Vocal communication in wild chimpanzees: a call rate study

Study site

We collected data for three months during the late dry season, between August and October 2019, in the Issa Valley, western Tanzania (Fig. 1). The study area of about 70 km² is comprised of a series of valleys separated by steep mountains and flat plateaus, with an altitudinal gradient ranging from 1,050 to 1,650 m above sea level. Vegetation is dominated by miombo woodland and also includes grassland, swamp and riparian forest. For analyses, we collapsed vegetation categories into ‘open’ (woodland, grassland, swamp) and ‘closed’ (riparian forest), see Fig. 1. The region is one of the driest and most open that is inhabited by chimpanzees (Moore, 1992) and characterised by two seasons: wet (November to April) and dry (May to October). Annual rainfall averaged 1220 mm per annum (range from 930 to 1,490 mm from 2009 to 2014) and temperatures ranged from 11 °C to 38 °C (Piel et al., 2017).

Study subjects

The study site covers the territory of at least two chimpanzee communities (one habituated community and one or more neighbouring communities). When the study began, the habituated Issa community comprised nine adult females, seven adult males, four subadult males, one subadult female, three juveniles and four infants (Table 1). During the study period, one female (AZ) was not yet fully habituated so we have not included data on her, as researcher presence likely influenced her natural behaviour (Crofoot et al., 2010; Nowak et al., 2014). Two individuals (adult female and infant) were killed and a female gave birth during the study period. The habituated community has a home range ≥55 km².

Data collection

We selected a focal chimpanzee (adult, subadult or juvenile) each morning and tried to follow him/her for the entire day i.e., from nest to nest). We conducted instantaneous focal animal sampling, with a scan defined as the behaviour of the animal recorded every five minutes, when we collected data on caller location (GPS), behaviour (travelling, feeding, resting, grooming, playing, other), vegetation (open or closed) and party size, defined as the number of juveniles, subadults and adults seen. We further noted all vocal behaviour ad libitum of the focal, including the type of vocalisation (pant grunt, grunt, hoot, pant hoot, bark, scream or combinations of different types—see e.g., Crockford, 2019; Goodall, 1986 for descriptions of chimpanzee vocalisations). Successive calls were considered as new events when separated by more than one second. We included only vocalisations that involved at least a scream, a pant-hoot or bark in the analyses, to match the calls that can be potentially recorded by an acoustic recorder deployed about 500 m away from the caller. We thus excluded grunts and other closed calls that do not propagate far. We included only hours of follow data with at least 10 scans/hour, which corresponds to at least fifty minutes of observation per hour.

Data analyses

We conducted all analyses in R v.3.6.1 (R Core Team, 2019). For each hour of focal follow, we determined the proportion of time spent travelling, feeding and other behaviours, in open or closed vegetation by the focal individual. We defined the proportion of time in e.g., open vegetation as the number of scans in which the focal was observed in open vegetation divided by the total number of scans where the focal was in view.

To model the number of calls per hour (call rate, CR) as a function of the covariates, we used a negative binomial distribution GAM with a log link function, that allowed us to account for overdispersion. Fixed covariates were (1) time (T, categorical with 11 levels: from 7am to 6pm), (2) mean hourly party size (PS, continuous), (3) presence of swelling female in the party (PrS, categorical with 2 levels: presence or absence), (4) proportion of time spent in closed vs. open vegetation (F, continuous), (5) proportion of time spent feeding (Fe, continuous), (6) proportion of time spent travelling (Tr, continuous) and (7) age-sex class (AS, categorical with 3 levels: adult-subadult female, adult-subadult male, juvenile). To incorporate the dependency among follows on the same day, we used ‘individual’ as a random intercept and to incorporate the dependency among observations of the same individual at the same time, we used “date” as a random intercept. We used the package mgcv (Wood, 2017) to fit the model. We centred and scaled continuous predictors.

We tested predictors for collinearity by calculating variation inflation factors (VIF) using the package car (Fox & Weisberg, 2018) in an equivalent linear model including only the fixed effects from each model fitted with the package MASS (Brian et al., 2020). Multicollinearity was not present (maximum VIF: PS = 1.53). We verified model assumptions by plotting residuals versus fitted values and QQ-plots. We ran a set of models and ranked them by AICc value.

Factors influencing the call rate

We did model averaging among models with ΔAICc < 2. The significant effects in the best averaged model are age/sex, proportion of time in the forest, proportion of time travelling, time-of-day, presence of swollen female, party size, and the two random intercept terms (individual and date) (Table 2).

Results of the GAM revealed that the age-sex class had a significant effect on call rate: overall, Issa chimpanzees exhibited a mean call rate of 1.91 with a 95% CI of [1.52−2.40]) calls per hour for the subadult and adult males, 0.84 with a 95% CI of [0.59−1.21] calls per hour for the subadult and adult females, and 0.50 with a 95% CI of [0.24−1.05] calls per hour for the juveniles. With 95% confidence, males call between 1.5 and 3.8 times as frequently as females and between 1.3 and 7.9 times as frequently as juveniles. Chimpanzees vocalised significantly more as the proportion of time spent in open vegetation and as the proportion of time spent traveling increased (Table 3). The smooth effect of time reveals call rates being highest in the morning decreasing thereafter, before increasing late afternoon (Fig. 2).

Call rate among demographic classes

Our results confirm early reports on sex differences between adult chimpanzee loud call production (Clark, 1996) and a recent study that subadult and adult male chimpanzees more than twice as often as subadult and adult females (Kalan, 2019). The sexual dimorphism seems to be even more pronounced for forest than savanna woodland dwelling chimpanzees. At Taï, males produced on average 2.5 ± 1.08 calls per hour and females produced 0.88 ± 0.32 calls per hour (Kalan, 2019), compared to 1.91 ± 0.12 calls per hour for the males and 0.84 ± 0.18 calls per hour for the females at Issa (this study). We propose two explanations for this sex difference: (1) necessity of maintaining strong bonds and/or (2) sexual selection. Chimpanzees are male-philopatric, where males stay in their natal territory, while females disperse at sexual maturity and migrate to neighbouring communities (e.g., Moore, Langergraber & Vigilant, 2015; Nishida, 1968). This sex difference in dispersal explains strong male-male relationships (e.g., Mitani, 2009). Males tend to develop stronger bonds with other male community members with whom they spend more time, forming a linear dominance hierarchy and developing affiliations and coalitions. East African female chimpanzees, on the contrary, are far less gregarious than males, travel less, and spend most of their time alone or with their offspring in their core home range (Wrangham & Smuts, 1980).

Ecological factors

We found that chimpanzees called most at two times of day (morning and evening) and from open (miombo woodland) compared to closed (riparian forest) vegetation. This temporal information is important for conservationists, who can for instance record sounds with PAM systems at these periods to maximise the likelihood of detecting calls and simultaneously limit the number of audio files to analyse. There are ecological and social reasons for these patterns. Temporally, the bimodal pattern that we found in this study is similar to what has been reported previously at Issa (Piel, 2018) and elsewhere (e.g., Wilson, Hauser & Wrangham, 2007; Wrangham, 1975). Vocalisations allow parties to coordinate movements, notably prior to arrival at their nesting sites (Fischer & Zinner, 2011) or feeding trees (Clark & Wrangham, 1993). Ecologically, callers may exploit ideal sound transmission conditions. The Acoustic Adaptation Hypothesis predicts that animals may adjust their vocal signals to maximize signal transmission and minimize sound degradation, specifically within each environment in which calls occur (Waser & Waser, 1977; Ey & Fischer, 2009; Brown & Waser, 2017). Environmental metrics, such as temperature gradients, humidity and wind, vary with time of day, and can degrade signals, resulting in structural changes to the primary sound (Waser & Brown, 1984). Relatedly, the natural habitat distorts signals as distance from the sound origin increases. Open vegetation, with increased wind-induced noises compared to closed vegetation, can further degrade signals but sounds attenuate faster in closed vegetation because of tree density (Brown & Waser, 2017). If chimpanzees rely more on close contact calls (not included in this study) to maintain party cohesion in closed, rather than open vegetation, the fact that chimpanzees vocalize significantly more often in open vegetation might be explained by their activity, rather than the external environment. Furthermore, the Issa landscape is dominated (>65%) by miombo woodland vegetation, thus it is unsurprising that they vocalize more from open than closed vegetation types. Calls made in open vegetation could serve multiple purposes. First, individuals may spread out more in these areas, given that food sources are more widely distributed here than in closed areas and so calls are functioning to preserve party cohesion. Additional data on inter-individual distance (e.g., party spread) in each vegetation type would help us resolve this. Second, calls travel further in open compared to closed vegetation, with fewer trees to attenuate sound (although see Waser & Brown, 2017), so individuals also may be calling to communicate with distantly located parties. Often counter calls are heard from these woodland pant hoots (A. Crunchant, 2019, pers. obs.), but we do not have comparative data from forests.

Travelling and feeding activities

We evaluated the effect of activity, namely time spent travelling and feeding, on call production rate. While previous studies have shown that chimpanzees frequently pant-hoot at food sources (Clark & Wrangham, 1993; Fedurek, Donnellan & Slocombe, 2014; Wrangham, 1977), call rate did not change with the proportion of time spent feeding in our study. However, our brief field season took place during the late dry season, when fruit availability is highest (McLester et al., 2019). The abundance of food might reduce competition among individuals and thus depress the need for calling. Furthermore, we did not incorporate which food items were consumed. Fedurek and colleagues (2014) have shown that male chimpanzees were more likely to pant hoot at high-quality food patches and thus we suggest that subsequent studies consider food type consumption during calling bouts.

The number of calls increased as the proportion of time spent travelling increased. It has long been demonstrated that loud calls facilitate fusion events and help regulate grouping dynamics and coordination among community members during travel (e.g., Goodall, 1986). In other communities, males travel significantly more after a pant hoot is produced (Mitani & Nishida, 1993) and males are more likely to repeat a call prior to, rather than after, fusion with other males (Fedurek, Donnellan & Slocombe, 2014).

Party size and presence of parous swollen female

We have shown that call rate was positively related to party size. Coordinating movement with a higher number of individuals requires more communication, especially for decision making and coordinating party fusion (Fischer & Zinner, 2011). Furthermore, a previous study has shown that the number of aggressive events is positively correlated to the number of males in a party (Muller, 2002), and thus the number of calls produced during agonistic events (i.e., barks, screams and pant hoots) would likely increase with the number of males in the party.

Finally, the presence of a swollen female in the party did also impact call production rate, as expected. It has indeed been shown that males prefer to mate with parous females and are more involved in male-male competition in their presence because parous females are attractive to males, which exhibit aggressive displays and courtship behaviour during these times (Muller, Thompson & Wrangham, 2006).

Implications and limitations

Variability in chimpanzee calling is well known, described in nearly every study on the topic; early work attributed variability to age, sex, context, party size, and community, among others (Goodall, 1986; Arcadi & Wrangham, 1993; Arcadi & Wrangham, 1994; Mitani et al., 1992; Mitani et al., 1994). For the most part, these studies reported variability in the context of call function. In the current study, however, we were not interested in call function. Whilst similar to some previous work, we report on context-specific variability, our primary interest is in the implications of call rate variability for conservation studies that rely on PAM to extract animal densities. For example, sex ratio, vegetation types and proportions, and activity budgets vary between communities. Our results here begin to reveal how call rate changes with each of these, which has direct bearing on how its parameterised in density analyses. Specifically, for communities that range across more open vegetation - Issa (Tanzania), Fongoli (Senegal), Semliki (Uganda) - call rate should reflect this in density models, varying with sensor location.

Those implications notwithstanding, we are cautious interpreting these results and conclusions should not be generalized to scenarios other than that where the data were collected. We sampled over a single and brief 3-month season and recorded only 487 h of follows (compared to Kalan (2019) who recorded 731.5 h of follows across 10 months). With food availability, party size and the number of swollen females highest at this time of year (A. Piel, 2021, unpublished data), it is likely that our data are inflated against annual means. More data collected across multiple seasons would resolve this uncertainty and would result in seasonal variation in call production, especially lower call rates in the early wet season when party size declines and fewer females exhibit maximum tumescence. We were also unable to equally balance data collection across the community, with some individuals being followed only 5 h and some more than 30 h. This was mostly due to the difficulty of following recently habituated females. Given that the call rate variation between two individuals is likely to be greater than the variation within the same individual, rather than over-sample fewer individuals, we chose to sample more individuals.

Future directions

Future studies can not only build off these results by adding more data, but also additional predictor variables that may influence chimpanzee call rate. Specifically, further analyses would benefit from evaluating the effect of rank on call rate. For instance, rank has been implicated in influencing call rate in chacma baboons (Papio cynocephalus ursinus) (Kitchen et al., 2003), gelada monkeys (Theropithecus gelada) (Benitez et al., 2016), orang-utans (Pongo pygmaeus) (Mitani, 1985) and also non-primate (fallow bucks (Dama dama) (Pitcher et al., 2014)) species, and strongly suggests that vocal communication is influenced by sexual selection. In support of this, multiple studies examining chimpanzee rank and call production reveal a positive relationship between male quality, testosterone, and rank (Clark, 1993; Clark & Wrangham, 1993; Mitani & Nishida, 1993; Fedurek et al., 2016). There is not yet empirical evidence however, demonstrating female preference for male vocalisations.

Moreover, we recommend that subsequent studies include a spatial component. For example, chimpanzees spend the majority of their time in the core of their home range, usually representing about 75–90% of their total territory (Wilson, Hauser & Wrangham, 2007). When males conduct patrols in high risk areas, e.g., croplands or territorial boundaries, call rate declines significantly (Wilson, Hauser & Wrangham, 2007). Chimpanzees from savanna habitats have far larger home ranges—e.g., ≥ 55 km² at Issa (C. Giuliano, 2021, unpublished data) or about 90 km² at Fongoli (Pruetz & Herzog, 2017)—than forest dwelling chimpanzees—e.g., 7 km² at Budongo (Newton-Fisher, 2003). Suzuki (1969) proposed that the low density and widely distributed foods in these savanna landscapes may promote more nomadism in savanna than in forest chimpanzees and Moore (1992) suggested that the sheer scale of these home ranges may make them indefensible. Forest-dwelling chimpanzees are well known to be highly xenophobic and respond aggressively to members from neighbouring communities (e.g., Goodall, 1986; Mitani, Watts & Amsler, 2010; Mitani & Watts, 2005; Watts & Mitani, 2001; Wilson & Wrangham, 2003). In contrast to forest dwelling chimpanzees, recent observations at Issa suggest that chimpanzees there may be more tolerant of neighbouring individuals, sharing large parts of their territory with non-community members (pers. obs). Similarly, as mentioned in the introduction, many species modulate their call rate in response to human disturbance—e.g., amphibians (Sun & Narins, 2005), birds (LaZerte, Otter & Slabbekoorn, 2017), and elephants (Soltis et al., 2014). For example, in Bili, northern DRC, chimpanzees call significantly less when they nest near human settlements (Hicks & Roessingh, 2010). The Issa study site lies not far from a human settlement and a road. Despite the fact that snare encounter rates increased with distance from the research station (Piel et al., 2015), the presence of cattle herding is still frequently observed in the area (AC, pers. obs.). We recommend that subsequent work evaluates whether Issa chimpanzees adjust their call rate near areas of increased human presence (e.g., road or cattle herders).