BEHAVIOR Behavior and Reproductive Status of Microsepsis armillata (Diptera: ) Away from Oviposition Sites

WILLIAM G. EBERHARD

Smithsonian Tropical Research Institute, and Escuela de Biologõ´a, Universidad de Costa Rica Ciudad Universitaria, Costa Rica

Ann. Entomol. Soc. Am. 93(4): 966Ð971 (2000) ABSTRACT The mystery of where virgin female Microsepsis armillata (Melander & Spuler) copulate has been solved with the discovery of sexual activity in small aggregations that are not tightly associated with the oviposition sites, where matings have been observed previously in this and other sepsid . Nonvirgin females also mate in these aggregations, and matings at such sites may help explain the otherwise puzzling variations in male precopulatory riding behavior in other sepsids. Approximately 25% of the unpaired females in aggregations carried an unlaid egg in her bursa, and larvae had hatched from some of these eggs. Retention of a bursal egg could prevent intromission by males.

KEY WORDS Microsepsis armillata, mating, aggregations, virgin females, Sepsidae

SEPSID FLIES ARE characterized by an unusual mating solution is suggested by observations of occasional system. Males wait near oviposition sites such as dung mating pairs of in large aggregations (Pont 1987) or carrion (Pont 1979), and mount females when they and among ßies in smaller groups away from dung arrive. But the males only mate with females after (Schulz 1999), and a brief note on apparent mating oviposition is Þnished (Parker 1972a, 1972b; Ward attempts by S. neocynipsea Melander & Spuler at an 1983; Eberhard 1999; Schulz 1999). The morphological apparent feeding site (Eberhard 1999). The current Þt between male and female genitalia during copula- article presents a detailed study of reproductive be- tion (Eberhard and Huber 1998) suggests why copu- havior away from oviposition sites. lation is generally postponed until oviposition is com- A second puzzle is that in some species, the males pleted (Eberhard 1996). Males of Archisepsis spp. and of some populations but not others ride the female Microsepsis spp. deposit a large spermatophore that before mating, thus apparently defending her against Þlls the anterior portion of the femaleÕs bursa, where other males (Schulz 1999). The reproductive payoffs each egg must pass on its way toward the ovipositor. to the male from this type of defense are probably At least in A. diversiformis (Ozerov) and M. armillata strongly affected by both the likelihood that the fe- (Melander & Spuler) sperm transfer from the sper- male will eventually mate with him after he has ridden matophore to the femaleÕs spermathecae and sper- her, and whether she is likely to mate subsequently mathecal ducts is not immediate (unpublished data; L. with another male away from this site before she lays Rodriguez, personal communication). Thus, if a male her next clutch of eggs. Mating does not always occur were to copulate with a female before she Þnished after riding in at least some species (40% and 92Ð97% laying the mature eggs in her ovaries, his spermato- of riding males failed to copulate in S. cynipsea (L.) phore would be driven out of her bursa by the next egg and two species of Archisepsis, respectively (unpub- as it passed down her reproductive tract, and little or lished data; Parker 1972b). Subsequent matings away no sperm would be transferred. Alternatively, if a from oviposition sites could also be important, espe- newly deposited spermatophore is difÞcult to dis- cially if the nearly complete second male sperm pre- place, the female would be prevented from laying cedence documented in S. punctum (F.) (Schulz further eggs until she dislodged the spermatophore. 1999) proves to be generally true for sepsids. A pair of puzzles regarding sepsid sexual biology are One possible type of mating site is a large aggrega- associated with this mating system. First, a femaleÕs tion such as those sometimes observed away from initial mating must occur before rather than after her oviposition sites (Pont 1979, 1987). Some aggregations Þrst oviposition bout (Pont 1979). Yet published de- in England were huge, containing up to possibly scriptions of mating only mention copulations follow- 100,000 Sepsis fulgens Meigen ßies (Pont 1987). Ag- ing oviposition, so the many observations of matings gregations this large are not often encountered, how- by females just after oviposition leave unanswered the ever, and detailed behavioral observations of ßies in question of where a femaleÕs Þrst copulation occurs them have never been made. It appears that most (Hafez 1947, Foster 1967, Parker 1972a, Ward 1983, individuals in these aggregations neither feed nor en- Ward et al. 1992, Zerbe 1993, Allen and Simmons 1996, gage in sexual interactions (Pont 1979, 1987). Another Blanckenhorn et al. 1999, Schulz 1999). A possible possible site of mating is a smaller group, such as the

0013-8746/00/0966Ð0971$02.00/0 ᭧ 2000 Entomological Society of America July 2000 EBERHARD:REPRODUCTION IN M. armillata 967

ßies associated with resources such as food or water. made when the female fulÞlled two conditions: she Katja Schulz (personal communication) reported ag- was found in an aggregation in which 90Ð100% of the gregations of S. neocynipsea with a few S. punctum males that were collected on the same or previous days along streams and in moist meadows in the mountains were M. armillata; and she was larger than the median of Arizona, and has also seen mating in several species size of all females collected from the same site. Be- of Sepsis at older dung, which serves for feeding but cause the other species found in these aggregations, M. not oviposition. Silva (1993) mentioned that adults of mitis, is smaller on average than M. armillata, selection Palaeosepsis sensu latu (includes Archisepsis and Mi- of larger females for dissection made it very probable crosepsis) have been captured on grasses, apparently that they were M. armillata. away from dung or carrion. Females to be dissected were either placed in 70% The current study uses behavioral observations and ethanol and dissected within a day, frozen and then dissections to determine both the reproductive con- dissected in saline solution, or placed in a refrigerator dition (ovary development and spermathecal con- at Ϸ5ЊC (where they were completely immobile) for tents) and the recent feeding history (intestine con- 1Ð2 d and then dissected in saline. I noted the follow- tents) of females of M. armillata captured away from ing: (1) the relative volume and color of the contents oviposition sites. Additional fragmentary data are of the enteric cecae or crop, and of the two loops of given for M. mitis (Curran) and A. discolor (Bigot). middle intestine lying between the two ovaries; (2) the contents of the vagina (spermatophore, egg, male genitalia, empty); (3) the length of a large egg in the Materials and Methods ovary; (4) the relative degree of Þlling of the large and Observations were made in pastures and early sec- small spermathecae with sperm (judged by isolating ond growth (shrubs and small trees) near San Antonio the spermathecae on a slide, squashing them under a de Escazu (elevation 1,400 m), San Jose´ Privince, coverslip, and observing sperm density under a com- Costa Rica between December 1998 (just before the pound microscope); and (5) the relative development beginning of the dry season) and the end of April 1999 of the ovaries, including both relative sizes of eggs and (Ϸ2 wk into the next wet season). Brief observations of the ovary itself (judged on the basis of the range of were also made on 10 and 11 April Ϸ5 km W of Atenas, development seen in dissections for a previous study) Alajuela Province, Costa Rica. Counts of ßies were (Eberhard and Huber 1998). Ovaries of females col- performed while walking or crawling very slowly to lected while copulating on dung and that had just laid minimize the number of ßies disturbed. Only ßies a clutch of eggs were judged to be Ϸ20% developed. Ϸ0.5Ð1.5 m ahead were counted, and ßies that were Females with no sperm in either spermatheca were seen landing (some of these were undoubtedly stirred presumed to be virgins. Sample sizes differed for dif- up by my movement) were not counted. The ßiesÕ ferent structures, as some structures were not exam- tendency to spend nearly all their time on the upper ined in some dissections. Means are followed by Ϯ1 rather than lower surfaces of leaves or other planar SD. surfaces, and to fan their wings and walk about actively made it likely that this survey technique included the Results large majority of the ßies present, despite their small size (Ϸ3Ð4 mm long). Flies were collected either by I found six aggregations of M. armillata that con- aspirating individuals or by swinging an net over tained from Ϸ50 to Ͼ2,000 individuals. Most ßies in their resting place. aggregations were on the upper surfaces of more or Aggregations were distinguished from scattered in- less horizontal leaves Ͻ15 cm above the ground. Many dividuals somewhat arbitrarily. An aggregation in- species of plants were occupied, and leaf sizes varied cluded a central area where there were several leaves widely (from Ͻ1toϾ1,000 cm2). Maximum densities with at least 3Ð4 ßies per leaf, and where interactions were only 1Ð2 ßies per square centimeter, and oc- between individuals occurred frequently. It is possible curred only on a few, relatively small leaves. The that some ßies judged to be scattered rather than in an largest aggregation was spread along Ϸ30 m of a small aggregation actually belonged to an aggregation stream. Although some aggregations persisted for whose center was missed. My search for ßies was more than a month, they were not stable at a small focused on the upper surfaces of leaves Ͻ1 m above scale. In two aggregations in an abandoned coffee the ground. Because two other species of Microsepsis Þeld, for instance, both the number of ßies and the and a similarly sized species of Archisepsis were also central area where most ßies were present changed present in both study areas, all species identiÞcations from day to day. were made by collecting ßies and examining them All sites but one were at least partially forested, but under a dissecting microscope. ConÞdent identiÞca- were near the edges of open pastures. The exception tions were only possible for males and for females that was at the edge of a tree-lined path between shaded were captured with a mounted male. Females of A. coffee Þelds where there were many fallen fruits of armata near Atenas, and of A. diversiformis near San Stemmadenia obovata (Hook & Arm.). Three of the Antonio de Escazu were also identiÞable because at four aggregations found and observed repeatedly dur- each of these sites this was the only species with dark ing the latter part of the dry season (March, early spots on the wings. Probable identiÞcations as M. ar- April) were near tiny streams whose width was as millata of unpaired females that were dissected was small as 15Ð30 cm in places. The largest of these ag- 968 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 93, no. 4 gregations (near Rio Agres) had vanished on 27 April, Ϸ1 wk after the rainy season began, and the other two stream-side aggregations were also gone when they were checked the next day. In contrast, two aggrega- tions that were found early in the dry season (De- cember) away from water (at least 100 m from the nearest creek) disappeared later in the dry season. In no case were the ßies nearly as densely clumped (so that the plants appeared black with ßies) as the ag- gregations of Sepsis described by Pont (1979, 1987). It was usually not obvious why one portion of an aggregation was more densely populated than others. In two cases with dense groups on leaves near water, however, I found a rotting fruit of S. obovata nearby with 10Ð30 ßies resting on it. The ßies in a third aggregation were within 3Ð5 m of a smashed, soft, and probably rotting root of a large aroid (probably Xan- thosoma) in the water. Flies on rotting fruit frequently appeared to be feeding because they repeatedly tapped the fruit with their extended mouthparts. One female also apparently oviposited, because she ex- tended her ovipositor with an egg visible inside, in- serted it into the fruit for several seconds, and then Þnally pulled rearward with the ovipositor as if pulling the eggÕs long respiratory tail from her body (Ward et al. 1992, Schulz 1999). This femaleÕs identity as M. Fig. 1. Relative numbers of male sepsids collected at an armillata was conÞrmed by raising her offspring. Ovi- aggregation near a small creek in secondary forest, at fresh position behavior in an aggregation was not seen in cow dung (Ͻ1 h old) Ϸ10 m from the creek in the forest, and any other case, however, so it is probably rare. An- at fresh cow dung in a pasture Ϸ20 m further away from the other aggregation was Ϸ2 m from a plant with a large creek. The relative abundances of all six species differed Ͻ population of membracids. The feces of the membrac- between the dung pats and the aggregation (P 0.01 with ␹2). ids accumulated on some leaves and under the plant, and were apparently fed on by several species of Collections of ßies drawn to fresh cow dung placed as well as by males and probably females of M. 10Ð20 m away from two different aggregations (one at armillata and M. mitis. I cannot rule out the possibility a stream, the other away from water) revealed that the that at other aggregations I missed attractants such as aggregations were far from a random sample of the small feces or small dead under the leaf litter. species of sepsids present in the area. Comparing ßies Scattered individuals were also found at three other at dung in forest Ϸ20 m from an aggregation at the side sites. Males of M. mitis were relatively more common of a stream, and at similar fresh dung Ϸ20 m farther among these individuals. Among 21 scattered males away in an open pasture (Fig. 1), M. armillata were collected on a hillside near a pasture, 90% were M. more abundant in the aggregation relative to males of mitis and 10% were M. armillata. Among 11 ßies col- the two other species of Microsepsis and three species lected along a small stream near Atenas, 73% were M. of Archisepsis collected (chi-square values were 60.5 mitis, 9% were A. discolor, and 9% were A. armata. and 179, respectively, both df ϭ 1, P Ͻ 0.001). Scattered ßies were so dispersed that there were no Those ßies found in pairs in aggregations were all interactions. copulating; in no case was the male only riding the Males in aggregations generally outnumbered fe- female as is common in other sepsids (Parker 1972a, males. The mean percentage of males in 517 ßies 1972b; Schulz 1999) and in M. armillata on dung. sampled from six aggregations was 60 Ϯ 7% (assuming, Copulating pairs were relatively rare. In seven surveys conservatively, that all females were M. armillata). at three different aggregations, only 0.38% of the ßies Spatial heterogeneity in sex ratios within aggregations were copulating (2,056 solitary ßies, four pairs). More was suggested by samples from an aggregation in an than 50 mounting attempts (all of which failed) were overgrown coffee Þeld: of 65 ßies in the 30- to 50-cm- observed, and several hundred brief aggressive inter- diameter central area of greatest concentration, 85% actions in which one individual (probably a male) were males (91% of the males were M. armillata, and struck his head against another, brießy tangled side- the rest M. mitis), whereas of 90 collected at the to-side, or pursued the other ßy to the edge of a leaf. periphery only 43% were males (56% of these were M. One mating pair that was encountered by another armillata and the rest M. mitis) (differences between individual (probably a male) was not attacked. center and periphery in both sex ratio and in species In total, 46 solitary and nine copulating females from of males were signiÞcant, respective chi-square values four aggregations were dissected. Copulating females were 27.0 and 17.6, both df ϭ 1 and P Ͻ 0.001). had less well-developed ovaries than solitary females. July 2000 EBERHARD:REPRODUCTION IN M. armillata 969

Fig. 3. Estimated degree of Þlling of large and small spermathecae in nonvirgin females. In all but two of 26 females both spermathecae had sperm; the large sper- matheca was nearly always as full or more full compared with the small one. Fig. 2. Two estimates of the readiness to oviposit of virgin and nonvirgin females found in aggregations. Nonvirgins of all degrees of ovary development were present, whereas with only small apparent eggs (0Ð0.7 mm), and full virgins all had relatively undeveloped ovaries. mid-intestines containing dark green-brown material (probably freshly consumed dung). Their crops were All nine copulating females had ovaries that were from one-quarter to completely full of clear, transpar- 20% or less developed, and 55% of them were Ͻ10% ent liquid. The clear contents of the crop were pre- developed; corresponding numbers for solitary fe- sumably nectar or other liquid that was appreciably males were 30.4% with ovaries Յ20% developed, and more viscous than the saline solution; when the crop 10.9% Ͻ10% developed (both P Ͻ 0.0004 with Fisher was broken, the contents ßowed slowly out and fell to exact tests) (Fig. 2). Copulating females were also the bottom of the dissection dish and gradually dif- more likely to be virgins than were solitary females (5 fused away. of 9 versus 2 of 46, P ϭ 0.01 with Fisher exact test). The Two females copulating in an aggregation had no spermathecae of the four nonvirgin copulating fe- appreciable material in their intestines (the same was males were about half to two-thirds full of sperm. true for one virgin M. mitis female), and another virgin Copulating virgins had little ovary development (0Ð female collected while copulating had a midintestine 2%). Solitary virgins also had relatively undeveloped full of dung. All virgin female M. armillata (both cop- ovaries (two were Ͻ5% one was 20%). The large sper- ulating and solitary) had clear liquid in their crops. matheca, whose duct is longer from the point where The same was true for two noncopulating virgin A. it branches from the basal, common spermathecal armata females and Þve M. mitis females collected duct, consistently had larger absolute masses of sperm, away from oviposition substrates. and was also usually relatively more full of sperm than the small spermatheca (Fig. 3). Discussion In 28.3% of the solitary females the bursa was not empty. It contained a mature egg (Þve individuals), a Female Feeding and Reproductive Biology. These just hatched from the egg (three individuals), or observations support the speculation, based on less a spermatophore (four individuals). Each of these direct data from S. neocynipsea, that virgin female females had sperm in her spermathecae. Those with sepsids sometimes copulate at sites not associated with spermatophores all had relatively undeveloped ova- oviposition (Eberhard 1999). This interpretation is ries (all Ͻ20%), whereas those with eggs or larvae all open to the objection that dissected females that had had relatively well-developed ovaries (all Ն60%). In completely exhausted their sperm supplies or had all but one female the egg in the bursa was clearly emptied their spermathecae during copulation were larger than any of the eggs in the ovaries. misclassiÞed as virgins. However, the consistent asso- Nine additional females copulating near fresh dung ciation of empty spermathecae and very small ovaries were dissected. Their reproductive and intestinal (Fig. 2) suggests that the females classiÞed as virgins tracts were quite uniform. All had sperm in at least one were indeed virgins. The same association occurred in spermatheca, moderately developed ovaries (10Ð30%) Þve probable virgin M. mitis females and two A. armata 970 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 93, no. 4 females that were collected away from oviposition tion), she could have moved this egg into her bursa. A substrates. It appears that in nature female M. armil- second possibility is that after oviposition the female lata copulate long before they are ready to oviposit, as matured a single egg more rapidly than all other eggs indicated by the lack of empty spermathecae among in her ovaries. This idea is not in accord, however, with females that had at least partially developed ovaries. the fact that no females were found with a single egg The contents of the digestive tracts of young virgin maturing more rapidly than the rest of the eggs in the females suggest that they usually feed on nectar or ovaries. Perhaps the most likely possibility is that some some other clear liquid, and occasionally on dung females do not mate immediately after oviposition, but before mating. instead induce the male to dismount (as commonly Nonvirgin female M. armillata also sometimes cop- observed in some Microsepsis and Archisepsis species) ulate away from oviposition sites, even though non- and then leave the oviposition substrate (as often virgin females follow the common sepsid practice of occurs in some other sepsids; Parker 1972a, 1972b; copulating at oviposition sites after laying their eggs. Ward 1983; Schulz 1999) with a single unlaid egg, Thus, both the site and the timing of copulations varies which is then moved into the bursa where it is fertil- in M. armillata. This sort of variation could help ex- ized and retained. Hatched larvae have been found in plain the variation in male escorting behavior and the bursae of many other groups of normally oviparous differences in the degree of sexual dimorphism that ßies (Meier et al. 1999). occur intra- and interspeciÞcally in Sepsis (Schulz The data in Fig. 3 have implications regarding the 1999). Other possibly important factors include dif- functional signiÞcance of having two spermathecae, ferences in the intensity of competition among males but further studies will be needed to determine on the oviposition substrate (Schulz 1999) and at whether the sperm in the two spermathecae have other sites, and differences in the femaleÕs tendency to different probabilities of being used for fertilizing accept male mounting and copulation while on the eggs. Such differential sperm usage occurs in both dung as opposed to while she is on nearby plants after Scathophaga stercoraria (Ward 1993) and, comparing copulation is Þnished. the seminal receptacle with the spermathecae, in Dro- Maintaining a mature egg or recently hatched larva sophila melanogaster (Fowler 1973). in the bursa has never been noted before in sepsids, Aggregations. The aggregations described here do and raises new questions regarding female receptivity not appear to correspond to the huge groups of Sepsis to copulation. A mature egg has also been found in the described by Pont (1979, 1987). Those of Sepsis differ bursa in females of both M. mitis and A. diversiformis in that they contain up to 100,000 ßies, are much collected away from oviposition sites (unpublished denser and have more sharply deÞned edges, never data). Both eggs and larvae, as well as the spermato- include mating pairs, sometimes include “frenetically phores that also occasionally occurred in the bursa of active” ßies, contain females that are all apparently female M. armillata, would prevent intromission by a virgins and nulliparous, and are not associated with male, because the morphology of copulation in this food or water. Most of the aggregations of M. armillata species is very similar to that in Archisepsis (Eberhard probably resulted, in contrast, from the ßies being and Huber 1998; unpublished data). Use of these ob- attracted to food or water (during the dry season). jects as plugs to prevent intromission would seem Probably the ßies, after feeding or drinking, spent time superßuous, however, because females of M. armillata in the general vicinity, perched on leaves or other also resemble Archisepsis in apparently being immune objects within 50Ð100 cm, as is common in many to rape, because the proctiger of the female must be sepsids near dung pats and carrion. If this interpreta- deßected dorsally by the female before intromission tion is correct, the striking lack of individuals of the can occur (unpublished data). other species present in the same immediate vicinity The fact that some females had a young larva in the of M. armillata aggregations (two other species of vagina indicates that the egg had been retained there Microsepsis, three species of Archisepsis) indicates that long enough for the larva to hatch. The minimum these other species must feed on as yet undetermined length of time the egg must have been held in the resources other than the rotting fruit and ßowers of S. bursa is suggested by the hatching times in other obovata. It may be signiÞcant in this context that S. species, which vary from Ϸ6hinS. impunctata at obovata is strongly poisonous (Pittier 1978). relatively high temperatures (Hafez 1939) to Ϸ1 d for Because it was not possible to identify females to S. cynipsea (Schulz 1989).The signiÞcance of the fact species, the quantiÞcation of biases toward males in that all seven females of M. armillata with a mature egg the aggregations of M. armillata is only tentative. Nev- or larva in the vagina also had relatively mature eggs ertheless, the larger numbers of males, the frequent in their ovaries (60Ð95% developed) is unclear. male-male aggressive interactions and mounting of Three possible explanations for how a mature egg females that occurred in aggregations, and the occa- came to be retained in the bursa occur to me. One sional presence of copulating pairs suggest that mating comes from the observation of a female M. eberhardi at such sites is frequent enough to play a signiÞcant Ozerov that had a single mature egg left in her ovary role in the sexual biology of this species. An additional when she was collected while mating near dung. If, conclusion is that because sex ratios of ßies raised in after the spermatophore that the male deposited was captivity were, if anything, biased toward females eliminated, presumably by the female digesting it as in (62.9% of 70 ßies in one clutch) (unpublished data), A. diversiformis (L. Rodriguez, personal communica- the apparent bias toward males in aggregations implies July 2000 EBERHARD:REPRODUCTION IN M. armillata 971 that females must spend more of their time at still Eberhard, W. G., and B. A. Huber. 1998. Copulation and undetermined sites. The lack of recognizable dung in sperm transfer in Archisepsis ßies (Diptera: Sepsidae) and the midintestines of females collected at aggregations the evolution of their intromittent genitalia. Stud. Dipt. and at scattered resting sites, when combined with the 5:217Ð248. observation that females at dung consistently had Foster, W. A. 1967. Co-operation by male protection of ovi- positing females in the Diptera. Nature (Lond.) 214: midintestines full of dung, suggests that females rarely 1035Ð1036. move directly from dung to aggregations or solitary Fowler, G. 1973. Some aspects of the reproductive biology resting sites. In contrast, the tendency to Þnd crops of of Drosophila: sperm transfer, sperm storage, sperm uti- even young virgin females at aggregations Þlled with lization. Adv. Genet. 17: 293Ð360. clear, viscous liquid indicates that they commonly Hafez, M. 1939. The life history of Sepsis impunctata Macq. feed on liquids, presumably from decaying fruit and Bull. Soc. Fouad Ier Entomol. 23: 319Ð325. roots or perhaps ßowers or other plant parts. Blanck- Hafez, M. 1947. Ecological and biological observations on enhorn et al. (1999) noted that individuals of S. cy- some coprophagous Sepsidae (Diptera). Proc. R. Ento- nipsea in captivity need sugar to survive, and speculate mol. Soc. Lond. 23: 99Ð104. that they forage for nectar in the Þeld. Females at all Meier, R., M. Kotrba, and P. Ferrar. 1999. Ovoviviparity and viviparity in the Diptera. Biol. Rev. 74: 199Ð258. stages of their reproductive cycles occurred in aggre- Parker, G. A. 1972a. Reproductive behaviour of Sepsis cy- gations, including young virgins, individuals with ma- nipsea (L.) (Diptera: Sepsidae) I. A preliminary analysis ture eggs that were probably about to oviposit, and of the reproductive strategy and its associated behaviour females that had probably just oviposited and had patterns. Behaviour 41: 172Ð206. large but empty ovaries. Parker, G. A. 1972b. Reproductive behaviour of Sepsis cy- Limitations of the Data. The survey techniques nipsea (L.) (Diptera: Sepsidae) II. The signiÞcance of the used may be biased toward one sex or the other if the precopulatory passive phase and emigration. Behaviour behavior of males and females differs in ways that bias 41: 242Ð250. their chances of being observed or captured. Males are Pittier, H. 1978. Plantas usuales de Costa Rica. Editorial Costa Rica, San Jose, Costa Rica. probably more active, and more likely to rest on the Pont, A. 1979. Sepsidae. Diptera Cyclorrapha, Acalyptrata. upper surfaces of larger leaves, rather than on the Handb. Ident.Br. Insects 10:5(c): 1Ð35. ground or smaller leaves, making them more easily Pont, A. 1987. The mysterious swarms of sepsid ßies: an sighted. Possible doubts regarding classiÞcation of fe- enigma solved? J. Nat. Hist 21: 305- 318. males as virgins were discussed above. Schulz, K. 1989. Vergleichende Untersuchung zur Biologie einiger kuhßadenbewohnender Arten der Gattung Sepsis (Diptera, Sepsidae). Diplomarbeit, Freien Universita¨t, Acknowledgments Berlin. I thank A. Ozerov for kindly identifying ßies, and K. Schulz Schulz, K. 1999. The evolution of mating systems in black for sharing her extensive knowledge of sepsid biology and for scavenger ßies (Diptera: Sepsidae). Ph.D. dissertation, commenting on a preliminary version of the manuscript. University of Arizona, Tucson. Financial support came from the Smithsonian Tropical Re- Silva, V. C. 1993. Revisa˜o da familia Sepsidae na regia˜o neo- search Institute and the Vicerrectorõ´a de Investigacio´nofthe tropical. III. Os geˆneros Palaeosepsis Duda, 1926, Universidad de Costa Rica. Archisepsis gen. n. e Microsepsis gen. n; chave para os geˆneros neotropicais (Diptera, ). Ilheringia Ser. Zool. Porto Alegre 75: 117Ð170. References Cited Ward, P. I. 1983. The effects of size on the mating behaviour of the dung ßy . Behav. Ecol. Sociobiol. 13: Allen, G. R., and L. W. Simmons. 1996. Coercive mating, 75Ð80. ßuctuating asymmetry and male mating success in the Ward, P. I. 1993. Females inßuence sperm storage and use dung ßy Sepsis cynipsea. Anim. Behav. 52: 737Ð741. in the yellow dung ßy Scathophaga stercoraria (L.). Be- Blanckenhorn, W. U., C. Morf, C. Mu¨ hlha¨user, and T. Re- hav. Ecol. Sociobiol. 32: 313Ð319. usch. 1999. Spatiotemporal variation in selection on Ward, P., H. Hemmi, and T. Ro¨o¨sli. 1992. Sexual conßict in body size in the dung ßy Sepsis cynipsea. J. Evol. Biol. 12: the dung ßy Sepsis cynipsea. Funct. Ecol. 6: 649Ð653. 563Ð576. Zerbe, F. 1993. Innerartliche Gro¨ssenvariabilita¨t und Paa- Eberhard, W. G. 1996. Female control: sexual selection by rungsverhalten bei Sepsis punctum (Fabricius, 1794) cryptic female choice. Princeton University Press, Prince- [Diptera, Sepsidae]. Diplomarbeit, Universita¨t Wurz- ton, NJ. burg, Germany. Eberhard, W. G. 1999. Mating systems of sepsid ßies and sexual behavior away from oviposition sites by Sepsis neocyipsea (Diptera: Sepsidae). J. Kans. Entomol. Soc. 72: Received for publication 12 October 1999; accepted 8 Feb- 87Ð88. ruary 2000.