Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Female promiscuity wikipedia , lookup
Sexual attraction wikipedia , lookup
Human male sexuality wikipedia , lookup
Sexual reproduction wikipedia , lookup
Body odour and sexual attraction wikipedia , lookup
Sexual coercion wikipedia , lookup
Sexual selection wikipedia , lookup
ANIMAL BEHAVIOUR, 2000, 59, 411–421 doi: 10.1006/anbe.1999.1310, available online at http://www.idealibrary.com on Courtship role reversal and deceptive signals in the long-tailed dance fly, Rhamphomyia longicauda DAVID H. FUNK* & DOUGLAS W. TALLAMY† *Stroud Water Research Center †Department of Entomology and Applied Ecology, College of Agricultural and Natural Resources, University of Delaware (Received 2 March 1999; initial acceptance 29 June 1999; final acceptance 27 September 1999; MS. number: A8423) We examined the function of secondary sexual characters in the role-reversed, lekking behaviour of female long-tailed dance flies, Rhamphomyia longicauda Loew (Empididae), to test the hypothesis that the degree of abdominal distention is an honest female signal about the state of egg development. Female Rhamphomyia cannot hunt for prey and they receive all of their protein from males by exchanging copulations for nuptial prey gifts. Females compete for male gifts within leks that are organized for a brief period each evening before dark. Before hovering within leks, females swallow air, inflating expandable pouches on the pleural margins of the abdomen. The result is a large saucer-like abdomen which is further exaggerated by wrapping scaled pro-, meso- and metathoracic legs along its pleural margins. Male preference for an enlarged abdomen was confirmed by suspending plastic models of varying size from monofilament lines and recording which models attracted the most males. There was a positive relationship between egg development and abdominal distention in a related species, R. sociabilis (Williston), which lacks inflatable abdominal pouches. Multiple regression showed that in R. longicauda, abdominal inflation completely masks the state of egg development. We conclude that female R. longicauda deceive mate-seeking males with the unreliable message that eggs are nearing maturation in order to obtain a protein meal in exchange for copulation. Males that fail to identify a female bearing mature eggs risk near-certain cuckoldry and an increased probability that the female will die before oviposition. should be skewed so that there are fewer males than females available for mating, females should compete through courtship behaviour for access to those males and males should become more selective about which of those females they mate with. These predictions concerning courtship role reversal have been upheld in a variety of vertebrate and invertebrate taxa (Gwynne 1991). Male reproductive investment triggering female competition/courtship behaviour and male choosiness comes in two forms: paternal care (vertebrates: Wells 1978; Reynolds 1985; Rosenqvist 1990; belostomatid giant water bugs: Ichikawa 1988; Smith 1997; Zygophachylus harvestmen: Mora 1990; Rhinocoris assassin bugs: Thomas 1995) and courtship feeding through the capture (empidid dance flies: Svensson et al. 1989; Cumming 1994) or production (stomatopod crustaceans: Hatziolos & Caldwell 1983; several orthopteran katydids: Gwynne & Simmons 1990; Simmons & Bailey 1990) of nutritious nuptial gifts. Paternal investment has created sexual selection with enough strength and directionality to produce permanently enhanced secondary Sexual selection is thought to operate when mating success is associated more with one particular phenotype than with others (Darwin 1871). In theory, the sex with the greatest investment in each offspring will have the lower rate of offspring production (Thornhill 1986) and the smaller number of sexually receptive individuals at any one time, and thus will stimulate greater intrasexual competition in the opposite sex (Williams 1966; Trivers 1972). Females are the more choosy, less competitive sex in most species because they typically exceed males in both pre- and postzygotic investment in offspring (Parker et al. 1972; Trivers 1972). Reversal of this courtship norm is predicted when reproductive investment by males exceeds that of females (Gwynne 1991) or when males control resources critical to female reproduction (Alcock & Gwynne 1991). In such cases, the operational sex ratio Correspondence: D. H. Funk, Stroud Water Research Center, 970 Spencer Road, Avondale, PA 19311, U.S.A. (email: [email protected]). D. W. Tallamy is at the Department of Entomology and Applied Ecology, University of Delaware, 216 Townsend Hall, Newark, DE 19717-1303, U.S.A. 0003–3472/00/020411+11 $35.00/0 2000 The Association for the Study of Animal Behaviour 411 2000 The Association for the Study of Animal Behaviour 412 ANIMAL BEHAVIOUR, 59, 2 Mating pairs Courtship swarm Hunting males Stream Figure 1. Diagrammatic depiction of hunting, courtship and mating interactions in R. longicauda. sexual characters in the females of role-reversed birds (Gill 1995), but such ornaments are rare among insects with reversed courtship roles, having arisen only among empidine dance flies (Cumming 1994). In Empis and Rhamphomyia (Empidinae), 28% of 583 species examined have developed some combination of enlarged wings, pinnate leg scales and eversible abdominal sacs on females alone. Although courtship behaviour has been studied in only 18 empidines with female secondary sexual characters, good evidence for role reversal has been found in all but two (Cumming 1994). What remains a puzzle is the function of secondary sexual characters in female dance flies. The role of empidine secondary sexual characters has been examined empirically only in Empis borealis, a dance fly in which the posterior distal margin of female wings is considerably expanded (Svensson & Petersson 1987, 1988; Svensson et al. 1989). In this species, males carrying prey enter female leks where they discriminate among hovering females on the basis of female wing size. Apparently wing size is an accurate index of both female size and age. The most common secondary sexual characters on female dance flies, however, are ventral eye facets that are much larger than the dorsal facets, large pinnate scales on the legs and bizarre eversible sacs on the pleural margins of the abdomen (Newkirk 1970; Cumming 1994). Although it has been presumed that pinnate scales and eversible sacs are the result of sexual selection via male choice (Cumming 1994), neither their role in female courtship nor their effect on male choice has been delineated. Here we used the long-tailed dance fly, Rhamphomyia longicauda Loew (Empididae), and a congener, R. sociabilis (Williston), to test the hypothesis that pinnate leg scales and eversible abdominal sacs increase the apparent size of a lekking female’s abdomen and that abdomen size is an honest signal of a female’s readiness to oviposit. Sex Role Reversal in Rhamphomyia Empidid dance flies are a large group of predators with some 3300 described species world-wide (Chvála 1976). Cumming (1994) argues that the plesiomorphic mating system of the largest subfamily, the Empidinae, is lek polygyny wherein males present females with a nuptial gift (a prey item) at the time of pair formation. Female empidines, having lost the ability to hunt on their own, rely for protein entirely on male-borne nuptial gifts received during frequent copulations (Downes 1970). It is because males control food resources critical to females for the development of their eggs that an increase in the frequency of sex role reversal is predicted in empidines (Cumming 1994). Role reversal in the form of female rather than male leks has been reported in eight species of Rhamphomyia (Cumming 1994). In R. longicauda, females gather at dusk to form leks in which they hover until dark (Steyskal FUNK & TALLAMY: RHAMPHOMYIA COURTSHIP REVERSAL METHODS Female Lekking Behaviour Figure 2. Uninflated (top) and inflated (bottom) R. longicauda females. 1941; Newkirk 1970). During this time males hunt and capture small insects (Fig. 1). As soon as they succeed, males individually fly with their nuptial gift into the female swarm, approaching from below. Before attempting to copulate, males fly repeatedly beneath the swarm, presumably to assess relative female quality. When a potential mate is selected, a male rises beneath her and hovers while the female drops to meet him. Prey exchange is then negotiated and copulation occurs on the wing. Males without nuptial gifts never enter female leks. Females return nightly to the lek to mate and feed while developing their eggs. The most curious aspect of female courtship in R. longicauda is that prior to lekking, females expand the size of their abdomen via pleural extensions of their integument (Newkirk 1970). A lekking female bears a laterally expanded abdomen three to four times the width of a nonlekking female (Fig. 2). Females also have long pinnate scales arranged on a single plane along the lateral margins of their legs and ventral eye facets that are larger than dorsal facets. Male R. longicauda have none of these traits. Because they collapse when punctured, Steyskal (1941) suggested that lekking females inflate themselves with air. We observed and manipulated courtship swarms of R. longicauda at 11 sites in deciduous hardwood edge habitats between 25 May and 1 July from 1985 to 1990 in Avondale, Chester County, Pennsylvania, U.S.A. We quantified sex ratios, mate selection by males, and the number of unmated individuals of each sex remaining in a swarm after the last successful pairing with a Panasonic WV-1854 high-resolution video camera with extended red and near-infrared sensitivity (<940 nm, usable illumination level 0.01 foot candles) fitted with a 50-mm f/1.8 lens. Supplementary illumination was used as needed, provided by a single 15-W incandescent light bulb in a darkroom safelight (Standard B&W safelight filter) at a minimum of 3 m distance from subjects. We analysed recordings on a Quasar VH5261 video cassette recorder. To determine the mechanism by which lekking females expand their abdomens, we performed dissections of expanded females fixed either in alcoholic Bouin’s solution (Carolina Biological Supply Company, Burlington, North Carolina, U.S.A.) or snap-frozen in flight within 95% ethanol cooled to approximately 80C with dry ice. In both cases the preservative was placed in a shallow, wide-mouthed (12-cm diameter) plastic jar. Females were then collected from within swarms directly into the jar. When females were snap-frozen, they were immediately transferred to a freezer (80C) for 12 h, then warmed to 10C and transferred to alcoholic Bouin’s for an additional 24 h before thawing to room temperature. This procedure permanently hardened the pleural regions of the abdomen in the expanded position. Females fixed directly in Bouin’s solution deflated slightly before permanent fixation, but were still easily dissected in the expanded position. To determine the occurrence of virgin females in leks over time, we sampled swarms periodically throughout the mating season. We caught and fixed females (N=88) in Bouin’s solution as before, and dissected and examined the spermatheca of each for the presence or absence of sperm. We also dissected females from one site (Buck’s Corner) early (31 May 1987; N=23) and midseason (16 June 1987; N=26) to assess the stage of egg development in lekking females over time. We arbitrarily chose six size classes of eggs and assigned the eggs of each dissected female to a single size class. Because Rhamphomyia displays gonotrophic concordance, all eggs of each clutch mature simultaneously. Thus, the size of any given egg can be used as an index of the stage of maturation of the entire clutch. We measured 10 eggs from each female. Male Mate Choice We quantified the effect of female size on male mate choice using two-dimensional models of inflated females. We created accurate models by photographing parts of inflated females that had been snap-frozen in flight as they swarmed. Photographs were made in silhouette (i.e. backlit) using Kodak Technical Pan film (2415) developed 413 414 ANIMAL BEHAVIOUR, 59, 2 either accepted or rejected each model as a potential mate after flying beneath it. A male was judged to have accepted a model if it hovered no more than 3 cm under it for several seconds. Males that did not pause beneath a model were scored as having rejected it. To avoid problems from pseudoreplication, males that consecutively returned to the same model were only scored once. To achieve homogeneity of variance, we log transformed the number of male acceptances of each model size and compared them by analysis of variance (ANOVA). We used Ryan’s Q multiple comparison procedure to separate means. Figure 3. Black and white composite print used as an artificial model of an inflated R. longicauda female for quantifying the effect of female size on male preference. for highest contrast (Kodak developer D-19). We substituted wing outlines filled with 50% ‘grey’ computergenerated pattern for actual wing images. We combined the various female parts (wings, legs, head, thorax, abdomen) in the darkroom into black and white composite prints that represented an intact female in flight (Fig. 3). We copied these using Technical Pan film, again developed for highest contrast. We then ‘printed’ negatives on the same type of film using a slide duplicator, yielding film positives. Printing was done at several magnifications including life size (i.e. actual size of average field-collected females), 0.75, 1.5 and 2.0. We cut the images from the film and trimmed them as closely as possible. Clear areas of film were transparent to wavelengths between 314 and 900 nm; black image areas were opaque at all wavelengths. We then glued (epoxy) models to 25-cm pieces of fine monofilament nylon thread and suspended them at approximately 5-cm intervals from another section of monofilament strung across swarm sites so that models hung at about 40 cm above the ground amidst actual lekking females (Fig. 4). We recorded activities on videotape as described above. We used two models of each of the four sizes with positions re-randomized each night data were collected. We quantified male response to these models during the entire swarming period (about 45 min) each night (N=9). Males Abdomen Size as an Honest Signal We tested the hypothesis that abdomen size is an honest signal of the state of egg maturation in Rhamphomyia. We examined the effect of body size, egg number and egg size (egg maturity) on abdomen size in female R. longicauda, which can misrepresent the actual size of their abdomens both by inflating their guts with air and by wrapping their scaled legs around their abdomens’ perimeter, and in female R. sociabilis, a western North American species whose abdominal morphology is argued to represent the unmodified ancestral condition under which the early courtship dynamics between R. longicauda males and females most likely developed; that is, female R. sociabilis lack the ability to inflate the abdomen and thus the ability to disguise the true size of the abdomen over the course of egg development. Although Evans (1988) refers to female R. sociabilis with swollen abdomens and Cumming (1994) lists the species as having extensive pleural membrane, our dissections provide no evidence of inflation. Female R. sociabilis also have pinnately scaled legs, but rather than position them concentrically around the abdomen as in R. longicauda, R. sociabilis females dangle them beneath their bodies during courtship flights (H. Evans, personal communication). Rhamphomyia sociabilis also differs from R. longicauda in that copulating pairs do not disperse from the lek site but rather form a subswarm near the swarm of unmated lekking females (Evans 1988). This feature of the Figure 4. The pattern in which various-sized models of inflated R. longicauda females were suspended from monofilament line among actual lekking females to quantify the effect of female size on male preference. FUNK & TALLAMY: RHAMPHOMYIA COURTSHIP REVERSAL First male observed hunting (N = 13) Mating interaction First male with prey swarming (N = 18) First female swarming (inflated) (N = 31) First mating pair observed (N = 5) Swarm dispersed (N = 27) –80 –60 –40 –20 0 Time relative to sunset (min) 20 40 60 Figure 5. The median and range of times at which mating interactions in R. longicauda began relative to sunset. mating system of R. sociabilis enabled direct measures of female traits associated with male preference. We obtained copulating (N=30) and courting (N=30) R. sociabilis females that had been fixed in Bouin’s solution from populations near Livermore, Larimer County, Colorado, U.S.A., with the help of Howard Evans, Colorado State University. Inflated R. longicauda females (N=30) were snap-frozen from swarms collected in Avondale, Pennsylvania, in the latter part of the swarming season. We measured 10 eggs from each ovary from all females of both species as well as total egg number and three measures of body size: head width (frontal view), wing length (annular incision to apex) and notal length (pronotum to tip of scutellum). We conducted a principal component analysis for both species to create a single index of overall body size based on the synthesized contribution of head width, wing length and notal length to body size. We then used this index (PC1) with the variables egg number and egg size in a stepwise multiple regression to determine which accounted for the majority of the variation measured in abdomen area. We also compared egg size, egg number and body size between R. sociabilis females from copulating swarms and females from courting swarms using t tests. We quantified total abdominal area when viewed two dimensionally from the venter (the view afforded a discriminating male) using the image analysis software IP Lab Spectrum 3.1.2a (Scanalytics, Inc., Fairfax, Virginia). Before dissections, the two-dimensional image of the abdomen of each female was video-captured on a Power Macintosh 7500 using a Sony CCD-IRIS colour video camera (model DSC-107A) mounted on a stereomicroscope. RESULTS Female Lekking Behaviour At our study sites, lekking in R. longicauda began about 10 min prior to sunset (2025 hours) and lasted about 40 min (Fig. 5) every night with suitable weather during the month of June. The first lek was observed on 25 May and the last on 2 July. Lek sites shared no obvious swarm marker on the ground, but all leks oriented directly under light gaps in an otherwise complete canopy. At dusk such gaps provided the essential ‘white’ background against which females could hover as conspicuous silhouettes. Leks varied in approximate size from 10 to 100 females, with little variation per site over the course of the season. Lek members slowly patrolled a horizontal plane about 0.5 m2 from 0.3–0.6 m above the ground or vegetation with their abdomens inflated and their scaled legs held concentrically around their abdomens. Females arrived at the swarm site inflated or uninflated. Uninflated females perched on nearby vegetation and proceeded to inflate their abdomens, a process which took about a minute. Dissection of snap-frozen, inflated females revealed that inflation was accomplished by pneumatic expansion of the midgut. Presumably females ‘swallow’ air to accomplish this. The hindgut remains uninflated, probably either due to inelasticity or the closing of a sphincter between it and the midgut. The result of this gut inflation was to push the abdominal contents laterally into the expandable pouches of the pleural membrane on segments two through seven (Fig. 2). The combination of gut inflation and the lateral position of the legs gave hovering females the appearance of having abdomens much 415 ANIMAL BEHAVIOUR, 59, 2 25 20 15 10 5 0 5 10 15 20 25 0.5 31 May N = 23 0.4 0.3 0.2 1 June 16 June Date 1 July Figure 6. Presence of unmated individuals (by sex) in nightly courtship swarms of R. longicauda following the last successful pairing in a given night. Combined observations from 1986, 1987, 1988 and 1990. larger than they actually were, especially when viewed from below as a silhouette against the sky. One or more (up to 20) flies of either sex remained unmated when courtship swarms dispersed (about 30 min after sunset) on the 21 nights on which such observations were made (Fig. 6). Very early and very late in the mating season there were more males with prey than females to evaluate, but during most of the mating season males with prey were in short supply and hovering females were the only individuals remaining at the lek site. Female behaviour at this time was marked by repeated sudden descents similar to those initiating pair formation in an apparent desperate attempt to secure a mating with (nonexistent or no longer visible) males. Only on two occasions were unmated members of both sexes present in the same swarm when interactions were terminated by darkness. Dissections proved that throughout most of the courtship season leks were comprised entirely of inseminated females; virgins were not found after the first week of June and even on the earliest dates of dissection never exceeded 4% of the females examined. As expected, eggs of most females examined at the end of May were in the early stages of vittelogenesis, but by mid-June all stages of egg development were recorded in nearly equal proportions (Fig. 7). Male Mate Choice Male R. longicauda began hunting up to an hour (median 40 min) prior to sunset either in the vicinity of courtship swarm sites or up to 30 m away (Fig. 5). While hunting, flight was characterized by a fast patrolling, often in a rough ‘figure of 8’ pattern. Males seemed to take any prey they could capture and subdue, but typically sought other swarming insects, especially flies (Diptera), mayflies (Ephemeroptera) and caddisflies (Trichoptera). Males with captured prey immediately flew to the swarm site (Fig. 5). If they arrived before females, males flew several passes through the swarm area before landing on nearby vegetation where they awaited the arrival of females. Once the female lek was organized, 0.1 Frequency Leftover individuals (by sex) Female Male 416 0 I II III IV V VI III IV V Egg developmental stage VI 0.5 16 June N = 26 0.4 0.3 0.2 0.1 0 I II Figure 7. Frequency of various stages of R. longicauda egg development at one lek site (‘Buck’s Corner’) early (31 May 1987) and midway (16 June 1987) through the mating season. males bearing prey flew beneath the swarming females. After several passes under the swarm, males would typically approach an individual female by ascending to a position 1–2 cm directly beneath a female and hovering briefly (0.5–2 s). A female thus approached would immediately drop to the male, whereupon the two would fly away together as they negotiated the transfer of the nuptial gift and the male climbed into the mating position above the female (Fig. 8). Detailed observation of this process was impossible because it occurred as the pairs flew rapidly away from the swarm, but during negotiation of prey transfer individuals appeared to be head to head. Occasionally pairs separated during this phase, but the vast majority of pairings appeared to be successful. Among pairs that separated, it was not clear whether separation resulted from rejection by one partner or merely a failure to transfer prey. Successful pairing usually took 10 s or less, after which the two flew slowly in copula, often in a rough ‘figure of 8’ 1.5–3 m above ground, directly beneath the lowest branch of a tree, and 2–10 m from the swarm site. Occasionally as many as four pairs were seen together, but mating pairs generally did not congregate and they were never observed to alight. FUNK & TALLAMY: RHAMPHOMYIA COURTSHIP REVERSAL 5.0 (a) 4.5 4.0 2 Area of abdomen (mm ) 3.5 Y = 0.005X + 2.356 r2 = 0.72 P ≤ 0.0001 3.0 300 200 400 (b) 16 14 Figure 8. Rhamphomyia longicauda in aerial copulation. Female below, feeding on a nuptial gift (small tipulid crane fly). 12 During prey transfer, and even throughout copulation, males maintained their grasp on the prey item with at least their prothoracic legs (Fig. 8). Nuptial flights lasted at least until it became too dark to see the pairs under ambient light (about 40 min after sunset). Neither males nor females returned to the swarm the same night of a successful mating. Males without prey were never observed in the courtship swarm. When R. sociabilis females from courtship swarms were compared to those from mating swarms the two groups did not differ in body size nor in the number of eggs they were carrying (Table 1). They did, however, differ in egg length; females that had been selected as mates by males carried eggs that were significantly larger (more mature) than the eggs in females that had failed to attract mates. A simple regression of abdominal area against egg length in lekking R. sociabilis females revealed a highly significant (P≤0.0001) and relatively tight (r2 =0.72) relationship (Fig. 9), suggesting that R. sociabilis males select mates that are close to oviposition by discriminating against females with unswollen abdomens. Manipulations with plastic models clearly showed that abdomen size also dictated male preference in R. longicauda. When given a choice between models of four sizes, males approached the largest model (twice life size) significantly more often Y = 0.011X + 10.349 r2 = 0.23 P = 0.0068 225 300 375 Egg length (µm) Figure 9. The effect of egg length on abdominal area in (a) R. sociabilis, a dance fly incapable of inflating its abdomen, and (b) R. longicauda, a species in which females disguise the actual size of their abdomen by swallowing air prior to joining a lek. than the other sizes (ANOVA: F3.24 =9.875: P=0.0002; Fig. 10). Abdomen Size as an Honest Signal In contrast to R. sociabilis, a simple regression of abdominal area on egg length in inflated R. longicauda females showed that abdominal area was a poor indicator of the state of egg maturation for mate-seeking males in this species (Fig. 9). Although the relationship was significant (P=0.007), only 23% of the variation in abdomen size was explained by egg length. The difference in the effect of egg maturation on abdominal size between R. sociabilis and R. longicauda was Table 1. Comparison of females in Rhamphomyia sociabilis courting swarms with those in mating swarms Trait Swarm type N Mean±SE t P PC1 (body size) Mating Courtship Mating Courtship Mating Courtship 20 20 19 17 30 30 −0.10±0.23 0.10±0.22 30.2±1.1 28.6±0.8 325.0±28.6 253.0±23.0 0.63 0.53 1.17 0.25 2.01 0.05 Egg number Egg length (µm) 450 417 ANIMAL BEHAVIOUR, 59, 2 entirely (Table 2). Instead, body size explains 67.1% of the variation in abdominal area. 7 Ln male approaches 418 6 5 A 309 DISCUSSION AB 4 117 3 BC 26 2 C 12 1 0 2.0 0.75 1.5 1.0 Model size Figure 10. The effect of female model size on male preference in R. longicauda. Bars represent 95% confidence intervals. Untransformed means to the right of log transformed means. Means sharing the same letter did not differ significantly (Ryan’s Q multiple comparison procedure; P=0.05). Table 2. Partitioned stepwise multiple regression of abdominal area against egg length, body size and egg number in Rhamphomyia sociabilis and R. longicauda Species Independent variable R. sociabilis Egg length+ PC1 (body size)+ egg number PC1 (body size)+ egg length+ egg number R. longicauda r2 (%) P 72.4 81.2 82.3 67.1 67.4 67.8 <0.0001 0.0241 0.2224 <0.0001 0.5266 0.5499 Table 3. Eigen values for principal component analyses of head width, notal length and wing length as measures of body size in Rhamphomyia longicauda and R. sociabilis Species R. longicauda R. sociabilis e1 e2 e3 e1 e2 e3 Eigen values Variance proportion 2.904 0.061 0.034 2.880 0.074 0.046 96.8 2.0 1.1 96.0 2.5 1.5 even more apparent when the relative effects of egg length, body size and egg number on abdominal area were examined via stepwise multiple regression (Table 2). Principle component analyses of body size measures provided strong evidence for a single principal component variable for size (PC1) in both R. longicauda and R. sociabilis (Table 3), which we used as our index of overall body size in the multiple regression. In R. sociabilis, the dance fly that is unable to inflate its abdomen, there was a highly significant relationship (P<0.0001) between abdomen size and the state of egg maturation, with egg length explaining 72.4% of the variation in abdomen size. This was not the case in R. longicauda; when lekking females artificially inflate their abdomens with air, the relationship between egg length and abdomen size disappears Our observations and manipulations confirmed that courtship roles in the long-tailed dance fly, R. longicauda, are reversed. Females hover in small- to moderately sized leks nightly where they compete with each other for access to males bearing nuptial prey. Selection to attract males is particularly intense because males provide the only known protein source for adult females during the nutritionally demanding period of oogenesis (Cumming 1994). Dissections showed that females become inseminated very early in the season, yet they apparently return to the lek nightly, using males as a source of food as well as sperm. Anecdotal evidence suggested that females are site faithful and return to the same lek each night: when females were removed from a lek for dissection, the lek diminished in size on subsequent nights by approximately the number of females removed. The need to attract mate-seeking males in empidine females has led to the evolution of three prominent secondary sexual traits present in various combinations and in a variety of species: broadened wings, long pinnate scales on the margins of the legs, and eversible membranes on the pleural margins of the abdomen (Cumming 1994). Female R. longicauda have pinnately scaled legs and pleural sacs on the abdomen and these work in concert to enable the female to misrepresent the actual size of her abdomen to males. Before joining a lek, females sit on nearby vegetation and ‘swallow’ air into their midgut. As a female inflates her gut, the normal contents of her abdomen (hemolymph, ovaries, fat bodies, malphighian tubules) are forced laterad into her eversible pleural sacs. Expansion of the abdomen occurs laterally rather than dorsoventrally, resulting in a round saucer-like abdomen. A fully inflated female then takes flight and hovers within the lek among her competitors. To accentuate the size of her abdomen further, a lekking female positions her scaled legs along the lateral margin of the abdomen. Such behaviour can only be appreciated when the female is viewed from the perspective of a mate-seeking male. Males that have succeeded in capturing a prey item enter a lek of females from below and fly slowly beneath the participants. Until this report, dipteran mating swarms (typically comprised of males) as well as female swarms in other sex-role-reversed empidines were thought to orient exclusively over terrestrial lek markers such as puddles, rocks, or distinctive vegetation (Sivinski & Petersson 1997). In contrast, R. longicauda and probably other Rhamphomyia with inflatable abdomens (Cumming 1994) use open patches of sky within an otherwise closed canopy as lek-site markers. This unique choice of a marker above the swarm rather than below it provides the backlighting for hovering females to display their deceptive sexual signals to their greatest advantage. A backlit hovering female appears only as a dark silhouette when viewed from below (Fig. 3). Not only does this advertise the inflated abdomen clearly, it also enables a female to FUNK & TALLAMY: RHAMPHOMYIA COURTSHIP REVERSAL deceive discriminating males about the actual size of her abdomen by folding her fringed legs concentrically around the abdomen’s perimeter. Under such conditions it is very difficult or impossible for a male to distinguish a female whose abdomen is full of mature eggs from one whose abdomen contains only undeveloped ovaries and air. The best strategy for a male would be to choose the female with the largest abdomen. Manipulations with two-dimensional models of varying size confirmed that R. longicauda males do, in fact, prefer to mate with larger females. Several hypotheses, some of which are not mutually exclusive, might explain this preference. (1) Large females may be more capable of laying larger egg clutches or clutches comprised of larger eggs (Svensson & Petersson 1987). (2) Females with the largest inflatable sacs may be preferred because they bear ‘good attractiveness’ genes (Eberhard 1996) and will thus produce ‘sexy’ daughters. (3) Inflated abdomens may compromise survivorship through lost manoeuvrability and thus convey ‘good viability’ genes via the handicap principle (Zahavi 1995; Zahavi & Zahavi 1997). (4) Females with larger abdomens may be further along in the gonotrophic cycle and thus be closer to oviposition than females with smaller abdomens. There are two reasons males might prefer females carrying eggs that are nearly mature. First, if ‘last-male’ sperm precedence is as important in R. longicauda as it is in a number of other flies (Ito & Yamagishi 1989; Otronen 1989, 1990; Ridley 1989; Opp 1990; Lorch et al. 1993), the last male to mate before oviposition has a distinct paternity advantage over males that precede him. Second, females whose eggs are mature are more likely to survive until oviposition than are females whose eggs still require the nutrients delivered by several additional matings. Our data suggest that the stage of egg development is a better predictor of male preference than is overall body size in Rhamphomyia dance flies. In R. sociabilis, the empidine that is not able to inflate its abdomen, egg length explained nine times more of the variation in abdomen area than did body size. Furthermore, courting females (those that had not yet attracted a prey-bearing male) were not significantly smaller in body size than mating females, but their eggs were significantly younger than the eggs of copulating females. Thus, R. sociabilis provides empirical support for the hypothesis that, in species without inflatable abdomens, Rhamphomyia males that prefer females with large abdomens couple with females that are closer to oviposition. Although confirmation must await reliable phylogenetic inferences, it seems likely male preference for females with large abdomens pre-dates the evolution of inflatable abdominal pouches in Rhamphomyia. Thus, our data also suggest that the evolution of inflatable abdominal pouches has enabled R. longicauda females to capitalize on a pre-existing male bias (sensu Ryan & Keddy-Hector 1992; Basalo 1990) towards egg-laden abdomens while simultaneously concealing the state of egg maturation in an extraordinarily effective way. Partitioning the variables in our stepwise multiple regression showed that, in contrast to R. sociabilis, overall body size rather than egg length accounted for most of the variance in abdomen area. Inflated females completely masked the effect that egg growth would otherwise have had on abdominal distention. Egg length in inflated females explained only 0.3% of the variation in abdominal area not accounted for by body size. A simple regression of abdominal area on egg length in R. longicauda suggests that egg length has a measurable effect on abdominal area, but we believe this effect is a consequence of the stronger relationship between overall body size and abdominal area in this species. As body size increases in R. longicauda females, there is more pleural membrane to inflate and thus larger females can inflate their abdomens to a greater size than can smaller females. As our manipulations with models demonstrated, larger females attract more males, each of which brings important food resources to the copulation. If larger females receive more food than smaller females, their eggs will develop faster; hence, larger females will, on average, carry larger eggs (the relationship seen in the simple regression of Fig. 9), not because the eggs distend the inflated abdomens of these females, but rather because larger females have the capacity to create larger abdomens and thus attract more male-borne food. The sample of R. longicauda females represented in Fig. 9 was taken about 3 weeks into the season, when this effect might be expected to be most pronounced. This conclusion is somewhat at odds with that of Svensson et al. (1989) who studied another dance fly, Empis borealis. As in R. longicauda and R. sociabilis, E. borealis displays complete sex role reversal, with females competing for male favour within leks (Svensson & Petersson 1987). But unlike Rhamphomyia species, female E. borealis differ from males in wing morphology alone; the distal margins of the female wing are greatly expanded and correlate well with female size (Svensson & Peterson 1988). Although Svensson et al. (1989) found that males prefer to mate with the largest females (i.e. those females with the largest wings), they also found that mated females had less wing wear than unmated females and thus were probably younger. If male E. borealis use wing size as a measure of female size when selecting mates, older females with more wing wear and thus smaller wings would be at a selective disadvantage. Yet older females may be more likely to bear larger, more mature eggs. It is also possible, however, that E. borealis females regularly produce a second clutch, and that older females with more wing wear are more likely to have already laid their first clutch and to offer only immature eggs of the second clutch to choosy males. More research is needed to be able to make this distinction. In comparison, dissections of R. longicauda suggest that (1) either some females are consistently rejected by males to the extent that they go long periods without proteinaceous food and consequently mature their first egg clutch very slowly, or (2) a substantial proportion of the females in a lek are seeking food for the development of their second clutch. Dishonest signalling about the state of egg development in R. longicauda can be labelled true behavioural deception as defined by Semple & McComb (1996) because the receiver of the signal, the mate-seeking male, 419 420 ANIMAL BEHAVIOUR, 59, 2 registers a situation that does not actually occur; that is, males that approach inflated females because they have swollen abdomens do not have the same probability of mating with a female whose eggs are near oviposition as do R. sociabilis males that approach females whose abdomens are distended with nearly mature eggs rather than air. As a result of inflation, R. longicauda females benefit from the acquisition of food, while males pay the cost of having hunted for and fed a female with a lower likelihood of surviving to oviposition and which will almost certainly cuckold him. One might argue that abdominal inflation in R. longicauda is not a dishonest signal because a regression indicates 23% of the variation in egg size is explained by abdomen size after inflation. However, this does not explain the evolution of abdominal inflation in the first place; comparisons with R. sociabilis clearly show that inflatable sacs could not have evolved to advertise egg size honestly because inflation decreases the accuracy of this signal over that conveyed by uninflated females by 49%! If signal accuracy drove selection of this trait, males preferring uninflated abdomens would have been able to identify females with mature eggs more often than males with a preference for inflated abdomens, and thus would have enjoyed the fitness advantages of higher paternity levels and greater offspring survivorship. It is more parsimonious to argue that inflatable sacs evolved precisely because males could not distinguish between abdomens inflated with air and abdomens swollen with mature eggs; that is, because inflatable sacs successfully deceived mateseeking males, females that bore them gained a considerable advantage over females that did not. Rather than indicating that there is no deception imparted by abdominal inflation, the fact that males continue to gain a small advantage by attempting to select the largest inflated abdomens instead suggests that the deception from abdominal inflation is not perfect. Deceptive sexual signals may carry important ecological costs for the Rhamphomyia females that employ them. One obvious consequence of inflating the abdomen to a flattened disc-like structure is to sacrifice aerodynamic competence in comparison to flies with the uninflated ancestral state. Inflated R. longicauda females do not appear to fly or manoeuvre as quickly as uninflated R. longicauda males or females of other sex-role-reversed empidines that have not evolved inflatable abdominal sacs. Inflated females, therefore, might be more vulnerable to bird predation. We suspect that, unlike other empidines that typically lek for long periods throughout the day (Aldrich & Turley 1899; Downes 1970; Svensson & Petersson 1987; Evans 1988; Cumming 1994), R. longicauda lek only during dusk as this is the only time of daylight during which most avian predators do not forage. Males of some empidine species have been reported to be relatively scarce towards the end of the swarming period (Laurence 1952; Chvála 1980, as cited in Cumming 1994; Svensson & Petersson 1987). Cumming (1994) suggests that this may be explained by a combination of protandry and a higher risk of mortality associated with hunting (compared with the relatively stationary, nonhunting females). However, Fig. 6 indicates that R. longicauda males predominate at lek sites both early in the season and very late in the season, suggesting both protandry and greater longevity in males. This observation is consistent with the idea that females of sex-role-reversed empidines might be at higher risk of predation because of the burden of ornamentation. Several hypotheses attempt to explain the evolution of female secondary sexual characters in dance flies (reviewed by Cumming 1994). It has been proposed that traits such as enlarged wings, pinnate leg scales and eversible abdominal sacs (1) enable lekking females to fly more aerodynamically (Gruhl 1963; Svensson & Petersson 1987), (2) act as premating isolating mechanisms (Downes 1970; Chvála 1990), or (3) evolved to attract small moths to communal swarms where they could be preyed upon by male dance flies (van der Goot & de Vos 1988). We agree with Cumming (1994), however, that secondary sexual traits in female dance flies function in the same way that secondary sexual traits do in the males of countless animal species; to signal, either honestly or deceptively, fitness while competing for mates. We conclude that male control of all proteinaceous resources in concert with selection for males to favour females with large abdomens has led to the reversal of normal male–female courtship roles in R. longicauda and intense intrasexual competition among females for access to prey-bearing males. Such intense selection, in turn, has favoured females with the ability to manipulate the size of their abdomen at will, artificially inflating it before competing with other females in leks and further accentuating its bizarre size with the judicious posture of fringed legs. Because the misrepresentation of abdomen size is best displayed in silhouette through backlighting, groundbased lek markers so common in other swarming Diptera have been abandoned in R. longicauda in favour of distinct patches of sky framed by dark canopy leaves. Against this background hovering females repeatedly conceal the true state of egg development from mate-seeking males. In so doing, even females whose eggs are many days from maturity acquire the protein required to complete oogenesis at least once before death. Thus, abdominal inflation in female R. longicauda appears to be one of the most successful dishonest sexual signals known. Acknowledgments We thank J. D. Newbold and E. L. White for statistical advice, R. G. Weber for field assistance, Darryl Gwynne and two anonymous referees for helpful comments on the manuscript, and the Stroud Foundation for providing facilities. We are especially grateful to Howard Evans for supplying the necessary R. sociabilis females for dissection and for helpful suggestions throughout the project. Published as paper No. 1658 of the Delaware Agricultural Experiment Station, contribution No. 716 of the Department of Entomology and Applied Ecology, University of Delaware, Newark. References Alcock, J. & Gwynne, D. T. 1991. Evolution of insect mating systems: the impact of individual selectionist thinking. In: FUNK & TALLAMY: RHAMPHOMYIA COURTSHIP REVERSAL Reproductive Behaviour of Insects: Individuals and Populations (Ed. by W. J. Bailey & J. Ridsdill-Smith), pp. 10–41. New York: Chapman & Hall. Aldrich, J. M. & Turley, L. A. 1899. A balloon-making fly. American Naturalist, 33, 809–812. Basalo, A. 1990. Female preference predates the evolution of the sword in swordtail fish. Science, 25, 808–810. Chvála, M. 1976. Swarming, mating and feeding habits in Empididae (Diptera), and their significance in evolution of the family. Acta Entomologica Bohemoslovaca, 73, 353–366. Chvála, M. 1990. Mating swarms of Diptera, effective ethological isolating mechanisms. Acta Universitatis Carolina, Biologica, 33, 449–454. Cumming, J. M. 1994. Sexual selection and the evolution of dance fly mating systems (Diptera: Empididae; Empidinae). Canadian Entomologist, 126, 907–920. Darwin, C. 1871. The Descent of Man and Selection in Relation to Sex. Princeton, New Jersey: Princeton University Press. Downes, J. A. 1970. The feeding and mating behaviour of the specialized Empidinae (Diptera); observations of four species of Rhamphomyia in the high arctic and a general discussion. Canadian Entomologist, 102, 769–791. Eberhard, W. G. 1996. Female Control: Sexual Selection by Cryptic Female Choice. Princeton, New Jersey: Princeton University Press. Evans, H. E. 1988. Observations on swarms of Rhamphomyia sociabilis (Williston) (Diptera: Empididae). Journal of the New York Entomological Society, 96, 316–322. Gill, F. B. 1995. Ornithology. New York: W. H. Freeman. van der Goot, V. S. & de Vos, R. 1988. Zwermen van Rhamphomyia marginata in de schemering (Diptera: Empididae). Entomologische Berichten, 48, 49–52. Gruhl, K. 1963. Sind Tanze von Weibchen bei Empis und Rhamphomyia (Diptera) an gewisse Sondermerkmale gebunden? Deutsche Entomologische Gesellschaft Mitteilungen, 22, 88–96. Gwynne, D. T. 1991. Sexual competition among females: what causes courtship-role reversal? Trends in Ecology and Evolution, 6, 118–121. Gwynne, D. T. & Simmons, L. W. 1990. Experimental reversal of courtship roles in an insect. Nature, 346, 172–174. Hatziolos, M. E. & Caldwell, R. L. 1983. Role reversal in courtship in the stomatopod Pseudosquilla ciliata (Crustacea). Animal Behaviour, 31, 1077–1087. Ichikawa, N. 1988. Male brooding behaviour of the giant water bug Lethocerus deyrollei Vuillefroy (Hemiptera: Belostomatidae). Journal of Ethology, 7, 113–117. Itô, Y. & Yamagishi, M. 1989. Sperm competition in the melon fly, Dacus cucurbitae (Diptera: Tephritidae): effects of sequential matings with normal and virgin or non-virgin sterile males. Applied Entomology and Zoology, 24, 466–477. Laurence, B. R. 1952. The prey of some Empididae and Dolichopodidae (Dipt.). Entomologist’s Monthly Magazine, 88, 156–157. Lorch, P. D., Wilkinson, G. S. & Reillo, P. R. 1993. Copulation duration and sperm precedence in the stalk-eyed fly Cyrtodiopsis whitei (Diptera: Diopsidae). Behavioral Ecology and Sociobiology, 32, 303–311. Mora, G. 1990. Paternal care in a neotropical harvestman, Zygopachylus albomarginis (Arachnida, Opiliones: Gonyleptidae). Animal Behaviour, 39, 582–593. Newkirk, M. R. 1970. Biology of the long-tailed dance fly, Rhamphomyia longicauda (Diptera: Empididae): a new look at swarming. Annals of the Entomological Society of America, 63, 1407–1412. Opp, S. B., Ziegner, J., Bui, N. & Prokopy, R. J. 1990. Factors influencing estimates of sperm competition in Rhagoletis pomonella (Walsh) (Diptera: Tephritidae). Annals of the Entomological Society of America, 83, 521–526. Otronen, M. 1989. Female mating behaviour and multiple matings in the fly Dryomyza anilis. Behaviour, 111, 77–97. Otronen, M. 1990. Mating behavior and sperm competition in the fly, Dryomyza anilis. Behavioral Ecology and Sociobiology, 26, 349– 356. Parker, G. A., Baker, R. R. & Smith, V. G. F. 1972. The origin and evolution of gamete dimorphism and the male–female phenomenon. Journal of Theoretical Biology, 36, 529–533. Reynolds, J. D. 1985. Philandering phalaropes. Natural History, 94, 58–65. Ridley, M. 1989. The incidence of sperm displacement in insects: four conjectures, one corroboration. Biological Journal of the Linnaen Society, 38, 349–367. Rosenqvist, G. 1990. Male mate choice and female–female competition for mates in the pipefish Nerophis ophidion. Animal Behaviour, 39, 1110–1115. Ryan, M. J. & Keddy-Hector, A. 1992. Directional patterns of female mate choice and the role of sensory biases. American Naturalist Supplement, 139, 4–35. Semple, S. & McComb, K. 1996. Behavioural deception. Trends in Ecology and Evolution, 11, 434–437. Simmons, L. W. & Bailey, W. J. 1990. Resource influenced sex roles of zaprochiline tettigoniids (Orthoptera, Tettigoniidae). Evolution, 44, 1853–1868. Sivinski, J. M. & Petersson, E. 1997. Mate choice and species isolation in swarming insects. In: The Evolution of Mating Systems in Insects and Arachnids (Ed. by J. C. Choe & B. J. Crespi), pp. 294–309. Cambridge: Cambridge University Press. Smith, R. L. 1997. Evolution of paternal care in the giant waterbugs (Heteroptera: Belostomatidae). In: The Evolution of Social Behavior in Insects and Arachnids (Ed. by B. J. Crespi & J. C. Choe), pp. 116–149. Cambridge: Cambridge University Press. Steyskal, G. 1941. A curious habit of an empidid fly. Bulletin of the Brooklyn Entomological Society, 36, 117. Svensson, B. G. & Petersson, E. 1987. Sex-role reversed courtship behaviour, sexual dimorphism and nuptial gifts in the dance fly, Empis borealis (L.). Annales Zoologici Fennici, 24, 323–334. Svensson, G. C. & Petersson, E. 1988. Non-random mating in the dance fly Empis borealis: the importance of male choice. Ethology, 79, 307–316. Svensson, B. G., Petersson, E. & Forsgen, E. 1989. Why do males of the dance fly Empis borealis refuse to mate? The importance of female age and size. Journal of Insect Behavior, 2, 387–395. Thomas, L. 1995. Parental care in assassin bugs (Reduviidae). Ph.D. thesis, University of Cambridge. Thornhill, R. 1986. Relative parental contributions of the sexes to offspring and the operation of sexual selection. In: Evolution of Animal Behavior: Paleontological and Field Approaches (Ed. by M. N. Nitecky & J. A. Kitchell), pp. 113–135. Oxford: Oxford University Press. Trivers, R. L. 1972. Parental investment and sexual selection. In: Sexual Selection and the Descent of Man (Ed. by B. Campbell), pp. 136–179. Chicago: Aldine–Atherton. Wells, K. D. 1978. Courtship and parental behavior in a Panamanian poison-arrow frog (Dendrobates auratus). Herpetologica, 34, 148– 155. Williams, G. C. 1966. Adaptation and Natural Selection. Princeton, New Jersey: Princeton University Press. Zahavi, A. 1975. Mate selection: a selection for a handicap. Journal of Theoretical Biology, 53, 205–214. Zahavi, A. & Zahavi, A. 1997. The Handicap Principle. Oxford: Oxford University Press. 421