming an eruciform shape, for the latter type exists, either in their allied and more primitive families or as primary larvae. Finally, in the Coleoptera Polyphaga, the mode of derivation appears to be similar to that of Trichoptera or Neuroptera, but it is intercalated by a staphylinoid-oligopod type.

    6. The Eucephalous-apod type. Eucephalous-apod larvae are characterised by the complete suppression of both thoracic and abdominal appendages but without marked modification in head size and structure. The type occurs in Diptera, in Siphonaptera, in Hymenoptera Apocrita, in certain families of Coleoptera, in the final instars of Strepsiptera and in a few leafmining Lepidoptera. In the majority of cases, it appears to have been derived from the oligopod types, more usually the eruciform-oligopod type, through the suppression of thoracic legs. But in Lepidoptera, the apod type follows directly the polypod type without the intercalation of any oligopod forms.

    7. The Hemicephalous-apod type. In this type, the head is markedly reduced, usually incomplete posteriorly and becoming more or less retractile into the thorax. The type is characteristic of the Diptera Brachycera, some Nematocera and many parasitic Hymenoptera. It represents the climax of larval specialisation attained by the Hymenoptera, but among the Diptera it is an intermediate type, leading from the more primitive eucephalous type to the much specialised cephalous larvae of Cyclorrhapha.

    8. The Acephalous-apod type. This type is the most highly specialised of insect larvae. It is represented by the so-called acephalous larvae of the Diptera Cyclorrhapha in which the head is vestigial, becoming fully and deeply retractile or invaginated within the thorax, and the usual mouthparts are not developed. The type is evidently derived from the eucephalous type through many intermediate stages.


    The precocious types are confined to the parasitic Hymenoptera and appear to have been independently acquired by the different groups. Three types, referable to the three earlier embryonic phases of Berlese, are to be recognised, namely, the vermiform-polypod type, the polymerous-protopod type and the oligomerous-protopod type. They all occur in early instars, being invariably followed by a final instar of the usual apod type that is characteristic of all Apocrita. Consequently, they are of more recent origin than the specialised types. It may be said that the Hymenopterous larvae, after attaining their eucephalous or hemicephalous apod state of organisation, have not gone any farther in specialisation than that reached in the Diptera, but new types have been evolved as a result of a shortening in the period of embryonic growth. It is highly improbable that, in the course of evolution, the three types would have developed at the same time; but several steps of premature birth are necessary before a larva can be produced at such an early stage of development as that represented by the oligomerous-protopod type. It is therefore logical to assume that the polypod type is the one that has appeared first, the polymerous-protopod type follows next through further precocious birth, and finally the oligomerous-protopod type which is the one most lately acquired by the Hymenoptera. Besides these well-marked types, intermediate forms are liable to occur, but our knowledge of the primary larvae of parasitic Hymenoptera is so scanty and imperfect that the true nature of a great many is not yet understood, being masked by superimposed adaptive or secondary modifications.

    The cause of premature birth is obscure. While paucity of egg yolk is generally considered as a factor of great importance, it does not seem to be operative in the present instances, since the parasitic embryos are generally enveloped by a trophic membrane which outweighs the deficiency of egg yolk. It is therefore not impossible that in these insects precocity of eclosion may be a result of precocious rupture or dissociation of egg envelopes, brought about by the unusual expansion of egg size. Many of the parasitic eggs, after deposition in the host bodies, are found to increase markedly in size until eclosion finally supervenes. Thus in the Ichneumon Meteorus dimidiatus Cresson, the newly laid egg measures 0.14 mm. x 0.04 mm., but it finally attains a maximum size of 1.2 mm. x 1.5 mm. (Strickland, 1923); in Dinocampus rutilis Nees, another Braconid, Jackson (1928) states that the egg may increase in cubic content over 1200 times. The chorion of these parasitic eggs is usually very thin or even membranous, or may be absent, so that it is possible for the egg to absorb, through diffusion or through the action of the inner trophic membrane, the body fluid of the host and to increase enormously its bulk up to the final rupture of the envelopes. If this be the case, then, these larval forms would be, primarily, veritable embryos which, after having been brought into the larval state through the precocious rupture of egg envelopes, have survived by reason of their endoparasitic mode of life, and have subsequently acquired superimposed secondary features.

    1. The vermiform-polypod type. The vermiform-polypod type occurs in the Cynipoid families Ibaliidae (Chrystal, 1930), Eucoilidae and Figitidae (James, 1928) and in the Proctotrypid Phaenoserphus viator Haliday (Eastham, 1929). Theoretically, it corresponds to Berlese's polypod phase of embryonic development in which the abdomen has just acquired its complete segmentation and full number of appendages. But in actual condition, they are more or less modified, notably in the abdominal appendages which are often subjected to reduction. In Ibalia Latreille and Phaenoserphus Kieffer, the type is characteristic of the first instar, but in Eucoilidae and Figitidae, it occurs in the second instar, while the first observed instar is of the polymerous-protopod type. These two families have attained, therefore, a more advanced stage in their larval evolution.

    2. The Polymerous-protopod type. This type is represented by the so-called eucoiliform and cyclopoid larvae of Eucoilidae and Figitidae and probably also by the caudate and teleaform larvae, found in certain genera of Ichneumonoidea, Chalcidoidea and Proctotrypoidea. They are to be compared with Berlese's protopod polymerous stage of embryonic development in which the abdomen has completed its segmentation but does not bear appendages, while the thorax bears rudimentary limbs; but many of them are actually so much modified that their true nature can hardly be recognised.

    3. The Oligomerous-protopod type. In this type, the larvae emerge in a very early ontogenetic stage comparable with the protopod oligomerous phase of Berlese; the metamorphosis is still incomplete, the abdomen is not segmented, and the internal organs are not fully differentiated. It is characteristic of certain larvae of Platygasteridae, among which the primary larva of Platygaster herrickii Packard affords a typical example.


Having characterised the principal types of insect larvae, we may proceed to trace their whole evolutionary sequence; this is illustrated by the accompanying diagram. It is evident that the specialised types, originated from the campodeoid-polypod type, have followed two lines of development: the one, starting from the sialoid-polypod type, culminates in the acephalous-apod type of Diptera; the other, beginning with the eruciform-polypod type, approaches its climax in the hemicephalous-apod type of Hymenoptera which, instead of further specialisation, is followed by the premature types. The first line of sequence may be designated the aquatic line and the second the terrestrial line. The former comprises the larvae of Neuroptera, Coleoptera, Strepsiptera, Trichoptera, Diptera and Siphonaptera; the latter, those of Mecoptera, Hymenoptera and Lepidoptera. We have seen that all the known sialoid-polypod larvae are aquatic in habit and we have further reasons (to be discussed in later sections) to believe that the orders of this line are primarily aquatic in origin, while those of the second line have been evolved on land. To summarise what has been discussed in the preceding section, the steps of the two lines of evolution are as follows:--

Aquatic Line Terrestrial Line
Acephalous Apod - further reduction and invagination of head within thorax, atrophy of typical mouth-parts(Specialized) Oligomerous Protopod - eclosion still further advanced -(Premature)
Hemicephalous Apod - reduction and retraction of head within thorax - (Specialized) Polymerous Protopod - eclosion further advanced - (Premature)
Eucephalous Apod - suppression of thoracic legs - (Specialized) Vermiform Polypod - moment of eclosion advanced - (Premature)
Eruciform Oligopod - assumption of a cylindrical shape of body - (Specialized) Hemicephalous Apod - reduction and retraction of head within thorax - (Specialized)
Staphylinoid Oligopod -disappearance of a distinct tarsal segment on the thoracic legs - (Specialized) Eucephalous Apod - suppression of thoracic legs - (Specialized)
Caraboid Oligopod -complete suppression of abdominal appendages -(Specialized) Eruciform Oligopod - complete suppression of abdominal appendages - (Specialized)
Sialoid Polypod -suppression of abdominal vesicles -(Specialized) Eruciform Polypod - suppression of abdominal styli, assumption of a cylindrical shape of body - (Specialized)
Campodeoid Polypod (Primitive)

The progress of larval specialisation is thus characterised by the progressive reduction of appendages, followed by the reduction and invagination of the head within the thorax combined with the change of body-shape from the flattened to the cylindrical. The most primitive type of insect larvae is, therefore, the type with most completely developed appendages, while the most specialised type has the appendages greatly reduced or completely suppressed. With regard to the premature types, precocity of eclosion being progressively brought about, the less developed protopod types follow the more highly developed polypod types in their evolutionary sequence.

The evolutionary history of the sawfly larvae may be taken as a partial illustration of the sequence. In this group of insects, the succession of several types of larvae is traceable from the typical eNciform-polypod type, through various phases of reduction, to the oligopod and apod types. The hypermetamorphosis of the Meloidae affords another example in which the first instar is of the staphylinoid-oligopod type, but the second instar assumes an eruciform-oligopod type which may be followed by a final instar of apod type. But if the sequence is generally followed by insects of different orders, lt is also subjected to curtailment. For instance, the apod larvae of Lepidoptera appear to have been directly derived from the eruciform-polypod type without the intercalation of the transitional eruciform-oligopod type. In the Diptera, the apodous larvae, before attaining their actual condition, might have passed through several transitional oligopod and polypod phases, but owing to the possibility of abbreviations in evolution, their exact history is unknown. On the other hand, in the hypermetamorphosis of the Hymenoptera, the ontogenetic sequence is exactly the reverse of the evolutionary sequence, but this is conceivable from their mode of origin. On the whole, the evolution of larval types from a common ancestral form, has followed two lines of sequence and two modes of origin. While this conclusion is justifiable from what has been discussed, its significance requires further elucidation.

  • In the first place, we perceive from the above study that a similar sequence of larval evolution towards reductive specialisation is followed in general by the most widely different orders of Holometabola irrespective of their origin and habits. The sequence appears thus to represent a definite trend of evolution. If we compare this remarkable trend towards reductive specialisation, or more especially the sequence of appendage-suppression with that of appendage development in the embryo, certain correlations may become apparent. In the embryo, the development of appendages proceeds, as a general rule, from the anterior backwards. The rudiments of antennae usually appear first; they are followed by those of mouthparts, of thoracic legs and finally by those of abdominal appendages (vide Dawydoff, 1928). Now, in the evolution of larval types, the reduction, or more especially the suppression, of appendages goes on precisely in the opposite direction; it is the abdominal appendages which disappear first, the thoracic legs, mouthparts and antennae following successively. It seems therefore that the reductive trend of larval evolution follows a general rule, that is, the later an organ is differentiated in the embryonic stage, the more it is susceptible to reduction or suppression in the larval stage. Furthermore, the trend continues to go on in the embryogeny: as the larva is progressively specialised, the time of inhibition becomes more and more advanced. Thus in Donacia Fabricius, all abdominal appendages except the first pair are suppressed from the embryogeny; and in the Diptera, all body appendages are no longer existent in the embryonic stage. In short, the sequence of larval types appears to be governed by a definite trend toward reduction of appendages, which in its turn is determined by the sequence of development of these appendages in the embryo.

  • Secondly, in following the definite trend of evolution, it will be found that larvae, while being progressively specialised, invariably become more and more differentiated in structure from the adult. Thus larval growth may be said to represent a deviation of morphogenesis from the original line of development, while larval evolution is characterised by leading that deviation to become more and more accentuated. Consequently, the more specialised types of larvae have to undergo greater changes in their transformation to the adult stage. But the most important feature of holometabolism is without doubt the retardation or inhibition in the growth of adult organs. This inhibition of growth, being progressive, is generally less extended in the primitive forms but gradually amplified in the specialised types. Moreover, the progression in the extent of inhibited growth is always accompanied by an increase in amount of change during metamorphosis. Hence we may conclude that the evolution of insect larvae through increasing deviation of morphogenesis is primarily determined by the tendency towards increasing retardation or inhibition of imaginal growth, of which the progressive reduction of appendages is but an outward expression.

  • Thirdly, it is evident from the foregoing discussions that the specialised types of larvae are not to be regarded as representing recapitulations of ancestral embryonic phages, for the assumption of a given larval type in a given insect is determined by the nature of its previously existing type of larva, or more fundamentally by the degree of imaginal inhibition attained by that insect. Moreover, in the course of evolution, there is actually a tendency towards progressive retardation in time of eclosion. We have come to this conclusion because: (1) in animal development, increasing complexity in evolution is generally accompanied by corresponding prolongation in embryonic growth; and (2) the existence of two or more types of larvae in certain cases of hypermetamorphosis (for instance in MELOIDAE) is a proof that in normal metamorphosis, the identity of the more primitive type of larva, in being modified to give rise to a new form, has habitually come to be passed in the egg stage and thus retarded the time of birth. The specialised types of larvae present, therefore, characters rather contrasting to the premature types. For we have seen that in the latter case, the larvae are primarily embryos and their sequence of evolution is followed by a progressive advancement in time of eclosion instead of their being retarded Thus the evolution of larval types is manifested by the following definite trends: the trend towards progressive reduction of appendages which is governed by the trend towards increasing inhibition of imaginal growth, and among the premature types, the trend towards progressive shortening in the period of embryonic growth.

Arriving at this point of our discussion, we may return to the introductory remark to see how the theories of Brauer and Berlese can be reconciled with the present view in regard to the significance of the diversity of larval types. Brauer's theory appears sound in considering all insect larvae to have been derived from a common ancestor; but the adaptability of larvae as a principal factor of evolution claimed by him is only of secondary importance. Berlese's theory explains well the origin of the premature types but fails to account for the specialised types which, by their universal occurrence in all orders, must be considered as the most important of larval forms.


The origin of insect larvae presents a problem of great interest, since it signifies the origin of the Holometabola or of holometabolism in insects. At the same time, several points for inquiry arise:

  • Firstly the mode of origin, that is how, in the course of evolution, the immature Hemimetabola in the form of one nymph has been transformed into the differentiated larva and pupa of Holometabola.
  • Secondly the milieu of origin: that is, in what habitat did the primitive larvae develop?
  • Thirdly the time of origin: that is, in what epoch of geological history did tbe first Holometabola appear ? All these points are to be dealt with in the present section.

It is almost universally admitted that the Pterygota are the descendants of Apterygota, and the Endopterygota (or Holometabola) the descendants of Exopterygota (or Hemimetabola). As insects are the only winged Arthropods they must, at some period in the history of their development, have passed from an originally wingless to a winged condition; and among the earliest winged insects so far discovered, the most generalised Palaeodictyoptera are evidently hemimetabolous. But the difficulty arises in explaining how the transition from the nymph of Hemimetabola to the differentiated larva and pupa has been brought about. In dealing with the problem, it is necessary to consider at first the nature of the ancestral nymphs. The most primitive of existing nymphs are found in the order Ephemeroptera. In this order, the nymphs exhibit many features which are undoubtedly of very primitive nature; thus the possession of a generalised of mouthparts with crustacean mandibles and well-developed superlinguae, the presence of a pair of long cerci and a median caudal filament on the eleventh abdominal segment and the retention of abdominal styli in the form of gills are all characters highly suggestive of those of Thysanura. But the Ephemerid nymphs, though primitive, are certainly already specialised; for instance, the abdominal vesicles which are characteristic of both Thysanura and primitive forms of larvae are notably absent. It appears therefore that the earliest nymphs, being ancestral to larvae and themselves derived from forms resembling existing Thysanura, should likewise possess abdominal appendages consisting of styli and vesicles. Hence we may visualise the primordial nymphs as forms belonging similarly to the campodeoid-polypod type, but differing from the primitive larvae in the possession of compound eyes, median ocelli well-developed antennae, long cerci and median caudal filament besides the external wing rudiments.

These considerations show that there was, originally, a general conformity in body form and abdominal structure among the three great divisions of the class Insecta. Consequently it appears that the larvae do not represent any earlier stage of morphological development than that reached by the nymphs as the theory of Berlese maintains; and that they would have been simply derived from the latter through the reduction of visual organs, antennae and caudal appendages--or in other words, through degeneration and parasitism as many authors believe. It may be noted, however, that the absence of compound eyes and well-developed antennae in larvae may well be the result of progenesis or precocious birth. In the early instars or primary nymphs of mayflies, which (as we shall see later) are evidently hatched at a very early stage of development, the compound eyes are represented by ocelli-like structures and the antennae are rudimentary. Moreover, in certain Hemimetabola (Aleyrodidae and Coccidae) in which a tendency towards holometabolism is clearly manifested, the young also emerge at an earlier stage than usual.

Nevertheless, the most important difference between the nymphs and larvae lies neither in the presence or absence of certain structural characters, nor in the stage of development they represent, but in the mode of development. In the typical nymphs of Hemimetabola as well as in the early stage nymphs of Ephemeroptera, growth is apparently directed towards gradual completion of imaginal features. In the larvae, many of the adult characters are more or less held in an embryonic state in the form of undifferentiated imaginal germs, and growth apparently proceeds in a different way; it leads to the pupal stage. The differentiation of immature Holometabola into the double forms of larva and pupa is, therefore, essentially a result of inhibition in the growth of adult characters in the larval stages. The origin of this inhibition of imaginal growth is obscure, but it must be sought for in the egg. It is interesting to note that the Palaeozoic holometabola were usually of much smaller size than the Hemimetabola of the same era, consequently it appears that they may have developed from much smaller eggs. The Palaeodictyoptera are generally recognised as the probable ancestors of Holometabola, and these primitive forms were very large insects; their wing-length normally ranges from 45 to 75 mm., rarely below 30 mm., and sometimes reaching the gigantic size of 160-190 mm. as in the genera Megaptilus Brongmart and Homoeoph1ebia Handlirsch (vide Handlirsch, 1908). On the other hand, the wings of Palaeozoic Holometabola attain only a length of about 10 mm. It is therefore possible that the sudden shortening of egg-size or egg-provisions may play an important role in determining the inhibition of imaginal growth. If this view were correct, we may figure the origin of holometabolism as follows: " In the course of evolution of the class Insecta, a Palaeozoic Hemimetabola, by certain genetic changes, has produced eggs which contain an unusually small amount of food reserves; the eggs eventually developed, giving rise to the first insect larva in which growth of some of the lately appeared characters was more or less inhibited and held undifferentiated in the form of imaginal germs." Thus the larval instar arises as a consequence of inhibition in imaginal growth which is primarily determined in the egg, and which leads to a deviation of morphogenesis from the original line of development. The pupal instar arises as a consequence of imaginal inhibition in the larval stage--it restores the original line of development of the insect by escaping from the inhibitory control.

Milieu of origin

The problem of the original milieu of insect larvae has been notably dealt with by Lameere (1899), who maintained that " l'holometabolisme est du a la penetration de l'insecte a l'interieur des tissues vegetaux." Consequently, according to Lameere, the oldest larvae must have had a burrowing or mining habit. But according to the foregoing studies on the larval sequence, we are led to adopt a quite different view that " insect larvae appear to have been originally aquatic in habit rather than phyto-parasitic or terrestrial." For not only the most generalised type of existing larvae are without exception aquatic, but many of the principal orders as well show evidence of aquatic origin.

In tracing the evolution of larval types, we have shown that the various types of insect larvae, being derived from a common ancestral form, have followed two ecological lines of sequence, one terrestrial and the other aquatic. The ancestral or primitive forms, actually exemplified by the campodeoid-polypod larvae of the neuropterous family Corydalidae, are all aquatic in habit. Of the derived types, while some have acquired a terrestrial habit and have been differentiated into the orders Mecoptera, Hymenoptera and Lepidoptera, others have been retained in the aquatic medium and have diverged in the course of evolution into the orders Neuroptera, Coleoptera, Trichoptera and Diptera. Thus in Neuroptera, all larvae belonging to the two most primitive families, Corydalidae and Sialidae, are found in water; in Coleoptera, out of the ten families included in the more primitive suborder Adephaga, five are exclusively aquatic; in Trichoptera, the early stages are almost without exception passed in water; in Diptera, the majority of the Nematocerous families are either entirely or preeminently aquatic in their larval stage. On the other hand, aquatic adaptation is always a rare and exceptional phenomenon in the orders of the terrestrial line, and whenever it occurs is invariably confined to the more specialised forms. As to the two other orders of the aquatic line, the Strepsiptera and Siphonaptera, their want of aquatic representatives is to be explained by their being secondary in origin.

It may be opposed, however, that in the (Coleoptera and Diptera, the terrestrial larvae of such families as Carabidae and Bibionidae are to be considered as forms of more generalised structure and hence of more primitive origin than the aquatic larvae of their respective allied families. But in these particular cases, generalisation of structure is not necessisarily correlated with primitiveness of origin. For insects as a whole are primarily terrestrial; if the primitive larvae were proved to be aquatic in habit (the existing facts are in favour of this view), it only means that the first Holometabola have arisen in a condition in which their larval stage was adapted to be passed in water. In other words, if insect larvae originated in the aquatic milieu, the terrestrial forms are to be regarded as having reacquired the original habit of the class. At the early stages of their evolutionary historv, the primitive aquatic larvae would not present much adaptive modification, whence the terrestrial forms arose and have retained a somewhat more generalised type of structure; whereas the aquatic forms might be further modified in adaptation to their habitat and exhibit physiognomy of a somewhat more specialised nature. The terrestrial forms thus derived were liable to be re-adapted to life in water, consequently the aquatic larvae are likewise of two different origins, including those which are primarily aquatic and those which are secondarily aquatic. By following these considerations, we can understand why the overwhelming majority of aquatic larvae are found in families of more primitive nature, and why these larvae, while belonging to more generalized types in evolutionary sequence, usually exhibit a much higher degree of adaptive modification than the secondary aquatic larvae, of similar habit, found in the more specialised families.

Besides the general reasons given above, we may examine more specially the two cases in question. In the case of Coleoptera, it is noteworthy that the sialoid-polypod larvae, while representing the most primitive type of the order, are without exception aquatic. They are found in the families Gyrinidae, Dytiscidae and Hydroptilidae. With respect to the latter family, it is also of interest to note that the larvae of the two subfamilies Limnebiinae and Hydraeninae are even considered by Boving and Craighead (1931) to be of more primitive nature than those of the Adephaga. The development of the two suborders Adephaga and Polyphaga, as we shall presently see, must have taken place at a very early stage in the evolution,of the Coleoptera. Consequently, it is conceivable that there should be a great number of aquatic forms present in the more primitive families of Polyphaga; whereas in the higher groups (Heteromera, Phytophaga, Rhynchophora and Lamellicornia), there are but few such cases found in the families Chrysomelidae and Curculionidae; these latter forms are evidently secondary in origin. In the case of Diptera, while the Bibionid larvae are structurally very primitive, their adult forms present evidence of specialisation in the reduction of wing veins and in the assumption of holoptic eyes in the males. And, if we were to accept the amphipneustic and metapneugtic conditions of respiration as features of aquatic adaptation, we might regard the whole suborder of Cyclorrhapha as having originated from thoge aquatic formg which have been already specialised.

In short, all the above facts seem to support the view that insect larvae were primarily aquatic in habit, and the transition from the hemimetabolous to the holometabolous type of development, which is essentially characterized by a changing mode of morphogenesis, has taken place in the aquatic milieu.

Time of origin

The Holometabola appear to be a very ancient group. The existence among their larval forms of a campodeoid-polypod type resembling so much the generalised Thysanura indicates that the group probably made its first appearance at a comparatively short time after the appearance of the Hemimetabola. This conception is wholly in accord with the palaeontological data, since in the Lower Permian times, authentic Holometabola were already abundant. The studies of Tillyard (1926, 1932) have shown that in the Lower Permian Beds of Kansas, U.S.A., not only the Mecoptera were rich in varieties and species, but the three suborders of Neuroptera, the Sialoidea, Raphidioidea and Planipennia, were represented by distinct and comparatively highly specialised types. It seems therefore that the earliest Holometabola must have developed not later than Upper Carboniferous times.

But the study of Tillyard (1926) has gone still further; he was able to show that the small Upper Carboniferous wing of Metropator pusillus, placed by Handlirsch in the order Palaeodictyoptera, is a true Mecopteran and, consequently, a true Holometabola. With regard to its mecopterous affinity, we shall have something to say in the following section in discussing the phylogeny of the Holometabola, but the small size and the venational features of the wing leave little doubt as to its holometabolic nature. Hence we have to look back for the origin of Holometabola to a more remote period in geological history. Now, the age of Metropator Handlirsch is of Lower Upper Carboniferous, and, seeing that fossilisation can only preserve an infinitely small part of the fauna or flora of a given period, while of those preserved, only an infinitely small part are available for study, we may safely conclude that at this age, the Holometabola must have already attained a certain degree of development, and consequently they must have originated at a time not later than the Lower Carboniferous.


In this section, we shall see how the views advanced in the foregoing pages may help us to explain the origin and phylogeny of the various orders of Holometabola. To do this, we have to appeal also to other domains of entomology among which palaeo-entomology stands first in importance as being possible to afford direct evidence. Fortunately, recent studies in fossil insects have brought to light a certain amount of data which may enable us to get some rough idea of the evolution of the Holometabola. The subject will, accordingly, be dealt with under the three following headings: (1) the nature of the primitive Holometabola, (2) the principal lines of holometabolous evolution and (3) the origin of the orders.

Nature of the primitive Holometabo1a

We have seen that the Holometabola made their first appearance possibly in the Lower Carboniferous period, but their earliest manifestations are unknown. The most primitive of Holometabola so far recorded is Metropator pusillus from the Upper Carboniferous, and this is the only ingect of holometabolous type known from that period. It is therefore of great interest to see what is the nature of this Carboniferous Holometabola and, together with that of the Lower Permian forms, how it may permit us to draw certain conclusions about the characteristics of the ancestral Holometabola.

The nature of the Metropator wing is liable to be disputed. While it is regarded by Tillyard as a true Mecopteron, its affinity to Neuroptera is none the less obvious, as the long subcosta and the broad costal space are both neuropterous features. The subject, however, may be better approached by considering at first the relative antiquity of the two orders in question. The Mecoptera are generally believed to represent the oldest order of Holometabola from which the majority of the other orders are capable of derivation. The abundance of these insects in the Lower Permian times and their venational affinities with the early types of many holometabolous orders are probably the reasons for this belief. But from many points of view, the order Neuroptera must be considered as the most primitive of all existing Holometabola, and consequently older than the Mecoptera. For ontogenetically, the Neuroptera present a larval type which is theoretically the most closely related to the primitive larvae, while the mecopterous larvae represent evidently a more specialised type. Ecologically, the neuropterous larvae are primitively aquatic a habit conforming in speculation to that of the earliest Holometabola, but those of Mecoptera appear to have been evolved on land. Palaeontologically, while the Mecoptera were abundant in the Lower Permian period, the fact that the Neuroptera were represented at the same period by very specialised types indicates an even more ancient origin. These considerations, supported by the venational features noted above, may well lead us to believe that Metropator pusillus is more closely allied to Neuroptera than to Mecoptera.

But if the Neuroptera were proved to be the most primitive of living Holometabola and Metropator pusillus to be a Neuropteron, we cannot yet conclude that the earliest Holometabola were of the Neuropteroid type. As Holometabola would have had a very remote origin, there should exist a long series of transitional forms leading from the earliest types to the Neuroptera. Moreover, the primitive forms of Coleoptera, or Protocoleoptera, being very different from the Neuroptera in venational features, are not to be regarded as having arisen from Neuropteroid ancestors. It is noteworthy, however, that with the exception of Coleoptera, all the main orders of Holometabola are referable to neuropterous ancestry (to be presently discussed); we may thus draw a preliminary conclusion that the ancestral form of Holometabola has given rise, in the course of evolution, to the early types of both Neuroptera and Coleoptera. Now the early types of both orders are characterised by a long subcosta and a broad costal space, and, as these features are also common to the Protorthoptera, we must sssume that the earliest Holometabola are the forms which exhibit certain relationships with the Protorthoptera. But the latter as a group are already more or less specialised, hence to account for the origin of the Holometabola, the most agreeable hypothesis is that " the Holometabola appear to have originated from a Paleodictyopteroid stock (or even a still earlier stock) which is more or less allied to, and probably contemporary with, that giving rise to Protorthoptera."

The earliest Holometabola are thus to be presumed as forms which exhibit certain relationships with the Protorthoptera, their larvae are probably aquatic in habit and possess a complete set of body appendages. From these early forms might have arisen the ancestral forms of Protocoleoptera as well as those of Neuroptera. The formation of these two primary lines of evolution must have taken place at a very early period, since the Neuroptera were proved of great antiquity and the ancestors of the Coleoptera, before attaining the form of Protocoleoptera with tegminous elytra would, originally, have membranous fore-wings. The ancestral forms of true Neuroptera, though not discovered, may be conveniently designated as the Protoneuroptera. The Protoneuroptera have probably given rise to two groups of insects, the Neuroptera and the Protomecoptera. The community of descent of these two groups is clearly manifested by the venational features of such forms as Choristosialis enigmatica (Tillyard, 1932) and Protomerope permiana (Tillyard, 1926). The Protomecoptera, in turn, appear to have given rise to the groups of Mecopteroidea and Trichopteroidea, as the venational affinities of their early forms tend to indicate. It seems therefore that the Protomeropidae and some of the other mecopterous forms discovered in the Lower Permian beds are not true Mecoptera, since their larvae should be of a more generalised type and aquatic in habit in order that, from them, both the aquatic Trichopteroidea and terrestrial Mecopteroidea might have originated. In other words, Protomecoptera and their derivatives, the true Mecoptera, must have co-existed in the Lower Permian and Pre-Permian times. Thus in their early history, the Holometsbola have probably followed four principal lines of evolution: namely, the Coleopteroidea, the Neuropteroidea, the Mecopteroidea and the Trichopteroidea. The Coleopteroid line represents an early offshoot and comprises actually the orders Coleoptera and Strepsiptera; the Neuropteroid line comprises the single order Neuroptera; the Mecopteroid line the orders Mecoptera, Hymenoptera and Lepidoptera; the Trichopteroid line the orders Trichoptera, Diptera and Siphonaptera.

Origin of the orders

  • The Coleopteroid orders. The Coleopteroid line of evolution is chiefly directed to the development of the order Coleopters. Although true Coleoptera elytra are not known until the Trisssic period, and the earliest fossils which may be recognised as generalised Coleopteroid types are not known until the Upper Permian age, the order appears to have had a more remote origin. For in Triassic rocks, fossil elytra were abundant, including already specialised types, some of which have been referred to the families Chrysomelidae and Curculionidae. We are thus led to believe that the tegminous remains of the Upper Permian are not the immediate ancestors of existing Coleoptera, but are to be regarded as specialised branches of the very ancient group Protocoleoptera which have extended to that period; and that true Coleoptera must have developed from the still earlier progenitors not later than or even before the Permian age. The Protocoleoptera of the Upper Permian are nevertheless of such great significance that they permit the following presumption with regard to the evolution of the Coleoptera. Firstly, the nature of these fossil wings tends to indicate that the Coleoptera, before the fore-wings had attained their present condition of structure, passed through a series of transitional stages in which these organs were tegminous and with clearly marked veins like those of many Hemiptera. This conclusion is quite in agreement with the view put forward above, that the Coleopteroid line represents an early offshoot of holometabolous evolution. Secondly, since some of the fossils, namely, the Permophilidae, present notable resemblance to the existing Hydrophilidae, we can at least suppose that the larvae and even certain adults of these ancient forms were aquatic in habit, and that the two suborders Adephaga and Polyphaga must have developed at a very early stage in the evolutionary history of the order. These propositions, as we have seen, are also conformable to the facts presented by living Coleoptera.

    The origin of the order Strepsiptera appears most obscure. While the assumption of a staphylinoid-oligopod type by their primary larvae may indicate certain relationship with the Coleoptera Polyphaga, they are so different in other respects as to render improbable the belief that they are directly derived from the latter. It is possible that they might have arisen from a common stock with the Coleoptera, but without further evidence all speculations on the subject are fruitless.

  • The Neuropteroid order. The Neuropteroid line of evolution is represented bv a single order Neuroptera, which is the most primitive of all living orders of Holometabola. These insects have certainly existed in the Upper Carboniferous or even at a time earlier than that. Their ancestral forms, together with those of Coleoptera, are probably the earliest branches of Holometabola. They have retained many primitive features in the subcostal venation of wing, the form of prothorax, the structure of larvae, etc. It appears certain that the order has never given rise to any other groups except the three which already existed in the Lower Permian, namely, the Sialoidea, the Raphidioidea and the Planipennia.

  • The Mecopteroid orders. The Mecopteroid orders here concerned are the Mecoptera, the Hymenoptera and the Lepidoptera; but in a broad sense, Trichopteroidea should also be included. The most generalised types of Mecopteroidea so far known are the Protomecoptera of the Lower Permian times. The venation of these early forms, with its profuse ramifications and widened costal space, is so closely allied to that of Neuroptera that the two groups most probably have arisen from a common stem, the theoretical Protoneuroptera. The Protomecoptera are further characterised by the development of Macrotrichia on the wing (observed by Tillyard, 1926) and by the retention of aquatic habit in the larval stage (discussed above). Both these points are of great significance, since they afford a basis from which the evolutionary tendencies of the Mecopteroid orders (sens. lat.) may be traced. It appears certain that all these orders are derivable from the Protomecoptera, yet their inter-relationships are obscure, more especially with regard to the Hymenoptera, of which no direct evidence is available. Now, on the basis of these two important characters, it is possible to show that the orders in question have followed two lines of evolution, the one (Mecopteroidea sens. str.) characterised by the reduction of wing trichia and the re-acquisition of terrestrial habits in the larval stage, the other (Trichopteroidea) by the increasing development of wing trichia and the retention of original larval habits. Consequently the Hymenoptera, by the conformation of wing structure, larval type and habit, must be regarded as being closely allied to the true Mecoptera in origin, like the Diptera to the Trichoptera. But the position of the order Lepidoptera becomes problematical, since it agrees, on the one hand, with the Trichoptera and Diptera in the development of wing trichia, and, on the other hand, with the Mecoptera and Hymenoptera in the re-acquisition of terrestrial habit by the larvae. The close affinity of the Lepidoptera and Trichoptera has long been recognised, and the development of wing trichia in both orders tends also to support the view. But the change of habit in the larval stage, accompanied by the loss of abdominal styli and the assumption of a uniform eruciform-polypod type, are characters common to all the three orders Mecoptera, Hymenoptera and Lepidoptera; and this conformity of characters must be considered to indicate community of descent. We have thus grouped the three orders together to represent a terrestrial line of evolution in contrast with the Trichopteroid orders which represent an aquatic line, both arising from the Protomecoptera.

    The true Mecoptera must have had a very remote origin, since they were abundantly represented, and co-existed with the Protomecoptera, in the Lower Permian times. The earliest known Hymenoptera occur in Upper Jurassic strata and those of Lepidoptera in rocks of Eocene age, but it is practically certain that both orders have appeared on the earth at a respectively earlier period. Although true Lepidoptera are of more recent origin than true Hymenoptera, yet on account of their affinity with the Trichoptera, their ancestral forms would appear to have become detached from the main stem of Mecopteroid evolution at an earlier period than those of the Hymenoptera.

  • The Trichopteroid orders. Included in this series are the orders Trichoptera, Diptera and Siphonaptera. The origin of the Siphonaptera is very obscure; they are generally considered to be related to the Diptera because their larval characters resemble those of certain Nematocera. On the other hand, the community of descent of the Trichoptera and Diptera is manifested in many ways, notably in the similarity of larval habit and of development of wing trichia are discussed above, and in the venational affinities of their possible ancestral forms (Paramecoptera, Paratrichoptera and Protodiptera) as observed bv Tillyard (1937). The two first-mentioned features, while characterising almost all the living Trichoptera, have been retained only in some of the more primitive groups of modern Diptera. As to the venational affinities of the three extinct groups, Tillyard's conclusions appear to be indubitable; but owing to the difference of larval habits, these groups cannot be regarded as having been derived from the true Mecoptera as he believed. The Paramecoptera, probably descended from the Protomecoptera, might be taken as the forerunners of Trichoptera; similarly the Protodiptera, evidently related to the Paratrichoptera, exhibit features ancestral to the Diptera. But between these extinct groups of the Permian times and the two modern orders the earliest fossils of which were recorded in the Jurassic rocks, no annectent forms have yet been discovered.


It seems appropriate to add here a brief discussion on the evolution of the nymphal forms. Nymphs, like the larvae of Holometabola, are referable to a number of types and have followed a similar sequence in evolution, although they have not attained such a diversity of form and structure as that found in the larvae. The following principal types may be recognised. (I) The Campodeoid-polypod type. This is presumably the most primitive type of nymph. It is no longer existent among the living groups of Hemimetabola; but from the reasons previously stated, we must believe that such a type did exist and was characteristic of the earliest forms. Nymphs of the present type were certainly more primitive in organisation than the primitive larvae, as we have seen, they would have very generalised mouth-parts, ten pairs of typical appendages on the first ten abdominal segments, a pair of long cerci and a median caudal filament on the eleventh segment, long antennae, well-developed compound eyes and three ocelli. They were literally Thysanura carrying wing-rudiments. The intercalation of a nymphal stage in insect development may be thus explained as the result of a progressive evolution in organisation (i.e. the acquisition of wings) without the corresponding prolongation in embryonic growth. (2) The Ephemeroid-polypod type. This type is characteristic of the nymphs of Ephemeroptera. It is found also among Plecopteroid insects in the existing family Eustheniidae and in the Permian order Protoperlaria (Carpenter, 1935). The type is evidently derived from the preceding by the suppression of abdominal vesicles. It is therefore comparable with the sialoid-polypod type of larvae and represents a parallel stage of evolution. (3) The Blattoid-oligopod type. By far the great majority of e~isting nymphs are referable to this type which is derived from the Ephemeroidpolypod type through further suppression of abdominal styli. The type parallels the caraboid-oligopod type of larva in stage of evolution and, according to Berlese, represents a post-oligopod phase of development. Besides these principal types which are characterised by the progressive suppression of abdominal appendages, there are others following different modes of evolution. For instance. the primary nymphs of mayflies evidently originated by way of progenesis. For these nymphs always issue from the egg at a stage in which the abdominal gills are not yet developed, the mouth-parts are still rudimentary, the antennae are only represented by a few segments, the compound eyes are mere ocelli-like structures and the internal systems are still incompletely differentiated (vide Lestage in Rousseau, 1921). They are comparable with the polymerous-protopod phase of embryogeny, although they are well advanced in organisation in certain respects, such as the development of thoracic legs. In the Thysanoptera, Aleyrodidae and Coccidae, the tendency towards inhibition of imaginal growth in early nymphal instars is clearly manifested so that the insects leave the egg not as ordinary nymphs but as incipient larvae which are finally followed by an incipient pupal stage.

There is thus a close parallelism in evolutionary tendencies between the nymphs of hemimetabola and the larvae of Holometabola. Moreover, the differentiation of the post-embryonic period of growth into larval and pupal stages so characteristic of the holometabola occurs also in some groups of Hemimetabola.

Finally, we may summarise here the various ways by which the different nymphal and larval forms were originated. These ways are: (1) by further complications in evolution without corresponding prolongation in embryonic growth, (2) by progenesis or shortening in period of embryonic growth, (3) by inhibition in imaginal growth and (4) by reductive specialisation; but in all cases, the evolution of a given form is often complicated by caenogenesis. The first-mentioned factor or process of evolution is of common occurrence in the animal kingdom and is especially characteristic of the lower groups of Metazoa such as Porifera and Coelenterata. It accounts for the origin of nymphs or of metamorphosis in insects, since the Apterygota are generally regarded as ametabolous. The existence of more than two types of larvae in the life history of many hypermetabolous insects (Meloidae, Mantispidae, etc.) is of similar origin. The second factor accounts for the origin of the primary larvae of many endoparasitic Hymenoptera and of the primary nymphs of Ephemeroptera. The third factor explains the origin of larvae or of holometabolism in insects as well as the origin of incipient larval instars in certain Hemimetabola. Lastly, the fourth factor, that of reductive specialisation, is the most characteristic feature of the evolution of the larval and nymphal types.

The various examples of hypermetamorphosis may be roughly classified under two types according to their mode of origin: (1) The meloid type, couprising the typical examples of hypermetamorphosis found in the Coleoptera, Strepsiptera, Neuroptera and Diptera, originated through the retention of more archaic type or types of larvae in the early instars; or, in other words, through the specialisation of the later stage larvae without corresponding prolongation in embryonic growth. (2) The figitid type, characteristic of many parasitic Hymenoptera, originated through the progenesis of embryonic forms.


In this paper, the author has discussed some aspects of the problem of larval evolution in insects and has arrived at the following conclusions.

  1. All insect larvae are derivable from a common ancestor.
  2. The primitive larvae are presumably of the campodeoid-polypod type, having three pairs of thoracic and ten pairs of abdominal legs; the latter bear each a vesicle and a stylus.
  3. From the primitive type, the various types of existing larvae appear to have been derived in two ways: by specialisation and by progenesis. Those of former origin are designated as specialised types; those of latter, as premature types.
  4. Specialisation of larval types has taken place chiefly through the reduction of appendages which proceeds from the posterior part of the body to the anterior. The progress in reductive specialisation is governed by the progress in degree of inhibition in imaginal growth. Consequently, the evolution of the specialised larval types follows a sequence proceeding from the polypod state to the apod, and the assumption of a given type by any given insect depends, primarily, upon the degree of imaginal inhibition attained by that insect.
  5. Progenesis of larval forms has been brought about probably by the precocious rupture of egg envelopes. The evolution of premature types follows a sequence which is the inverse of the embryonic sequence in normal insect development; the assumption of a given type of premature larva by any given insect depends, primarily, upon the stage in organisation in which eclosion takes place.
  6. The larva arises as a consequence of inhibition in imaginal growth which is initiated in the egg and which leads to a deviation of morphogenesis from the original line of development. The pupa arises as a consequence of imaginal inhibition in the larval stage, it restores the original line of development of the insect concerned by escaping from the inhibitive control.
  7. The earliest larvae were probably aquatic in habit; they appear to have developed at a time not later than the Lower Carboniferous and from a Palaeodictyopteroid stock, more or less allied to that giving rise to the Protorthoptera.
  8. From the primordial Holometabola might have arisen the ancestral forms of the Protocoleoptera as well as those of the Protoneuroptera, the two earliest lines of holometabolous evolution. The Protoneuroptera gave rise to the early types of Neuroptera and Protomecoptera; the Protomecoptera to Mecopteroidea and Trichopteroidea; the former comprises the orders Mecoptera, Hymenoptera and Lepidoptera, the latter the orders Trichoptera, Diptera and Siphonaptera.
  9. The nymphs of Hemimetabola, like the larvae of Holometabola, are referable to a number of types and exhibit similar tendencies in evolution. The primordial nymph was theoretically campodeiform and polypodous, resembling exactly a generalised Thysanuran carrying wing-rudiments and differing from the primitive larval forms in the possession of compound eyes, median ocelli, well-developed antennae and long caudal appendages as well as in the wing-rudiments.


BERLESE A., 1913, Intorno alle metamorfosi degli insetti. Redta 9: 121-38.

BOVING, A. G., and CRAIGHEAD, F. C., 1931, Larvae of Coleoptera. Brooklyn.

BRAUER, F., 1869, Betrachtung ueber die Verwandlung der Insekten in Sinne der Descendenz-Theorie. Verh. zool.-bot. Ges. Wien. 19: 299-318.

CARPENTER, F. M., 1935, The Lower Permian Insects of Kansas. Part 7. The Order Proterperlia. Proc. Amer. Acad. Arts Sci. 70: 101 46.

CARPENTER, G. H., 1921, Insect Transformation. London

CHRYSTAL, R. N., 1930, Oxf. For. Mem. 11.

DAWYDOFF, C., 1928, Traite' d'Embryologie comparee des Inverte'bre's. Paris.

EASTHAM, L. E. S., 1929, The post-embryonic Development of Phaenoserphus viator Hal. Parasitology 21: 1-21.

HANDLIRSCH, A., 1908, Die Fossilen Insekten. Leipzig.

HEYMONS, R., 1896, Uber die abdominalen Korperanhange der Insekten. Biol. Zbl. 16: 855-64.

IMMS, A. D., 1937, Recent Advances in Entomology. London.

___ , 1938, A General Textbook of Entomology. London.

___ , 1942, Outlines of Entomology. London.

JACKSON, D. J., 1928, The Biology of Dinocampus (Perilitus) rutilus Nees. Proc. Zool. Soc. lond. 1928: 597-430.

JAMES, H. C., 1928, On the life-histories and economic status of certain Cynipid parasites of Dipterous larvae. Ann. appl. Biol. 15: 287-316.

LAMEERE, A., 1899, Discours sur la raison d'etre des metamorphoses chez les Insectes. Ann. Soc. ent. Belg. 43: 619-36.

LESTAGE, J. A., 1921, in Rousseau, larves et Nymphes aquatiques des Insectes d 'Europe. Brusselles.

PACKARD, A. S., 1898, A Textbook of Entomology. New York.

SHARP, D., 1895, Insects. Camb. Nat. Hist. 5.

SNODGRASS, R. E., 1931, Morphology of the Insect Abdomen. Smithson. Misc. Coll. 85, No. 6.

STRICKLAND, E. H., 1923, Bull. Agric. Can. (N.S.) 26.

TILLYARD, R. J., 1926, Kansas Permian Insects. Part 7. The order Mecoptera. Amer. J. Sci. 11: 133-64.

___ , 1932, Kansas Permian Insects. Part 14. The order Neuroptera. Amer. J. Sci. 23: 1-30.

___ , 1937, The ancestors of the Diptera. Nature 139: 66-7.

YUASA, H., 1922, A Classification of the Larvae of the Tenthredinoidea. Illinois Biol. Monog. 7, No. 4.