How small you can go: Factors limiting body miniaturization in winged insects with a review of the pantropical genus Discheramocephalus and description of six new species of the smallest beetles (Pterygota: Coleoptera: Ptiliidae)

The recently described and originally monotypic genus Discheramocephalus Johnson, 2007 from the Solomon Islands is revised. Six new species are described, illustrated and keyed: Discheramocephalus brucei sp. n. (Cameroon), D. elisabethae sp. n. (Cameroon), D. mikaeli sp. n. (Tanzania), D. stewarti sp. n. (Bolivia), D. jarmilae sp. n. (Bolivia), D. minutissimus sp. n. (Indonesia). Adults of D. minutissimus have a body length of about 400–426 μm, which is at the lower limit among non-eggparasitoid insects. Evidence is provided that an egg size large enough to produce a viable larva is the main factor limiting miniaturisation of female insects. Females and males of egg-parasitoids are able to overcome the 400 μm threshold and reach limits of 180 μm and 130 μm, respectively. Brain size is likely the second most important factor limiting miniaturisation in insects.


INTRODUCTION
Miniaturisation in multicellular animals, or evolution of extremely small body size, is a remarkable natural phenomenon. As they become smaller, organisms gain access to new niches, acquire new food sources and avoid predation. Miniaturisation is among the key reasons for extensive radiation in several evolutionary successful lineages where animals have less-than-the-average body size, e.g. shrews (Soricomorpha: Soricidae) in mammals, hummingbirds (Apodiformes: Trochilidae) in birds and featherwing beetles in insects. There are, however, limiting factors preventing organisms from becoming indefinitely small.
Physical body size is the most commonly used concept of "size" in eukaryote organisms. In multicellular metazoans there are at least two other parameters, both representing different concepts of what "size" might mean in biology. Metazoans are made up of cells that also have their own linear limits (Faisal et al., 2005). Genome size of a haploid nucleus (C-value) is yet another way of measuring "size" (Johnston et al., 2004;Gregory, 2007). There is apparently a positive correlation between genome and cell size (Gregory, 2001), and, likely, between body size and cell size (Rensch, 1948). Thus the true concept of "biological" size of a multicellular organism is a significantly more complex issue than just physical body size (Hanken & Wake, 1993).
Attention to animal miniaturisation has focused mainly on vertebrates (Alexander, 1996) or on individual insect organs such as the brain (Beutel et al., 2005). Different factors named as those limiting the process of miniaturisation in animals include disruption of thermoregulation or respiration, water loss, difficulty to fly, water surface tension (Novotny & Wilson, 1997), cell and tissue size, and egg or brain size (Polilov, 2005).
The goal of this paper is to revise the genus Discheramocephalus Johnson, 2007 of minute featherwing beetles (Ptiliidae; see Hall, 2005, for a recent overview of the family), whose species are among the smallest non-eggparasitoid insects known. Evidence is presented suggesting that egg size is the main factor limiting body miniaturization in female winged insects, while brain size might be the second most important factor. Other factors, such as relatively greater air viscosity and weight reduction of articulated appendages and their muscles, are briefly considered. Different biological scenarios related to the phenomenon of body size miniaturization in Pterygota are discussed.

MATERIAL AND METHODS
Beetles were extracted from forest floor leaf litter or collected by flight intercept traps (FIT). Leaf litter was pre-sifted to separate the fine fraction containing micro-arthropods, and then placed in Winkler or Berlese funnels to extract small invertebrates. Captured animals were preserved in 70% ethanol and subsequently sorted under a dissecting microscope. Beetles were later removed from ethanol, cleared in hot 10% KOH solution and mounted on microscope slides in Euparal. One specimen per species was glued by the dorsal side of its abdomen to the point of an entomological pin. This pin was inserted into a horizontally-oriented holder in a scanning electron microscope and rotated at 360 degrees around its axis for different views of the specimen. Two specimens of D. brucei were dehydrated in gradient solutions of acetone and embedded in epoxy resin for serial sectioning at 1 µm on a Leica RM 2165 rotation microtome and stained with methylene blue and acid fuchsin. Transmitted light microscopy images were taken on a compound microscope with attached Nikon digital camera; those using scanning electron microscopy (SEM) were taken of uncoated specimens with a Philips XL30 ESEM. SEM specimens were measured using their images with a superimposed scale generated by the attached computer; specimens on microscope slides were projected through a camera lucida on a previously projected image of a 1000 µm scale subdivided into one hundred 10 µm-long increments. Images of free or whole-mounted beetles were captured by a digital camera and then assembled in Photoshop onto a single plate. Some images were captured at different focal depths to create a combined image with the maximum depth of focus using CombineZ5 software (Hadley, 2006).
All specimens dealt with in this work, including types of the new species (unless otherwise stated), are stored in the Canadian National Collection of Insects, Arachnids and Nematodes, Ottawa, Canada (CNCI). Most of the terms used in this paper are those generally adopted for Ptiliidae (see, for example, Sörensson, 1997). Terms "mesoventer" and "metaventer" are used, however, for the misapplied terms "mesosternum" and "metasternum" (as explained in Lawrence, 1999); "meso"-and "metasternal lines" of Ptiliidae are therefore called "meso"-and "metaventral" lines. Diagnosis. Members of this genus are easily distinguished from all other Ptiliidae by having a deep transverse groove on head behind eyes, which crosses dorsal surface of head (Figs 4,20), and extends laterally to ventral surface (Figs 9,22). Possession of a transverse row of two or more deep cavities on sternite VIII is unique among Ptiliidae (Figs 13,(92)(93)(94)(95)(96)(97)(98). Species of Discheramocephalus also have the following unique character combination: mesoventral keel with deep, horizontally oriented fossa on each side (Fig. 10); mesoventer with deep and vertically oriented fossa on each anterior corner (Fig. 11); pronotum with at least two deep longitudinal grooves (Figs 1,16,25,37,46,55); metacoxae transverse and contiguous (Fig. 3).

Genus
Description (these characters, including "absent" characters, are standardized to the format used in generic descriptions in a more inclusive taxonomic treatment of a new tribe of Ptiliidae, which includes the genus Discheramocephalus; see Grebennikov, 2008). Body between pronotum and elytra in dorsal view constricted; body behind pronotum not swollen laterally and vertically; longitudinally oriented micro-ridges on ventral surface of prothorax present; elytral setae forming seven or eight longitudinal rows; transversely oriented depression behind eyes present as groove extending laterally and ventrally; transverse band of 100-200 small punctures on head behind eyes absent; apical antennomere not dumbbell-shaped; longitudinal depressions on labrum absent or present; posterior edge of pronotum at middle concave; depressions on pronotum present as grooves at least half as long as pronotum or ornamented with varying number of longitudinal furrows (type species only); longitudinal keel on scutellum present, sharp, about as long as scutellum; transverse row of round punctures on base of elytra and scutellum absent; two fossae on scutellum absent; hind wings present; meso-metaventral suture between mesocoxae not visible externally; mesometaventral suture laterally of mesocoxae not visible externally, present as internal thickening of cuticle; posteriorly pointed serration along meso-metaventral suture laterally of mesocoxae absent; metaventral longitudinal lateral lines absent; horizontal perforation of mesoventral keel round and transparent in lateral view; mesoventer without transverse grooves; middle of alacrista of metathorax without setae along margins; metacoxae transverse, contiguous, separated by about 1/15 of metaventral width; posteriorly oriented projection of metaventral plate between metacoxae without two lateral teeth; transversely oriented group of about 50-70 closely adjacent round micropores along posterior edge of tergite VIII absent; cavities on abdominal sternum VIII present; single elongate internal sclerite alongside aedeagus present (Figs 75,(93)(94)(95)(96)(97)(98); spermatheca weakly developed and nonsclerotised except for sperm pump (Fig. 99), markedly similar in shape in all species.
Composition and geographical distribution. Johnson (2007) described this genus based on a new species from the Solomon Islands and mentiond that "…undescribed species are known to me from Sri Lanka, Madagascar, Congo and Brazil…". Dybas (1980) illustrated an undescribed congeneric species from Florida and referred to this (then undescribed) genus as "widespread tropical". I saw undescribed species of Discheramocephalus from Australia (Queensland) and Madagascar. The six new species are from Bolivia, Cameroon, Tanzania and Indonesia. These data suggest that Discheramocephalus is a speciose and pantropical genus.
Monophyly and phylogenetic relationships. The genus Discheramocephalus is most likely a monophyletic group because both the deep transverse groove on head behind eyes (Fig. 9) and the transverse row of two or more deep cavities on sternite VIII (Figs 93-98) are unique among Ptiliidae and not known to occur in its sister-family Hydraenidae. Johnson (2007) suggested that the genus Millidium Motschulsky, 1855 might be its closest relative. Our phylogenetic analysis (Grebennikov, 2008) suggests that among known Ptiliidae taxa, Discheramocephalus is most closely related to, or derives from within, the genus Skidmorella Johnson, 1971 from the tropical and subtropical parts of the Oriental zoogeographical region. These two taxa are hypothesised to form a monophyletic group supported by presumably synapomorphic presence of deep longitudinal furrows on pronotal disk, although superficially similar and likely nonhomologous pronotal grooves also occur in other Ptiliidae, such as Millidium Motschulsky, 1855 and some Ptilium Gyllenhal, 1827.
Bionomics. All specimens of Discheramocephalus examined were collected in tropical forests either by flight intercept traps or by sifting leaf litter. Diagnosis. Body length 0.45-0.48 mm (n = 4); transverse postocellar groove on head apparently closed laterally, bridged by posterior extension of eye joining temple; posterior part of pronotal disk with three short longitudinal furrows on each side adjacent to midline and not longer than 40% of pronotal length (Johnson, 2007: Fig. 23); each side of abdominal sternite VIII with small cavity about 10% of sternal width; metaventral longitudinal lines absent; posterior serration on pygidium absent; spermatheca was not described. Male unknown.
Material. No specimens were studied; description is adapted from Johnson (2007). Holotype (female) and three paratypes (females) are deposited in the Natural History Museum, London.   598´, 6.v.2006, 2,200 m, V. Grebennikov; six of them in 70% ethanol and two others are mounted together in Euparal on one microscope slide. Two more specimens from the same locality (Mt. and Lake Oku) were sectioned on a microtome for histological studies and not designated as paratypes.

Remarks.
A single specimen of a Discheramocephalus species closely resembling, or possibly conspecific with, D. brucei was collected in the Korup National Park, Cameroon (Fig. 70). It differs from D. brucei, however, in shorter body, differently shaped abdomen (in dorsal view), and pronotum having obtuse angulation at the middle of lateral edges (Fig. 70). Whereas all other African Discheramocephalus species were sifted out of leaf litter in the mountain forests at altitudes above 1,000 meters, this single specimen was taken by a flight intercept trap in the lowland forest at 300 meters above sea level. This is the only record of Discheramocephalus from the Korup National Park, although extensive sifting of the leaf litter was undertaken there. Additional material is needed to clarify the identity of this population. The specimen is stored in 75% ethanol and bears the following label data: "Cameroon: S.-West province, Korup N.P., Rengo Camp, N05°02.194´ E008°49. 769´, 12.-16.v.2006, 300 m, V. Grebennikov leg". Remarks. Aedeagi of D. elisabethae males are remarkably long and amount to about 1/3 (32%) of the body length (Fig. 75). This relative length might be among the largest in Coleoptera. each side adjacent to midline and not shorter than 90% of pronotal length, flanked laterally by one longitudinal basal depression; tarsi not broad, about 15× as long as wide; longitudinal ridges of microsculpture on mesoventer absent or weakly developed; abdominal sternite VIII with transverse row of 5-7 small cavities each not wider than 5% of sternal width, two most lateral of them about twice width of medial ones; metaventral longitudinal lines absent; transverse rows of cuticular serration on abdominal sternites present; posterior serration on pygidium absent; aedeagus Fig. 95; spermatheca as in Fig. 99 Diagnosis. Body length 0.84-0.88 mm (n = 2); transverse postocellar groove on head closed laterally and fully bridged by posterior extension of eye joining temple; longitudinal grooves on labrum absent; disc of pronotum with one longitudinal furrow on each side adjacent to midline and not shorter than 90% of pronotal length, flanked laterally by two longitudinal basal depressions; tarsi broad, about 6-7× as long as wide; longitudinal ridges of microsculpture present on mesoventer; abdominal sternite VIII on each side with one large cavity  Type material. Holotype (%) mounted in Euparal on a microscope slide: Bolivia, Cochabamba Department, Villa Tunari, Hotel El Puente, S16°59.02´ W65°24. 50´, 357 m, 15.-27.xii.2005, rainforest FIT, #05-45, S. & J. Peck. Paratypes: Eight specimens with the same data as holotype; two in separate vials with 70% ethanol (one of them has been used for SEM images) and six more mounted on five Euparal microscope slides. Etymology. The specific epithet is the Latin adjective minutus, -a, -um (little, small, minute) in superlative form and refers to the minute body size of this species.

DISCUSSION ON INSECT MINIATURIZATION
It is difficult to propose a universally applicable measurement for determining the absolute size of animals. For practical purposes, body length is often used, measured along the longest axis of a specimen's body and excluding appendages or other body extensions, like the ovipositor, which do not significantly contribute to body volume. Body size has the major advantage of being relatively easy to measure and to compare. Still, there are some assumptions and approximations routinely used when dealing with this parameter. Body volume (or weight) might be the single best parameter to compare, although it is subject to extreme modifications even within a single organism depending on how dehydrated, starved, etc. it is. Thus, body length is normally the most convenient measurement to use when deciding which is smaller or larger. Immature stages are generally smaller in size than adults and, therefore, it is customary to use the maximum body length of a fully grown and sexually mature specimen. Even this simplified measurement poses considerable difficulties, because animals with telescopic or jointed body parts are subject to shrinking or bending, which affects the accuracy of measurement.
The widely cited body length of 260 µm for the smallest beetle (Crowson, 1981) is a measurement error . 101 -longitudinal section through the whole beetle; a -spermatheca; bmaturing egg; c -horizontally oriented fossa through mesoventral keel; d -foregut merging with midgut. 102 -longitudinal section through posterior part of thorax and abdomen; e -Malpighian tube; f -metendosternite; g -maturing egg; h, i -coil of midgut; jwing; k -metathoracic ganglion; l -mesocoxa. 103 -cross section through the head; m -brain (cephalic ganglion); n -foregut; ostipes; p -transverse external groove on head; q -eye. 104 -cross section through thorax and abdomen between meso-and metacoxae; r -maturing egg; s -elytral interlocking device; t -coil of midgut; u -basal part of wing; v -spiracle. Scale bars: 100 µm. 105-106: Dwarf and likely blind male with body length 280 µm (left, indicated by arrow) and fully developed much larger female (right) of Errolium sp. (Hymenoptera: Platygastridae) inside host egg (likely Inglisia sp., Coccidae; see Masner & Huggert, 1989) under low (105) and high (106) magnification. Note that the larger female does not have fully developed wings, which expand later after emergence from the host egg. recently corrected by Sörensson (1997). This author gave the real minimal body length for beetles as about 400 µm, although later Hall (1999) cited 300-350 µm for some of his new species of Nanosellini (Ptiliidae). The newly described Discheramocephalus minutissimus further suggests that 350-400 µm might be close to the true lower limit of body length in Coleoptera.
There is no evidence that extreme miniaturization of adult and larval body size of featherwing beetles is achieved through significant morphological reduction and structural simplification, as exemplified in some hymenopteran egg-parasitoids (Mockford, 1997) and suggested as a general trend for all minute animals (Hanken & Wake, 1993). Discheramocephalus species described herein do not demonstrate significant internal or external simplification compared to medium-sized beetles. There is only some reduction, such as the presence of only one ovary (Polilov, 2005) and complete lack of a heart (Fig. 104). Minute larvae of Ptiliidae are known to possess the normal chaetotaxy and musculature of the head despite their extremely small size (Grebennikov & Beutel, 2002).
Discheramocephalus minutissimus, along with some other minute species of featherwing beetles, has a body length of about 400-420 µm and, therefore, is among the smallest winged insects. Besides Ptiliidae, only a few groups of hymenopteran egg-parasitoids, accounting for less than 1% of insect species diversity, are known to reach, and even to exceed, this size threshold, and are the smallest insects known. No species from groups of winged insects other than Ptiliidae and egg-parasitic Hymenoptera is reliably known to be as small as 350-400 µm in body length in the adult stage.
Two facts related to the process of insect body miniaturization are of general biological interest. One is that the size of about 400 µm seems to be the lower limit of body length for free-living insects, as exemplified by featherwing beetles. This suggests that there is a lower limit to further miniaturization. Another fact is that hymenopteran egg-parasitoids are able to overcome this threshold and become significantly smaller. Either eggparasitoids might bypass the limit to miniaturization in non-egg-parasitoid insects or, alternatively, this main limitation acts differently in egg-parasitoids compared to other insects. Polilov (2005Polilov ( , 2007 hypothesized that the main constraint for the process of body miniaturization in adult insects is the limit imposed by the necessity to produce eggs large enough to permit development of a viable larva. Three implications result from this hypothesis. Implication 1: Selective pressure on minute organisms will be directed towards supporting females laying fewer eggs per time unit, which ultimately results in only one egg maturing in a female at one time. A by-product of this implication is that only a single ovary per female might be expected to exist in the smallest organisms. This is indeed true in Ptiliidae and in many other insect (and noninsect) groups with extremely small species, like eriophyoid mites (Nuzzaci & Alberti, 1996). Implication 2: Due to their unique biology hymenopteran egg-parasitoids might be exceptionally capable of miniaturization. The reason is that eggs of egg-parasitoids are laid directly into eggs of host species and, therefore, their newly hatched larvae have a unique opportunity to utilize the stored resources of the host egg. Thus, the egg of an egg-parasitoid can afford to be markedly less supported with resources than an egg of any other insect. This permits the egg-parasitoid female to achieve a smaller size compared to insects with free-living larvae. Many other insects have larvae that develop inside a plant or animal host and thus are surrounded by an abundance of food; however, none of them are able to match eggparasitoids in body miniaturization. The reasons for this may be that the egg is particularly rich in resources easily metabolized by the newly hatched parasitoid larva compared to the tissues of hosts in later developmental stages. Another explanation is a purely physical constraint: the maximum body size of an egg-parasitoid adult is determined by the size of the host egg, e.g. very small eggs of Psocoptera or Thysanoptera can only yield a very small adult parasitoid (Doutt & Viggiani, 1968).
Implication 3: The hypothesis of a limit to female miniaturization leaves males not accounted for. This suggests that males could have a greater ability to become smaller then females. The reason is that the female has to produce an egg, whereas the male has to produce only fertile sperm, which does not normally require as much resources as an egg. This is in line with Rensche's rule that females of smaller animals tend to be larger than males, whereas in larger animals the opposite is true, as convincingly illustrated for flower mites and humming birds (Colwell, 2000); two groups holding records of body miniaturization in Acarina and Aves, respectively.
The most convincing illustration that egg size limits female insect body miniaturization is the smallest known insect, Dicopomorpha echmepterygis Mockford, 1997, a mymarid egg-parasitoid (Mockford, 1997). Females of this species are 386-550 µm in body length, winged, fully functional, and are known to produce a single egg at a time, 106-110 µm in length. Males, however, are minute, blind, wingless and 139-240 µm in length, which is the record for small body length in the Insecta (Mockford, 1997). It should be noted that having half the body length of the female, the body volume of males is an eighth of that of females.
This phenomenon of dwarf and structurally simplified males is relatively widespread in minute Hymenoptera. In Trichogrammatidae, Prestwichia aquatica Lubbock, 1863 has apterous males (Enock, 1896), which copulate with females inside the host egg (Enock, 1898), while its freshly hatched larva is a minute non-segmented creature lacking all major tissue systems but for a few muscles used to feed on a host egg of Dytiscidae (Coleoptera) (Ivanova-Kazas, 1950, 1962. Masner & Huggert (1989) cited E. Valentine's observation that a dwarf, apterous and blind male was observed inside the same host egg skin as was a female of an undescribed Errolium species (Platygastridae) in New Zealand . Unlike the degenerate blind and apterous males of D. echmepterygis (Mymaridae), Errolium males are known in both degenerate and fully developed form (E. Valentine in Masner & Huggert, 1989). Similar blind, apterous and morphologically simplified (although not dwarfed) males can be found in Melittobia Westwood, 1847 (Hymenoptera: Chalcidoidea: Eulophidae), gregarious parasites of Aculeata prepupae (Dahms, 1984). Adults of this genus are not extremely small, ranging between 1.0 and 1.5 mm in length. This example demonstrates the existence of other factors, besides miniaturization, which might lead to morphological simplification of males in parasitic Hymenoptera.
Polilov (2005,2007) hypothesized that brain size might be the second most important factor limiting insect miniaturization. Neuron cell body size reaches a lower limit of 2 µm in diameter in the smallest insects, which is at the lowest recorded limit in multicellular animals (Beutel et al., 2005;Polilov, 2005). The smallest brains still have to maintain all circuits of sensory input necessary for the organism to function (Kaas, 2000). This is particularly true for the adult stage compared to larvae, because adults may need a larger brain to handle more complex behaviour and associated necessary sensory input. Another restricting mechanism might be the necessity to maintain a certain proportion between neuron body size and axon diameter. The latter have to be greater than 0.1 µm in diameter to permit transmission of a signal (Faisal et al., 2005). The third limiting mechanism might be genome size, which limits volume of the nucleus by filling it up with chromatin. The nucleus occupies 80-90% of a neuron cell body in Ptiliidae (Polilov, 2005), thus genome size, brain size and linear body size are possibly linked together as correlated parameters.
Any object moving through air is subject to the laws of fluid mechanics. The hypothesis of Horridge (1956) that small insects "have abandoned altogether the aerofoil action and that they literally 'swim' in the air" is currently not supported (Ellington, 1999). As insects become smaller, the viscosity of the air increases proportionally, which leads to a higher energy cost for active flight. This physical phenomenon might be another factor partly limiting insect miniaturization. Existence of this flightimposed constraint is partly suggested by the fact that smallest flying insects like Ptiliidae or Mymaridae have a similar, light-weight wing structure with the wing membrane markedly reduced in width and substituted by a fringe of trichia, which increase flapping surface and lift. Wing weight reduction is likely an aspect of a small arthropod's structural need to reduce weight of articulated appendages and associated muscles.
Some non-flying arthropods are markedly smaller then the smallest Pterygota. Eriophyoid mites, for example, have an average adult body length ca. 200 µm, and range from 80 µm to 500 µm (Lindquist, 1996). The smallest eriophyoid mites normally only have a single egg per female at any one time (Gondim & Morales, 2003) and their males are smaller than the females (Colwell, 2000). These mites, however, are not egg-parasitoids but freeliving organisms that feed on plant tissue. There is no convincing explanation why these organisms achieve a level of body miniaturization not matched by insects, although it is plausible that the markedly different structure of the brain and the simplicity of their behaviour (Polilov, 2005) along with reduction of locomotory muscles due to passive wind dispersal and lack of need to move around on their host plants might account for this phenomenon.