вход по аккаунту


Diet traditions in wild orangutans.

код для вставкиСкачать
Diet Traditions in Wild Orangutans
Meredith L. Bastian,1,2 Nicole Zweifel,3 Erin R. Vogel,4 Serge A. Wich,5,6
and Carel P. van Schaik1,3*
Department of Evolutionary Anthropology, Duke University, Durham, NC 27708-0383
Gunung Palung Orangutan Project, Ketapang 78851, Kalimantan Barat, Indonesia
Anthropologisches Institut & Museum, Universität Zürich, CH-8057 Zürich, Switzerland
Center for the Advanced Study of Hominid Paleobiology, Department of Anthropology,
George Washington University, Washington, DC 20052
Great Ape Trust of Iowa, Des Moines, IA 50320
Behavioural Biology, Utrecht University, 3508 Utrecht, The Netherlands
culture; food selection; geographic variation
This study explores diet differences
between two populations of wild Bornean orangutans
(Pongo pygmaeus wurmbii) to assess whether a signal of
social learning can be detected in the observed patterns.
The populations live in close proximity and in similar
habitats but are separated by a river barrier that is impassable to orangutans in the study region. We found a
60% between-site difference in diet at the level of plant
food items (plant species–organ combinations). We also
found that individuals at the same site were more likely
to eat the same food items than expected by chance.
These results suggest the presence of diet (food selection)
traditions. Detailed tests of three predictions of three
models of diet acquisition allowed us to reject a model
based on exclusive social learning but could not clearly
distinguish between the remaining two models: one positing individual exploration and learning of food item
selection and the other one positing preferential social
learning followed by individual fine tuning. We know
that maturing orangutans acquire their initial diet
through social learning and then supplement it by years
of low-level, individual sampling. We, therefore, conclude
that the preferential social learning model produces the
best fit to the geographic patterns observed in this study.
However, the very same taxa that socially acquire their
diets as infants and show evidence for innovation-based
traditions in the wild paradoxically may have diets that
are not easily distinguished from those acquired exclusively through individual learning. Am J Phys Anthropol
143:175–187, 2010. V 2010 Wiley-Liss, Inc.
Recent field studies of nonhuman primates have
revealed geographic patterns in population-specific foraging techniques and social signals that have been interpreted in terms of socially learned innovations (Whiten
et al., 1999; Boesch, 2003; Perry et al., 2003a,b; van
Schaik et al., 2003; but see Laland and Janik, 2006).
These field studies receive strong support from experimental laboratory studies demonstrating the presence of
observational forms of social learning needed to acquire
these innovations (e.g., Whiten et al., 2005) and from
observational field studies that are consistent with the
operation of such processes (Lonsdorf et al., 2004; Perry
and Ordoñez, 2006; Jaeggi et al., 2010). Taken together,
these studies have strongly suggested that culture, previously considered as uniquely human (e.g., Kroeber and
Kluckhohn, 1963; but see Hallowell, 1963), is built on
traditions found among nonhuman animals (Whiten and
van Schaik, 2007).
Experimental interspecific cross-fostering studies in
nature have suggested that simple forms of social learning play a role in how individuals acquire the list of food
items they select from among the many more potentially
available to them (e.g., Rowley and Chapman, 1986;
Slagsvold and Wiebe, 2007). These forms of social learning may be due to mere gregariousness or due to
enhancement, observational conditioning, or socially
induced affordance learning (sensu Whiten et al., 2004).
Such experiments suggest the presence of dietary traditions in a wide range of birds and mammals. Indeed, pri-
matologists have long speculated that geographic intraspecific variation in diet is traditional (Nishida et al.,
1983; Chapman and Fedigan, 1990; Chapman and
Chapman, 2002; Panger et al., 2002; Ganas et al., 2004;
Boesch et al., 2006; Russon et al., 2009). In addition,
computer simulations suggest that gregariousness alone
can produce variable food traditions across groups in
identical habitats (van der Post and Hogeweg, 2006,
2008). However, it is not clear to what extent this also
C 2010
Additional Supporting Information may be found in the online
version of this article.
Grant sponsors: A.H. Schultz Foundation, American Society of
Primatologists, Denver Zoological Society, Duke University Graduate School, L.S.B. Leakey Foundation, National Geographic Society,
and Netherlands Organization for Scientific Research (NWO); Grant
sponsor: National Science Foundation; Grant numbers: 0452995 and
0643122; Grant sponsor: Wenner-Gren Foundation for Anthropological Research; Grant number: 7330.
*Correspondence to: Carel P. van Schaik, Anthropologisches
Institut & Museum, Universität Zürich, Winterthurerstrasse 190,
CH-8057 Zürich, Switzerland. E-mail:
Received 13 May 2009; accepted 2 February 2010
DOI 10.1002/ajpa.21304
Published online 6 May 2010 in Wiley Online Library
TABLE 1. Expected patterns in diet between two sites according to different models of diet acquisition
Variation in diet
Between populations
Within populations
Exclusive individual learning
Heterogeneous, especially
for fallback foods
Exclusive social learninga
Supplemented social learninga
Homogeneous in theory,
heterogeneous in practice,
especially for fallback foods
Equally heterogeneous for all items
Heterogeneous, more so for fallback foods
Assuming individuals are equally connected within the social network.
happens under natural conditions, because social learning may merely serve to speed up the acquisition of species-specific diet choices or may get overruled by individual experience (Galef and Whiskin, 2001). The latter is
plausible because individuals may use a set of innate
rules that evolved because food selection is critical for
growth, survival, and reproduction (Stephens and Krebs,
1986). Thus, whether or not geographic variation in primate diets is traditional remains unclear, despite the
evidence for much more complex skill traditions in the
same set of species.
The goal of the present study is to document differences in diet (food selection) between two wild populations
of Bornean orangutans (Pongo pygmaeus wurmbii) and
to determine whether the observed patterns are compatible with the observed modes of diet acquisition. We
selected a pair of nearby study sites that have similar
habitats but are separated by a wide, impassable river,
allowing for social learning within sites but not between
them. Some orangutan studies have already suggested
that variation in particular skilled feeding techniques,
including the use of tools, reflects local traditions
(van Schaik and Knott, 2001; van Schaik et al., 2003;
Fox et al., 2004). Recent studies have also provided the
first strong evidence that social learning, including
observational learning, plays a critical role in the diet
acquisition process of wild infant orangutans (Jaeggi
et al., 2008, 2010).
We aim to distinguish between three models (Table 1).
Under exclusive individual learning (Model 1), maturing
individuals acquire their diets independently because
they do not have or ignore opportunities to learn socially.
Under exclusive social learning (Model 2), they acquire
their diets exclusively by adopting the food choices of
others through social learning as immatures and then
retain their preferences due to reluctance or inability to
explore individually (orangutans: Rijksen, 1978; rats:
Galef et al., 2008; chimpanzees: Hrubesch et al., 2009;
Jaeggi et al., 2010). Under supplemented social learning
(Model 3), they initially rely on social learning, as shown
for maturing orangutans (Jaeggi et al., 2010), but may
subsequently adjust diet composition through individual
experience (e.g., Galef and Whiskin, 2001).
These models make different predictions concerning
within-population diet variation. Assume, as in Model 1,
that individuals decide independently whether an unfamiliar item should be included in the diet by sampling a
small portion and then assessing its quality through sensory feedback cues (Dominy et al., 2001) or through the
interaction between taste and the consequences of food
ingestion (Provenza, 1996; Ueno, 2001). This strategy
should lead to within-site diet variation, because whenever a food item provides little or no distinct sensory
feedback (or even negative feedback but is nonetheless
edible), is eaten in small amounts along with many
American Journal of Physical Anthropology
others, has low nutritional value or has some combination of these properties, the animal cannot distinguish
its profitability (cf. MacArthur and Pianka, 1966) from
that of the many other food items eaten on a given day.
These problems have been demonstrated empirically (see
extensive discussion in van der Post and Hogeweg,
2006). In addition, any costs or risks to sampling should
produce dietary conservatism, leading animals not to
switch easily once they have made a choice. Individuals
will, therefore, develop within-group differences in diet
for those items for which they fail to accurately assess
their relative profitability. This is especially likely to
happen for two classes of items, which are not mutually
exclusive: i) low-quality food items that are eaten when
preferred foods are not available (fallback foods sensu
Marshall and Wrangham, 2007, i.e., items of relatively
poor nutritional quality in high abundance eaten particularly during periods when preferred foods are scarce),
because assessment of their profitability is hampered by
low quality and small meal sizes; ii) items that are hard
to obtain or require specialized processing techniques
[e.g., inner bark, spiny rotan (Palmae spp.)], because initial experience or chance may determine whether an
individual recognizes the item as profitable. In support
of this, Baritell et al. (2009) found that the least profitable food items in the diets of wild orangutans (i.e., inner
bark, leaves, and vegetative material), measured by
energy gain per unit time, were characterized by low
preference and shorter feeding bouts relative to more
profitable food items (i.e., fruits and flowers). They also
observed a significant positive correlation between profitability and preference for all food items in the diets of
wild orangutans. Model 1, therefore, predicts that the
average fallback food item will be eaten by a smaller
number of individuals than the average preferred item
within each site (Prediction 1).
Alternatively, under Model 2 (exclusive social learning), we expect within-site homogeneity in diet provided
that all individuals are directly or indirectly linked in a
social learning network. Model 2 also does not predict
any major difference in homogeneity between preferred
and fallback foods, once possible artifacts due to insufficient observation time are removed, so that there should
be no difference in the number of individuals eating a
preferred or fallback food item. Under Model 3 (supplemented social learning), individual learning may override previous social learning when the feedback signals
are clear. However, this is most likely for highly profitable food items that were, for whatever reason, not initially in the diet, and least likely for the fallback food
items for which profitability is low and difficult to estimate (and which therefore will largely remain outside
the diet). Thus, Model 3 would also predict that food
choices of individuals converge and, therefore, that
within a site, the diets should be largely homogeneous,
regardless of whether they are preferred or fallback
foods, although perhaps not as strictly as under Model 2.
In sum, then, within-site diet variation among individuals is expected to be greater for fallback than preferred
foods only if all food choices are based on individual
learning, as under Model 1 (Prediction 1) but not under
Model 2, whereas Model 3 predicts an intermediate pattern but one closer to Model 2.
Turning now to between-site diet differences, the models predict different patterns if the sites are ecologically
identical. Under Model 1 (exclusive individual learning),
the two sites should produce identical diets, whereas
under Models 2 and 3 (exclusive or supplemented social
learning, respectively), we expect differences between
the sites due to the social influences on food choice at
each site. However, because the sites are inevitably not
ecologically identical, there will be between-site diet differences under all models because differences in availability of food items may create differences in the optimum diet (MacArthur and Pianka, 1966), and the same
plant species may also occasionally have food items of
different chemical composition at different sites, and,
thus, different food properties that may affect food selection (Hladik, 1977; Glander, 1982). Model 1 predicts that
the same randomness that characterized individual variation in diet within sites will also play a role in diet differences between sites, and as a result, the choices of
top-ranked, preferred food items (mainly fruit) will show
strong between-site convergence, whereas those of fallback foods will not to the same extent. Thus, Model 1
predicts greater between-site diet differences for fallback
foods just as it does at the level of individuals within
sites. Model 2 (exclusive social learning) predicts clear
between-site differences, equal for all items; whereas,
Model 3 predicts that the between-site diet differences
will be most likely for the items whose profitability is
difficult to assess (i.e., fallback foods and items that
require processing innovations) for which local traditions
can more easily develop than for items for which it is
easy to assess profitability, such as preferred foods.
Thus, both Models 1 and 3 predict that between-site differences in the diet list are greatest for fallback foods,
unlike Model 2 (Prediction 2). We will test these predictions using estimates of between-site diet differences and
comparing differences for fallback and preferred food
We can also develop a third prediction. If items that
are available at both the sites are eaten only at one site,
then this suggests that they are not easily recognized as
profitable by orangutans (and thus are also more likely
to be fallback foods). Thus, if animals discover such
items independently, then the number of individuals eating them should be lower than the number eating the
more obvious items shared between sites, even if we
restrict ourselves to fallback foods (Model 1). On the
other hand, if there is a major role for social learning,
the number of individuals eating these unique fallback
items should be as large as the number eating the
shared food items (Model 2; Prediction 3), for the very
reason that these foods are hard to recognize individually and social information should therefore override
individual feedback. Under Model 3, we expect an intermediate pattern but closer to Model 2 because the items
independently acquired are likely to be acquired by most
individuals through positive feedback from these items.
In this prediction, one needs to correct for extreme differences in abundance of items, if the reason that some
Fig. 1. Systematic approach for comparison of dietary differences of groups or populations based on increasingly conservative criteria for inclusion of specific items in the comparison.
items are not obviously recognized as food and thus
unique to one site is that they are extremely rare.
Testing these predictions requires a clear definition of
fallback foods and a procedure for the estimation of the
dietary difference. We follow Marshall and Wrangham’s
(2007) definition of fallback foods as widely available,
low-quality resources (based on energy/g dry weight, see
Knott, 1998; Baritell et al., 2009) that are consumed in
inverse proportion to the availability of preferred foods
(see also Knott, 1998; Vogel et al., 2008). Thus, we identified fallback foods as a food item whose consumption is
negatively correlated with the availability of preferred
foods. Our procedure for estimating diets and diet differences is laid out in Figure 1. First, we established the
diets at each of the two sites by systematically collecting
a large set of feeding records but excluding items that
were only briefly sampled or eaten only once. We then
identified those items available at both the sites (as
revealed by phenology or other records) but eaten at
only one site. This list is the first, maximum, estimate of
the diet difference between the two populations. To
assess the validity of the between-site differences produced in this list, we also corrected for two possible artificial reasons for the absence of a food item from the diet
at a particular site: i) insufficient observation time and
ii) low abundance of the item at that site. We used these
corrections to produce a minimum estimate of the
between-site diet differences.
Study sites
Data were compiled from Tuanan (28 090 06.1"S; 1148
260 26.3"E) and Sungai Lading (028 150 49"S; 1148 220
43.1"E) located 12.6 km apart on either side of the
Kapuas River in Central Kalimantan, Indonesian Borneo
(see Fig. 2). The Kapuas River is 100–150 m wide in the
American Journal of Physical Anthropology
Fig. 2. Location of Tuanan and Sungai Lading study areas.
study region, making it impossible for orangutans in one
population to cross the river or to observe the feeding
behavior of orangutans in the other population. Tuanan
is a well-established research area (see van Schaik et al.,
2005) with a 950 ha trail system. Sungai Lading was
established especially for this study and covers 200 ha.
Both the sites primarily consist of swamp forest lying
on shallow peat of varying thickness up to 2 m disturbed
by intensive selective logging. Comparisons of vegetation
plots at the two sites show very similar diversity of tree
species (Sungai Lading: 86 species from 29 families, 56
genera, n 5 1,537 trees, x diameter at breast height
(dbh) 5 17 cm; Tuanan: 98 species from 34 families, 64
genera, n 5 1,612 trees, x dbh 5 16.4 cm) and high overlap at all levels (family: 83%, genus: 75%, species: 70%).
Sørenson’s index of similarity was well above the values
normally considered to represent high species similarity
between communities (Mueller-Dombois and Ellenberg,
1974) at all taxonomic levels (Table 2).
Fluctuations in temperature and rainfall and in the
phenology of leaves, flowers, and fruit are virtually identical, although the mean production rate of fruiting trees
(10 cm dbh) is significantly higher at Sungai Lading
(Bastian, 2008). Both forests have high orangutan densities (Tuanan: 4.5 indiv/km2; Sungai Lading: 7.7
indiv/km2) by Bornean standards (van Schaik et al.,
2005; Bastian, 2008), especially at Sungai Lading,
largely because the study area has recently become
hemmed in by burnt forest. The only major ecological
difference between the two sites is more regular flooding
at Sungai Lading, due to its location within the fresh
water flood zone of the Kapuas (Bastian, 2008).
Genetic differences between populations may cause
diet differences through differences in food preferences.
Large rivers serve as dispersal barriers for orangutans
and may, therefore, cause genetic differentiation of the
American Journal of Physical Anthropology
TABLE 2. Summary of botanical composition similarity between
Tuanan and Sungai Lading at each taxonomic level
Taxonomic level
Sørenson’s index
Elenberg’s index
populations living on opposite banks (Jalil et al., 2008).
However, orangutans may be able to cross the Kapuas
River closer to the headwaters, and gradual migration of
genes down its course may reduce genetic population differentiation between the opposite banks. Preliminary
results of genetic research indicate clear overlap in the
mtDNA haplotype spectrum (M. Krützen, personal communication). Thus, the Kapuas is more likely to act as a
barrier to cultural transmission than to gene flow.
Study subjects
A total of 39 and 16 independent or semi-independent,
individually recognized were encountered at Tuanan and
Sungai Lading, respectively. Only individuals followed at
least twice at each site for which we have a minimum of
3 h of feeding data and whose identities were confirmed
genetically (Tuanan: 36; Sungai Lading: 13) were used in
analyses. During the full 4.5 years of data collection in
Tuanan (July 2003–February 2007) and 20 months in
Sungai Lading (July 2005–February 2007), a total of
17,068 and 3,370 h of data were collected continuously
at each site, respectively (Supporting Information Tables
1 and 2). A total of 9,474 and 1,585 h of feeding data
were collected at Tuanan and Sungai Lading, respectively. Range overlap among adult females was between
54–77% at Tuanan and 72–96% at Sungai Lading (see
Bastian, 2008 for further details). Mean party size for
adult females, a standardized method for summarizing
the overall sociality of an individual orangutan (van
Schaik, 1999), was 1.13 for Tuanan and 1.03 for Sungai
Lading. Individuals at Tuanan spent 15.1% (independent
females: 12.4%; independent males: 24.6%) of the time in
associations with other individuals compared with the
2.82% (independent females: 1.73%; independent males:
5.70%) at Sungai Lading.
Sampling methods
We used standard focal animal sampling of orangutan
behavior using methods previously standardized across
orangutan sites (;
see also van Schaik, 1999), allowing direct comparison
between the sites. All food items eaten by orangutans
were carefully recorded. A food item is defined here as
the specific combination of food type and plant species
(e.g., the bark of Koompassia malaccensis) consumed by
an animal. The following food types were distinguished:
fruit (all stages of ripeness; including seeds), flowers,
bark (all references to bark refer to inner bark, i.e.,
phloem and cambium), leaves (young and mature), nongreen vegetable matter (e.g., pith, soft wood, and roots),
and insects (insects were not subdivided, as it is difficult
to distinguish different species or estimate their availability). Identification of plant species and categorization
of food items were extensively cross-checked to guarantee consistency. Our study was entirely observational
and complies with the Code of Ethics of the American
Association of Physical Anthropologists for the ethical
treatment of research subjects.
Botanical surveys
To determine botanical composition of the two sites,
data from a single large phenology plot (Tuanan: 2 ha;
Sungai Lading: 1.5 ha) and 20 smaller (0.5 ha total) vegetative plots were examined at each site. A phenology
plot was placed along transects in the center of each of
the study areas. All trees with 10 cm dbh within 5 m
from either side of the transect were monitored monthly
for their fruit, flower, and young leaf phenology. Twenty
small vegetative plots (5 3 5 m2 each) were placed randomly throughout each site at trail crossings to gain
additional information about the presence or absence of
nontree species (e.g., lianas, epiphytes, and ground vegetation) and trees \10 cm dbh or those too patchily distributed to enter the phenology plots.
Plants in the plots as well as food samples were regularly collected for identification. Morphospecies (distinct
plant species to which we may or may not have been
able to assign to a species as well as genus) were identified using local names and regularly updated based on
the information from the phenology plots and from the
other site. The scientific names of these morphospecies
were identified, mostly to the species level, using floras
and in consultation with botanists from the Wanariset
Herbarium (Balikpapan, East Kalimantan). Furthermore, samples collected in the field were cross-checked
with herbarium voucher specimens.
Diet composition
Diet composition was assessed for each site as the
total number of distinct food items (see Supporting Information Tables 3 and 4 for details of monthly diet compo-
sition per individual orangutan). The following criteria
were used to establish the diet at a site. First, we
excluded items eaten only once or for \6 min throughout
the duration of the entire study period to avoid the
inclusion of foods only briefly sampled (when longer feeding bouts were usually possible) and to minimize sampling artifacts due to incorrect species identification by
observers (Fig. 1a). Second, only items eaten by independent or semi-independent individuals (orangutans
regularly making and sleeping in night nests [50 m
from their mother) were included; thus, infants were
excluded. Third, items eaten exclusively by Sumi, a
female orangutan in the Tuanan population, were also
excluded, as she often foraged outside the transect system in open habitat with very different species composition, where no other individuals ranged. A further justification for this decision was that in no case was Sumi
the only Tuanan orangutan to eat an item that was also
eaten by orangutans at Sungai Lading.
Diet clustering
To assess the extent to which the consumption of
food items present at both sites was distributed randomly among individuals regardless of site affiliation,
or instead clustered according to site affiliation (which
would indicate site-dependent diet choices and thus a
social learning signal), we calculated the expected
number of items eaten at one site only as the
Qi ¼ ð1 Pi ÞNT þ ð1 Pi ÞNS
the probability that all individuals observed eating item
i happened to reside in a single site by chance, NT 5
total number of individuals observed at Tuanan, NS 5
total number of individuals observed at Sungai Lading,
and Pi 5 the observed proportion of individuals eating
item i for both sites combined (i.e., for NT 1 NS). In this
expression, (1 2 Pi)NT gives the probability that item i is
eaten by none of the individuals at Tuanan (and thus
entirely by animals at Sungai Lading only), (1 2 Pi)NS
gives the equivalent for Sungai Lading, and the combination, therefore, refers to the probability that an item
is eaten at only one site. The summed Qi over all items i
gives the expected number of items eaten at only one
site. Thus, the number was compared to the observed
number, and using the expected and observed number of
items eaten at both sites, we performed a v2 goodness-offit test. This test was done on the full data set, and on a
reduced set, where only individuals followed between
165 and 500 h were included. We used this criterion
because 1) it maximized the number of individuals we
could include in the analyses at each site and 2) provided a comparable number of individuals and observation hours at each site.
A cluster analysis was executed to determine
whether the differences in diet were at the level of the
individual or the population. Whether an item was
eaten or not eaten by each individual was entered as
binary data into a matrix in which columns represent
items (present at both the sites) and rows represent
individuals. Only individuals with overlapping and
comparable observation hours at both the sites
(between 165 and 500 h) were included to standardize
the comparison.
American Journal of Physical Anthropology
Preferred and fallback foods
To test Prediction 1, we needed a distinction between
preferred and fallback foods. We calculated Vanderploeg
and Scavia’s selectivity coefficient (Vanderploeg and
Scavia, 1979) for all food items (e.g., species and item of
that species) consumed in the diet at each site in a given
month. Specifically, the following equation was used:
r =p
Pi i
ðri =pi Þ
where ri is the proportions of food item (i) of a given species in the diet and pi is the relative availability of the
food item of a given species in the environment. For ri,
the percentage of time feeding on each item of each species was calculated for each month following the methods outlined in Harrison et al. (2009). For pi, the relative
abundance of each species was calculated by taking the
number of productive stems of species X per month for a
given species divided by the total number of producing
stems in the phenology plots for that month. This index
was calculated for each food item consumed in a given
month. To examine dietary preference for each food type
(e.g., fruit, bark, leaves, and flower), the average preference was calculated for each food type from these items
per month, and these monthly food type averages were
used in the analysis that compared preference among
food types. Note that if a food type was not eaten in a
given month but was available, it had a zero for preference (see Supporting Information Table 5 for details of
individual food item availability per month). Preferences
for both the Tuanan and Sungai Lading populations
were calculated using data collected during the study
period when data were collected at Sungai Lading.
Number of individuals eating some classes of
food items
Prediction 1 required testing the number of individuals eating preferred and fallback food items at each site.
It is obvious that an individual’s estimated diet will
increase with observation time, but it is not obvious how
this should bias the likelihood of including either preferred or fallback food items into the diet. However, to
examine this possible bias, we repeated tests of this prediction with a subset of the data set using the same
criteria used for the cluster analysis.
Maximum estimate of dietary difference
Prediction 2 required that we estimate the dietary difference between sites. Once a complete diet list was produced for both sites, we determined the presence of all
items at each site. Following our logic (see Fig. 1), food
that is present at both the sites but only eaten at one of
them represents the maximum possible dietary difference between the two populations. Presence was documented on both species and item level, since a certain
species could be highly abundant but never produce a
particular item during the observation period, so that
this item was not available at a particular site for possible consumption. We knew that a species was present
whenever it occurred either on the phenology or vegetative plots, when we observed the species elsewhere in
the study area or it was observed to be eaten by oranguAmerican Journal of Physical Anthropology
tans. Leaves, bark, and other vegetative items were considered available whenever the species was present.
Fruits and flowers were only considered available if they
were recorded as present during the monthly phenology
monitoring, if they had been recorded as eaten, or if we
had a photograph or sample for positive identification.
Minimum estimates of dietary difference
A certain food could be recorded as eaten at one site
but not at the other for artificial reasons. First, items
that were recorded as present but not eaten at a particular site could in fact be eaten at such a low frequency
that they were missed due to insufficient observation
time. Second, an item may be so rare in the habitat of
the site where it was not observed eaten that we missed
its consumption by orangutans or that orangutans did
not eat it because they encountered it too rarely.
To correct for the possible effects of differential follow
time, we used the Poisson distribution to calculate the
probability that an item should be eaten (where k 5 0
denotes an item was not eaten and k 1 denotes an
item was eaten at least once) at the site, say site A,
where it was not observed to be eaten,
Pðk 1Þ ¼ 1 Pðk ¼ 0Þ ¼ 1 em
We based the estimate of m on information from the
other site (where the item was eaten), say site B, where
mB 5 pB 3 NB, pB is the probability that an individual
eats item i per follow day at site B (i.e., proportion of follow days on which the item was eaten at least once) and
NB is the number of follow days for this animal (provided 6 h follow time was available for that day). We
then calculated the mean pB over all individuals at site
B that had a nonzero value of pB. Then, for site A, the
site where the item was not seen to be eaten, we calculated the mA 5 pA 3 NA, where NA is the actual follow
time at site A (as number of days with 6 follow hours)
and pA is the expected probability of use at site B [Eq.
(3)]. When the expected probability of observing that the
item at site A was eaten exceeded 0.90, we assumed that
the item could have been eaten at A but was not for reasons unrelated to limited follow time.
To correct for the possible effect of reduced abundance
of the food item at the site where it was not eaten, we
removed all the food items from the dietary difference
list for which the item had a [10-fold higher abundance
at the site where it was eaten. To allow for the inclusion
of those species consumed but not present in phenology
plots because they are rare, we assigned a value of 1 to
these species and also added 1 to the abundance of each
species within the plot (Tuanan 3.9% of food items; Sungai Lading 2.67% of food items). All species that were
unique to each site were encountered in the phenology
plots. On the basis of these sensitivity tests, we produced
three minimum estimates of the dietary difference
(Fig. 1e).
Statistical analysis
Frequency data were analyzed using v2 or Fisher’s
exact probability tests depending on sample sizes,
whereas continuous data were analyzed using MannWhitney U (MWU) and Wilcoxon signed–ranks tests
(WSRT) and Spearman’s rank order correlations. Oneway analysis of variance (ANOVA) or the nonparametric
equivalent (WSRT) was used to examine the differences
in dietary selectivity coefficients. If the homogeneity of
variance assumption was violated, we report a Welch
ANOVA statistic. If significance was detected, we used
Tukey-Kramer HSD to determine pairwise differences
among items.
The cluster analysis was computed using R 2.50. All
other statistics were computed with JMP-SAS 6.0.3-7.0
or StatView 5.0. The probability levels for all tests were
two-tailed and a was set at 0.05.
Identification of fallback foods
To test predictions concerning the causes of withinand between-site diet differences, it was necessary to
first identify fallback and preferred foods. Within
Tuanan, there was significant individual variation in
preference among the major food types (Welch ANOVA,
F4,98 5 58.87, P \ 0.0001). A post-hoc multiple means
comparison revealed that fruit was preferred over all
other food categories (Tukey-Kramer HSD, q* 5 2.73,
P \ 0.05). Flowers were more preferred than leaves,
inner bark, and nonleafy vegetable matter. Significant
variation in preference was also detected among the
major food types at Sungai Lading (Kruskal-Wallis: v2 5
55.68, df 5 4, P \ 0.0001). The post-hoc comparison
revealed that at Sungai Lading, as at Tuanan, fruit was
preferred over leaves and vegetable matter, although
there was no statistical difference among fruit, flowers,
and bark (Tukey-Kramer HSD, q 5 2.58, P \ 0.05),
which may owe to small sample sizes and high variance
in the bark and flower categories.
Orangutans at both sites were primarily frugivorous,
spending a majority of their total foraging time feeding
on fruits (overall monthly means: Tuanan: 71%; Sungai
Lading: 61%) whenever they were available. During
periods of fruit scarcity, orangutans at Tuanan fed primarily on flowers, whereas orangutans at Sungai Lading
fed mostly on bark, leaves, and other vegetative matter,
a difference explained by the greater availability of flowers at Tuanan. Because the consumption of bark, leaves,
and nonleafy vegetative matter correlated negatively
with the prevalence and consumption of preferred fruits
at both sites (Bastian, 2008; Vogel et al., 2008), all three
food item types were designated as fallback foods. We
left flowers unclassified, since their relative abundance
differed so dramatically between sites but ran analyses
both with and without flowers included as fallback foods,
considering only inner bark, leaves, and nonleafy vegetation to be true fallback foods, to test the robustness
General patterns
Despite high floristic similarity between Tuanan and
Sungai Lading, we detected a difference between the two
sites in orangutan diets. A total of 228 items from 140
plant species were eaten (for 6 min per item) at the
two sites combined (182 items eaten at Tuanan, 106 at
Sungai Lading). In addition, orangutans fed on 95 food
items (Tuanan: 65; Sungai Lading: 50) for \6 min total
per item. These items were considered sampled only
(because longer feeding times were generally possible)
and were not included in the diet list. From the 228 food
items eaten at either site, 150 items were present at
both sites and could potentially have been shared. Of
150 items, 90 were actually eaten at only one site or the
other, yielding a maximum estimate of the between-site
diet difference of 60% between Tuanan and Sungai
Lading at the level of food items. The large size of this
difference is consistent with a role for social learning in
diet selection.
A formal v2 test of the probability with which items
were observed versus expected to be eaten at one site
revealed that more individuals eating a particular item
happened to live at the same site than expected if the
inclusion of dietary items by individuals was independent of the site where they lived (v2 5 73.33, df 5 1, P \
0.0001). This result indicates that an individual orangutan’s food choice at a site is statistically dependent on
that of others, suggesting a possible role for social learning to explain the observed patterns. If this same analysis is repeated only for fallback foods, the result indicates that fallbacks are also highly clustered by site
(v2 5 41.11, df 5 1, P \ 0.0001).
One might argue that false absences due to insufficient observation time, rather than social learning, created larger between-site diet differences than expected
by chance. The total number of food items in the diet per
individual does indeed showed a positive relationship
with observed feeding time (Tuanan: Spearman’s q 5
0.923, P \ 0.0001, n 5 36; Sungai Lading: Spearman’s
q 5 0.912, P 5 0.0004, n 5 13). We, therefore, repeated
the v2 analysis including only individuals for whom we
had comparable and reasonably complete feeding observations, that is, we took all individuals, regardless of
site, for whom we had between 165 and 500 h of feeding
time. The outcome of this analysis, although based on a
much smaller data set, still showed a trend toward statistical significance (v2 5 2.83, df 5 1, P 5 0.0925).
Therefore, the between-site clustering of diets is not an
artifact of differences in observation intensity between
the two sites.
A cluster analysis of dietary overlap between dyads in
which both members had between 165 and 500 h of feeding observation time generated a dendrogram that split
individual dietary profiles into two major groups (see
Fig. 3). These groups sorted exactly according to population affiliation, illustrating the earlier result that the
greatest observed differences in diet are between populations, rather than due to the idiosyncratic foraging
behaviors of a few individuals within one or both sites.
Taken together, these results provide suggestive evidence of distinct diet traditions at the two sites.
Prediction 1: Within-site dietary variation
Exclusive individual learning (Model 1) predicts that
within each site, each individual preferred food item will
be consumed by a greater number of individual orangutans relative to each fallback item. Results of MWU tests
confirmed this prediction for Tuanan, regardless of
whether flowers were or were not considered as fallback
foods in the analysis (Table 3). For Sungai Lading, there
was a trend that preferred food items were consumed by
a greater number of individuals if flowers were included
in fallback category, but this trend did not exist when
flowers were not considered fallback foods. However, it is
important to note that the results for Sungai Lading,
while not reaching statistical significance, were in the
predicted direction. No change was detected in either the
pattern or statistical significance of these results after
correcting for observation time by using only data from
American Journal of Physical Anthropology
Fig. 3. The dendrogram resulting from a cluster analysis based on individual dietary profiles, indicating that individual orangutans cluster according to population affiliation; includes individuals from both sites with 10,000–30,000 min observation time during
overlapping time periods; TU, Tuanan individual; SGL, Sungai Lading individual. Terminal ends represent individuals and vertical
distances between individuals reflect the magnitude of differences in dietary repertoire. Clustering specifications: manhattan metric, complete linkage.
TABLE 3. Number of individuals within each site eating preferred and fallback foods
Preferred foods (fruit) vs. ‘‘true
(bark, leaves, nonleafy
vegetative matter)
Mann-Whitney U
n1 (# preferred foods)
n2 (# fallback foods)
Median (Mean) # individuals
each preferred food item
Median (Mean) # individuals
each fallback food item
P value
Preferred foods (fruit)
vs. all fallbacks
(including flowers)
Mann-Whitney U
n1 (# preferred foods)
n2 (# fallback foods)
Median (Mean) # individuals
each preferred food item
Median (Mean) # individuals
each fallback food item
P value
Sungai Lading
Before correcting
for observation
After correcting
for observation
Before correcting
for observation
After correcting
for observation
12.5 (14.72)
2 (2.32)
4 (4.77)
2 (2.33)
4 (8.14)
1 (1.12)
3.5 (4.13)
2 (2.04)
Preferred [ fallback
Preferred [ fallback
Preferred [ fallback
Preferred [ fallback
12.5 (14.72)
2 (2.32)
4 (4.77)
2 (2.33)
4 (7.20)
1 (1.09)
3 (3.97)
2 (1.92)
Preferred [ fallback
Preferred [ fallback
Preferred [ fallback
Preferred [ fallback
individuals within a similar range of follow hours (Table
3). This pattern of results is more favorable to Models 1
and perhaps 3 than Model 2.
Prediction 2: Between-site dietary variation
While Models 1 (exclusive individual learning) and 3
(preferential social learning) predict greater between-site
American Journal of Physical Anthropology
differences in fallback compared with preferred foods,
Model 2 (exclusive social learning) predicts no such difference. The between-site difference in dietary items
present at both sites but eaten at only one site was
based on the maximum estimate, as shown in Figure 4.
This difference was smaller for the preferred food type,
fruit (34.8% of 46 items), than for the fallback foods
(leaves: 68.8% of 48; nonleafy vegetable matter: 70.6%
into the diet. After excluding the 19 items with a [10fold higher abundance at the site where the items were
consumed, we still found that fallback items accounted
for a greater proportion of the diet differences than did
preferred items (v2 5 4.29, df 5 1, P 5 0.05), confirming
the first analysis.
Correcting for differences in observation time
Fig. 4. Percentage difference in dietary items present at
both sites but part of the diet at one site only (based on maximum dietary differences between sites) is heterogeneous (v2 5
20.36, df 5 4, P < 0.0001). Abbreviations: bk, inner bark; fl,
flowers; fr, fruit; lv, leaves; veg, non-leafy vegetable matter.
of 17; inner bark: 73.3% of 15) and for flowers (84.2% of
19). A v2 test indicated that this distribution was heterogeneous (v2 5 20.36, df 5 4, P \ 0.0001). A Fisher’s
exact probability test indicated that the three fallback
food types showed no variation in their between-site diet
difference (P 5 0.665), allowing us to combine them into
a single category. The difference between fruit and these
combined fallback foods was highly significant (v2 5
14.79, df 5 1, P 5 0.0002). This result favors Models 1
and 3 over Model 2. We submitted this result to two sensitivity tests, correcting for the rarity of food items and
insufficient observation time.
Correcting for variation in relative abundance of
food items
To examine the possible effect of food item abundance
on inclusion in the diet, we compared the abundance of
an item not eaten at a site with its abundance at the
site where it was eaten. Of the 90 food items present at
both sites but eaten at only one, 65 belonged to tree species present on the phenology plot at the site where they
were not observed eaten. However, for these items, no
significant differences were found between the overall
densities of items consumed versus those not consumed,
within either Sungai Lading (WSRT: v2 5 0.01, n1 5 53,
n2 5 58, p 5 0.919), or Tuanan (WSRT: v2 5 0.12, n1 5
98, n2 5 13, P 5 0.731), or for both sites combined
(WSRT: v2 5 0.755, n1 5 156, n2 5 66, P 5 0.385; median for eaten items 5 0.002, for items not eaten 5
0.001). These abundance effects were also assessed separately for each food type. No significant differences were
found between the abundances of consumed versus nonconsumed food species-item combinations for fruit,
leaves, bark, or vegetative items, although consumed
flowers (median density 5 0.001) were more common
than those not eaten (median \ 0.0001) across sites
(WSRT: v2 5 5.34, n1 5 19, n2 5 15, P 5 0.021). This
suggests that local differences in consumption of food
types do not reflect differences in item abundance, with
the possible exception of flowers. Hence, between-site differences in diet cannot be explained by variation in item
Nevertheless, we also examined the possibility that
extreme differences in abundance might affect inclusion
It might be argued that the maximum difference list
contains false zeroes. If the between-site difference in
diet composition were driven entirely by differences in
observation intensity, we would expect all 65 testable differences to disappear when we correct for observation
time. We found, however, that 36 items had a greater
than 90% probability of being observed eaten by at least
one individual at the site where they were not observed
eaten during the study period. Further, after removing
the likely false zeros due to insufficient observation
time, fallback foods still made up a statistically greater
proportion of dietary differences between sites than did
preferred items (v2 5 5.58, df 5 1, P 5 0.024). This suggests that correcting for differences in observation time
does not affect the conclusion of the original analysis of
the pattern in diet differences between sites.
Features of minimum between-site dietary
One way to distinguish between the power of each proposed model to account for the observed patterns
reported in this article is to make qualitative observations relating to specific food items on the most conservative minimum differences list. This list (Table 4) includes
only items with differential consumption between sites
that pass rigorous criteria, correcting for both differences
in observation time and in extreme differences in item
abundance. Out of the 15 items on this most conservative dietary differences list, we identified four-item differences that are especially difficult to explain with an
argument based on exclusive individual learning. All
four of the following food items (species-item combinations) were present at both sites but consumed during a
minimum of 23 different follow days at one site only, despite being more abundant at the site where they were
not consumed: ‘‘Manggis hutan’’ (two similar Garcinia
morphospecies, difficult to distinguish) leaves (eaten by
n 5 19 individuals), ‘‘Maruang’’ (Myristica lowiana) fruit
(eaten by n 5 23 individuals), and ‘‘Tarantang’’ (Campnosperma coriaceum) fruit (eaten by n 5 21 individuals)
were all consumed at Tuanan only and ‘‘Bengaris’’ (K.
malaccensis) bark was consumed only at Sungai Lading
(n 5 13 individuals). These examples are inconsistent
with Model 1 and are consistent with Models 2 and 3.
Prediction 3: Consumption of unique and
shared foods
Model 1 predicts that within each site, unique food
items (those eaten only at one site, despite being available at both) will be eaten by a smaller number of individuals than shared food items (eaten at both sites)
because they are harder to recognize as food. Results of
MWU tests confirmed this prediction, reaching statistical
significance for both sites when the analysis was carried
out using the full diet list (Table 5). Results remained
highly significant for Tuanan but, though in the same
direction, were no longer significant for Sungai Lading
American Journal of Physical Anthropology
TABLE 4. Dietary difference list correcting for observation time and extreme abundance differences
Local name
Item type
Site where
item eaten
# Days item
Pr (K 1)c
Licania splendens
Ilex cymosa
Pouteria cf. malaccensis
Pouteria cf. malaccensis
Garcinia spp.
Myristica lowiana
Antidesma cf. cuspidatum
Palaquium spp.
Campnosperma coriaceum
Koompassia malaccensis
Blumeodendron kurzii
Nephelium mangayi
Diospyros siamang
Xanthophyllum discolor
Pometia pinnata
Mangis hutan
Karandau putih
Pinding pandan
Sungai Lading
Sungai Lading
Sungai Lading
Sungai Lading
Sungai Lading
Sungai Lading
fr, fruit; lv, leaves; fl, flowers; bk, inner bark; veg, nonleafy vegetable matter.
Density ratio 5 density at site where item eaten/density at site where not eaten. Density calculated as # stems carrying item
X/total # stems on phenology plot.
Pr (K 1) 5 Poisson probability (%) of a particular food item being observed eaten by at least one individual at the site where
the item was not observed eaten during the study period.
TABLE 5. Number of individuals within each site eating shared and unique foods
Full diet
Mann-Whitney U
n1 (# shared foods)
n2 (# unique foods)
Median (mean) # individuals eating
each shared food item
Median (mean) # individuals eating
each unique food item
P value
Sungai Lading
Before correcting
for abundance
After correcting
for abundance
Before correcting
for abundance
After correcting
for abundance
14 (16.05)
14 (15.80)
4 (4.87)
4 (4.76)
4 (6.76)
5 (6.36)
2 (3.33)
3 (4.33)
Shared [ unique
Shared [ unique
Shared [ unique
Shared [ unique
‘‘True fallbacks’’
All fallbacks
‘‘True fallbacks’’
All fallbacks
9 (10.52)
4 (4.67)
4 (4.44)
4 (6.31)
2 (3.42)
2 (3.35)
Shared [ unique
Shared [ unique
Shared [ unique
Fallbacks only (after correcting for abundance)
Mann-Whitney U
n1 (# shared foods)
n2 (# unique foods)
Median (Mean) # individuals eating
9.5 (10.75)
each shared food item
Median (Mean) # individuals eating
4 (6.73)
each unique food item
P value
Shared [ unique
after excluding items with a greater than 10-fold higher
abundance at the site where the items were consumed.
When the analysis was performed only for fallback
items, after correcting for extreme site differences in
item abundance, no change was detected in either the
pattern or statistical significance of the results when
flowers were included in the fallback category for
Tuanan. However, when flowers were excluded from the
fallback food category, a strong trend remained in the
predicted direction (Table 5). For Sungai Lading, while
the results did not reach statistical significance, there
was a trend that shared items were consumed by more
individuals than unique items, regardless of whether
flowers were included in the fallback foods category.
These results strongly favor Models 1 and perhaps 3
over Model 2.
American Journal of Physical Anthropology
This study of intraspecific variation in diet is to our
knowledge the first to compare two physically separated
primate populations, living in similar habitats but
exchanging no migrants, to evaluate the possibility that
social learning is, at least partly, responsible for geographic variation in diet, i.e., that there are local diet
traditions. We also developed rigorous tests against the
alternative that between-site diet differences are due to
individual exploration and learning of diet choices.
Various patterns were examined to test predictions of
the scenarios presented in the introduction. First, we
found clear-cut and large differences in the diet of
the two sites, with 60% (90) of the 150 potentially
shared items actually eaten at only one site. Even after
excluding all those differences that could potentially be
due to local differences in item abundance or observation
time, a diet difference of 25% (16 of 65 testable differences) nonetheless remained. Another general finding was
that food items, including fallback foods, were not distributed independently of site affiliation among the individuals but instead showed significant clustering by site.
These two findings provide prima facie support for the
models invoking social learning, supporting the numerous
studies that have shown such differences and interpreted
them as diet traditions (references in introduction).
Tests of the three detailed predictions were not favorable to the model assuming exclusive social learning
(Model 2), but could not distinguish unambiguously
between models of individual diet acquisition (Model 1)
and social learning combined with individual exploration of diet post weaning (Model 3). We found that preferred items were on average eaten by a greater number of individuals at a site than were fallback items,
consistent with Models 1 (exclusive individual learning)
and 3 (supplemented social learning), that fallback food
items showed greater dietary differences between the
two sites than did preferred items, consistent with
Models 1 and 3 (preferred, but not exclusive social
learning), and that unique food items (in the diet at
only one site though available at both) were eaten by
fewer individuals on average than shared food items,
consistent with Models 1 and 3. Thus, these tests support Models 1 and 3, suggesting that both individual
and social learning play major roles in diet selection
among orangutans, and may explain major diet differences between sites.
Differentiating between Models 1 and 3 based on the
observed diet patterns alone is difficult. Their predictions differ only quantitatively, leading to a lack of resolution. However, there are some reasons to believe that
Model 3 actually provides the closer fit. First, we can
rely on behavioral observations of the actual modes of
learning to assess the importance of social learning.
Thus, we know that infant orangutans at Tuanan acquire a core food set from their mothers via vertical
social learning, much of which is very simple and classifiable as enhancement, such that the diet repertoires of
mother and infant are virtually identical at around
weaning (Massen, 2004; Dunkel, 2006; Jaeggi et al.,
2010). Other field studies of primates have suggested
very similar patterns (Tarnaud, 2004; Rapaport and
Brown, 2008). After weaning, young orangutans begin to
range more independently not only continuing to copy
the food choices of conspecifics with whom they regularly
associate (van Noordwijk and van Schaik, 2005) but also
individually sampling potential food items. This could
explain why the (independent, i.e., weaned) individuals
in this study were observed to sample 95 food items. Second, we know that innovated feeding techniques, including some based on tool use differ between orangutan
populations, leading to at least some between-site diet
differences that are entirely cultural (van Schaik and
Knott, 2001; van Schaik et al., 2003). We suspect that
the extreme cases found in this study (Table 4) require
either some innovative processing or, more plausibly, a
socially induced override of the initial sensory feedback
provided by these items. Finally, we may underestimate
the role of social learning in this study because the
orangutans at the two sites are not gregarious enough to
warrant local cultural uniformity (cf., Mitra Setia et al.
Taken together, these considerations strongly suggest
that the results, while consistent with both Models 1
and 3, are nonetheless most consistent with a combination of initial social learning followed by some a continuous, low-level individual exploration and evaluation (i.e.,
Model 3). However, having established that Model 3 provides a somewhat better fit than Model 1, it should also
be noted that their predictions are quite close. This
means that dietary traditions should be hard to detect in
permanently gregarious organisms, such as nonhuman
primates. The reason for this is individual sampling,
which results in the convergence of diets between sites.
Why does this convergence happen when it does not
seem to work in the simulation models of van der Post
and Hogeweg (2006)? We suspect that it is because
orangutans start out with a profitable diet inherited
form their mothers and then, being long lived, have
many years to try out food items at low rates. When
sampling of food items is low level, not exceeding once a
day, it may be possible for animals to develop some estimate of an item’s profitability, since it is the only thing
different compared to the diet the day before. This interpretation requires that sampling is low level; this prediction will be tested in subsequent work. An important
theoretical reason to expect some continued low-level exploration is that exclusive reliance on social learning
(Model 2) would inevitably lead to an erosion of local
diet breadth. Thus, some sampling is required, but the
older immatures and adults that do the sampling may
be better able to avoid the costliest mistakes, and by
adding items one by one to an existing diet they may be
better able to evaluate each item’s profitability. Sampling
also allows individuals to buffer changes in their habitat
by shifting or supplementing their initial core food set as
necessary (Russon, 2002).
This process will tend to produce the patterns
observed in this study; whereas most of the diet is
acquired socially, subsequent individual exploration and
evaluation produces great convergence among individuals and sites. Thus, initial social acquisition of the diet,
followed by individual fine-tuning (Model 3) ends up producing nearly the same geographic pattern in diet as
purely individual exploration and learning (Model 1), at
least in long-lived organisms that have time to sample
food items not in the diet.
Exceptions will remain for two kinds of food items.
First, whenever special processing is required to make
the item ingestible, those populations where the innovation did not take place or was not maintained through
social transmission would not have this food item in
their diet. This exception can account for the known
cases of tool-based exploitation of highly nutritious food
sources in some sites but not others (e.g., van Schaik
and Knott, 2001) but probably for several more as well
that are less striking because they do not involve tool
use. Second, when aversive sensory information from a
food item strongly suggests low profitability or even toxicity, when in fact they are profitable, this knowledge
may be discovered and transmitted socially in one population but not in another. This needs to be tested in
future work.
In conclusion, these results suggest that comparisons
of diet patterns, even if done with great care to remove
possible statistical artifacts and account for the inevitable ecological differences, will not yield strong evidence
for diet traditions, unlike the long-standing expectation
in the literature (e.g., Nishida et al., 1983) and our own
American Journal of Physical Anthropology
intuition based on evidence for geographic variation in
socially transmitted innovations (e.g., van Schaik et al.,
2003, 2006). The mismatch between expectation and
finding may arise because those animals most likely to
socially acquire food choices, intelligent, long-lived animals with long parent–offspring associations and thus
numerous opportunities for social learning, are the very
ones that have numerous opportunities for subsequent
adjustments of these choices through years of low-level
sampling, in a way that is neither too risky nor uninformative. Diet differences between populations are, therefore, only expected where novel food-processing techniques must be invented or where obvious negative feedback (e.g., bad taste) wrongly suggests unpalatability.
The authors thank the BOS Foundation for permission
to work at MAWAS; Universitas Nasional (UNAS) for
sponsoring the Tuanan and Sungai Lading projects; the
Director General of the Departamen Kehutanan
(PHKA), the Indonesian Institute of Sciences (LIPI), the
Direktorat Fasilitasi Organisasi Politik dan Kemasyarakatan, Departamen Dalam Negri, and the BKSDA
Palangkaraya for permission to work in Indonesia.
The authors thank the numerous field staff, students,
and assistants who have participated in the projects and
Pak Ambriansyah and Pak Arbainsyiah of the Waniriset
Herbarium, Pak Haji Achmad Ilas, Tono bin Guan, Idun
bin Guan, Awan bin Daut, Pak Ihing, Pak K. Odom, Pak
Nadi, Pak Linandi, Pak Rahmat, and Pak Kade Sidiasa
for their help with the identification of botanical specimens. The authors thank Maria van Noordwijk for managing the Tuanan orangutan database and Karin Isler,
Mark Harrison, Christine Drea, Ken Glander, Tracy
Kivell, Josh Linder, and Cheryl Knott for fruitful discussion. The authors thank Willie Smits for his generous assistance with the Tuanan field station and providing us
with the satellite image. This study was conducted
within the framework of a Memorandum of Understanding between UNAS and the Anthropological Institute
(AIM) of the University of Zürich. The authors especially
thank Tatang Mitra Setia and Sri Suci Utami Atmoko
for fruitful long-term collaboration.
Baritell KV, van Schaik CP, Vogel ER. 2009. Foraging efficiency
in wild orangutans (Pongo pygmeaus wurmbii) at the Tuanan
field station in Central Kalimantan, Indonesia. Am J Phys
Anthropol 44:82–83.
Bastian ML. 2008. Effects of a riverine dispersal barrier on cultural similarity in wild Bornean orangutans (Pongo pygmaeus
wurmbii), Ph.D. dissertation, Duke University, Durham, NC.
Boesch C. 2003. Is culture a golden barrier between human and
chimpanzee? Evol Anthropol 12:82–91.
Boesch C, Bi ZBG, Anderson D, Stahl D. 2006. Food choice in
Tai chimpanzees: are cultural differences present? In: Hohmann G, Robbins MM, Boesch C, editors. Ecology in apes and
other primates: ecological, physical and behavioral aspects.
New York: Cambridge University Press. p 183–201.
Chapman CA, Chapman LJ. 2002. Foraging challenges of red
colobus monkeys: influence of nutrients and secondary compounds. Comp Biochem Physiol 133:861–875.
Chapman CA, Fedigan LM. 1990. Dietary differences between
neighboring Cebus capucinus groups: local traditions, food
availability or responses to food profitability? Folia Primatol
(Basel) 54:177–186.
American Journal of Physical Anthropology
Dominy NJ, Lucas PW, Osorio D, Yamashita N. 2001. The sensory ecology of primate food perception. Evol Anthropol
Dunkel LP. 2006. Development of ecological competence in Bornean orangutans (Pongo pygmaeus), with special reference to
difficult-to-process food items, Unpublished Master Thesis,
University of Zurich, Switzerland.
Fox EA, van Schaik CP, Wright DN. 2004. Intra- and interpopulational differences in orangutan (Pongo pygmaeus) activity
and diet: implications for the invention of tool use. Am J Phys
Anthropol 125:162–174.
Galef BG, Dudley KE, Whiskin EE. 2008. Social learning of food
preferences in ‘‘dissatisfied’’ and ‘‘uncertain’’ Norway rats.
Anim Behav 75:631–637.
Galef BG, Whiskin EE. 2001. Interaction of social and individual learning in food preferences of Norway rats. Anim Behav
Ganas J, Ribbins MM, Nkurunungi J, Kaplin B, McNeilage A.
2004. Dietary variability of mountain gorillas in Bwindi
Impenetrable National Park, Uganda. Int J Primatol
Glander KE. 1982. The impact of plant secondary compounds on
primate feeding behavior. Yearb Phys Anthropol 25:1–18.
Hallowell AI. 1963. Personality, culture, and society in behavioral evolution. In: Koch S, editor. Psychology: a study of a science. New York: McGraw Hill. p 429–509.
Harrison ME, Vogel ER, Morrogh-Bernard HC, van Noordwijk
MA. 2009. Methods for calculating dietary composition compared:
a case study using orangutans. Am J Primatol 71:353–358.
Hladik CM. 1977. A comparative study of the feeding strategies
of two sympatric species of leaf monkeys: Presbytis senex and
Presbytis entellus. In: Clutton-Brock TH, editor. Primate ecology: studies of feeding and ranging behavior in lemurs, monkeys and apes. New York: Academic Press. p 324–353.
Hrubesch C, Preuschoft S, van Schaik CP. 2009. Skill mastery
inhibits adoption of observed alternative solutions among
chimpanzees (Pan troglodytes). Anim Cogn 12:209–216.
Jaeggi AV, Dunkel LP, van Noordwijk M, Wich SA, Sura AAL,
van Schaik CP. 2010. Social learning of diet and foraging
skills by wild immature Bornean orangutans: implications for
culture. Am J Primatol 72:62–71.
Jaeggi AV, van Noordwijk MA, van Schaik CP. 2008. Begging
for information: mother-offspring food sharing among wild
Bornean orangutans. Am J Primatol 70:1–9.
Jalil MF, Cable J, Sinyor J, Lackman-Ancrenaz I, Ancrenaz M,
Bruford MW, Goossens B. 2008. Riverine effects on mitochondrial structure of Bornean orangutans (Pongo pygmaeus) at
two spatial scales. Mol Ecol 17:2898–2909.
Knott CD. 1998. Changes in orangutan caloric intake, energy
balance, and ketones in response to fluctuating fruit availability. Int J Primatol 19:1061–1079.
Kroeber AL, Kluckhohn C. 1963. Culture: a critical review of
concepts and definitions. New York: Random House.
Laland KN, Janik VM. 2006. The animal cultures debate.
Trends Ecol Evol 21:542–547.
Lonsdorf EV, Eberly LE, Pusey AE. 2004. Sex differences in
learning in chimpanzees. Nature 428:715–716.
MacArthur RH, Pianka ER. 1966. An optimal use of a patchy
environment. Am Nat 100:603–609.
Marshall AJ, Wrangham RW. 2007. Evolutionary consequences
of fallback foods. Int J Primatol 28:1219–1235.
Massen JJM. 2004. The development of feeding competence in
immature Bornean orangutans (Pongo pygmaeus), Unpublished Masters Thesis, Utrecht University, Utrecht, NL.
Mitra Setia T, Delgado RA, Utami Atmoko SS, Singleton I, van
Schaik CP. 2009. Social organization and male-female social
relationships. In: Wich SA, Utami Atmoko SS, Mitra Setia T,
van Schaik CP, editors. Orangutans: geographic variation in
behavioral ecology and conservation. Oxford: Oxford University Press. p 245–253.
Mueller-Dombois D, Ellenberg H. 1974. Aims and methods of
vegetation ecology. New York: Plenum Press.
Nishida T, Wrangham RW, Goodall J, Uehara S. 1983. Local differences in plant-feeding habits of chimpanzees between the
Mahale Mountains and Gombe National Park, Tanzania. J
Hum Evol 12:467–480.
Panger MA, Perry S, Rose L, Gros-Louis J, Vogel E, Mackinnon
KC, Baker M. 2002. Cross-site differences in foraging behavior of white-faced capuchins (Cebus capucinus). Am J Phys
Anthropol 199:52–66.
Perry S, Baker M, Fedigan L, Gros-Louis J, MacKinnon KC,
Manson JH, Panger M, Pyle K, Rose L. 2003a. Social conventions in wild white-faced capuchin monkeys: evidence for traditions in a neotropical primate. Curr Anthropol 44:241–267.
Perry S, Ordoñez Jiménez JC. 2006. The effects of food size, rarity, and processing complexity on white-faced capuchins’ visual attention to foraging conspecifics. In: Hohmann G,
Robbins MM, Boesch C, editors. Feeding ecology in apes and
other primates. Cambridge: Cambridge University Press.
Perry S, Panger M, Rose LM, Baker M, Gros-Louis J, Jack K,
MacKinnon KC, Manson J, Fedigan L, Pyle K. 2003b. Traditions in wild white-faced capuchin monkeys. In: Fragaszy
DM, Perry S, editors. The biology of traditions: models and
evidence. Cambridge: Cambridge University Press. p 391–425.
Provenza FD. 1996. Acquired aversions as the basis for varied
diets of ruminants foraging on ragelands. J Anim Sci
Rapaport LG, Brown RB. 2008. Social influences on foraging
behavior in young nonhuman primates: learning what, where,
and how to eat. Evol Anthropol 17:189–201.
Rijksen HD. 1978. A field study on Sumatran orangutans.
Wageningen, The Netherlands: H. Veenman and Zonen.
Rowley I, Chapman G. 1986. Cross-fostering, imprinting and learning in two sympatric species of cockatoo. Behaviour 96:1–16.
Russon AE. 2002. Return of the native: cognition and site-specific expertise in orangutan rehabilitation. Int J Primatol
Russon AE, van Schaik CP, Kuncuro P, Ferisa A, Handayani
DP, van Noordwijk MA. 2009. Innovation and intelligence in
orangutans. In: Wich S, Utami-Atmoko S, Mitra Setia T, van
Schaik CP, editors. Orangutans: geographic variation in behavioral ecology and conservation. Oxford: Oxford University
Press. p 279–298.
Slagsvold T, Wiebe KL. 2007. Learning the ecological niche.
Proc R Soc B 274:19–23.
Stephens DW, Krebs JR. 1986. Foraging theory. Princeton, NJ:
Princeton University Press.
Tarnaud L. 2004. Ontogeny of feeding behavior of Eulemur fulvus in the dry forest of Mayotte. Int J Primatol 25:803–824.
Ueno Y. 2001. How do we eat? Hypothesis of foraging strategy
from the viewpoint of gustation in primates. In: Matsuzawa T,
editor. Primate origins of human cognition and behavior. Tokyo: Springer. p 104–111.
Vanderploeg HA, Scavia D. 1979. Calculation and use of selectivity coefficients of feeding: zooplankton grazing. Ecol Model
van der Post DJ, Hogeweg P. 2006. Resource distributions and
diet development by trial-and-error learning. Behav Ecol Sociobiol 61:65–80.
van der Post DJ, Hogeweg P. 2008. Diet traditions and cumulative cultural processes as side-effects of grouping. Anim
Behav 75:133–144.
van Noordwijk MA, van Schaik CP. 2005. Development of ecological competence in Sumatran orangutans. Am J Phys
Anthropol 127:79–94.
van Schaik CP. 1999. The socioecology of fission-fusion sociality
in orangutans. Primates 40:73–90.
van Schaik CP, Ancrenaz M, Borgen G, Galdikas B, Knott CD,
Singleton I, Suzuki A, Utami SS, Merrill M. 2003. Orangutan
cultures and the evolution of material culture. Science
van Schaik CP, Knott CD. 2001. Geographic variation in tool
use on Neesia fruits in orangutans. Am J Phys Anthropol
van Schaik CP, van Noordwijk MA, Wich SA. 2006. Innovation
in wild Bornean orangutans (Pongo pygmaeus wurmbii).
Behaviour 143:839–876.
van Schaik CP, Wich SA, Utami SS, Odom K. 2005. A simple alternative to line transects of nests for estimating orangutan
densities. Primates 46:249–254.
Vogel ER, van Woerden JT, Lucas PW, Utami Atmoko SS, van
Schaik CP, Dominy NJ. 2008. Functional ecology and evolution of hominoid molar enamel: Pan troglodytes schweinfurthii
and Pongo pygmeus wurmbii. J Hum Evol 55:60–74.
Whiten A, Goodall J, Mcgrew WC, Nishida T, Reynolds V,
Sugiyama Y, Tutin CEG, Wrangham RW, Boesch C. 1999. Cultures in chimpanzees. Nature 399:682–685.
Whiten A, Horner V, de Waal FBM. 2005. Conformity to cultural
norms of tool use in chimpanzees. Nature 437:737–740.
Whiten A, Horner V, Litchfield CA, Marshall-Pescini S. 2004.
How do apes ape? Learn Behav 32:36–52.
Whiten A, van Schaik CP. 2007. The evolution of animal ‘‘cultures’’ and social intelligence. Philos Trans R Soc Lond B
American Journal of Physical Anthropology
Без категории
Размер файла
322 Кб
diet, tradition, wild, orangutans
Пожаловаться на содержимое документа