Austroconops Wirth and Lee, a Lower Cretaceous Genus of Biting Midges Yet Living in Western Australia: a New Species, First Description of the Immatures and Discussion of Their Biology and Phylogeny (Diptera: Ceratopogonidae)

Abstract The eggs and all four larval instars of Austroconops mcmillani Wirth and Lee and A. annettae Borkent, new species, are described. The pupa of A. mcmillani is also described. Life cycles and details of behavior of each life stage are reported, including feeding by the aquatic larvae on microscopic organisms in very wet soil/detritus, larval locomotion, female adult biting habits on humans and kangaroos, and male adult swarming. Austroconops annettae Borkent, new species, is attributed to the first author. Cladistic analysis shows that the two extant Austroconops Wirth and Lee species are sister species. Increasingly older fossil species of Austroconops represent increasingly earlier lineages. Among extant lineages, Austroconops is the sister group of Leptoconops Skuse, and together they form the sister group of all other Ceratopogonidae. Dasyhelea Kieffer is the sister group of Forcipomyia Meigen + Atrichopogon Kieffer, and together they form the sister group of the Ceratopogoninae. Forcipomyia has no synapomorphies and may be paraphyletic in relation to Atrichopogon. Austroconops is morphologically conservative (possesses many plesiomorphic features) in each life stage and this allows for interpretation of a number of features within Ceratopogonidae and other Culicomorpha. A new interpretation of Cretaceous fossil lineages shows that Austroconops, Leptoconops, Minyohelea Borkent, Jordanoconops Szadziewksi, Archiaustroconops Szadziewksi, and Fossileptoconops Szadziewksi form a monophyletic group. Within this assemblage Leptoconops and Minyohelea are sister groups and Austroconops and Jordanoconops are monophyletic (Austroconops is possibly paraphyletic in relation to Jordanoconops). All are considered to be members of Leptoconopinae Noè and the subfamily Austroconopinae Borkent, Wirth and Dyce is a new synonym of Leptoconopinae. Extant and fossil distributional records suggest that Austroconops was displaced from its previously broad distribution by the emergence of Culicoides Latreille. Larval feeding on microorganisms is plesiotypic within the Ceratopogonidae and Chironomoidea, and living in small aquatic habitats is plesiotypic within each family of the Culicomorpha. Outgroup comparisons further suggest that diurnal feeding by adult females is plesiotypic within the Ceratopogonidae and Chironomoidea.


INTRODUCTION
Perhaps only fellow entomologists could appreciate the thrill of the following sadomasochistic scenario: the first author and his wife standing still on the edge of a golf course in Yanchep National Park, just north of Perth in Western Australia, being badly bitten on the face by hundreds of biting midges, with exclamations of joy and wonder. We realized that we were being attacked by a species considered to be an extremely rare "living fossil", a member of a genus which was surviving only in this small corner of the globe but which was once a widespread and diverse lineage during the Early Cretaceous, 121 million years ago. Previously known from only a limited series of adult specimens of a single species and a moderate number of fossils, our study of this genus uncovered a wealth of new information: the discovery of the previously unknown eggs, larvae, and pupa; the presence of a second extant species (including its immatures); and observations of female biting behavior, male swarming, and more.
There are good reasons why we were excited to pursue these small beasts, and we committed ourselves to an extensive expedition to Western Australia to seek them out. In recent years cladistic analyses and an abundance of fossil material in amber of different ages have resulted in an increasingly detailed understanding of the phylogenetic relationships of the biting midges. The genus Austroconops Wirth and Lee plays a pivotal role in this analysis. Together with Leptoconops Skuse, these two ancient genera represent the earliest extant lineage in the family. However, unlike the rather modified mem-bers of Leptoconops, species of Austroconops have a number of plesiomorphic features, allowing for an enhanced interpretation of character states within the family. Simply put, Austroconops is a group of primitive "living fossils''. Further to this, knowing that the Early Cretaceous species of Austroconops were morphologically similar to the extant species suggested that study of the biology of the living species would provide special insight into the adaptations of Ceratopogonidae existing 121 million years ago. Of particular interest and importance was the discovery of the eggs and larvae of the two extant species, as well as the pupa of one of these. One of the reasons that nematocerous Diptera are such excellent candidates for phylogenetic studies is that the different life stages show a remarkable degree of morphological divergence and therefore provide independent sources of character states to test phylogenetic hypotheses.
When Austroconops and its single species A. mcmillani was first described by Wirth and Lee (1958) only 39 females were known from four localities in the vicinity of Perth, Western Australia. These authors noted that the species seemed closest to Culicoides Latrielle, well within the Ceratopogonidae, but mentioned that it shared some features with the early lineage Leptoconops. They were uncertain as to which subfamily the genus belonged. The only information on the species' biology was that females were recorded as biting humans, particularly on the eyelids. Then, in 1985, W.W. Wirth and A.L. Dyce collected the first males of the species, and study of these three specimens (plus an additional 10 females) led to a cladistic analysis of the early lineages of the family, demon- 2004 strating that Austroconops belonged near the very base of the phylogeny of the Ceratopogonidae (Borkent et al., 1987). Since then, six fossil species have been described from Early to Late Cretaceous amber (from Lebanon, France, and Siberia), providing, within a cladistic analysis, further support that the genus indeed represents an early lineage of biting midge (Borkent, 2000a). Within the Ceratopogonidae, there is a striking correlation between cladistic patterns and the distribution of fossils in different ages of amber, with increasingly older fossils representing increasingly earlier lineages (Borkent, 2000a). Our study here further refines the phylogenetic position of Austroconops, interprets characters states of the newly discovered immatures, analyzes the implications for our interpretation of features of other Ceratopogonidae, and provides an interpretation of historical zoogeographical and bionomic features of early lineages of Ceratopogonidae and other Culicomorpha.

MATERIALS AND METHODS
The collecting trip to Western Australia was undertaken by the first author and his wife with the purpose of discovering further material of Austroconops. Sampling was primarily with a sweep net in a wide array of wet or aquatic habitats from October 12 to November 23, 2001, covering the area from Yanchep N.P., about 47 km north of Perth, south to Augusta, east to Albany, returning through Darkan to Augusta and then heading north to Yanchep N.P. once again. Mud samples were taken from wet habitats near the sites of where adult Austroconops were collected and immatures searched for (unsuccessfully) either manually or by floating debris and substrate using a saturated sugar solution. Female adult A. mcmillani were collected with a net and aspirator by Len Zamudio during December, 2001-January, 2002and October, 2002-February, 2003 while the midges were attacking humans in Yanchep National Park.
To obtain eggs of A. annettae, an adult female was captured by sweeping and placed directly into a vial with mud/detritus (about 1 cm deep) taken from the spot where the adults of this species had been collected. BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 3 Subsequent care was as described below for A. mcmillani. Females of A. mcmillani were allowed to attack the face at site A (see fig.  22C), complete their feeding, and then were captured either by placing a small vial over the female until she finished feeding or by sticking one's head in an insect net and gently aspirating the fully fed females after they left the skin surface and flew into the net. These captured females of A. mcmillani were then cooled briefly in a refrigerator and when torpid, quickly placed in six groups of 7-12 females each and put into separate 2-or 3dram vials containing about 1 cm of wet mud/detritus taken from the partially inundated margin at the south end of Loch McNess (fig. 22C;site B). Females of both species received a small smear of honey on the inside surface of the vial at about midheight (none were seen to feed) and each vial was stoppered with a cotton ball. A drop of water was added periodically to the mud when it began to dry (so the mud remained quite wet). The vials were kept at more-orless ambient temperature in a small cooler that was always kept in the shade. Females of A. mcmillani were transported from Australia to Canada with the addition of a large bottle of very warm water to the cooler to keep adults from becoming torpid and sticking to the mud surface. The eggs of A. annettae, which had been laid while in Australia, plus first instar larvae were transported to Canada in the same cooler. In Canada the eggs and larvae of A. annettae and the female adults, eggs and larvae of A. mcmillani were kept at about 23-28°C during the day and about 18°C at night. Most of the females of A. mcmillani that were transported died en route, probably because of excessive shaking of the vials during air travel; of 58 original females, only 12 survived the trip to Canada. Of these, at least one of three females laid eggs.
Eggs were allowed to develop in the vials, and after all had hatched, the mud contents with the larvae of each species was transferred to small (SO X 7 mm) plastic petri dishes with lids. The larvae could be easily observed moving or lying against the bottom of the dish by inverting the clear dish under the dissecting microscope. Vials and dishes were examined at least twice a day for the -AMERICAN MUSEUM NOVITATES great majority of larval and pupal development.
The mud/detritus substrate initially contained some small, unidentified microorganisms, but nearly all died shortly after obtaining the samples. When it became clear that the first-instar larvae were not feeding, between hatching and December 8, 2001, the following items were added as potential food at different times to the Petri dish with larvae of A. mcmillani-Euglena sp., yeast, a freshly killed ostracod and Chaoborus Lichtenstein larva, several species of unidentified green algae and cyanobacteria (from local fish tanks and ponds), dried blue-green algae (Spirulina) soaked in water, and the following cultured cyanobacteria and green algae: Anabaena flosaquae, Oscillatoria sp., and Selenastrum capricornutum. An unidentified species of rotifer was added between December 31, 2001 andJanuary 31, 2002. A small amount of water was added to agar plates with a culture of the nematode Caenorhabditis elegans and the bacterium Escherichia coli, swirled around, and 2-5 drops of the water with C. elegans were added every 2--3 days to the mud/detritus in the Petri dishes with A. mcmillani and A. annettae starting on December 31. These nematodes died within 12-24 hours. Beginning on December 17, 2-4 drops of an infusion made from fresh goat feces and water (first made on December 15) were added every 2-3 days to both dishes; these drops regularly included a small clot of feces. Water was added to the infusion to replace loss through evaporation (the infusion was not sealed), and a fresh infusion was made on August 9, 2002.
Five first-instar larvae of A. mecmillani were added to an agar plate with a culture of C. elegans and E. coli on December 14 and were seen feeding on clumps of E. coli, but all died within a few days. At least two of these larvae had their guts partially filled with white material, almost certainly E. coli. On June 28, 2002, three third-instar A. mcemillani were placed in a separate dish with some of the original mud/detritus, put into a dark cold room for four days at 13°C, and then placed in a refrigerator at 5-8°C. By August | all larvae had died and were preserved in 70% ethanol.
Some living larvae of different instars NO. 3449 were examined under both dissecting and compound microscopes by placing them in a well slide in which they were studied while warm and after they were cooled by ice. Cold larvae were examined for mouthpart and proleg movement. Park. Biting times by female A. mcmillani were determined by observing a female land, waiting for initiation of feeding, and then carefully inverting a vial over the female until it ceased feeding, when it invariably flew up into the vial.
Specimens were examined, measured, and drawn using a Wild M3 dissecting microscope and a Zeiss Jenaval compound microscope. Larval head capsule length was measured from the posterolateral margin to the anterolateral margin of the head capsule (near the base of the mandible). A male and female of both species were cleared and compared in temporary glycerine slide mounts. These and other adults, as well as most eggs and larvae were mounted on microscope slides using the technique described by Borkent and Bissett (1990). Larvae to be studied with the scanning electron microscope were killed with hot water. Some eggs and larvae were critical point dried, sputtercoated at 30 angstroms with chrome, and studied with a Jeol JSM 6301 FXV scanning electron microscope.
Terms for structures follows those used in the Manual of Nearctic Diptera (McAlpine, 1981). More specific larval and pupal terms follow Lawson (1951). The kangaroos referred to in the text are all western grey kangaroos (Macropus fuliginosus). The name Austroconops annettae is attributed solely to the first author.
Additional material was borrowed from the following museums: mere 12 with or without subbasal constriction, flagellomere 13 with subbasal constriction, flagellomere 1 with two patches of short sensilla trichodea, without sensilla coeloconica. Mouthparts moderately short to moderately long. Mandible elongate (ending near apex of labrum), with several slender terminal spicules. Lacinia elongate, simple. Palpus ( fig. 1C, D) with 4-5 segments, segment 3 ovoid to elongate, slender, with capitate sensilla scattered on mesal surface or perhaps in pit, at least extant species with membranous area between segment 3 and 4 + 5. Thorax Legs: Femora, tibiae slender. Legs lacking armature except in some with pair of stout setae on apex of tarsomeres 1-4 of all legs. TR of foreleg/hindleg = 1.7-2.2. Fore-and midleg trochanter without pair of thick setae. Midleg tibia with or without apical spur. Hindtibia apex with two rows of spines. Hindleg first tarsomere with or without thick basal spine, with scattered slender or stout setae. Claws on fore-, mid-, hindleg equal in size, both equal on each leg, each claw simple or toothed ( fig. 11, J). Empodium slender. Genitalia: Two large, one markedly smaller spermathecae. Sternite 9 continuous medially. Segment 10 with pair of setae. Cerci short to moderately elongate.
Egg ( fig. 2A): Very light yellowish brown, appearing nearly white during development of larva. Elongate and slender. With 9-10 longitudinal rows of short, minute tubercles, each tubercle slender at base, expanded distally ( fig. 2B). Anterior end more rounded than posterior end. Eggshell opening a single dorsal slit along most of length.
Mandible (figs. 2I, 3E, 10A) curved, apical half tapering to sharp, darkly pigmented, point, with dorsal and ventral grooves; with large to very small lobe (not visible in some specimens) on inner surface; with subbasal seta and minute peg in pit on outer margin; with well developed apodeme attached to dorsal margin of mandible, extending into posterior half of head capsule. Maxilla (figs. 10B, 16B, 17D) well developed, wide; palpus elongate; large lobelike, apically tapering plate dorsal to base of palpus; with posteromedial elongate seta; palpus with 4 apical or subapical sensilla, two lateral sensilla each with elongate slender nib; group of 7 thick sensilla lateral to palpus: sensillum 1 extending to about half length of palpus; sensillum rounded apex; lateral seta elongate. Pharyngeal complex (figs. 2J, 3F) well developed, epipharynx with two lateral arms, apparently lacking combs (present but difficult to discern in some other Ceratopogonidae). Hypopharynx with lateral arms articulating with lateral apices of lateral arms of epipharynx. Thorax, abdomen: Cuticle unpigmented, Abdominal segment 9 with uncertain number of setae; dorsal seta o at least slightly thicker than other setae on segment. Hemolymph unpigmented or very pale pink.
Third-instar larva: The only species in which females have been observed to bite are those of A. mcmillani, which feed on kangaroos and humans. However, the finely serrate mandible and retrorse lacinial teeth of the adult females of A. annettae strongly indicate that these too feed on vertebrates (Borkent, 1995: 129-132). The morphology of the claws of the female of A. annettae additionally may indicate that these feed on birds (see discussion below under that species). The mouthparts of fossil Austroconops are not visible (Borkent, 2000a), and therefore it is uncertain what they fed upon. The presence, however, of vertebrate blood-feeding in A. mcmillani and the phylogenetic position of the genus as an early lineage within the family where vertebrate blood-feeding is plesiotypic strongly suggest that all extinct species fed on vertebrates (Borkent, 1995(Borkent, , 2000a. Details of male swarming in A. mcmillani are given below. First-instar larvae of A. annettae and A. mcemillani were both successfully reared to fourth-instar larvae (and A. mcmillani to the pupal stage) in very wet soil, with regular (generally every second day) additions of nematodes and a fecal infusion. All instars of the aquatic larvae were clearly attracted to fresh drops of fecal infusion as shown by the concentrations of larvae directly under these drops and they were rarely seen to feed on nematodes. This phenomenon, and the presence of well developed, finely toothed scopae (figs. 3A, 19) suggest that they are likely associated with feces or concentrated decomposing vegetation in nature (producing an abundance of microorganisms). Some second, nearly all third, and all fourth-instar larvae had pink or red hemolymph, indicating the presence of hemoglobin, which additionally suggests that they may be associated with an oxygen deficient wet habitat. Larvae cannot swim but use a combination of their anterior and posterior prolegs and, especially in later instars, a relatively slow serpentine body motion to move through wet substrate. Further details of behavior and habitat are provided below for each species. Borkent et al. (1987) suggested that the blue-green pigmentation of live adult A. NO. 3449 mcmillani may have indicated that the then unknown larvae were feeding on algae because this coloration has been observed in some Ceratopogonidae such as some Culicoides species of the schulzei species group and some Dasyhelea Kieffer species which feed on algae as larvae. Our observations here show that the relationship between algal feeding and adult color is not substantiated for at least A. mcmillani as larvae matured to adulthood without feeding on algae. Austroconops annettae was also successfully reared to the fourth-instar without algae.

TAXONOMIC DISCUSSION
We consider the elongate, pale egg of Austroconops species unique within the family. Eggs laid by other Ceratopogonidae are initially pale but soon turn dark. However, of 103 extant genera, eggs have been described for only 18 genera and of those, 4 genera are known only as eggs described from within the female abdomen (so their final color cannot be determined).
We identified the four larval instars of both species based on the following evidence. First-instar larvae are easily identified by their darkly pigmented egg burster (figs. 2H, 7A, B, 8B). Because of the small number of measured second and fourth-instars of reared Austroconops larvae it was initially difficult to identify the instars 2-4. The fourth-instar of A. mcmillani was confidently identified because one of these molted to the pupal stage. Although the head capsule of this individual was not retrieved from the mud and therefore not measured, observations before it pupated showed that it was clearly close to, or within, the range of measurements reported here. Head capsule length increments of all instars followed Dyar's Law, increasing by a factor of 1.32, 1.31, and 1.34 for A. mcmillani and 1.33, 1.32, and 1.25 for A. annettae. The latter value, the factor of change for third to fourth-instar, may be due to the small sample size of fourth-instar A. annettae or perhaps to less than optimal feeding conditions (and hence smaller individuals). These values are similar to those reported for species of Culicoides, which are about 1.4 (Kettle and Lawson, 1952;Kettle and Elson, 1975).
There are significant statistical differences in egg characteristics (table 3) and larval head capsule lengths (table 4) of the different instars between A. mcmillani and A. annettae. Considering, however, that these immatures were obtained from eggs laid by a very few females, and that the larvae were reared under laboratory conditions, the size differences noted here may be artifacts. Otherwise, eggs and larvae of the two species could not be distinguished from one another.
Larvae of Austroconops, when compared to other Ceratopogonidae, are missing head capsule sensilla z, k (sensory pit), and q.
Also, there are two additional dorsolateral setae on larval abdominal segment 9 ( fig. 3H), here labeled as 1, and 1l,, which have not been reported in other Ceratopogonidae.
The larval antennal sensilla of Ceratopogonidae have never been described in sufficient detail to determine most homologies within and outside the family. Larvae of Austroconops species have a well-defined basal segment bearing 6 apical sensilla and a further second and third segment (figs. 9A, B, 14A, B, 20A). Based on position, relatively large size, and a uniformly porous surface, the most elongate, porous sensillum is homologous to the "large lobe"' present in Culicoides (Murphree and Mullen, 1991) and most other Ceratopogoninae (Borkent and Bissett, 1990;Borkent and Craig, 1999). Based on position and some details of structure, the following sensilla are likely homologous to sensilla in Chironomidae: most elongate, porous sensillum = blade; mesal short sensillum = accessory blade; short bilobed sensillum labeled sensillum 3 here = fused style and peg sensillum. The remaining three short sensilla do not have readily apparent homologies with Chironomidae; they are labeled here as sensilla 1, 2, and 4. Sensillum | has an open apex bearing a tiny peg. Sensillum 4 has a porous surface.
The labrum of Austroconops shows a number of similarities to other Ceratopogonidae. Sensilla 1-4 are clearly present as a group on the labra of most other Ceratopogonidae (Hribar and Mullen, 1991, labeled as *'sensillum styloconicum'"'). In many Chironomidae the two sensilla SIVA amd SIVB on the anterior portion of the labrum are morphologically very similar to S2 and S4, and BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA pa these are likely homologous. Chironomidae S1, just dorsal to the labral lamellae, probably is homologous to our S10, and the scopae present in Austroconops and Ceratopogoninae are probably homologous to the labral lamellae (Wiederholm, 1983). Further detailed comparisons require further study of these structures in numerous taxa (both SEM and study of nerves). Borkent et al. (1987) reported an "Mshaped apodeme"' in the female genitalia of A. mcmillani. In further study we are puzzled as to the nature of this structure, which varies in both A. mcmillani and A. annettae from a faint M-shaped sclerotization to the presence of a pair of laterally positioned, pigmented sclerites. Some other Ceratopogonidae also have sclerotized structures anterior to the fused or separated sternite 9 (e.g., some Leptoconops, some Culicoides, Alluaudomyia Kieffer, and others). Careful histological study is needed to better interpret this feature. Cladistic analysis below indicates that, among extant taxa, Austroconops is the sister group of Leptoconops and that together they form the sister group to all remaining extant Ceratopogonidae. Eight species of Austroconops are now known: two of these are extant and six are Cretaceous fossils: A. annettae, n.sp. Borkent, this publication. Australia (Western Australia) A. borkenti Szadziewski and Schliiter, 1992: 78. France. Upper Cretaceous A. fossilis Szadziewski, 1996: 38 A. sibericus Szadziewski, 1996: 40. Russia. Upper Cretaceous The genus Jordanoconops Szadziewski was proposed as a monotypic genus to include a Lower Cretaceous Jordanian amber fossil, Jordanoconops weitschati Szadziewski (Szadziewski, 2000). It may actually be a member of Austroconops (see below under   ) and at times were attacking humans at the rate of more than 50 per minute! Further observations on biting are given below. Swarming males were collected directly southeast of the tee-off area at the southwestern end of the fairway, which was the driest area on the fairway. The southeastern margin of the fairway had some very wet soil about halfway down its length, and the northeastern end included a ditch with open water. The vegetation included the following trees at the western drier end: Agonis flexuosa (Weeping Peppermint; native to Southwest of W.A., but introduced into the Park), Allocasuarina fraseriana (Common Sheoak), Banksia attenuata (Yellow Candle Banksia), Banksia grandis (Bull Banksia), and Banksia menziesii (Firewood Banksia). The shrubs and sedges were the same as listed for the eastern (wet) end as given below, but were sparse. Trees at the wet eastern end of the fairway were Banksia littoralis (Swamp Banksia);  west of where a small stream enters the lake. The adult population was significantly smaller than at site A, and on November 18-20 females attacked, at best, at the rate of about 1-2/minute. One male was swept from the surrounding vegetation. This wooded area had a narrow innundation zone with very wet soil and mud and some small pools that partially filled after a short rain. The predominant tree at this specific site was Melaleuca rhaphiophylla (Swamp Paperbark) with On November 20 from about 10:50 AM to 1:15 pM the first author and his wife walked completely around Loch McNess ( fig. 22C), stopping 14 times at more-or-less even intervals around the lake for 3 minutes at a time, breathed deeply (to produce maximum CO,), and waited for attacking females. The only spots where females were found were at sites B and C ( fig. 22C), suggesting that the species is indeed restricted to some specific localities in the park. Females attacked within 5 minutes at site B when arriving at 2:55 PM  All sites shared the presence of very wet soils, which corresponds well with the behavior of the reared larvae which actively burrowed through wet mud/detritus.
Additional biting females were collected by Len Zamudio at fairway 3 on January 3 and 9, 2002 by a small creek which traverses the fairway about halfway down its length The paratype specimens of A. mcmillani collected at Yanchep N.P. were collected on December 23, 1954 by B. McMillan "'at the opening of a small cave to the right of the track leading north from the golf course" (A. L. Dyce, personal commun.) and this would represent another place in the park where A. memillani are known. However, the total park area has been progressively drying over the past number of years and it is unknown what impact this might have on the distribution of A. mcmillani in the park since 1954.
EGG LAYING, LARVAL AND PUPAL BEHAVIOR AND DEVELOPMENT Eggs were laid by female A. mcmillani in a scattered pattern on the surface of wet mud in a vial or on the sides of the vial just above the mud where moisture had condensed on the vial wall. Three eggs of A. mcmillani that were laid on the sides of the vial were at least somewhat dried after they were laid and had partially collapsed but when immersed in water and rehydrated, they subsequently hatched. The sequence of egg laying, hatching, and larval development is given in table 5.
First-instar larvae burrowed very actively through moderately packed, quite wet mud/ detritus with a serpentine motion. They used different means of propulsion depending on the nature of the substrate and whether they had located food. When larvae were in water, but with substrate to crawl upon, they used their anterior proleg in an anteroposterior motion, dragging the rest of the body along to produce a slightly jerky anterior movement. Larvae that were in very loose detritus and that wanted to move through the water from one clump of detritus to another used their posterior proleg as an anchor, extending  In thicker substrate, at least some movement was facilitated by serpentine movements of the anterior half of the body with the abdomen being dragged behind and with the posterior proleg hooks withdrawn into the anus.
In dense substrate (such as thick mud/detritus or agar), the head capsule would additionally be extended anteriorly, bent somewhat down to take hold of the substrate and then bent farther anteroventrally, helping to bring the body forward; these larvae also withdrew their anterior proleg into the body in the crease between the cervix and remaining prothorax. The proleg was withdrawn much like the inverting of a shirt sleeve, with the terminal hooks inverting first into the interior of the extended leg and being withdrawn toward the body and the rest of the proleg following suit. This inversion is almost certainly the result of muscles that insert at the apex of the proleg ( fig. 4B). Extension of the proleg was the reverse of this sequence. Larvae removed from substrate and placed in a drop of water on a slide never withdrew the anterior proleg as they actively sought a substrate to latch on to, and it is clear that this structure is important to their locomotion.
As larvae were moving through substrate, their head capsules would move rapidly in a flicking motion left and right and dorsoventrally, apparently testing for potential food. The larvae would also periodically stop, at times anchoring themselves with their posterior proleg and poke their heads rapidly here and there among the detritus, again in an apparent search pattern. The larvae rested regularly and most often used their extended posterior proleg hooks as an anchor on the substrate. Such resting larvae were generally in a more-or-less elongate position (with some curves or bends in the body), but occasionally they would assume a tight Ushaped position, with the posterior proleg hooked into detritus.
When first-instar larvae were removed from very wet mud/detritus to water, they < BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 55 immediately became very active, writhing and attempting to find a substrate upon which to take hold. Upon doing so, they immediately burrowed into the substrate and clearly preferred more solid clumps of detritus than fine, looser particles to burrow into. Larvae in detritus that were harassed with a probe quickly escaped with rapid use of the anterior prolegs and, in thick substrate, some undulations of their bodies. Five first-instar larvae of A. mecmillani were observed feeding on E. coli, and when doing so they had their posterior proleg firmly hooked into the substrate and worked over acceptable substrates with their mouthparts. The head capsule moved back and forth, appressed to the agar, and the mouthparts moved rapidly, with the labrum contracting and apparently moving food into the mouth cavity. The pharyngeal complex periodically moved food into the gut. The gut contents appeared white in these individuals (same color as the E. coli). When feeding, the antennae actively moved independently of one another. After feeding on EF. coli for 3 days, all five larvae died.
When drops of the fecal infusion, nematodes, and rotifers were introduced to the larvae in the mud/detritus substrate, further observations were made of feeding. Larvae were obviously attracted by the drops of fecal infusion and were often very concentrated directly underneath these small pools of feces. Occasionally, larvae would be seen lashing at a passing Paramecium or a slower rotifer. A few second-instars were seen to successfully attack and eat passing rotifers but missed most that were nearby. The strong impression of all larval instars was that they could at best only attack slow prey.
Second-instar larvae became increasingly fat and slower and by the third-instar and especially by the fourth-instar, larvae moved more cumbersomely, with a slow, serpentine motion. Serpentine movements of the body were obviously the most important basis for movement as the larvae worked their way through the mud/detritus substrate. To retext. Numbers refer to the fairways at the golf course. D. Map of the type locality of Austroconops annettae at 5 km SSW of Forest Grove, Western Australia. verse direction, a larva either simply reversed its serpentine movement or, by turning its head capsule and anterior portion of the thorax in a posterior direction, doubled back along the length of its body The anal hooks were withdrawn into the anus when a larva moved and were often (but not always) exserted when the larva rested; in some instances the hooks were attached to substrate, but in more sparsely distributed substrate were simply exserted into the water without being anchored to anything. The anterior proleg also appeared to be rarely used by third-and fourth-instar larvae and apparently was exserted only when the larva had difficultly finding any substrate ahead. Second-, third-and fourth-instar larvae always extended their anterior proleg in a groping motion when placed in a drop of water on a microscope slide and writhed helplessly, very similar to the more ponderous movements of Dasyhelea larvae. Larvae which were harassed with a probe quickly escaped with rapid serpentine undulations of their bodies through the surrounding substrate. When the substrate was shaken and larvae severely dis-   turbed, fourth-instars often assumed a curledup position (this was not attempted with earlier instars). The darting, probing search pattern of the head capsule continued in these stages. Larvae sometimes moved their antennae and, when ingesting food, moved their mandibles synchronously. Some second-instar and most third-instar larvae took on a pinkish hue. A few third-instar larvae remained unpigmented and translucent. All fourth-instar larvae were pink or reddish. The day before the single fourth-instar larva pupated, its thorax was strikingly broad, the respiratory organs of the pharate pupa could be clearly seen, and the eyespots had moved posteriorly in the larval head capsule. The pupa, which completely shed the larval exuviae, was initially very pale, with red hemolymph, and, at 22-25°C, was very sluggish, although capable of very slow circular movements of the abdomen. It was buried in the wet substrate at about a 50-60° angle from the surface, with the tip of one respiratory organ just sticking out of the water; later, the tips of both respiratory organs were exposed. Throughout its development the pupa had the apex of either or both respiratory organs exposed to air. After 24 hours, the pupal cuticle appeared nearly transparent (except for the brown respiratory organs), with an inner core of reddish tissue surrounded by completely clear fluid and a few subcutaneous bluish-green patches in the thorax and abdomen. It was virtually motionless, NO. 3449 and flooding for 16 minutes (so that the respiratory organs were submerged) did not result in any pupal movement, nor did any very slight jarring of the surrounding substrate. At one point, upon turning on a bright microscope light, the pupa briefly moved its abdomen in small circular motions. After 48 hours, the pupa did not look much different from 24 hours earlier. After 72 hours, the interior of the pupa became somewhat darker. The blue-green patches were also present in the head (this may have been true earlier but the head could not be seen clearly then). The pupa was very lethargic and moved only slightly when bright light was suddenly shone on it. At 84 hours the developing adult eyes became visible. At 96 hours, the developing adult had darkened further, with the blue-green patches more extensively developed. At 108 hours, the dark adult appeared to be more-or-less completely developed and its black head and thoracic cuticle could be clearly seen. Shortly before the adult male emerged, at 120 hours after pupation, the pupa slowly wriggled its way out of its circular burrow and lay on the wet mud surface. The adult emerged during daylight hours (early afternoon) but failed to completely expand its wings (they were entrapped by the wet mud surface); in all respects it appeared normally developed and a hindleg first tarsomere length of 176 wm (males in nature = 178-214 wm, N = 13) indicated that the rearing conditions and food provided to the larva was of sufficient quality to produce a typical adult.

BITING ACTIVITY AND BEHAVIOR
Of those female adults of A. mcmillani biting humans at Yanchep National Park, most bit on the face ( fig. 21D, F); less than 15% of individuals fed elsewhere on the body (mostly shoulders, arms, legs). Females on the face clearly favored areas without hair and, of those attacking the face, were primarily close to the eyes and directly on the ears, with fewer on the cheeks and forehead. A few fed among hairs right at the hairline, including the scalp, eyebrows, eyelashes, and the first author's beard. We are unable to suggest the cues used by females to preferentially attack the face. When bare arms were 2004 held up and appressed against the face, the face was still preferred; however, in such a position a few females attacked the armpit (no deodorants used), suggesting that secretions of the skin may be involved as an attractant. Slightly elevated heat from these areas cannot be excluded but this seems unlikely, since females were not attracted to handheld cups of warm or hot water, as are females of some species of Culicoides and Culicidae. It is probable that female A. mcmillani were attracted by CO,. At the south end of Loch McNess ( fig. 22C; site B) when the frequency of biting decreased, the number of incoming females could be increased within 3-5 minutes by doing on-thespot exercises or by intentionally breathing deeply and rapidly.
Females flew about the head for a period before landing. Those landing on the hair nearly always quickly flew off the surface to resume hovering. After landing on _ skin, nearly all females began feeding immediately (very littke walking about on the skin surface).
Kangaroos were closely observed during the first sampling period of October 16-20, 2001, before the emergence period of A. mcemillani. They very rarely scratched or rubbed themselves. When the first author and his wife were experiencing very intense biting on the golf course in Yanchep N.P. on November 20-21, it was clear that the kangaroos were also being severely bothered by midges (figs. 21B). Many rubbed their eyes and scratched at their forelegs and the shins of the hindlegs with their foreleg feet (fig. 22A). Female kangaroos carrying well developed joeys (baby kangaroos) also often scratched at the front of their extended pouches. The midges attacked the eyes and areas with a minimum of hair. The first author chased kangaroos with an aerial net during these observations and collected numbers of A. mcmillani females, but it was uncertain if these specimens were attacking the kangaroos, were attracted to the first author, or were merely free flying in the area.
Females assumed a biting posture similar to those of Leptoconops and Culicoides that the first author has examined, with the body virtually parallel to the skin surface but very slightly elevated posteriorly ( fig. 21D, F) 22C; site B), they ceased attacking by 5:45 pM as the light was fading and the temperature was 24°C. Females in vials became lethargic at about 18-19°C and at about 16-17°C were nearly torpid. It appeared that stronger winds dampened activity, although females continued to bite in sheltered areas, such as the leeward sides of trees, a behavior typical of many biting ceratopogonids. High temperatures (generally above 24°C) and increased humidity appear to greatly increase biting rates (L. Zamudio, personal commun.). Cloudy weather and even a light rain did not stop females from attacking, although a steady drizzle did stop biting activity on the afternoon of November 18, 2001. from 6:45 to 9:15 Am. At 6:45 the temperature was only 15°C (in shade, warmer in sun) and calm. At 7:50 AM and 22°C a single, small (about 30 cm in diameter), more-orless spherical swarm was seen about | meter above the grass and about 6 meters from the margin of short trees at the edge of the fairway. A light wind began blowing. Several similarly situated swarms were seen during the next 40 minutes. The swarms started to increase in size, became more elongate vertically, and moved to within 3 meters of the leeward side of the trees, just south of the tee-off area. No obvious swarm markers were evident, other than being on the leeward side of the trees. By 8:05 AM the bottom of the swarms were 2-3 meters above the ground and larger (about 50 < 100 cm) and withstood some wind, albeit being periodically blown about and then regrouping. Swarming activity ceased in this immediate area at about 8:30 AM but continued a short distance farther east along fairway 2 (where it was a bit cooler) and then began in the first area again at 8:37 AM for at least 30 minutes. One swarm collected at 9:00 AM with a rapidly swept aerial net consisted of 160 males and 10 females (the latter may have been captured because females had begun biting by then). By 9:15 AM swarming had ceased and did not resume for the next 30 minutes, at which time observations were terminated.
A male and a female of A. mcmillani were collected in copula at 8:47 Am during these observations by sweeping from ground level to about | meter in height in the nearby vi- NO. 3449 cinity of the swarms; the pair separated after about 20 seconds while in the net. While still attached the pair stood on the net mesh facing in opposite directions, showing that the male can at least opportunistically rotate his genitalia 180° (they are otherwise generally held at about 90°). Small white sheets (about 1 X 1 m) were placed on the ground under the swarms of males but no mating pairs settled on these. The fact that females captured immediately after they had completed feeding were able to lay fertile eggs proves that mating occurs before feeding.

OTHER OBSERVATIONS
On November 21, 2002 females did not attack the first author on fairway 2 in Yanchep N.P. before 7:50 AM. From 6:15 until 6: 45 AM females were sampled by dropping a net over one or more fresh (wet) kangaroo feces and then disturbing the feces with a stick slipped under the margin of the net that was appressed against the grass. Some females were sampled by kicking at the feces and then immediately sweeping briskly over the area. In virtually every patch, some females were collected, with numbers varying from O to 8 individuals (58 collected in 30 minutes). Three attempts at collecting females from dry kangaroo feces failed to produce any specimens. This suggests that the females were imbibing nutrients from the wet feces but further observations are required. That reared larvae obtained nutrients from a fecal infusion in the laboratory might suggest the possibility that the females were laying eggs on or near the kangaroo feces, but most of the feces were in an area with normal soils, and the clearly aquatic larvae of A. mcmillani could almost certainly not survive there.
Of 280 preserved male A. mcmillani collected from swarms in Yanchep National Park on November 21, 2001, one specimen had a phoretic female mite intertwined between its legs. The mite belongs to an undescribed species of Blattisocius Keegan (Ascidae) (E. Lindquist, personal commun.). This mite genus is virtually cosmopolitan, and 11 of the 12 known species are, or are likely to be, predators (the twelfth is an ectoparasite in noctuid moth ears). No obvious 2004 internal parasites (e.g., no nematode infestation) were seen among the specimens examined.
Female adults clean their antennae with the foretibial spurs. The wings were cleaned by opening the wings very slightly and hooking one hindleg over the anterior margin of the wing and pushing it alongside and under the abdomen where both hindlegs rubbed their tarsi against each corresponding wing surface (so that the left leg rubbed the dorsal surface of the left wing and the right leg rubbed its underside). After wing cleaning, the wings were sometimes left at an angle but shortly after "'clicked"' back to the typical overlapping position over the abdomen. The hindleg tarsi were cleaned by rubbing them against each other under the body.

TAXONOMIC DISCUSSION
The adult male and female of A. mcmillani were described in some detail by Borkent et al. (1987). There is some confusion about the exact location of the type locality, recorded on the labels of the type series as "'National Park, Perth, W.A., 21-XII-1954''. There is not, and has never been, a national park in Perth, although the renowned 4-km? Kings Park may have been a candidate in the mind of the collector. More likely, the labels may refer to John Forrest National Park, the first national park in Western Australia, located about 22 km northeast of Perth city center. An old "'Caltex'' roadmap in the possession of the first author, probably from the 1950s or 1960s, indicates John Forrest N.P. merely as 'National Park'', perhaps confirming that this is actually the type locality. The original collector, B. McMillan, is deceased (A.L. Dyce, personal commun.). The holotype is housed in the ANIC. Wirth and Lee (1958) recorded the following original material, all females: holotype and 16 paratypes from ''Perth National Park'', 9 paratypes from Yanchep National Park, 11 non-paratypes from 10 miles SE of Darkan and two nonparatypes from Armadale. We were unable to locate 5 paratypes from "Perth National Park'', 1 paratype from Yanchep National Park, and 10 of the specimens from 10 miles SE of Darkan, and these appear to be lost (A.L. Dyce, personal com-BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 43 mun.). The specimens we examined from Armadale and 10 miles SE of Darkan are labeled as paratypes, although they were excluded from the type series in the original publication. Furthermore, a specimen from **Perth National Park'? (ANIC) was not labeled as a paratype although it almost certainly is one. The slide label is written in W.W. Wirth's handwriting and the specimen is mounted in his typical fashion; we have added a paratype label to indicate this. The female adults from Yanchep N.P. were larger than most of the specimens from farther south. The following are the ranges and means of wing lengths for the limited specimens available: types, 1 female not labeled but likely a paratype (new paratype label added) (3, USNM; 1, ANIC).
In addition to the above material, more eggs, eggshells, and specimens of the different larval instars were studied but were either left to develop or were subsequently lost. Therefore, the numbers in tables 3 and 4 recording numbers of specimens does not match that listed above.

DERIVATION OF SPECIFIC EPITHET
The species was named by Wirth and Lee (1958)    Two females found on November 2, 2001 2004 at the type locality were collected between 3: OO and 4:00 PM when the ambient temperature was 20°C (but substantially warmer in the sun). When the single female retained live in a vial was at 17°C, she became very lethargic (nearly torpid), but when the vial was warmed up, she again became very active. A third female collected on November 4 was lost but was collected at 12:50 PM and at 22°C. One male was collected on November 13 at 1:15 PM at 33°C, and one female was found on the same date and temperature at 3:30 PM. Fervent and repeated sweeping all around this immediate site on each of the above dates failed to produce any further specimens, which suggests that the few adults were indeed concentrated at the small pool that appeared to be the only open water (although very small) in a large area. Samples of mud from the edge and bottom of this small pool failed to produce any immatures. The distinctive claw of female A. annettae ( fig. 1J), with a pronounced inner basal tooth more-or-less situated in the same plane as the rest of the claw, may indicate the type of host upon which the female feeds. Within the Simuliidae, females of species with a similarly shaped claw are restricted to those which feed on birds (Crosskey, 1990). Within the Ceratopogonidae, the only species of vertebrate feeders with variation regarding the presence or absence of an inner basal tooth are those in the genus Leptoconops (Forcipomyia Meigen (Lasiohelea Kieffer) and Culicoides all have simple claws with at most, a small, very slender spicule). Nearly all Leptoconops which bite mammals have a simple claw or a claw with a small basal spicule. Species with a claw with a pronounced inner basal tooth feed either on humans (e.g., L. spinosifrons (Carter) and L. siamensis Carter; Chanthawanich and Delfinado, 1967) or on birds (L. werneri Wirth and Atchley; Wirth and Atchley, 1973), indicating that the shape of the claw may not strictly indicate host type in Leptoconops. Other species which have a claw shape virtually identical to A. annettae are L. freebor-  The female allotype collected on November 13 laid eggs scattered separately on the surface of wet mud in a vial or on the sides of the vial just above the mud. The sequence of egg laying, hatching, and larval development is given in table 6. Larval behavior was indistinguishable from that of A. mcmillani (see above). In addition to the feeding observations associated with the fecal infusion described under A. mcmillani, one first-instar larva A. annettae was observed to eat a small live nematode whole (in less than 2 seconds).
A second larva, upon encountering a somewhat moribund large nematode, pierced it at midlength. This was followed by rapid movement of the pharyngeal complex and ingestion of part of the nematode; it did not complete feeding on the nematode. In spite of these observations, the first-to fourth-instar larvae mostly ignored the large number of nematodes added to the Petri dishes every 2-3 days and, similar to larvae of A. mcmil-46 AMERICAN MUSEUM NOVITATES lani, congregated near fresh drops of the fecal infusion, suggesting that they too primarily feed on microorganisms.

TAXONOMIC DISCUSSION
Austroconops annettae is very similar to A. mcmillani in all its stages (pupa not known) and we were unable to distinguish differences in male or female genitalia, eggs, or the different larval instars (except for some size differences in the immatures, which may have been due to restricted sample size or laboratory rearing conditions; tables 3, 4).
The first-instar larvae are described as having nine undivided segments ( fig. 2C). However, in A. mcmillani, early first-instar larvae also appear to have undivided segments, whereas older first-instar larvae have divided segments. The same is likely true for A. annettae.
The female from Augusta was included as a specimen of A. mcmillani in the analysis by Borkent et al. (1987).
In addition to the above material, more eggs, eggshells, and specimens of the different larval instars were studied but were either left to develop further or were subsequently lost. Therefore, the numbers in tables 3 and 4 that record specimens do not match that listed above.

NO. 3449 DERIVATION OF SPECIFIC EPITHET
This species is named after the first author's wife, Annette Borkent. She shared in virtually every aspect of the six-week expedition to study Austroconops, and the results of this paper would not have been possible without her continuous support.

PHYLOGENETIC CONCLUSIONS
Due to its phylogenetic position in the family, the genus Austroconops plays a pivotal role in the interpretation of character states (both morphological and bionomic) within the Ceratopogonidae. Because it has remained morphologically conservative (unlike its divergent sister genus Leptoconops), the character states in this group allow for a far better understanding of the diversification of features of other ceratopogonids. Furthermore, considering how similar the extant species are to Lower Cretaceous fossils of Austroconops 121 mya, it seems very likely that the biology of the extant species are also similar to those species from so long ago. In short, information on this genus is a window on how at least some Ceratopogonidae were living at a very early stage of their evolution. There are not many families of extant metazoans which include such long-lived genera.
In this section we interpret the phylogenetic relationships between early lineages of ceratopogonids, followed by an analysis of their zoogeography and bionomic divergence. As further characteristics of Ceratopogonidae and other Culicomorpha become better known, Austroconops will remain a critically important group to interpret such features.

CLADISTIC ANALYSIS
In this section we discuss our hypotheses of the relationship of species of Austroconops to each other and the relationship between Austroconops and extant and extinct genera of Ceratopogonidae. Borkent et al. (1987) provided the first cladistic analysis of the relationships between the early lineages of Ceratopogonidae. They concluded that, of extant lineages, Leptoconops was the sister group to all other Ceratopogonidae and that Austroconops was the sister group of all Cer- 2004 atopogonidae other than Leptoconops. These authors, recognizing some conflicting evidence, provided an alternative hypothesis, suggesting that Leptoconops and Austroconops were sister groups and that, together, they formed the sister group of all remaining Ceratopogonidae. As shown below, this alternative hypothesis is now strongly supported.
Since Borkent et al. (1987) provided a cladogram of early lineages, a wealth of fossil ceratopogonid material has been described from virtually every insect-bearing amber deposit in the world (see summary in Borkent, 2000a). Some of those fossils, particularly those from the Cretaceous, represent other early lineages in the family and their cladistic relationships to each other and to extant Austroconops and Leptoconops have been analyzed by Borkent (2000a). Members of species of both Austroconops and Leptoconops are present in the oldest amber, from 121 million years ago (Borkent, 2000a(Borkent, , 2001.
The discovery of a second extant species of Austroconops and, in particular, the immatures of the extant species provides the means to further test and improve our understanding of these relationships as presented below.

RELATIONSHIPS BETWEEN SPECIES OF AUSTROCONOPS
In this section we consider the relationships between the two extant and six extinct species of Austroconops ( fig. 24A). As is the nature of fossils, some character states are missing for some species, and this makes our conclusions rather tentative. Nevertheless, our hypothesis provides a framework for further analysis. The Cretaceous fossils Jordanoconops weitschati and A. borkenti are each known as a single female and both have the derived state of character 1 below. However, further features are either not known or are not informative and the two species are not considered further in the analysis (but see comment section below). The following character states were considered.
1. Wing with r-m transverse or oblique (plesiomorphic); r-m more-or-less parallel to R, and R; (apomorphic). This feature ap-BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 47 pears to be unique within the Culicomorpha and is restricted to members of the genus Austroconops ( fig. 1K, L). The condition in Leptoconops is impossible to score because all members of this genus lack r-m. Therefore, this derived condition may, in fact, have evolved in the ancestor of the extant Austroconops + Leptoconops with subsequent loss in Leptoconops. However, we consider the fossil genus Minyohelea Borkent, with an oblique r-m, to be the sister group of Leptoconops ( fig. 25A), and this would indicate that the derived condition was not present in the ancestor of Leptoconops. 2. Male tergite 9 with well developed apicolateral processes (plesiomorphic); tergite 9 with very short or without apicolateral processes (apomorphic). Apicolateral processes have been repeatedly lost within the Ceratopogonidae. Nevertheless, they are present in at least some members of all early lineages for which males are known. They are present in the earliest lineage of Ceratopogonidae, the 121 million-year-old Lebanoculicoides (Borkent, 2000a). In Leptoconops apicolateral processes are present in most species but are absent or very reduced (leaving only one seta arising from the posterior margin of tergite 9) in the two subgenera Styloconops Kieffer and Megaconops Wirth and Atchley. Apicolateral processes are also present in the fossil genera Archiaustroconops Szadziewski and Protoculicoides Boesel and are present or absent in Minyohelea (Borkent, 2000a;Szadziewski, 1996). Within extant genera, they are present in Dasyhelea, absent or reduced to a single seta in Forcipomyia and Atrichopogon Kieffer (a very few species with a secondarily derived pair of small projections), present in the Culicoidini, and present or absent in the remaining Ceratopogoninae.
Further outgroup comparisons were provided by Borkent (1995: char. 3). The presence of apicolateral processes is likely a synapomorphy of the Ceratopogonidae (see character 9 below in the analysis of generic relationships). Their absence (or reduction to a group of strong setae) in some species of Austroconops is therefore considered to be derived, although there is homoplasy in various lineages. 3. Wing with costa extending well be-yond apex of vein R, (plesiomorphic); costa not extending or only slightly extending beyond apex of vein R, (apomorphic). In Thaumaleidae, Simuliidae, and early lineages of Chironomidae (e.g., all Podonominae, many Tanypodinae, some Telmatogetoninae; Brundin, 1966;Wiederholm, 1989) the costa extends to and, in early lineages in these families, past the apex of R,,;. In Culicoidea, the costa appears to continue along the posterior margin of the wing. Within the Ceratopogonidae, Lebanoculicoides Szadziewski has a short extension past the apex of R,,5; there is an extension in the earliest lineage of Leptoconops (subgenus Palaeoconops Borkent), but there is none in extant species of this genus, it is present in the fossil genus Protoculicoides; it is short to long in the fossil genus Archiaustroconops; it is absent to long in the fossil genus Minyohelea; it is absent in the monotypic fossil genera Fossileptoconops Szadziewski and Alautunmyia Borkent; and it is absent in virtually all extant Ceratopogonidae (Borkent, 1995: char. 46;Borkent, 1998).
It appears, therefore, that a costal extension is present in at least early lineages of other families of Chironomoidea and is present in at least some members of early lineages of Ceratopogonidae. We therefore consider the absence of the costal extension in some Austroconops as derived.
4. Midleg tibia with apical spur (plesiomorphic); midleg tibia without apical spur (apomorphic). Borkent (2000a) used this feature (as character 13) as evidence for the monophyly of Ceratopogonidae other than Leptoconops and Austroconops. It was recognized that the feature was homoplastic in all early fossil lineages (other than the monotypic Fossileptoconops and Alautunmyia, Which either lack the spur or have a very small one). This suggests that the presence of the apical spur is likely plestomorphic within the family, with repeated reductions or losses. All Lebanese amber species of Austroconops have the spur, and the state is unknown in A. sibiricus and A. borkenti. It is definitely absent in both extant species of Austroconops. The loss of a midleg tibial spur is therefore considered a synapomorphy of these two species, although it may actually include both A. sibiricus and A. borkenti. NO. 3449 5. Gonostylus with apical spine (plesiomorphic); gonostylus without apical spine (apomorphic). This feature is difficult to interpret, mostly because of the difficulty of observing the feature in fossil material (Borkent, 2000a: char. 8). The feature is present in at least some species of all early lineages, other than Lebanoculicoides where the condition is uncertain. Within Austroconops, the condition is known only in A. fossilis and A. sibiricus where the spine is present. The absence of the spine is therefore considered derived for the two extant species A. mcmillani and A. annettae.

DISCUSSION
This tentative analysis ( fig. 24A) indicates that there is a close correspondence between the age of species of Austroconops and the pattern of cladogenesis within the genus. The oldest species in Lebanese amber are also the earliest lineages within the genus. The two extant species are sister species; they are also morphologically very similar, with the male and female genitalia and many other features (including all larval features) being presently indistinguishable. The congruence between fossil lineages and patterns of cladogenesis is similar to that present within the entire family (Borkent, 2000a).
The Jordanian Cretaceous amber fossil Jordanoconops weitschati (Szadziewski, 2000) may actually be a member of Austroconops, as it shares with Austroconops the distinctive wing modification described in character 1 above. The sole feature distinguishing it from Austroconops is the loss of one of the radial cells, and as an autapomorphy, this loss may have occurred as a sister group of Austroconops, but equally well may have occurred within the genus. Jordanoconops weitschati belongs between synapomorphy 1 and 3 on the cladogram (characters 2, 4, and 5 are unknown for this fossil). Recognizing Jordanoconops as a valid genus leaves Austroconops without any defining synapomorphy grouping all extant and extinct species. For the time being, we are considering it as the sister group of Austroconops although there is no logical reason to do SO.

RELATIONSHIPS BETWEEN THE EARLY
LINEAGES OF CERATOPOGONIDAE Borkent (2000a) provided the latest cladistic analysis of the relationship between Austroconops and other early lineages of Ceratopogonidae, including those known only as fossils. Our results differ primarily in our conclusion that among extant taxa, Leptoconops and Austroconops are sister groups ( fig. 24B) and that among genera known as fossils, Leptoconops, Minyohelea, Austroconops, Jordanoconops, and Archiaustroconops form a monophyletic group (fig.  25A).
The relationships between early lineages of extant Ceratopogonidae are illustrated in figure 24B. As is always the case, fossils permit study of only a subset of potential character states important to the interpretation of their relationships to each other and to the extant fauna. We therefore provide a separate cladogram ( fig. 25A) indicating the relationships between all early lineages, including both extant Austroconops and Leptoconops and some fossil genera of Ceratopogonidae, using those characters from below which can be utilized in their interpretation.
As in the previous analysis by Borkent (2000a), the monotypic fossil genus Alautunmyia is not included here. It is known from two females in New Jersey Cretaceous amber, but some critical character states are obscure (Borkent, 1996(Borkent, , 2000b. Its phylogenetic placement must await further specimens, including males.
The following character states were considered in our analysis. For a given node, characters are arranged sequentially according to eggs, larvae, pupae, males, and females and dorsal to ventral and anterior to posterior.
1. Egg short, oval to somewhat elongate (plesiomorphic); egg very elongate (apomorphic'); egg short (apomorphic"). Most Ceratopogonidae have very elongate eggs, with a length/width ratio of more than 5.0 ( fig. 2A). Other Culicomorpha have short to somewhat elongate eggs, with some Chironomidae having a maximum L/W ratio of 4.6 (Nolte, 1993). There are two exceptions within the Ceratopogonidae. Smith and Lowe (1948) reported that the egg of L. arnaudi BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 49 Clastrier and Wirth (as Holoconops kerteszi Kieffer) had a L/W ratio of only 3.4, and they drew the egg as short and more-or-less oval but reported its shape as '"'bananashaped''. They indicated that L. carteri Hoffman (as L. torrens (Townsend)) had a similar egg. However, papers on seven other species of Leptoconops reported that these have the similar elongate egg as shown here for Austroconops. We suspect that Smith and Lowe (1948) misidentified the eggs as Leptocon-Ops. Forcipomyiinae and Dasyhelea have short eggs and we consider this a secondarily derived condition within the Ceratopogonidae (apomorphic").
2. Larval head capsule setae o distant from one another (plesiomorphic); setae o closely associated, with bases nearly abutting (apomorphic'); setae o represented by a single seta (apomorphic"). All Ceratopogonidae larvae other than Leptoconops have a pair of setae called '"'o" on the anteroventral area of the head capsule, and in those taxa with two setae, they are close to one another (figs. 2K, 3G). Larvae of Leptoconops appear to have only one o seta ( fig. 23C). Some Forcipomyiinae may also have only one o seta but others are reported as having two; further study is needed.
The interpretation of this character is somewhat difficult. We have been unable to accurately homologize the setae on the head capsule of Ceratopogonidae with those of other Culicomorpha; nevertheless we are confident that this is possible, especially through the study of first-instar larvae and innervation patterns. However, other Culicomorpha do not appear to have a closely associated pair of setae on the anteroventral area of the head capsule, although the head capsule setae S9 and S10 of Chironomidae (Saether, 1980) may be homologous to the o setae illustrated here. In some chironomids the two setae S9 and S10 are close to one another. We therefore consider the paired setae o of all Ceratopogonidae, other than those of Leptoconops, a synapomorphy for the family.
The chaetotaxy of the head capsule of Leptoconops has never been homologized with those of other Ceratopogonidae, likely because of the rather modified nature of their head capsules. Based on the illustration of L. gallicus Clastrier by Clastrier (1972; identified as L. kerteszi) we provide a tentative identification of the setae ( fig. 23A-C). Even if we have not correctly identified seta o in Leptoconops, it is appears that there are no setal pairs on the head capsule of any Leptoconops and this is likely derived. The larvae of Leptoconops require further study with scanning electron microscopy.
3. Larval pharyngeal complex present and acting as a sieve or absent; not acting as a crushing structure (plesiomorphic); pharyngeal complex a grinding and sifting structure (apomorphic). Borkent et al. (1987: char. 2) discussed this character and predicted that the then unknown larva of Austroconops would also have a pharyngeal complex (as described here: figs. 2J, 3F). However, the pharyngeal complex of Austroconops apparently lacks brushes or teeth as do some, but not all Leptoconops (Clastrier, 1971;Hribar and Mullen, 1991;Smith and Lowe, 1948). Brushes or teeth are present in larvae of other ceratopogonid genera but can be difficult to detect. 4. Larval anal papillae simple (plesiomorphic); anal papillae bifid (apomorphic). Within Ceratopogonidae, the anal papillae are bifid in Austroconops (figs. 3B, 11B), nearly all Forcipomyia and Atrichopogon, all Dasyhelea, and in all Ceratopogoninae (poorly known). Leptoconops have simple papillae (Aussel, 1991;Ishigami, 1959;Smith and Lowe, 1948); we consider this as a reversal to the plesiomorphic condition. Most other Culicomorpha appear to have simple anal papillae. Simuliidae have anal papillae that are apically simple although many species of Simuliini have anal papillae which bear a few to many small basal lobules, clearly a secondarily derived state. Chironomidae, Thaumaleidae, and Culicoidea larvae have simple anal papillae. These outgroup comparisons support the presence of bifid anal papillae as a derived condition.
5. Pupal respiratory organ with apical porous plate (plesiomorphic); respiratory organ with open pores arranged in a row (apomorphic). The apomorphic condition is nearly unique within the Culicomorpha and is therefore likely derived. Chironomoidea other than Ceratopogonidae have a plastron NO. 3449 and Culicoidea have an apical sieve plate (with larger holes). Simuliidae and some Chironomidae have respiratory organs which function as a physical gill and obtain their oxygen directly from aerated water (not from the surface). One group of Corethrella Coquillett species in the New World has remarkably flattened respiratory organs, each of which has a row of pores along the outer edge. The pores are broadly spaced and each has a trachea which gradually merge together near the base of the organ. This condition is likely independently derived in Corethrella.
6. Setae on vertex of adult head capsule scattered or in dorsolateral arrangement (plesiomorphic); in addition to other setae on vertex, a single seta located medially, just dorsal to where the eyes meet or, in groups where the eyes are separated dorsomedially, between these (apomorphic). This character was discussed by Borkent (2000a: char. 9) as evidence that Leptoconops was the sister group of all other Ceratopogonidae, including Austroconops (.e., the derived condition is absent in Leptoconops). This feature requires further analysis and study since the discovery that Austroconops and Leptoconops are sister groups.
Within the Ceratopogonidae, the condition is not known in the fossil genus Lebanoculicoides. Nearly all Ceratopogonidae other than Leptoconops and Fossileptoconops have the apomorphic condition (a few derived species groups have secondarily lost the medial seta; Borkent, 2000a). We are reluctant to interpret the significance of the absence of the medial seta in the single known amber specimen of Fossileptoconops; if true and if the median seta is present in Lebanoculicoides, the absence in Fossileptoconops may be a synapomorphy shared with Leptoconops. If so, it would draw into question the synapomorphies proposed here for Leptoconops + Minyohelea (characters 20 and 21).
We have surveyed further extant species of Leptoconops for this character. It is indeed true that all are lacking the median seta, which is here considered a secondary loss, perhaps related to the adult female habit of burying itself in sand to rest and lay eggs.
As hypothesized here, the presence of the median vertex seta is a synapomorphy of all Ceratopogonidae and has been secondarily 2004 lost in Leptoconops (and possibly Fossileptoconops). For fossil taxa, we have placed this feature on the same node as synapomorphies 8-10, even though it is unknown for Lebanoculicoides ( fig. 25A). 7. Wing with well developed R,,; (plesiomorphic); R,,; thin and faint, very poorly defined or absent (apomorphic). This was discussed by Borkent (2000a: char. 4), with further modification by Borkent (2001: 8). There are some puzzling patterns here. It is clear that there is a strongly and well developed R,,; in the Lebanese amber fossil genus Lebanoculicoides, and this provides good evidence that this genus forms the sister group of all other Ceratopogonidae, fossil and extant. However, the lesser but still clearly evident R,,; in the Lebanese amber subgenus Leptoconops (Palaeoconops) is more difficult to interpret. As placed here, the reduction of R,,; occurred in the ancestor of all other Leptoconops, and its complete loss occurred independently in at least three other lineages ( fig. 25A) and possibly more, depending on the relationships of unresolved lineages.
9. Male adult tergite 9 without apicolateral process or, if present, lacking setae (plesiomorphic); pair of apicolateral processes present and each bearing at least one seta (apomorphic). This character is discussed above as character 2 in the analysis of the relationships between species of Austroconops. 10. Genital fork (sternite 9) of female genitalia well developed, with vaginal apodeme extending along dorsal wall of common oviduct (plesiomorphic); vaginal apodeme very reduced or absent (apomorphic). This character was discussed by Borkent et al. (1987: char. 1). The feature was not clearly visible in any Lebanese amber fossils. It will be particularly important to discover if it is present in the Cretaceous lineage Lebanoculicoides (as the earliest lineage of Ceratopogonidae, fig. 25A).
No other Culicomorpha larvae have divided segments. Several lineages within the Psychodomorpha have independently acquired the derived condition but these differ in structural details, further indicating their independent acquisition of this feature. The larvae of Anisopodidae have secondarily divided abdominal segments but with the anterior portion shorter than the posterior portion for each of these. Larvae of Trichoceridae have secondarily divided first and terminally divided segments with abdominal segments 2-6 divided into three divisions. Some Psychodinae also have secondarily divided abdominal segments with associated sclerotized plates to indicate their position. Although some Leptoconops larvae have only abdominal segments 1-8 subdivided (Aussel, 1991;Laurence and Mathias, 1972;Smith and Lowe, 1948), other species appear to have further divisions of thoracic and/or abdominal segments 1971, 1972. All Leptoconops larvae have a well developed cervix as a subdivision of the prothorax, as do most (all?) Ceratopogoninae.
12. Fourth-instar larvae without hemoglobin in hemolymph (plesiomorphic); with hemoglobin (apomorphic). Fourth-instar larvae of Austroconops have a red hemolymph which is assumed here to be due to the presence of hemoglobin. This feature appears to be unique with the Ceratopogonidae. However, at least some larvae of Leptoconops have orange hemolymph, which is here considered to be homologous to that of Austroconops: L. americanus Carter (as L.
Culicomorpha larvae with red or orange hemolymph are otherwise known only in some Chironomidae (some Tanypodinae, many Chironominae), where this feature is almost certainly independently derived.
Among fossil taxa, this feature is present in males of species of Archiaustroconops. Males of Jordanoconops and Fossileptoconops are unknown. Minyohelea is considered the sister group of Leptoconops on the basis of other character states (see characters 20 and 21 below) and is therefore considered to have lost the derived condition in the scenario presented here.
14. Male and female adult with five palpal segments (plesiomorphic); with four palpal segments (apomorphic). Extant adult Austroconops and Leptoconops have only four palpal segments (figs. 1C-E 23D, E), with segments 4 and 5 being fused (as evidenced by the landmark capitate sensilla on segment 3). Culicomorpha other than Ceratopogonidae have 5 palpal segments with only a few representatives of Chironomidae, and many Culicidae having a reduced num- NO. 3449 ber. Nearly all other Ceratopogonidae have five segments but there have been a number of reductions in unrelated taxa (listed by Borkent et al., 1987: 598); also in Diaphanobezzia Ingram and Macfie, in some Forcipomyia, and in some Atrichopogon).
Within fossil lineages there are reversals to the plesiomorphic condition in one of the six fossil species of Austroconops (A. gondwanicus) and in four out of the six known species of Archiaustroconops (A. szadziewskii Borkent, A. hamus Borkent, A. bocaparvus Borkent, A. alavensis Szadziewski and Arillo).
15. Male and female adult with palpal segments entire (plesiomorphic); apex of palpal segment 3 and base of segments 4 + 5 membranous (apomorphic). The adults of both species of Austroconops and nearly all species of Leptoconops have a membranous connection at the apex of segment 4 and base of segment 4 + 5 ( fig. 23D, E). This condition is not known elsewhere in the Culicomorpha and is therefore a synapomorphy of these two genera. The derived condition, although present in nearly all species of Leptoconops, appears to be absent in a few species (e.g., L. asilomar Clastrier and Wirth, L. stygius Skuse), which we consider to be reversals.
The feature unfortunately is not clear in the fossil material the first author has examined and therefore cannot be presently interpreted for those taxa.
16. Larval mandible positioned laterally on head capsule (plesiomorphic); mandible positioned more medially (apomorphic). The mandibles of Leptoconops are unique in the family in lying side by side ( fig. 23C). All others are more separated medially. Other Culicomorpha larvae have moderately to widely separated mandibles with most operating in opposition or at an oblique angle.
17. Larval mandible with short to moderately elongate apodeme, simply attached to base of mandible (plesiomorphic); mandibular apodeme very elongate, anteriorly broken into separate parts (apomorphic). Larvae of Leptoconops have an unusual and elongate apodeme extending from near the base of the mandible posteriorly to inside the prothorax, and this feature is unique in the 2004 Culicomorpha. The apodeme has been referred to as a ventrolateral rod by previous authors. The anterior portion of the mandibular apodeme of Leptoconops appears to be separated into two or more different portions intimately associated with the base of the mandible, identified by Smith and Lowe (1948) as the mandibular lever, by Clastrier (1971Clastrier ( , 1972 as "plaques de l'arc génien"' (sclerites of the genial arch), and as the subgenal sclerite by Laurence and Mathias (1972). These small apodemes may actually be part(s) of the mandible itself. This complex of apodemes at the base of the mandible is also unique within the Culicomorpha. The homology of the "'ventrolateral rod'' with the mandibular apodeme of other Culicomorpha is confirmed by its intimate connection with the base of the mandible (via the small anterior separate portions) and by the presence of large adductor muscles attached to the 'ventrolateral rod'? which serve to adduct the mandible (Laurence and Mathias, 1972).
The larva of Austroconops also has a relatively long mandibular apodeme (figs. 21, 3E). Larvae of other Ceratopogonidae have either a very short apodeme or lack them altogether. Chironomidae are variable, but the Podonominae we examined had short apodemes. We have not yet made further outgroup comparisons but these may show a more elongate apodeme to be a synapomorphy of Leptoconops and Austroconops, with the apodeme becoming longer and more modified in Leptoconops larvae.
18. Larvae with four anal papillae (plesiomorphic); with two anal papillae (apomorphic). Leptoconops larvae are unique within the Culicomorpha in having only two anal papillae (Aussel, 1991;Ishigami, 1959). Larvae of all other Ceratopogonidae and nearly all other Culicomorpha have four of these structures. The larvae of Simuliidae have three and those of Chironomidae have either two, four, or six, but nearly all have four (almost certainly the plesiomorphic condition in that family). Thaumaleidae and all Culicoidea have four anal papillae.
19. Pupal apicolateral processes pointed (plesiomorphic); apicolateral processes rounded (apomorphic). Leptoconops is the only ceratopogonid genus whose pupae have rounded apicolateral processes. Because those of all other Ceratopogonidae, we consider the condition in Leptoconops as derived. Outgroup comparisons support this interpretation. Chironomidae have a wide array of apicolateral processes from pointed to membranous; the pointed condition, however, is present in many, including early lineages (Brundin, 1966;Wiederholm, 1986).
Simuliidae, Dixidae, Corethrellidae, and most but not all Thaumaleidae pupae have pointed apicolateral processes (called terminal spines in Simuliidae). Chaoboridae and Culicidae have articulated, membranous pupal paddles in place of the apicolateral processes. The rounded, lobelike apicolateral processes of Leptoconops appear to be unique in the Culicomorpha. 20. Wing with long radial cells (plesiomorphic); radial cells very short (apomorphic). This character was discussed by Borkent (2000a: char. 12) but was presented as a derived condition independently evolved in Leptoconops and Minyohelea. We here consider it to be a synapomorphy of these two genera. This character may merely be another expression of synapomorphy 21. However, there are other Ceratopogonidae (derived lineages) in which reduction of the radial cells does not produce a stigma (e.g., Baeohelea Wirth and Blanton, Baeodasymyia Clastrier and Raccurt, and Nannohelea Grogan and Wirth).
21. Female wing with R, and R, joining costa without thickening (plesiomorphic); female wing with R, and R, joining costa as a strong, thickened stigma (apomorphic). Borkent (2000a: char. 6) considered this feature to have evolved independently in Leptoconops and Minyohelea. Because of the discovery that Austroconops is the extant sister group of Leptoconops, we considered them to be homologous here. The condition is better developed in some Minyohelea species than in others (Borkent, 2000a: figs. 19f, 20a). 22. Wing with r-m (plesiomorphic); wing lacking r-m (apomorphic). The wing of Leptoconops, in lacking a r-m, appears to be unique within at least the Culicomorpha.
24. Posteromedial margin of female sternite 8 with or without semicircular concavity, without semicircular row of stout setae (plesiomorphic); with semicircular concavity bearing four or more stout setae on its margin (apomorphic). The apomorphic condition is unique within at least the Culicomorpha. All female Leptoconops examined have the feature (Clastrier and Boorman, 1987;Herzi and Sabatini, 1983;Wirth and Atchley, 1973). 25. Female cercus short, rounded in lateral view (plesiomorphic); cercus elongate, laterally compressed (apomorphic). This feature was discussed by Borkent (1995: char. 6) but as a synapomorphy for only some Leptoconops. Subsequent discoveries of Lebanese amber Leptoconops (Borkent, 2001), which are the sister group of all extant Leptoconops, showed that elongate cerci are basal to the genus and that the short cerci of the females in the subgenera Styloconops and Brachyconops Wirth and Atchley are reversals to the plesiomorphic condition.
The condition is unique within the Ceratopogonidae, and within the Culicomorpha occurs only in members of the Telmatogetoninae (Saether, 1977), which are clearly specialized (marine) Chironomidae. The elongate cerci of some Leptoconops species are likely used to oviposit in sand. Females of at least some Leptoconops in the subgenus Styloconops, which have short cerci, lay their eggs directly on the sand surface (Aussel, 1993). 26. Larval antenna short, with short blade (plesiomorphic); antenna long, with elongate blade (apomorphic). Within the Ceratopogonidae, larvae of nearly all species have short antenna and a short, squat, somewhat bulbous blade. Larvae of Austroconops have an elongate antenna and a strikingly elongate antennal blade (figs. 2H, 3C, G, 9A, B, 14A). The larvae of most species of Forcipomyia and all Atrichopogon have very elongate antennae, but we considered these NO. 3449 to have evolved independently; their antennae are each situated on a cuticular tubercle and either do not appear to have any sensilla or have very short sensilla restricted to the very apex of the antenna (Jeu and Rong, 1980). Within Forcipomyia only the distinctive larvae of the subgenus Phytohelea Remm have short antennae, which seems almost certainly to be a secondary reduction. Saunders (1957;as F. (Trichohelea)) stated that the antennae are reduced to"' windows" which bear minute sensilla. Larvae of species of Brachypogon Kieffer have somewhat elongated blades (not nearly as long as those of Austroconops), but this likely represents convergence.
Outgroup comparisons indicate that other Culicomorpha have elongate antennae and many Chironomidae have an elongate blade, perhaps indicating that the condition in Austroconops is plesiomorphic and that the short antenna and blade in Leptoconops is independently derived from those of Dasyhelea and Ceratopogoninae.
27. Larval head capsule with seta q (plesiomorphic); without seta q (apomorphic). As discussed above, the lack of seta q in Austroconops (figs. 2H, 3C) may be a misinterpretation. It may be that seta t is missing and that q is actually more anteriorly placed in Austroconops. Regardless, either seta q or t is absent in Austroconops and, because both of these setae are present in other Ceratopogonidae, this is likely a synapomorphy. The homology of these setae with those in other Culicomorpha is not yet understood.
28. Larval head capsule with posterior seta of setae p simple, tapering (plesiomorphic); with posterior seta of setae p subapically expanded (figs. 7A, B, 12A, 18) (apomorphic). Within the Ceratopogonidae the apomorphic condition is otherwise known only in some Forcipomyia species where it almost certainly secondarily derived; the shape of the single seta p (See character 45 below) is very diverse in different species of Forcipomyia. Chironomidae, Simuliidae, and Culicoidea do not have apically expanded dorsal head setae and only some Thaumaleidae have plumose setae.
29. Apex of pupal palpus extending to level of or posterior to apex of labium (plesiomorphic); apex of palpus anterior to 2004 apex of labium (apomorphic). This feature is nearly unique within the Culicomorpha ( fig. 4C). In Chironomidae the palpus is directed laterally and therefore cannot be scored here, although they are long enough that if they were directed posteriorly, they too would extend beyond the apex of labium.
30. Pupa with simple setae (plesiom- The apomorphic condition appears to be unique in A. mcmillani (figs. 4A, 5A). The position of the two strong anterodorsal setae abutting against the respiratory organs clearly functions to support the respiratory organs and keep them from bending either medially or posteriorly. Why this might be necessary in nature is not known.
32. Wing veins with r-m transverse or oblique (plesiomorphic); r-m more-or-less parallel to R, and R,; (apomorphic). The feature appears to be unique within the Culicomorpha and is restricted to members of the genus Austroconops and the extinct Cretaceous genus Jordanoconops. See character 1 and the comment section of the analysis between species of Austroconops for further discussion.
33. Female cercus with short to moder-BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA ye) ately long setae (plesiomorphic); with very elongate setae (apomorphic). The derived condition, present only in Fossileptoconops, appears to be unique within the Culicomorpha. 34. Male adult antennal pedicel with large basal foramen (plesiomorphic); pedicel with moderately sized or narrow basal foramen (apomorphic). This feature, discussed by Borkent (1995: char. 17), could be seen clearly in only one specimen of the fossil genus Protoculicoides (Borkent, 2000a) and it had the plesiomorphic condition. Unfortunately, this character is impossible to observe in most fossils.
35. Pupa with hindleg sheath not visible or barely visible along lateral margin of wing sheath (plesiomorphic); hindleg sheath clearly visible (apomorphic). Within the Ceratopogonidae, all pupae other than those of Austroconops ( fig. 4A) and Leptoconops have the hindleg sheath clearly visible along the lateral margin of the wing sheath. Blanton and Wirth (1979: fig. 35B, C) provided a good illustration of a Culicoides pupa showing this feature as the sclerite between the "'wing pad"' and '"'abdomen 2"'. In Austroconops and Leptoconops (Clastrier, 1972: fig. 4B;Ishigami, 1959: fig. 8;Laurence and Mathias, 1972: fig. 9) there is only membrane present between the wing sheath and second abdominal segment, and the hindleg is completely under the wing sheath (except for its apex medially).
Other Culicomorpha exhibit some variation. Within the Chironomidae, virtually all have the wing sheath not visible. However, in the Podonominae genus Boreochlus Edwards the hindleg sheath is apparently present (Brundin, 1966: 307); other Podonominae have the sheath hidden. In other Culicomorpha, the sheath is either not visible or is present (e.g., Simuliidae). We consider the situation in Boreochlus and Simuliidae to be convergent with that present in the Forcipomyiinae + Dasyheleinae + Ceratopogoninae.
36. Adult thoracic anapleural suture well developed, extending to anterior margin of anepisternal cleft (plesiomorphic); anapleural suture short, extending to posterior margin of anepisternal cleft (apomorphic). This character was discussed by Borkent (2000a: char. 14). The feature is probably an excellent indicator of relationship; no instances of homoplasy are known. All Lower Cretaceous ceratopogonids have the plesiomorphic condition.
38. Adult midleg tibia with spur (plesiomorphic); midleg tibia lacking spur (apomorphic).This feature was discussed by Borkent (2000a: char. 13). No member of this lineage has a midleg tibial spur. The derived condition is susceptible to homoplasy in the outgroup, with numbers of losses (see character 4 in the analysis between Austroconops species above because other Forcipomyia and all Atrichopogon are hypognathous, these are probably reversals. It seems likely that the completely hypognathous condition is actually a synapomorphy of the Forcipomyliinae, but a cladistic analysis of the Forcipomyiinae is required to confirm this. Outgroup comparisons indicate that all other Culicomorpha are prognathous, except for the larvae of Thaumaleidae, which are also hypognathous, and this must certainly be due to homoplasy. 40. Larval mandibles articulating with head capsule (plesiomorphic); mandibular articulation reduced (apomorphic). Lieven (1998) showed that the mandibles of larvae of Forcipomyiinae and Dasyhelea are uniquely disarticulated. 41. Larval head capsule with pharyngeal complex moderately developed (plesiomorphic); pharyngeal complex massive (apomorphic). This feature (Lieven, 1998) is nearly unique within the Ceratopogonidae. Larvae of species of Culicoides (Monoculicoides Khalaf) also have large pharyngeal complexes (Murphree and Mullen, 1991), NO. 3449 but this is certainly due to homoplasy. Other Culicomorpha do not have a large pharyngeal complex in the center of their head capsules.
42. Larval salivary duct opening near margin of hypostoma (plesiomorphic); salivary duct opening near center of head capsule (apomorphic). Lieven (1998) showed that, within the Chironomoidea, this feature is restricted to Forcipomyiinae and Dasyhelea. It would be valuable to study additional taxa within and outside of the family to confirm this. Furthermore, this character may be another expression of character 41, in which the pharyngeal complex is situated in the middle of the head capsule.
43. Male antenna with setae on flagellomere 1 of similar length to those on subsequent flagellomeres (other than those few terminal flagellomeres which have shorter setae) (plesiomorphic); setae on flagellomere 1 much shorter than those on more distal flagellomeres (apomorphic). This feature was discussed by Borkent (2000a: 403). Although there is some homoplasy in higher lineages within the Ceratopogoninae, this feature appears to be a good indicator of relationship. 44. Larvae with short, slender setae (plesiomorphic); with prominent, strong setae (apomorphic). Forcipomyiinae larvae are unique within the family in having prominent, strong setae on the head, thorax, and abdomen. Many taxa have these setae on raised tubercles. Within the Culicomorpha, strong setae are also present in Thaumaleidae, which is likely an independently derived feature.
This character was discussed by Borkent et al. (1987: char. 9) as "'setae arising from cuticular projections''. Some Forcipomyia do not have cuticular projections, although these are almost certainly reversals. 45. Larval head capsule with two setae p (plesiomorphic); with one seta p (apomorphic). Larvae of all Ceratopogonidae, other than those of Atrichopogon and nearly all Forcipomyia, have two setae p. Only two species in the highly modified subgenus F.
(Phytohelea) have two setae p and this must be considered a reversal (Chan and LeRoux, 1971; F. nicopina Chan and LeRoux, F. grandis Chan and LeRoux). Further outgroup 2004 comparisons to other Culicomorpha are not yet possible because we are uncertain of homologies.
46. Larval antenna close to anterior margin of head capsule (plesiomorphic); antenna on basal half of head capsule (apomorphic). Forcipomyiinae are nearly unique in the Culicomorpha in having posteriorly displaced antennae. Within the genus only members of the modified F. (Phytohelea) have anteriorly placed antennae, almost certainly a case of reversal. Within the Culicomorpha, only Thaumaleidae, which superficially resemble Forcipomylinae, have posteriorly positioned antennae.
This character may be an expression of the head capsule having become hypognathous (see character 39 above) and perhaps is not an independent indicator of relationship.
47. Adult terminal flagellomere tapering or rounded apically (plesiomorphic); terminal flagellomere with basally constricted nipple at apex (apomorphic). This feature is unique within the Culicomorpha. However, some species of Atrichopogon do not have the derived condition, which we interpret as a reversal to the plesiomorphic state.
48. Male adult tergite 9 with pair of apicolateral processes present and each bearing at least one seta (plesiomorphic); without apicolateral processes or each represented merely by a single seta (apomorphic). The plesiomorphic feature is considered a synapomorphy of the Ceratopogonidae (character 9 above). Many Forcipomyia have a pair of setae on the posterior margin of tergite 9 but others lack even these. Forcipomyia generally do have posterior projections, sometimes drawn as arising from the posterior margin of tergite 9, but in fact these arcitnescerci.
55. Larva with posterior proleg (plesiomorphic); without posterior proleg (apomorphic). Larvae of Leptoconops and Ceratopogoninae are the only members of the family without a posterior proleg, and we consider these as independent losses. Posterior prolegs with hooks are present in all other Chironomoidea. In Corethrellidae and Chaoboridae they are reduced so that only apical hooks are present (Cook, 1956). 56. Pupae with at most one seta and one pore on anterior portion of abdominal tergite 4 (plesiomorphic); with two setae (apomorphic). Nearly all Ceratopogoninae pupae have two anterior setae on abdominal tergite 4 (called dasm for Ceratopogonidae).
The only exception are species of Parabezzia Malloch, where only one seta is present; pupae of this genus are otherwise quite modified. Other Ceratopogonidae have either one seta, one pore, or one seta and one pore ( fig.  5G). Other Culicomorpha generally have only one or no seta. Only within Chironomidae are there some taxa with more setae scattered on the tergite (Wiederholm, 1986), but at least within the Podonominae, there appears to be only one anterior seta (Brundin, 1966;personal obs.). 57. Sternite 9 of female terminalia forming a continuous band ventrally (plesiomorphic); sternite 9 discontinuous medially, forming two halves (apomorphic). This feature was discussed by Borkent (1995: char. 26).

When Wirth and Lee (1958) first described
Austroconops and puzzled over its classification, they stated that ''Final disposition of the genus Austroconops would best await the discovery of the male or, perhaps better still, the immature stages.'' Our discovery and interpretation of the eggs, larvae, and pupae amply confirm their prediction! Among extant taxa, we demonstrate here that Austroconops and Leptoconops are sister groups. Of these two genera, Leptoconops is overall a highly derived group, while Austroconops is generally plesiomorphic.
In the first cladistic analysis of the Ceratopogonidae, there were two character states proposed by Borkent et al. (1987) to unite Austroconops with all Ceratopogonidae other than Leptoconops. We now consider the first character state, the presence of a medial adult head capsule seta in Austroconops and all Ceratopogonidae other than Leptoconops, to have been secondarily lost in Leptoconops (see character 6 above). The second character state is problematic. Borkent et al. (1987) pointed out that the postoccipital ridge was relatively poorly sclerotized and that the posterior portion of the tentorium was directed in a more posterodorsal angle in adult Leptoconops and that this was also the condition present in Chironomidae (hence plesiomorphic). Austroconops and remaining Ceratopogonidae have a more sclerotized postoc- NO. 3449 cipital ridge and a horizontal tentorium and this was interpreted as evidence of the sister group relationship between these two lineages. This character state is in conflict with the conclusions here and requires further study, especially of the state in further Chironomidae and other Culicomorpha.
Our new analysis also results in significant changes in the interpretation of fossil genera, recognizing for the first time the following as a monophyletic group: the extant Austroconops and Leptoconops and the fossil genera Minyohelea, Jordanoconops, Archiaustroconops, and Fossileptoconops ( fig. 25A). Of these, Leptoconops and Minyohelea are now considered as sister groups. There is one previously published character state which suggested different phylogenetic relationships. Borkent (2000a) and Szadziewski (1996) considered a hindleg tarsal ratio/foreleg tarsal ratio of =1.40 as derived and evidence for the monophyly of the Austroconopinae, at that time including Austroconops, Archiaustroconops, and Minyohelea. Considered to be a nearly unique feature within the family, Borkent (2000a) noted there was some homoplasy within Forcipomyia. Since then, further instances of homoplasy with values as high as 1.8 have become evident within some extant and some fossil Leptoconops (Szadziewski, in press). We now therefore no longer recognize any synapomorphies for Austroconopinae.
Other relationships shown in figure 24B are those that have been hypothesized earlier (Borkent, 1995), but we have added substantial support through the addition of further characters, especially from the immatures. In particular, the monophyly of Forcipomyiinae and Dasyhelea is confirmed. This is especially important because two characters given by Borkent (1995) have since been shown to be suspect (Borkent, 2000a: 403). In addition, the Ceratopogoninae have further support as a monophyletic group. A recent molecular study of the mitochondrial cytochrome oxidase subunit 2 of a variety of Ceratopogonidae by Beckenbach and Borkent (2003) supported all the relationships shown in figure 24B.
In our analysis here we have excluded the monotypic genus Washingtonhelea Wirth and Grogan, which may form the sister group 2004 either of the Forcipomyiinae + Dasyhelea, of the Ceratopogoninae, of Forcipomyiinae + Dasyhelea + Ceratopogoninae, or potentially of a subgroup of the Ceratopogoninae (Borkent, 1995). Female Washingtonhelea frommeri Wirth and Grogan have broadly serrate mandibles, a character state restricted to a subgroup within the Ceratopogoninae, suggesting that the genus may be a member of the Ceratopogoninae. It will be fascinating to examine the features of the immatures of this genus once they are discovered.
It is striking that we have been unable to identify a synapomorphy for Forcipomyia. We suspect that the genus will be found to be paraphyletic in relation to Atrichopogon. There is a wealth of character states in both of these genera, and a cladistic analysis is long overdue. Such an analysis would further test or refine some character states used in our analysis here (characters 2, 26, 39, 45-48).
Because of their phylogenetic conclusions, Borkent et al. (1987) proposed that Austroconops be placed in the new subfamily Austroconopinae. The discovery here that the extant Leptoconops and Austroconops and the extinct Minyohelea, Jordanoconops, Archiaustroconops, and Fossileptoconops form a monophyletic group negates the necessity of recognizing a separate subfamily Austroconopinae, and we here place all these genera in the subfamily Leptoconopinae (new synonym).
There are several features of ceratopogonids discussed in earlier publications which require some comment. Borkent (1995Borkent ( , 2000a argued that the presence of an anterior larval proleg must be plesiomorphic within the Ceratopogonidae because it is present in all other Chironomoidea and first-instar Corethrellidae (Borkent and McKeever, 1990). As predicted by Borkent (1995: 103--104), Austroconops larvae have an anterior proleg. Therefore, within the Ceratopogonidae the anterior proleg has been independently lost by Leptoconops, by Dasyhelea and by the Ceratopogoninae (first-instar larvae of Culicoides species have the feature but it is lost in the second-instar; Jobling, 1953): Borkent (2000a: char. 17) indicated that the row of palisade setae on the first tarso-BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 59 mere of the hindleg was a synapomorphy of all Ceratopogonidae other than those known as Lower Cretaceous fossils. That was in error and the feature, as discussed here, is restricted to those Ceratopogoninae which form the sister group of the Culicoidini. The male adults of A. mcmillani and A. annettae have permanently erect antennal plumes. All other male Ceratopogonidae have erect plumes only when they are in a sexually aroused state during swarming periods. Indeed, this is the basis for the name of the family: Ceratopogon Meigen means 'bearded horn'', referring to the decumbent male antennal plume. In other Culicomorpha, most families with males possessing a plumose antenna (Chironomidae, Corethrellidae, Chaoboridae) have the plumes permanently erect. Male Anopheles Meigen also have decumbent plumes but other mosquitoes have permanently erect plumes. These outgroup comparisons suggest that the decumbent state has either evolved twice in Ceratopogonidae (in Leptoconops and most Ceratopogonidae) or the plesiomorphic condition has been reacquired by Austroconops. Nijhout and Sheffield (1979) described the mechanism by which male Anopheles erect their plume, and it would be a test of homology to see if the same is present in Ceratopogonidae.
In our analysis we reported only 12 synapomorphies for the genus Leptoconops. In reality, it would be possible to report many more by interpreting each detail of the larval and pupal chaetotaxy and more details of the larval mouthparts (maxillae, buccal cavity, etc.) and other modifications. It is clear that Leptoconops is highly derived in many of its features.
Our proposal that Leptoconops and Minyohelea form a monophyletic group suggests that the extinct Minyohelea had elongate eggs (character 1), larvae with secondarily divided abdominal segments (character 11), and hemolymph with hemoglobin (character 12), as well as those features shared by all Ceratopogonidae (characters 2-4).

ZOOGEOGRAPHY
Although the two extant species of Austroconops are presently restricted to the southwestern corner of Western Australia, the genus was much more broadly distributed in the past, with fossils recorded from Lebanese, Spanish, French, Siberian, and Burmese ambers. There are two interesting distributional features of these fossils. First, Austroconops fossils are known only from Old World ambers. Second, all Austroconops fossils are known only from the Cretaceous, dating from 85-121 mya ( fig. 25B). By 44 mya, Austroconops had been eliminated from at least northern Europe, as suggested by its absence in Baltic amber (with 1103 specimens and 101 ceratopogonid species recorded; Szadziewski, 1988Szadziewski, , 1998. Ceratopogonidae from other Old World Tertiary deposits either remain unidentified or include too few specimens to be interpreted here. Austroconops, therefore, was diverse in Old World Cretaceous ambers and sometime since its last record 85 mya it became restricted to Western Australia. What might be the reasons for the present day relict distribution of Austroconops? We suggest that the answer lies in an understanding of the nature of the adaptations of the early lineages of Ceratopogonidae. It is striking that the genus Leptoconops, as the sister group of Austroconops, is known from every significant amber deposit (Old World and New), except for the relatively recent deposits on the Dominican Republic (which otherwise have a reduced ceratopogonid fauna; Borkent, 2000a), and is found today worldwide. Leptoconops is an example of an ancient lineage that has survived and thrived by developing a peculiar life cycle with an accompanying wide array of autapomorphies. Species are restricted to areas of beach (or alkaline soils) where the larvae burrow through the sand, and forage on microorganisms; in many places they are the dominant macroorganism in this specialized habitat. Adult females of most (perhaps all) species rest by burrowing into sand, either at night for unfed females or both day and night after blood-feeding. The larval, pupal, and adult stages all have unusual features and have diverged so much that it is difficult to homologize many of the larval and pupal features with those of other ceratopogonids or Culicomorpha. On the other hand, Austroconops is in many ways the quintessential "'living fossil', preserving many plesiomorphic fea- NO. 3449 tures in each life stage. Indeed, this was a primary motivation to discover the immatures; the adults were earlier recognized as conservative, and it was expected that the immatures would be the same. True to expectations, the features of the immatures were easily homologized with those of other Ceratopogonidae, especially those of Dasyhelea and Ceratopogoninae (Forcipomyiinae larvae and pupae are rather divergent). We suggest that this is why Austroconops now has such a restricted distribution-the Ceratopogoninae, and especially species of Culicoides, have retained many of the basic features of Austroconops and have replaced Austroconops throughout most of its historic range. There are two pieces of evidence supporting this assertion. First, there is a general pattern of mutual exclusion in the fossil record ( fig. 25B). Austroconops, as noted above, is restricted to amber 85-121 million years old. The earliest records of the Ceratopogoninae, including the genus Culicoides, are in New Jersey amber (88-93.5 mya), which does not contain Austroconops. Siberian amber from Yantardakh (83-88 mya) includes the last fossil record of Austroconops and the first Old World record of Culicoides, and we suggest that this provides a snapshot of the transition from the one genus to the other. As noted earlier, Baltic amber at 44 mya includes an abundant and diverse group of Ceratopogoninae and no Austroconops.
Second, mutual exclusion between Austroconops and Culicoides is also suggested in the present day local distribution of Austroconops. Although Culicoides is otherwise diverse in Western Australia, we did not encounter any biting Culicoides at Yanchep National Park during the entire adult season of A. mcmillani (sampled for 2 years). Neither did we encounter biting Culicoides at the type locality of A. annettae, and this further suggests that, to this day, the two lineages exclude one another.
Considering that the oldest New World amber is from New Jersey (88-93.5 mya) ( fig. 25B), it is quite possible that Austroconops was present in the New World before they were, as hypothesized here, eliminated through competition by the evolution of the Ceratopogoninae (particularly Culicoides). If ceratopogonid-bearing amber older than 95 2004 mya is discovered in North America, we suggest that it will include Austroconops. In examining maps of land distributions from the Cretaceous (Barron, 1987), it is clear that the Cretaceous islands of Europe that preserved amber were close to coastal New Jersey (with its 88-93.5 mya amber) and there were also connections between eastern Russia and western North America at that time. All amber is preserved in coastal environments and it would be hard to suggest what factor could keep early to mid-Cretaceous Old World Austroconops out of the New World. Finally, it will be particularly fascinating to see if Austroconops shows up in the 53 millionyear-old Eocene amber of Le Quesnoy, France (Nel el al., 1999) or the 57-65 million-year-old Sahkalin, Russia amber (Szadziewski, 1990). Our hypothesis here suggests that Austroconops should not be present in any Tertiary amber deposits in the Holarctic region because Culicoides was well established throughout the region by that time.
If the above pattern is correctly interpreted, we suggest that there may be a good basis for the mutual exclusion between Austroconops and Culicoides. It was striking to us that larvae of Austroconops moved relatively slowly, both in seeking out food and in response to disturbance, when compared to those of most species of Culicoides. In addition, larvae of Austroconops could not swim, again in contrast to the distinctive, remarkably fast swimming motion of Culicoides (and most Ceratopogoninae). Such swimming abilities allow Culicoides larvae a rapid escape response, as well as dispersal to other nearby habitat. We speculate that Culicoides larvae may have a distinct selective advantage to those of Austroconops but, of course, this needs to be tested.

BIONOMIC DIVERGENCE
The larval feeding habits of Ceratopogonidae are quite diverse, ranging from grazing on microorganisms to being obligate predators, eating either whole prey or attacking and invading the larval bodies of much larger Chironomidae (Hribar and Mullen, 1991; Mullen and Hribar, 1988). The discovery that Austroconops larvae could be reared on microorganisms present in a fecal infusion fits BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 61 the general pattern within the family (fig .  26A). The larvae of Leptoconops, Forcipomyiinae, Dasyhelea, and many early lineages within the Ceratopogoninae (most Culicoides, many Ceratopogonini) feed on microorganisms, fungi, and algae. Some Dasyhelea also feed on dying or decayed invertebrates and one species is a leaf-miner (personal obs.). Many members of the Culicoidini and Ceratopogonini appear to include both microorganisms and small prey organisms in their larval diet. Ceratopogonid species that are completely predaceous as larvae appear to be restricted to the lineage composed of the Heteromyiini + Sphaeromuni + Palpomyiini + Stenoxenini. It is clear therefore that feeding on microorganisms is plesiotypic within the family. Within the Chironomidae all early lineages feed on microorganisms and predation is restricted to all Tanypodinae, some Chironominae, and a few Orthocladinae. The larvae of all Thaumaleidae, Simuliidae, and Dixidae are also browsers ON Microscopic organisms and small organic particles. Within the Culicoidea it appears that predation evolved at least four times: in Corethrellidae, Chaoboridae, and, within the Culicidae, at least some Culicinae (e.g., some Psorophora Robineau-Desvoidy) and all Toxorhynchitinae. It thus seems likely that feeding on microorganisms is plesiotypic for the entire Culicomorpha.
Although we were unable to discover the larval habitat of Austroconops species in nature, it was clear from laboratory rearings that the larvae require wet, loose detritus/ mud in which to move about and feed. If the mud compacted at all, the larvae experienced difficulty in moving about. When placed in free-standing water the larvae became frantic, looking for attachment on any piece of detritus; upon contact with a substrate they immediately began burrowing. This strongly suggests that the larval habitat in nature must be a permanently wet substrate that would not be disturbed by either current or wave action. Such a restricted aquatic habitat would be congruent with the pattern otherwise present within the family. Ceratopogonid larvae occur in a wide variety of semiaquatic and aquatic habitats, ranging from decaying vegetation and wet mosses to rivers and lake bottoms. However, nearly all mem-bers of the early lineages ( fig. 26A) are present in small or restricted habitats, such as rock pools, epiphytes, and wet soils at the margins of lotic and lentic habitats. The immatures of Leptoconops are generally found in wet or damp sand or sandy soil near freshwater or marine beaches, although a few species are known from seepage areas in desert oases, the wet margins of salt flats, the margin of vernal ponds in desert areas, in halomorphic, calcareous soil, or in cracked clay soils. Forcipomyiinae and Dasyhelea are either in small aquatic habitats or live in or on wet vegetation (many feeding on algae). All of the early lineages within the Ceratopogoninae are either in small aquatic or very wet habitats (e.g., wet soil and pools) or on the margins of rivers and lakes. Only some species of the lineage Heteromyiini + Sphaeromiini + Palpomyiini + Stenoxenini truly occur in rivers and lakes. Small aquatic habitats and/or wet areas appear to the plesiotypic habitat for the Ceratopogonidae. Outgroup comparisons suggest that this feature is actually plesiotypic for the entire Culicomorpha. The pattern within the Chironomidae is somewhat uncertain for two reasons. First, the most recent phylogenetic interpretations of the family (Saether, 2000) included character states that have not been polarized using outgroup comparisons. Second, the Chironomidae are markedly diverse in their larval habitats. We agree with Cranston et al. (1987) that the earliest Chironomidae were present in small lotic and lentic habitats and that presence of spiracles in these taxa is plesiomorphic. Further outgroup comparisons are more clear. The earliest lineage of Simuliidae, the genus Parasimulium Malloch, is restricted to small seeps, and Thaumaleidae are restricted to wet rock faces. Within the Culicomorpha, nearly all are restricted to small aquatic habitats. Only some species in the derived Chaoboridae genus Chaoborus are found in lakes; presence in lakes is likely a plesiotypic feature of Chaoborus because a number of the larval features are clearly related to predation pressures by fish. Other genera of Chaoboridae, representing earlier lineages in that family, are restricted to small pools and ponds. This pattern shows that presence in small aquatic habitats is a plesiotypic feature of the entire Culicomorpha. NO. 3449 Borkent (2000a) pointed out that both Leptoconops and Austroconops species are xerically adapted (only a very few species of Leptoconops occur in more mesic areas). Considering that the two genera are now known to be sister groups, it is likely that this adaptation was present in their common ancestor as well as in the extinct genus Minyohelea ( fig. 25A).
There is a pattern of dispersal capabilities among the early lineages of Ceratopogonidae. Borkent (1991) showed that the ceratopogonid fauna of volcanic oceanic islands worldwide is virtually always restricted to species of four genera: Forcipomyia, Atrichopogon, Dasyhelea and Culicoides. These genera include many species which live in ephemeral habitats, such as small bodies of water and wet, decaying vegetation, and it was concluded that these temporary habitats selected for taxa which were good dispersers. All four genera are distributed worldwide. They also are early lineages within the Ceratopogonidae ( fig. 24B), suggesting that frequent dispersal was an early characteristic of the ancestor of the Forcipomyiinae + Dasyheleinae + Ceratopogoninae (Culicoides is an early lineage within the Ceratopogoninae). The lineage Leptoconops + Austroconops shows a different pattern. The absence of Leptoconops species on nearly all islands reflects the permanency of its primary habitat on marine and freshwater beaches, even though it has an extensive fossil record extending to 121 mya. The restriction of Austroconops to Western Australia likewise suggests that it is a weak disperser and that its larval habitat, once found in nature, will be a predictable, stable environment. The ancestor of Leptoconops and Austroconops therefore was likely in a stable habitat as well and was a poor disperser.
One of the striking features of female adult Austroconops is that they are diurnal biters. To interpret this feature phylogenetically, figure 26B shows the distribution of the times when female adult Culicomorpha bite. Virtually every lineage includes species which do not feed at all. Within the Ceratopogonidae, female adult Leptoconops, Austroconops, and Forcipomyia (Lasiohelea) (the only Forcipomyia that feeds on vertebrates) all feed during the day. Other Forcipomyia and 2004 Atrichopogon feed on larger insects, and limited records indicate that they too attack their hosts during the day. Female Dasyhelea do not suck blood, and Culicoides females may attack at any time (species' behaviors vary within the genus). Other biting Ceratopogoninae feed on insects of approximately the same size (up to two to three times their own length), with most entering male swarms of other Nematocera, males of their own species, or, for a few species, attacking Ephemeroptera. They may be diurnal or crepuscular. This pattern within the family strongly suggests that diurnal feeding is probably plesiotypic within the family and that the crepuscular and nocturnal feeding periods of some Culicoides and other Ceratopogoninae are a derived behavior.
Outgroup comparisons suggest that diurnal feeding is plesiotypic to at least the Chironomoidea. The feeding period of the only biting Chironomidae (Archaeochlus Brundin, Austrochlus Cranston) are unknown, but Simuliidae bite only during the day. The Culicoidea are either nocturnal feeders (Corethrellidae bite calling frogs), feed at variable times (Culicidae), or do not feed at all (Dixidae, Chaoboridae).
What might be the significance of diurnal feeding in Simuliidae and early lineages of Ceratopogonidae? The oldest fossil Culicomorpha are members of the Chironomidae and are recorded from the Upper Triassic (Krzeminski and Jarzembowski, 1999). As the Chironomidae are the sister-group of the Ceratopogonidae, the latter must be considered equally as old, and the earlier divergence of the Simuliidae shows that the blackflles must have been present in the Upper Triassic as well. This pattern suggests that the diurnal biting habit shared by early lineages of Ceratopogonidae and Simuliidae originated during or, more likely, earlier than the Upper Triassic. The Middle and Upper Triassic was the time during which the dinosaurs and mammals, respectively, first arose. The diurnal feeding habit may, therefore, reflect an early stage of bloodfeeding, in which Ceratopogonidae and Simuliidae depended primarily on visual cues and detection of heightened CO, production by their daytime hosts. Apparently extant Simuliidae do not use heat to detect their endo-BORKENT AND CRAIG: BITING MIDGES FROM AUSTRALIA 63 thermic hosts (Crosskey, 1990). Perhaps the original hosts of biting Culicomorpha were ectotherms and detection of their hosts was only possible when hosts were active during the day (producing more CO, and visually apparent). Similar to Culicidae (Bock and Cardew, 1996), female adult Culicoides, including those that feed nocturnally, use a combination of olfactory, visual, and thermal cues to detect their hosts (Kline and Lemire, 1995). It would be interesting to know what else attracts Leptoconops and Austroconops females to their hosts, other than their known attraction to CO, (Brenner et al., 1984) and especially if they are attracted by heat. Our interpretation here would predict that they are not attracted by thermal cues. Both species of Austroconops almost certainly overwinter in the larval stage (tables 5, 6) and this is by far the most prevalent mode of Ceratopogonidae in temperate areas. Only a few species of Culicoides appear to overwinter in the egg stage (Jobling, 1953;Parker, 1950). Overwintering in the larval stage is therefore plesiotypic within the Ceratopogonidae.

SUGGESTIONS FOR FUTURE RESEARCH
As always in science, more information means more questions. This study indicates several areas where further investigation would be fruitful. @ Locate the larvae and pupae of Austroconops species in their natural habitat. This would give a better understanding of the adaptations of these species and enhance the likelihood of conserving these "'living fossils''. @ Describe the pupa of A. annettae to test the purported pupal synapomorphies detailed here. @ Investigate the presence of female adults of A. mcmillani on wet kangaroo feces and determine the significance of that behavior. @ Determine whether Austroconops transmits any diseases between its hosts. The host of A. annettae is unknown. @ Determine further homologies with other Culicomorpha. Comparison of the first-instar larvae of Culicomorpha would be particular valuable because these seem to generally be more conservative. Within the Ceratopogonidae, an SEM study of the larva and pupa of a species of Leptoconops would allow ho-64 AMERICAN MUSEUM NOVITATES mologies of this divergent group to be better understood. @ Study of Forcipomyiinae cladistically. @ A comparative morphological study of the larvae and pupae of Ceratopogonidae is basic to good keys to the genera and would provide further phylogenetically important characters (Borkent, in prep.).

ACKNOWLEDGMENTS
Annette Borkent, the precious wife of the first author, not only supported this trip with finances but bravely withstood the onslaught of biting midges, smiled while being bitten and photographed, and shared numerous aspects of the exciting expedition to Australia that led to the information reported here. The first author wonders what he would do without her! The expedition to Australia was supported financially by four other sources, and the first author is grateful for their substantial help: the Dipterological Fund (Canada), the Smithsonian Institution and S.W. Williston Diptera Research Fund, CanaColl Foundation (Canada), and the Schlinger Foundation.
The Australian National Park service was incredibly helpful and supportive. In particular, John Wheeler (Ranger in Charge, Yanchep N.P.) and Paul Tholen (Yanchep N.P.) provided logistical support and a wealth of information without which this study could not have taken place. Additionally, park rangers in other national parks provided excellent support and good advice concerning collecting sites. The Department of Conservation and Land Management (CALM) provided the first author with collecting permits in Western Australia.
Dr. Terry Houston at the Western Australia Museum in Perth readily lent much appreciated field equipment during the stay in Australia and gave important advice concerning collecting sites and we thank him for this effort. David T. Pike (Wildflower Society of