Introduction

Recognition and awareness of death was long considered unique to our own species (Anderson et al. 2016). Understanding how the other animals respond to their dead, particularly our closest relatives, can contribute to elucidating the evolutionary origins of human responses to death (Anderson et al. 2016, 2018; Gonçalves and Carvalho 2019). However, how nonhuman primates detect and react to death is still understudied; our current knowledge is largely based on a handful of (mainly) opportunistic observations. Although probably lacking explicit biological knowledge about causes of death, some species may possess a basic understanding about potentially deadly events (e.g., falls, attacks; Anderson 2019). A combination of visual and olfactory cues together with the lack of usual reactions by corpses may help animals to acquire some components of death awareness. One of our closest evolutionary relatives, the chimpanzee, shows several behaviors that recall human responses to the death of kin or a close acquaintance (e.g., spatial proximity, cleaning the corpse, mourning; Anderson et al. 2010). Reactions to a dead conspecific may change with contexts; for example chimpanzees may stay close to and touch a clearly injured or dead conspecific, whereas they are more likely to avoid “unexplained” corpses (Teleki 1973; Boesch and Boesch-Achermann 2000; Boesch 2012; Watson et al. 2015).

One basic reaction to a dead conspecific is necrophobia, the avoidance of corpses. Numerous species show aversion to fatty acid necromones (e.g., cadaverine and putrescine), chemical cues that signal the presence of a corpse (invertebrates: Haskins and Haskins 1974; Wilson et al. 1958; fish, Hussain et al. 2013; Oliveira et al. 2014, mammals: Peterson and Batholomew 1967; Prounis and Shields 2013; Wisman and Shrira 2015). Necrophobia is adaptive in terms of avoiding predators and reducing disease transmission risk, and thus in this sense, awareness of death may also be adaptive (Gonçalves and Biro 2018; Gonçalves and Carvalho 2019). In both terrestrial and aquatic social animals, however, individuals may also stay close to the corpse, touching and maintaining physical contact with it or—in the case of dead infant cetaceans and primates—carrying it until decomposition (cetaceans: Reggente et al. 2016, 2018; elephants: e.g., Moss 1988; Douglas-Hamilton et al. 2006; primates: e.g., Anderson et al. 2010, 2018; Watson and Matsuzawa 2018; Porter et al. 2019).

Together with visual and chemical cues, tactile information may help in assessing the precise state of the injured or dead individual, and even the cause of death (Harzen and dos Santos 1992). In primates, immediate responses to a dead group member may include aggressive grabbing and biting, exploratory behaviors (smelling and touching) and grooming (e.g., rhesus Macaques: Buhl et al. 2012; chimpanzees: Anderson et al. 2010; Stewart et al. 2012; eastern gorillas: Porter et al. 2019). Aggression toward a corpse or an anesthetized individual may reflect an attempt to rouse it, frustration caused by the failure of it, or even curiosity triggered by the lack of responsiveness (Anderson et al. 2010; Stewart et al. 2012; Porter et al. 2019; Douglas-Hamilton et al. 2006). Through these behaviors, conspecifics can collect information about the state of responsiveness of the target individual (Cronin et al. 2011), potentially contributing towards an understanding of irreversibility of death, a concept that arises in humans by 5–6 years of age (Anderson et al. 2018). Conceivably, such understanding is shared at least with our closest relatives, the chimpanzees (Anderson et al. 2018), who show evidence of higher cognitive and empathic abilities (Hare et al. 2001; Schmelz et al. 2011; Yamamoto et al. 2012; Devaine et al. 2017). Chimpanzees also sometimes physically interact with corpses of other species, suggesting that their discrimination between “living” and “dead” applies across species (e.g., Hirata et al. 2001; Carvalho et al. 2010; Hockings et al. 2012; Cibot et al. 2017; Boesch and Boesch 1989; Zamma 2002).

The death of an infant solicits particular and peculiar behaviors. In primates and cetaceans, post-mortem parental carrying and care may last from hours to months (Biro et al. 2010; Reggente et al. 2016, 2018; Anderson 2011; Gonçalves and Carvalho 2019). However, in monkeys and apes, contradictory behaviors, notably filial cannibalism, have also been reported (e.g., Trapanese et al. 2019). Several hypotheses have been proposed to explain these various behaviors (Watson and Matsuzawa 2018). One is that prolonged care and carrying reflect the maternal-bond strength and influence of maternal hormones (Biro et al. 2010; Li et al. 2012; Warren and Williamson 2004; Hrdy 1999). Alternatively, mothers simply might not ‘realize’ that their infant is dead (Sugiyama et al. 2009); they might believe it is simply unconscious (Hrdy 1999). It has also been suggested that prolonged carrying and care are more likely in the case of non-traumatic death (e.g., through illness) than death from injury or infanticide (Watson and Matsuzawa 2018). Another hypothesis is that extended carrying may arise through cultural transmission within families or groups (e.g., Biro et al. 2010). Finally, extremes of climate (arid, cold) might delay decomposition, making extend carrying more likely (Fashing et al. 2011).

Recently, responses toward the corpses of intra-group and extra-group individuals in eastern gorillas (Gorilla beringei) were described, including a general curiosity about death (Porter et al. 2019). The gorillas showed various affiliative and investigative behaviors toward corpses, but only silverbacks and blackbacks responded aggressively to them (Porter et al. 2019). All we know about reactions to death in western gorillas (Gorilla gorilla) comes from observations on a captive group (Hoff et al. 1998), in which increased aggression levels continued for 3 months in adult females after the death of the silverback (Hoff et al. 1998). Here, I report reactions in a wild, habituated western gorilla group to an unanimated infant after an almost-fatal accident, and to a dead red river hog. These observations can further help to shed light on perception of death in gorillas and on what underlies prolonged carrying and care of dead infants by mothers.

Methods

Observations were carried out at Bai Hokou (2º50′N, 16º28′E) in the Dzanga sector of the Dzanga-Ndoki National Park (1222 km2) in south-west Central African Republic. The Dzanga-Ndoki National Park is part of the Sangha Trinational (TNS), a transboundary World Heritage Site where the Central African Republic, the Republic of Cameroon, and the Republic of Congo meet. The climate is seasonal with a dry season of less than 100 mm monthly rainfall between December and February and a peak rainy season in September and October. The events described here are the only two death-related cases observed during the 8-month study period, which included half-day and occasional full-day observations between April 2008 and June 2008, and from November 2008 to March 2009 (NDays = 193 NFocalHours = 738; Masi et al. 2009, 2012, 2015; Masi and Breuer 2018). The author (SM) and trained Aka trackers collected data on individuals of all age–sex classes belonging to a habituated group of western gorilla (Makumba group), using continuous focal animal sampling (Altmann 1974). During the study, group size ranged between 11 and 13 individuals (one silverback, three adult females, one blackback, zero to two sub-adult females, three juveniles, and three infants; following Breuer et al. (2008) due to emigration by the two sub-adult females. In addition to focal animal sampling, SM conducted instantaneous scan sampling at 10-min intervals (Altmann 1974) on each visible individual. For each scan, the distance between the focal individual and any other visible individual was recorded as one of three categories: 0–5 m, 6–10 m, and ≥ 10 m. Observations pertinent to the death-related events were recorded using a combination of quantitative and qualitative records of all occurrences of affiliative, investigative, and agonistic behaviors of visible individuals.

Because multiple factors can affect inter-individual proximity (i.e., season, food availability, infant developmental stage, etc.), comparison of the proximity of the two adult females involved in the first observation used only scans from the month in which the event occurred (March 2009), and in which either of the two females was the focal individual (Nscans = 64). Moreover, if in one scan distance from the non-focal female was not recorded, a distance of more than > 10 m was assigned (Nscans = 30). To test for differences in distances between the two adult females, I used Friedman test for paired samples for three groups (the three distance categories) and Conover pairwise multiple comparison as post hoc test, with P values adjusted by the false discovery rate (FDR) method of Benjamini and Hochberg (1995). For analyses, we used R software version 3.4.3 (R Development Core Team 2017), R package for the Friedman rank sum test (friedman.test {stats}) and the R package PMCMR (Pohlert 2016) for the post hoc tests.

To compare the frequency of chest beating and hand clapping in the three immatures involved in the second observation, the frequency of these two behaviors was calculated from their focal animal sampling data, respectively (minutes of focal sampling per individual: NEtefi = 1677, NMio = 1165, NMossokabouli = 3239). Given the low frequency of these behaviors, monthly rates were calculated for each individual, and the high number of zeros allowed only qualitative comparisons.

The study adhered to Principles for the Ethical Treatment of Primates, the protocols and legal requirements of the Central African Republic. The data collection protocols were approved by the Ministries of Education and Water and Forests of the Central African Republic.

Results

Reaction to the apparent death of an independent infant

On March 15, 2009, the silverback (Makumba) and other group members were approximately 25 m up in an Angylocalyx pynaertii tree, eating young leaves. SM and one Aka tracker were following a juvenile female, Mossokabouli, who was on the ground feeding within 10 m of the tree. Suddenly, at 10:17 a 2-year-old infant female, Mobangui, fell out of the tree, from a height of more than 15 m. She screamed just before she hit the ground, whereupon she looked dead to us. Almost immediately, her mother, Bombe, descended silently from the tree. Although visibility in the canopy was poor, we neither saw nor heard any reaction from the silverback, who stayed in the tree. An adult female, Malui, continued to feed on the ground, approximately 6–7 m from the unanimated Mobangui. Bombe approached Mobangui and tried to lift her arm as she normally would to help her to climb onto her back for carrying, but Mobangui was completely unresponsive, reinforcing our impression that she was dead. Bombe repeated this gesture twice, after which she stopped and stared for a couple of seconds at the unanimated infant. At 10:21, Malui’s 1-year-old son (Tembo) approached to within 2 m, looked at Mobangui for 1–2 s, and then started to climb the tree from which she had fallen. At 10:22, approximately 3 min after her arrival, Bombe lifted Mobangui and carried her in her arms. We saw no blood or wounds, but Mobangui still gave no visible (to us) signs of life. Bombe walked approximately 5 m and then placed the infant on the ground next to her, before lying down beside it. At 10:24, a juvenile (Essekerende, Mobangui’s eldest full sister) came and sat nearby. At 10:26, Malui stopped feeding and moved unusually close to Bombe. She watched as Bombe licked a wound on her own arm that had appeared the previous day. For more than 40 min, Malui and her two offspring (Mossokoabouli and Tembo) lay resting unusually close—within 1 m—from Bombe and Mobangui. Overall, during the same month, these two adult females spent very little time in close proximity, usually being more than 10 m apart (Fig. 1; Friedman rank sum test: X2 = 6.781, df = 2, P = 0.034, N(0–5m) = 13, N(6–10m) = 21, N(> 10m) = 30; all post hoc analysis provided P < 0.0001). At 10:31, Bombe laid her hand on the still-unanimated Mobangui. At 10:33, we started to suspect that the infant was actually not dead, as she appeared to have moved slightly. At 10:35, Mobangui extended an arm vertically, but Bombe did not react to this. One minute later, the infant clearly moved then lay down with her eyes open, a few centimeters from her mother. At 10:35, the silverback descended from the tree and rested approximately 8–10 m from the injured infant. He was followed out of the tree by a third adult female (Mopambi), and her infant (Bokata), but none of these three individuals approached closer. At 10:38, in response to an accidental touch by Essekerende, Mobangui turned slightly, but she remained lying down. At 10:43, she moved her left leg for the first time. At 10:53, the youngest infant in the group, Tembo, approached Mobangui playfully; the latter opened her eyes and turned onto her side, at which time Bombe emitted a close contact call (Salmi et al. 2014; Hedwig et al. 2014); Mobangui remained immobile. At 11:01, Bombe started looking closely at her infant, with whom she was still in physical contact. Mobangui did not move again until 11:09, when she moved one arm and her left leg slightly. The group moved away at 11:15, causing Bombe to help Mobangui to climb up onto her back and start following. After 2 min, Mobangui climbed down from her mother’s back and walked independently, slowly and limping, her right leg and right wrist frequently giving way. Bombe carried her again for 1 min, after which Mobangui resumed walking/limping behind the group. She repeatedly laid down suddenly on the ground and rested. Her mother never moved more than 5 m from her.

Fig. 1
figure 1

Percentages of March 2009 scans showing relative distance between the mother (Bombe) of the unanimated infant (Mobangui) and the adult female (Malui) who stayed within 1 m of them after the infant fell from a tree. Error bars indicate standard errors. ***Shows significance at P < 0.0001. See text for details and for sample size per category

Full size image

The following day at 10:45 we found the group after a long downpour. Everyone was feeding in a tall tree (50 m high), with Mobangui in the lower canopy at a height of 15–20 m. After 15 min, Mobangui climbed up and walked normally to join the rest of the group. After this day, she never showed any signs of long-term injury.

Reaction to a dead red river hog

On June 27, 2008, SM and two Aka trackers were following the blackback of the study group (Kunga) when, at 14:33, he encountered a dead red river hog (Potamochoerus porcus). He stopped 2.5 m from the corpse and stared at it for 1.6 min. The corpse was intact, quite fresh, with no signs of blood but starting to be covered by flies and larvae (Fig. 2). The cause of death was not evident, and we estimated the corpse at less than 24 h old. There was no strong odor or bloating of the body, which in fact appeared emaciated (Fig. 2). Kunga slightly changed direction and made a detour around the corpse, never getting closer than about 2.5 m from it. He walked away, but we did not follow him because we noticed some young members in the group starting to behave unusually. At 14:35, a sub-adult female (Mio) climbed up a small tree within 5 m of the corpse. At the top of the tree, she chest beat while staring at the corpse (Fig. 3a). Her younger sister, the juvenile Mossokoabouli, climbed up a neighboring tree and displayed the same behaviors (chest beating, staring). At the same time, another sub-adult female (Etefi), half-sibling of the first two, climbed into the same tree as Mio (Fig. 3b). At 14:42, the three young gorillas still appeared agitated: still in the trees, they chest beat and turned to look in alternation at each other and at the corpse. Mio had performed 12 bouts of chest beating, Mossokoabouli six, and Etefi two; some bouts were accompanied by urination and defecation (Fig. 3c). At 14:52, Mio chest beat for the final time. While still in the trees, none of the three gorillas tried to get closer to the corpse, and no other group members were seen on the ground nearby, although the silverback emitted a close contact call in the vegetation nearby (Salmi et al. 2014; Hedwig et al. 2014) just after a single hand clap by Mio (Fig. 3d). At 14:53, the three youngsters climbed down to rejoin the group who had moved far; upon which we left to find Kunga.

Fig. 2
figure 2

The corpse of the dead red river hog encountered by the Makumba group. Flies and white fly larvae are clearly visible, the latter particularly on the muzzle of the corpse, which appears emaciated

Full size image
Fig. 3
figure 3

a–d The sub-adult female, Mio, staring at the corpse of the red river hog (a). Another sub-adult female, Etefi, joined Mio on the tree to look at the corpse (b). The gorillas seemed agitated and worried by the presence of the corpse, defecating and urinating (c) while chest beating, hand clapping (d) and looking in alternation at each other and at the corpse

Full size image

Notably, the youngsters very rarely chest beat or hand clapped for the rest of the study period; only in February 2009 (Fig. 4) was it comparable with that of June 2008, when the dead hog encounter took place. However, the high frequency in February was due entirely to Mio, with nine chest-beating bouts and one hand clap on the same day, during rest and play.

Fig. 4
figure 4

Monthly frequencies of chest beating and hand clapping during the study period for the three immatures observed when they encountered the dead hog in June 2008. This graph is based on combined focal animal data for the three individuals (see text for details)

Full size image

Discussion

This paper describes the reaction of a western gorilla mother to the apparent death of her independent infant, and the first observed reactions of wild western gorillas to a dead red river hog. Documenting such rarely seen events can contribute positively to the growing field of evolutionary thanatology (Anderson et al. 2018).

Given the heavy biological investment of mammalian mothers in their offspring, that the unanimated infant regained consciousness supports the hypothesis that it may be adaptive to continue carrying and caring for an unanimated infant, at least for a short time (Appleby et al. 2013; Hrdy 1999). On a proximate level, the mother may perceive the infant’s condition as ambiguous, or she may anticipate recovery, thus she continues caregiving. This should be particularly true for experienced mothers (Biro et al. 2010; Sharma et al. 2010), and here, Mobangui was at least the third infant of Bombe. On an ultimate level, too readily abandoning a temporarily unresponsive but potentially viable infant would be a costly error. However, accidents resulting in unconsciousness are probably rare, whereas prolonged carrying and care of dead infants is widespread (reviewed in Watson and Matsuzawa 2018). Social learning may play a role in maintaining this behavior within a population (Biro et al. 2010): witnessing dead-infant carrying by others may promote the behavior in mothers experiencing their own infant death (Biro 2011). We know that the gorilla Bombe had seen another female transporting her dead infant at least once previously (Cipolletta, pers. comm.).

The question arises though why a mother having witnessed prolonged carrying of a obviously dead infant, would reproduce this costly and “useless” behavior. One possibility is that the recovery of infants that get stunned or injured during accidents (see van de Rijt-Plooij and Plooij 1987) may act as a positive reinforcement for such behavior. Indeed, the extraordinary capacity of wild animals to recover from near-fatal situations, as observed here, may partly explain at least short-term continued caring and transport of dead infants. Moreover, in some cases, unconsciousness may be difficult to distinguish from “possibly dead”. Therefore, ascertaining the irreversibility of the state before abandoning the corpse may be profitable. In our observation, the unanimated infant was 2 years old, which rules out birth-related hormones as the trigger for the mother’s behavior. Infant corpse-carrying by non-mothers also suggests that a hormone-based hypothesis cannot fully explain prolonged carrying and caring (Trapanese et al. 2019; Watson and Matsuzawa 2018; Tokuyama et al. 2017; Fashing et al. 2011; Tian et al. 2016; Warren and Williamson 2004; Kano 1992).

In contrast to the mother’s immediate response, the initial lack of reaction by the silverback to the infant’s fall and scream was surprising. For example, infant screams often cause silverbacks to intervene, for example to prevent infanticide (Harcourt and Stewart 2007; Masi and Bouret 2015). It is noteworthy that the same silverback also showed no reaction when another adult female in his group vocalized (barking) in distress repeatedly while stuck in a tree for more than an hour with her newborn infant; she finally jumped 2–3 m to get back down (SM, pers. obs.). Did the silverback fail to perceive that another group member might be in danger, or did he lack empathy or complex Theory of Mind (see Devaine et al. 2017)? The apparently fatal accident did not elicit much response from the other group members either, other than a brief look from a curious 1-year-old infant. This contrasts sharply with what is described for eastern gorillas seeing dead conspecifics (Porter et al. 2019), and chimpanzees after a community member died on falling out of a tree (Teleki 1973). In these cases, the other individuals remained close to the corpse, displaying and vocalizing loudly. Chimpanzees typically show higher social tolerance and social complexity than western gorillas (Masi et al. 2009, 2012; Goodall 1986; Lonsdorf 2006; Boesch 1991; Humle et al. 2009). Moreover, compared to eastern gorillas, western gorillas rarely exchange grooming (except in specific situations such as infant care; Masi et al. 2009), and adults do not usually stay in close proximity each other, particularly adult females (Harcourt 1979; Masi et al. 2012). Another possibility is that the gorillas rapidly perceived that the infant was still alive after the fall. This could also explain why the mother did not react to the infant’s post-unconscious first movement. In any case, the only other adult female who witnessed the accident from nearby also did not react to the infant’s fall or scream, although she and her offspring remained unusually close to mother–unanimated infant pair. Eastern gorillas tend to gather closely around a dead conspecific (Porter et al. 2019) or in dangerous situations; again more observations are required to compare different gorilla populations.

When the gorillas met a non-gorilla corpse, a dead river hog, their behaviors were more similar to the alarmful and agonistic reactions of chimpanzees and eastern gorillas to conspecific deaths (Teleki 1973; Boesch and Boesch-Achermann 2000; Anderson et al. 2016; Porter et al. 2019). Aggression towards a dead conspecific has been interpreted as an attempt to rouse it (Cronin et al. 2011; Wisman and Shrira 2015), a sign of frustration at failure to arouse it (Anderson et al. 2010), or in the case of a dead extra-group gorilla, a demonstration of physical strength (Porter et al. 2019). Gorillas are usually calm and silent in the presence of living red river hogs, in contrast to the reaction to the corpse. Climbing into the tree to observe it, and urination and defecation while clapping and displaying might reflect both fear and threatening the corpse (Harcourt and Stewart 2007), possibly to test its responsiveness.

Table 1 presents all published reports of primates interacting with dead allospecifics. I include only reports in which there was no evidence that the animal was alive prior to the interaction (thus cases in which chimpanzees clearly captured live animals were excluded, e.g., Hirata et al. 2001). The most striking commonality is that only immatures (mainly adolescent and juveniles) showed particular reactions to and interest in allospecific corpses. This is understandable, as the encounters might be their first contacts with dead animals. As in humans (Bonoti et al. 2013), direct experience of death in childhood may facilitate maturation of the understanding of death in nonhuman primates. The reaction of the blackback male, who stared at the hog corpse before making a detour around it (even if the others were loudly alarmed), suggests prior experience and knowledge of corpses; after that he showed no interest or curiosity. Curiosity typifies all cases reported in Table 1, but chimpanzees also displayed more affiliative behaviors, whereas gorillas reacted more aggressively. Young chimpanzees and bonobos are reported to interact with dead animals or captured prey in playful or exploratory ways (Cibot et al. 2017; Hockings et al. 2012; Carvalho et al. 2010; Hirata et al. 2001; Boesch and Boesch 1989; Zamma 2002; Sabater Pi et al. 1993), whereas gorillas seem more likely to avoid physical contact with or even proximity to corpses. However, all chimpanzee observations concern interactions with small vertebrates (e.g., owls and hyraxes), not a large mammal such as the red river hog. Chimpanzees may also be more used to contacting bodies of other species since they hunt mammals and other small vertebrates (e.g., Boesch and Boesch 1989; McLennan 2010); gorillas do not (e.g., Masi et al. 2015; Doran et al. 2002; Doran-Sheehy et al. 2006; Remis 1997). Interestingly, although eastern gorillas touched and even groomed the corpse of an extra-group conspecific (Porter et al. 2019), western gorillas were more circumspect and fearful toward an allospecific corpse. Possibly, closely inspecting conspecific corpses might provide valuable information about the identity of the dead individual (e.g., kin, or a known competitor) or about a possible danger for their own species, which outweighs the potential risk of contamination (Porter et al. 2019).

Table 1 Published reports on the reaction of primates interacting with allospecifics corpse
Full size table

The wide animal variation in response to corpses reflects the relationship of the deceased to the living, the cause of death, and the condition of the corpse, the latter two from the compiling of different sensory cues (reviewed in Gonçalves and Biro 2018). The presence of flies and larvae on the hog corpse may have provided the gorillas with cues to the dead state of the animal. In the case of the unanimated infant female, her mother attempted three times to pull her arm to her shoulder like she would have done under normal circumstances. Perhaps she was testing the infant’s responsiveness, similarly to what chimpanzee mothers do when testing their dead infants (Cronin et al. 2011). Presence of odor, wounds, infestation, and absence of movement, sound, or body heat might all be used by primates to evaluate a corpse (Gonçalves and Biro 2018).

Given the existence of infanticide in gorillas (Harcourt and Stewart 2007; Masi and Bouret 2015), several individuals in a group may be familiar with permanently nonfunctional infant bodies. Moreover, in eastern gorillas, premature death inflicted by a non-dominant silverback on an old, ill adult female or to a young silverback during inter-group encounters, shows that gorillas may experience other deadly contexts (Veit 1982; Williamson 2014). In addition, it is not unusual to encounter dead animals in tropical forests, giving the opportunity to accumulate experiences with both living and dead individuals of a given species. Therefore, it is likely unnecessary to hunt or kill animals to acquire and develop an awareness of death. Most published work on great ape behavior around corpses has focused on mothers carrying their dead infants, consumption of prey, or deaths of older conspecifics. The cases described here of responses to an unanimated conspecific infant body and an allospecific corpse offer different perspectives for better understanding primates’ response to fatal events and related cues.