Below, we describe the findings of studies that tested the ASH and re-evaluate them in the light of the above criteria. Specifically, we seek evidence for (1) between-individual differences in pre-copulatory cannibalism and in voracity towards prey, (2) repeatability of these behaviours and (3) correlation between tendency to engage in pre-copulatory cannibalism and voracity towards prey. Furthermore, the ASH states that highly aggressive females exhibit larger adult body sizes and higher propensity for pre-copulatory cannibalism independent of their mating status, hunger level and mate quality (Arnqvist & Henriksson 1997). Hence, we look for (4) the correlation between female propensity for pre-copulatory cannibalism and her adult body size. To re-evaluate the support for the ASH and to allow for additional or alternative explanations, we ask whether (5) the incidence of pre-copulatory cannibalism depends on factors such as female hunger (if females reared under low food conditions, more often employ pre-copulatory cannibalism), mating status (if mated females more often employ pre-copulatory cannibalism) and/or mate quality (if females more often cannibalise less preferred male phenotypes). Finally, we ask whether (6) cannibalistic females have a reduced reproductive success as predicted by the ASH (Table 1).
Pre-Copulatory Cannibalism in Fishing Spiders (Dolomedes fimbriatus, D. triton)
The ASH by Arnqvist & Henriksson (1997) ensued from the instances of pre-copulatory cannibalism in the fishing spider, Dolomedes fimbriatus. In this species, virgin females may attack and attempt to cannibalise males during courtship and mating (Gerhardt 1926; Schmidt 1953; Arnqvist 1992). Arnqvist & Henriksson (1997) reported that the high rate of sexual cannibalism in this protandric species results in dramatic declines of the male population when females mature. In laboratory trials with unknown sample sizes, they found that in 78% of partly repeated interactions, females attacked courting males from a distance (Arnqvist & Henriksson 1997). In 11% (Arnqvist & Henriksson 1997) or 3% (Arnqvist 1992) of such attacks, males were killed, resulting in a relatively low risk of sexual cannibalism at this stage of courtship (the actual numbers of killed males are not reported but can be estimated to be a single male in Arnqvist 1992). The male risk amounts to 7% at the next pre-copulatory stage, that is, during mounting of females (Arnqvist 1992). To explain these observations, the authors tested and excluded the effects of female mating status (the number of palpal insertions a female had received), male size, and food availability on aggressive behaviour of females (Arnqvist & Henriksson 1997). However, they found that smaller males were more likely to be cannibalised after an attack, which might support the mate choice and the mate size dimorphism hypothesis (MSD) (e.g. Wilder & Rypstra 2008; Roggenbuck et al. 2011). The MSD hypothesis suggests that females are indiscriminately aggressive towards males, and that their actual killing depends on the female's physical strength relative to the male's ability to defend his life, which both depend on body size (Wilder & Rypstra 2008).
Arnqvist & Henriksson (1997) interpreted their findings on the fishing spider as follows: adult size depends on juvenile growth, and thus, juvenile food consumption should determine female fitness. Food consumption is positively related to foraging aggression, and hence, aggressive behaviour is selectively favoured during development. Aggressive juvenile females consume more prey, acquire larger adult sizes and should produce more offspring than non-aggressive juveniles. However, because of genetic constraints on aggressive behaviour, adult females exhibit high aggression and low discrimination towards prey including male conspecifics and may remain unmated; this reasoning led Arnqvist & Henriksson (1997) to develop the ASH as a conceptual model.
Unfortunately, the data on D. fimbriatus do not allow a direct estimation of the risk of remaining unmated because females are known to consume unfertilised egg batches, and hence, field samples of egg-carrying females are biased towards fertilised females (Arnqvist & Henriksson 1997). Suggestive evidence comes from 18 field-collected, egg-carrying females that were used to determine the fertilisation rate of their egg sacs (Arnqvist & Henriksson 1997). All egg sacs contained fertilised eggs, but in a third of the inspected clutches, fertilisation rates were below 50%. It is possible that the latter group of females received only one palpal insertion, plausibly due to female aggression (Arnqvist & Henriksson 1997). Hence, the reason for the low fertilisation rates is based on the authors' assumption (Arnqvist & Henriksson 1997), not data, and so the actual costs of female aggressiveness remain unknown. Furthermore, it is unknown how these fertilisation rates compare with the rates in other, less cannibalistic species.
To further complicate the outcomes of the study, only a small proportion of the experimental females laid eggs (Arnqvist & Henriksson 1997). Causes for reproductive failure are dubious but at least six females received no copulation although it is not clear whether this was due to female aggressiveness (Arnqvist & Henriksson 1997). If more than 80% remained unmated, this would contradict personality theory assuming high between-individual variability in personality traits (e.g. Koolhaas et al. 1999). However, it is possible that the experimental conditions including the males' options to escape or the observation duration of only 45 min might be at least partly responsible for the observed low mating success and/or high cannibalism frequencies. In cannibalistic species, males often approach females very cautiously, and it may take more than an hour for copulation to occur, as for example, in Dolomedes triton (Johnson & Sih 2005), in Argiope keyserlingi (Herberstein et al. 2002) and the redback spider (Stoltz & Andrade 2010; Neumann & Schneider 2011). Also, copulation latency might be longer for aggressive vs. non-aggressive females, as, for example, in Anelosimus studiosus (Pruitt & Riechert 2009a). Hence, if a realistic risk of sexual cannibalism is to be estimated, mating trials should allow sufficient time. Although males in this study commenced courtship early during trials, over 70% were attacked from a distance. Although most of them successfully evaded female attack, they may subsequently become particularly cautious and require even more time and space for another approach (Arnqvist & Henriksson 1997). Furthermore, even though the ASH predicts reduced reproductive output, Arnqvist & Henriksson (1997, p. 259) found no difference between cannibalistic and non-cannibalistic females in this respect.
Foraging success will generally result in larger size and weight, and females' larger size and weight results in higher fecundity, a correlation that has been shown in many insect and spider species (Higgins 1992; Spence et al. 1996). Arnqvist & Henriksson (1997) did not directly measure aggressiveness in foraging but predicted that fecundity selection for large female body size would also favour high aggressiveness towards prey as this will increase foraging success. Johnson & Sih (2005) directly tested aggressiveness in a foraging and mating context in a congener, the American fishing spider, D. triton. As predicted by the ASH, D. triton females exhibited consistent individual differences in aggressiveness towards prey through ontogeny, which suggests a genetic control of aggressive behaviours in the foraging context (Johnson & Sih 2005). Unfortunately, the authors did not control for female hunger – but should – to test the ASH. Voracity towards prey should generally be compared between individuals of the same state and motivation (age, sex, hunger level and mating status). Nevertheless, in accordance with the ASH predictions, aggressiveness towards prey correlated positively with feeding rate, adult size and fecundity (Johnson & Sih 2005). The authors report a positive correlation between voracity in foraging and the tendency to engage in pre-copulatory attacks. Moreover, 46% of the males that were attacked (attacks occurred in 32% of all courtship events across the whole experiment) were killed in pre-copulatory attacks, and a female's propensity to kill and consume a male was well explained by her tendency to engage in pre-copulatory attacks. Surprisingly, the same measure of aggressiveness was not correlated with the number of palpal insertions she ultimately received or to the fertilisation rate of her egg sac, suggesting aggressive females did not suffer reproductive costs (Johnson & Sih 2005). Hence, while the predictions of the spillover model hold in the system, the presumed costs of pre-copulatory sexual cannibalism to females were not present. Interestingly, consuming one or more males had no effect on female fecundity, which suggests that pre-copulatory sexual cannibalism entails neither costs nor benefits to females.
To sum up, the ASH has a good potential to explain the high frequency of pre-copulatory sexual cannibalism in fishing spiders (genus Dolomedes). However, as pointed out by Johnson & Sih (2005) and by Arnqvist & Henriksson (1997), longitudinal studies would be called for as well as heritability estimates. Furthermore, realistic estimations of the true frequencies and costs of pre-copulatory sexual cannibalism to females are highly desirable.
Pre-Copulatory Cannibalism in a Subsocial Spider (Anelosimus studiosus)
To some degree, studies of A. studiosus support the ASH as an explanation of pre-copulatory cannibalism. These subsocial spiders are polymorphic in social (aggressive) behaviours (Pruitt et al. 2008). The findings that social females are consistently less aggressive towards predators and prey and less often cannibalise males during courtship than aggressive females (7.5% vs. 30%; (Pruitt & Riechert 2009a,b; Pruitt et al. 2008, 2011) are in accordance with the ASH predictions. In line with the ASH, docile females exhibit higher reproductive output and are preferred by males over aggressive females (Pruitt & Riechert 2009a). Other results may or may not support the ASH. In the staged experimental trials, successful copulation significantly more often occurred between aggressive and docile phenotype (aggressive female–docile male; aggressive male–docile female) than between two aggressive or two docile types (Pruitt & Riechert 2009a). However, aggressive females are more likely to kill large aggressive than small docile males, which is adaptive; this pair combination has an increased reproductive success (Pruitt et al. 2011). It is possible that aggressive females use sexual cannibalism as a mate choice strategy, when they preferentially reject and devour large aggressive males. Furthermore, the latency to copulate is longer in aggressive females, which may further support the notion that aggressive females are choosier than social ones (Pruitt & Riechert 2009a). Alternatively, males may be more cautious when attempting to copulate with aggressive females. These questions remain open. In summary, sexual cannibalism in A. studiosus is not an invariant element within a broader syndrome, and hence, it cannot be considered merely a spillover of aggressive personality, but it seems to be a result of interaction between social and sexual selection (Pruitt et al. 2011).
Pre-Copulatory Cannibalism in a Funnel Spider (Agelenopsis pennsylvanica)
As in the above studies, pre-copulatory cannibalism in a funnel spider, A. pennsylvanica, could not be solely explained by the ASH (Berning et al. 2012). Pre-copulatory cannibalism occurred in 36% of all matings of virgins; however, none of the tested females killed two males in succession (Berning et al. 2012). The females that killed the first male, later copulated with the second male (Berning et al. 2012). The authors showed that pre-copulatory cannibalism in A. pennsylvanica depended on the female's general aggressiveness, measured as voracity towards prey, as well as on hunger state: the most aggressive females, and those deprived of food, were most likely to consume their mates. These results suggest that pre-copulatory cannibalism in this system can be explained with the ASH and with a foraging strategy. In contrast to the ASH predictions, however, virgin females that consumed the first potential mate prior to copulation, exhibited increased reproductive success: they gleaned more offspring from heavier egg sacs (Berning et al. 2012). This study clearly demonstrates the adaptive consequences of pre-copulatory sexual cannibalism, which can be a result of multiple mechanisms acting in concert.
Furthermore, despite females exhibiting consistent individual differences in aggression in the foraging context, the fact that females never cannibalised two males consecutively implies the lack of repeatability of pre-copulatory cannibalism. These results contradict the ASH and instead suggest that pre-copulatory cannibalism might be related to mate choice. However, pre-copulatory cannibalism was independent of male size and mass. To further investigate mate choice and foraging as explanations, we believe that further experiments should test for other male qualities such as aggressiveness and courting intensity (Pruitt & Riechert 2009a; Pruitt et al. 2011; Kralj-Fišer et al. 2012), in particular because females that rejected the first male had higher reproductive success. To conclude, pre-copulatory cannibalism in A. pennsylvanica may be explained as a foraging strategy and partly by the ASH – voraciousness levels of virgin females were positively related to incidences of pre-copulatory cannibalism in their first mating trial.