Orangutan Cultures and the Evolution of Material Culture

See allHide authors and affiliations

Science  03 Jan 2003:
Vol. 299, Issue 5603, pp. 102-105
DOI: 10.1126/science.1078004


Geographic variation in some aspects of chimpanzee behavior has been interpreted as evidence for culture. Here we document similar geographic variation in orangutan behaviors. Moreover, as expected under a cultural interpretation, we find a correlation between geographic distance and cultural difference, a correlation between the abundance of opportunities for social learning and the size of the local cultural repertoire, and no effect of habitat on the content of culture. Hence, great-ape cultures exist, and may have done so for at least 14 million years.

Among the numerous definitions of culture, the idea that it is a system of socially transmitted behavior is particularly useful for comparative purposes (1). Because the creation of culture under experimental conditions illuminates neither the extent of culture among wild animals nor its content, documenting culture's existence in nature remains essential. Unfortunately, this task is not easy; even if a study lasts long enough to show that a newly observed variant is an innovation, it remains difficult to demonstrate convincingly that the variant's acquisition by others is guided by social transmission. However, recent work on chimpanzees has shown geographic patterns in many behavioral variants that are consistent with the operation of cultural processes. A variant is considered cultural if it is customary (shown by most or all relevant individuals) or habitual (shown by at least several relevant individuals) in at least one site but is absent in at least one other ecologically similar site (2, 3). Intraspecific genetic variation is almost certainly not responsible for these patterns (4).

Critics have stressed that the geographic approach may generate a type I error, spuriously leading us to conclude that cultures exist, when in fact unrecognized ecological differences between sites have produced within-population convergence and between-population divergence through individual learning (5–7). Hence, further tests are essential to increase our confidence in a cultural interpretation (8, 9).

Orangutans (Pongo pygmaeus) showing variation in two forms of tool use consistent with culture (10, 11) provide an opportunity for further testing. Here, we systematically apply the geographic approach to six different wild orangutan populations in Borneo and Sumatra (12) and test additional predictions derived from a cultural interpretation.

Table 1 lists three categories of geographic variants (13): (i) very likely cultural variants, which are behaviors present in at least one site at customary or habitual levels and absent elsewhere without clear ecological differences; (ii) likely cultural variants [as in (i) above] for which ecological explanations for absence, though unlikely, cannot be excluded; and (iii) rare variants that are unlikely to be maintained by social transmission. We shall refer to the first two as “putative cultural variants.”

Table 1

Geographic variation in orangutan behavior patterns. C, customary; H, habitual; R, rare; P, present with unknown frequency, probably rare; E, absent for ecological reasons; A, absent; ?, unknown.

View this table:

The list of putative cultural variants at the six sites (Fig. 1) contains 24 elements; an additional 12 local variants did not spread to customary or habitual level at any site. Data from additional sites would expand the list (14), as it does for chimpanzees (3). Of the putative cultural variants, 10 involve specialized feeding techniques, including tool use, and 6 are alternative forms of social signals, such as kiss-squeaks. As in chimpanzees (2,3), some variants may come close to reflecting shared meaning based on arbitrary symbols. In particular, the “raspberry” vocalizations, emitted in the final phase of nest building in Suaq Balimbing (11) and just before its start at Lower Kinabatangan (Table 1), seem to announce that the sender is bedding down for the night. Putatively cultural geographic variation in orangutans, therefore, is very similar to that in chimpanzees.

Figure 1

Locations of the six study sites with long-term orangutan data included in this study.

Human cultures show geographic patterning reflecting innovation and diffusion, and they incorporate more elements if they are open to influences from other societies (successful local innovation being comparatively rare), and within fairly broad limits show only a moderate effect of habitat on their content (15,16). If these generalizations also apply to orangutans and chimpanzees, it would increase our confidence in the cultural interpretation and the heuristic used here and elsewhere (2, 3, 17).

First, the innovation-and-diffusion hypothesis suggests that a behavioral variant often occurs at a site because it was brought there by animals dispersing from the site where it originated. We found the predicted correlation between geographic distance and cultural difference for the putative cultural variants (Fig. 2), i.e., those variants that spread well within at least one locality should therefore diffuse between localities as well. This relationship is unlikely to be an artifact because it is not found for the local variants that do not reach customary level at any site and therefore should diffuse poorly (Mantel tests: chimpanzees, P = 871/5039 = 0.17; orangutans, P = 434/719 = 0.60).

Figure 2

The relationship between geographic distance (in kilometers) and cultural difference (as the percentage of difference over all habitual and customary local variants) in orangutans [open circles (from Table 1)] and chimpanzees {closed circles [from (2)]}. For orangutans, the relationship is significant [r = +0.601; Mantel test: a more extreme result in 21 of 719 possible permutations (29);P = 0.029]; for chimpanzees, it is marked (r = +0.576; Mantel test: P = 334/5039 = 0.067).

Second, the size of the local cultural repertoire is the balance between the rates of origination (due to innovation or diffusion from elsewhere) and extinction (due to failed social transmission). Thus, greater size of the local repertoire may reflect (i) higher rates of origination, which in turn reflect greater need for innovation due to marginal ecological conditions (necessity) or more opportunities for playful exploration (free time) (18); or (ii) higher rates of retention due to better conditions for diffusion between sites and social transmission within sites, caused by frequent tolerant proximity (19, 20). Statistical power is insufficient to establish patterns if both influences are important, but if one predominates it should receive significant support. Across orangutan sites, no support exists for the necessity hypothesis (using percent of feeding time on tree cambium as an index of food scarcity and food-related local variants, cultural or not, as the response variable:r = –0.812, n = 6, P< 0.05, which is opposite to prediction), nor does statistical support exist for the free-time hypothesis (using total minutes in the day spent resting and total number of local variants, respectively:r = –0.910, P < 0.05, which is opposite to prediction). On the other hand, we did find support for the opportunities for social learning hypothesis as suggested by the pattern in humans: The number of customary and habitual variants in both orangutans (from Table 1) and chimpanzees [from table 1 in (2) ] is predicted by the percentage of time that nondependent animals spend in association, which is used to index opportunities for learning from individuals other than the mother (21). Moreover, this relationship is stronger when limited to the customary and habitual variants that are related to feeding (Fig. 3), as expected, because acquisition of these variants should, on average, depend more on close-range socially biased learning than does acquisition of other variants. Hence, the size of the cultural repertoire at a given site is best predicted by the opportunities for oblique and horizontal social transmission during development.

Figure 3

The relationship between time spent in association (30–32) and the local repertoire of customary (C) and habitual (H) variants related to feeding in both chimpanzees (closed circles) (2, 3) and orangutans (open circles). Association effect: F[1,7] = 31.87, P < 0.001; species effect: F[1,7] = 18.29, P < 0.01; interaction effect:F[1,7] = 11.32, P < 0.05. (No quantitative association estimate is available for Kutai orangutans, but the value would be low, supporting the trend).

Third, habitat may facilitate predictable individual learning and thus may facilitate convergent variant repertoires in separate localities, overriding the effects of the historical process of innovation and diffusion. No habitat effect is found, however, in orangutans, where pairs of sites with similar types of habitats do not have more similar variant repertoires than those with different types of habitats (sea-level floodplains versus mainly dryland forests) (Mantel test:P = 503/719 = 0.70; in chimpanzees, habitat effects are difficult to evaluate independently because they coincide with geographic differences).

These additional tests support a cultural interpretation of geographic variation in great-ape behavior and indicate fundamental similarities to human culture. However, because culture, as defined above, may be common among vertebrates (1, 17,22), finer distinctions are needed for meaningful evolutionary reconstruction. Differences in cultures should reflect variation in the complexity of innovation and the mechanisms of socially biased learning. Thus, cultural elements may be (i) labels, where food preferences or predator recognition are socially induced (5, 7, 23) and which generally involve little innovation; (ii) signals, involving socially transmitted arbitrary innovations as variants on displays, such as kiss-squeaks on leaves or song dialects (17); (iii) skills, involving rare innovations (including tool use), whose complexity depends on the nature of socially biased learning [which affects the degree of ratcheting (6)]; and (iv) symbols, probably derived from signal variants that became membership badges of the social unit or population (6, 15).

Species are expected to vary in the kinds of cultural elements they display. Only humans have all four kinds of cultural elements, whereas, unique among nonhuman primates, chimpanzees and orangutans show the first three (2, 3, 8–11), which are made possible by innovative abilities and sophisticated forms of socially biased learning (24–26). Human cultures, therefore, differ from those of great apes in having unambiguously symbolic elements (6, 27), far more complex skills, and far greater repertoire sizes, made possible by cognitive differences affecting innovation or observational learning (1, 5, 6). The presence in orangutans of humanlike skill (material) culture pushes back its origin in the hominoid lineage to about 14 million years ago, when the orangutan and African ape clades last shared a common ancestor (28), rather than to the last common ancestor of chimpanzees and humans.

Important tasks for the future include documenting the possible interdependence among these different kinds of cultural elements, identifying the conditions favoring their evolution, and assessing whether they all show the geographic and social correlates known for humans and demonstrated here for great apes.

  • * To whom correspondence should be addressed. E-mail: vschaik{at}

  • Present address: Orangutan Foundation International, 4201 Wilshire Boulevard, Suite 407, Los Angeles, CA 90010, USA.


View Abstract

Navigate This Article