DISCUSSION
The detailed observations of female responses during male-takeovers across fifteen years suggest that living in a multi-level society means that females can mate with extra-unit males, form joint defense with other females in the same OMU against male attacks, and are able to successfully transfer to other OMUs to remove their unweaned offspring from infanticidal males. These counter-strategies appear effective, as in response most males would tolerate the dependent infant and/or stop attacking any further infants when some mothers left with infants, and thus only a small number of infants present during take-overs were killed by males (4/31). Males in turn gain from changing their own strategy, as tolerating unweaned infants means their mothers stay with him, and tolerant males would have a more chance of siring offspring with them in the future with longer tenure. These findings suggest that social constraints might be important in shifting the balance of sexual conflict either towards females or towards males.
In response to the threat of infanticide by males, females have been argued to exhibit the Bruce effect, where they terminate pregnancies after exposure to unfamiliar males as an adaptive strategy as it is highly likely that a female’s offspring will be killed after its birth (Roberts et al. 2012). This counter-strategy has been confirmed in other wild mammals including a wild primate, the gelada (Theropithecus gelada ) with 80% of pregnancies terminated in the weeks after a dominant male is replaced. In other studies of Bruce effect (Berger 1983; Hackländer & Arnold 1999), more than 90% pregnant females induce abortions following male takeover. However, the rate of fetal death (25%) is much lower than fetal death rates reported in those studies, suggesting that fetal loss in study population is not an obligate response to male takeovers. Therefore, abortions in golden snub-nosed monkey might be the result of sexually selected feticide rather than the Bruce effect in instances where the new males are aggressive towards females and their offspring, as has been shown in savannah baboon (Zipple et al. 2017).
In golden snub-nosed monkeys, the Bruce effect has likely failed to evolve due to several reasons. First, there is no additional benefit for females to terminate pregnancy before August. The females could not be conceptive shortly after abortions because golden snub-nosed monkeys are strictly seasonal breeders with most birth occurring from March to May and most conceptions occurring between late August and early November (Xiang et al. 2017). Second, the risk of male infanticide is relatively low because females have more versatile and effective counter-strategies. Therefore, the best evolutionary strategy is not to terminate gestation because the pregnant females have many other counter-strategies to reduce the risk of male infanticide.
Female primates appear best able to defend their offspring against male infanticide if they have the flexibility to interact with a range of social partners and rely on multiple counterstrategies. Multi-male mating by females in mammals has been indicated to be one of the most effective strategies to prevent infanticide, as it obscures paternity such that males refrain from killing offspring if they might be their father (Wolff & MacDonald, 2004; Lukas & Huchard 2014). In snub-nosed monkeys, extra-OMU mating and extra-OMU paternities occur between females and males residing in other OMU (Li & Zhao 2007; Guo et al. 2010; Yang et al. 2014). Furthermore, we observed two cases where females were accepted into a unit with a resident male with whom she had previously been observed mating although there no paternities were identified; another one case was also reported in a recent study (Qi et al. 2020). In contrast, females very rarely mated with males residing in the all-male unit, even though all males taking over OMUs came from this unit because resident males from multiple OMUs collectively show intense aggression towards these outside males in golden snub-nosed monkeys (Xiang et al. 2014), presumably limiting opportunities for females to mate with them and to reduce the risk of infanticide potentially even further as it does in some polygynandrous societies.
As there is frequently pronounced dimorphism between females and males in primates, it has been suggested that mothers would be unable to prevent infanticide physically (Ebensperger 1998; Palombit 2012). In this study, we find that, while females might be unable to resist a male attack on their own, in nearly 85% of attacks females formed joint defenses that were effective in preventing males from killing their offspring in that instance. Therefore, although female joint defenses against male attacks has sometimes been considered ineffective because females might be unable to completely prevent males from killing offspring (de Waal 1997), our findings suggest that in snub-nosed monkeys it might create the crucial delay for females to exhibit other counterstrategies, in particular keeping their offspring alive until they might have the opportunities to transfer from their unit. This effect was more pronounced in larger units, potentially because more adult females in an OMU can lead to stronger female-female coalitions, vigilance and risk detection might also be more efficient (Palombit 2012).
By transferring from their social unit mothers can remove their unweaned offspring from a potential killer male (Hrdy 1997). Establishing a new unit with the ousted male, as was observed in four cases here, is presumably more likely to occur in species in which the habitat is not saturated with groups that defend territories. Alternatively, females can attempt to transfer into an already established social unit where a male would be unlikely to kill the offspring (Steenbeek et al. 1997). In multi-level societies such as in golden snub-nosed monkeys, the potential costs of transferring into a new unit are likely to be reduced compared to other societies. Some females already disperse as sub-adults, and both natal and secondary dispersal of females mostly occurs into other social units within their natal band (Guo et al. 2015). This means that females frequently have female relatives residing in neighbor OMUs, which might help them to join and be tolerated in these units. Some females have sexual experienced with other resident male int the band, this might also offer her and her unweaned infant a refuge when there is high risk of infanticide. In addition, food competition among the band member in this species is low because ubiquitous lichens and leaves form a major dietary constituent throughout the year (Li 2006), facilitating units to be in close proximity to potentially reduce predatory attacks through travelling, feeding, and resting within the same areas (Grueter et al. 2020).
The relative low costs associated with female transfers appear to influence male behavior, too. From the perspective of males taking over a social unit, emigration of adult females means that they would lose reproductive opportunities completely. While infanticide might increase current reproductive success if its mother remains in the OMU, it appears that on average males are likely to gain higher fitness if they refrain from attacking the unweaned infant. It is in the case in the study band when they do not kill their current offspring because male tenures are significant longer than the female reproductive cycle, tolerant males even have longer tenure than those aggressive males, and males are likely to sire future offspring of females as the low probability of extra-unit paternity. Similarly, a resident male of an OMU might not only accept an immigrating lactating female because he previously mated with her, but also because he can increase his future reproductive success if he remains dominant for at least another reproductive cycle of that female.
Additionally, we confirm that females snub-nosed monkeys are able to mount versatile and effective counterstrategies against male infanticide including extra-unit mating, joint defense and transfer, leading to no significant difference in the mortality of infant present during male replacement to the base-line mortality of infants in stable groups (Fig 2). Furthermore, mortality from infanticide for those infants involved male replacement in the study population was also much lower than the mortality observed in some Asian colobines in which females live in single-male or multi-male societies. For example, in an uni-male society of white-headed langurs (Trachypithecus leucocephalus ), all eleven infants younger than 6 months died of possible infanticide after male takeovers (Yin et al. 2013). In uni-male and multi-male society of hanuman langurs (Winkler et al. 1984; Sommer 1987; Borries 1997), death rates of infant following male replacement were 37.9% (11/29), and 21.6% (8/37) which also higher than golden snub-nosed monkeys (12.9%, 4/31). These data might implicate that social organization or mating system would have an influence on the efficacy of female counterstrategies against infanticide.
Finally, we want provide a tentative explanation for why older males are not likely to kill the dependent infant although it would obviously increase their fitness, they might have not enough life span to wait another reproductive cycle of lactating female as those younger males. We argued that it might be explained by the male’s personality which often termed as consistent individual differences in behavior across time and contexts. Based on the insight that the trade-off between current and future reproduction, animal personalities could be given an adaptive explain in polymorphic population (Wolf et al. 2007). In snub-nosed golden monkeys of a multilevel society, where multiple OMUs coordinate and forage together as one larger social band (Grueter et al. 2020), social complexity would allow polymorphic personality of resident male in the band. In this study, we have observed two kinds of male personalities towards unweaned infants following male replacement. An aggressive male would obviously obtain immediate reproductive success by infanticide, and this personality may not only benefit to the females in his OMU when competed with patch resources (i.e., one tree with more fruit or better for sleeping) but also benefit to the population when confronted a predator because more aggression to conspecific also mean bolder in interspecific fight to a predator in many species (Sih et al. 2004). A tolerant male would obviously obtain future reproductive success through longer tenures (Fig 4b) and monopolize paternity by attracting females’ residency. Therefore, aggressive males are favored by natural selection, and tolerance males might favor by sexual selection in this multilevel society. In fact, we have observed an extreme example which show the male’s personality might be stable. For instance, in case #8 and #17, the male (NN) continue attacking the dependent infant which result two infants were killed (Case #8) or all two lactating females emigration (Case #17). In the first case, the male obtained current and future reproductive success, however, he lost all lactating females in later case.