Most closely related species of insects differ exclusively or predominantly with regard to sexually dimorphic traits 12. This observation implies that these traits evolve fast and are likely to be involved in speciation. Understanding the evolution of sexual dimorphisms is thus crucial for understanding the impressive species diversity of the Insecta, the most speciose clade of multicellular animals 3. This has made the study of sexual dimorphisms also one of the most important goals of modern evolutionary biology.

Much of the literature focuses on the theoretical issues that drive the sexual selection that is responsible for the origin and maintenance of sexual dimorphisms. Traditionally, the fast evolution of male ornaments had been attributed to selection via female choice 4. Some empirical data and genetic models had shown that this choice could be driven by direct benefits that increase survivorship 56 or indirect genetic benefits that could improve viability and/or the attractiveness of the offspring 789101112. However, there is an alternative explanation for the fast evolution of sexual dimorphisms. The unequal investments in gametes by males and females leads to a conflict of interest between the sexes ("sexual conflict" 1314). Males have a higher interest in multiple matings than females and may thus try to coerce females into being polygamous. As a result both sexes can become involved in a sexually antagonistic arms race and evolve mechanisms to gain control over reproduction, regardless of the costs to the other sex 1516. Evidence for such morphological coevolution between male and female reproductive characters has been found across a number of insect groups like Coleoptera (e.g.: Bruchidae and Dytiscidae), Hemiptera (e.g.: Gerridae) and Diptera (e.g.: Empididae and Scathophagidae) 171819202122232425.

However, comparatively little is known about the function of many sexually dimorphic structures and even less is known about how the function of these structures changes over evolutionary time. One reason for this gap in our knowledge is that studying the evolution of function requires concurrent information on the morphology, behavior and phylogeny of multiple species. Only then can functional changes be mapped onto a phylogenetic hypothesis. Unfortunately, for many sexually dimorphic features, only the morphology has been documented in detail and it is often unclear how they are being used. This is particularly so for male genitalia whose function is difficult to study because they interact with internal elements of the female reproductive tract 2627. Even if the function for a particular structure has been studied in great detail, information is usually only available for one or a few species 28 and if there is information for several species, the phylogenetic relationships are frequently unknown. As a consequence there are only few studies that have documented the evolution of the function of sexually dimorphic structures 182930313233. Here, we use a wealth of information on the morphology and function of the male forelegs for 31 species of Sepsidae (Diptera: Cyclorrhapha) to document how sexual dimorphisms are correlated with mating behavior over evolutionary time.

Over the past 30 years, sepsid flies have become model organisms in the study of behavior in general and sexual selection in particular 34353637383940414243. Sepsids are ideal models because some of the structures involved in the interactions between males and females are easily accessible and display stark sexual dimorphism 4445. Furthermore, sepsids can be maintained in the laboratory and some aspects of their mounting and copulation behavior can be video-taped and studied without dissection. The most interesting sexually dimorphic structure that is easily accessible is the forelegs. In most sepsid species, the males use their forelegs for grasping the female wing base prior to copulation. The forelegs are often strongly sexually dimorphic, with female legs being unmodified while the male legs display a wide variety and species-specific cuticular outgrowths, indentations, and/or modified bristles (see Figures 1 and 2). The males of most sepsid species will indiscriminately jump on females and attempt to clamp to the female's wing base. It is here that the morphological modifications of the male forelegs play a major role. They tightly mesh with the wing veins and cells of the female wing 4445 and help the male during the next stage of mating, a rodeo-like struggle that is found in most species and that starts after the male has mounted the female 34464748. The females of many sepsid species use vigorous shaking in order to dislodge and/or test the stamina of the male 40434950. In addition, females usually bend their abdomen ventrally so that the males cannot establish genitalic contact without female consent 3442. Copulation can only be initiated once the female lifts the tip of her abdomen, which may or may not occur in response to male behaviors 385152. Many copulation attempts by males are ultimately aborted either because the males voluntarily dismount or because males are dislodged through shaking 38.

Sexually dimorphic forelegs are found in the vast majority of sepsid species. However, monomorphic legs are known for six species: Orygma luctuosum Meigen 1830, three species of Perochaeta Duda 1926, Sepsis duplicata Haliday 1838, and S. secunda (Melander et Spuler 1917). Here, we document the foreleg morphology and mounting behavior for two outgroup species, 3 species of Sepsidae with monomorphic legs and 28 species of Sepsidae with sexually dimorphic legs. We reconstruct the phylogenetic relationships among all species based on sequence data for 10 genes, map foreleg morphology and mounting behavior onto the phylogenetic tree, and use a concentrated change test 53 to establish whether the gains and losses of dimorphic forelegs are correlated with changes in how the forelegs are used during mounting.

Mounting behavior in Sepsidae is strictly and statistically significantly correlated with the origin and losses of sexual dimorphisms. The "typical" mounting position of Sepsidae involving the wing grasp only evolved once and within the family (see Figure 2; red hatchmark) and the origin of this mounting position coincides with the origin of sexually dimorphic forelegs (see Figure 2; green hatchmark). The males of the three sepsid species with monomorphic forelegs, Orygma luctuosum, Perochaeta dikowi, and Sepsis secunda place the forelegs on different parts of the female body. Given its placement on the tree, the mounting behavior and lack of sexual dimorphism is plesiomorphic in O. luctuosum. In fact, not only do the forelegs of O. luctuosum closely resemble the legs of the outgroups (see Figure 2) but the mounting behavior is also very similar to the behaviors of the outgroups Gluma nitida 66 and Willistoniella pleuropunctata (see Figure 4). For instance, male coelopids have been observed to mount females and wrap their legs around the female thorax and as observed in O. luctuosum, the females will 'kick' the males with their midlegs and even roll over 6768. Other acalyptrate flies like Protopiophila litigata (Diptera: Piophilidae) and Dryomyza anilis (Diptera: Dryomyzidae) also exhibit similar mounting behaviors. For example, upon mounting, P. litigata males rest their fore tarsi on the female postpronotum 69 while in D. anilis, the male forelegs are placed on the female head 70.

However, the monomorphic legs of Perochaeta dikowi and Sepsis secunda are due to the convergent loss of the male leg armature in two distantly related clades of Sepsidae. A study of the mating behavior reveals that in both species, the forelegs are no longer used for grasping the female wingbase. One can imagine many different ways of replacing the typical sepsid mounting position with a new mounting technique. Yet, P. dikowi and S. secunda have convergently invented near identical, new behaviors. In both species, the male approaches the female from behind. In both species, the males curl their abdomens forward in order to touch the female abdomen with their claspers. In both species, it is only afterwards that the male legs play a minor role in mating: in P. dikowi, males rest their fore tarsi on the female postpronotal callus whilst the males of S. secunda rest their fore tarsi mid-way along the female wing margin. Convergence in morphology is often driven by similarities in habitat and environment. However, there are no aspects of the known autecology of these two species that could explain the convergent evolution of the mounting behavior. Both are typical scavenger flies that use cow dung as breeding substrate. Sepsis secunda is mostly found on open pastures in hot and dry areas of North America, while P. dikowi is only known from forests at mid-elevation sites in the tropics 64. Furthermore, both species co-occur in the same habitat with other sepsid species that have the "normal" wing grasp.

Evolutionary biologists have long predicted and documented convergence in sexually dimorphic structures 45. Some authors attribute the convergent evolution of correlated traits to highly conserved patterns of pleiotropy as a potential mechanism for channeling sexual shape dimorphism in Diptera 71. These hypotheses address the proximate causes for the convergent evolution of sexual dimorphisms and should be complemented with studies that can address the ultimate causation. We suggest that the ultimate reasons for why similar genetic architecture may be utilized to produce convergent dimorphisms are convergent changes in mating behavior. In our case the convergent loss of the wing clamp leads to the convergent loss of the leg dimorphisms. Yet, more intriguing is our finding of convergent evolution of a very specific new mounting behavior which may suggest that highly conserved patterns of pleiotropy and a conserved genetic architecture may also exist for behavior.

We had previously outlined that there is only scant information on how the function of sexually dimorphic traits change over time. Our study helps to fill the void. In the sepsid outgroups and Orygma luctuosum the fore femora and tibiae of the males function like the same structure in most flies. However, in most species of Sepsidae the male forelegs acquire an additional function, i.e., the clamping of the females' wingbase and as a result a new sexual dimorphism evolves. Subsequently, the male forelegs undergo extensive and fast modification as revealed in Figure 1 where all 28 species with sexually dimorphic legs have species-specific modifications. However, as P. dikowi and S. secunda document sexual dimorphisms can also disappear once a structure loses its role in mating behavior.

There are still relatively few studies that document how a behavioral change is correlated with a morphological change 18293132. However, a similar case is also known from seaweed flies (Diptera: Coelopidae) where the male foretarsus touches the female antenna during mating 72. A new morphological structure, a thumbnail-like process, evolved on the basitarsus of males and it must be evolving rapidly because it differs between the different species of Coelopidae 73. The antlered flies in the genus Phytalmia (Diptera: Tephritidae) provide another example. Males have cuticular projections on the head (antlers) that are used in male-male combat for territory and the complexity of the antler shapes differs quite drastically between species 74. Species with complex antlers have simple behavioral repertoires whilst those with simpler antlers evolved more complex behaviors involving the fore, mid and hind legs (e.g.: 'wing-locking' and 'stilting' 33).

The novel mounting technique in Perochaeta dikowi and Sepsis secunda is not only associated with the loss of sexually dimorphic legs, but also associated with the gain of new sexual dimorphisms. In P. dikowi the males have modified fourth or fifth sternites that form brushes. These are used during copulation 64. In S. secunda, the males have sexually dimorphic 4^th tergites. Normally, this tergite is monomorphic which is not surprising given that the dorsal side of the male abdomen is not involved in the mating behavior of most sepsids. However, in S. secunda the male tergites touch the female when the male abdomen is curled forward [see Additional file 8] (or visit the following website 65). A new dorsal, bump-like extension of the tergite four evolves and this tergite becomes strongly sexually dimorphic (see Figure 5; SEM picture). In addition, S. secunda has an elongated abdomen (elongated segments I & II). In particular, the posterior part of the segment II is strongly lengthened and connected to a distinctly constricted tergite three (see Figure 5; red arrow). This constriction is likely related to the new mating behavior that requires the male to bend his abdomen forward. Indeed, this constriction closely resembles the wasp waist of parasitoid wasps that need to have similar flexibility in order to use their ovipositor for injecting poison and/or eggs into a host. It appears likely that these morphological changes in S. secunda occurred rapidly because they are not found in its sister group, S. duplicata, and the two species are very similar with regard to the barcoding gene COI (1.97%; normal distances in most closely related Diptera species > 3% 75).

Throughout the discussion, we often implicitly assumed that the change in behavior preceded the loss of the sexual dimorphism although technically the strict correlation between the two traits could also imply the reverse order. However, based on functional considerations it appears unlikely that the reduction of the male leg armature preceded the behavioral change given that the leg armature has been experimentally shown to be important for successful mating in sepsids 37445176. Reversing the order of events would also not alter our finding of a strict correlation between mounting behavior and the losses and gains of sexual dimorphisms in Sepsidae.

Several groups of Diptera are important models for investigating sexual conflict. A number of studies have documented coevolution between male and female reproductive characters. For instance, Minder et al. (2005) show that the sperm length in males evolves with the length of spermathecal ducts in females across several species of yellow dung flies (Scathophagidae) 22. In Drosophila melanogaster (Drosophilidae) female resistance evolves in response to the harm caused by male accessory gland proteins that reduce female fitness and are transferred during copulation 77787980. Furthermore, Bonduriansky and Rowe (2005) argued that intralocus sexual conflict reduces the heritability of sexual traits in Prochyliza xanthostorna (Piophilidae) and Otronen (1994) suggested that inter-sexual conflict over mating patterns led to repeated copulations in Dryomyza anilis (Dryomyzidae) 81. In Sepsidae, some authors argued for sexual conflict 384782 while others claimed that the evolution of sexual dimorphisms is driven by female choice 3537.

In our study, we have identified several elements that are characteristic of an arms race: morphological variation among close relatives as well as a correlation between morphology and behavior. However, we only documented this coevolution within the male sex and in order to document sexual conflict, we would also need to document coevolutionary responses in the females (see 45). In the absence of this evidence, we cannot attribute the diversity of male foreleg armature to sexually antagonistic coevolution.

One of the most powerful indirect techniques in evolutionary biology is using the comparative method to investigate convergent evolution 83. By combining behavioral, morphological, and phylogenetic data, we are here able to show that sexual dimorphisms evolve fast and convergently in Sepsidae. Homoplasy is not restricted to sexual dimorphisms, but also affects mounting behavior. Our study furthermore documents how important it is to study the function of sexually dimorphic structures.

NP conducted all the mating behavior experiments in this paper. Both NP and KF–YS were responsible for the morphological documentation of the various sexually dimorphic and monomorphic structures. The molecular work i.e. DNA matrix assembly was primarily carried out by KF–YS with help from Ms. Sujatha N. Kutty. RM contributed to the conception of this research and performed all the phylogenetic analyses in the paper. All authors read and approved the final manuscript.