Orangutan males make increased use of social learning opportunities, when resource availability is high

Summary Humans’ colonization of diverse habitats relied on our ancestors' abilities to innovate and share innovations with others. While ecological impacts on innovations are well studied, their effect on social learning remains poorly understood. We examined how food availability affects social learning in migrant orangutan unflanged males, who may learn from local orangutans through peering (i.e., observational social learning). We analyzed 1,384 dyadic associations, including 360 peering events, among 46 wild Sumatran orangutan and 25 Bornean orangutan males, collected over 18 years. Migrants’ peering rates significantly increased with higher food availability and time spent in proximity to others. Furthermore, migrants in the more sociable Sumatran population exhibited significantly higher peering rates compared to the Borneans, suggesting intrinsic and/or developmental effects of food availability on social learning. These findings emphasize the importance of investigating ecological effects on social learning on the immediate, developmental, and intrinsic levels for our understanding of cultural evolution.


INTRODUCTION
Innovations, i.e., novel, learned behavior patterns acquired by an individual, 1,2 are the source of cultural knowledge.4][5] Social learning includes any learning that is influenced by the observation of or interaction with another individual. 6,7Culture, hence, at its most basic definition can be described as ''socially transmitted innovations,'' 1 and more specifically as the sum of ''all behaviors and knowledge that are acquired and passed on within and between generations through social learning.'' 80][11][12][13][14] Innovations played a crucial role during early humans' migration into unknown habitats, 15,16 yet they also came at costs such as time and energy investments, 17 as well as risks in the form of predation and poisoning. 18,19Horizontal social learning from individuals outside of the family/matrilineal/cultural unit, i.e., ''strangers,'' can lead to the acquisition of new skills and knowledge and thus might have allowed individuals to adapt to new ecological niches, without carrying the costs of individual exploration. 20,21Learning from strangers was thus likely a pivotal element in our species' migratory history.
While the effects of prevailing ecological conditions on innovation rates are well studied in non-human primates, much less is known about such effects on social learning and transmission.Broadly speaking, there are two scenarios on how ecological conditions can affect innovation rates in wild primates: on the one hand, ecological resource availability can modulate individuals' innovation probabilities via the availability of substrates and materials needed for certain innovations (e.g., hammer and anvil stones needed for tool use, ''opportunity hypothesis'' 22 ) and, on the other hand, ecological resource scarcity may force individuals to come up with new innovations that increase the exploitation of available resources (''necessity hypothesis'' 23,24 ).Innovation propensity has been further linked to dispersing individuals, who have a higher need to invent solutions to deal with new problems when leaving familiar natal areas, 14,25,26 but see Schnoell et al. 27

OPEN ACCESS
However, in contrast to individual learning, most forms of social learning require associations between individuals. 7,28Forms of social learning that allow for high-fidelity transmission of information, such as observational social learning (e.g., contextual imitation and production emulation) and interactive social learning, 4,29 require close spatial proximity within these associations. 30Furthermore, high-fidelity forms of social learning, i.e., being allowed to watch or to interact, may require even higher levels of social tolerance than just being close by. 31Because associations and in particular close associations bear significant costs and risks to individuals, ecological conditions may heavily affect social learning opportunities (see in the following).1][42][43][44] In contrast, mutual benefits of social learning between strangers are the increased likelihood of the exchange of new, valuable knowledge and skills that have critical fitness consequences, 45,46 the potential for cooperative relationships, 47 and the prospect of future mating opportunities. 48Accordingly, there is likely strong selective pressure on traits that allow individuals to balance the costs and benefits of associating and social learning from an unfamiliar individual.
Flexible shifting from social to individual learning and back, modulated through ecological resource availability and associated competition over resources, has been studied in the theoretical context through agent-based models 49 but also in data-based studies, e.g., in wild birds. 50,51These studies found that harsher environments overall favor less reliance on social learning in individuals, likely to avoid resource competition with conspecifics.However, it has also been argued that fluctuation in resource availability during human evolution might have favored the development of tolerant behaviors between nongroup individuals. 346][57] Since geographical distance positively correlates with ecological and cultural differences, 58 long-distance and long-term movement likely provide the most diverse experiences to individuals.Accordingly, migrating individuals are expected to show increased social learning propensity for optimal resource exploitation and adaptation.
5][66] Accordingly, orangutan sociality is directly linked to the productivity of the habitat they live in: whereas Bornean orangutans live in habitats characterized by low and fluctuating food availability and are semi-solitary, many populations of Sumatran orangutans (particularly the ones in northwest Sumatra) live in forests with higher productivity. 61Accordingly, Sumatran populations show higher levels of sociability, in that they spend more time with conspecifics. 67Both Sumatran and Bornean orangutans show individual-based fission-fusion social dynamics, 68 which allows researchers to track opportunities for social learning.In contrast to highly sociable species living in fission-fusion societies like chimpanzees or bonobos, 34,69 social learning might be easier to study in semi-solitary species since associations are more overseen and it is thus easier to track who is paying attention to whom (but see limitations part, on potential confounding effects of sociability on social learning patterns).The higher mean association frequency and spatially closer associations of orangutans in northwest Sumatra likely provide individuals with more social learning opportunities from a wider range of conspecifics compared to Bornean individuals.However, how they utilize these social learning opportunities in the wild remains to be investigated.Orangutans are also highly neophobic 19 and rely heavily on social learning for their skill and knowledge acquisition. 31,70,71This high reliance on social learning is thought to be the source of their vast and complex cultural repertoires. 58,72,73s means of social learning, orangutans use i) peering behavior, i.e., the attentive close-range watching of the activities of conspecifics 31,57,[74][75][76] and ii) food solicitations. 70,779][80][81][82] Both peering and food solicitations require close spatial proximity between association partners (i.e., social learning opportunities).Not all peering and food soliciting may lead to social learning, as it likely depends on how the peered-at or solicited information relates to the knowledge repertoire of the peering individual.However, peering and food solicitations open a window for knowledge transfer and thus represent likely utilizations of social learning opportunities.
The results of our recent study suggest that adult orangutan migrants use observational social learning in the form of peering to learn about local foods, including complex feeding skills, from knowledgeable resident orangutans. 57Generally, adult male orangutans disperse over large distances [83][84][85][86] which means that during the dispersal process, males likely encounter different types of habitats and host populations. 87Mo ¨rchen et al. 2023 57 found that migrant males' peering decreased with increasing residency (i.e., the duration spent in the area) and thus likely familiarity with the area.Furthermore, migrant males interacted significantly more with the peered-at item after the peering event, than before, providing further evidence that their peering is means to social learning. 57rangutans.We used a large longitudinal dataset on dyadic associations with and without peering events by 71 migrant unflanged males.These data were collected in the highly sociable population of Sumatran orangutan (Pongo abelii) at Suaq Balimbing and in the less sociable population of Bornean orangutan (Pongo pygmaeus wurmbii) at Tuanan.Both populations live in peat-swamp forests that differ in terms of food availability due to differences in forest productivity between Sumatran and Bornean forests (see earlier text).These differences in food availability are thought to have led over evolutionary time to differences in social tolerance between the two orangutan populations.Furthermore, on a smaller scale, in both populations, local food availability fluctuates within and throughout the year. 62,63

Rationale and predictions
Ecology may affect social learning on different levels: firstly, prevailing ecological conditions can affect social learning opportunities and the utilization thereof on the immediate proximate level.Thus, we predict that migrant male peering rates will increase with increasing food availability and with increasing time spent in close spatial proximity to peering targets (prediction 1).If the effect of food availability on peering is visible even after correcting for differences in close association (time the peerer and peering target spent within 2 m), we can conclude that food availability affects social learning beyond spatial proximity, likely via effects on social tolerance.
Secondly, ecological conditions experienced during ontogeny may affect the development of social learning propensity and thus result in differences in the tendency to use social learning during later life.Because of the difference in forest productivity between the two populations, males in West Sumatra that dispersed to Suaq experienced significantly higher food availability during development prior to dispersal compared to males at Tuanan, in Borneo.Similarly, the general differences in ecological conditions may affect the genetic fixation of social learning propensity over evolutionary time.Thus, between the populations, we predict that (as a result of developmental or genetic effects, or a combination thereof) migrant males in Suaq will show significantly higher peering rates than migrant males in Tuanan, even when controlling for the prevailing differences in food availability in the study sites, suggesting they make greater use of their social learning opportunities (prediction 2).

RESULTS
The comparison between the full and null models showed that as a whole the predictor variables of our model had a significant impact on migrants' peering rates (likelihood ratio test [LRT]: X 2 = 70.79,degrees of freedom [df] = 3, p < 0.001).In line with our predictions, we found that first, within populations, migrant orangutan males' peering rates significantly increased with higher prevailing food availability and the amount of time spent at close spatial distance to residents (Table 1; Figures 1 and 2).Second, we found that migrant orangutan males of the more sociable population of Suaq peer significantly more than the migrant orangutan males in Tuanan (Table 1; Figure 3).In the model we controlled for variables that previously have been shown to affect migrant orangutan males' peering rates, like the age-sex class of the peering target and the peerer and the number of months a male has been in the study area upon peering date 57 (Table 2).Furthermore, we controlled for dyad id and year, by including them as random effects (Table 2).

DISCUSSION
In our study, we looked at the effect of food availability on peering rates as a proxy of observational social learning in two orangutan populations and species that differ in levels of sociability 68,86 due to differences in forest productivity. 61In contrast to other studies, by investigating peering behavior as an indicator of social learning, we analyzed social learning by directly looking at the moment of potential knowledge transmission, rather than indirectly by studying its potential results. 73This allowed us to narrow down the mechanism underlying social transmission.The setup of our study allowed us first to differentiate between the immediate proximate effects of food availability on the utilization of social learning opportunities and second to assess the effects of experienced food availability on the development of social learning propensities, even though differentiating between developmental and genetic effects was not possible.
We found several lines of evidence that ecological conditions modulate observational social learning by dispersing orangutan unflanged males from resident orangutans: First, in line with our predictions, we found that migrant orangutan males' peering rates significantly increased with increasing local food availability.Further, the migrant orangutan males' peering rates also significantly increased with increasing time they spent in close spatial proximity to resident orangutans.These results complement the results of previous studies that found that increased food availability can lead to increased association time between individuals. 51,68,83,88Strikingly, our results also showed that the positive effect of food availability on the males' peering persisted even after correcting for the time the males spent in close association, suggesting that food availability might affect the migrants' social learning rates beyond spatial tolerance.This effect was observed in both populations.Second, we found that even when controlling for the known study sites' differences in food availability, migrant orangutan males of the more sociable population of Suaq show significantly higher peering rates than their counterparts in Tuanan, suggesting differences in social learning propensities between migrant males of the two species.All in all, our results suggest that, in addition to increased social learning opportunities provided by more frequent associations, 51,68,83,88 migrant orangutan males also make increased use of these opportunities when prevailing ecological conditions are favorable.
Similar results have been found in theoretical and empirical studies, 49,50 implicating that favorable environments foster social learning, while in harsher environments individuals will shift to individual learning, likely to avoid resource competition with conspecifics.Especially for individuals which do not share a genetically anchored interest to be tolerant to each other, prevailing ecological conditions are likely to affect the levels of social tolerance.This in turn will influence social learning opportunities, the utilization thereof, and ultimately social transmission of information.This means that when learning from unrelated individuals becomes critical, for example after migration, 5,56 ecological conditions likely heavily affect individuals' social learning via modulating the levels of social tolerance and spatial distance between the individuals involved.Mo ¨rchen et al. 2023 57 found that peering rates of migrant males were highest upon arrival in the new area and significantly decreased with increasing time spent in the area.This is consistent with the idea that the ''necessity'' for socially learning new knowledge is particularly elevated when migrants arrive in a new area.Here we found that, when statistically controlling for the time present in the area, migrants' peering rates increased in times with increasing food availability.The findings of Mo ¨rchen et al. 2023 57 and the current study showed that the necessity to acquire information, but also the ecological conditions that favor the opportunities to do so, affects social learning in migrant orangutan males.Thus migrant males are likely balancing a trade-off between socially learning new information from local orangutans, while simultaneously avoiding increased levels of competition with non-related conspecifics in a challenging phase of their lives. 89These results complement other studies that found contrasting evidence for effects of ecological conditions on rates of innovations. 22,23In combination, these findings strongly suggest that both ecological ''necessity'' and ''opportunity'' can modulate individuals' social and individual learning.As such, these studies underline the importance of the effects of ecological conditions on cultural transmission.We propose to expand the existing framework on how ecology, specifically food availability, affects social learning and knowledge transmission between adult, non-related individuals, by also incorporating the effects of ecological ''necessity'' and ''opportunity'' on the propensities for social learning, rather than solely focusing on actual individual learning occurrences. 31,70,71he differences in peering rates between the two populations persisted after controlling for differences in food availability.Similar differences in propensities for observational social learning have been found between immatures from the same study populations: immature orangutans in Suaq peer significantly more than their counterparts in Tuanan, 90 even after controlling for the known differences in association time between the species. 68,83,86,88These differences in social learning opportunities and propensities could be the result of either developmental effects, i.e., a consequence of growing up under different ecological conditions (see in the following), or of genetic differences between the species that have split approximately 674 kya ago, 91 or a combination of both. 92Shown are variance, standard deviation, sample size of control variables and random slopes of age-sex class of the peering target, dyad id, and year over the number of months the migrant peerer had been present in the study area (''Present Month Area'') at the day of associating with the peering target.

Suaq
Higher rates of individual learning, as evident in more frequent exploratory behavior, have previously been documented in the more sociable population at Suaq compared to the less sociable population at Tuanan. 90As a likely result of the higher individual and social learning performances in Suaq, individuals show larger sets of learned skills and overall more learning-intense diets (i.e., a diet that requires more intense pre-ingestive processing), compared to Tuanan. 90The results of these studies in combination with the results of the current study are in line with the developmental version of the cultural intelligence hypothesis.The cultural intelligence hypothesis states that growing up with increased opportunities for social learning will first lead to higher individual learning abilities, and second to higher social learning abilities, through the larger sets of affordances that individuals can draw from when being confronted with a novel problem. 93Indeed, in the same populations as studied here, immature orangutans' exploration rates (with exploration being a measure of individual learning ability) are correlated with past experienced sociability during the first years of their lives but were not affected by prevailing food availability. 94When housed under comparable conditions in zoos, where food availability is high and stable, Sumatran orangutans outperform their Bornean counterparts in problem-solving tasks. 92These differences in cognitive performance despite similar developmental and immediate conditions suggest that the differences in individual and social learning abilities between Sumatran and Bornean orangutans are at least to some degree genetically fixed (evolutionary version of the cultural intelligence hypothesis).
Our finding of frequent opportunities and high propensities for social learning in migrants may affect their cognitive and social learning abilities.Applying the predictions of the developmental effects of social learning opportunities on learning abilities to migrants, who experience frequent and diverse social learning opportunities when, e.g., traveling through unfamiliar habitats, one would predict that migrating individuals may show advanced learning capacities, in terms of social and individual learning abilities.In fact, when human brains are repeatedly exposed to new environments, lasting physical changes like an increase in hippocampal gray matter density can be measured.This is predicted to stand in connection to spatial cognition, 95 which is significantly increased in individuals that have higher demand for using the brain repeatedly in navigation contexts.Additionally, psychological changes can occur, like an increase in overall social tolerance and empathy, through witnessing other cultures and lifestyles when traveling. 96,97In humans this can lead to knowledge spillovers driven by migrants in multicultural groups, which can increase productivity and innovativeness and eventually stimulate long-lasting cultural diversification. 45,98,99iscovering effects of prevailing food availability on social tolerance and social learning in the least sociable great ape species that is most distantly related to humans points toward a deep evolutionary origin of ecological effects on social learning and learning propensities in the hominid lineage 100 and potential presence in other lineages.All great apes, but also early hominins, such as Australopithecus africanus and Paranthropus robustus, 101,102 show varying forms of sex-biased dispersal 48,86 and thus show patterns of migration during the lives of the individuals of the dispersing sex.We thus would expect evolutionary continuity in the ability of migrating individuals to use social learning from resident individuals to adapt to new habitats, when environmental conditions are favorable for mutual knowledge exchange.
Based on the results of our study, we hypothesize that possessing two specific traits provided a survival advantage to migrating individuals in our hominid lineage.First one is the ability to decide whether or not and how close to associate with an unfamiliar conspecific under the prevailing ecological conditions, given that associations bear potential risks and that these risks increase with increasing physical proximity. 36,40,41Favorable ecological conditions would allow then for close associations and subsequent high-fidelity social learning which could lead to valuable knowledge exchange. 45,46,99However, during times of scarcity, it may pay off more to avoid competition and disease transmission by increasing physical distance. 44Secondly, natural selection likely favored individuals with an increased social learning propensity, which likely improved their overall learning performance which then again allowed for more complex innovations to emerge, such as material technologies during hominin evolution. 10,103Thus, resource availability likely affects the developmental and genetic foundations underlying the propensity to attend to ecological information accessible through local residents, which over time played out to be a pivotal role in the history of human migration and the emergence of our unmatched culture. 10,12,104,105mitations of the study and future work Although the difference in migrant peering rates between the two studied populations persisted after controlling for differences in prevailing food availability, with this study, we are unable to differentiate between the developmental and genetic effects on the migrant orangutans' social learning propensities.This will remain challenging because, e.g., relocation and cross-fostering experiments that would allow to disentangle innate from learned behaviors during ontogeny, are not feasible in great apes like orangutans, for several reasons, including ethical and conservation concerns.
Further, we did not explore the potential of alternative functions of peering behavior, e.g., that peering might be used as a social tool for migrant males, to get access to local females.In a recent study, 106 it was reported that dispersing male vervet monkeys (Chlorocebus pygerythrus) are capable to adjust their levels of sociality to the sociality of the group they migrate to, underlining the importance of social integration after dispersal.Orangutans females are thought to have a certain degree of leverage over males. 107Therefore, peering behavior, which also resembles infantile behavior since it is most commonly observed in immatures, 31 might be one way for males to advertise nonthreatening behavior and thus be allowed close by.Whereas this and other potential functions do not negate the possibility of knowledge transfer and thus learning function of peering, it may explain remaining variation in our dataset (see for potential social functions of peering 108 ).Indeed, for all analyses shown in the paper, the confidence intervals are larger for the Suaq than the Tuanan population, which means that our model explains less of the variation in the dependent variable (peering probability) in Suaq than in Tuanan.This might correspond to the higher levels of sociability in the Suaq population (including larger and more frequent associations) which may mean that other factors that we did not include in our model (such as social factors) affect peering probabilities in the Suaq population.Especially social factors may have a more pronounced effect on peering at Suaq, the more sociable population.
Another limitation of our study is that at both populations the geographic origins of the migrant males are unknown, and thus it is not known if, and to what extent, their natal areas' habitat differs from the habitats in Suaq or Tuanan.Genetic studies suggest long-distance dispersal, 85 which makes it likely that the natal habitats differ from the host habitats.However, we cannot assess how much an individual male must learn upon arrival.Generally, Suaq is surrounded by different habitat types such as dryland hill forests, and further away higher-altitude forests, whereas Tuanan is surrounded by similar swamp forests and has no direct connection to different habitat types.Thus, the chances that a migrant male grew up in a very different habitat and therefore is highly unfamiliar with the locally available food items is likely higher for Suaq, than for Tuanan.This may also contribute to the larger confidence intervals for Suaq in our model.However, when investigating migrant orangutan males' peering behavior in the feeding context, migrant males from both populations peer significantly more at orangutans feeding on skill-intense food items, than at easy-to-process ones, and peered significantly more at rare food items (rarely consumed in the local orangutans' diet) than at common ones. 57Further, migrant peering rates would significantly decrease with increasing time spent in the area.These findings suggest overall, in both populations, migrating individuals need to learn about locally available foods after dispersal.
Additionally, in our study, we used food availability to assess the effects of ecology on social learning.However, many other ecological factors likely affect the behavior of individuals, such as the nutritional composition and monopolization potential of available food, rain, topography, or temperature, which were not targeted by this study.
Finally, the details of how migrant individuals make actual use of the knowledge, which they witnessed through peering behavior, remain to be investigated.Follow-up studies could tackle this question by assessing the development of the migrant males' individual behavioral repertoires, and the inclusion of new, learned behaviors with increasing time they spent in association with role models, and the overall time spent in the local area.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Figure 1 .
Figure 1.Effects of time spent in close spatial proximity on migrants' peering Migrant orangutan males' predicted peering probabilities over the time they spent within 2 m of peering targets.Shown are the model predictions (horizontal lines) and confidence intervals (vertical lines) with all other variables of the model kept constant at their mean.The occurrence of the actual migrant males' peering events is depicted by transparent smaller open circles with either 0 (no peering) or 1 (peering event) happening, for each daily dyadic association.

Figure 2 .
Figure 2. Effects of prevailing food availability on migrants' peeringMigrant orangutan males' predicted peering probabilities over prevailing habitat food availability, ranging from low (0) to high(20).Shown are the model predictions (horizontal lines) and confidence intervals (vertical lines) with all other variables of the model kept constant at their mean.The occurrence of the actual migrant males' peering events is depicted by the transparent smaller open circles with either 0 (no peering) or 1 (peering event) happening, for each daily dyadic association.For Suaq the actual FAI values ranged from 3.89 (minimum) to 17.40 (maximum), with a mean FAI value of 10.01.For Tuanan the actual FAI values ranged from 0.27 (minimum) to 10.05 (maximum), with a mean FAI value of 3.76.

Figure 3 .
Figure 3. Differences in peering probabilities between the two orangutan populationsMigrant orangutan males' peering probabilities at the more sociable Suaq population in Sumatra (Pongo abelii), and at the less sociable Tuanan population in Borneo (Pongo pygmaeus wurmbii), with all other variables of the model kept constant at their mean.Shown are the model predictions (horizontal lines) and confidence intervals (vertical lines) for each site.

Table 1 .
Effects on migrant orangutan males peering ratesThe results of the GLMM with Poisson family distribution, analyzing the effects of site, FAI, and time spent in close distance of 2 m to the peering target, on migrant males' peering behavior.Model estimates, standard errors (SEs), confidence intervals (CIs), Chi-square (c2), and degrees of freedom (df) are provided, based on N = 1,384 daily dyadic observations of migrants associating in distances of 0-50 m with targets of all age-sex classes.The conditional pseudo delta R2 was 0.85.Significant p values are marked in bold font.

Table 2 .
Results of the random effects of Model 1

TABLE
d RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability d EXPERIMENTAL MODEL AND SUBJECT DETAILS B Institutional permission information for higher vertebrate model d METHOD DETAILS B Data collection d QUANTIFICATION AND STATISTICAL ANALYSIS