Data collection

Ethical note

This study was purely observational and did not involve any interventions, apart from daily visits to the two study communities. Researchers and field assistants follow strict hygiene, quarantine, and observation distance rules to prevent disease transmission, as detailed in the project guidelines (www.budongo.org). Permission to study the chimpanzees has been given by the University of St. Andrews Animal Welfare and Ethics Committee, as well as the Uganda National Council for Science and Technology, following the advice of the Uganda Wildlife Authority, and National Forest Authority.

(a) Hunting behaviour: Prey species hunted and success rates

Sonso.

The Sonso chimpanzees fed on the meat of six species, including four primates, following 203 hunts, 182 of which were successful over a 17-year observation period (89.7% success; Table 1). These numbers are an underestimate, partly because it is impossible to document all hunting events throughout the year. This is particularly true for unsuccessful hunts, which are more difficult to document. When we examined data from a 22-month period (January 2010 to October 2011) in which CM collected all evidence of hunting activity during bouts of focal animal sampling, we found significantly lower success rates (focal sampling hunt success: 8/19 hunts, 42.1% success; long-term data 182/203, 90%, p<0.0001, 2-tailed Fisher’s exact test).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Specialization of hunting in the Budongo chimpanzees, 1999–2017.

https://doi.org/10.1371/journal.pone.0178065.t001

Results are displayed as number of each species hunted, followed by the percentage that this represents of all animals hunted in parentheses. Adult body mass of prey species is marked in kilograms [66] with male and female mass separate where there is sexual dimorphism. Events per species are given for both observed attempts (includes all hunting activity irrespective of success) and successful hunts (successful hunts include events where the hunt was not observed but the subsequent meat sharing was). Note: as hunting behaviour was not studied directly until 2008 unsuccessful hunts frequently went unobserved, and even when observed were unreported. Data from unsuccessful hunts are reported in order to provide additional information on the frequency with which different species were hunted, rather than as an accurate indication of hunting success. *Data taken from Newton-Fisher et al. [5]; only successful hunts were reported.

Guereza colobus monkeys (Colobus guereza occidentalis) constituted 74% of all prey items, significantly more than all other species combined (attempted hunts: p<0.0001; successful hunts: p<0.0001; binomial tests, two-tailed), indicating a clear bias towards this species (see Table 1). On a single occasion Sonso chimpanzees were observed to chase a red duiker (Cephalophus natalensis), apparently with the intent of capturing it, but the attempt was unsuccessful. Juvenile and infant chimpanzees were observed to interact playfully (chasing without aggression or attempts to kill or injure) with both red and blue duikers (C. monticola), as well as to groom and play with three monkey species: blue monkeys (Cercopithecus mitis), red tail monkeys (Cercopithecus ascanius), and olive baboons (Papio anubis).

87.6% of Guereza colobus monkey hunts involved multiple chimpanzees (N = 89 records with sufficiently detailed information on identity of hunters). Solo colobus hunts, while rare (n = 11), were not less successful than group colobus hunts (n = 78) (success rates: solo hunts: 9/11; group hunts: 60/78; p = 1.00, 2-tailed Fisher’s exact test). In non-colobus prey species (three primates and blue duiker; N = 24 records with detailed information on identity of hunters), solo hunts were significantly more common than for colobus prey (solo rates: non-colobus: 22/24, colobus: 11/95; p<0.0001; 2-tailed Fisher’s exact test). If analysing non-primate species only, 100% of hunts were solo (n = 10).

(b) Rates of hunting in Sonso

Given the bias towards reporting successful hunts described in section (a) and other possible confounds in a long-term data set such as observer bias or experience, and the effects of chimpanzee habituation, we are cautious in exploring ‘rates’ of hunting in the long-term data set. For example: if we apply the success rate from the 22-month specific study of hunting (42%) to the long-term data set (n = 182 successful hunts) this would predict n = 433 hunts over the 17-years, more than twice the n = 203 actually reported. The hunting rate before the correction would be 11.9 times a year (roughly once a month); after it would be more than double that at 25.5 times a year.

Moreover, hunting rates may have changed substantially over the years. Newton-Fisher et al. [5] report a successful hunting rate of n = 17 in 8-years, or 2.1 hunts per year. If we again estimate a success rate of 42%, we obtain a rate of 5.1 total hunts per year. We can compare this with the section of long-term data that follows immediately on from this period (14-years, 2003–2017, n = 175 successful hunts), a rate of n = 13 successful hunts per year, or (with a 42% success rate), an estimated n = 30 total hunts per year, which corresponds to an increase of 6-times that reported in the earlier period. Nevertheless there may still be large year-to-year variation in the annual hunting rate; only 11 hunts were recorded in 2015, and only 3 hunts in 2016, a sharp decline from the 19 recorded in 2014, or the 17 recorded in 2013.

Within a given year, we can look at hunting rates from month-to-month. Given our findings above we were extremely cautious about drawing conclusions across years. Instead, we tested for variation on a larger scale by comparing the pattern of hunting month-to-month during the first 9-years of our full data set (1999–2007) with the subsequent 9-years of our full data set (2008–2016). We found that the distribution of hunting frequency month-to-month varied between the earlier and later data on hunting in Sonso for colobus prey (chi square: Χ² = 29.79, df = 11, p = 0.002) but not for non-colobus prey (chi square: Χ² = 11.86, df = 11, p = 0.375) between the earlier and later data on hunting in Sonso. As indicated above, however, when we examined the trends from year to year we see striking differences, so that while the period from April–Jul appears to represent a period of relatively low hunting frequency over the past 9-years (see Fig 1, panel B), we see that in 2008 this pattern was reversed, with the majority of hunting that year recorded between May and August (Fig 1, panel C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Variation in month-to-month hunting of colobus and non-colobus prey by the Sonso chimpanzees.

Monthly rates in panel A and B were calculated by calculating the hunting rate per month for each year, and then averaging across these to control for year-to-year variation within the data sets. Panel A shows the hunting behaviour in the 9-year period from 1999–2007 inclusive; Panel B shows the hunting behaviour in the 9-year period from 2008–2016 inclusive. Panel C and D show the colobus and non-colobus hunting behaviour respectively in the 9-year period from 2008–2016 for each individual year (note that here the axis represents total hunts per month across the 9 years, rather than mean hunts per month, to allow year to year comparison).

https://doi.org/10.1371/journal.pone.0178065.g001

(c) Abundance and density of potential prey species

The Mean Encounter Rate for mammals, including both primate and non-primate species, monitored appeared higher in Sonso than in Waibira; however, the only statistical difference we found was between blue monkey populations, which appear higher in Sonso (see Table 2). Although Red duiker and Bushbuck were combined here due to difficulties in reliably distinguishing dung, Red duiker were seen on n = 5 occasions in Sonso and n = 4 occasions in Waibira; and Bushbucks were seen on n = 4 occasions in Sonso and n = 1 occasion in Waibira during the survey. Population densities of the three primate prey-species appeared higher in Sonso than in Waibira (see S1 Table); however, as the coefficients of variation (CV) were high (CV range = 17.4 to 39.6%) the sample sizes were likely too small for reliable estimation of density.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Mean encounter rate of potential prey species.

https://doi.org/10.1371/journal.pone.0178065.t002

(d) Meat-eating: Monopolization and sharing

Variation in prey-species hunted

At many chimpanzee study sites, hunting efforts appear biased to red colobus monkeys (Piliocolobus spp.), including areas where Guereza colobus monkeys are also present [3, 67, 68]. It appears that in the absence of red colobus, which are not present in the Budongo Forest, Sonso chimpanzees have instead specialized in hunting Guereza colobus monkeys. They show a clear bias, in terms of the frequency of both attempted and successful hunts. Feeding on the meat of other species appears less preferred, with individuals even discarding the carcass uneaten.

In contrast, the first observations of the Waibira chimpanzees show no such bias: in addition to Guereza colobus, they hunt duiker with equal frequency and show similar social behaviour during meat-eating of both species. While Sonso chimpanzees have been observed to occasionally hunt blue duiker, they have never been observed to successfully hunt a red duiker in 27-years of almost continuous observation. Red duikers are large (around double the body mass of the blue duiker) and Waibira chimpanzees (including females who immigrated from Sonso) show similar behavioural levels of excitement during red duiker-meat eating, comparable to Sonso chimpanzees feeding on colobus. However, to date the number of hunts and meat-eating events recorded in Waibira remain relatively small; as a result it is very difficult to establish the absence of a particular behaviour which may be relatively infrequent, for example the abandoning of uneaten carcasses seen in Sonso.

Genetic differences between these two neighbouring communities are unlikely to explain the current variation in prey species hunted. There is evidence of regular genetic exchange between the groups, with at least four Sonso-born females known to have immigrated to the Waibira community in the past 5-years [69]. Ecological differences are harder to rule out. While the two communities share the same continuous forest habitat, it remains possible that minor variations in, for example: canopy structure [11, 18] may impact colobus hunting, or may impact encounter rates with the different prey species in the two communities. However, we also see a remarkably similarity in the pattern of hunting frequency and prey species hunted in Waibira today compared to that in an early study of Sonso hunting between 1994 and 2002 [5]. In both cases, hunting was reported only infrequently and with no bias towards colobus prey. While there has been limited illegal logging within the relatively protected range of the Sonso community, there is unlikely to have been a substantial change in canopy conditions within Sonso in the years before and after 2002. It seems more likely that human presence during the habituation of a new community contributes to the observed differences in colobus hunting: either between early and later Sonso data, or between current Waibira and Sonso data. The habituation hypothesis suggests that human presence may impact chimpanzee behaviour for an extended period of time, and should be examined as a potential ecological factor before cultural explanations can be considered.

Successful colobus hunts usually require multiple hunters [23, 43]. This requirement is less true for other prey species, which can be captured through opportunistic solo hunting. It is possible that the process of habituating chimpanzees disrupts complex social hunting attempts, and, in doing so, biases the prey-record in hunting observations towards solo-hunted species. It is difficult to test this hypothesis directly (as there are no prey-records from non-habituated communities); however, hunting, in particular group hunting, is frequently one of the last behaviours to be recorded in a new community. At all major chimpanzee study sites, hunting rates tended to be low during the first 5–10 years of observations (Tai [18]; Gombe [2]; Mahale: [8, 70]). At the same time, prey spectra were distributed across many species, particularly those that could be opportunistically grabbed by individual hunters [1, 8, 9]. At Sonso, where habituation started in 1990, only 17 hunts were recorded in the 8-years between 1994 and 2002 [5]. During this time the frequency of colobus to non-colobus hunting was the same as currently seen in Waibira today: around one third of all hunting attempts. It is only after 2002 that the Sonso prey bias towards colobus emerged. In Mahale researchers have described a very similar long-term shift in the prey profile, with an increase in the proportion of colobus prey only after 16-years of researcher presence (from 14% in the first 16-years to 56% in the next 7-years, and finally up to 83–84% consistently for the past 14-years [70]). In this study we compare hunting rates in the two halves of our 17-year data set–showing a striking difference in both the frequency of hunting and the distribution of prey species hunted. As a result, we caution against describing ‘typical’ hunting behaviour from the Budongo communities by averaging over long-term data, even once chimpanzees appear habituated. Hunting rates in Sonso in 1994–2002 were extremely low (average successful hunts 2 per year [5], estimated average total 5 per year). Over the past 14-years they are comparable to other east African communities (Sonso: average successful hunts = 12 per year, estimated average total = 27 per year); higher than some (e.g. Kanyawara [52]), lower than others (e.g. Kasekela [52]). In addition, even once habituation is extremely high, we see fluctuations in both the annual rate and apparent seasonal variation. Other factors, such as the presence of particular individuals who act as ‘impact hunters’ [52] or specific environmental events that may impact diet (e.g. El Nino years [71]), may also impact hunting behaviour and account for the variations observed in Budongo.

If the habituation hypothesis for the variation in colobus specialization seen in current Waibira and Sonso hunting (or early and late Sonso hunting) is correct, one striking question is, why does it take so long for groups to resume colobus hunting after the beginning of habituation? During the period of the first Sonso study (1994–2002) the majority of the mature individuals in the group were considered habituated, in that they could be followed without any obvious signs of stress or disruption to their daily activity [6]. Nevertheless, researchers with an interest in hunting behaviour who were present at this time rarely observed any signs of it (n = 17 in 8 years, [5]). Sonso chimpanzees predominantly hunt colobus prey as a group, whereas most hunts of other primates and all hunts of non-primate species were solo hunts. It would be interesting to examine whether early records of colobus hunting in Sonso contained a greater proportion of solo and opportunistic hunts compared to present day hunting behaviour. In Waibira, four of the six non-successful colobus hunts were all solo chases. However, it is difficult to obtain reliable data from early records because hunting behaviour, as opposed to subsequent meat-eating behaviour, is harder to observe during the early phases of habituation. One possible explanation is that the easier ‘fall-back’ option of solo non-colobus hunting becomes normalised for the generation of individuals who matured during the first few years of habituation. Some colobus hunting remains throughout but, given chimpanzees’ dietary conservatism [70, 72], it may take time for the prey bias to re-establish following the forced disruption from habituation and reincorporate species into their ‘prey profile’.

A concrete test of this would be to perform dietary analyses on faecal samples of non-habituated chimpanzee groups; if habituation does disrupt hunting we would predict that the prey preferences and hunting frequency in unhabituated groups should resemble those of communities in which hunting has resumed. Similarly, while not yet widely employed, it may be possible in the future to investigate unhabituated chimpanzee behaviour through the use of passive acoustic monitoring systems. Recently used to establish chimpanzee and other primate species density, ranging, and territory use [73, 74], the ability to triangulate the co-occurrence of both chimpanzee hunting barks and prey-species alarm calling may provide more detail on encounter rates and unsuccessful hunting attempts in unhabituated communities.

Neither ecological conditions nor the habituation hypothesis is likely to fully explain the absence of red duiker hunting in the Sonso communication. The Waibira group has repeatedly and successfully hunted red duiker, while the Sonso community has never been observed to do so. We found no differences in ungulate abundance between the sites, and although we were unable to distinguish Red duiker dung from Bushbuck dung reliably, Red duiker were encountered during surveying in both areas at a similar rate. As found in our own data, a recent large-scale survey suggested that there is little variation in duiker populations between the Sonso and Waibira areas [75].

Instead, this variation in duiker preference between communities may be a possible example of a ‘cultural phenomenon’ comparable with the group differences in food preferences already found between commonly available plant foods [76]. Interestingly, the four confirmed Sonso-born females who had recently emigrated to Waibira have all been observed begging for and feeding on red duiker meat, despite having never had the opportunity to do so in their natal community. While this is particularly surprising given the evidence for behavioural conservatism in chimpanzees [72, 77], recent research suggests that young immigrant females may rapidly conform to behavioural variations within their new community [78].

Harassment and meat sharing

The apparent variation between neighbouring communities in meat-possession and sharing behaviour is of interest, but must also be interpreted with caution. A frequently encountered issue with long-term data is the tendency to ignore observations considered ‘typical’ and the non-systematic ad libitum nature of the entries. To date, our long-term data suggest that the ‘typical’ behaviour of the Sonso chimpanzee community when feeding on meat is that higher-ranking individuals harass and aggressively attack lower-ranking individuals to obtain possession of a carcass. Peaceful monopolization of a carcass by a low-ranking individual would be considered highly atypical in Sonso, but we find no recorded cases of it. We were unable to document the same pattern in the Waibira community, where lower-ranking sub-adult individuals maintained possession of any prey type for long periods with only non-contact or low-contact begging (classed as minimal harassment).

Here, as with prey-preference, it remains possible that differences in habituation could account for the observed behavioural differences. For example, sub-adult individuals tend to habituate more quickly than other age-classes [79] and are typically lower-ranking than adult individuals; however, higher-ranking individuals were present throughout our observations and in close proximity to the individual in possession of the carcass. It seems unlikely that a lower habituation level alone would inhibit aggressive behaviour towards lower-ranking individuals, but not inhibit the tendency to approach or remain in proximity to them. In addition, there are no records of similarly peaceful, extended, monopolization of large portions of a desirable carcass by low-ranking individuals in the presence of high-ranking individuals in the Sonso community, despite 182 hunting observations in an 18-year period. Even in the early period covered by the 2002 study, when habituation levels were lower, it was typical for high-ranking Sonso individuals to monopolise the carcass (Newton-Fisher, pers. comm.). A study of Fongoli chimpanzees in Senegal recently described observations of regular female meat monopolization, procured without actively hunting for it, and subsequent successful refusal of male begging behaviour [16]. The authors suggest that this supports a social bonding hypothesis, with the low levels of male aggression towards female meat possessors representing a long-term social investment [16]. If, as Pruetz et al. [16] suggest, this behaviour is part of a long-term social strategy, then there may be particular social and/or demographic factors that promote this, for example the ratio of adult males to adult females, resulting in its more regular use.

Our observations of the Sonso and Waibira community hunting behaviour suggest that 1) neither ecological nor genetic factors satisfactorily explain variation in chimpanzee hunting preferences, indicating that behavioural differences may represent responses to human observers or, in the case of red duiker prey, socially learned traditions and that 2) the energetic costs imposed by harassment alone may not provide a complete explanation of meat sharing in Budongo chimpanzees, to which social factors appear to make an important contribution.