Dermaptera (earwigs) is a relatively small polyneopteran order with approximately 2200 described species. They are characterized by a pair of forceps, which are hardened, unsegmented cerci at the caudal end of the abdomen. In most species, males have more exaggerated forceps than females, indicating an effect of sexual selection on them. Earwigs also exhibit astonishing diversity in the number, laterality and size of both male and female genital components. This characteristic has promoted the study of postcopulatory sexual selection in several representative species. Here, previous studies of earwigs that examined pre- and postcopulatory sexual selection are reviewed in detail. Related topics included here are sexually antagonistic coevolution, evolution of laterally asymmetrical morphologies, and developmental aspects of intra-sexually dimorphic traits. A new terminology system for male genitalia is also proposed.
Dermaptera (earwigs) is a polyneopteran insect order (Grimaldi & Engel 2005) with approximately 2200 described species from mainly tropical and warm temperate regions (Sakai 1982, 1985, 1987a, 1990, 1991, 1992, 1993, 1994, 1995a,b,c,d; Steinmann 1986, 1989a,b, 1990, 1993; Popham 2000; Haas et al. 2012). Most earwig species are considered omnivorous or carnivorous insects that live in various types of natural or seminatural environments (Günther & Herter 1974; Renz & Kevan 1991). Although several species are considered pests in gardens and agriculture (e.g. Nala lividipes (Dufour) (Labiduridae); e.g. Cooper 1992) or pest control agents (e.g. Labidura riparia (Pallas) (Labiduridae); e.g. Kharboutli & Mack 1993), most earwig species have no direct relationships with human activities (e.g. Günther & Herter 1974; Renz & Kevan 1991; Costa 2006). Accordingly, basic bionomics have been studied in only a few common species and published information on the ecology of most species is extremely scarce. However, the following three aspects of Dermaptera make this relatively small and poorly understood insect order a suitable model for ecological and evolutionary studies. First, all of the species examined to date show various degrees of maternal care of offspring (cf. Shimizu & Machida 2011a). Devoting more than 30 pages in his book, Costa (2006) provided a contemporary and comprehensive review of parental care in earwigs. Nevertheless, many important studies published in recent years require updates. They include studies on the costs and benefits of maternal care (Suzuki et al. 2005; Kölliker 2007; Kölliker & Vancassel 2007; Miller et al. 2011; Meunier & Kölliker 2012a; Miller & Zink 2012), cannibalism among family members (Suzuki et al. 2005; Dobler & Kölliker 2010; Miller & Zink 2012), provisioning behavior and its regulation (Staerkle & Kölliker 2008; Mas et al. 2009; Suzuki 2010, 2011; Mas & Kölliker 2011; Meunier & Kölliker 2012b; Wong & Kölliker 2012), evolution of the number of reproductive cycles (Meunier et al. 2012), possible viviparity in a tropical species (Kočárek 2009) and reproduction in poorly known primitive representative species (Matzke & Klass 2005; Shimizu & Machida 2009, 2011a, b). Second, earwigs are characterized by their forceps, which are specialized cerci located at the caudal end of the abdomen (Fig. 1A–E). Although the forceps are usually heavily sclerotized and unsegmented, segmented cerci have been reported during the larval stages of Diplatyidae (e.g. Shimizu & Machida 2011b) and Karschiellidae (at least in the distal part; Haas et al. 2012). Like the exaggerated horns of horned beetles and the mandibles of stag beetles, the forceps of earwigs, especially those of males, are sometimes conspicuously exaggerated and therefore have attracted the interest of many biologists. Third, the order Dermaptera is also characterized by a high level of diversity in genital structures, including the size, number and laterality of both male intromittent organs and female sperm receptacle/storage organs (spermathecae), making them valuable study models of genital evolution.
In this paper, I review studies related to the second and third earwig characteristics that were conducted in relation to sexual selection. Sexual selection is a mode of natural selection in which some individuals have an advantage over others of the same sex in a population because they are better at securing mates (Darwin 1871). Darwin (1871) recognized two major categories of sexual selection that operate in precopulatory phases: male–male competition (intrasexual selection) and female choice (intersexual selection). Because of the much higher production rate of sperm than ova, males usually compete for access to females, while females are usually choosy about male mates (Andersson 1994; Parker & Simmons 1996). These processes cause evolution in weaponry traits in males or genetically correlated evolution between male ornaments and female preferences, given heritable variation in these traits (Fisher 1958; Zahavi 1975; Pomiankowski 1988; Pomiankowski et al. 1991). Therefore, sexual selection can be operationally defined as differences in fecundity (including the number of grandchildren) caused by heritable differences in access to mates (Arnold & Wade 1984; Arnold 1994). The processes of sexual selection continue into the postcopulatory phases when females potentially mate with two or more males. Spermatozoa from multiple males compete to fertilize ova (sperm competition, hereafter SC; Parker 1970). Since Waage's (1979) seminal work, evidence mounts that males develop many behavioral and morphological adaptations for sperm competition, including mate-guarding, copulatory plugs and displacement of rival sperm from female sperm storage organs (Simmons & Siva-Jothy 1998; Simmons 2001). On the other hand, females sometimes terminate mating before full insemination, eject or consume sperm after insemination and control the pattern of sperm usage, actively influencing paternity consequences (reviewed by Eberhard 1985, 1996). If heritable variation in male abilities to overcome such female-born “trials” exists, polyandrously mated females “automatically” gain genetic benefits by having sons that inherit the superior abilities of their fathers (“sexy-sons” effects; Keller & Reeve 1995). In addition, when such abilities are positively and genetically correlated with other male or female fitness measures, further benefits arise (“good-genes” effects; Yasui 1997). The term cryptic female choice (CFC, coined by Thornhill 1983) or sperm choice refers these processes, which can be seen as the postcopulatory version of sexual selection through female choice. In this paper, I adopt the broadest definition of CFC; that is, CFC is any female morphology, physiology or behavior that works during or after copulation and increases the chance of having superior offspring. This definition includes any indirect (following the definition of Wiley & Poston 1996) postcopulatory processes that lack discriminative responses to sperm or sperm donors but which cause a correlation between a focal female trait and obtained sperm quality. Although the dichotomous categorization into pre- (male–male competition and female choice) and postcopulatory (SC and CFC) sexual selection is convenient, as discussed below, evolutionary changes in precopulatory processes can affect postcopulatory sexual selection, and vice versa. Growing evidence is demonstrating the importance of another type of sex-related selection, sexual conflict, as a compelling force of antagonistic coevolution between male and female sexual characters (sexually antagonistic coevolution, or SAC; Hosken & Stockley 2004; Arnqvist & Rowe 2005; Parker 2006). Usually, males try to mate at a higher frequency than females; therefore, they sometimes develop specialized adaptations to forcibly mate with unwilling females. Like female choice and CFC, this process can result in correlated evolution between male (offensive) and female (defensive) traits (SAC), and genetic correlations between them (Arnqvist & Rowe 2005; Eberhard 2010). However, unlike CFC, multiple matings by females are not a prerequisite for SAC (Parker 2006). In this review, studies on earwigs that examined these sex-related selection processes or the evolution of sexually dimorphic characters are discussed in detail.
Taxonomic overview of earwigs
Dermaptera are usually subdivided into three taxa: Hemimerina, Arixeniina and Forficulina. Members of the former two taxa live on mammals (hamster rats Cricetomys spp. in Africa and bats Cheiromeles torquatus Horsfield in Asia, respectively) and show many adaptations to their phoretic–epizoic lifestyle (Nakata & Maa 1974). According to the classification system of Sakai (1982), Forficulina comprises eight families of typical free-living earwigs: Pygidicranidae, Diplatyidae (=Diplatyinae of the Pygidicranidae (Steinmann 1989b)), Anisolabididae (=Carcinophoridae (Steinmann 1989b)), Apachyidae, Labiduridae and Spongiphoridae (=Labiidae (Steinmann 1989b)), Chelisochidae and Forficulidae. Recent studies (e.g. Popham 2000; Haas & Klass 2003; Jarvis et al. 2005) have treated Karschiellinae of the Pygidicranidae in Sakai's (1982) system as a ninth family Karschiellidae; here, I follow this view. More recent studies have suggested that both Hemimerina and Arixeniina are in-group members of Forficulina (Klass 2001; Haas & Klass 2003; Jarvis et al. 2005), but their phylogenetic placements in Forficulina have not been resolved. To my knowledge, no comprehensive study has been conducted for any aspects of sexual selection in these epizoic earwig taxa (except Ashford 1970 and Burr & Jordan 1913, which provide descriptions of the unique mating postures of Hemimerina and Arixeniina, respectively). Therefore, all of the studies discussed here in detail examined free-living, typical earwigs that belong to Forficulina.
Although the phylogenetic relationships among forficuline families have not been resolved, previous studies based on morphological and/or molecular traits strongly support the monophyly of Eudermaptera (Spongiphoridae + Chelisochidae + Forficulidae; Sakai 1987b; Haas 1995; Haas & Kukalová-Peck 2001; Colgan et al. 2003; Haas & Klass 2003; Kamimura 2004b) and higher Forficulina (Apachyidae + Anisolabididae + Labiduridae + Eudermaptera; Popham 1985; Sakai 1987b; Haas 1995; Haas & Kukalová-Peck 2001; Haas & Klass 2003; Jarvis et al. 2005). Figure 2A shows the presently most elaborate phylogenetic hypothesis of Forficulina (modified from Haas & Klass 2003). Several studies have suggested possible paraphyly for Diplatyidae (Haas 1995; Haas & Kukalová-Peck 2001; Haas & Klass 2003) and Pygidicranidae (Haas & Kukalová-Peck 2001; Haas & Klass 2003), and polyphyly for Labiduridae (Haas & Kukalová-Peck 2001) and Spongiphoridae (Jarvis et al. 2005).
As I introduce its mating biology in some detail below, the European earwig Forficula auricularia Linnaeus (Forficulidae) is the most intensively studied earwig species for almost all biological aspects (Sakai 1995c; Costa 2006). Wirth et al. (1998) revealed that two sibling species are present in the European earwig: species A is semelparous (produces an egg batch only once), except for Mediterranean populations, and lacks an overwintering diapause. Species B has two reproductive cycles with reproductive diapauses (Wirth et al. 1998; Guillet et al. 2000). Both of these species invaded and became established in North America (Wirth et al. 1998). Crossing experiments revealed not complete but sizable reproductive isolation between these two cryptic species (Wirth et al. 1998). To my knowledge, no taxonomical rearrangement has been published for these two sibling species, including the determination of which species corresponds to the species that Linnaeus (1758) had originally described as F. auricularia. Unfortunately, not only studies published before Wirth et al. (1998), but many recent studies, lack an explicit declaration of which sibling species (A or B) was used. Thus, in this review, unless explicitly indicated in the original article, I simply refer to F. auricularia for the F. auricularia species complex.
Precopulatory sexual selection and evolution of sexually dimorphic forceps
Because of their usually nocturnal and elusive nature, male–male and male–female interactions in earwigs are believed to be largely dependent on tactile (or short-distance chemical) stimuli. One exception is the vibration of substrates (leaf sheathes of corn plants) by male Doru taeniatum (Dohrn), which may function as a short-distance courtship song (Briceño & Schunch 1988). Accordingly, almost all of the studies on precopulatory sexual selection in earwigs have focused on the function of the forceps, which are typically sexually dimorphic in shape. In most cases, the forceps of adult males are more exaggerated and ornamented with teeth or other structures, while those of adults females and nymphs are rather simple (Fig. 1A–E). These exaggerated male forceps can be used as weaponry during male–male competition for mates. Alternatively, they can be displayed to females during precopulatory courtship or used as mate-holding or copulatory courtship devices during copulation. Although many researchers have examined the exaggerated forceps of earwigs, comparative or experimental studies that tested explicit hypotheses have only commenced relatively recently. Only fragmentary and anecdotal descriptions of the uses of earwig forceps in relation to mating were reported before several groups started to publish important references almost simultaneously in the mid-1990s (see Briceño & Eberhard 1995 for a more detailed historical account). To examine the significance of forceps during male–male contests, some groups used staged mating experiments with two males without a female in an arena (♂–♂). However, male–male combat observed in this experimental design is not necessarily related to sexual selection. On the other hand, by using an experimental setting with two or more males and one or more female (♂–♂–♀), researchers can directly observe combat among males for securing a mate. However, any resulting bias in mating success could be caused by male–male competition or precopulatory female choice. In other words, compared to a “one male and one female” (♂–♀) design, a ♂–♂–♀ design allows a female to compare and choose between two males. In most earwig species, male–male or male–female interactions start with direct tactile contact between individuals, usually by the antennae (e.g. Briceño & Eberhard 1995). Thus, for such purposes, it is usually necessary to put a female and two or more males into a single enclosure that allows unconstrained interactions among them (cf. van Lieshout & Elgar 2009). Accordingly, the results of ♂–♂–♀ experimental designs need to be interpreted with caution.
Observations of social interactions in the spongiphorid Vostox apicedentatus (Caudell) by Moore and Wilson (1993) represent one of the first comprehensive studies on sexual selection in earwigs. Using virgin adults that emerged from field-collected nymphs, they observed interactions among individuals, especially interactions that involved forceps, under various social conditions: ♂–♂, ♂–♂–♀, ♂–♀ and ♀–♀. They found that one male tended to dominate in male–male contests and these males were also successful in mating, although no apparent indication of active female choice was noted. During combat, males extensively used their forceps to stroke or grab competitors. Females also used their forceps in ♂–♀ and ♀–♀ interactions, but much less frequently. As reviewed below, forceps length is usually positively correlated with male body size. Moore and Wilson (1993) did not measure any male or female trait sizes, including forceps length; thus, factors that affected the probability of winning in combat and mating success are unclear for this species.
Briceño and Eberhard (1995) conducted similar observations and described male–male combat, courtship behaviors and copulations in 13 Neotropical species belonging to four families (Pygidicranidae, Anisolabididae, Spongiphoridae and Forficulidae). Although their work was observational and did not involve any manipulation of the forceps, they identified seven male-specific forceps characteristics in three species that apparently function as weapons (or threating devices) against other males in male–male competition. For example, males of Skalistes inopinata (Burr) (Forficulidae) use a dorsal triangular projection of their forceps to threaten combatants. The forficulid genus Ancistrogaster is characterized by lateral extensions of the male's posterior abdominal tergites (Steinmann 1993; Sakai 1995a). Although it is not a forceps trait, in Ancistrogaster scabiosa Steinmann, Briceño & Eberhard (1995) observed that males raise their abdomen to display and make direct contact with their abdominal extensions. Similarly, the peculiarly shaped pygidium of Paralabellula dorsalis (Burmeister) (Spongiphoridae; referred to as Paralabella dorsalis in Briceño & Eberhard 1995; cf. Kevan & Vickery 1997), which develops between the base of the forceps in sex-specific shapes, apparently functions as a weapon during male–male combat. In contrast, Briceño & Eberhard (1995) recognized only two male-specific characteristics that function as courtship and mate-holding devices among the 13 species examined: the basal inner tooth of male forceps in P. dorsalis is used for courtship, and the widely curved male forceps in Pseudomarava prominens Steinmann (Spongiphoridae) are used to hold the mate during copulation. Because the forceps are located at the caudal end and male and female genitalia are located on the ventral side of the abdomen, male earwigs usually rotate their abdomen nearly 180° and walk backward to establish an end-to-end copulation posture (Fig. 1F). Female quiescence is usually necessary to establish genital coupling; thus, male earwigs usually cannot coercively mate with an unwilling female. However, males of P. prominens bite the antennae or legs of females with their mouthparts during the first stage of courtship. Females respond aggressively to this grabbing by directing their forceps at the males. Then, the males pinch the base of the female forceps with their widely curved forceps to establish genital coupling (Briceño & Eberhard 1995). Judging from their description, this species is unique in the occurrence of coercive mating by males. Recently, Shimizu and Machida (2011a) observed copulations in Apachyus chartaceus (De Haan) (Apachyidae). During copulation, the males also clasp the postabdomen of females with their forceps, suggesting a similar mate-holding function. The end-to-end mating posture of this species is also unique because the males do not twist their abdomen, and the whole bodies of a mating pair face in dorsoventrally opposite directions. Such a mating posture was reported for only another earwig species Sparrata bolivari Bormans (Spongiphoridae), which is also characterized by the dorsoventrally flattened body as Apachyus (Briceño & Eberhard 1995).
Within-sex variation in forceps
Moore and Wilson (1993) and Briceño and Eberhard (1995) showed that several male-specific characteristics of forceps do function as weaponry, courtship or mate-holding devices. However, these observational studies can not answer the question of whether males with larger, well-developed forceps enjoy enhanced mating success. For D. taeniatum, Briceño and Schunch (1988) compared forceps length between solitary males and males found with a female under leaf sheathes of corn plants in the field. They found that paired males had significantly longer forceps than solitary males. Although this result suggests that sexual selection was working on male forceps under natural conditions, forceps length was positively correlated with body size (measured as elytra length) and the observed trend may be confounded by the effect of body size. As shown below, positive correlations between forceps length and body size are common in earwigs.
Intraspecific (or static) allometry is usually described in terms of an allometric slope (b), based on the equation Y = aXb, or log(Y) = b log(X) + log(a), where Y and X are indices of trait size and body size, respectively. Isometry (b = 1) occurs when the ratio of a trait to body size remains constant across the range of body size. Positive (hyper-) allometry occurs when larger individuals have relatively larger traits (b > 1). To avoid underestimation of the true allometric slope (b), use of geometric (Model II) regression such as the reduced major axis (RMA) is recommended instead of ordinary least-squares (OLS) regression (Green 1999). Sexually selected characters used in precopulatory processes, such as exaggerated and bizarre structures used in combat or display, are usually characterized by positive allometry (Petrie 1988, 1992; Green 1992; Emlen & Nijhout 2000). However, sexual selection itself does not necessarily cause positive allometry in the selected trait, and positive allometry can evolve for reasons other than sexual selection (Bonduriansky & Day 2003; Bonduriansky 2007; Bertin & Fairbairn 2007). Using a simple allocation trade-off model, Bonduriansky and Day (2003) demonstrated that various types of allometric relationships, from negative ones (b < 1) to positive ones, can evolve in sexually selected traits depending on the exact modes of selection that are working on body and trait sizes. Their analysis specified the condition for evolution of positive allometry as a marginal fitness gain from an increase in relative trait size that is greater for larger than smaller individuals.
For 42 earwig species belonging to Apachyidae, Labiduridae, Spongiphoridae, Chelisochidae and Forficulidae, Simmons and Tomkins (1996) examined allometry in male forceps and elytra (or tegmina; forewings), which are considered to be sexually and a naturally selected traits, respectively, on body size (pronotum width). For species with dimorphic male forceps, only brachylabic morphs (males with shorter forceps) were measured and analyzed. Their analysis revealed that the slope was usually larger for forceps allometry than for elytra. In addition, species with more exaggerated male forceps tended to have a larger allometric slope for the forceps. This finding remained true after considering the phylogenetic relationships among the species (based on Sakai's 1982 system). Although the slope showed high variation, among the 42 species examined, only four species had negative slopes (b < 1; major axis regression), among which only one species (Labidura truncata Kirby; Labiduridae) showed significant negative allometry. Thus, according to Bonduriansky and Day (2003), their overall results support the general view that exaggerated male forceps are likely to confer greater fitness benefits to larger than smaller males. Tomkins and Simmons (1996) examined forceps allometry in six species that show dimorphisms in their male forceps (for F. auricularia, they examined two populations). Again, male forceps usually showed positive allometry, whereas no consistent differences in slopes and intercepts were found between brachylabic and macrolabic males (for discrimination of the two morphs, see below).
Van Lieshout and Elgar (2009) reported isometric relationships between body size (measured as pronotum width) and both right and left forceps length in an anisolabidid Mongolabis brunneri (Dohrn). (Van Lieshout and Elgar 2009 placed this species in the genus Euborellia; cf. Srivastava 1999). They also demonstrated that male forceps length is a significant determinant of male–male combat success in this species, while it did not affect precopulatory female choice (under ♂–♂–♀ conditions), mating frequency (under ♂–♀ conditions) or mating duration (under ♂–♀ conditions). As one possible cause of the isometry in the forceps of M. brunneri, van Lieshout and Elgar (2009) noted the species was present at a relatively low density, which could result in a low encounter rate for males in this species in the field. A reduction in the frequency of combat would likely result in a decrease in the benefit that larger males derive from having longer forceps. As Bonduriansky and Day (2003) pointed out, such variation in the strength and exact mode of sexual selection may explain the interspecific variation in allometric slope revealed by Simmons and Tomkins (1996).
The observed strong positive correlations between forceps length and body size indicate that the former is a condition dependent trait. An organism's condition can be defined as the pool from which its resources are allocated (Rowe & Houle 1996). This refers not only to the nutritional state of an individual, but may also include other factors such as variation in metabolic efficiency, immunocompetence or the ability to perform behavioral tasks (Bonduriansky & Day 2003). However, body size is frequently used as a convenient covariate of condition (Blanckenhorn 2000). Using simple mathematical models, Gross (1996) showed that a dimorphism can evolve in a condition- (or state-) dependent trait when the relative fitness values of two morphs (tactics), as a function of the condition, alternate at a certain switch point (see also Tomkins & Hazel 2007 for further refinement of the models). For more than 120 years (Bateson & Brindley 1892), F. auricularia has been well known for having two distinct morphs in its male forceps, which are usually termed macrolabic (=long forceps) and brachylabic (=short forceps) morphs (see Figure 1A,B for examples of a congener Forficula mikado Burr). Similar or other types of polymorphisms in male forceps have been reported in many other earwig species (e.g. Steinmann 1986, 1989a, 1990, 1993; Simpson & Mayer 1990; Tomkins & Simmons 1996).
Eberhard and Gutiérrez (1991) developed a statistical method to discriminate two alternative morphs within a character, such as the horns of horned beetles and the forceps of earwigs. Their method involves three steps: (i) testing the linearity of the allometry of the focal trait to body size; (ii) searching for a switching point where the slope of the regression changes; and (iii) testing the change in the slope and noting the presence of a gap in the trait value at the switch point identified during the second step. For the second and third steps, linear measurements are used to estimate regression relationships instead of log-transformed values. Eberhard and Gutiérrez (1991) applied their method to forceps length in four earwig species (three forficulids, D. taeniatum, F. auricularia, Metresura ruficeps (Burmeister) and one spongiphorid, P. dorsalis). None of the four species was found to be dimorphic in their forceps, in contrast to beetles in which 10 of 13 examined species showed statistically discernible dimorphisms in horn length. Among the four earwig species, Eberhard and Gutiérrez (1991) recognized two distinct morphs in F. auricularia, as indicated by an apparently bimodal distribution in forceps length. However, because of extensive overlap in body size between the two morphs, their method could not detect significant nonlinearity in the allometric relationship. Accordingly, in subsequent studies of the two male morphs in this species (reviewed in the following two subsections), the morphs have usually been separated at the valley in the bimodal distribution of forceps length with no statistical treatment. To overcome this problem, Kotiaho and Tomkins (2001) proposed an alternative method to statistically detect a switching point in forceps length itself (Y), instead of a switch in body size (X). Of course, this method can not determine a switching point in body size.
Experimental studies on F. auricularia
As noted above, large earwig males generally have relatively longer forceps, which are sometimes morphologically distinct from those of small males. Unexpectedly, almost all of the experimental studies conducted to determine the cause of this trend have used the single “species” F. auricularia, usually without any indication of which sibling species was used (Table 1). The first comprehensive study of the function of male forceps in F. auricularia was conducted by Radesäter and Halldórsdóttir (1993b). Using laboratory-reared virgin (or naïve) adults from a Swedish population, they conducted various experiments. In their first experiment, they placed two brachylabic and two macrolabic males together with a single virgin female (♂m–♂m–♂b–♂b–♀) in a Petri dish and observed their interactions. Although no significant difference in mating success (proportion of males mated) was noted between the two male morphs, macrolabic males copulated longer than brachylabic males (P = 0.004; however, repeated copulations by the same male were counted as independent events in their analysis). They also observed male–male interactions in detail under a ♂–♂ experimental setting (both morphs mixed). Even without any resources (food, shelter or females) in the arena, fights took place between the two males in 47 trials. The winners had both longer bodies and forceps than the losers, and a significant positive correlation was observed between these two traits. To determine the effect of forceps length while controlling the effect of body size, Radesäter and Halldórsdóttir (1993b) used a multiple regression with “win” or “lose” (coded as “1” or “0,” respectively) as the dependent variable and found that only forceps length gave a partial-F value that contributed significantly to the regression model (P = 0.01). However, they included both the winner and the loser of each replicate as independent data, and used a standard linear regression model instead of a logit-link function (logistic regression). Thus, caution is needed when interpreting this result. The third experiment was designed to examine female choice of male forceps under a ♂–♀ setting (♂b–♀ or ♂m–♀). Body length, forceps length or the male morph did not affect latency time to copulation. In the final experiment, Radesäter and Halldórsdóttir (1993b) observed staged encounters between two males that were matched for body size but had different forceps types (♂b–♂m–♀) and demonstrated an advantage for macrolabic males in mating success: 60 min after the initiation of observation, more macrolabic males were observed copulating with females than brachylabic males (29 macrolabic : 15 brachylabic; although they used a χ2-test and obtained P = 0.04, a binomial test is more appropriate (P = 0.049), while no such bias was found at 10 min after the initiation of observation. They noted male–male competition using forceps for taking over mating, and concluded that intrasexual selection was working on the male forceps of F. auricularia, whereas they found no evidence of active female choice.
Table 1. Studies of sexual selection working on forceps length of Forficula auricularia
Tomkins and Simmons (1998) first studied the effect of sexual selection on male F. auricularia using an artificial manipulation of forceps length (and forceps asymmetry; discussed below). In this study, they thoroughly adopted a ♂–♀ staged mating design, which was completely free of male–male interactions. In the first experiment, they examined the effects of male forceps length in brachylabic males (without manipulation) on the duration of courtship behavior in four populations in the United Kingdom (UK). With the exception of one population, males with longer forceps, and with forceps that were longer relative to body size (pronotum width), were more readily accepted by females. In the second experiment, they removed 10% (“Long” treatment) or 25% (“Short” treatment) of the forceps from fourth (=last) instar nymphs. The control group was treated the same as the treatment groups but the forceps were left unmanipulated. After emergence, this manipulation successfully produced the desired variation in male forceps length, without producing any signs of the manipulation in the adult forceps or causing any significant differences in body size or forceps asymmetry among the three groups. Each female was tested with a male from each of the three groups (in a randomized order) on three consecutive days. They repeated this experiment for two different UK populations. The mean duration of courtship up to the solicitation of mating from the female did not significantly differ among the three groups in either population. In only one population (Bownsman), in which the control males did not actively court females, females more frequently copulated with “Long” males than “Short” males (P = 0.03).
Styrsky and Van Rhein (1999) also studied sexual selection on male forceps length in F. auricularia using an artificial ablation technique, but with a totally different experimental design (♂–♂). They used field-collected adults from an Illinois (USA) population. In their first experiment, they introduced a male with longer forceps (upper 50% of the distribution) and a male with shorter forceps (lower 50%) in a stage without any resources and observed the outcome of the resulting combat. Judging from the distribution of forceps, both macrolabic and brachylabic morphs were used. As a result, the winners had significantly longer and wider forceps, while no differences in body size (pronotum length) were observed between the winners and losers. This result provides further support for the conclusion of Radesäter and Halldórsdóttir (1993b) that forceps length is important in male–male contests. Their second experiment involved manipulating male forceps (50% were cut at the adult stage vs an unmanipulated control). The result showed that the original (pre-manipulation) forceps length was a significant determinant of combat success, while post-manipulation forceps length was not. This result strongly indicates the presence of confounding factors, such as aggression, correlated with male forceps length. As discussed below (in the subsection “Directional asymmetry in forceps”), the winner (or loser) effect caused by previous male–male interactions also needs to be taken into account.
Using ♂–♀ and ♂–♂–♀ experimental settings and adult insects from a Virginia (USA) population, Walker and Fell (2001) recorded and analyzed the courtship behaviors of F. auricularia in great detail. They recognized 16 male and 10 female behaviors during courtship interactions, of which six and three, respectively, involved usage of the forceps. Behavioral sequences were not highly stereotyped: among almost 100 different behavioral transitions for males alone, only three occurred significantly more frequently than expected by chance. Similar to Radesäter and Halldórsdóttir (1993b), Walker and Fell (2001) observed interruptions of matings and takeover by non-copulating males during their ♂–♂–♀ observations. Nevertheless, Walker and Fell (2001) failed to detect any advantage of male body size or forceps length on mating success. They also conducted a manipulation experiment using males whose forceps had been completely ablated at their base. Although they detected no sign of behavioral abnormalities following the ablation treatment, none of the manipulated males succeeded in mating with a female under competitive conditions with unmanipulated males. Compared to former studies (Tomkins & Simmons 1998; Styrsky & Van Rhein 1999) in which forceps-ablated males (up to 50% of the length) succeeded in mating, at least on some occasions, this result strongly indicates that male forceps, at least a certain length of forceps, are essential for establishing a mating position (see also Kuhl 1928).
Although they lacked any artificial manipulations of forceps, a series of experiments by Forslund (2000, 2003) provided some of the most reliable information about the significance of male forceps during the precopulatory phases. In these studies, he both experimentally controlled the effect of body size (measured as fresh body weight) by matching similar-sized males in each contest and statistically controlled the residual variation in body size using a multiple logistic regression model. Using samples from a Swedish population, Forslund (2000) studied the effects of body size and forceps length on male mating success (total time of copulation) using a ♂–♂–♀ design. Three experiments, the “forceps experiment,” “weight experiment” and “size experiment,” were carried out. By matching two males with similar body weights or forceps lengths, the first two experiments examined the effects of forceps length and body weight, respectively, on mating success. In the “size experiment,” both forceps length and body weight were allowed to vary. The multiple logistic regression analysis detected a significant positive effect of male relative body weight on copulation success in the “weight” and “size” experiments, while the effect of relative forceps length was not significant in all three experiments. Using essentially the same experimental design, Forslund (2003) examined in greater detail morph-specific factors that determined male mating success. In this study, a noncompetitive session for each male (♂1–♀ and ♂2–♀) was conducted before matching two males in a ♂1–♂2–♀ arena. As in the previous study (Forslund 2000), Forslund (2003) detected a significant effect of body weight on mating success in brachylabic males. However, in macrolabic males, the possession of longer forceps was more important than being heavier. Based on observations of behaviors, including interruptions during mating and taking over, he concluded that the observed bias in mating success was mainly due to male–male contests rather than active female choice. According to this view, the benefits of body weight (in brachylabic males) and forceps length (in macrolabic males) were not observed in the preceding ♂–♀ session. However, this finding contradicts the findings of Tomkins and Simmons (1998) that having longer forceps is important for soliciting mating in brachylabic males. Both of these studies used, at least partly, field-collected adults, which may have been non-virgin and non-naïve. Strong effects of past experiences, which were suggested by the results of Styrsky and Van Rhein (1999) mentioned above, might explain this inconsistency in results. Regardless of the many discrepancies among studies, many studies have shown that male forceps in F. auricularia are used in male–male contests as weaponry. In particular, in conditions with one female and two or more males, interruptions of mating and take over by non-copulating males are commonly observed (Forslund 2000, 2003; Walker & Fell 2001). As shown by Walker and Fell (2001), forceps are necessary for establishing mating and many male courtship behaviors include the use of forceps. However, evidence of active female choice for longer forceps is not convincing.
Although more detailed and refined studies are needed to fully explain disagreements among the results of the studies reviewed above, this general conclusion does not contradict the view that male forceps in F. auricularia are a condition-dependent trait. Tomkins (1999b) studied the effects of rearing density and diet quality on the growth of forceps in two UK populations of this species. Food condition, source population, and their interaction term all significantly affected the absolute length of male forceps and the composition ratio of the two morphs, while rearing density had little or no effect on these measures (see also Kuhl 1928). Almost identical results were obtained for forceps length relative to body length. In addition to this interpopulation genetic difference, Tomkins and Simmons (1999) detected a significant level of additive genetic variance for male forceps (but not very large, and only significant after pooling dam effects with the residual variation in a post hoc analysis) using sib analysis with a half-sib mating design. Thus, as in many other cases of sexually selected traits (reviewed by Pomiankowski & Møller 1995; Rowe & Houle 1996), the evolvability of male forceps length has not been depleted despite the effects of directional sexual selection working on it. Condition dependence of forceps length may explain this apparent paradox, since the condition of an individual is likely to be affected by many genetic loci as well as environmental factors (Rowe & Houle 1996).
Tomkins and Brown (2004) compared the proportions of macrolabic morphs and body sizes that corresponded to switching points between the two male morphs (using the method of Eberhard & Gutiérrez 1991 introduced above) among 35 island and 11 mainland populations in the UK. Tomkins and Brown (2004) found that population density was positively correlated with the proportion of macrolabic morphs among island populations, while the effects of distance from the mainland and food availability (estimated as the biomass of ground-nesting birds per square meter since earwigs feed on both guano and the carcasses of dead birds there) were not significant. Their analysis revealed that the observed increase in the proportion of macrolabic males in high-density island populations was mainly caused by decreasing evolutionary changes in the switch point rather than by (evolutionary or environmentally caused) increases in body size. They interpreted these results as follows. Under high-density conditions, unpaired males are expected to frequently encounter males that are guarding a female, and high-conditioned males with long forceps benefit from being able to break up mating pairs. Notably, in the rearing experiment by Tomkins (1999b), rearing density under laboratory conditions had little effect on the proportion of macrolabic morphs. Thus, the observed interpopulation variation in the switching point likely represents rapid evolutionary diversification.
Understanding the mechanism underlying the development of earwig forceps also provides important insights into the evolution of weaponry traits. As a member of hemimetabolous insects, F. auricularia has also been used as a model organism for this aspect of biology, together with a holometabolous representative, the dung beetle Onthophagus taurus Schreber (Coleoptera: Scarabaeidae). One fundamental question is whether reprogramming of developmental processes is responsible for the two male morphs. In other words, can two alternative male morphs be explained by exponential growth following a single allometric function with an extraordinarily steep slope? A statistical analysis by Tomkins et al. (2005b) indicated that reprogramming of developmental processes at a certain threshold (also termed “developmental decoupling”) created the two forceps morphs in F. auricularia. Moreover, this result is in contrast to O. taurus, for which Tomkins et al. (2005b) could fit a single allometric function (cf. Moczek 2006; Tomkins 2006). Thus, the growth of male forceps is likely to follow different developmental programs, either depending on the body size beyond a threshold (macrolabic morph) or not (brachylabic morph). However, Tomkins and Moczek (2009) showed that variation in the allometric slope of the brachylabic morph is correlated with the population's threshold (=switching point body size), which indicates that the evolution of the two morphs is not totally independent. The evolution of exaggerated weaponry is also unlikely to be independent of the evolution of other body parts. The investment of limited resources into weaponry may be accomplished at the cost of reductions in the relative sizes of other body parts (trade-off). On the other hand, some body parts may also be developed to support exaggerated weaponry (sexually selected trait compensation (SSTC)). Tomkins et al. (2005a) examined partial-correlation structures among the sizes of various body parts in F. auricularia. They found no evidence of morphological trade-offs, but that hind-femur length covaries positively with morph type, suggesting SSTC.
Fluctuating asymmetry in forceps
Since the seminal book by Ludwig (written in 1932; see Ludwig 1970), the evolution of asymmetry in animal and plant body parts has remained a fundamental topic of biology. Asymmetries or deviations from bilateral symmetry can be classified into two categories (Van Valen 1962; Palmer 2005). The first category includes two different types of conspicuous asymmetries: directional asymmetry (DA) and antisymmetry (AS). In populations exhibiting DA, most individuals (>95%) are behaviorally or structurally asymmetrical (i.e. favoring one side), being either right-handed (dextral) or left-handed (sinistral). On the other hand, right-handed and left-handed individuals are almost equally common in populations showing AS (antisymmetry in a strict sense; see Palmer (2005) for a more detailed classification and terminology). In contrast to these conspicuous asymmetries, fluctuating asymmetry (FA) refers to small random deviations from perfect symmetry in bilateral traits. FA has been considered an indicator of developmental stability, or an organism's ability to cope with genetic and environmental stresses during development (see Swaddle 2003; Leamy & Klingenberg 2005; Polak 2008). A negative correlation might be expected between a measure of genetic quality (or condition) and the extent of FA, based on which females can choose male mates to obtain genetic benefits (the FA-sexual selection hypothesis). However, no general support currently exists for the view that FA is heritable and that females gain genetic benefits by choosing symmetrical males (Tomkins & Simmons 2003; Polak 2008).
With the exception of interspecific comparisons based on dried specimens from museums (Tomkins & Simmons 1995, 1996; see below), ecological or genetic studies on FA have only been conducted for male forceps in F. auricularia. Radesäter and Halldórsdóttir (1993a) examined the effect of FA in male forceps on mating success using laboratory-reared virgin or naïve adults. In this study, no indication of which morph (macrolabic or brachylabic) used for the mating experiments was determined, although the forceps size range (fig. 1 of Radesäter & Halldórsdóttir 1993a) and another study based on the same population (Radesäter & Halldórsdóttir 1993b) suggest that both morphs were represented in the sample. In their population, FAs in forceps and elytra were highly negatively correlated with body condition, which was measured as residual body weight regressed on body length in males (P = 0.0003 and 0.0005, respectively). However, no such correlations were observed in females. FA in forceps was also negatively correlated with forceps length in males, but the correlation was positive in females. In their first mating experiment, they paired males with symmetrical (S) or asymmetrical (A) forceps individually with a virgin female (♂S–♀ or ♂A–♀; each with 10 replicates). In 13 pairs in which mating occurred (eight and five cases for ♂S and ♂A, respectively), latency until mating was significantly shorter in ♂S than ♂A. In the second experiment, they placed a ♂S and a ♂A with a virgin female (♂S–♂A–♀). More ♂S individuals were observed copulating with females than ♂A individuals (7 vs 1 case); Radesäter and Halldórsdóttir (1993a) detected a significant effect (P = 0.01) using a Fisher's exact probability test. However, they incorrectly treated cases of “symmetric male wins” and “asymmetric male loses” as independent data. This treatment spuriously inflated the degrees of freedom, and reanalysis by Ghent (1998) revealed that the result was not significant (binomial test, P = 0.07). Furthermore, although they used similar-sized males for this experiment, the residual effects of body size and forceps length were not statistically controlled.
For the same species, Tomkins and Simmons (1998) conducted two experiments to examine the effects of FA in forceps on female mate choice, using both natural and artificially induced variations in forceps FA. In their first experiment, they divided the adult males into two groups, “S (symmetrical)” and “A (asymmetrical),” depending on the original asymmetry (within ±1% of the forceps length, or not). Then, the tip of one branch of the forceps was cut to make asymmetrical males symmetrical (S–a), or symmetrical males asymmetrical (A–s). Including unmanipulated controls (S and A), no difference was observed in latency to solicit mating from a female among the four groups in a ♂–♀ setting. However, unmanipulated males were more successful at establishing copulation, suggesting possible detrimental effects of forceps manipulation in the adult stage. Accordingly, they conducted a similar experiment by bilaterally ablating the forceps during the fourth nymphal instar. No significant effects of the original (pre-manipulation) or resultant (post-manipulation) forceps FAs were detected on either latency to mating or mating success. These results contradict the findings of Radesäter and Halldórsdóttir (1993a) that males with a low forceps FA mate sooner than males with a high FA. Tomkins and Simmons (1998) also failed to detect a negative correlation between forceps FA and body condition in their two UK populations, which was found in the Swedish population studied by Radesäter and Halldórsdóttir (1993a). Therefore, variation in the extent of condition dependence in male forceps FA may explain the inconsistency between these studies.
Tomkins (1999a) examined the ontogeny of forceps FA in F. auricularia by repeatedly measuring the length of right and left forceps, once per nymphal instar (second to fourth) and at the adult stage. As a result, from the third to the fourth instar, and from the fourth instar to adults, a trend of compensatory growth was observed. That is, when the right branch of the forceps was longer, the left branch tended to grow more than the right one after molting. Consequently, although absolute FA did not decrease during nymphal development, the magnitude of FA relative to forceps length decreased in the adult stage compared to immature stages. Although the compensatory process is imperfect, leaving FA in the adult stage, Tomkins (1999a) considered that the observed FA in the adult stage does not directly reflect male genetic quality and thus may not be used for female choice. In accordance with this view, Tomkins and Simmons (1999) did not detect significant genetic variation in forceps FA in male F. auricularia, but they confirmed additive genetic variation in forceps length and pronotum width based on the same samples.
The same conclusion was drawn from interspecific studies. Using specimens that had been preserved in museums, Tomkins and Simmons (1995) examined FA in male forceps in 26 species belonging to Apachyidae, Labiduridae, Spongiphoridae, Chelisochidae and Forficulidae. If males in high body condition develop long forceps, a negative relationship between forceps length and the extent of FA can be expected. However, significant negative relationships were only found in two of 26 species examined (F. auricularia, which is discussed in detail above, and Spongiphora croceipennis Audinet-Serville). Across species, the extent of FA relative to forceps length was positively correlated with residual forceps size (forceps size controlling for body size) as a measure of forceps exaggeration. This positive relationship also held after controlling for phylogenetic constraints using the generic relationships proposed by Sakai (1982). Also, in Apachyus faea Bormans (Apachyidae), the extent of FA was highly positively correlated with body size. In a congener A. chartaceus, Shimizu and Machida (2011a) reported that male forceps are used to grasp the female postabdomen during copulation. The observed large forceps FA in large-sized A. faea males, which suggests possible antisymmetry, may be related to a similar mate-holding function.
Applying the same logic for correlations between forceps length and FA, brachylabic males with smaller body size can be expected to show a higher degree of forceps FA. Tomkins and Simmons (1996) examined the following six earwig species that show dimorphisms in male forceps to determine the relationships between forceps size and asymmetry: Spongovostox assiniensis (Bormans) (Spongiphoridae), Proreus ludekingi (Dohrn) (Chelisochidae), F. auricularia (Forficulidae), Elaunon bipartitus (Kirby) (Forficulidae), Timomenus aeris (Shiraki) (Forficulidae) and Oreasiobia stoliczkae (Burr) (Forficulidae). However, except for the macrolabic morph in one of the two examined populations of F. auricularia, the expected relationship was not generally observed. In conclusion, for earwig species that have essentially symmetrical forceps, no consistent support exists for the view that FA in male forceps reflects body condition and that females choose males based on this FA.
Directional asymmetry in forceps
Several earwigs are characterized by having conspicuously asymmetrical forceps that show DA. Typically, only male forceps show conspicuous DA. Examples come from almost all of the forficuline families except for Eudermaptera, for which only a few cases have been described: Karschiellidae (all the species described), Pygidicranidae (e.g. Tagalina spp.), Diplatyidae (e.g. Haplodiplatys major (Brindle)), Anisolabididae (many genera) and Labiduridae (e.g. Allostethus spp.) (e.g. Sakai 1985, 1987a, 1990, 1991, 1992, 1993, 1994, 1995a,b,c,d; Steinmann 1986, 1989a, 1990, 1993). Functional studies have only been conducted for male-specific forceps DA in several anisolabidid earwigs, in which the right branch is typically more strongly curved than the left one (Fig. 1D).
Briceño and Eberhard (1995) observed male–male combat, courtship and mating behaviors in four anisolabidid species, Carcinophora rosenbergi (Burr), Carcinophora gagatina (Klug) (Carcinophora robusta in Briceño & Eberhard 1995; cf. Srivastava 1999), Anisolabis americana (Palisot de Beauvois) (Carcinophora americana in Briceño & Eberhard 1995; cf. Srivastava 1999) and Anisolabis maritima (Bonelli). In these four species, males were often observed pinching opponent males, and in A. maritima and C. gagatina, pinching resulted in punctures to the male abdomen (one case for each species). Such severe damage was not observed in the other 11 earwig species they studied. They proposed that the strongly curved right branch represents an adaptation to produce powerful pinches when the forceps are used as a weapon. On the other hand, they found no apparent function of the forceps asymmetry in courtship and mating.
Van Lieshout and Elgar (2009) observed precopulatory female choice and combat behavior for a resource (a shelter without a female) between two males of M. brunneri under various experimental settings (♂–♂, ♂–♀ and ♂–♂–♀). Although pronotum width, body weight and forceps length were significant positive determinants of combat success, the extent of the asymmetry did not affect precopulatory fitness components (male–male combat and female choice). However, males with less-curved forceps copulated more frequently and for longer total durations under ♂–♀ experimental settings that were preceded by ♂–♂ and ♂–♂–♀ sessions, although the reason for this is currently unclear.
Munoz and Zink (2012) studied male–male aggression in A. maritima using field-collected adult males. They released two male contestants that had been starved for several days into an arena with food. Thus, although combats were not observed in the context of sexual selection, the observed male–male contest behaviors agreed well with those described by Briceño and Eberhard (1995), which were observed under a setting without any resources. To control possible confounding effects of body size, Munoz and Zink (2012) divided the male sample into large- and small-sized groups based on pronotum width. Three types of trials, large vs large, large vs small and small vs small males, were conducted to examine the effect of body size and forceps asymmetry on combat outcome. Forceps asymmetry was a significant positive predictor of the probability of winning only in the small vs small trials. In the other two types of trials, larger males had the advantage in combat. They noted a lateral bias in contest behaviors: males of A. maritima twist their bodies to the left significantly more than to the right to deliver strikes with their forceps. This direction of abdomen twisting results in contact between the strongly curved right branch and the dorsal side of the opponent. Munoz and Zink (2012) repeated their size-matched contests with manipulated males, from which approximately 15% of the right or left side forceps had been cut off. Although the result was not significant, they found a trend whereby males with large pre-manipulated asymmetries won contests, while post-manipulated asymmetries did not affect the outcome in small vs small contests. As with the study of F. auricularia discussed above (Styrsky & Van Rhein 1999), this result suggests the presence of unknown factors that are correlated with forceps morphology. A possible candidate is the effect of past experience. For M. brunneri, van Lieshout et al. (2009) reported a strong loser effect, with almost all losers of previous combats losing against new rivals in contests that were held immediately after the first one. However, no winner effect was observed in this species (van Lieshout et al. 2009). Similar effects have been reported in V. apicedentatus. For this species, Moore and Wilson (1993) noted that 72 h of isolation was sufficient to minimize the effects of previous interactions. Considering the diversity in mating ecology (e.g. Briceño & Eberhard 1995), future studies should examine the extent and duration of winner and loser effects in each species.
Postcopulatory sexual selection and evolution of genitalia
Overview and genital terminology
Animal genitalia used for copulation and sperm transfer, especially male genitalia, are often more complex than is necessary for sperm transfer alone and seem to evolve more rapidly than other structures (reviewed by Eberhard 1985; Hosken & Stockley 2004). Such extravagant features include elongation, gigantism, spines and multiplication of the genital apparatus. Accordingly, the identification of closely related species often requires examinations of genital morphology by specialist taxonomists. Although female insect genital structures are usually assumed to show less prominent interspecific diversity than males, careful examinations of female genitalia in some insects have provided evidence that these structures have coevolved with male genitalia (e.g. Rönn et al. 2007; McPeek et al. 2009; Kamimura 2012). Earwigs are not exceptions. Male genitalia are commonly used for species descriptions and identification, and female genital characteristics are being increasingly recognized to contain important information for phylogenetic studies (Klass 2003). Four major hypotheses, the lock-and-key, SC, CFC and sexual conflict hypotheses, have been proposed to explain the rapid and sometimes correlated evolution between male and female genital morphologies. Debate is ongoing about the relative importance of these hypotheses for rapid and correlated evolution of genital morphologies (Eberhard 2010). The lock-and-key hypothesis assumes stabilizing selection on male genital morphology (the key) in the form of pre-insemination discrimination of conspecific males with genitalia that provide an exact structural or tactile fit with the female genitalia (the lock) (Dufour 1844). This hypothesis predicts that corresponding genital morphologies function to avoid interspecific matings, which result in the production of unviable or sterile hybrid offspring. At present, only a few examples in which tactile or sensory processes may play important roles support this view (Coyne & Orr 2004; Masly 2012). The SC and CFC hypotheses are inextricably related to each other. As introduced above, these postcopulatory sexual selection processes are only relevant to genital evolution when females (at least potentially) mate with multiple males. For example, males can remove the sperm of rival males using genitalia with a morphology that fits the shape of the female sperm storage organ. When mated with multiple males, a female may increase her fitness by biasing paternity toward sires of higher genetic quality (CFC). This control of paternity by manipulating sperm can also be achieved by female sensory mechanisms; thus, in considering CFC, we need to extend the meaning of correlated morphologies to include the nervous system (Eberhard 1996; Cordero & Eberhard 2003). Growing evidence is showing the importance of sexual conflict as a compelling force of antagonistic coevolution between male and female genitalia (Hosken & Stockley 2004; Arnqvist & Rowe 2005; Kamimura 2012). Usually, males try to mate at a higher frequency than females, and male genitalia are often used as devices for coercive matings with unwilling females (e.g. Arnqvist & Rowe 2002; Vahed 2002).
Earwigs show a high level of diversity in male genital morphology, including the size, number and laterality of intromittent organs (Fig. 2), making them valuable models of genital evolution. Although the structure of male earwig genitalia is relatively simple, confusion exists regarding genital terminology in the Dermaptera, especially for the definition of “penis” (Table 2). Considering the recent increase in studies of the functional morphology, here I propose the following terminology system, which may help comprehension of their functional integrity (Table 2). Although male forceps and the pygidium are also sometimes used to establish genital coupling (Briceño & Eberhard 1995; Briceño 1997; Shimizu & Machida 2011a), these structures are usually not included in the category of male genitalia. In male earwigs, sperm are produced in a pair of testes and then transferred to a spherical seminal vesicle for storage. From the seminal vesicle, one or two ejaculatory ducts lead to the male genitalia (sensu stricto), which is located in a space (termed genital chamber) that is dorsal to the subgenital plate (=last abdominal sternite). The virga is a variously shaped, sclerotized tube that is directly connected to the ejaculatory duct. As introduced below, a virga is usually inserted into the female storage organ (spermatheca) to transfer sperm. At the junction between the ejaculatory duct and the virga, a dilation termed the basal vesicle is found in many groups (although several phylogenetic studies (e.g. Haas 1995; Haas & Kukalová-Peck 2001) treated Anisolabididae as a taxa lacking basal vesicles, this structure has also been found in several anisolabidids (e.g. Hudson 1973; Sakai 1987a; Kamimura 2000)). The most caudal part (or entire length) of each virga is surrounded by a membranous penis lobe (here I define a penis as a virga and its penis lobe). Species with paired ejaculatory ducts thus have paired penises, each of which contains a virga. Lateral to the single or paired penises, a pair of parameres (=external paramere) of various shapes are located at the caudal end of the genitalia.
Table 2. Genital terminology proposed in this study with alternative names in major references
Males in six families have two laterally paired ejaculatory ducts and virgae (and hence two penises), with few exceptions (Fig. 2A–C). Among the basal Dermaptera (Karschiellidae + Pygidicranidae + Diplatyidae), both the right and left penises symmetrically point to the head when not in copulation in Pygidicranidae and Diplatyidae (status 1 of Fig. 2A), while the left penis is largely reduced and vestigial in Karschiellidae (e.g. Sakai 1985; Steinmann 1986). Hereafter, the term “resting state” is used to describe penises in these positions when not in copulation. In the other three taxa (Apachyidae, Anisolabisidae and Labiduridae), the resting states of their penises are laterally asymmetric: the left and right penises point in opposite directions (status 2 of Fig. 2A; Popham 1965; Kamimura & Matsuo 2001; Kamimura 2006). Because a penis should point posteriorly during mating, organs that point posteriorly when resting can be considered ready for mating. Earwigs in taxa with asymmetric genitalia show antisymmetry; i.e. R- and L-ready males are almost equally frequent (reviewed by Kamimura 2006). One exception, however, is the labidurid L. riparia (Kamimura 2006; see subsection “Evolutionary reduction in penis number” below for details). Members of Eudermaptera have only one penis, without exception (status 3 of Fig. 2A; Steinmann 1990, 1993). Although the proposed phylogenetic relationships among the forficuline families are still quite inconsistent, all previous studies have assumed that paired virgae represent the ancestral state (Popham 1985; Sakai 1987b; Haas 1995; Haas & Kukalová-Peck 2001; Colgan et al. 2003; Haas & Klass 2003; Kamimura 2004b; Jarvis et al. 2005). In addition, to my knowledge, no examples of reversals, from status 3 to status 2, or from status 2 to status 1 of Figure 2A, have been reported. Thus, the number and direction of penises are considered stable characters and are used to define higher taxa in Dermaptera (Fig. 2A; Burr 1915a,b, 1916; Popham 1965; Steinmann 1986, 1989a, 1990, 1993).
Evolutionary origin of paired penises
Two fundamental questions arise concerning evolutionary changes in the number of earwig penises. Why did the ancestor of earwigs acquire paired penises? Why did one of the paired penises degenerate in several earwig groups, and which penis was lost? Both of these questions have being explored in relation to postcopulatory sexual selection. In anisolabidids such as A. maritima and Euborellia plebeja (Dohrn), males have two functionally competent elongated virgae, although they use only one of them during a single genital coupling (Kamimura 2000; Kamimura & Matsuo 2001). Since the virgae are fragile, they sometimes break off during mating; therefore, the remaining counterpart in paired virgae functions as a spare (Kamimura & Matsuo 2001). This phenomenon can explain the modern function of paired penises in this specialized group. However, because not all earwigs, especially those in basal groups, have thin and elongated virgae, this spare function is not likely sufficient to explain the evolution of paired virgae in the ancestor of earwigs. Examinations of genital couplings and postcopulatory sexual selection in basal Dermaptera are especially important for exploring this question. Among them, members of the family Diplatyidae are of special interest because they are considered to be the most basal offshoot of Forficulina, together with Karschiellidae (Fig. 2A). Diplatyids retain many presumably primitive characteristics, such as blattoid-type necks and segmented nymphal cerci (Haas 1995; Haas & Kukalová-Peck 2001; Klass 2003). One such primitive characteristic is the possession of multiple sperm storage organs (spermathecae) in females. Females of all diplatyid species examined to date possess multiple (from 2 to 6) spermathecae with independent orifices (Diplatys macrocephalus (Palisot de Beauvois) (Popham 1965; Klass 2003); Diplatys flavicollis Shiraki (Kamimura 2004a); Haplodiplatys orientalis Steinmann (Klass 2003)). Most female earwigs have a single spermatheca and multiple spermathecae have been reported only in Diplatyidae, Karschiellidae and several Pygidicranidae (Popham 1965; Mariani 1994; Klass 2003; Kamimura 2004a). The paired virgae found in male diplatyids (except in the genus Cylindrogaster) and in two subfamilies of Pygidicranidae (Pyragrinae and Esphalmeninae) are bifurcated, and each has two gonopores that yield four exits of sperm per male (e.g. Brindle 1984; Sakai 1985; Steinmann 1986), while a single virga with one gonopore per penis lobe is found in the remaining earwigs (e.g. Popham 1965; Steinmann 1986, 1989a, 1990, 1993). Based on observations of the four independent spermathecal openings in D. macrocephalus, Popham (1965) predicted a one-to-one correspondence between male gonopores and female spermathecae during copulation. Therefore, one plausible hypothesis is that the common ancestor of earwigs used two virgae simultaneously to inseminate multiple spermathecae, and that the Diplatyidae have retained this primitive method of genital interaction, whereas other families have changed to using only one virga to inseminate a single spermatheca.
Kamimura (2004a) performed a detailed investigation of male and female genital structures, mating behaviors and insemination processes in D. flavicollis. As in other diplatyids, males of this species have two gonopores on each of two virgae, whereas females have four to six independent spermathecae, each with a spermathecal duct and an opening. Rapid fixation of mating pairs and the insemination success of males from which one virga had been removed clearly revealed that only two gonopores of one virga are used during a single genital coupling and are usually sufficient for inseminating multiple spermathecae. This finding rejects the one-to-one correspondence between male gonopores and female spermathecae predicted by Popham (1965). However, the virgae of diplatyid males show considerable variety in size and shape (Burr 1915a; Hincks 1955; Sakai 1985; Steinmann 1986). Therefore, studying mating in other diplatyids, such as Haplodiplatys, which Haas (1995) treated as the basal genus of the family Diplatyidae, is warranted to determine whether the observed use of a single penis is a common feature of this family. Several recent studies have supported a sister relationship between Dermaptera and Plecoptera (stoneflies) among the polyneopteran insect orders (Kjer 2004; Yoshizawa & Johnson 2005; Misof et al. 2007; Ishiwata et al. 2011; Yoshizawa 2011). Moreover, several stonefly species are also characterized by having paired orifices in ejaculatory ducts that apparently function as sperm exits (Brink 1956a, b). Thus, the origin of paired earwig penises may date back to the common ancestor of these two orders.
Why do females of Diplatys develop multiple spermathecae? Theory predicts that having multiple sperm stores can be a powerful device for CFC (Hellriegel & Ward 1998). Differential storage of sperm from multiple males has been reported in several animals having multiple sperm storage organs (e.g. Ward 2000, 2007; Snow & Andrade 2005; Herberstein et al. 2011). The following two points have implications for the possibility of CFC in D. flavicollis (Kamimura 2004a). First, partial insemination was observed in one female in which only three of the four spermathecae received sperm, and no abnormalities were detected in the openings and ducts of the spermathecae. Thus, the male may have poorly transferred sperm or was partially rejected as a sperm donor. Second, the number of spermathecae is largely independent of female body size (an index of female condition), in contrast to the total volume of spermathecae, which is primarily determined by female body size (Kamimura 2004a). A similar phenomenon was reported for the dung fly Scathophaga stercoraria Linnaeus (Ward 2000). As in the case of the dung fly, the observed variation in spermathecal number may have a genetic basis, and thus could reflect different female strategies for dividing sperm storage into multiple packages. Future studies should examine the environmental and genetic determinants of spermathecal number in diplatyid females as well as storage patterns of sperm derived from multiple males.
A unique feature of genital couplings in D. flavicollis is that the virga is not placed directly into the spermatheca (Kamimura 2004a). The male gonopores are much wider than the spermathecal openings and ducts; thus, they must be incapable of being directly inserted into the latter. Instead, each of two gonopores on a virga fits closely to spermathecal openings (2–3) that are arranged in two rows. For other species belonging to Anisolabididae, Labiduridae, Spongiphoridae and Forficulidae that have been examined to date, a closed system of sperm transfer (i.e. direct insertion of a virga into a spermatheca) has been reported (F. auricularia Popham 1965; L. riparia Arora & Bhatnagar 1961; Singh et al. 1982; P. dorsalis Briceño 1997; A. maritima Kamimura & Matsuo 2001; E. plebeja Kamimura 2000). Thus, an evolutionary change from an open to a closed system may have occurred in the forficline phylogeny.
One other feature of D. flavicollis is the presence of three differently shaped accessory sclerites, specifically U-, rod- and saber-shaped sclerites, in each penis (Kamimura 2004a). In an established genital coupling, two lateral pockets in the female genital chamber receive the U- and rod-shaped sclerites, while the saber-shaped sclerite contacts the female subgenital plate (Kamimura 2004a). Although similar accessory structures are frequently observed in other earwigs (e.g. Steinmann 1986, 1989a, 1990, 1993), to my knowledge, the functions of these structures have only been studied in one species, P. dorsalis, by Briceño (1997). In this species, a sclerotized horn-shaped structure surrounds a virga. Briceño (1997) observed that the tip of the horn is located near the spermathecal opening during copulation. The male penis in this species also possesses several areas with hardened spines (referred to as “toothed plates” in Briceño 1997). During copulation, the membranous penis is inflated and these spines contact areas around the female gonopore and the lateral wall of the vagina. Females of P. dorsalis also have sclerotized spines around the gonopore (the orifice of the common oviduct), and some of the teeth on a male toothed plate mesh with them during copulation. For this species, Briceño (1997) also observed that the parameres are used to grasp the female pygidium. In Labidura and Euborellia, parameres are deeply inserted into the female genital chamber during copulation, and contact between male parameres and the female pygidium is physically impossible at the established copulation stage (Y. Kamimura, pers. obs., 2012). Thus, the observed contact between male parameres and the female pygidium in P. dorsalis likely does not represent a common feature among earwigs.
Evolutionary reduction in penis number
The second question is why the ancestors of several earwig groups lost one of their paired penises. A reduction or degeneration of one of the paired penises occurred independently multiple times, at least in the common ancestors of Karschiellidae and in that of Eudermaptera (Spongiphoridae, Chelisochidae and Forficulidae). Among them, the latter case is the most important because the single penis status is shared by more than 1000 extant species of Eudermaptera, which are considered to be monophyletic, without any exceptions (e.g. Steinmann 1990, 1993; Sakai 1991, 1992, 1993, 1994, 1995a,b,c,d). Although two males that emerged from gamma-irradiated nymphs of Proreus simulans (Stål) (Chelisochidae) possessed paired virgae in their genitalia (Kamimura 2007), such a malformation has not been reported in natural populations of Eudermaptera. All of the species of Eudermaptera examined to date have a vestigial and blind ejaculatory duct at the left side of the seminal vesicle, while the right-side ejaculatory duct, which leads to the virga, is functionally competent (Ramamurthi 1958; Popham 1965; Kamimura 2007). This indicates that a common ancestor of Eudermaptera lost the left ejaculatory duct and hence the left penis. The family Labiduridae is especially important for exploring the question as to why the ancestor lost the left penis because previous studies have suggested a sister-group relationship between this family and Eudermaptera (Sakai 1987b; Haas 1995; Wirth et al. 1999; Haas & Kukalová-Peck 2001; Colgan et al. 2003; Haas & Klass 2003; Kamimura 2004b; Jarvis et al. 2005).
Kamimura (2006) examined laterality in penis morphology and its usage in five Japanese populations of the labidurid L. riparia (note that several authors treat L. riparia in Japan as a separate species, Labidura japonica (De Haan); Steinmann 1989a,b). The study revealed that the male L. riparia predominantly uses the right penis for insemination, although both penises are functional without any detectable morphological differences between them. Such a behavioral bias in virga use also manifested in the directionally biased asymmetry in the penile resting state toward R-ready. This laterality was established without any mating experience, indicating an underlying genetic control. In wild-caught males, only about 10% were left-handed; within this group, abnormalities were frequently observed in the right penis. These lines of evidence indicate that the left penis is merely a spare intromittent organ, which most L. riparia males will likely never use.
Why are male L. riparia predominantly right-handed? No credible answer is presently available for this question. As Huber et al. (2007) pointed out, no general explanation has been proposed to date for the evolution of asymmetric genitalia. Asymmetric copulatory positions and asymmetry in female genitalia are possible causes for the evolution of asymmetric male genitalia (Ludwig 1970; Huber et al. 2007; Huber 2010). The first scenario is not likely applicable to L. riparia because copulatory position (i.e. males twist their abdomen in a clockwise or counterclockwise direction) was not correlated with the penis use pattern (Kamimura 2006). The structure and position of the spermatheca in female L. riparia, which receives sperm directly from the male virga, may explain the extreme right-handedness in males. Although the spermatheca lies on the dorsal side of the common oviduct running along the median line of the female body, Hudson (1973) described a peculiar coiling structure in the basal part of the spermatheca having an opening that points to the lateral side of the body. Such asymmetrical structures in females could generate differential insemination success between right- and left-handed males. However, sexual selection working on handedness may be not very strong because Kamimura (2006) detected no significant deference in one-day-pairing insemination success between the two types of males. In future studies, female–male genital contacts should be studied in greater detail.
Based on a phylogenetic analysis of asymmetry variation among animals, Palmer (1996) pointed out an evolutionary trend by which AS tends to precede DA in asymmetrical traits expressed in postlarval phases. The case of penis-handedness in earwigs is not an exception. According to the most reliable current phylogeny (Fig. 2A), the following evolutionary transitions in genital laterality occurred in Forficulina: symmetry (basal Dermaptera) → AS (Anisolabididae and possibly Apachyidae) → DA (L. riparia and possibly shared by the other labidurids), and finally, the evolutionary loss of one penis in Eudermaptera (Kamimura 2006; see also Palmer 2006). In populations exhibiting AS, the handedness of each individual may be determined by either environmental or genetic factors. Peculiarly, with only a few possible exceptions (e.g. Hori 1993; Hata & Hori 2012), previous studies have generally failed to detect any evidence of inheritance in the direction of asymmetry in organisms showing AS (Palmer 2004, 2005, 2009). Thus, in marked contrast to the genetically determined direction of DA, handedness in AS traits appears to be controlled epigenetically. Considering the general trend that AS tends to precede DA (Palmer 1996), the evolutionary AS → DA transition would be accompanied by a genetic assimilation process, that is, the replacement of an external environmental handedness trigger with a genetic trigger (Palmer 1996, 2004, 2005, 2009). Even with the comparatively high statistical power of the half-sib breeding design, Kamimura and Iwase (2010) detected no significant additive genetic, dominance or maternal effects on genital laterality in male E. plebeja, which shows AS in penis use. Considering the moderate heritabilities that have been detected for other traits using the same sample set, this is not likely due to insufficient statistical power. This result supports the trend that AS variations observed in animals are generally not inheritable, in contrast to genetically determined DA variations. No earwigs with genetically determined AS have been reported at present, and the evolutionary transition from AS to DA in the earwig penis would be accompanied by a genetic assimilation process.
As discussed above, animal genitalia are often more complex than is necessary for sperm transfer alone, and they seem to evolve more rapidly than other structures (for example, Fig. 3). In marked contrast to this interspecific variation, intraspecific variation in animal genitalia is generally less variable than that of nongenital traits. Generally, male genital measurements are characterized by negative allometry (b < 1), indicating that large individuals have relatively small genitalia. Although several exceptions have been reported in vertebrates (e.g. Lüpold et al. 2004), low levels of phenotypic variation and negative allometry are the norm for the sclerotized genital parts of male arthropods (Eberhard et al. 1998; Eberhard 2009). Strong directional or stabilizing selection via SC, CFC or sexual conflict is presently the most plausible cause of genital hypometry (e.g. Hosken & Stockley 2004; Eberhard 2009, 2010; Eberhard et al. 2009; Simmons et al. 2009).
Genital allometry has been studied in only a few species of earwigs. Eberhard et al. (1998) revealed the general trend for hypometry in the genital parts of many species of insects and arachnids. Their analysis included one species of earwigs, P. dorsalis. In this species, the entire length of the male genitalia (trait no. 23 in Eberhard et al. 1998) showed hypometry (OLS regression slope, b = 0.393), while the correlation and regression (log–log) coefficients were not significant for four individual genital parts. Using male samples (n = 622) of E. plebeja in a quantitative genetic study, Kamimura and Iwase (2010) measured the length of elongated genitalia and pronotum width as an index of body size. The resulting allometric relationship was negative with a slope of 0.653 (major axis regression; 95% confidence interval of the slope 0.567–0.755). For the same species, a previous study with a much smaller sample size (n = 43) did not detect a significant correlation between the two traits, apparently because of the low allometric slope (Kamimura 2000). Similarly, in another anisolabidid, M. brunneri, male genital length showed only a weak, positive correlation with body size (measured as pronotum width; n = 15–92; van Lieshout 2011; van Lieshout & Elgar 2011a).
Although testes are usually treated as a component of the internal reproductive organs rather than the male genitalia, Tomkins and Simmons (2002) found that the allometric slope of log-testis mass on log-body mass (excluding testes mass) was almost isometric (RMA slope, b = 1.077) in F. auricularia. Analysis of covariance (ancova) found no significant difference in the allometric relationship between macro- and brachylabic males (Tomkins & Simmons 2002). In contrast, van Lieshout and Elgar (2011a) reported a positive allometry (RMA slope 1.54; 95% confidence interval 1.28–1.85) for the size of the seminal vesicle in M. brunneri. Showing that hypometry is not unique to genital traits, Eberhard et al. (2009) pointed out a common problem of comparing absolute slopes with the usual reference value of 1.00 for inferring underlying mechanisms of allometry. The studies reviewed here indicate that much lower allometric slopes are common for genital traits of earwigs compared with forceps allometry discussed above.
Irrespective of the exact selection mechanism of genital evolution, the strong directional or stabilizing selection forces are likely to deplete not only phenotypic variation in male genital traits (as manifested in the negative allometric slope), but also additive genetic variation in traits. The results of several empirical studies do not agree with this prediction: levels of genetic variation are similar to levels of variation in general morphological traits (Arnqvist & Thornhill 1998; Preziosi & Roff 1998; House & Simmons 2005; Andrade et al. 2009; Higgins et al. 2009). In E. plebeja, although the allometry was negative (b = 0.653), and thus phenotypic variation in male genitalia was restricted compared to that of body size, these two morphological traits showed similar levels of narrow-sense heritability and coefficients of additive genetic variation (Kamimura & Iwase 2010). As discussed in the next section, strong sexual selection is likely to be working on virgal length in Euborellia. Nevertheless, genetic variation in genital length was not depleted. Although the allometric slope is typically low for genital traits, they are still positively correlated with body size, which indicates their condition dependence. Thus, as with the weaponry traits discussed above, many loci that affect an individual's condition may function as perpetual sources of genetic variation (Rowe & Houle 1996).
Functional studies on elongated genitalia
Several groups of earwigs are characterized by having elongated male genitalia, which are sometimes longer than the male body. Such conspicuously elongated genitalia are found sporadically in Spongiphoridae (Ramamurthi 1958), but many examples can be found in Anisolabididae (e.g. Hudson 1973; Sakai 1987a). The functional significance of elongated genitalia, in relation to SC, has been studied in only two representative species, E. plebeja and M. brunneri.
Euborellia plebeja is a small-sized anisolabidid with vestigial wing pads on the thorax. Both males and females of the species are highly promiscuous (Baijal & Srivastava 1974; Kamimura 2003a,b, 2005). Under laboratory conditions, females mate several to dozens of times during 15 h (Kamimura 2005). Several (usually less than three) copulations are enough to saturate an elongated, fine-tubed spermatheca. After saturation, females continue to have frequent repeat copulations with males, which can not be explained as a mere sperm-supply function (Kamimura 2005). These habits suggest severe SC in this species. Flash-fixation experiments of mating pairs have revealed that males use one of their paired virgae, which are as long as their bodies (on average 15.8 mm; Fig. 3), to remove rival sperm from the spermatheca (Kamimura 2000, 2003a). The mechanism of sperm removal is as follows. First, a male inserts the virga deeply into the spermatheca without ejaculating. It then extracts the virga while ejaculating semen from its tip and simultaneously removing rival sperm using a fringe-like projection on the virgal tip (Kamimura 2000). However, the spermatheca in this species is twice the length of the female body (on average 33.6 mm; Kamimura 2000, 2005). As predicted from the considerable difference in their lengths, males can remove only a portion of the stored sperm (Kamimura 2000), and a paternity-analysis experiment revealed that the paternity gain is only about 20% (mean ± SD = 0.193 ± 0.182) for a single copulation with a female whose sperm storage organ has been saturated by a rival male (Kamimura 2005; here, the paternity gain from a single copulation with a sperm-saturated female is termed last male paternity, Plast). Another experiment, in which the mating order of two males was not controlled, also showed that the best fit between predicted and realized paternity success is obtained when assuming Plast = 0.22 (Kamimura 2005). Females undergo repeated oviposition (approximately 40 eggs) at intervals of about 27 days (Kamimura 2003b) and thus are likely to collect sperm from several to dozens of males before depositing each clutch. Field-caught females lay egg batches with low genetic relatedness as a result of mixed paternity, clearly showing that female promiscuity and incomplete sperm displacement are not laboratory artifacts (Kamimura 2003b). A staged mating experiment with a paternity success analysis revealed that large males dominated male–male competition for burrows housing females, which resulted in repeated copulation with the same female. Despite the low paternity gain per copulation, such repeated copulations resulted in a significant increase in paternity for larger males with higher resource-holding potential (RHP; Kamimura 2013). The copulation frequency of males increased with male body size, which resulted from repeated copulations by large males rather than an increase in the cohabitation rate (Kamimura 2013). Thus, male–male competition is an important factor in the mating system of E. plebeja. On the other hand, no obvious precopulatory mate choice or refusal behavior has been detected (Kamimura 2013). These findings indicate that male body size and genital length are subject to sexual selection. Nevertheless, a quantitative genetic study revealed that both male body size and genital length show sizable heritability (narrow sense heritability h2 = 0.41 and h2 = 0.45, respectively; Kamimura & Iwase 2010). Although longer virgae seem to confer benefits during SC (see van Lieshout & Elgar 2011a, discussed below), elongation of virgal length may be counter-selected by increased fragility and the risk of breaking during copulation (Kamimura & Matsuo 2001; Kamimura 2003a).
Van Lieshout and Elgar (2011a) examined the benefits of having longer genitalia, which is almost as long as their body length, for male M. brunneri. Male genitalia of this species are essentially similar to that of E. plebeja. They allowed virgin females to mate sequentially with two different males for two consecutive days (24 h for each male). To determine the paternity success of the two contestants, one of the two males, the first or second one, had been sterilized by exposure to gamma-radiation from a 60Co source. As a result, they detected a moderate median P2 = 0.57 (the paternity success of the second mate but without controls of mating frequencies: note that “P2” here is used not as in the common definition), with very large variation, but most observations were distributed above parity. They used a generalized linear mixed model (GLMM) to explore the determinants that explained the variation in P2 among virga length, age and the size of the seminal vesicle. In their samples, body size was weakly correlated with virga length, but it was not included in the analysis. They detected two interaction terms, between the virgal length of the first and second males, and between the age of the first and second males, as significant determinants of P2. Their post hoc analyses revealed that when preceded by a male with long virgae, a second male with short virgae earned a reduced P2. Similarly, old second males that followed young first males had a low P2. In other words, older males or males with shorter virgae could not displace as much sperm from a spermatheca that was filled with sperm from the first male. Van Lieshout (2011) further analyzed the relationships between male morphology and mating behavior in M. brunneri. Although he did not measure paternity success, his analysis revealed that both virga length and body size (no significant correlation was detected between these two traits) were positively correlated with total copulation duration in this species. This observation suggests that not only virgal length itself, but also copulation duration (and possibly copulation frequency), which is correlated with it, may cause the observed defensive benefits during SC in male M. brunneri with longer virgae (van Lieshout & Elgar 2011a). As discussed above, sperm removal in E. plebeja is incomplete and males must repeatedly copulate to achieve high fertilization success. Judging from the mixed paternity after multiple matings by second males (van Lieshout & Elgar 2011a), a similar phenomenon likely occurs in M. brunneri. However, van Lieshout and Elgar (2011a) did not measure the copulation frequency of each male and they only measured the total duration of copulation for a subset of the samples. The latter did not explain variation in P2. Additional comparative studies are required to resolve these inconsistencies among anisolabidid species.
Because the elongated spermatheca, which is longer than the sperm-removal organ (virgae), is the proximate cause of the observed incomplete sperm removal in E. plebeja, we must consider this phenomenon from the female perspective. Similar incomplete sperm removal by a single copulation has also been reported in several other animal species (Siva-Jothy & Tsubaki 1989; Cordero et al. 1995; Wada et al. 2005). However, the functional significance of this phenomenon for females is generally unclear. Among many possible explanations (see Kamimura 2013 for details), including the procurement of fresh young sperm while minimizing the risk of mating with an infertile male, CFC hypothesis provides a plausible explanation for incomplete sperm manipulation by males if it promotes SC and increases the chance of having genetically superior offspring. To determine if female E. plebeja realize their optimal Plast to accumulate sperm from males with high RHP (genetic benefits), a simulation study was conducted using real data sets for mating patterns observed under a ♂–♂–♀ experimental design (Kamimura 2013). In the simulation, males were randomly resampled based on their cohabitation frequencies. For each cohabitation, mating frequency was randomly determined by resampling from the real data set of mating frequency per cohabitation for the chosen male. By repeating this random resampling for a given number of sampled males (1, 5, 10, 15 or 30 males), mating sequences for 10 000 females were determined (e.g. a small male (1.3 mm pronotum width) comes and copulates two times, the female then copulates four times with a larger male (1.7 mm pronotum width), and so on). For a given value of Plast (from 0 to 1) and a mating sequence, the resultant paternity share of each male can be calculated. For example, when Plast = x, a male gains x of the paternity share from its first copulation, while 1 − x will be sired by former rival males. If the male repeats copulation once more with the same mate, the expected paternity is x + (1 − x)x, where the first and second terms represent the paternity gain by the second and first copulations, respectively. Then, given the paternity share values and the heritability of male body size, the expected average body size of sons can be calculated for each female. The simulation showed that a moderately low value of Plast (≈0.2) is optimal for producing large-sized sons when a female E. plebeja mates with several to several dozens of males before oviposition. Given that male body size in E. plebeja, which is positively correlated with RHP, is a heritable trait (h2 = 0.41; Kamimura & Iwase 2010) and that the actual Plast is approximately 0.2, the genetic benefit was estimated to be a 1.4% increase in the mating success of sons, which corresponds to a change from 1.540 to 1.544 mm in the pronotum width of sons. These genetic benefits disappeared when males were assumed to copulate only once per cohabitation. Why can female earwigs gain genetic benefits by allowing only incomplete sperm displacement? If females allow almost complete sperm removal (Plast ≈ 1) or they do not allow it at all (Plast ≈ 0), almost all of the offspring will be sired by only the last or the first male, respectively, that copulated with the female. In E. plebeja, large males with high RHP do not necessarily visit females frequently, and the simulation confirmed that such sperm-storage strategies, which are vulnerable to stochasticity in the arrival order of males, produce no genetic benefits. Instead, large males of E. plebeja have an advantage in contests over shelters with females; thus, they can repeatedly copulate with the same female during prolonged cohabitation in a shelter. The elongated spermatheca of E. plebeja, which moderately restricts the amount of sperm displaced per single copulation, can properly reflect this variation in male RHP to the resultant paternity shares.
The discussion above also shows that although mating frequencies and sequences are determined by precopulatory processes, moderately restricted sperm displacement (a postcopulatory process) is an essential factor that allows female E. plebeja to accumulate genes from superior males. In other words, properly restricted sperm displacement increases the chance of accumulating sperm from superior males, which satisfies the definition of CFC adopted here. In the absence of both detectable costs of mating and direct benefits from frequent remating, CFC is also thought to explain the propensity for promiscuity in E. plebeja females (Kamimura 2013).
Conclusions and future directions
Substantial evidence indicates that sexually dimorphic earwig forceps are used as weaponry in male–male contests. In contrast, although forceps are frequently used in courtship, evidence that females choose males based on forceps shape is quite limited, without reproducible results from multiple study groups. In particular, no study has succeeded in conclusively showing that FA in forceps is used during precopulatory female mate choice. For the evolution of earwig genitalia, including the number, laterality, size and allometry of genital parts, previous studies have demonstrated the importance of postcopulatory sexual selection (SC and CFC) as a driving force. However, as reviewed here, only a few representative species have been studied for each aspect of their reproductive biology. The following five points are especially important for future studies.
Reducing taxonomic bias
As reviewed here, among the more than 2000 extant earwig species, only a few have been studied extensively. Partly because of their circumtropical distribution, ecological studies on the basal Dermaptera (Karschiellidae, Diplatyidae and Pygidicranidae), Apachyidae, Chelisochidae, and the epizoic earwigs (Arixeniina and Hemimerina), are quite rare. Together with further refinements in the estimated phylogeny of earwigs, reproductive biology should be studied in additional representative species from various dermapteran taxa. Fortunately, earwigs are usually easy to rear under laboratory conditions; thus, they have the potential to become new model organisms for evolutionary biology.
Untangle the complicated evolution of weaponry or ornamental traits
Unlike genital traits, forceps length and forceps morphs are strongly correlated with body size. Accordingly, many studies on the functions of forceps have tried to use various experimental and/or statistical techniques to control the confounding effects of body size (e.g. Table 1). However, for two distantly related earwig species, F. auricularia (Forficulidae) and A. maritima (Anisolabididae), original, pre-manipulated forceps length has been reported to affect the outcome of male–male contests (Styrsky & Van Rhein 1999; Munoz & Zink 2012). These studies used field-collected adults, and thus the observed trends could have been caused by long-lasting effects from past experiences in the adult stage. Alternatively, other presently unknown factors correlated with forceps size may play important roles in sexual selection. One possible candidate is aggression. As reviewed here, male–male combats in many earwig species occur easily over female mates, food resources, shelters, or even when no resources exist in an arena. Although immature stages usually cannot be sexed, such contest behaviors are also likely to occur between male nymphs. Studies on the ontogeny of aggression and combat behaviors are therefore also warranted. Such studies are not feasible for model organisms in holometabolous insects such as the dung beetle O. taurus. In a population of F. auricularia, Rantala et al. (2007) showed that a trade-off existed between male relative forceps length and a measure of immunocompetence, specifically lysozyme activity of the hemolymph against a bacterium species. This measure of immunocompetence increased with overall body size in both sexes, and also with relative forceps length in females. This correlation between sexually selected forceps and other fitness components indicates the need for further comprehensive studies on sexually selected traits. The presence of presently unknown factors correlated with forceps length may also explain the large observed overlaps in body size between two distinct male morphs in many earwigs.
Genetic basis of the evolution of genital laterality
As reviewed above, the genetically determined right-handedness in penis use in L. riparia likely evolved from an ancestor that showed AS with no genetic control of handedness in penis use. Although this evolutionary pattern (from random AS to genetically controlled DA) is common among animals (Palmer 1996), the underlying processes remain unclear. Recently, a theoretical model was proposed to explain the apparent absence of a genetic basis in empirical examples of AS (Kamimura 2011). The theory predicts the coexistence of two types of genes, handedness genes that determine chirality (right- and left-morphs) and randomization genes that randomize chirality, in a population in which negative frequency-dependent selection works on the two morphs. To date, genital AS and its genetic basis have been studied in only a single anisolabidid species. Together with the development of noninvasive observation methods for determining penis laterality, further studies are required to explore the genetic and developmental controls of AS in penis use, as well as possible negative frequency-dependent selection working on it.
Sexual conflict before and after the establishment of genital coupling
In contrast to sexual selection, for which many studies have clarified its importance in the evolution of weaponry and genital traits, no comprehensive study has examined sexual conflict and SAC in any earwig species. One plausible cause of this bias is the superiority of females in determining of mating frequency: female quiescence is usually necessary for the establishment of genital coupling. However, as introduced above, using mouthparts and forceps, males of P. prominens seem to be able to conduct coercive matings with unwilling females (Briceño & Eberhard 1995). The spined areas of male and female genitalia in P. dorsalis, which mesh with each other during copulation, also suggest the importance of SAC for genital evolution in earwigs.
Interrelationships between pre- and postcopulatory sexual selection
The studies reviewed here showed that interruptions of mating pairs by non-copulating males and the takeover of mating by them are common in earwigs; examples were from F. auricularia (Radesäter & Halldórsdóttir 1993b; Forslund 2000, 2003; Walker & Fell 2001), M. brunneri (van Lieshout & Elgar 2011b) and E. plebeja (Kamimura 2013). Interruptions of mating by a second male were also reported in D. taeniatum and P. prominens (Briceño & Eberhard 1995). Because males can not use forceps for male–male combat during mating, smaller or weaker males of M. brunneri can sometimes break up the mating of a male with a higher RHP (van Lieshout & Elgar 2011b). Similar phenomena have been reported in F. auricularia (Forslund 2000) and E. plebeja (Kamimura 2013), which indicates that females find it difficult to assess male quality based solely the outcome of combat. In contrast with animals that mate in open spaces, many terrestrial earwig species mate in narrow and dark habitats. Postcopulatory processes may provide a robust method for collecting superior sperm under conditions when it is difficult to execute precopulatory mate choice (Kamimura 2013). The development of molecular markers for paternity analyses (e.g. Guillet & Deunff 2000) will facilitate additional comparative studies of the interrelated evolution between pre- and postcopulatory sexual selection processes in earwigs.
I thank C-Y Lee (Universiti Sains Malaysia) and his students for providing me with an ideal intellectual environment during the writing of this review, as well as M Nishikawa (Ehime University) for important references and valuable information on the taxonomy of the Dermaptera. I am also grateful to Joseph L Tomkins (University of Western Australia) for invaluable comments on an early draft. This research was partly supported by a Grant-in-Aid for Scientific Research from JSPS (No. 22770058).