Ecological variation in adult social play reveals a hidden cost of motherhood for wild chimpanzees

Summary:

Though common among humans, social play by adults is an uncommon occurrence in most animals, even between parents and offspring^1-3. The most common explanation for why adult play is so rare is that its function and benefits are largely limited to development, so that social play has little value later in life^3-6. Here we draw from 10 years of behavioral data collected by the Kibale Chimpanzee Project to consider an alternative hypothesis: that despite its benefits, adult play in nonhumans is ecologically constrained by energy shortage or time limitations. We further hypothesized that, since they may be the only available partners for their young offspring, mother chimpanzees pay greater costs of play than other adults. Our analysis of nearly 4000 adult play bouts revealed that adult chimpanzees played both among themselves and with immature partners. Social play was infrequent when diet quality was low but increased with the proportion of high-quality fruits in the diet. This suggests that adults engage in play facultatively when they have more energy and/or time to do so. However, when diet quality was low, most adult play fell to near zero whereas it persisted between mothers and offspring. Increased use of play by and between adult chimpanzees during periods of resource abundance suggests that play retains value as a social currency beyond development, but that its costs constrain its use. At the same time, when ecological conditions constrain opportunities for young to play, play by mothers fills a critical role to promote healthy offspring development.

Graphical Abstract

eTOC Blurb:

• Sabbi et al., report that, unlike other forms of adult play, mother chimpanzees continue playing with their offspring even when diet quality is poor. In this way, mothers might balance the energetic costs of grouping with the developmental needs of their young, and, like humans, become critical play partners to their offspring.

Results and Discussion

Humans share the playfulness of youth with many other species, but we are among very few who continue playing regularly into adulthood^1-3. The most prominent explanation for the decline of social play with age is that many of its functions relate to social and motor development^3-5, which benefit immature individuals specifically. However, playing with competent adult partners provides valuable learning opportunities for younger, inexperienced players that they might not be able to get from their peers^5-7. This is particularly important for humans, where play between parents and offspring is critical to infant and child development^2,7-8. Further, playing can also have direct benefits for adults, for instance, by reducing tension and reinforcing social bonds^3,9-11. Despite these benefits to peers or younger partners, adults of most nonhuman species only sporadically play with each other or even with their own offspring^3,11-12, opting for less intensive and more utilitarian affiliative behaviors such as grooming. This suggests that the costs of play usually override its benefits once development is complete. Chimpanzees (Pan troglodytes), one of our closest living relatives, are notable exceptions for which adult social play, including maternal play and play among adults, is regularly observed^13-15. The variation in adult play in this species provides a rare opportunity to evaluate whether social play is ecologically constrained.

We examined 3891 adult play bouts recorded in over a decade of observations of the Kanyawara community of chimpanzees living in Kibale National Park, Uganda (2010-2019, Table S1). Adults played with each other (10.8% of bouts), with unrelated immatures (51.4% of bouts), and with their own offspring (37.8% of bouts). Here, we test the prediction that play by adults decreases when dietary quality is low. Since play between mothers and offspring should be specifically beneficial for offspring development, we predicted that this type of play would be less constrained than play with other partners.

Adult social play is constrained by diet quality for wild chimpanzees

Social play incurs energetic and time costs^4,16-18, resulting in reduced play when food is less available across primates (e.g. Chlorocebus pygerythrus¹⁹, Saimiri sciureus²⁰, Alouatta palliata²¹) and other mammals (e.g. white-tailed deer, Odocoileus virginianus²², meerkats Suricata suricata²³, Long-Evans rats²⁴). As a fission-fusion species, chimpanzee sociality is strongly dependent on food availability^25-28, and the use of costly behaviors, such as hunting^29-30 and boundary patrols³¹, declines when fruit availability is low. We estimated diet quality by the proportion of non-fig fruits in the chimpanzee diet during a given study month. At Kanyawara, non-fig (drupe) fruits are generally preferred over other food types and have high nutrient value^28,32-33, resulting in higher energy balance³⁴, reduced feeding time³³, and larger social groupings²⁸ in the Kanyawara chimpanzees. Thus, high diet quality could facilitate play by increasing the time or energy available to spend on play or by increasing the availability of partners.

Adult social play decreased significantly with diet quality. We observed adult play at least once in every month of the study, and while it occurred on 38% of observation days overall, days with at least one adult play bout tended to include multiple play bouts. Therefore, we used GLMM³⁵ (lme4 package ver. 1.1.31) to examine two monthly measures of the frequency of adult play: the proportion of observation days that adult play was recorded (range: 2-97%) and the total number of adult play bouts (range: 1-241). In our analyses, we controlled for observation effort (number of observation hours) and the opportunity to play by including the average daily party size (including all adults and immature individuals) as there is potential for increased social interaction with more individuals. We found that adults played on fewer days (β=0.18+0.04, p<0.001, VIF=1.13-3.26, Table 1, Figure 1) and there were fewer adult play bouts (β=0.34±0.07, p<0.001, VIF=1.17-2.56, Table S2, Figure S2a) in months when chimpanzees consumed less non-fig fruit. This was also true when we restricted our analyses to examine only play between adults (n = 421 bouts on 256 days; Figures 1 and S2b, Table S1 and S2a). Thus, low diet quality constrained play among adults and their willingness to join younger players.

Table 1:

Model results for the relationship between monthly diet quality and days with adult social play

		Predictors of Interest			Interaction Terms		Control Variables
Model	Int.	Diet Quality	Adult Player				Party Size	Obs. Days	VIF
All Adult Play	2.45±0.03***	0.18±0.04***	-		Diet Quality x Party Size	−0.06±0.03*	0.31±0.04***	0.25±0.03***	1.13-3.26
Adults with Adults	0.12±0.13***	0.80±0.13***	-		Diet Quality x Party Size	−0.23±0.10*	0.48±0.16**	0.29±0.11**	1.03-2.74
Adults With Immatures 1	1.85±0.05***	0.19±0.04***	Sex of Adult (Males < Females)	−0.24±0.08**	Sex of Adult x Diet Quality	−0.05±0.07n.s	0.024±0.05***	0.38±0.04***	2.19-3.25
Adult females with immatures 1	1.68±0.06***	0.18±0.06**	Relationship (Non-mothers < Mothers)	−0.52±0.08***	Relationship x Diet Quality	0.17±0.05***	0.20±0.06*	0.40±0.05***	1.45-2.29

Results of each best-fit GLMM according to AIC. All models were fit using scaled numerical predictors. Values for model terms are grouped starting with targeted predictors: Diet quality (the avg. proportion of non-fig fruit in the diet) and Type of Adult Player (as appropriate for each model). All interactions that improved model fit were retained in final models. All models included average party size and number of observation days to account for opportunity for play to occur and observation effort. Models were fit with count data using a Poisson distribution except for¹ which indicates negative-binomial error distribution to account for zero-inflation. Raw data for days with play shown in Figures 1, 2, and S1. See Table S1 for descriptive statistics about each model dataset and Table S2 for model results using total play bouts and accompanying raw data in Figure S2.

Asterisks (*) denote significance (p <0.1°, p < 0.05*, p <0.01**, p<0.001***). Non-significant terms are denoted using (n.s.).

Figure 1: Adult play occurred more often as diet quality increased.

The proportion of days that at least one adult played (green, dashed line and circles) increased with the proportion of non-fig fruits in the diet among wild chimpanzees after controlling for opportunity through party size and observation days. This was also true when considering only play between two adults (yellow, solid lines and diamonds). Each point represents raw data from one observation month. See Table S1 for descriptive statistics of each dataset. Also see Table S2 and Figure S2 showing similar results using total bouts per month.

Mothers play with their offspring even when diet quality is low

As predicted, play was common between mothers and their offspring. Though this manner of play has been reported elsewhere for chimpanzees^14,26,36-37, play between mothers and offspring is usually rare in nonhuman animals, including primates (e.g. howler monkeys, Aloutta spp.^38-40, rhesus macaques, Macaca mulatta⁴¹), and even some of the other great apes (orangutans, Pongo spp^42-44, western lowland gorillas, G. g. gorilla⁴⁵). On the one hand, since mother chimpanzees are more constrained than nonmothers due to the high costs of lactation and the inefficiencies of juvenile travel^13,46-47, their social behavior should be particularly sensitive to diet quality. However, as mothers limit their associations with other chimpanzees when food availability is low^13,48-49, they also limit available play partners and opportunities for their offspring.

Both adult males and females played with immature partners less often in periods of lower diet quality, but adult females played more compared to males overall (days: β = −0.24±0.08, p = 0.004, VIF: 2.19-3.25, Figure S1, Table 1; bouts: β = −0.46±0.13, p < 0.001, VIF: 1.71-3.21, Figure S2, Table S2). Importantly this was not explained by underlying sex differences in the relationship between party size and diet quality (Figure S2A). Instead, it was driven largely by play between mothers and their own offspring which occurred more often than female play with non-offspring (β = 0.52±0.08, p < 0.001, Figure 2, Table 1 and S2). Further, mother-offspring play did not decrease as steeply in response to declining diet quality (β = 0.17±0.05, p < 0.001, VIF: 1.45-2.29). Thus, while all types of play were sensitive to environmental quality, increasing when high quality resources were abundant, mothers continued to play with their offspring even under the lowest quality conditions.

Figure 2: Mother-offspring play occurred on more days and declined less with decreasing diet quality.

Play between mothers and their offspring (maroon, dot-dashed line and circles) occurred more frequently and persisted at higher rates when diet quality was low. Shown in comparison to play between adult females and non-offspring partners (solid, pink line and diamonds). Each point represents raw data from one month of the study. See Table S1 for descriptive statistics of each dataset. Figure S1 displays sex differences in adults’ play with young partners. Also see Table S2 and Figure S2 showing similar results using total bouts per month.

The ecological context of adult social play

Like humans, and unlike most other wild mammals, wild chimpanzees continue to play regularly into adulthood both among themselves and with immature partners. Our results show that, like other aspects of chimpanzee sociality, social play by adults was constrained when diet quality was poor but increased with the proportion of high-quality fruits in the diet. While grooming is the more common social currency among adult chimpanzees^13,26, every adult in this community played at least once over the course of our study, and play was recorded on up to 97% of observation days when non-fig fruit consumption was at its highest. Thus, social play likely retains unique benefits for adults that are worth investing in whenever time and energy budgets allow.

Despite ecological constraints, mother chimpanzees engaged in play with their own offspring disproportionately, even when feeding conditions were poor. This suggests that mothers compensate for constraints on grouping by providing stable play partners for their offspring. Female chimpanzees’ avoidant social strategies may reduce feeding competition with other group members^13,49, but they also limit partner availability for dependent offspring. Due to non-seasonal breeding and long interbirth intervals, opportunities to play with peers can also be limited by community demography even when resources are abundant. Firstborn offspring, in particular, may be limited to only their mothers as play partners until their younger sibling is born up to 5-7 years later^50-51. Indeed, although maternal play has only rarely been documented in non-human animals, those species where it is commonly observed (e.g. bottle-nosed dolphins, Turisops truncates^52-53, Yucatecan spider monkeys, Ateles geoffroy⁵⁴) tend to share similar life history characteristics (e.g. extended development) and social systems (e.g. fission-fusion) to chimpanzees.

Conversely, orangutan mothers, another of the great apes, only rarely play with their offspring despite the fact that they are more often alone and have even longer birth spacing compared to chimpanzees^42-44. This may be consistent with ecological constraints, as orangutans exhibit extreme and unpredictable seasonal changes in resource availability^42,55-56. However, this pattern could also be explained by reduced benefits of social play as orangutans are largely solitary as adults and therefore inhabit a less complex social world compared to other great apes^{42, 55}. Notably, a comparative study of captive orangutan and chimpanzee mothers with ready access to consistent, nutrient-dense diets reported that both species played with their infants equally⁵⁷. In the wild, there is evidence that the sparse associations female orangutans do make facilitate play opportunities for their offspring with other juveniles while females forage⁵⁸.

The functions and benefits of play likely vary depending on the age of the partners, and different types of partners may confer different benefits^3-4,59-60. Human parents are the first play partners for their infants and play between parents and offspring not only supports physical and cognitive development, but also helps to build socio-cognitive and motor skills for playing with other partners^2,5-8,61. Bottle-nosed dolphins follow a similar pattern, playing with their mothers before expanding toward other partners, including peers^52-53. In sanctuary-housed chimpanzees, mother-reared individuals had longer play bouts with peers that were less likely to end in aggression compared to orphaned chimpanzees⁶². This suggests that, like humans and bottle-nosed dolphins, infant chimpanzees may gain unique developmental benefits from playing with their mothers, specifically, before branching out toward a wider network of play partners. Indeed, in our study, mother-offspring play was the most common form of adult social play and, in addition to maintaining play when diet was poor, mothers’ play with offspring increased with diet quality, even after controlling for the availability of other partners. Together, our results signal that, beyond simply filling the role of most available partner, playing with mothers specifically, may be more critical to sustaining the cognitive, social, and motor development of offspring than previously suspected in chimpanzees.

While adult chimpanzees did play with other adults, they played with young partners more frequently, even those unrelated to them. This contrasts with grooming, where adults are more likely to groom one another than younger individuals^13,63-65. While the benefits of playing with older, more experienced partners may be obvious for immature animals^2,5,7,60, it is less clear what unrelated adults gain from investing in young partners. One possibility is that playing with younger partners helps to build or reinforce social bonds that will last into adulthood. Enduring bonds like this have been observed in gorillas⁶⁶ and chimpanzees^64,67, but the role of play in these relationships is unknown.

Over ten years, wild adult chimpanzees’ play was constrained by diet quality, but occurred frequently when high-quality, non-fig fruits were eaten in high proportions. In contrast, mothers’ play with offspring was much less responsive to diet, remaining high even as non-fig fruits were scarce in the diet. Together, both results suggest that play does not simply lose its function and/or benefits during adulthood for chimpanzees. Rather, it is likely that social play retains specific functions and/or benefits for adult players that, in most cases, they can only afford to capitalize on under ideal dietary conditions. Whether adult chimpanzees play to build relationships^1,9-10,12,68 or support learning opportunities for younger group members^2,5-8,61, evaluations of the potential functions and evolution of play are incomplete without consideration of its costs. That mother chimpanzees bear these costs even when ecology constrains other aspects of their social behavior most, reveals a hidden cost of motherhood for this species where, like humans, mothers are critical play partners for their dependent young.

STAR Methods

Resource Availability:

Lead Contact:

Any requests for further information and resources should be directed to Kris H. Sabbi (ksabbi@fas.harvard.edu), the lead contact on this paper.

Materials Availability:

This study did not generate any new unique materials e.g. reagents.

Data and Code Availability:

All data used in this study have been deposited and are publicly available for download at the DOI listed in the resources table. Code for all tests was adapted from publicly available packages in R⁶⁹ and is available upon reasonable request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental Model and Subject Details:

Adult play data were extracted from long-term observational behavior records of the Kanyawara community of chimpanzees in Kibale National Park, Uganda from Jan. 2010 to Dec. 2019. This wild community of chimpanzees has been followed by researchers and local field assistants of the Kibale Chimpanzee Project since 1987 and all individuals are well-habituated and individually known⁷⁰. Following previous work, females were considered adults after the date of their first maximally tumescent genital swelling, and males were considered adults after the age of 15 y.⁷⁰. Over the course of the study, the average community size was 50 chimpanzees (range: 45-54) including an average of 16 adult females (range: 14-18), 10 adult males (range: 7-12), and 24 subadults (range: 20-27). This study includes 89 unique players, including 27 adult females and 19 adult males.

Method Details:

Observational Behavior Data Collection:

Detailed descriptions of all core data collection at KCP are reported in Emery Thompson et al.,⁷⁰. To summarize protocols used for this study: each day, teams of two to four highly trained local field assistants completed nest-to-nest follows of chimpanzees. While one team member (or more) recorded focal data on the activities of a single chimpanzee, another recorded group-level data including party membership, location, and group activity at 15-min intervals, as well as all-occurrence records of social interactions within the party⁷⁰. Thus, all occurrences of adult social play, defined as any that involved at least one adult party member, that occurred among chimpanzees in the focal party were recorded using descriptive long-hand notes. For each play bout, researchers recorded the time that the bout was first noticed, and the identity of any players involved as well as play behaviors that occurred during the bout. Over the course of the study, all chimpanzees were individually recognized, and field assistants were regularly tested for interobserver reliability. Data collection protocols were approved by the Institutional Animal Care and Use Committees at Harvard University, the University of New Mexico, and Tufts University, as well as the Uganda Wildlife Authority and the Ugandan National Council for Science and Technology.

Quantification and Statistical Analysis:

Data Extraction and Calculations:

We extracted all play bouts that included at least one adult player. While the goal of the research team is to follow chimpanzees from nest-to-nest, collecting 10-13 hours of data per day, sometimes this is not possible due to conditions in the field⁷⁰. For this reason, and because adult social play occurred more rarely compared to other social behaviors (1 bout per 8-9 observation hours on average across the study, but only on 38% of observation days), we excluded observation days that were unusually short (under 6 hours) to reduce bias that was due to observation effort.

Our outcome variable, diet quality, was based on scan samples of feeding recorded every 15 minutes throughout the observations. For each month, we omitted scans in which no feeding occurred and defined diet quality as the proportion of feeding scans in which chimpanzees fed on non-fig drupe fruits.

Due to chimpanzee’s flexible fission-fusion social system^13,26, the size and composition of the group of chimpanzees (i.e. the party) that is observed on a given day fluctuates and, though rare, some days follow only a single chimpanzee. Therefore, we also excluded days when fewer than two chimpanzees were observed because a play bout inherently requires multiple chimpanzees. This yielded 3891 play bouts for our analysis. As our different models addressed differing subsets of data (e.g., play between adults and immatures), we similarly adjusted the inclusion criteria to days when it was possible for that type of interaction to occur (Table S1). On days when multiple eparate parties of chimpanzees were followed, we considered each distinct party separately as its own “observation day” therefore multiple observation days may share the same calendar date. While it is possible that individual chimpanzees moved between parties during the day, individuals could only be included in one party at a time, by definition.

Statistical Analysis:

All statistical analyses were completed in R⁶⁹ (ver. 4.2.2) using RStudio⁷¹ (ver. 2022.07.02 Build 756). We used GLMM to examine the relationship between diet quality and two measures of adult play (the number of days that play occurred and the total number of bouts) per month. As the likelihood of social behavior occurring is correlated with the number of available partners, each model controlled for the average party size on the observation day to account for the number of potential partners. This included all adult and immature individuals. While overall community size varied moderately over the course of the study (see Study Site and Subjects), our preliminary analyses showed that party size (i.e., number of partners immediately available) was a better predictor of play rates than community size. Therefore, as these two predictors are tightly intercorrelated, we chose to include party size in our models to control for opportunity. We also included observation hours to control for differences in opportunity to record play across months.

We created four pairs of models to examine the relationship between diet quality and (1) adults’ overall participation in social play with any partner, (2) play between only adult partners, (3) adult males’ and females’ play with younger partners (under 10 y.o.), and (4) adult females’ play with dependent offspring versus non-offspring (under 10.y.o.). For all models, we initially created GLMMs with a Poisson error structure (lme4 package ver. 1.1.31³⁵) and used the performance package⁷⁰ (ver. 0.10.4) to check for zero inflation. When zero inflation was indicated, we adjusted to a negative binomial error structure.

We included interactions between diet quality and other predictors in all initial models and then removed this interaction if interactions did not improve model fit⁷³ (e.g. decrease AIC by >2). In subsequent similarly examined interactions we also used the check_collinearity function of the performance package⁷² to confirm that predictors were not collinear (VIF<5). Due to large differences in scale, we rescaled all continuous predictors before creating each model. In all models, we included calendar month nested within year as random effects to help capture temporal variation that might be due to longer term changes in ecological or social factors.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
Compiled data for each analysis included in the manuscript	This paper; Mendeley Data	Reserved DOI: 10.17632/j85c4b8djz.1

Highlights:

• Adult chimpanzees played more often with all partners when diet quality was high

• Unlike other dyads, only mother-offspring play persisted when diet quality was low

• Playing with offspring helps balance the costs of grouping with offspring needs

• Like humans, parent-offspring play may be crucial to development for chimpanzees