Mate with the young, kill the old: reversed sexual cannibalism and male mate choice in the spider Micaria sociabilis (Araneae: Gnaphosidae)
- First Online:
- Cite this article as:
- Sentenská, L. & Pekár, S. Behav Ecol Sociobiol (2013) 67: 1131. doi:10.1007/s00265-013-1538-1
- 524 Views
Female mate choice is regarded as a strong selective force that significantly affects male mating success. In extreme cases, mate rejection can result in sexual cannibalism. However, males may choose between their partners as well. The killing of potential female mates, i.e. reversed form of sexual cannibalism, may be related to male mate choice. We examined male mate choice in the spider Micaria sociabilis, focusing on the roles of female mating status (virgin/mated), size and age. Reversed cannibalism reached its highest frequency in the period of generation overlap, i.e. when young males from the summer generation met old(er) females from the spring generation. These results suggest discrimination against old(er) females. The frequency of cannibalism was not affected by female mating status or female size. However, larger males from the summer generation were more cannibalistic than smaller males from the spring generation. We conclude that reversed sexual cannibalism might be an adaptive mate choice mechanism and can be explained in the context of the aggressive spillover hypothesis.
KeywordsCannibalismReversedFemale ageSelective males
Sexual cannibalism occurring before, during or after mating is considered to be an extreme case of sexual conflict resulting from different interests of males and females (Schneider and Lubin 1998). Several hypotheses have been proposed to explain the evolution and persistence of this behaviour.
In many species of spiders, sexual cannibalism occurs before the male has managed to transfer sperm; this has very different implications for males and females (Elgar 1992; Wilder et al. 2009). Newman and Elgar (1991) assumed that premating sexual cannibalism is the result of female foraging strategy. Their “economic” model predicts that the frequency of cannibalism is related to the availability of prey and potential mates. In situations where prey is scarce but the opposite sex is abundant, males might be more valuable as prey than as sexual partners. On the other hand, sexual cannibalism represents a highly disadvantageous strategy when males are scarce (Newman and Elgar 1991). The aggressive spillover hypothesis (Arnqvist and Henriksson 1997) suggests that the result of an encounter with a potentially cannibalistic female depends on her individual aggression level as a personality trait. Exhibiting a high level of aggression is an advantageous strategy during juvenile stages if food is scarce because the aggressive juveniles attack and consume more prey than less aggressive ones and therefore obtain larger body sizes (Johnson 2001; Johnson and Sih 2005). However, high aggression favoured in juvenile females may spill over into the adult stage and lead to sperm limitation due to lack of copulation (Arnqvist and Henriksson 1997, but see Elgar and Schneider 2004).
Sexual cannibalism may also be a mechanism of female mate choice (Elgar and Nash 1988) whereby high-quality males are accepted as sexual partners while low quality ones are rejected and cannibalized. The male can express his quality during courtship (Schneider and Lubin 1998), or the female can evaluate it directly from male physical characteristics, such as size (Elgar and Nash 1988; Prenter et al. 2006), age (Elgar 1992; Morse and Hu 2004) or secondary sexual structures (e.g. Persons and Uetz 2005) and/or from male personality traits (e.g. aggressiveness; Kralj-Fišer et al. 2012). However, it is often not clear whether females choose actively (direct mate choice) or whether the determination of certain males as prey is a simple by-product of differences between the physical attributes of males and females (indirect mate choice). For example, bigger males are not necessarily preferred by females directly; they are simply better able to resist her attacks than smaller ones (Prenter et al. 2006; Roggenbuck et al. 2011).
The hypotheses mentioned above were developed for the situation in which the male becomes a victim of the female. However, in some species both sexes face the danger of being killed by their sexual partner, a situation that has been observed in spiders (Jackson 1982; Jackson and Pollard 1990; Schütz and Taborsky 2005; Cross et al. 2007, 2008) and crustaceans (Tsai and Dai 2003). A few studies have reported mating systems in which only the males are cannibalistic. This reversed form of sexual cannibalism has been observed in the spider Allocosa brasiliensis Petrunkevitch (Aisenberg et al. 2011) and in two species of crustaceans (Dick 1995; Haddon 1995). Here sexual cannibalism is restricted to time before copulation (but see Haddon 1995).
The hypotheses proposed for classical precopulatory sexual cannibalism could also be applied to the reversed form, yet evidence for this reversed form is still rare. Male mate choice is usually found in systems where males contribute more than sperm to their offspring—sometimes more than females themselves. In such systems, the reversal of sexual roles is expected. Male choosiness may lead to consumption of the potential female mate in order to minimize unnecessary reproductive costs (Bonduriansky 2001). However, male mate choice (and potentially also a reversal of sexual cannibalism) can be found even in systems where males provide only sperm. Male choosiness can be the result of the female-biased operational sex ratio (i.e. when males are rare, females are under stronger selection; Kvarnemo and Ahnesjö 1996) and/or intensive male–male competition (Bel-Venner et al. 2008), or it can simply be explained as a response to high variability in female quality (Bonduriansky 2001; Gwynne 1991).
Males can choose their mates using characteristics similar to those used generally by females. In arthropods, female size is directly linked with fecundity and therefore serves as an indirect indicator of female quality (Head 1995; Bonduriansky 2001). Female quality may also vary with age, as younger females have a longer life expectancy and greater future reproductive potential than older ones (Rutowski 1982). Further, the presence of female sperm-storage organs (spermathecae) in combination with polyandry imposes strong sperm competition (Bonduriansky 2001). In many entelegyne spiders, internal morphology favours the sperm of the first male and, therefore, males of some species prefer mating with virgins (e.g. Austad 1982; Herberstein et al. 2002; Gaskett et al. 2004; Stoltz, et al. 2007). However, this preference for a virgin mate does not necessarily represent the general trend as morphology may differ in other species or the sperm may be mixed in the spermatheca (Elgar 1998; Eberhard 2004). For males it would be also disadvantageous if the mated female did not survive to the stage of producing egg sac and/or to offspring independence, resulting in total reproductive failure. It has been found that males try to minimize the mortality risk of the potential partner by mating with females close to oviposition (Rittschof 2011; Rittschof et al. 2012).
We have observed that males of the entelegyne spider Micaria sociabilis Kulczyński (Gnaphosidae) attack females prior to mating. This species is a tiny myrmecomorphic spider with diurnal activity and inhabits the bark of old oaks in association with its model ant, Liometopum microcephalum Panzer. These spiders live among their ant hosts, but they do not attack ants or take any food resource tended by them. Juveniles and adults (including males) capture tiny invertebrates, such as flies or springtails, living on tree surfaces or hidden in bark crevices (Pekár and Jarab 2011). Here, we describe mating and reversed sexual cannibalism in M. sociabilis. We focused on the male mate choice as an explanation for sexual cannibalism and thus sought factors explaining its occurrence. We expected males to be cannibalistic towards low-quality females and therefore tested the effects of mate qualities, such as female size, female age and mating status, on the frequency of cannibalism. To accomplish these tests, we studied the phenology of the species and investigated cannibalism over three seasons to reveal whether the frequency of cannibalism varied among months. With respect to the aggressive spillover theory as an alternative explanation, we also measured male size as this characteristic could represent an indirect measure of spider aggression.
All individuals of M. sociabilis were collected using a pooter from the tree bark of dozens of trees in Lednice, Czech Republic from March to September 2008, 2009 and 2010. Captured spiders were housed individually in glass tubes (diameter 15 mm, length 60 mm) with a layer of plaster at the bottom, which was moistened with few drops of water at 2-day intervals to maintain the required humidity. Spiders were kept at room temperature (approx. 22 °C) and 40 % relative humidity (RH) and under a natural long-day regime. They were fed fruit flies (Drosophila melanogaster Meigen) and springtails (Sinella curviseta Brook) to satiation regularly at 2-day intervals. To exclude the occurrence of cannibalism due to hunger, spiders were fed on the day preceding each experiment.
All experiments were performed in the laboratory. All trials were carried out in transparent containers (diameter 35 mm, height 40 mm) in order to both maximize encounter rate and provide sufficient space for escape (spiders were able to run on the container sides and lid). Spiders were released into the container one after the other (female before male). Once the pair was released into the container the occurrence of the following behaviours was recorded: classical cannibalism (when the female kills the male), reversed cannibalism (when the male kills the female) and copulation. When neither copulation nor cannibalism had occurred after 20 min from the first contact between an adult male and an adult female, the spiders were replaced. All statistical analyses were performed in the R environment (R Development Core Team 2010).
Generation and size
The effect of generation and body size on the outcome of the interaction between male and female was evaluated from the set of trials conducted in 2008, 2009 and 2010. In these trials we used individuals captured in the field in the adult stage. The generation class (spring or summer) of males and females was attributed to collected individuals by comparing it with observed phenology (Fig. 1).
In these trials, a single young adult male was paired with a single adult female either from the same generation (young female) or from the previous generation (old female). The former treatment group included young males and females occurring in approximately the same period in the field (spring generation males were collected in April, and spring generation females were collected in April and May; summer generation males were collected in July and August, and summer generation females were collected from August onwards). The latter group represented the pairing of young males with old females; such females were collected during or after the spring mating period (in June and July). These females from the spring generation were paired with young males from the summer generation (collected in July). As juveniles and subadults hardly ever moult in the laboratory, all individuals used in these trials were collected from the field in the adult stage; consequently, their mating history was not known. The trials were performed as described above. Both sexes were used repeatedly to determine whether they are polygamous. The males (N = 121) were used between two and seven times to maximise the number of various size ratios. Male identity was included as a factor into the analyses. Males who mated two times and/or cannibalised the female once were not used in subsequent trials. Females (N = 121) were also used repeatedly until copulation or cannibalism occurred. Trials where neither cannibalism nor copulation occurred were excluded from the analyses because the proportion of cannibalism to copulation was our main interest. Moreover, cases without copulation or cannibalism could represent several different situations (male ignorance, female resistance, unsuccessful attempt to copulation/cannibalism). We also analysed whether the sizes of the male and female differ between those trials in which copulation or cannibalism occurred and trials in which none of these behaviours were observed.
Prosoma length in all adult individuals used in the trials was measured using an ocular micrometer on a stereomicroscope (model SZX 9; Olympus, Tokyo, Japan). The size of the prosoma was measured instead of total body size because it does not change with the amount of ingested food. Due to the existence of two generations in each season, both sexes were categorised according to the generations as follows: the spring generation of males was found from April to May and the summer generation from July to September; the spring generation of females was found from March to the end of July and the summer generation from August to September.
The sizes of males and females were compared between generations using GLM with Gamma error structure and log link with interaction between sex and generation due to heteroscadistic data (Pekár and Brabec 2009).
Binary data on the occurrence of cannibalism were analysed using Generalized Estimation Equations (GEE) with binomial error structure and logit link (Pekár and Brabec 2012). GEE is an extension of GLM and used to model repeated measurements by specifying the covariance structure of the residuals. The association structure was autoregressive AR(1) due to temporal pseudoreplications. The linear predictor included interaction between size (either absolute or relative) and generation. Relative size was estimated by dividing the length of the male prosoma by the length of the female prosoma.
To study the effect of generation, every male was included in the analysis only once, i.e. only the outcome of the first trial with each male was included (except for trials where neither cannibalism nor copulation occurred, which were excluded from analyses). The trials performed in 2009 were excluded from this analysis because no trials were performed in July and this data set was not complete for this season. The frequencies of cannibalism observed monthly during each season in 2008 and 2010 were compared using the Generalised Linear Model with binomial error structure and logit link (GLM-b) for each year separately (Pekár and Brabec 2009). The proportion test was used to compare the frequency of cannibalism between the two months, one with the highest observed frequency of young and the other with the highest frequency of old females.
To determine the effect of female mating status (virgin/mated) on the frequency of cannibalism, similar mating trials were carried out in August 2009 using males and females from the same generation. Virgin females (N = 17) were collected in the subadult stage and raised to adulthood under laboratory conditions as described above. Adult males (N = 14) and adult females (N = 26) were then collected in the field and mated once in the laboratory. Each male was randomly selected and consecutively paired with two virgin and two mated females in separate trials. Males were chosen to be bigger than females in all trials, with the size ratio between partners ranging from 1.05 to 1.3. Half of the males were first paired with virgin females and the other half with mated ones. After 7 days, the trials were repeated so that the first half of males was paired first with mated females and then with virgin ones. In total, four different females were provided to each male; no female was provided more than once to the same male. The data thus obtained were analysed using the McNemar test due to the paired design.
Description of mating and cannibalism
In laboratory trials, copulation occurred soon after the first contact (median 23s, Q25 = 3 s, Q75 = 124 s, N = 159). No courtship was observed during these trials. After the first contact, the male chased the female in attempts to mount her. The female usually struggled while the male was trying to reach the copulation posture. Spiders mated in type III posture (Foelix 1996) with the male being positioned dorsally by the side of the female. After the male reached this posture, the female usually stopped fighting and allowed the male to insert his palp. However, the female was often running around the Petri dish during copulation. The copulation ended when the female began to struggle vigorously and shook off the male. Males and females mated multiple times when offered more partners: on average males mated [mean ± standard deviation (SD)] 2.7 ± 2.3 times (N = 56), and females mated 2.4 ± 1.5 times (N = 46).
Cannibalism also took place early after the first contact (median 17 s, Q25 = 3 s, Q75 = 56 s, N = 49). The female was chased by the male, defending herself vigorously as observed at the beginning of the copulation. The attack always took place before the male mounted the female. Cannibalism was always female directed, and no case of classical sexual cannibalism was observed.
Frequency of mating and cannibalism during a season
In 2010, mating occurred in 52 % of trials and reversed cannibalism in 20 % (N = 92). Trials where neither cannibalism nor copulation occurred were excluded from analyses (28 %, N = 92). The frequency of reversed cannibalism also differed among months (GLM-b χ262 = 66.6, P = 0.01), with the highest frequency in July (63.6 %, N = 11). The frequency of cannibalism in April, May and August was 20.8 % (N = 24), 12.5 % (N = 24),and 42.9 % (N = 7), respectively ((Fig 2.). The frequency of cannibalism differed significantly between April and July (Proportion test χ21 = 4.3, P = 0.03).
In mating trials with virgin females (N = 28), reversed cannibalism occurred in three trials (copulation in 9 trials). In mating trials with mated females (N = 28), reversed cannibalism occurred in two trials (copulation in 6 trials). The frequency of reversed cannibalism with both virgin and mated females was not significantly different (McNemar test χ21 = 0, P = 1).
By pairing males with females of different qualities such as size, age and mating status, all being related to potential future reproductive success, we examined whether the reversed form of sexual cannibalism in M. sociabilis is an adaptive mechanism of male mate choice. If males are able to estimate female qualities related to probable reproductive future, their rejection of low-quality females via sexual cannibalism should be adaptive, hence maintained through evolution. Alternatively, male sexual cannibalism could be explained in the context of the aggressive spillover hypothesis.
In accord with our hypothesis that males will kill low-quality females, we found the highest frequency of cannibalism during the period when young males from the summer generation met old females from the previous spring generation. This increased frequency of cannibalism of old females suggests male mate choice based on female age. In general, the number of oocytes in females is limited, and after its depletion females experience reproductive senescence (Mangel and Heimpel 1998). Male mate choice based on age has been documented, for example, in the spider Lycosa tarantula Linnaeus (Moya-Laraño 2002) where the authors found that old females of this species laid smaller egg sacs and produced fewer spiderlings; thus, males preferred to mate with younger females. However, in this study the rejection of older females was not followed by cannibalism as was observed in our study. We did not examine whether the ovaries of females from the spring generation were truly depleted in July. In the laboratory, we observed that females laid more egg sacs (Pekár and Jarab 2011) shortly after mating. Nonetheless, even if the females were still potentially able to produce offspring, we can consider them to be low-quality mates in comparison with young females as age is, in general, correlated with an increase in the probability of death due to physiological changes (Rutowski 1982). Therefore, we presume that males would do better to reject old females and wait for young ones from the new generation regardless of whether the ovaries of older females are depleted or not. Cannibalism during the period when two generations of M. sociabilis overlapped seems to be an advantageous strategy. By killing low-quality females from the previous generation, the cannibalistic males do not lose a valuable mating opportunity as females from the new generation are available within a few days.
The body size difference between sexes in the summer generation is not easy to explain. However, larger male sizes in the summer generation may result from selection for a period of prey scarcity and the abundance of old low-quality females. Alternatively, the sizes of males from the summer generation could be related to maternal effects of early adult females on summer sons to sire offspring with females that can defend themselves, which would be in agreement with our results of relative body size. It is also possible that summer males are bigger and more capable of attacking the females due to inbreeding avoidance, which would confirm the higher frequency of cannibalism observed in period of generation overlap. Nevertheless, we did not know the degree of relatedness between the partners. Although even smaller males were sometimes able to kill the female, we found bigger individuals to be more successful in their attacks in the summer generation, when the frequency of cannibalism increased with increasing male size. This pattern was not found in the spring generation and may suggest that aggression is related to male size, as large males were more cannibalistic. Larger individuals also have higher energetic requirements and therefore attack more prey (Johnson 2001). Attacking the female is potentially risky, as she might kill the male while defending herself. On the other hand, in terms of body size, females may represent substantial meal for a male so there may be a conflict between costs and benefits in terms of whether to kill the female or not.
We do not know whether cannibalistic males eat only females or whether they also hunt other prey in the field. In the laboratory, cannibalistic males accepted different food, such as flies and springtails. Moreover, we have never observed males to be killed by females.
Male individual aggression also seems to play an important role, as only a portion of the males were actually cannibalistic. In spiders, which were shown to exhibit individual differences (personalities) in aggression, aggression level is often positively related to body size (Pruitt and Riechert 2009; Pruitt et al. 2011); hence, body size might be an indirect measure of spider aggressiveness level (low or high). The higher frequency of cannibalism of larger males could be explained in the context of aggressive spillover hypothesis (Arnqvist and Henriksson 1997).
Even if differences in body size played an important role in the summer generation, male mate choice based on female body size was not confirmed for any generation. Absolute female size did not affect the frequency of cannibalism, even though size is considered to be a reliable indicator of female quality (Head 1995). These results suggest that males did not actively prefer larger females over smaller ones, at least not directly. On the other hand, relative size and male absolute size were important factors in the summer generation and significantly affected the frequency of cannibalism, which may suggest that males choose their mates indirectly (Prenter et al. 2006). A similar pattern has been observed in the water spider Argyroneta aquatica Clerck, where the presence or absence of cannibalistic attack by the male was determined by relative size in the pair (Schütz and Taborsky 2005). Unfortunately, in M. sociabilis we were unable to evaluate whether the male was trying to copulate or kill the female because of the overall forced and quick nature of mating. Therefore, we could not determine whether the male tried to attack only females of a particular size or whether he attacked them indiscriminately. If only relatively larger females were able to escape or survive to copulation and if only sufficiently large males could overpower them, this would indicate indirect male mate choice through struggle (Prenter et al. 2006).
As male mate choice is expected to predominate in systems where males contribute considerably to the offspring (Bonduriansky 2001), this mechanism was found to explain reversed sexual cannibalism in the wolf spider Allocosa brasiliensis, whose males render their burrow to the female after mating (Aisenberg et al. 2009, 2011). However, in our study species, M. sociabilis, no such investment from males has ever been observed. Nevertheless, males can also be choosy if the variability in female quality is high (Bonduriansky 2001). In spiders, the quality of the female can be estimated from her physical attributes (Head 1995; Bonduriansky 2001), but her mating history is also important (Eberhard et al. 1993; Uhl et al. 2010). The male preference for virgin females is widespread among entelegyne spiders (e.g. Austad 1982; Herberstein et al. 2002; Gaskett et al. 2004; Stoltz et al. 2007, but see Elgar 1998; Eberhard 2004). Reversed sexual cannibalism in the wolf spider Allocosa brasiliensis seems to fit this pattern, as males preferentially copulate with virgin females, while mostly killing those which had already mated (Aisenberg et al. 2011). In our study on M. sociabilis, however, males did not seem to choose females according to their mating status, as the frequency of cannibalism was similar for both mated and virgin females. In our trials the frequency of mating was also very low, possibly due to a reluctance to copulate with a certain female or male due to mate choice (Bonduriansky 2001; Prenter et al. 2006).
Our study provides an insight into an unusual mating system, which differs significantly from the general model. Such a model considers males as active non-discriminatory machines trying to copulate with every female encountered. Female mate choice is considered to be a strong selection force on male behaviour and morphology (Andersson and Iwasa 1996). However, even males may choose their potential partners (Bonduriansky 2001; Huber 2005) and apparently, in some cases, they can present their choice as extremely as females do by cannibalising unpreferred mates.
To conclude, we found males of M. sociabilis to be highly cannibalistic during the period of generation overlap when young males meet old females. It is tempting to explain this behaviour by male mate choice based on female age. However, in our study cannibalism was not restricted solely to the period of overlapping generations. It would appear that also individual aggression level represented by male body size plays an important role. We also expect that a number of other factors are involved. Ecological conditions, such as prey and mate availability, may also play important roles in M. sociabilis. The effect of these factors will be presented in a forthcoming paper.
This study was supported by grant no. MUNI/A/0937/2012 provided by MU. We would like to thank to S. Toft and T. Bilde for providing the Collembola culture used to feed the species.