In gonochoric species, male reproductive success is typically limited by access to unfertilised eggs while female reproductive success is limited by resources to produce eggs (Bateman's principle). This principle is assumed to apply likewise to male and female functions in hermaphrodites [1,11,12]. If the reproductive success of the female function is limited by the availability of resources for egg production and not by the availability of allosperm (sperm received from a donor), this may lead to a conflict of interest between the sperm donor and the sperm receiver regarding the use of sperm [1]. For example, most if not all received sperm might be digested rather than be used for fertilization [10]. In addition, if individuals mate repeatedly as a female before egg fertilization, sperm from different donors will compete for fertilization [6,12-15]. Under these conditions, individuals should on the one hand persuade their partners to mate with them, but on the other hand suppress their willingness to remate later, adding to the mating conflict.

Repeated mating and sperm competition in combination with sperm digestion have been predicted to lead to increased male investment [16], at least under certain conditions [17]. The resulting high costs associated with sperm donation may explain why many internally fertilising hermaphrodites apparently hesitate to donate sperm and prefer mutual insemination [18]. Conditional reciprocity of sperm transfer, i.e. sperm trading, was first described for the opisthobranch Navanax inermis [19,20]. In N. inermis mating partners repeatedly alternate the female and the male role [19,20]. This results in balanced overall durations of sperm transfer between partners during a mating sequence, consistent with sperm trading [21]. The only experimental evidence for sperm trading thus far has been obtained from the sea slug Chelidonura hirundinina by controlling a slugs' ability to donate sperm without affecting other aspects of mating behaviour [22]. Here, focal animals more rapidly abandoned partners that did not transfer sperm.

Observing bilateral insemination does not suffice to infer conditionality since reciprocity can also result from a mutual, but unconditional willingness to donate sperm [5]. Several studies concluded against sperm trading despite explicit reciprocal insemination [23-25]. In a study of the snail Lymnaea stagnalis reciprocal sex role alternation only occurred when both mating partners had been isolated, which led the authors to question conditional reciprocity as a per se explanation for reciprocal mating patterns [26]. In addition, body size influences insemination durations and the degree of reciprocity in at least three different sea slug species [15,27,28] which makes the question of conditional reciprocity even more complicated.

The present study first describes the as yet unique mating behaviour of Elysia timida. We then examine whether exchange between partners is balanced and if there are indications for conditional reciprocity.

Figure 1 Mating sequence of E. timida. A Head-to-head contact B Copulation position C Hypodermic transfer, showing extended penises (p, marked red) D Start of circling E Circling F End of circling G Sperm transfer into female aperture. Drawings by V. Schmitt from (more ...)

Figure 2 Two E. timida individuals in copulation position. (from video recording).

Twenty-two out of 25 pairs started penis protrusion immediately and simultaneously during initial head-to-head contact or first copulation position. The remaining pairs everted their penis soon after initial contact. In twenty pairs both partners kept their penis everted throughout the complete mating sequence.

When partners had taken up copulation position, they either stayed in position for a hypodermic transfer attempt (Fig. 1C, Video 2) or they started to circle (Fig. 1D, Fig. 3, see below).

Figure 3 Mating E. timida individuals performing the stereotypic anticlockwise circling movement. (from video recording).

If starting a hypodermic transfer attempt, partners stretched their penis up to one third of their body length to reach the dorsal surface of their partner (Fig. 1C). The movable penis tip touched the partner's dorsal surface in an attempt to insert. In 25.9% of insertion attempts the penis transmitted a white substance on or into the dorsal surface of the partner. Substance transfer by hypodermic injection lasted from 3 s to 32.25 min ( = 6.47 ± 7.63 min, N = 44 individuals). Several such hypodermic transfers occurred per mating sequence ( = 2.68 ± 2.30, range: 0 – 9, N = 50 individuals).

After each hypodermic injection the penis was withdrawn from the partner (but as a rule not fully retracted) and individuals started a synchronous, stereotypic circling movement in a counter-clockwise direction (Fig. 1D, Fig. 3, Video 1). If only one individual started circling, its partner usually responded by circling as well (Fig. 1E and Fig. 3, Video 1). Most (81%) circling phases involved a single circle only (1.06 ± 3.04 min, N = 50 individuals). In the other cases individuals kept circling, in one case up to 40.8 min. After resuming copulation position, a new hypodermic transfer attempt was started (Fig. 1B–C). Transfer attempts and circling movements were alternated 1 to 34 times during a single mating sequence ( = 12.7 ± 6.6, N = 25 pairs).

The hypodermic injection and circling phase described above represented the longer part of the mating sequence. It was followed by a short final phase in 21 out of 25 matings. Here, individuals took up copulation position again, but inserted their penis into the partner's female aperture (Fig. 1G, Video 3). Mates inserted mutually and almost simultaneously and transferred sperm in a short time period ( = 1.61 ± 0.75 min, range: 0.57 – 4.12 min, measurements available for N = 41 individuals). Out of the 21 matings with a final insemination phase 20 mating sequences ended with a reciprocal final insemination. Only one single mating ended with a unilateral final insemination.

After reciprocal final sperm transfer slugs withdrew and separated simultaneously, gliding off into opposite directions. Complete mating sequences lasted 42.9 ± 14.7 min (range: 10.8 to 79.5 min, N = 25). After insemination and separation a white blob of semen remained attached to the outside of the female aperture. Investigation under a microscope revealed that sperm had a simple thread-like shape without a prominent head and showed vigorous spiralling movement of the middle section. In structure and movement the sperm of E. timida correspond strongly to that described for E. patina [42].

Figure 4 The main area of hypodermic transfers in the dorsal notum. (highlighted).

Although E. timida shows hypodermic injection, no special penis armature in the form of a stylet or hook(s) was found on inspection with SEM and stereo-microscopy. One animal that had received four hypodermic transfers notum was fixed 5 h after copulation. On inspection with SEM four bumps were identified in the main insertion area. An individual kept in isolation had a very smooth notum without bumps. These observations confirm our assumption that hypodermic injection results in the transfer of a substance.

Figure 5 Range plot of total hypodermic transfer duration per mating for both partners (represented by the minimum and maximum) for 25 pairs, ranked by the minimum value per pair. Dashes indicate pairs with identical values for both partners. There is a high degree (more ...)

Figure 6 Range plot of number of hypodermic transfers per mating for both partners (represented by the minimum and maximum) for 25 pairs, ranked by the minimum value per pair. Dashes indicate pairs with identical values for both partners. There is a high degree (more ...)

The duration of final sperm transfer was also balanced between partners but differed significantly between pairs (Kruskal-Wallis χ² = 45.1, d.f. = 24, P = 0.006; Fig. 7), coinciding with the synchrony with which these transfers were performed by both partners. The four matings in which partners did not transfer sperm into the female aperture occurred in pairs in which only one partner had transferred hypodermically while the other had not.

Figure 7 Range plot of the duration of sperm transfer into the female aperture per partner (represented by the minimum and maximum) for 25 pairs, ranked by the minimum value per pair. Dashes indicate pairs with identical values for both partners. There is a high (more ...)

Figure 8 Average durations of single unilateral and bilateral hypodermic transfers per mating in 11 individuals (see text). Individuals transferred clearly longer by bilateral than by unilateral transfers.

There are also possible indications for conditional reciprocity in the final mating phase. Individuals often touched the partner's female aperture with their penis several times before final sperm transfer and seemed to "wait" for the partner to react reciprocally. Final sperm transfer was reciprocal and approximately simultaneous except for two cases. Here, only one partner transferred sperm, which was followed by an exceptionally long circling phase of about 30 min. This is in contrast to the immediate separation after reciprocal sperm transfer. In one of these two cases, partners finally transferred reciprocally and then separated immediately. Additionally, both pairs displayed repeated, alternating penis retractions and protrusions while circling, which was never the case in other matings. This aberrant behaviour suggests a conflict over failed reciprocity.

From the perspective of the sperm donor, prostate fluids can represent a manipulating or stimulating cocktail that changes the physiology of the receiver in the interest of the sperm donor [5,52-54]. In a pre-insemination context, the injections might serve to manipulate the partner to accept the less desirable sex role, as seems to be the case in the gastropterid sea slug Siphopteron quadrispinosum [44]. In E. timida, pre-insemination manipulation may be required as a part of a ritualised courtship aiming at synchrony and reciprocity. In a post-insemination context, prostate secretions can possibly influence sperm usage and increase the relative paternity of the sperm donor as is achieved by the love dart in Helicid snails [55,56]. Similar mechanisms are likely to exist in other hermaphrodites [5,33]. It is important to stress that – even if hypodermic injection may have evolved from a donor-specific benefit, we do not exclude that receivers may receive some form of benefit from being injected – as suggested by the "cooperative" and ritualised behaviour lacking signs of avoidance.

Hypodermic transfer may, however, also involve sperm transfer. Support for this comes from the observation that hypodermic insemination is widespread among sacoglossans [15,34,39-43], but transfer of fluid without sperm has never been reported. In Alderia modesta it was shown that hypodermic insemination anywhere in the body leads to successful egg fertilization [15]. The copulation position and insertion area described for Elysiasubornata, which inseminates exclusively hypodermically [42], are very similar to what we describe for the hypodermic transfers in E. timida. This similarity includes the position and appearance of the bumps seen after hypodermic injection. These observations lend further support to the possibility that E. timida is unique in that it combines standard with hypodermic sperm transfer. All the possibilities for stimulation and manipulation before and after sperm transfer discussed above remain in effect under this scenario.

If it is sperm that is injected hypodermically, it is important to stress that the pattern of sperm exchange is not based upon alternating insemination as in Navanaxinermis [21] or Chelidonura hirundinina [22], but on repeated, mostly simultaneous transfer. Simultaneous transfer is one possible type of sperm trading [18], but has never been described as a repeated process. Other unusual characteristics of E. timida mating behaviour such as slow movement and circling may facilitate reciprocity. Altogether the mating behaviour of E. timida is the exact opposite of that of the "hit-and-run" species in which mating individuals attempt to inseminate a partner quickly and unilaterally [57]. E. timida partners rather seem to "sit-and-wait" to assess the degree of synchrony and reciprocity they receive from their partner. A measure indicating this hesitance is the long duration of the whole mating sequence.

The mating behaviour of E. timida represents a hitherto undescribed combination of repeated hypodermic injections and standard sperm transfer into a female genital aperture. If the hypodermic injections transfer sperm, this mating system could represent a transition between the two insemination types and a key system to analyse the evolution of insemination mechanisms.

Sexual behaviour was very synchronised and balanced between partners, leading to a high degree of reciprocity. The synchrony within pairs versus the differences between pairs indicates that mates adapt their behaviour in response to that of their partner. Hypodermic injections were significantly shorter when performed unilaterally than bilaterally. The final sperm transfers were always reciprocal except for only two non-reciprocal cases in which individuals reacted with aberrant behaviour indicating a conflict over non-reciprocity. Combined, our findings suggest that mating in E. timida is likely to be based on conditional reciprocity or sperm trading.

Each pair was observed until partners had separated completely after a successful mating sequence. From the first five complete mating sequences an ethogram of the species specific mating behaviour was defined and a code for behavioural elements was developed. The occurrence and duration of these behavioural elements were recorded. Data were obtained for 25 complete mating sequences involving 50 different individuals.

The present study was a part of VS's diploma project. VS carried out the behavioural studies in Banyuls sûr mer, performed the statistical analyses, and drafted the manuscript. NA carried out histological analyses and assisted in data interpretation and manuscript editing. NKM supervised the study, participated in its design and coordination and helped to analyze the data and draft the manuscript. All authors read and approved the final manuscript.

We would like to thank the Westphalian Wilhelms-University of Muenster for providing financial support to VS for the diploma project in France. Many thanks are due to the Laboratoire Arago, Banyuls sûr mer, for providing laboratory facilities. Furthermore we thank Joe Lange for his work at the SEM. Many thanks also to Heike Wägele and one unknown referee for their valuable comments on the manuscript.