Introduction

Behavioural diversity or ethodiversity (Cordero-Rivera 2017a) is a fundamental level of phenotypic variability that can promote population stability in changing environments (e.g. Dingemanse et al. 2004), increase adaptability at the intraspecific level (Berger-Tal and Saltz 2016) and inform about evolutionary processes. Ethodiversity can thus play an important role in species resilience and consequently in how we can manage species for conservation strategies. Unfortunately, while many conservation strategies address the extinction of species and its possible cascading effects across trophic levels (Pérez-Méndez et al. 2016), the disappearance of behaviours is rarely considered (Caro and Sherman 2012). Behaviours that are exclusive of a limited number of individuals, populations or species are relevant from a conservation perspective, because they are in higher risk of extinction (Caro and Sherman 2012). For instance, sexual conflict can be an engine of speciation, because it may trigger antagonistic coevolution between the sexes, potentially leading to rapid divergent evolution of the characters related to reproduction (e.g. Cordero-Rivera 2017b). Therefore, the study of reproductive behaviours is relevant for our understanding of diversification and speciation.

Insects from the order Odonata (dragonflies and damselflies) represent a clear example of rare behavioural diversity, regarding their copulatory behaviour. In most insects, copulation is performed with both partners oriented in opposite directions (Downes 1969; Rutowski 1982) (Fig. 1a). However, the reproductive behaviour of odonates is unique because during copulation, male and female remain attached by two points, forming the so-called “copulatory wheel” (Fig. 1b). In this position, the male anal appendages located at the tip of the abdomen grasp the female thorax (or head), and the female genitalia located at the distal part of the abdomen contacts with the male secondary genitalia, positioned in the proximal part of the abdomen. Such copulatory position occurs because the male intromittent organ is situated in the second and third abdominal segments (secondary genitalia), while the male primary genitalia are located in the ninth abdominal segment. Therefore, before insemination, the male must translocate sperm from the primary to the secondary genitalia, a behaviour called intra-male sperm translocation (hereafter, ST) (Fig. 1c–f). The ST behaviour is not found in any other insect (Shuker and Simmons 2014; Córdoba-Aguilar et al. 2018), although indirect insemination is also found in the cephalopods (octopuses, squids, cuttlefishes, nautilus and allonautilus; with the exception of some deep-sea species) and some arachnids from the order Araneae (spiders), subclass Acari (mites) and the order Solifugae (sunspiders). The ST is thus a clear example of diversity at the behavioural level with important evolutionary implications, because the viability and quality of the sperm, and hence fertility, will depend on the timing of ST relative to copulation (Rivas-Torres et al. unpublished).

Fig. 1
figure 1

Insects usually mate with both sexes facing opposite directions, like in butterflies (a Hipparchia sp. Fabricius, 1807; Satyridae). However, in odonates (b Orthetrum brunneun (Fonscolombe, 1837), Libellulidae), the sexes are orientated to the same direction, allowing flying in copula. The intra-male sperm translocation can take place after the male has grasped the female in tandem, like in c Neurobasis chinensis (Linnaeus, 1758) (Calopterygidae), with or without precopulatory genital touching. In some species, this behaviour is repeated up to seven times in the same copulation, like in d Coenagrion scitulum (Rambur, 1842) (Coenagrionidae). Finally, in some species, males translocate sperm alone before copulation, like in e Diphlebia lestoides (Selys, 1853), or after copulation, very close to the female as in f Euphaea masoni Selys, 1879 (Euphaeidae). Pictures were from A. Cordero-Rivera, except panel e, courtesy of Reiner Richter

Full size image

However, little is known about its evolution, its relation with the evolution of the copulatory wheel, and particularly its diversity within the order. Our main aim was therefore to explore the variation of ST behaviour from an evolutionary perspective and discuss its possible connection with the copulatory wheel. We performed a literature review on the diversity of ST across Odonata, and we used phylogenetic comparative analyses to investigate the evolution of the ST behaviour within this insect order. Our literature review revealed differences in ST behaviour within and between species, and the comparative phylogenetic analyses suggested ST in tandem and before copulation, as the ancestral behaviour in Odonata. We discuss potential evolutionary routes for all the ST behaviours found, which could help to better understand behavioural evolution in this insect group, as well as certain aspects of the evolution of sexual behaviours and divergence in other animals.

Material and methods

Literature review of ST behaviour

Published data on ST in odonates were searched by querying Google scholar (http://scholar.google.com) for “sperm translocation” and “odonat*” or “intra-male sperm transfer” and “odonat*”. Searches were carried out in December 2016. Additional searchers were completed in November 2018 looking for ST behaviour in papers written in Spanish, French, German and Italian and manually looking in old English references, where the behaviour could be described in different terms (e.g. “filling of the sperm vesicle”). Japanese references were also included thanks to the help of colleagues. We also screened manually the papers published in Martinia (from 1991 to 2016), and the Journal of the British Dragonfly Society (from 1995 to 2016), which are non-digitised odonatological journals, and furthermore, we manually searched non-digitised issues of Odonatologica and the International Journal of Odonatology. We collated all available observations of ST, including how this behaviour was performed, and its duration (with sample size and standard error). We compiled information for a total of 176 species (see “Results”). Full details of the literature reviewed, variables considered and species included in the analyses are given in Supplementary Materials 1 and 2. For the purposes of our study, we have categorised ST in four characters: (i) before or after copula, (ii) alone or in tandem, (iii) non-repeated or repeated, and (iv) with or without precopulatory genital touching (see Table 1 and “Results”). Since the description of the precopulatory genital touching might not be stated explicitly in the revised papers, we established that when something similar to this behaviour was described just before the ST, we considered it as precopulatory genital touching.

Table 1 Summary of the diversity of sperm translocation behaviour in the Odonata (N number of species, % percentage of variants of ST behaviour)
Full size table

Phylogenetic tree reconstruction

We aimed at understanding the evolution of ST using phylogenetic comparative analyses (Harvey and Pagel 1991). Therefore, we first constructed a phylogenetic tree including the odonate species for which data on ST were available in the literature. We searched GenBank (Clark et al. 2016, https://www.ncbi.nlm.nih.gov/genbank/) for sequences of the mitochondrial genes COI and 16S, and the nuclear genes 18S and 28S. In total, we retrieved sequence data for 129 species out of the 176 for which we had data on ST. We also retrieved sequences of the same genes from 31 species without information on ST, to increase the resolution of the tree, and for six species of Ephemeroptera, to be used as outgroups in the analysis. The final dataset included a total of 160 Odonata species (including 31 without data for ST), which represented 14 families of Zygoptera and 11 families of Anisoptera (see Supplementary Material Table S2).

All sequences were imported into Geneious version 9.1.7 (http://www.geneious.com, Kearse et al. 2012) for visual inspection before alignment. Sequences were aligned using the ClustalW algorithm (Thompson et al. 1994), as implemented in Geneious. We used BEAST version 2.4.8.0 (Bouckaert et al. 2014) to reconstruct the phylogenetic relationships among the study species. The phylogenetic reconstruction was performed with the nucleotide substitution models selected for each gene by the bModelTest package version 1.0.4 (Bouckaert and Drummond 2017), and a strict clock model and the Yule speciation model as priors. The analysis was run for 10 million generations and sampled every 1000 generations. The output was examined with Tracer version 1.6 (Rambaut and Drummond 2014) to assess convergence of the Markov-chain Monte Carlo onto a stationary distribution through the analysis of trace plots and effective sample sizes (ESS) of the model parameters (ESS > 200 was considered acceptable). TreeAnnotator version 2.4.8, included in the BEAST package, was used to build a maximum clade credibility tree after discarding 10% of sampled trees as burn-in. The consensus tree was visualised and edited with FigTree version 1.4.2 (available at: http://tree.bio.ed.ac.uk/software/figtree/). This tree was pruned to contain only those species with available ST information using the R package ape (Paradis et al. 2004), and the pruned tree was subsequently used for the phylogenetic comparative analyses.

Phylogenetic comparative analyses

We estimated the ancestral states of our study discrete characters related to sperm transfer using the function ace in the R package ape (Paradis et al. 2004). Our character states were ST performed (i) before vs. after copula, (ii) in tandem vs. alone vs. both in tandem and alone, (iii) with repetition vs. with no repetition vs. with and without repetition, and (iv) with precopulatory genital touching vs. without precopulatory genital touching vs. with and without precopulatory genital touching (Robertson and Tennessen 1984). We note that the combined states (i.e. both in tandem or alone) correspond to intra- or interpopulation variation of the ST behaviour. We used maximum likelihood estimation with equal rates of transition (Pagel 1994), and the likelihood of the ancestral states was computed using a joint estimation procedure (see Pupko et al. 2000). Finally, we also investigated possible correlated evolution between pairs of characters. For this analyses, we first recoded the study characters to binary. We coded the combined states to the least common variant of ST for each character. Our rationale behind this decision is that we wanted to give more weight to the least common behaviours, which are already under-represented in our dataset. Even though species with more than one type of ST have not stably evolved only one ST behaviour, they could offer hints on the evolution of ST across odonates. We used Pagel’s (1994) function to detect correlated evolution as implemented in the R package phytools (Revell 2012), with the function fitDiscrete from the package geiger (Harmon et al. 2008).

Results

Variability of ST behaviour

We found a total of 123 papers that provided information on ST from 176 species of odonates, belonging to 22 families (Table 1, Supplementary Materials 1 and 2). The ST is a behaviour that usually lasts for some seconds but shows high variability (mean ± SE 16.7 ± 2.7 s, range 0.2 to 150 s, N = 82 species) (Table 1).

At least five different variants of ST were described across species: (i) male alone, before the formation of precopulatory tandem; (ii) in precopulatory tandem, without genital touching, only once per copula; (iii) in precopulatory tandem after genital touching, only once per copula; (iv) in tandem, repeated during copulation; and (v) male alone, after copula (see Table 1). However, in some species [e.g. Libellula quadrimaculata (L., 1758) or Erythemis simplicicollis (Say, 1839)], we found a combination of the ST variants due to intra- or interpopulation variability (Fig. 2).

Fig. 2
figure 2

The variants of intra-male sperm translocation behaviour, mapped on a phylogeny of the studied Odonata. The ancestral behaviour of sperm translocation was estimated to be before copula, in precopulatory tandem, non-repeated and without genital touching. Family names have been added to the main branches of the tree

Full size image

Overall, the majority of species (66%) perform ST in precopulatory tandem (Table 1), hence before copulation and only once per copula. Anisoptera males perform ST alone, before tandem formation in a higher proportion (31%) compared with Zygoptera (9%). The ST in tandem and repeated was found for 4% of species of Zygoptera, but no in Anisoptera (Table 1). Finally, ST alone, after copula, was found for 6% of the species of Zygoptera and 3% of Anisoptera (Table 1).

The families Coenagrionidae and Libellulidae are the most speciose and also display more variants of ST behaviour, including all five in the first family. The majority of coenagrionids studied performed ST in precopulatory tandem, and only once per copula (58%; Table 1). The second most common variant was the ST in tandem, but only after precopulatory genital touch (32%; Table 1). In libellulids, ST in tandem, only once per copula, was also the commonest variant (61%; Table 1), and the second commonest ST was male alone, but in this case, before tandem formation (33%; Table 1). ST by the male alone, after copulation, is a rare behaviour but is the dominant behaviour in the Euphaeidae (57%) and Polythoridae (67%), and it is also performed by the only living representative of Pseudolestidae.

Evolution of ST behaviour

DNA sequences were available for 129 out of the 176 species for which information on ST behaviour was available in the literature (Supplementary Material Table S2). The consensus phylogenetic tree was congruent with the currently accepted and relatively well-established phylogeny of the order (e.g. Lorenzo-Carballa and Cordero-Rivera 2014, but see Dijkstra et al. 2013), and the support for the main clades was high in most cases (posterior probability support higher than 0.6 and in many cases close or equal to 1; see Supplementary Material Fig. S1).

In our dataset, the majority of species (95%) perform ST before copula (Fig. 2). Six species perform ST after copula (four Zygoptera and two Anisoptera). According to the ancestral reconstruction for this trait, the variant of ST after copula has evolved independently in all cases except for the group Euphaea/Anisophaea (Zygoptera). Regarding ST in tandem and/or alone, most species perform it in tandem (76%), while 22 species do it alone (17%; seven Zygoptera and 15 Anisoptera), and nine species use both strategies (7%; three Zygoptera and six Anisoptera). These variants have independently evolved in all cases except for the group Euphaea/Anisophaea (Zygoptera), and the clade Nannothemis/Pachydiplax/Erythemis and for the species of Leucorrhinia (Anisoptera). In the case of ST performed repeatedly or not, we found that the majority of species (96%) perform it just once. Only within Zygoptera, we found three species that repeat sperm transfer (2%) and two that perform both variants (2%; corresponding to five independent origins). Regarding ST performed with or without genital touching, only five species have ST with genital touching (4%) and three perform both variants (2%; all of them within Zygoptera). These cases correspond to independent origins except for the group of Ischnura ramburii, I. graellsii and I. elegans. In summary, the ancestral state of ST would be before copula, in precopulatory tandem, non-repeated and without genital touching.

Finally, we did not find any significant correlated evolution between the different variants of ST (P > 0.05).

Discussion

Our review of the literature indicates that there are at least five variants of ST behaviour among Odonata. Our work also highlights the lack of basic information on the ST behaviour for the vast majority of species within this insect group: The total number of odonate species is estimated around 6000 (Lorenzo-Carballa and Cordero-Rivera 2014); however, we only found data on ST behaviour for 176 species.

Variation of ST behaviours

In most of the odonate species studied, males perform ST after having grasped the female in precopulatory tandem and once per copula (variant ii). This variant of ST is the ancestral state, according to our comparative phylogenetic analyses. Variant (i), the completion of ST by the male alone, before finding a female (Table 1), could be advantageous in species where females are rarely encountered and copulations are brief. Males that performed ST alone would be ready to copulate as soon as they grasp a female in tandem.

Variant (iii) involves performing ST only after precopulatory genital touching (Table 1), a behaviour that presumably signals female receptivity to the male (Robertson and Tennessen 1984). This behaviour might derive from the basal behaviour (i.e. variant ii), if males commonly encounter unreceptive females. High densities of unreceptive females would be predicted in populations/species whose females remain near the reproductive site in the maturation phase, like many Ischnura (Cordero et al. 1998). When females give no refusal signs, males can remain in tandem for very long periods, even for a full day in the laboratory (Cordero et al. 1992). Genital touching would therefore be adaptive for males, avoiding wasting sperm with unreceptive females, but also for females, because they would be released faster if signalling their unreceptiveness. In odonates, it has been assumed that females cannot be forced to copulate (e.g. Fincke 1997), although forced matings are possible in populations with high male densities and females ovipositing unguarded (Cordero-Rivera and Andrés 2002).

Only two species, Coenagrion scitulum (Rambur, 1842) and Megaloprepus caerulatus (Drury, 1782), are known to routinely perform variant (iv), i.e. repeated ST in one copulation (Table 1, Fig. 1d). This behaviour has been also recorded occasionally in three other zygopteran species: Lestes barbarus (Fabricius, 1798) (Lestidae), Ischnura aurora (Brauer, 1865) (Coenagrionidae) and Perissolestes remotus (Williamson & Williamson, 1924) (Perilestidae) (see Supplementary Material 1). In the case of C. scitulum, the repetition of ST during the copulation has been interpreted as a mechanism evolved in the context of sperm competition (Córdoba-Aguilar and Cordero-Rivera 2008). In this species, males are apparently unable to remove a significant portion of sperm from rivals using their genital ligula (Cordero et al. 1995). Therefore, by repeating ST and insemination, they might over-compete rival sperm. However, other species of odonates [e.g. Hypolestes trinitatis (Gundlach, 1888)] are also known to have limited ability to remove sperm, but ST is not repeated (Torres-Cambas and Cordero-Rivera 2011), and hence the link between both phenomena is not straightforward.

The most intriguing variant for ST is variant (v), i.e. the translocation of sperm by the male alone, after copulation (Fig. 1f, Table 1). This behaviour was first reported for the coenagrionid Mortonagrion hirosei Asahina, 1972, from Japan (Naraoka 2014), and then observed in nine additional species, including four Euphaeidae, and Pseudolestes mirabilis Kirby, 1900 (Cordero-Rivera and Zhang 2018). Translocating sperm after copulation could be explained if these species have evolved physiological mechanisms to maintain sperm alive for long periods of time (until next mating). However, at least for the first copula, males have to fill their sperm vesicle before copulation. Males could routinely perform ST each morning to be prepared for the next mating, but this has never been observed in P. mirabilis, the only species studied in detail (Cordero-Rivera and Zhang 2018).

In some species, we found several of the ST variants due to intra- or interpopulation variability. For example, males of Hetaerina americana Fabricius, 1798 were reported performing ST alone before copula in one population, in tandem without precopulatory touch in another population and in tandem with precopulatory touch at a third locality (Supplementary material Table 1). Two ST variants were also reported for different populations of I. aurora and Libellula quadrimaculata Linnaeus, 1758, and between males of a single population of Aeshna cyanea Muller, 1764 (Supplementary material Table 1). Intraspecific variability seems rare, and unlikely to be detected in short-term studies. For instance, only 5% of the 137 ST observed in two Enallagma species occurred before tandem formation (Logan 1971; cited by Corbet 1999). The reasons behind this intraspecific diversity of ST are unknown, but these species are excellent candidates for further investigation of the diversification of ST strategies within odonates.

Why has ST evolved?

The evolution of ST behaviour is likely related to the atypical copulation position in the Odonata. One plausible scenario for the evolution of ST behaviour is that ancestors of modern odonates produced a spermatophore, and deposited it on the substrate, a behaviour currently observed in arachnids, myriapods and wingless hexapods (Proctor 1998). The thick cerci of Namurotypus Brauckmann and Zessin, 1989 males (ancestor of the carboniferous Odonata) could have been used to firmly grasp the female behind her compound eyes. The male could then have directed the female over the spermatophore (Bechly et al. 2001), in a way similar to the “drag off” behaviour observed in whipspiders (Amblypygi) (Weygoldt 1969). Attaching the spermatophore to the male body is clearly more efficient and could be the selective pressure required for the evolution of secondary genitalia in odonates (reviewed in Cordero-Rivera and Córdoba-Aguilar 2010), and hence, the ST behaviour.

An alternative, yet more speculative hypothesis for the evolution of ST, could be related to sexual cannibalism, since females of several zygopterans are known to sometimes attack and eat mature conspecific males (Cordero 1992). We are aware of only one case of sexual cannibalism described during copulation: Robertson (1985) observed a female of Ischnura ramburi (Selys, 1850) that was repeatedly chased and grasped by a male until she finally attacked him and ate out his thorax, but the male succeeded initiating copulation before dying in copula. Other predatory animals, like some arachnids or cephalopods, have a specialised appendix used as an intromittent organ to introduce the spermatophore in the female reproductive organs. These two groups also show sexual cannibalism (Ibáñez and Keyl 2010; Li et al. 2012). Therefore, it could be hypothesised that the tandem and wheel position of odonates during copulation, along with the intra-male ST, might allow males to avoid sexual cannibalism (Chapman et al. 2003; Schneider 2014). From this idea, we propose the hypothesis that when the risk of sexual cannibalism is high, selection favours the evolution of secondary mechanisms to inseminate females safely for the males. This could be tested with a review of the insemination behaviours across different animal taxa and within a phylogenetic context.

Conclusions

Some of the ST variants seem to have evolved several times, but this behaviour needs further investigation in a larger number of species. No other insect group shows a behaviour equivalent to the ST of odonates (although some similarities can be found with a special appendage of some arachnids and the hectocotylus of cephalopods), and consequently we lack comparative evidence to understand the evolution of this enigmatic behaviour. ST is sometimes performed very fast, and therefore careful video recording of this behaviour, particularly in Anisoptera, is needed to avoid confusing movements to clean the abdomen with true ST. We currently lack information for a number of families of both Anisoptera (e.g. Austropetaliidae, Neopetallidae and Macromiidae) and Zygoptera (e.g. Chlorocyphidae, Amphipterygidae and Lestoideidae), and most of the species with information on ST are from temperate zones and fewer from tropical areas. This bias can be explained by the scarcity of field studies in tropical regions, and also due to the low sexual activity observed in tropical species (e.g. Sanmartín-Villar and Cordero-Rivera 2016). Although, tropical odonata usually appear at very low densities, they are priority for future research, because we expect a higher diversity of behaviours in tropical areas (Cordero-Rivera 2017a), and rare alternatives might be more common in these tropical families.

Our study emphasises also the relevance (and scarcity) of detailed natural history observations for most species. We expect that this review will encourage the scientific community towards more research in diversity of reproductive behaviours, with a special focus on tropical species. This applies not only to odonates, but to other animal taxa. This information, combined with modern molecular techniques and phylogenetic hypotheses, is fundamental to understand relevant questions about behavioural evolution as well as behavioural diversity (ethodiversity). This is a necessary step to increase awareness on the importance of conserving not only species but also behaviours (Cordero-Rivera 2017a).