High Variation in Last Male Sperm Precedence and Genital

applyparastyle “fig//caption/p[1]” parastyle “FigCapt”

Biological Journal of the Linnean Society, 2020, 130, 497–506. With 2 figures.

High variation in last male sperm precedence and genital morphology in the emerald damselfly, Lestes sponsa Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 FRANK JOHANSSON1,*, DAVID BERGER1, JACOB HÖGLUND1, YVONNE MEYER-LUCHT1, PATRIK RÖDIN-MÖRCH1, SZYMON SNIEGULA2 and PHILLIP C. WATTS3

1Department of Ecology and Genetics, Animal Ecology, Uppsala University, Uppsala, Sweden 2Department of Ecosystem Conservation, Polish Academy of Sciences, Krakow, Poland 3Department of Biological and Environmental Science, University of Jyväskylä, Jyväskylä, Finland

Received 24 February 2020; revised 30 March 2020; accepted for publication 30 March 2020

In organisms in which individuals mate multiply, knowledge of the proportion of offspring sired by the last male

to mate (P2) under field conditions is important for a thorough understanding of how sexual selection works in

nature. In many insect groups, pronounced intraspecific variation in P2 is commonplace. Interestingly, however, in

stark contrast to these observations, compilation of P2 data in dragonflies and damselflies (Odonata) indicates that

a high P2, seldom below 0.95, is a feature of this taxon. Here we used double digest restriction-site associated DNA sequencing to generate a panel of single nucleotide polymorphisms (SNPs) with which we could determine paternity

and estimate values of P2 in the offspring of 19 field-collected pairs of the emerald damselfly Lestes sponsa. We also

estimated the relationship between P2 and male genital shape of 16 males using geometric morphometric analysis.

P2 was variable (range = 0.0–1.0; mean = 0.5), and there was a marginally non-significant (P = 0.069) relationship

between genital shape and P2, suggesting that males with a high P2 had an aedeagus with a broader tip. We suggest

that the high P2-values reported in past studies in Odonata are partly due to the methods used to infer paternity. Use

of SNPs to determine patterns of paternity and P2 in odonates is needed for a better appraisal of fitness in odonates, and would open many future avenues for use of odonates as models of sexual selection.

ADDITIONAL KEYWORDS: ddRADseq – last male sperm precedence – P2 – sexual selection – aedeagus.

INTRODUCTION (Parker, 1970; Gwynne, 1984) and in several species, females can gain control over fertilization by storing Sexual selection represents competition between sperm in specialized organs, potentially intensifying members of the same sex to gain a reproductive both sperm competition and sexual conflict (Wigby & advantage (Hosken & House, 2011; Lindsay et al., Chapman, 2004). 2019). In certain instances, intersexual conflict is Odonates (dragonflies and damselflies) are an insect an apparent outcome of mating behaviour, such as taxon in which sperm competition is predicted to be territoriality (or sneaky mating) where males attempt common, as female often mate multiply (e.g. Alcock, to monopolize (or gain surreptitious) access to females. 1979; Fincke, 1984, Lowe et al., 2009; Thompson et al., When females mate multiply, sexual selection and 2011) and can store sperm in specialized organs (bursa cryptic female choice may operate through sperm copulatrix and spermatheca). A characteristic feature competition (Parker, 1970; Córdoba-Aguilar & Cordero- of many odonate species, however, is that males Rivera, 2008; Cordero-Rivera & Córdoba-Aguilar, 2010) use their genitals (often adorned with specialized and males need to utilize strategies to improve the appendages) to remove sperm from previous matings odds that their sperm takes priority to fertilize eggs. from the female’s reproductive tract before copulation Sperm competition is common in many insect taxa (Waage, 1979, 1982, 1983; Cordero-Rivera, 2017). A capacity for sperm removal presents a potential mechanism to bypass sperm competition, especially *Corresponding author. E-mail: [email protected] when this behaviour is accompanied by male guarding

© 2020 The Linnean Society of London, Biological Journal of the Linnean Society, 2020, 130, 497–506 497 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected] 498 F. JOHANSSON ET AL. of females during oviposition that would prevent Variation in fertilization success might be caused by sperm removal by a rival male. intraspecific variation in male genital morphology, but One key parameter in understanding sperm so far few studies have been done on Odonata. competition is the proportion of offspring sired by the Our study species, Lestes sponsa (the emerald last male to mate (P2) (Parker, 1970). Using carefully damselfly or common spreadwing), is a zygopteran designed experiments, the pattern of P2 can be used to that has a wide Palaearctic distribution, most identify adaptive male phenotypes, or female patterns typically in lentic habitats (Boudot & Kalkman, 2015).

of sperm utilization and the opportunity for post- Mature males and females are most common in July Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 mating mechanisms such as cryptic female choice and August. Copulation by L. sponsa typically lasts (Firman et al., 2017). Last male sperm precedence between 23 and 42 min and after mating the males has been studied in many insect taxa where the guard the females for the entire period of oviposition, typical pattern is pronounced intraspecific variation albeit with strong interference competition (Stoks in P2 (Harvey & Parker, 2000). In odonates, the et al., 1997). Stoks (2001) found that as much as 52% volume of a previous male’s sperm that is removed of L. sponsa males never mate during their visits to from the female reproductive tract by a male during a breeding site, and that lifetime mating success was copulation varies considerably, within and among determined mainly by the ability to acquire matings species (reviewed by Corbet, 1999, 2004; Cordoba- during visits. With this type of behavioural study, an Aguilar et al., 2003). However, this variability in sperm important assumption is that behavioural observations removal effectiveness has been regarded as of little reflect a male’s potential fitness because the observed importance for the paternity of the clutch of eggs (and mating at the breeding site results in a fertilized hence the opportunity for sperm competition) that clutch of eggs. However, this might not necessarily be is laid immediately after copulation. For example, in so (Alcock, 1994). Values of last male sperm precedence Corbet’s (1999, 2004: 521) review of dragonfly biology are not known for L. sponsa. Here, we use a double it is stated that: ‘Recorded values for P2 in eggs laid digest restriction-associated DNA sequencing protocol immediately after copulation are seldom less than (ddRADseq) to generate a panel of single nucleotide

95%’. Low values of P2 have only been associated with polymorphisms (SNPs) which we used to determine clutches that are laid long after the last mating; for paternity in clutches of larvae of L. sponsa. We find P2 example, P2 = 0.5 7–8 days in Mnais pruinosa (Siva- to be variable, and often less than 1. We discuss these

Jothy & Tsubaki, 1989) and P2 = 0.5 after 6 days in data in light of potential drivers of last male sperm Ischnura senegalensis (Sawada, 1998). Cordero & precedence, including the genital morphology of the Miller (1992) also found that the second male sired 0, genus Lestes. 22, 38 and 100% in four crosses examined. The reasons for this temporal contrast might be that females have exhausted the sperm in the bursa copulatrix and use MATERIAL AND METHODS sperm that has been stored in spermathecae and/ or have had sufficient time to mix sperm (Cordoba- Paternity Aguilar et al., 2003; Corbet, 2004), or that females can Twenty copulating pairs of L. sponsa were collected affect paternity if complete sperm removal does not from two locations in Uppsala, Sweden (Gnista: occur, and thus favour some males over others (Andrés 59.839105°N, 17.730887°E and Gränby: 59.873389°N, & Cordero-Rivera, 2000). 17.679381°E), on 10–21 August 2018. The females

Another reason for variation in P2 among taxa might used for the collection of eggs/offspring were a mixture be due to morphological variation in male genital of mating and tandem guarding pairs, and the pairs shape (Simmons, 2013). This variation in genital collected had been in tandem or mating for at least morphology among taxa holds also for odonates 1 min before we caught them. The males were caught (Miller, 1991; Cordoba-Aguilar et al., 2003) and has and preserved in 99% alcohol in the field, whereas been hypothesized to account for differences in sperm the females were transferred to 200-mL plastic cups removal effectiveness (Cordoba Aguilar et al., 2003). with a lid. The lid had small holes for air circulation

With this in mind, the high value of P2 that apparently and the inside was lined with moist filter paper. characterizes odonates (Corbet, 1999: 521), despite the The plastic cups were transferred to a laboratory at observed variation in genital morphology, is a special Uppsala University, Sweden, and put in a climate case. However, before firm conclusions can be drawn room, with a temperature 20 °C and a photoperiod of about P2 in odonates, studies on more species are a 14:10-h day–night cycle. Females oviposited eggs needed. Here we provide an example of P2 estimates in into the filter after 2–4 days. Thereafter the females a damselfly using a modern molecular technique and were preserved in 99% alcohol. The filter papers with also provide an estimate of the relationship between eggs were transferred to 1-L containers with 0.5 L of

P2 and genital morphology at the intraspecific level. aged non-chlorinated tap water. Eggs started to hatch

© 2020 The Linnean Society of London, Biological Journal of the Linnean Society, 2020, 130, 497–506 SPERM PRECEDENCE AND GENITAL MORPHOLOGY IN A DAMSELFLY 499 after 2–3 weeks. We waited until all eggs had hatched (Wang, 2004; Jones et al., 2009). Paternity of clutches and then we randomly picked 20 offspring from each was assigned on the basis that maternal–offspring female’s egg clutch and reared larva individually relationships were known (i.e. females laid clutches for 3 weeks in 1-L containers. Larvae were fed 300 of eggs and the maternal genotype is known for each laboratory reared brine shrimp each day. After 3 weeks, clutch). We allowed 10% of SNPs to mismatch between the larvae were preserved in 99% alcohol. parents and offspring, allowed relatively high (10%) DNA was extracted from thoraxes of parents and the error and allele dropout rates (to be conservative),

entire larvae using a high-salt protocol (Paxton et al., and we assumed that there was an 80% chance that Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 1996). Sufficient high-quality DNA was available the father’s genotype was present in the sample of for nine (Gränby) and ten (Gnista) families, with 15 candidate fathers. As we cannot know whether we offspring from each family (except for one family from have sampled all the potential fathers, the choice of Gnista where 14 offspring were available). Genotyping 80% chance that we sampled the father was thought was achieved using ddRADseq (Peterson et al., 2017), to reflect the putative high probability that the male as described in Johansson et al. (2017). Briefly, libraries caught in tandem with a female had sired many (if not were created by digesting DNA with EcoRI-HF and all) of the offspring in that female’s clutch, while still MseI, after which adaptors with individual tags were allowing for the occurrence of prior, cryptic mating. ligated with T4 DNA ligase, and PCR was conducted Fathers were assigned on basis of best maximum- (replicated four times per sample) with Q5 DNA likelihood scores, using the full likelihood option (with polymerase (all New England Biolabs, Ipswich, MA, high precision) in COLONY as this has been identified USA). After PCR, samples were pooled and size- as the most accurate method (Wang, 2012). selected on agarose gel (see Johansson et al., 2017 for We also assigned paternity using CERVUS v.3.0 details of methods). Libraries were sequenced on a (Kalinowski et al., 2007), which uses categorical single lane of an Illumina HiSeq2500 flow-cell (125- allocation based on maximum-likelihood assignment bp paired end) at SciLifeLab, Uppsala, Sweden (www. tests to identify the most likely parent–offspring scilifelab.se), to generate 180 million read pairs. The combinations (Marshall et al., 1998; Jones et al., 2009). STACKS v.2.2 (Catchen et al., 2011, 2013) denovo-map We specified the known mother–offspring relationships. pipeline was used to reconstruct loci and call SNPs. We Critical values of delta (the difference between LOD set the number of mismatches allowed between stacks [logarithm of multilocus likelihood] scores for best within individuals (-M) and between individuals (-n) and second best father–mother–offspring triads), used to 2. The final data set was filtered in the populations to determine the most likely father, were calculated program of STACKS, retaining only the first SNP of using parental genotypes to estimate population each RAD-tag that could be found in 80% of all the allele frequencies, and by simulating 50 000 offspring, individuals at a minor allele frequency ≥0.05. At the with 100 candidate fathers, 0.58 proportion of loci end of the procedure, we identified 581 (Gnista) and typed (to account for the missing genotype data) and 3184 (Gränby) variable sites. 0.05 proportion of mistyped loci (to allow genotyping For samples from Gnista, raw SNP data were filtered errors). Paternal assignments were accepted when using VCFtools v.0.1.13 (Danecek et al., 2011) to retain they bettered the appropriate critical value of delta at only bi-allelic loci that had a minimum genotype a relaxed (80%) confidence limit, a procedure that may quality of 30, and a minimum and maximum read increase the number of type I errors, but nonetheless depth of 4 and 10 reads respectively, and no more than increases the chance that one of the sampled ‘true’ 50% missing data. For Gränby, SNP data were filtered fathers is also accepted as a father compared with a as described above but with no more than 40% missing 95% confidence limit (i.e. type II errors are reduced). data. This filtering procedure left 139 and 210 SNPs available for paternity assignment in the families from Gnista and Gränby respectively; this number of Correlation between paternity and genital SNPs for paternity analysis on L. sponsa is within the morphology typical range of SNPs (60–200) suggested to confer To examine if there was a relationship between reasonable power for parentage analysis (Flanagan & P2 values from the paternity analyses and genital Jones, 2019). morphology we used the program tpsRegr (Rohlf, Paternity of offspring was assigned using COLONY 2003). This program explores the relationship between v.2.0.6.5 (Jones & Wang, 2010), which uses maximum morphology (both uniform and non-affine shape likelihood to group offspring genotypes into sibships components) and one or more independent variables, that are clustered into maternal and paternal families. in our case P2. Male genital was captured by dissecting Candidate parents are then assigned to clusters, and the male penis (also named aedeagus) and taking if no suitable candidate parents are found COLONY a photo from the ventral side using a dissecting will reconstruct appropriate parental genotypes microscope equipped with a side-mounted camera

© 2020 The Linnean Society of London, Biological Journal of the Linnean Society, 2020, 130, 497–506 500 F. JOHANSSON ET AL. lucida (LeicaR MZ8; Leica Microsystems GmbH, infrequently one of the other males within our sample Wetzlar, Germany). Thereafter, nine landmarks (Fig. 1) of available male genotypes (e.g. 3% of the paternity of the aedeagus were captured from the photos with assignments in Gränby and 25% of assignments at the program tpsDig (Rohlf, 2017). After landmarks had Gnista, where two sample males were assigned high been captured they were subjected to a Generalized paternity in the two clutches 4 and 5, Table 1). Given Procrustes Analysis (GPA; Rohlf & Slice, 1990) in the the many other males at breeding sites, it seems program TpsRegr (Rohlf, 2003). This program was unlikely that our small sample of males would have

also used to examine, statistically and visually, the (by chance) been the male that sired offspring in the Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 relationship between aedeagus morphology and P2. To other clutches within our sample. The incidence of last examine whether male body size affected P2 we also male sperm precedence (P2) was comparatively low performed a regression on P2 against male size (total (Table 1). For example, only four out of the 19 (21%) length). families examined had a P2 > 0.85 and in four families the candidate last male was never assigned as the

father (P2 = 0.00). Indeed, although values of P2 in L. sponsa varied among individuals, average (±SD) P RESULTS 2 was quite low: 0.433 ± 0.36 (Gnista) and 0.511 ± 0.35 aternity P (Gränby). Hence, the mean P2 for both populations was Average (±SD) read depth at the locations of SNPs 0.47, and excluding the four males with 0% paternity (averaged over individuals) called by STACKS was resulted in a P2 of 0.60. 10.2 ± 1.5 (range = 5.5–11.2 reads; Gnista) and 8.4 ± 1.2 We also made paternity assignments using (range = 6.6–18.8 reads; Gränby). Due to missing SNPs that were called from sites with higher read data, the mean (±SD) number of SNPs available per depth (8–20), as this may allow better calling of individual genotype for paternity assignment was heterozygotes. Calling SNPs at a higher read depth 61.8 ± 12.0 (range = 38–114; Gnista) and 77.2 ± 19.0 yielded 105 variable sites in Gnista (allowing 50% (range = 51–154; Gränby). Genotype error rate was low, missing data) and 248 SNPs in Gränby (allowing 40% with 2.4% (5/210) and 0.7% (1/139) loci showing one or missing data). This procedure altered the estimated more mismatches in the comparisons of maternal and values of P2, but the estimates of P2 were nonetheless offspring genotypes from Gränby and Gnista samples variable (between 0 and 1) and rarely close to 0.95 respectively. Hence, the respective estimated error or above (Supporting Information Table S1). While rate across loci calculated by CERVUS was 0.095% there was a significant positive correlation (r = 0.699, and 0.014%. n = 19, P < 0.05) between the estimates of P2 based on The power of these panels of SNPs to differentiate the paternity assignments made using COLONY and among the candidate fathers was good. The candidate CERVUS, the paternity assignment from CERVUS father assigned by COLONY to the sample of offspring almost always supported the hypothesis that some within a clutch almost always reflected the known last- father other than the candidate male (i.e. the individual male to copulate or a reconstructed genotype and only caught in tandem) had sired offspring within a clutch. Hence, there is comparatively little evidence for high last male sperm precedence using CERVUS to assign fathers to offspring within clutches, where the average

(±SD) values of P2 were 0.154 ± 0.15 in Gnista and 0.385 ± 0.32 in Gränby (Table S2). A corollary of these estimated, often low and variable,

values of P2 is an apparent high incidence of multiple paternity in the clutches of eggs laid by L. sponsa, with an estimated average of two fathers siring offspring in the sampled clutches, and only four of 19 broods sired by a single father (Table 1).

Correlation between paternity and genital morphology Figure 1. Ventral view of the male genital (aedeagus) We were able to dissect the genitals from 16 of the males of the emerald damselfly Lestes sponsa showing the nine used for paternity analyses. Hence the estimate of the landmarks used to capture the morphology. Note the small relationship between P2 and aedeagus morphology spines (black dots) within the area of the six landmarks to was based on these individuals. The multivariate the left. regression of morphology on P2 was marginally

© 2020 The Linnean Society of London, Biological Journal of the Linnean Society, 2020, 130, 497–506 SPERM PRECEDENCE AND GENITAL MORPHOLOGY IN A DAMSELFLY 501 non-significant (Generalized Goodall F-test: P = 0.069, belief that odonates are characterized by high last F = 1.65, d.f. = 14,196). Visualization of deformation male sperm precedence. By genotyping the offspring grids (i.e. shape of the aedeagus) shows that with a P2 of mating pairs of the emerald damselfly Lestes of 0 the aedeagus has a narrow tip while if P2 is 1.0 the sponsa we found that P2 was variable and often low. aedeagus has a much broader tip (Fig. 2). There was no In addition, our results suggest that variation in male significant relationship between male body size (total genital morphology is associated with variation in P2. 2 length) and P2 (P = 0.22, r = 0.11). High (>0.95) values of P2 have been reported for nine

species of odonate (reviewed by Cordoba-Aguilar et al., Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 2003) and last male sperm precedence is considered to be typical in odonates (Corbet 1999, 2004: 521), DISCUSSION except when there has been a sufficient period of time Odonates have diverse mating strategies and provide for mixing of sperm stored in the bursa copulatrix an excellent model to study components of both pre- (Siva-Jothy & Tsubaki, 1989; Cordero & Miller, 1992; and post-copulatory intrasexual competition and mate Sawada, 1998). Indeed, the frequent occurrence of choice. However, few studies have used molecular half siblings of the damselfly Coenagrion puella at genetic markers to examine the paternity of odonate a breeding site was hypothesized to be the result of clutches, potentially as a corollary of the persistent females mating (with a different male) and oviposition

Table 1. Paternity assignment (using COLONY v.2.0.6.5) in 19 families of the emerald damselfly Lestes sponsa from two sites in Europe. Noff, number of offspring that were genotyped using ddRAD; NLM, number of times that the copulating

(last) male was assigned as a parent of an offspring in the clutch; P2, proportion of clutch sired by the last male; Nfath, estimated number of fathers for the clutch.

Gnista Gränby

Family Noff NLM P2 Nfath Noff NLM P2 Nfath

1 15 0 0.000 1 15 3 0.200 2 2 15 15 1.000 2 15 10 0.667 2 3 15 5 0.333 2 15 5 0.333 3 4 15 5 0.333 3 15 11 0.733 2 5 15 0 0.000 3 15 6 0.400 2 6 15 9 0.600 2 15 6 0.400 2 7 15 13 0.867 2 15 0 0.000 2 8 15 10 0.667 2 15 15 1.000 1 9 14 0 0.000 3 15 15 1.000 1 10 15 8 0.533 2 Mean 0.433 2.2 0.511 1.9 SD 0.363 0.8 0.356 0.6 Max. 1.000 3 1.000 3 Min. 0.000 1 0.000 1

Figure 2. Deformation grids showing the difference in morphology between morphology of genitals (aedeagus) from males with a P2 of 0.0, 0.5 and 1.0. Note that the deformation grid for P2 = 0.5 shows the consensus morphology not the actual morphology. The dots show the landmarks, and the tip of the aedeagus is to the left (see Fig. 1).

© 2020 The Linnean Society of London, Biological Journal of the Linnean Society, 2020, 130, 497–506 502 F. JOHANSSON ET AL. on different days (Lowe et al., 2009). However, our data and genital morphology if handling disturbance had indicate that female L. sponsa can, and often do, use a major impact on patterns of paternity. The females sperm from several males when ovipositing the clutch used for the collection of eggs/offspring were a mixture that is laid. This behaviour is comparable with data of mating and tandem guarding pairs, but we had no on paternities of broods of the damselfly Ischnura information on how long the pairs had been in the actual elegans where values of P2 for females immediately mating act or mate guarding (i.e. tandem). However, (0 days) after mating varied from 0.59 to 1.00 (Cooper all tandem and mating pairs had been in tandem or

et al., 1996). Moreover, considerable among-individual copula for at least 1 min, and we never caught any Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 variation in last male sperm precedence has been tandem pairs where we saw the actual formation of reported for Enallagma hageni (Fincke, 1984) and for the pair. We hence assume that we did not collect any Leucorrhinia intacta when males had short (<240 s) precopula tandem pairs. To explore this further, we copulation durations (Wolf et al., 1989). also calculated the mean P2 when excluding the four Some studies on odonates have implicated the use males with 0% paternity. This provides an estimate for of spermatheca for storing sperm from several males P2 in cases where insemination was not completed by

(e.g. Siva-Jothy & Hooper, 1995), potentially to allow the last males. This new P2 value was 0.60, and thus long-term use by females (Nakahara & Tsubaki, still quite low. 2007), whereas the bursa copulatrix is thought to be It would be interesting in future studies to compare packed with sperm from the last male to mate and P2 in females that are in tandems during oviposition used to fertilize eggs that are oviposited immediately (and then collect the eggs laid) with females that are after copulation (Cordero & Miller, 1992; Nakahara not in tandem during oviposition (and then collect the & Tsubaki, 2007). Females of many species within eggs). Females ovipositing without a guarding male the genus Lestes have a rudimentary (or even lack) could potentially penalize the male that deserted spermatheca (Cordoba-Aguilar et al., 2003) and sperm the female before egg laying (Eberhard, 1994). An from all matings are thus stored in the bursa copulatrix. additional factor that could have contributed to the

In addition, in species of Lestes, males typically lack low P2 in our study is that the lag of time between specialized appendages to remove sperm (Cordoba- copulation and egg laying that occurred in the

Aguilar et al., 2003). Species of Lestes have been laboratory could reduce P2 by biasing the fertilization estimated to remove between 50% (L. vigilax, Waage, of the eggs towards the stored sperm. 1982; L. barabarus, Cordoba-Aguilar et al., 2003) and Another experimental feature of our data is the 100% (L. viridis, Cordoba-Aguilar et al., 2003) of sperm use of genetic markers to determine paternity. Most from the bursa copulatrix. Because of the rudimentary studies (n = 8) of P2 in odonates have quantified spermatheca and the lack of specialized appendages paternity using a sterile (irradiated) male technique, to remove sperm, it is possible that male L. sponsa in contrast to the few studies that have used molecular cannot remove all of a previous male’s sperm during markers such as RAPDs (n = 2) or microsatellites copulation and that the rudimentary bursa copulatrix (n = 1) (Cordoba-Aguilar et al., 2003: table 1). The use allows females to rapidly (i.e. before sperm depletion) of irradiated males involves an experimental contrast mix sperm: hence, there is variable P2. Equally, it could between just two males (one of which is irradiated) be argued that the simple female genital tract of Lestes (Fincke, 1984), whereas the use of genetic markers to should allow efficient sperm removal. Moreover, the determine paternity allows an assessment of mating fact that male L. sponsa guard females throughout patterns by wild females. In the harlequin beetle-riding oviposition (Stoks et al., 1997) could be taken as some pseudoscorpion (Cordylochernes scorpioides), a sperm indication that sperm removal in this species is (at precedence experiment (with paternity assessed using least reasonably) effective and that mate guarding is genetic markers) that allowed females to mate with needed to achieve reproductive success by stopping only two males revealed last male sperm precedence another male from removing sperm. (high P2), whereas the broods of females that had An experimental reason for low and variable last male mated with three males had mixed paternity that was sperm precedence in our study may be that individuals typical of mating behaviour by wild pseudoscorpions were disturbed during sampling, given the apparent (Zeh & Zeh, 1994). P2 is also reduced in Tenebrio negative association between copulation duration molitor and Drosophila melanogaster when females and P2 (see Siva-Jothy & Tsubaki, 1989; Wolf et al., mate more than twice (Drnevich, 2003; Laturney et al., 1989; Sawada, 1998). The probability of experimental 2018). Hence, early studies of odonate last male sperm effects (such as disturbing the male’s efficiency at precedence, which largely preceded the widespread sperm removal during mating) seems low because use of highly polymorphic genetic markers to study of the careful handling of pairs that were sampled. animal behaviour in wild populations, may have been

Also, we would not have expected to find a marginally predisposed to find high values of P2 via the ‘double non-significant relationship (P = 0.069) between P2 mating’ experimental design [but cf. Fincke (1984)

© 2020 The Linnean Society of London, Biological Journal of the Linnean Society, 2020, 130, 497–506 SPERM PRECEDENCE AND GENITAL MORPHOLOGY IN A DAMSELFLY 503 who found the last male sired between 53% and behaviour by an odonate in nature, rather than an 95% of broods using irradiated males]. In the wild, experimental bias. female odonates often mate multiply (e.g. Fincke, Evidence for low P2 raises the following question: how 1984; Stoks et al., 1997; Lowe et al., 2009; Thompson many fathers sired the clutch? It is difficult to assess et al., 2011) and the double mating design is unlikely the number of fathers as the specific genotyping error to recapitulate natural mating behaviour. Future rate is unknown, but the estimated error rates will studies that use genetic markers to quantify sperm have a notable effect on the number of putative fathers

precedence in odonates are required to test whether reconstructed by COLONY. Nonetheless, the tentative Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 low last male sperm precedence is commonplace in estimated average of two fathers (and between one and odonates. Moreover, use of genetic markers (also see three) per clutch is comparable to data on I. elegans, Cooper et al., 1996) to better estimate parameters where a minimum of between one and six copulations other than P2, such as the number of times a female per female were estimated (Cooper et al., 1996). has mated, and thus examine whether multiple mating Multiple paternity within an odonate female’s clutch is a function of certain behaviours (e.g. sperm removal, is compatible with odonate behaviour, where multiple guarding) might be interesting. matings at oviposition sites are observed (e.g. Stoks

Another reason for variation in P2 among the males et al., 1997; de Block & Stoks, 2005; Lowe et al., 2009; in our study, not mutually exclusive from the points Thompson et al., 2011). Studies that use behavioural discussed above, might be differences in male genital observations at oviposition sites will underestimate morphology, which has been associated with paternity the number of mates achieved by individuals given that effects in previous studies on diverse taxa (Arnqvist many odonates engage in copulation in more locations & Danielsson, 1999; Hotzy et al., 2012), including away from the oviposition site (e.g. Siva-Jothy, 1987). odonates (Cordero-Rivera, 2017). Male genital Given our evidence for variable last male sperm morphology may have been affected by selection for precedence, it is unclear whether estimates of fitness avoiding sperm competition through the removal of and/or the strength of sexual selection for odonates sperm stored from previous matings by the female. that are based on observations at an oviposition site

Higher P2 in males with a broader aedeagus apex is alone are meaningful (Khelifa 2019). Cryptic multiple indicative of some fitness consequences associated mating and low P2 can explain why estimates of male with intraspecific variation in odonate aedegus fitness based on observations of tandem pairs in the morphology. In L. sponsa the apex of the ventral field are a poor reflection of realized fitness (Thompson side of the aedeagus is covered with numerous small et al., 2011). spines [Fig. 1; see also Jödicke (1997: 21, figure 18) Given many benefits of using odonates as models for scanning electron micrographs], and we suggest of the sexual selection (Cordoba-Aguilar et al., 2003; that the small spines are used for removing sperm. Bybee et al., 2016), it is not clear why there have been Although having much larger spines, the length of so few studies of sperm competition using genetic spines affects fertilization success in the seed beetle markers in this taxon. One possible reason may be Callosobruchus maculatus (Hotzy et al., 2012), and that the development of microsatellite loci has been a larger area with spines might be more efficient at problematic for odonates, with initial attempts (e.g. sperm removal in L. sponsa. However, more detailed in I. elegans) yielding just two loci (Cooper et al., and mechanistic studies are needed to test this 1996). Subsequent studies typically isolate between hypothesis. ten and 20 microsatellite loci per species (Watts, The comparison of maternal and offspring genotypes 2009; Johansson et al., 2013; da Silva-Mendez et al., indicates that the genotyping error rate was low (~2% of 2013). However, the poor PCR-amplification success loci show mismatches between maternal and offspring of microsatellite loci among odonate species (Watts genotypes, and there was an overall estimated mean et al., 2004; see also Johansson et al., 2013) means that mistyping rate of <0.02%). Given (1) the apparently new microsatellite loci have to be developed for many low rate of mistyping, (2) that we allowed a reasonable species, incurring substantial time and cost, especially number of mismatches between the candidate father’s as microsatellite loci in odonates often suffer from null genotype and the offspring genotypes before rejecting alleles (Watts, 2009; da Silva-Mendez et al., 2013). paternity (to be conservative about rejecting the The use of genomic techniques is needed to explore hypothesis that last male sperm precedence is high), (3) the rich behavioural ecology of odonates (Bybee et al., that paternity assignment achieves greater confidence 2017). More widely, still only a handful of studies have when there are genotypes from known mothers and (4) used next generation sequencing methods to genotype that we observed general agreement among the two individuals and estimate paternity in wild populations methods of paternity assignment, we are confident (reviewed by Flanagan & Jones, 2019). Here, we show that the low and variable values of P2 reflect mating that ddRADseq provides a rapid and cost-effective

© 2020 The Linnean Society of London, Biological Journal of the Linnean Society, 2020, 130, 497–506 504 F. JOHANSSON ET AL. means to generate many SNPs in a species that Bybee S, Córdoba-Aguilar AM, Duryea C, Futahashi R, otherwise lacks extensive genomic information (see Hansson B, Lorenzo-Carballa MO, Schilder R, Stoks R, also Boyle et al., 2018), to offer exciting prospects to Suvorov A, Svensson EI, Swaegers J, Takahashi Y, study behaviours associated with sexual selection. Watts PC Wellenreuther M. 2016. Odonata (dragonflies In summary, our data, together with studies by and damselflies) as a bridge between ecology and evolutionary Fincke (1984), Wolf et al. (1989) and Cooper et al. genomics. Frontiers in Zoology 13: 46. (1996), indicate that last male sperm precedence (P ) is Catchen J, Hohenlohe P, Bassham S, Amores A, 2 Cresko WA. 2013. Stacks: an analysis tool set for population

not an inherent feature of odonates. This evidence for Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 low P and multiple mating by females implies that the genomics. Molecular Ecology 22: 3124–3140. 2 Catchen J, Hohenlohe P, Cresko W, Amores A, opportunity for post-copulatory sexual selection, and in Postlethwait J. 2011. Stacks: building and genotyping loci particular cryptic female choice, in odonates is large. de novo from short-read sequences. G3: Genes, Genomes, Accordingly, there are diverse opportunities for future Genetics 1: 171–182. work into processes that drive intra- and interspecific Cooper G, Miller PL, Holland PWH. 1996. Molecular genetic variation in P 2. For example, are high or low values of analysis of sperm competition in the damselfly Ischnura P2 typical of odonates, and does P2 have a phylogenetic elegans (Vander Linden). Proceedings of the Royal Society pattern, or is sperm precedence associated with London B 263: 1343–1349. variation in mating behaviour or morphology of genitals Corbet P. 1999. Dragonflies: behavior and ecology of Odonata. as we found some evidence for in this study? How Colchester: Harley Books. important is cryptic mating in terms of male success? Corbet P. 2004. Dragonflies: behavior and ecology of Odonata, 2nd edn. Colchester: Harley Books. Cordero A, Miller PL. 1992. Sperm transfer, displacement and precedence in Ischnura graellsii (Odonata: Coenagrionidae). ACKNOWLEDGEMENTS Behavioral Ecology and Sociobiology 30: 261–267 F.J. was funded by the Swedish Research Council; Córdoba-Aguilar A, Cordero-Rivera A. 2008. Cryptic S.S. was funded by the Institute of Nature female choice and sexual conflict. In: Córdoba-Aguilar A, ed. Conservation, Polish Academy of Sciences; and Dragonflies and damselflies. Model organisms for ecological P.C.W. was funded by the Finnish Academy (287153, and evolutionary research. Oxford: Oxford University Press, 324602) and is especially grateful for access to the 189–202. computing facilities provided by CSC Finland (https:// Cordoba-Aguilar A, Uhıa E, Cordero Rivera A. 2003. www.csc.fi/). We thank two anonymous reviewers for Sperm competition in Odonata (Insecta): the evolution very helpful comments. of female sperm storage and rivals’ sperm displacement. Journal of Zoology, London 261: 381–398. Cordero-Rivera A. 2017. Sexual conflict and the evolution of REFERENCES genitalia: male damselflies remove more sperm when mating with a heterospecific female. Scientific Reports 7: 7844. Alcock J. 1979. Multiple mating in Calopteryx maculata Cordero-Rivera A, Córdoba-Aguilar A. 2010. Selective (Odonata: Calopterygidae) and the advantage of non-contact forces propelling genitalic evolution in Odonata. In: guarding by males. Journal of Natural History 13: 439–446. Leonard J, Córdoba-Aguilar A, eds. The evolution of primary Alcock J. 1994. Postinsemination associations between males sexual characters in animals. Oxford: Oxford University and females in insects: the mateguarding hypothesis. Annual Press, 332–352. Review of Entomology 39: 1–21. Danecek P, Auton A, Abecasis G, Albers CA, Banks E, Andrés JA, Cordero-Rivera A. 2000. Copulation duration DePristo MA, Handsaker R, Lunter G, Marth G, and fertilization success in a damselfly: an example of cryptic Sherry ST, McVean G, Durbin R; 1000 Genomes female choice? Animal Behaviour 59: 695–703. Project Analysis Group. 2011. The variant call format and Arnqvist G, Danielsson I. 1999. Copulatory behavior, genital VCFtools, Bioinformatics 27: 2156–2158. morphology, and male fertilization success in water striders. De Block M, Stoks R. 2005. Fitness effects from egg to Evolution 53: 147–156. reproduction: bridging the life history transition. Ecology 86: Boudot J-P, Kalkman VJ, eds. 2015. Atlas of the European 185–197. dragonflies and damselflies. The Netherlands: KNNV Drnevich JM. 2003. Number of mating males and mating Publishing. interval affect last-male sperm precedence in Tenebrio Boyle JH, Martins DJ, Pelaez J, Musili PM, Kibet S, molitor L. Animal Behaviour 66: 349–357. Ndung’u SK, Kenfack D, Pierce NE. 2018. Polygyny does Eberhard WG. 1994. Evidence for widespread courtship not explain the superior competitive ability of dominant during copulation in 131 species of insects and spiders, ant associates in the African ant‐plant, Acacia (Vachellia) and implications for cryptic female choice. Evolution 48: drepanolobium. Ecology and Evolution 8: 1441–1450. 711–733.

Fincke OM. 1984. Sperm competition in the damselfly Svensson O, Edwards S. 2019. Endless forms of sexual Enallagma hageni Walsh (Odonata: Coenagrionidae): selection. PeerJ 7: e7988. benefits of multiple mating to males and females. Behavioral Lowe CD, Harvey IF, Watts PC, Thompson DJ. 2009. Ecology and Sociobiology 14: 235–240. Reproductive timing and patterns of development for Firman RC, Gasparini C, Manier MK, Pizzari T. 2017. the damselfly Coenagrion puella in the field. Ecology 90: Postmating female control: 20 years of cryptic female choice. 2202–2212. Trends in Ecology and Evolution 32: 368–382. Marshall TC, Slate J, Kruuk LEB, Pemberton JM. 1998. Flanagan SP, Jones AG. 2019 The future of parentage Statistical confidence for likelihood-based paternity inference analysis: from microsatellites to SNPs and beyond. Molecular in natural populations. Molecular Ecology 7: 639–655. Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 Ecology 28: 544–567. Miller PL. 1991. The structure and function of the genitalia Gwynne DT. 1984. Male mating effort, confidence of paternity, in the Libellulidae (Odonata). Zoological Journal of the and insect sperm competition. In: Smith RL ed. Sperm Linnean Society 102: 43–73. competition and the evolution of animal mating systems. New Nakahara M, Tsubaki Y. 2007. Function of multiple sperm- York: Academic Press, 117–149. storage organs in female damselflies (Ischnura senegalensis): Harvey IF, Parker GF. 2000. ‘Sloppy’ sperm mixing and difference in amount of ejaculate stored, sperm loss, and

intraspecific variation in sperm precedence (P2) patterns. priority in fertilization. Journal of Insect Physiology 53: Proceedings of the Royal Society London B 267: 2537–2542 1046–1054. Hosken D, House CM. 2011. Sexual selection. Current Parker GA. 1970. Sperm competition and its evolutionary Biology 21: R62–R66. consequences in the insects. Biological Review 45: 525–567. Hotzy C, Polak M, Rönn JL, Arnqvist G. 2012. Phenotypic Paxton RJ, Thorén PA, Tengö J, Estoup A, Pamilo P. 1996. engineering unveils the function of genital morphology. Mating structure and nestmate relatedness in a communal Current Biology 22: 2258–2261. bee, Andrena jacobi (Hymenoptera, Andrenidae), using Jödicke R. 1997. Die Binsenjungfern und Winterlibellen microsatellites. Molecular Ecology 5: 511–519. Europas. Magdeburg: Westarp Wissenschafte. Peterson BK, Weber JN, Kay EH, Fisher HS, Hoekstra HE. Johansson F, Halvarsson P, Mikolajewski DJ, Höglund J. 2017. Double digest RADseq: an inexpensive method for de 2017. Genetic differentiation in the boreal dragonfly novo SNP discovery and genotyping in model and non-model Leucorrhinia dubia in the Palearctic region. Biological species. PLoS ONE 7: e37135. Journal of the Linnean Society 121: 294–304. Rohlf FJ. 2003. tpsRegr, shape regression, version 1.28. Johansson F, Berger D, Höglund J, Meyer-Lucht Y, Stony Brook: Department of Ecology and Evolution, State Rödin-Mörch P, Sniegula S, Watts PC. 2020. High University of New York at Stony Brook. variation in last male sperm precedence and genital Rohlf FJ. 2017. tpsDig, tpsDig2 version 2.31. Stony Brook: morphology in the emerald damselfly Lestes sponsa. Dryad Department of Ecology and Evolution, State University of Dataset. doi.org/10.5061/dryad.s4mw6m945. New York at Stony Brook. Johansson H, Stoks R, Nilsson-Örtman V, Rohlf FJ, Slice DE. 1990. Extensions of the Procrustes Ingvarsson PK, Johansson F. 2013. Large-scale patterns method for the optimal superimposition of landmarks. in genetic variation, gene flow and differentiation in five Systematic Zoology 39: 40–59. species of European Coenagrionid damselfly provide mixed Sawada K. 1998. Sperm precedence in the damselfly Ischnura support for the central-marginal hypothesis. Ecography 36: senegalensis (Rambur): is prolonged copulation advantageous 744–755. to spermprecedence (Zygoptera: Coenagrionidae)? Jones AG, Small CM, Paczolt KA, Ratterman NL. 2009. A Odonatologica 27: 425–431 practical guide to methods of parentage analysis. Molecular da Silva-Mendez G, Lorenzo-Carballa MO, Ecology Resources 10: 6–30. Cordero Rivera A, Watts PC 2013. Microsatellite loci for Jones OR, Wang J. 2010. COLONY: a program for parentage two threatened dragonfly (Odonata: Anisoptera) species: and sibship inference from multilocus genotype data. Oxygastra curtisii (Dale, 1834) and Macromia splendens Molecular Ecology Resources 10: 551–555. (Pictet, 1843). Conservation Genetics Resources 5: 1171–1174. Kalinowski ST, Taper ML, Marshall TC. 2007. Revising how Simmons LW. 2013. Sexual selection and genital evolution. the computer program CERVUS accommodates genotyping Austral Entomology 53: 1–17. error increases success in paternity assignment. Molecular Siva-Jothy MT. 1987. Variation in copulation duration and the Ecology 16: 1099–1106. resultant degree of sperm removal in Orthetrum cancellatum Khelifa R. 2019. Females “assist” sneaker males to dupe (L.) (Libellulidae: Odonata). Behavioural Ecology and dominant males in a rare endemic damselfly: sexual conflict Sociobiology 20: 147–151. at its finest. Ecology 100: e02811. Siva-Jothy MT, Hooper RE. 1995. The disposition Laturney M, van Eijk R, Billeter JC. 2018. Last male and genetic diversity of stored sperm in females of the sperm precedence is modulated by female remating rate in damselfly Calopteryx splendens xanthostoma (Charpentier). Drosophila melanogaster. Evolution Letters 2–3: 180–189. Proceedings of the Royal Society London B 259: 313–318. Lindsay WR, Andersson S, Bererhi B, Höglund J, Johnsen Siva-Jothy MT, Tsubaki T. 1989. Variation in copulation A, Kvarnemo C, Leder EH, Lifjeld JT, Ninnes C, Olsson duration in Mnais pruinosa pruinosa Selys (Odonata: M, Parker GA, Pizzari T, Qvarnström A, Safran RJ, Calopterygidae) 1. Alternative mate-securing tactics and

sperm precedence. Behavioural Ecology and Sociobiology 24: Wang J. 2012. Computationally efficient sibship and parentage 39–45. assignment from multilocus marker data. Genetics 191: 183–194. Stoks R. 2001. Components of lifetime mating success Wang JL. 2004. Sibship reconstruction from genetic data with and body size in males of a scrambling damselfly. Animal typing errors. Genetics 166: 1963–1979. Behaviour 59: 339–348. Watts PC. 2009. Characteristics of microsatellite loci Stoks R, de Bruyn L, Matthysen E 1997 The adaptiveness in Odonata. International Journal of Odonatology 12: of intense contact mate guarding by males of the Emerald 275–286. Damselfly, Lestes sponsa (Odonata, Lestidae): the male’s Watts PC, Wu JH, Westgarth C, Thompson DJ, Kemp SJ perspective. Journal of lnsect Behavior 10: 289–298. 2004. A panel of microsatellite loci for the Southern Damselfly, Downloaded from https://academic.oup.com/biolinnean/article/130/3/497/5843780 by Beurlingbiblioteket user on 16 October 2020 Thompson DJ, Hassall C, Lowe CD, Watts PC. 2011. Coenagrion mercuriale (Odonata: Coenagrionidae). Field estimates of reproductive success in a model insect: Conservation Genetics 5: 117–119. behavioural surrogates are poor predictors of fitness. Ecology Wigby S, Chapman T. 2004. Sperm competition. Cell 14: Letters 14: 905–913. PR100–PR102. Waage JK. 1979. Dual function of the damselfly penis: sperm Wolf LL, Waltz EC, Wakeley K, Klockowski D. 1989. removal and transfer. Science 203: 916–918. Copulation duration and sperm competition in white-faced Waage JK. 1982. Sperm displacement by male Lestes vigilax dragonflies (Leucorrhinia intacta, Odonata: Libellulidae). Hagen (Zygoptera: estidae). Odonatologica 11: 201–209. Behavioral Ecology and Sociobiology 24: 63–68. Waage JK. 1983. Sexual selection, ESS theory and insect Zeh JA, Zeh DW. 1994. Last-male sperm precedence breaks behaviour: some examples from damselflies (Odonata). down when females mate with three males. Proceeding of the Florida Entomologist 66: 19–31. Royal Society London B 257: 287–292.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s website. Table S1. Paternity assignment (using COLONY v.2.0.6.5) in 19 families of the emerald damselfly Lestes sponsa from two sites in Europe. Table S2. Paternity assignment (using CERVUS v.3.0) in 19 families of the emerald damselfly Lestes sponsa from two sites in Europe.

SHARED DATA The data from the study are available from the Dryad Digital Repository (Johansson et al., 2020).