Micromalthus debilis 33 , the sole constituent of the family Micromalthidae (Coleoptera), has been an enigma since its description over a century ago. Of about 17 extant independent haplodiploid clades, only one – M. debilis – consists of only a single species ( 35 ) and of seven known independent cyclically parthenogenetic clades, again only one – M. debilis – contains only one species ( 20 ). It is generally known that various components of this insect's life cycle are unique among Coleoptera and indeed unique among Metazoa. Although most major work on M. debilis occurred before 1950, the major features of the beetle's life cycle are well established, although they are not widely known and they are far from being well understood. Here we provide a general introduction to M. debilis , review the history of studies of its life cycle, briefly compare it to other species with somewhat similar life cycles, and suggest a novel evolutionary interpretation.

Since then, on the basis of two different character systems, namely the wing‐folding pattern of adults and larval mouthparts, Micromalthidae has been placed with Cupedidae in the primitive suborder Archostemata ( 15 ; 7 ). This placement has remained essentially unaltered, reinforced by the fact that 12 , 13 , 14 ), 29 ), 31 , 32 ), and 30 ) supported the archostematan affinities of Micromalthus . A detailed study of the larva of Micromalthus is given by 5 ), whose phylogenetic analysis of larval characters places the genus as the sister group of Cupedidae, within Archostemata. Somewhat earlier, however, 2 ) removed Micromalthidae from Archostemata and placed it at the end of Cantharoidea, based on absence of notopleural sutures in the adult. It is tempting to compare the paedogenetic larva of M. debilis with a neotenic female adult of Lampyridae (as will be discussed below) and this may have accounted for Arnett placing Micromalthidae immediately after Lampyridae. This new placement, however, was based on little supporting data. More plausibility was added to a non‐Archostematan alliance of M. debilis by 21 ) who stated that only symplesiomorphies were used by Crowson to group Micromalthidae with the other three families of Archostemata. As is typical with enigmatic taxa, for some authors it was easier to say where Micromalthidae did not belong, than it was to say with certainty where the family did belong. 21 ) stated that M. debilis may be a simplified cantharoid or lymexyloid. Despite this, there seems solid consensus that Micromalthus belongs in Archostemata, near Cupedidae (authors given above).

The first elucidation of the life cycle, although somewhat superficial, was presented in two papers by 3 a, 4 ). These will be discussed more fully in the section on the life cycle of M. debilis , but an important taxonomic change occurred in 4 ), who suggested that the family Micromalthidae be erected to accommodate M. debilis . However, Barber failed to state where this new family should be placed in the classification of Coleoptera, although presumably he would have placed it near Lymexylidae.

Surprisingly, in the same year and journal volume, 23 ) provided the first description and figure of the cerambycoid larva of M. debilis , as well as the first illustrations of the adult. Hubbard compared the characters of the larvae of Micromalthus and Hylecoetus and supported LeConte's placement of Micromalthus in Lymexylidae based on similarities in larval antennae and mouthparts. From an historical point of view, it is interesting that the level of analysis presented for the larval stage far surpassed that for the adult.

33 ) described the genus Micromalthus , and the species M. debilis , based on material collected in rotting wood in Detroit, USA. He placed the genus, with some question, in the family Lymexylidae because of certain morphological similarities between Micromalthus and Hylecoetus Latreille, 1806, the latter of which is a `typical' lymexylid. 33 : p. 613) mentioned that the species is `feeble and ill‐developed', and that M. debilis would be expected to have lost the peculiar characters of the maxillary palpi present in all other members of Lymexylidae.

Seemingly major differences between Hong Kong and American specimens of M. debilis were noted by 34 ). For example, they found the triungulin larva to have a single stemma on each side of the head capsule, whereas all previously published descriptions of the triungulin stage from the United States or South Africa did not mention the presence of any stemmata. Despite these differences, some of which are more readily reconciled than others, none of the above authors described a new species of Micromalthus based on specimens from South Africa or Hong Kong. However, 30 ) mentioned that presence of stemmata in the Hong Kong form may be justification for recognition of a species distinct from M. debilis .

According to the latest classification of Coleoptera by 32 ), Micromalthidae is composed of the single species, M. debilis . Several papers have appeared concerning the differences between North American, South African and Hong Kong specimens of M. debilis . Both 38 ) and 36 ) listed differences between South African and American specimens of M. debilis , in the triungulin and cerambycoid larva, and in the adult female. One of the criteria used by 36 ) was the slightly larger size of the South African specimens. Clearly, this difference has little taxonomic or systematic significance; 1 ) suggested that a varied intrapopulation size is common in Coleoptera whose larvae cannot control their nutritional environment. As an example, a tree‐boring species of Cerambycidae may exhibit three times the intrapopulation body length variation as in a free‐ranging species of Dytiscidae or Carabidae.

As an explanation for the presence of M. debilis in South Africa, 38 : 276–77) offered two possibilities. First, the original introduction may have occurred in pine lumber imported from North America. However, as 3 a) mentioned, M. debilis inhabits very old wood in the red rot stage of decay. Therefore, since the imported lumber was to be used for constructing underground mine shafts, it is doubtful that badly decayed logs would have been included in the shipment. The other explanation offered by Pringle is that M. debilis may also (naturally?) occur in some of the more moist forests of South Africa. It is difficult to speculate on the validity of either hypothesis, although it is generally believed that M. debilis expanded its range in historical times through transport of infested lumber and wood products ( 29 , 30 ).

Although it was originally described from, and is apparently native to, the eastern United States ( 29 ), M. debilis subsequently has been found in South Africa ( 38 ; 36 ), Hong Kong ( 34 ), Cuba, Brazil and Hawaii ( 29 ), and British Columbia, New Mexico, Florida and Gibraltar ( 30 ). This range expansion apparently represents passive dispersal by humans, as 38 ) stated that M. debilis was found in structural timber at a depth of 6000 feet below the surface of the ground in a South African mine. The distribution of M. debilis has been examined in detail by 37 ), who stated that a recently collected specimen from Belize might represent part of the natural range of the species. Miocene fossil triungulins of M. debilis have been reported from Mexico ( 39 ). Earlier, the range of the genus Micromalthus was wider, as a fossil Micromalthus has been reported from Cretaceous Lebanese amber ( 14 ).

Life cycle: an historical review of the evidence

Nothing was stated in the original descriptions of the adult and larvae by 33) or 23), respectively, that gave any indication of the bizarre and complex life cycle of this beetle. Had either of these authors known what now is known about M. debilis, certainly more importance would have been placed on consideration of its taxonomic and evolutionary significance.

3 a, 4 ) is credited with first having observed M. debilis in detail, and with discovering its basic life cycle. His remarks were based on examination of a field colony and on careful laboratory observations. Barber was given a vial of beetle larvae for identification and noticed three distinct types. One form he recognized as M. debilis from the description in 23 ), but two other distinct forms were present in the sample. These, in fact, were the triungulin and paedogenetic larva of Micromalthus . 3 a: p. 33) thought the robust larva from one of his colonies to have been a prepupa, but `this hypothesis was shattered…when embryos began issuing alive, but in an oval shape, from the ventral surface, close to the tip of the body of one that had shortly before been isolated in a small vial'. Barber had discovered the paedogenetic component of the life cycle, as well as the triungulin (caraboid, or legged larva), cerambycoid and reproductive larvae. As a summary of the life cycle, 3 a: p. 35) listed five distinct forms: (1) viviparous larviform, reproductive stage, giving birth to (2) legged larvae which molt into (3) legless larvae, giving rise to either (4) pupae or (5) winged adults. Barber compared the situation discovered in M. debilis to the extreme sexual dimorphism exhibited in the beetle family Phengodidae. However, at that time, the fact that M. debilis reproduced parthenogenetically was unknown, and Barber speculated how the paedogenetic larvae could be fertilized while deep inside wood, or underground.

Significant additions to knowledge of the life history of M. debilis were made in a second paper by 4). The timing of the various larval instars was noted, based on extended laboratory observations. The caraboid or triungulin larva feeds for about a week, after which it moults to the cerambycoid larva. This form may moult once or twice additionally without significantly changing form. It then bores through the wood for several months, during which the ovaries of the next instar become apparent. After becoming quiescent, the cerambycoid larva moults to the paedogenetic larva, or, very rarely, to a pupa. The young larvae are born in two weeks and average 10 in number.

For the first time, the sex‐determination mechanism, or at least the sex segregation mechanism, was described by 4). Some paedogenetic larvae die apparently without giving birth, and others produce a single, large egg which remains attached to the outside of the mother larva. In about 10 days, this egg hatches to a form unlike any other; it was called the curculionoid larva. This larva inserts its head into the exit system of its mother and devours her body contents. Once fed, it moults into what Barber called the metrophagous larva, after which pupation occurs. Only adult, winged males are produced in the above manner. 4) felt that this radically divergent life cycle was a hindrance to inbreeding because it was much easier to produce a winged female than a winged male. Therefore, by the time the latter is achieved, the females would be either dead, or otherwise unavailable for mating with their male siblings.

To more fully explore the hypermetamorphosis of M. debilis larvae, 4) isolated 21 caraboid larvae and 2 months later, the contents of the vials were examined. Of 16 survivors, the following number of different forms were found: seven cerambycoid larvae, two of which were very close to moulting to the paedogenetic larva; four paedogenetic larvae without apparent embryos; two were eaten after having given birth to larvae; two paedogenetic larvae with a male egg on each; and one pupa of the adult female. It is important to note here, as Barber did, that since the wood for all these larvae was at a similar stage of decay and was kept under the same conditions, the variety of forms arising from the caraboid larvae could not simply be attributed to action of environmental factors present in the rearing medium. In his two papers, Barber made significant initial advances in knowledge of M. debilis. The only life stages he did not observe were eggs and larvae produced from an actual mating between normal males and females. The diagram which 4) used to illustrate the life cycle of M. debilis is reproduced (after 38) in Fig. 1.

Figure 1 Open in figure viewerPowerPoint . Diagrammatic representation of the life cycle of Micromalthus debilis LeConte according to 4 ) (after 38 )

Surely these two early papers, with their descriptions of the very strange life cycle of M. debilis, must have been met with some skepticism. Although Barber was correct on almost every point, 41) attributed the distrust of Barber's findings to the latter's lack of illustrative evidence. One critic who actually published his disagreement was 10), who never once actually studied M. debilis himself. Caillol's major points of contention were as follows (from 38): (1) the caraboid larva of Barber should properly be called a triungulin, as the hypermetamorphosis of M. debilis is basically similar in design to that of Meloidae; (2) the paedogenetic larva is actually a degenerate, wingless, parthenogenetic adult female. 38) presented a diagrammatic representation of Caillol's alternative hypotheses; this is given in Fig. 2.

Figure 2 Open in figure viewerPowerPoint . Diagrammatic representation of the life cycle of Micromalthus debilis LeConte according to 10 ) (after 38 )

38 ) agreed that the first instar larva of M. debilis was not truly caraboid, because it lacked the prominent urogomphi and typical carabid‐like antennae. He agreed that `triungulin' was a more appropriate name for this larval stage. However, on the second point of contention, Pringle did not agree with Caillol, and gave three arguments against calling the paedogenetic larva a neotenic female ( 38 : p. 274). Since, in M. debilis , the larvae of the next generation are already partially developed in the cerambycoid larva, the form which succeeds it must be a paedogenetic larva. Also, the mouthparts of the paedogenetic larva, although reduced, resemble closely those of the cerambycoid larva. Finally, there is no evidence of a pupal, or other resting stage between the cerambycoid and paedogenetic larva. In other groups of Coleoptera exhibiting neoteny or paedomorphosis such as Phengodidae and Lampyridae, a female pupal stage is present ( 14 ). Therefore, we agree with Pringle's assessment that the larviform reproductive stage of M. debilis is in fact a larva, and not a neotenic female in the sense of the same stage in several families of Cantharoidea. Although 38 ) paper added nothing new concerning the life cycle of M. debilis , relatively detailed descriptions and illustrations were given of the three `female' larvae and the pupa of M. debilis .

Perhaps the most important and illuminating examinations of Micromalthus were published by A. C. Scott in a series of papers from 1936 to 1941. No previous worker had actually examined the cytogenetics, spermatogenesis, or structure of the gonads of M. debilis when in fact, its life cycle is at least in part, merely a manifestation of these internal features.

40 ) discovered, and first documented, that males are haploid in their germ line cells throughout development whereas females remain diploid. From examination of cleavage nuclei in developing eggs, Scott found the diploid number of chromosomes to be 20. However, since the counts were taken early in cleavage, the possibility of chromosome elimination was not discounted. A detailed study of spermatogenesis revealed unipolar spindle fibres, and an abortive first division resulting in only two spermatozoa being produced from each primary spermatocyte. Based on these findings, 40 ) challenged several previously accepted theories about meiosis and spindle fibre formation.

The life cycle of M. debilis was most clearly documented by 41), who considered internal reproductive structures worthy of study. Five reproductive forms were named by 41: 635): (1) thelytokous (female‐producing) paedogenetic larva; (2) arrhenotokous (male‐producing) paedogenetic larva; (3) amphit[erot]okous (female‐ and/or male‐producing) paedogenetic larva; (4) adult female; and (5) adult male. The three types of paedogenetic larvae are impossible to distinguish until the middle of the penultimate instar. At this time, the thelytokous larva can be identified by the presence of elongating eggs and young embryos. The male‐producing larva is distinguished by its opaque white colour, and its more cylindrical shape with distinct abdominal segmentation.

4 ) thought that the paedogenetic larva was occasionally barren, i.e. had not given birth to any larvae. 41 ) discovered an amphitokous form; these larvae were dissected and found to be basically arrhenotokous, but not to have successfully shed their male eggs. When this male egg is aborted or otherwise fails to develop, thelytokous eggs are subsequently produced by such larvae. This may explain why 4 ) found four larvae without apparent embryos, in his isolation experiment described above. 41 ) mentioned that the ovaries of each type of paedogenetic larva are distinctive, but that the characteristics of each are attained only gradually. From the small, undifferentiated ovary of first stage female larvae, three outcomes are possible ( 41 : p. 642): (1) the small, possibly abortive, ovary of the adult female with only one or several eggs; (2) the ovary of the arrhenotokous larva with relatively larger eggs and robust pedicels; or (3) the large ovary of the thelytokous larva with cylindrical eggs. Each follicle of type 3 serves as a brood chamber for a single embryo. In all paedogenetic larvae, there is no uterus, vagina, or spermatheca ( 41 ). The adult female, however, has a more typical reproductive system with these three structures present.

The amphitokous larva held much interest for Scott and in 1941, he published a treatment of the various components of production of males in M. debilis, including the reversal from male to female production. Various questions were posed: (1) why is only one male produced from the arrhenotokous larva when its ovaries may contain several eggs? 41) indicated that of approximately 200 arrhenotokous larvae examined, 1% had one egg in both ovaries, 71% had two, 23% had three and only 4% had four eggs; (2) does the relative age and/or position of the male egg within the ovariole influence its chances of being used?; and (3) if the larva of the shed egg is prevented from eventually consuming its mother, can the arrhenotokous larva release another male egg?

42 ) could not answer the first question, and found no evidence to support his hypothesis that either age or position had an influence on an embryo's chances of being shed. It was thought that since a male larva's sole source of food is the paedogenetic mother, multiple births would cause competition for this resource. 42 ) envisioned a `physiological cooperation' between the shed male embryo and its mother. A possible mechanism was suggested to be a hormone, the release of which may be triggered by exit of the egg. This hormone would then cause the muscular contractions, which normally expel the eggs, to cease.

If the male embryo is removed from the female or does not complete development, a new brood of female larvae is formed, after about 4 weeks. The development and birthing processes of these reversed female larvae are exactly the same as those of larvae produced for ordinarily thelytokous paedogenetic larvae. The brood size of this reversed arrhenotokous larva is intermediate between that of a normal thelytokous and a normal arrhenotokous paedogenetic larva. 42) found that occasionally, amphitokous larvae produced new female larvae even though male embryos had not been shed. Therefore, the mechanical removal of the male embryo is not the reason for development of the new batch of female larvae. The oogonia which eventually contribute to development of female embryos are present in the amphitokous larvae before the male embryo is shed. These proliferate in undeveloped ovarioles which do not fully form during growth of the male eggs.