Sexual conflict and the evolution of genitalia: male damselflies remove more sperm when mating with a heterospecific female

In Calopteryx damselflies, males remove rivals’ sperm stored by the female, thereby reducing sperm competition. This behaviour may create a sexual conflict, because females could lose the sperm stored in the spermatheca, used for long-term storage. Comparative evidence suggested antagonistic coevolution between sexes, which might prompt the evolution of narrow spermathecal ducts, or longer spermathecae, hindering sperm removal. Calopteryx haemorrhoidalis and C. splendens coexist and sometimes hybridize. Therefore, here I predicted that if females coevolve with conspecific males, heterospecific males should have an advantage when interspecific matings occur because females will show less resistance to them than to conspecific males. By hand-pairing females to males of both species, I found that in intraspecific and interspecific matings, sperm was almost completely removed from the bursa (97–100%), but only partially from the spermathecae, with more spermathecal removal in interspecific (63–71%) than intraspecific matings (14–33%). This suggests that heterospecific males are more efficient in sperm removal as predicted by a sexually-antagonistic coevolutionary scenario. Furthermore, in most cases, only the left spermatheca was emptied, suggesting that the evolution of more than one spermatheca might also be a female counter-adaptation to regain control over fertilization.

studies have used cross-species matings to test for genital coevolution. In Ohomopterus carabid beetles, interspecific matings result in genital damages to males and females, due to mechanical incompatibilities, supporting a lock-and-key mechanism 14 . In Drosophila, micron-scale differences between species in their genitalia also result in wounding and infection to females, suggesting sexually antagonistic coevolution between the sexes 15 .
Here I conducted experimental copulations between species of the damselfly genus Calopteryx to study the mechanisms behind male-female coevolution of genitalia. In these damselflies, an elaborated pre-copulatory courtship prevents interspecific matings, which are rare even when two or more species coexist 16,17 . Calopteryx species show little morphological differentiation in male genitalia, differing mainly in size 18 . The copulatory mechanisms of Calopteryx are well known. Females store sperm in a large bursa copulatrix and a "Y-shaped" spermatheca. At the moment of oviposition, the egg passing down contacts with a group of sensilla situated in the vaginal plates, and this elicits sperm ejection for fertilization 19 . Such communication between sensilla and the spermathecae has been corroborated when neural microsurgery of sensilla was carried out, which impeded spermathecal sperm emptying after sensilla stimulation 20 . Males have a penis head that removes sperm from the bursa copulatrix and two lateral horns, which are used to remove sperm from the spermatheca 21,22 . During the first part of copulation (stage I), males remove sperm from the bursa and (in some species) also from the spermatheca, whereas in stage II the sperm is transferred to the female 23 . The spermatheca is used for long-term sperm storage, and in some species males apparently cannot remove sperm from this organ 19 , resulting in higher genetic diversity in spermathecal sperm 24 . For these reasons, the sperm in the spermatheca has been proposed as the centre of sexual conflict in Calopteryx 25 . Males are expected to develop mechanisms to remove spermathecal sperm, and females are expected to evolve resistance, such as a reduction of spermathecal lumen to impede the introduction of male genitalia in that organ or an increase in spermathecal length 25 . Males could also elicit females to eject sperm by mimicking with their ligula the movements of eggs during fertilization, stimulating mechanical sensilla embedded in the female vaginal plates 26,27 . This mechanism could be interpreted as evidence for cryptic female choice, due to male stimulation. If this were the case, then more sperm is expected to be ejected with increasing number of vaginal sensilla 28 .
In this paper, I bypass the precopulatory courtship by "hand-pairing" 29 individuals of C. haemorrhoidalis and C. splendens. The studied population of C. haemorrhoidalis shows spermathecal sperm removal, with a minimum of sperm remaining after 60 movements of stage I 25 . The ability of males of C. splendens to remove spermathecal sperm is unclear, with evidence against 24 and in favour 30 , and was therefore also addressed in this study. If sexually antagonistic coevolution is the main mechanism behind genital evolution, the expectation is that females should be able to resist conspecific male attempts to remove spermathecal sperm, and maintain control over fertilization. When females are mated to heterospecific males, the prediction is that males should be able to remove more sperm from the spermatheca, because in this situation, females could not evolve resistance mechanisms. If cryptic female choice is of relevance, then more sperm should be ejected by females having a higher number of vaginal sensilla.

Results
Anatomy of genitalia. Male genital ligula ends in a head with two lateral processes (horns) covered by spines, and was very similar between the studied species ( Fig. 1 and Table 1). In both species, the right horn was longer, but only significantly in C. splendens (Table 1).
Female genitalia was, nevertheless, clearly different between species, with C. splendens females having longer spermathecae ( Fig. 1 and Table 1). In both species, the width of male horns is narrower than the width of the common duct of the spermatheca (Table 1), indicating that direct sperm removal from the spermathecae using male genital horns is possible in intra-and interspecific matings. Females of C. splendens had more sensilla in the vaginal plates than females of C. haemorrhoidalis (on average five more; Table 1, Fig. 1), and in both species there were no differences between sides. Therefore, if sperm is ejected by male stimulation, C. splendens females should expel more sperm.
Copulatory behaviour. Copulation lasted 1.9-2.3 minutes in intraspecific matings, and also when males of C. splendens were mated to females of C. haemorrhoidalis (Fig. 2). Nevertheless, copulation was shortened to 1.2 minutes when males of C. haemorrhoidalis were mated to females of C. splendens (h-s matings). A two-way ANOVA was used to analyse the effect of male and female species and their interaction on copulation duration. In the case of stage I, there was a significant effect of the male species (F 1,79 = 6.06, p = 0.016), with shortened duration by male C. haemorrhoidalis (Fig. 2). The female had no significant effect (F 1,79 = 2.34, p = 0.130), but there was a highly significant effect of the male × female interaction (F 1,79 = 9.52, p = 0.003), because interspecific matings were shorter. The duration of stage II was not significantly affected by male identity or the interaction (p > 0.209), although there was a marginally significant effect of female species (F 1,79 = 3.17, p = 0.079).
Half of h-s matings monitored (6 out of 11) had great difficulties to start, with several attempts before a secure genital contact was achieved. This mating direction was difficult to obtain, because many females did not cooperate, and refused to bring their abdomen in contact with male genitalia, and other pairs were unable to engage genitalia after several attempts and separated. On the other hand, in one mating between a male C. splendens and a female C. haemorrhoidalis, excluded from the above averages, copulation lasted 14.55 minutes, because the pair was for several minutes unable to disengage genitalia, indicating that also this mating direction may have some mechanical incompatibilities.
Sperm removal in C. splendens. The volume of sperm in the bursa copulatrix was drastically reduced (almost to zero) after only 20 movements of stage I in C. splendens matings (Fig. 3) and the differences were therefore highly significant among groups (ANOVA, F 5,27 = 6.74, p < 0.001). The volume of the spermatheca was also reduced with increasing number of movements in the interrupted females (regression analysis, F 1,23 = 4.34, p = 0.048), but the reduction was small, so that differences between groups were not statistically significant (ANOVA, F 5,27 = 1.09, p = 0.388; Fig. 3).
An ANOVA analysing the effect of species identity of the male and the female, their interaction and the number of sensilla as a covariate on the spermathecal volume after 60-120 movements (N = 28 females), suggests that both spermathecae are emptied with different efficiency (Fig. 5). On the left spermatheca, the only factor with a significant effect is female species (F 1,21 = 4.53, p = 0.045), with a smaller sperm volume in C. haemorrhoidalis (Fig. 5). The species identity of the male, male × female interaction and the covariate had no significant effect Note that the tridimensional position of the horns is not natural, because they were placed in a plane by a coverslip, to allow measuring their length. There were no significant differences between species in horn length (from the basis to the tip) and width (measured as indicated by the red segment on the middle of the horn). The lower figures show the female genital tract and sperm storage organs of mated females at the same scale as male genitalia. The sperm stored is the dark mass inside the bursa and the spermathecae. The spermathecae (L = left, R = right) were significantly longer in C. splendens, suggesting this is a counter-adaptation to difficult male sperm removal from that organ. (Scale bar = 0.5 mm, common for all images).
(p > 0.140). On the right spermatheca, the volume of sperm remaining was positively correlated with the number of sensilla (F 1,20 = 5.63, p = 0.028), with a marginally significant effect of male (F 1,20 = 3.54, p = 0.075) and no effect of female species or the male × female interaction (p > 0.120; Fig. 5).

Discussion
This experiment clearly shows that males of Calopteryx are able to remove sperm more efficiently when they mate with a heterospecific female, supporting a scenario of male-female coevolution in genital traits by sexual antagonism 2 . I have found no evidence of cryptic female choice by sperm ejection (see also ref. 30).
Experiments on artificial selection for high and low mating rates have shown that genitalia evolves very fast, supporting the antagonistic coevolution as one likely mechanism 31,32 . The theory of sexually antagonistic coevolution is based on the premise that repeated interactions between sexes determine an arms-race due to the reciprocal evolutionary change that each sex imposes to the other 33 . Females may develop resistance to male adaptations, particularly when these produce injuries during sexual interactions, such as copulatory wounds 34 or develop anti-clasping mechanisms 35 . Alternatively, females may develop tolerance mechanisms, to minimize the negative effects of male adaptations, such as behavioural 36 , anatomical 37 and physiological responses 38 . My results indicate that in Calopteryx damselflies, genital coevolution is better explained by sexually antagonistic mechanisms, rather than by cryptic female choice. These results are in concordance with the higher success of heteropopulation (but conspecific) males in sperm competition experiments with the yellow dung fly (Scatophaga stercoraria), also interpreted as a case of sexually antagonistic coevolution due to allopatry 39 .
The genus Calopteryx has developed one of the most elaborate precopulatory courtships of damselflies, with males displaying wing colour, abdomen colour and oviposition substrates to potential mates 40 . Ethological  Numbers indicate sample size. Interspecific matings were significantly shorter, due to reduction in stage I, the phase when sperm removal occurs.
divergence is clear 41 , but morphological difference negligible 18 . When C. haemorrhoidalis and C. splendens coexist in the same stream, a situation particularly common in Central Italy, interspecific interactions are frequent, but interspecific matings are rare, due to the ethological isolation promoted by precopulatory courtship. Nevertheless, some hybridization occurs 16 . My results indicate that heterospecific males may have higher fertilization success than conspecific males, if sperm volumes translate into fertilization rates, and may explain why some hybrid males are found in the field 16 .
It is interesting to note that males of the two Calopteryx species studied here show little divergence in the anatomy of the penis head, the part of the genitalia used to remove the sperm from previous matings 21 , a result which contrasts with the typically larger divergence in male genitalia between species 3 . Allopatric populations of Calopteryx diverge in their genital morphology more than in non-genitalic traits 25 , as expected from male-female coevolution. In some populations, males are able to physically remove sperm from the bursa copulatrix and the spermatheca, but in other populations, the spermathecal lumen is so narrow that direct sperm removal seems impossible 27 . In fact, a study of a French population of C. xanthostoma (considered sometimes as a subspecies of C. splendens) showed no evidence for spermathecal sperm removal 24 , while German populations of C. splendens do remove spermathecal sperm 30 , as the Italian population studied here. When males cannot physically remove spermathecal sperm, it has been shown that they sometimes exploit a pre-existent sensory bias in females by stimulating the vaginal sensilla (used by females in egg fertilization), which elicits sperm ejection by the female 42 .  . The volume of sperm in the bursa copulatrix and spermatheca of double mated females after intraspecific and interspecific matings (postcopula treatment; the male species is indicated first in the legend), compared to the volume stored by females mated to a conspecific male and then to a conspecific or heterospecific male, but interrupted at the end of stage I (60-120 movements), before insemination. The volume of sperm in the bursa is reduced by 97-100%. The volume of the spermatheca in intraspecific matings is reduced by 14% in C. haemorrhoidalis and 33% in C. splendens. In contrast, volume reduction is 63-71% in interspecific matings.
In the two species studied here, male genital horns are narrower than the duct of the spermatheca of conspecific and heterospecific females, indicating that they can physically remove sperm from the spermathecae 25,30 . Given that copulation duration is under male control in damselflies 43,44 , one possible explanation for the short duration of interspecific copulations is that males emptied the spermatheca very fast, because only stage I is shortened (Fig. 2). The duration of stage II, when males inseminate, did not change. The sperm volume of the spermathecae was clearly reduced. This is easily explained by physical sperm removal. The possibility of sensory stimulation to elicit sperm ejection as the mechanism 42 , is not supported by the asymmetry in sperm removal, and the positive relationship between the volume of sperm remaining and the number of vaginal sensilla, when the expectation from this mechanism is a negative relationship 28 .
I have found clear differences in spermathecal sperm volume when comparing conspecific and interspecific matings (Fig. 4). In contrast with other interspecific copulations, where genital damage is common 34 , no evidence for genital damage was found in this experiment. Under the hypothesis of reciprocal evolution of genitalia by sexual selection, females are expected to coevolve with males and therefore be able to better resist in a scenario of sexual conflict 45 . In Calopteryx, the source of conflict is the sperm stored in the spermatheca 7 . In agreement with my predictions, males remove more sperm from heterospecific females, and this occurs in both directions. This pattern may be explained by antagonistic coevolution: females cannot coevolve with heterospecific males, and are therefore less able to resist to male manipulations. Males might also stimulate vaginal sensilla of heterospecific females more efficiently than conspecific males, and as a result females eject more sperm 26 , but I have found no evidence for this mechanism. In fact, even if the number of sensilla on C. splendens females is clearly higher than in C. haemorrhoidalis, there was no more sperm removal/ejection from this species (see below). Females of C. splendens have evolved long spermathecae, clearly longer than male genital horns, in contrast to C. haemorrhoidalis, where male and female genital sizes are almost identical. The evolution of these long spermathecae might be an example of a female counter-adaptation to make sperm removal more difficult 25 . The fact that males of C. haemorrhoidalis are able to remove 63% of spermathecal sperm of C. splendens females (after 60 movements) is therefore surprising, particularly compared to the 33% achieved by conspecific males (after 60-120 movements), and again suggests that sexually antagonistic coevolution is the explanation for this pattern.
Probably the most intriguing pattern is the asymmetrical sperm removal found in this experiment. This asymmetry had previously found in the same population of C. haemorrhoidalis 25 and more recently in Calopteryx cornelia, from Japan 46 . It seems that the morphology of penis head precludes the insertion of both horns simultaneously into the spermathecal common duct, and males specialise in using only one horn for spermathecal sperm removal 46 . The fact that left and right horns are similar in length (albeit the right one is longer) needs further research. The duplication of the spermatheca in Calopteryx might be another example of female counter-adaptation to male sperm removal ability 25 . It would be rewarding to examine whether asymmetric sperm removal is common in damselflies with Y-shaped spermatheca, which is a key prediction of this hypothesis.

Methods
Fieldwork was carried out at two localities near Pontecorvo (Frosinone, Central Italy), where C. haemorrhoidalis and C. splendens have large populations, in August 2004. In the river Forma Quesa, C. haemorrhoidalis was the dominant species, whereas in the river Melfa, more specimens of C. splendens were found. Both rivers are tributaries of the river Liri, and are separated by about 10 km. Damselflies were captured with a hand net, individually marked with and indelible pen, and introduced to an outdoor insectary (2 × 2 × 2 m), mounted on the shore of the river. To obtain matings, females were hand-paired to a male 29 , by allowing the male to grasp the prothorax of the female with his anal appendages, and then releasing the female. Most pairs mated a few seconds after this manipulation, usually on the hand, which allowed a close observation of copulatory activity.
To study mating duration, females were hand-paired to a conspecific male (N = 19 females of C. haemorrhoidalis and 37 females of C. splendens) or to a heterospecific male (N = 16 females of C. haemorrhoidalis and 11 females of C. splendens) and allowed to complete copulation. For each mating, I measured the duration of the stage I, the number of pumping movements and the duration of stage II, to the nearest second. Using an ANOVA, the duration of stage I, stage II and number of movements were compared including the male species, the female and their interaction as explanatory variables. Response variables were normalised by a Box-Cox transformation before analysis.
To study sperm removal, a reciprocal hand-pairing experiment was designed. To control for previous female mating history, all females were first mated to a conspecific male. This was made to make sure that all females had sperm before the experimental mating. Previous research indicates that after 60 movements of stage I, males of C. haemorrhoidalis from this population have removed most of the sperm stored by females from previous matings 25 . In the case of C. splendens, the ability of males to remove bursal sperm was known for other populations 24 , but not for the study population. Therefore sperm removal was studied by comparing the volume of sperm stored by mated females (N = 8) with the volume remaining in females interrupted after 10, 20, 40, 60 and 120 movements of stage I (N = 5 in all cases) of their second mating.
After the first (intraspecific) mating, females were allowed to mate with a second male of the same species, or with a male of the other species, but copulation was interrupted after 60 movements of stage I in the case of male C. haemorrhoidalis (N = 4 females of each species). For C. splendens, 10 females interrupted after 60-120 movements from the previous experiment, were used for the intraspecific matings. A further group of 10 females of C. haemorrhoidalis were mated first to conspecific males and then to males of C. splendens and interrupted after 60-120 movements. In all cases, these copulations were interrupted before insemination. Specimens were immediately preserved in 70% ethanol and stored in individual plastic vials. To minimize the number of specimens used in the experiments, some males were used once as sperm donors and once as sperm removers. In the laboratory, females were dissected and the sperm storage organs extracted. In Calopteryx, these organs are a large bursa copulatrix, which receives the male genital ligula during copulation, and a "Y-shaped" spermatheca. Female genitalia were mounted on a slide and two insect pins of 0.2 mm of diameter placed over the slide, to allow the cover-slide to remain at this distance. The volume of sperm (in mm 3 ) was estimated by calculating the area of the sperm mass using an image analysis software (ImageJ, https://imagej.nih.gov/ij/) and multiplying by 0.2 mm 47 . One sample was damaged during dissection and could not be measured. Final sample size for each group was 4-19 females (see Fig. 4). The volume of sperm was compared between treatments (mated females and experimentally interrupted females after 60-120 movements of stage I) and between intra-and interspecific matings using a two-way ANOVA with the interaction term. No transformation of sperm volume was needed to meet the normality assumption.
The sperm vesicle of 14 males of C. haemorrhoidalis and 17 males of C. splendens whose copulation was unsuccessful or interrupted during stage I was also dissected and the sperm mass mounted on a microscope slide, to measure its area and estimate volume. This represents the amount of sperm that males inseminate, because after copulation the sperm vesicle was empty (N = 4 males of C. haemorrhoidalis and 3 of C. splendens). Male genitalia was measured as in previous studies 48 . Results are reported as mean ± SE (N) throughout the text. Statistical analyses were done with xlStat (www.xlstat.com) and Genstat software 49 .