orangutan locomotion

American Journal of Primatology 00:1–15 (2012)

RESEARCH ARTICLE

Forest Structure and Support Availability Influence Orangutan Locomotionin Sumatra and Borneo

KIRSTEN L. MANDUELL∗, MARK E. HARRISON, AND SUSANNAH K. S. THORPESchool of Biosciences, University of Birmingham, Edgbaston Birmingham, United Kingdom

The influence of habitat structure and support availability on support use is an important aspect ofunderstanding locomotor behavior in arboreal primates. We compared habitat structure and supportavailability in three orangutan study sites—two on Sumatra (Pongo abelii) in the dry-lowland forestof Ketambe and peat swamp forest of Suaq Balimbing, and one on Borneo (Pongo pygmaeus wurmbii)in the disturbed peat swamp forest of Sabangau—to better understand orangutan habitat use. Ouranalysis revealed vast differences in tree and liana density between the three sites. Sabangau hada much higher overall tree density, although both Sumatran sites had a higher density of largertrees. The two peat swamp forests were more similar to each other than to Ketambe, particularlywith regard to support availability. Ketambe had a wider variety of supports of different sizes andtypes, and a higher density of larger lianas than the two peat swamps. Orangutans in all three sitesdid not differ substantially in terms of their preferred supports, although Sumatran orangutans hada strong tendency to use lianas, not observed in Sabangau. Differences in observed frequencies oflocomotor behavior suggest the homogeneous structure of Sabangau limits the locomotor repertoire oforangutans, with high frequencies of fewer behaviors, whereas the wider range of supports in Ketambeappears to have facilitated a more varied locomotor repertoire. There were no differences among age-sex classes in the use of arboreal pathways in Suaq Balimbing, where orangutans selected larger treesthan were typically available. This was less apparent in Sabangau, where orangutans generally usedtrees in relation to their environmental abundance, reflecting the homogeneous nature of disturbedpeat swamp forest. These results demonstrate that forest architecture has an important influenceon orangutan locomotion, which may become increasingly important as the structure of orangutanhabitat continues to be altered through human disturbance. Am. J. Primatol. 00:1–15, 2012. C© 2012

Wiley Periodicals, Inc.

Key words: primate locomotion; forest structure; peat swamp forest; dry-lowland forest; supportuse

INTRODUCTIONCant [1992] identified four key habitat-related

problems that arboreal primates must deal withto resolve energetic challenges associated with ar-boreal locomotion: straightening the path of move-ment, negotiating large supports, crossing gaps be-tween trees, and increasing speed along the pathof movement. Gross canopy structure and the typesand diameters of supports available for weight bear-ing have considerable influence on the possible so-lutions primates can employ to resolve these prob-lems. A number of studies have demonstrated thatthe characteristics (e.g. type and diameter) of thesupports used for weight bearing have substantialinfluence on the expressed locomotor repertoire ofarboreal primates [e.g. Cant, 1987; Cartmill, 1985;Hunt, 1992; McGraw, 1996; Thorpe & Crompton,2005]. However, our understanding of the influenceof habitat structure, and support availability vs. sup-

port use, on primate locomotion remains remarkablyunderdeveloped [Dagosto & Yamashita, 1998; Mc-Graw, 1996; Warren, 1997; Youlatos et al., 2008], es-pecially given its importance in avoiding erroneousinferences about species differences in locomotionthat may actually result from animals inhabitingstructurally different environments [e.g. Cant, 1992;Thorpe & Crompton, 2009].

Previous studies of the positional behavior oforangutans (Pongo spp.) imply that, while the types

∗Correspondence to: Kirsten L Manduell, School of Biosciences,University of Birmingham, Edgbaston, Birmingham, B15 2TT,United Kingdom. E-mail: [email protected] 16 April 2012; revised 16 July 2012; revision accepted18 July 2012

DOI 10.1002/ajp.22072Published online in Wiley Online Library (wileyonlinelibrary.com).

C© 2012 Wiley Periodicals, Inc.

2 / Manduell et al.

of orangutan arboreal locomotion employed do notdiffer substantially between species (beyond greaterarboreality in Sumatran orangutans—P. abelii—that probably relates to the presence of a large,ground-dwelling predator [Cant, 1987], the rela-tive frequencies of positional behaviors do differ,with higher levels of pronograde (horizontal trunk)and compressive locomotion, and lower levels ofsuspensory locomotion and tree sway in Suma-tra compared to P. pygmaeus in Borneo [Manduellet al., 2011; Thorpe & Crompton, 2006, 2009]. Thesedifferences are probably related to habitat struc-ture [Manduell et al., in preparation] since Suma-tran orangutans exhibited distinct patterns of as-sociation between the type, diameter, and numberof supports used and locomotion [Thorpe & Cromp-ton, 2005; Thorpe et al., 2009]; whereas, in Borneo,primary activity type (feeding or traveling) had thestrongest influence on locomotion, and support typeand diameter were most strongly associated withthe height of the animal in the canopy and the age-sex class of the individual, respectively [Manduell etal., in preparation]. Manduell et al. [in preparation]proposed that these associations indicate that thevaried habitat structure of the primary, mixed dipte-rocarp forest study site in Sumatra allowed Suma-tran orangutans to use preferred locomotion/supportcombinations, whereas the homogeneous nature ofthe Bornean site studied (logged peat swamp for-est) led to Bornean orangutans being forced to usethose supports that were most prevalent in theenvironment.

Orangutans inhabit a number of different for-est types, including dry lowland and hill dipterocarpforest, peat swamp forest, freshwater swamp forest,alluvial forest, and heath (kerangas) forest [Hussonet al., 2009; Knott, 1998; Morrogh-Bernard et al.,2003; Rodman & Mitani, 1987]. These forest typesdiffer substantially in terms of tree species com-position, productivity and structure, and betweenthe same habitats on Sumatra and Borneo. Primaryproductivity is likely to be substantially lower inBorneo than in Sumatran forests because of thelatter’s younger, more fertile volcanic soils [Mar-shall et al., 2009; Wich et al., 2011]. Mixed dipte-rocarp forests are tall forests, with the top of thecanopy typically reaching 45 m [Whitmore, 1984].Alluvial forests are also species rich but have a lowercanopy than dipterocarp forests [Proctor et al., 1983].Undisturbed lowland peat swamp forests have lowertree species richness and a generally medium (35–40 m) to low (15–25 m) closed canopy layer, with“mixed” peat swamp forest, such as that found inthe orangutan research area in Sabangau, havinga closed canopy layer between 15 and 25 m [Pageet al., 1999]. Freshwater swamp forests have a var-ied structure that can range from low scrub withtrees 10 m in height, to a structure similar to mixedlowland forest [MacKinnon et al., 1996]. Heath (or

kerangas) forests are found on white sand soils thatare nutrient poor, highly acidic and free draining,and are frequently covered in a superficial peatlayer. Although the most productive heath forestscan resemble lowland dipterocarp forests, heath for-est structure generally tends to be characterizedby shorter, smaller trees with a low single-layeredcanopy and heath forests share numerous featureswith peat swamp forest including a large degree ofspecies overlap [MacKinnon et al., 1996]. The levelof past and contemporary human disturbance alsohas an important impact on forest structure, sincelogging often results in large gaps in the contin-uous upper canopy layer, which in turn increasesthe quantity of vegetation in the lower canopy, re-sulting in a more rugose and discontinuous forestcanopy [Vogel et al., 2009]. To date, detailed studiesof positional behavior have only been conducted onorangutans inhabiting lowland dry forest [Ketambe,Sumatra; Thorpe & Crompton, 2005, 2006, 2009;Thorpe et al., 2007, 2009] and mixed peat swampforest [Sabangau, Borneo; Manduell et al., 2011,in preparation].

Lianas are woody vines that are flexible in com-pression, yet strong in tension, and are an importantstructural component of tropical forests, typicallyconstituting around 25% of the woody stem densityand species diversity [Appanah et al., 1992; Gentry,1991]. In addition, lianas have an essential role inmany aspects of forest dynamics, including suppress-ing tree regeneration, increasing tree mortality, pro-viding an important food source for forest faunaand, crucially, providing pathways for arboreal an-imals that link trees together [Emmons & Gentry,1983; Grand, 1983]. It has been noted that there isa difference in orangutan liana use both within Bor-neo, and between Borneo and Sumatra [Cant, 1987;Manduell et al., in preparation; Thorpe & Cromp-ton, 2009]. Cant [1987] found a higher proportion ofliana use at Mentoko (P. pygmaeus morio), Borneo,compared to subsequent studies in other forests, andalso describes “curtains of lianas” in the forest. Useof lianas by orangutans in mixed dry forest in Suma-tra was also found to be high, especially when en-tering emergent feeding trees [Thorpe & Crompton,2005]; whereas Manduell et al. [in preparation] ob-served a very low frequency of liana use in Sabangau,Borneo. Peat swamp forests are likely to have lowerdensities of lianas than mixed dry forests, as lianadensity is associated with nutrient availability andpeat soils contain lower available nutrients [Whittenet al., 1987].

To understand the effect of habitat variation onorangutan locomotion, we quantified forest structureand support availability at the two study sites forwhich orangutan locomotion and support use arewell documented: Sabangau (disturbed peat swampforest, Central Kalimantan, Borneo, P. pygmaeuswurmbii) and Ketambe (mixed dry forest, Leuser

Am. J. Primatol.

Habitat Effects on Orangutan Locomotion / 3

Ecosystem, Sumatra, P. abelii) [Manduell et al.,2011; Thorpe & Crompton, 2005, 2006, 2009; Thorpeet al., 2007, 2009]. We also obtained new locomo-tor and habitat data for orangutans at Suaq Bal-imbing, an undisturbed peat swamp forest in theLeuser Ecosystem, Sumatra [van Schaik, 1999]. Thisallowed for comparison of the relationship betweenhabitat structure and locomotion within a singlespecies (P. abelii), and comparison of differenceswithin a single habitat type (peat swamp forest)between species, helping us to tease apart the rel-ative influence of species vs. habitat on orangutanlocomotion.

Within each study site, we also investigatedwhether support use mirrored support availability orwhether supports were selected because of propertiesthat made them preferable for locomotion. In lightof previous primate studies of orangutan locomo-tor behavior [Manduell et al., 2011, in preparation;Thorpe & Crompton, 2005, 2009; Thorpe et al., 2009],we hypothesize that (1) orangutans in Sumatra(Ketambe and Suaq Balimbing) will show strongerpreference/avoidance strategies given the moreheterogeneous nature of the forest, whereas (2)orangutans in disturbed peat swamp forest (Saban-gau), which is likely to be more homogeneous, willbe less selective over their substrate use.

Orangutans are well known for their use of arbo-real pathways [Cant, 1992; Mackinnon, 1974; Thorpe& Crompton, 2005], which might explain the lim-ited influence of age-sex class on locomotor behavior[Thorpe & Crompton, 2005]. However, how travelroutes are selected and whether these strategies dif-fer between species or as a consequence of habitatvariation is largely unknown. Numerous features ofthe canopy could potentially influence the selectionof travel routes; for example, the connectivity of treecrowns will affect the size and type of gaps betweentrees, which would be expected to influence how anorangutan might traverse these gaps. To investigatethis, we compared the structural features of treesused during travel in Suaq Balimbing and Saban-gau, in order to understand intersite differences be-tween the two peat swamp study sites and intrasitedifferences between age-sex classes. Given the useof arboreal pathways by orangutans, we further hy-pothesize that there will be: (3) little variation intravel trees used between the age-sex classes; (4)differences in arboreal pathways between Suaq Bal-imbing and Sabangau, as a result of differences inforest structure between the sites; (5) less variationin travel trees used in the Sabangau compared tothose typically available in the environment, giventhe apparently more homogeneous nature of this for-est; whereas (6) in Suaq Balimbing, orangutans willselect for trees that have attributes that are likely toreduce vertical displacement (e.g. greater connectiv-ity between crowns), given the seemingly more openand discontinuous canopy.

METHODSStudy Sites

Research took place in three study sites, two ofwhich are located in Sumatra and one on Borneo.This research adhered to the American Society ofPrimatologists principles for the ethical treatmentof primates and permission to conduct this researchwas provided by the University of Birmingham’s an-imal care committee and adhered to the legal re-quirements of Indonesia. Ketambe is situated inthe northeast of the Leuser Ecosystem, Sumatra(3◦ 41′N, 97◦ 39′E), and mainly comprises primarymixed dry lowland rainforest. Forest structure wasquantified during the period May 2010 to July 2010.Suaq Balimbing is situated in the western coastalplain of the Leuser Ecosystem, Sumatra (3◦42′N,97◦ 26′E). The site mainly comprises peat swampforest, in which the peat layer increases in thick-ness with increasing distance away from the river[Wich et al., 2009]. All data were obtained here dur-ing the period August 2010 to April 2011. The Nat-ural Laboratory for the Study of Peat Swamp For-est research station, in Sabangau, southern Borneo(2◦ 03′S, 113◦ 54′E) is also peat swamp forest. Thestudy site was selectively logged under concessionfrom 1966 to 1996, and then illegally logged from1996 to 2004. Data collection for this study was un-dertaken between March 2007 and September 2007,and April 2009 and January 2010. An overview ofthe three study sites is provided in Table I.

Habitat SurveyTo characterize forest structure, 20 100-m-long

transects were established in each of the sites. Thelocation and orientation of each transect were ran-domly selected within the orangutan study grid ateach site, were sufficiently far apart (≥25 m) toensure that trees were never sampled twice, andwere oriented so that no two transects intersected.Sample points were taken at 25 m intervals alongeach transect (five points per transect) using thepoint-center-quarter method [PCQM, Cottam & Cur-tis, 1956], which has been widely used in previ-ous primate studies [e.g. Balko & Underwood, 2005;Cannon & Leighton, 1994; Marsh & Loiselle, 2003;Teelen, 2007; Villard et al., 1995]. The distance to thenearest tree from the point center (DTPC) was mea-sured in each quadrant, as defined by the transect di-rection and its perpendicular. Within each quadrant,the diameter at breast height (DBH, 1.3 m above theground) of the nearest tree for each of the diame-ter classes was measured, allowing quantification oftree density for each size class.

To quantify support availability, 40 points wererandomly selected from the 100 points along thePCQM transects. In each quadrant, for the nearesttree ≥10 cm DBH and ≥4 cm DBH (n = 160 trees at

Am. J. Primatol.

4 / Manduell et al.

TABLE I. Overview of Study Sites

Sabangau Catchment Suaq Balimbing Ketambe

Location Central Kalimantan, Borneo Gunung Leuser, Aceh, Sumatra Gunung Leuser, Aceh, SumatraSpecies Pongo pygmaeus wurmbii Pongo abelii Pongo abeliiOrangutan

standardizeddensity (/km2)a

2.35 ind/km2 7.44 ind/km2 3.24 ind/km2

Forest type Peat swamp forest Peat swamp forest Mixed dry forestCanopy heightb,c 15–25 m 15–25 m 35–40 mDisturbance Logged Unlogged UnloggedProductivity Low High HighRainfalld 2,790 mm 3,400 mm 3288 mmMean elevationa,d 10 masl 10 masl 320 masl upwardsaHusson et al. [2009].bPage et al. [1999].cWhitmore [1984]dWich et al. [2009].

each site for each size class), we measured DTPC andDBH, plus tree height and bole height (height to thefirst main bough) using a clinometer. For these trees,we also measured crown diameter using a tape mea-sure along the ground; crown shape (narrow cone,wide cone, umbrella, monopodial; Fig. 1); crown con-nectivity [a four-point scale was used to indicate theposition of the crown relative to neighboring crowns,in terms of contact with or proportion overlappingneighboring crowns: 1 = 0–25%, 2 = 26–50%; 3 =51–75%; 4 = 76–100%, after Whitten, 1982]; andcrown volume, calculated using crown height anddiameter incorporating correction factors specific tothe crown shape [Coder, 2000]. Support availabilitywas quantified by counting the number of boughsand branches for all classes greater than 2 cm di-ameter within the crown. The number of smallerbranches (≤2 cm diameter) was difficult to countaccurately and therefore a semilogarithmic scalewas used to estimate the number of these supports(Table II). The number and size of lianas present ineach tree crown were counted precisely. Forest pro-file diagrams presented in Figure 1 provide an im-pression of the overall structure of each of the sitesused in this study.

Support UseOrangutan positional behavior observations in

Ketambe were made by a single observer (SKT),and all observations in Suaq Balimbing and Saban-gau were made by a single observer (KLM) duringnest-to-nest follows of wild orangutans, followingthe same methods. Instantaneous samples on the1-min mark were used to obtain detailed data of sup-port use during locomotion in nest-to-nest follows ofwild orangutans. Data collected at each sample pointincluded the support type (branch, bough, trunk,liana); support diameter (<4 cm; 4–10 cm, 10–20 cm,>20 cm); and the number of weight-bearing supports

(1, 2, 3, 4, >4). These methods have been describedin detail elsewhere [Manduell et al., 2011; Thorpe& Crompton, 2005]. For observations of support useduring locomotion; 1,762 observations were obtainedfrom orangutans in Suaq Balimbing (this study) and2,037 in Sabangau [Manduell et al., 2011]; and 1,783in Ketambe [Thorpe & Crompton, 2005].

Travel TreesDuring nest-to-nest follows of wild orangutans

in Suaq Balimbing and Sabangau, the trees in whichfocal animals traveled were marked with ribbon andtheir GPS positions taken, in order that they couldbe returned to at a later date. These trees were thenrelocated and the same measurements taken as de-scribed in the previous section for the random sam-ple. In addition, we measured the trunk-to-trunkdistance and the gap distance or degree of crownoverlap (measured as projected to the ground, us-ing a tape measure in the direction of travel to thenext tree). Unfortunately, it was not possible to ob-tain these data for Ketambe, as the locomotor studywas carried out a number of years earlier than thecurrent work [Thorpe & Crompton, 2005]. Thesevariables were compared between age-sex classes[flanged males, sexually active females, nonsexuallyactive females, and unflanged males, as defined byMorrogh-Bernard et al., 2009] to see if there wereany differences in the structural features of treesused during locomotion; between the trees used forlocomotion and those from the random sample (i.e.support use vs. environmental availability); and be-tween sites.

Statistical AnalysisA t-test or one-way ANOVA (with Tukey post

hoc) was used where data did not violate assump-tions of normality and heterogeneity of variance;

Am. J. Primatol.


Fig. 1. Approximate representation of the forest profileat the three sites. Profiles are based on distances anddensities obtained from a randomly selected transect ateach site using the PCQM, together with actual measure-ments for structural attributes (e.g. tree height, bole height,crown diameter, crown shape, connectivity). (i) wide cone;(ii) narrow cone; (iii) umbrella; (iv) monopodial [after Cantet al., 2001, 2003].

otherwise variables were compared using Kruskal–Wallis nonparametric test and the Mann–WhitneyU-test post-hoc. Given the need for multiple compar-isons in the Mann–Whitney U test, the significancelevel was lowered according to Bonferroni probabil-ities [dividing the Type I error rate, e.g. 0.05, bythe number of comparisons; Field, 2005]. Categor-ical data were compared using Chi-squared tests.

Overall tree and liana density (number ofstems/ha) was calculated by dividing 1 by the squareof the mean of all distances, measured in meters,and then multiplying by 10,000 to convert the figurefrom m2 to ha2 [Cottam & Curtis, 1956]. Similarly,density was calculated for each sampling point and

TABLE II. Scale Used for the Estimation of the Num-ber of Branches (≤2 cm Diameter)

Number of branchesa Mid-value used in calculations

1–5 35–10 811–25 1826–50 3851–100 75101–500 300501–1,000 7501,001–2,000 1,5002,001–4,000 3,0004,001–6,000 5,0006,001–8,000 7,0008,001–10,000 9,000aScale established by Morrogh-Bernard et al. [2009] for long-term pheno-logical monitoring in the Sabangau.

these values were used to compare the three sites.Vacant quarters were corrected for using correctionfactors detailed in Warde and Petranka [1981].

Jacobs’ D value [Jacobs, 1974] was used as anindex to assess preference for different main weight-bearing supports across the three study sites. Thisindex has been used in a number of primate studiesboth for canopy selection [e.g. Cannon & Leighton,1994; Machairas et al., 2003] and for support prefer-ence [Warren, 1997; Youlatos, 2008; Youlatos et al.,2008]. Although a variety of alternative electivity in-dices do exist, comparisons of these have found that,with the exception of Strauss’ L, all the indices arebroadly comparable and are useful measures of pref-erence [Lechowicz, 1982]. Jacob’s D is calculated as:

Jacobs D = (r − p)/(r+p − 2rp)

where r is the relative use of the support and p is therelative availability for the support within the forest.This method standardizes the relationship betweensupport use and support availability to between +1and −1, where +1 indicates maximum preferenceand −1 indicates maximum avoidance, and is sym-metrical around 0, indicating neutrality of choice (i.e.use in direct relation to abundance).

RESULTSTree and Liana Density

Ketambe, Suaq Balimbing, and Sabangau werefound to have significantly different tree densi-ties (Table III). Small trees dominated in all threeforests, but Sabangau had a significantly higher den-sity of trees <20 cm DBH compared to Ketambeand Suaq Balimbing, and Suaq Balimbing had asignificantly higher density of these trees than didKetambe (Table III). In contrast, for medium-sized

Am. J. Primatol.

6 / Manduell et al.

TABLE III. Comparison of Tree and Liana Densities in Three Study Sites

Mann–WhitneyStem density U-test (U)

Kruskal–Wallis (H) Ketambe Suaq Sabangau 1 2 1Variable 1 vs. 2 vs. 3 (1) (2) (3) vs. 2 vs. 3 vs. 3

TreesDensity <4 cm DBH 118.165*** 1,182 1,942 4,505 *** *** ***Density ≥4–<10 cm DBH 130.056*** 654 1,000 2,011 *** *** ***Density ≥10–<20 cm DBH 101.874*** 249 426 687 *** *** ***Density ≥20–<40 cm DBH 25.328*** 126 181 214 ** ns ***Density ≥40 cm DBH 54.121*** 67 59 17 ns *** ***

LianasDensity <2 cm DBH 38.351*** 1,025 1,352 924 *** *** nsDensity ≥2–<4 cm DBH 26.778*** 202 177 141 ns *** *Density ≥4 cm DBH 23.387*** 119 75 53 ** ns ***

Kruskal–Wallis test (df = 2), post hoc Mann–Whitney U-test (df = 1).Kruskal–Wallis: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.Mann–Whitney U-test (according to Bonferroni probabilities): *P ≤ 0.017; **P ≤ 0.0003; ***P ≤ 0.00003.

trees (20–40 cm DBH), densities were similar in bothpeat swamp forest sites, and significantly higherthan in dry forest. For larger trees (>40 cm DBH),however, densities were similar in the Sumatransites (Ketambe and Suaq Balimbing), but signifi-cantly lower in the Borneo site (Sabangau). SuaqBalimbing had the highest density of small lianas(<2 cm diameter), whereas the density of smalllianas was similar in Ketambe and Sabangau. Thedensity of medium lianas (2–4 cm) was higher inKetambe and Suaq Balimbing than in Sabangau. Fi-nally, for large lianas (>4 cm), densities were sig-nificantly higher in Ketambe than in the two peatswamp forest sites, where densities were similar(Table II).

Canopy Variables and Support AvailabilityThe mean DBH of trees >10 cm was significantly

higher in Ketambe than in Suaq Balimbing and Sa-bangau, but similar between Suaq Balimbing andSabangau (Table IV). Crown volume, crown width,and tree height were also significantly higher in Ke-tambe than in Suaq Balimbing and Sabangau, butno significant difference was found between the twopeat swamp forest sites (Table IV). There was a sig-nificant difference between the three study sites interms of crown connectivity (χ2 = 87.196; df = 6;P ≤ 0.001, Fig. 2), with connectivity in Sabangau >

Ketambe > Suaq Balimbing.The forests at Ketambe, Suaq Balimbing, and

Sabangau also differed in terms of the numberof different-sized supports in the forest canopy(Table IV). Ketambe had a significantly larger num-ber of supports than the other two sites for the major-ity of support classes. Overall, the availability of sup-ports at Suaq Balimbing and Sabangau was found tobe relatively similar, although Suaq Balimbing had

Fig. 2. Frequency distributions of crown connectivity for trees>10 cm DBH across three study sites.

a significantly higher density of 4–10 cm and >10 cmbranches, and <2 cm boughs (Table IV). There wasa significantly higher number of lianas in Ketambethan in the two swamp forests, but no significantdifference was found between Suaq Balimbing andSabangau (Table IV).

Support UseThere were considerable differences in the size of

supports used by orangutans between the three sites(χ2 = 616.72; df = 16; P ≤ 0.001; Fig. 3). Themost striking result was that orangutans in SuaqBalimbing used the multiple supports of the verysmallest diameter (<4 cm) much more than was ob-served in both Ketambe and Sabangau. In Ketambe,orangutans used single larger supports (10–20 cmand >20 cm diameter) more than was observed inthe two peat swamp sites. Orangutans in all threestudy sites used single supports of 4–10 cm diame-ter with similar frequencies, although orangutans in

Am. J. Primatol.


TABLE IV. Comparison of Canopy Variables and Supports for Trees >10 cm DBH across Three Study Sites

Mean Mann–Whitney U-test (U)

Kruskal–Wallis (H) Ketambe Suaq Sabangau 1 2 1Variable 1 vs. 2 vs. 3 (1) (2) (3) vs. 2 vs. 3 vs. 3

n = 160 n = 160 n = 160DBH 15.846*** 29.69 20.54 17.40 * ns ***Crown volume 27.472*** 270.11 93.94 84.18 ** ns **Crown width 36.907*** 7.47 5.29 5.32 *** ns ***Tree height 18.414*** 20.33 17.03 16.51 *** ns ***Number lianas 24.791*** 7.11 4.31 1.62 ** ns ***Boughs

<2 cm 13.456*** 15.43 27.45 14.67 ** ** ns2–4 cm 2.011ns 10.22 7.97 9.45 ns ns ns4–10 cm 19.212*** 6.57 3.05 3.67 *** ns **>10 cm 41.830*** 3.58 1.23 0.45 *** ns ***

Branches<2 cm 21.864*** 1,600.29 1,046.44 892.06 *** ns ***2–4 cm 12.125** 19.17 9.37 8.08 ** ns **4–10 cm 20.188*** 4.70 1.92 1.80 ** * ***>10 cm 28.375*** 1.78 0.78 0.03 ** *** ***

Kruskal–Wallis test (df = 2), post hoc Mann–Whitney U-test (df = 1).ns: not significant.Kruskal–Wallis: *P ≤ 0.05, **P ≤ 0.01; ***P ≤ 0.001.Mann–Whitney U-test (according to Bonferroni probabilities): *P ≤ 0.017; **P ≤ 0.0003; ***P ≤ 0.00003.

Fig. 3. Frequency distribution for support diameters used byorangutans during locomotion across three study sites.

Sabangau used multiple supports of this size muchmore than was observed in the Sumatran sites. TheSabangau orangutans also employed multiple sup-ports of 10–20 cm more often than elsewhere.

There was also a significant difference in thetypes of supports used between the three sites(χ2 = 2495.49; df = 18; P ≤ 0.001; Fig. 4). Themost notable differences were that orangutans in Ke-tambe used both single and multiple branches, andsingle and multiple lianas, more than observed atother sites, whereas orangutans in Sabangau usedboth single and multiple trunks more than observedin the two Sumatran forests. Orangutans in SuaqBalimbing used single boughs more than observedelsewhere. In peat swamp forest, multiple boughsand mixed tree supports (i.e. any combination of

Fig. 4. Frequency distribution for supports used by orangutansduring locomotion across three study sites.

trunk/branch/bough) were used more often than inmore than in dry forest.

Assessment of PreferenceOrangutans at all three sites had broadly sim-

ilar profiles of preferred supports (Fig. 5), althoughsome differences were apparent. While orangutansat all three sites showed strong avoidance of thesmallest branches and lianas (<2 cm diameter),the pattern for the smallest boughs did not followthe same trend. Although orangutans in Ketambeshowed strong avoidance for boughs of greater than2 cm diameter, in Sabangau, they showed only slightavoidance whereas in Suaq Balimbing, they showed

Am. J. Primatol.

8 / Manduell et al.

Fig. 5. Jacobs D value for preference and avoidance of supportsduring locomotion. Bo, bough; Br, branch; Tr, trunk; Li, liana.

a preference. Orangutans in both Suaq Balimbingand Sabangau used boughs 2–4 cm much more thanwas observed Ketambe. All orangutans used trunksgreater than 20 cm DBH in similar proportionsto their availability in the environment, however,preference values were slightly positive in the twopeat swamp forests (Sabangau and Suaq Balimbing)but negative in dry forest (Ketambe). In all threesites, orangutans showed a slight preference for trees>20 cm DBH. Orangutans in the two Sumatransites showed a strong tendency for using lianasin the 2–4 cm and >4 cm diameter categories,whereas in the Bornean site, orangutans used lianas2–4 cm diameter in similar proportions to their avail-ability and showed a slight avoidance of larger lianas>4 cm.

Travel TreesIn Sabangau, there was a significant difference

in the trees used for travel between the age-sexclasses in terms of DBH (F = 12.368; df = 3; P ≤0.001), crown width (F = 4.419; df = 3; P ≤ 0.01),and tree height (F = 7.450; df = 3; P ≤ 0.001):sexually active females used larger trees than theother age-sex categories (Tukeys post hoc, TableV). There was also a significant difference amongage-sex classes in the trunk-to-trunk distance be-tween travel trees (F = 6.859; df = 3; P ≤ 0.001) inSabangau, with the mean distance being the great-est for flanged males and the smallest for nonsex-ually active females. However, there was no signif-icant difference in the degree of crown overlap orgap size. In Suaq Balimbing, there were no signif-icant differences in trees used for travel betweenthe age-sex classes in any of the variables analyzed(Table V).

The trees used by orangutans for travel in SuaqBalimbing differed significantly from the randomsample of trees. “Travel trees” had a larger DBH,crown width, and crown volume, and were taller

than the random sample (Table VI). However, in Sa-bangau, travel trees did not differ significantly fromthe random sample in any variable measured, ex-cept for tree height, which was taller for travel trees(Table VI). There was also a marked difference inthe number of supports found in travel trees in SuaqBalimbing, which had significantly more branchesand boughs of all sizes, whereas in Sabangau, thenumber of supports in the travel trees was similarto those obtained in the random sample (Table VII).In both sites, the number of lianas in the crowns oftravel trees was similar to the random sample withthe exception of the smallest lianas in Suaq Balimb-ing (Table VII).

Comparison between the two peat swamp for-est sites revealed a larger DBH of trees used fortravel in Suaq Balimbing than Sabangau (27.7 cm vs.12.6 cm; t = 12.342; df = 851; P ≤ 0.001), even thoughthere was no significant difference in the meanDBH of the random samples between the two sites(12.8 cm vs. 11.4 cm; t = 1.226; df = 318; P = 0.221;trees > 4 cm DBH); plus a greater mean height oftrees used (16.4 m vs. 13.7 m; t = 6.827; df = 824;P ≤ 0.001) and greater crown width in SuaqBalimbing (5.5 m vs. 4 m; t = 9.124; df = 848; P ≤0.001). The trunk-to-trunk distance between consec-utive travel trees was also larger in Suaq Balimbingthan Sabangau (4.2 m vs. 2.9 m; t = 7.5; df = 746; P ≤0.001), as was the degree of crown overlap (2.31 mvs. 1.33 m; t = 7.350; df = 540; P ≤ 0.001). There wasno difference in the mean gap size between the twosites (1.4 m vs. 1.2; t = 1.500; df = 176; P = 0.118).

LocomotionTorso-orthograde suspensory locomotion, partic-

ularly orthograde clamber, dominated orangutan lo-comotion in all three study sites (Table VIII), al-though it was most frequently observed in Sabangau.Torso-pronograde suspension and bipedalism weremore common in the two Sumatran sites than it wasin Borneo. Tree sway was more commonly observedin peat swamp, whereas bridge was slightly morecommon in dry lowland forest. Quadrupedalism wasobserved at a higher frequency in dry lowland forestthan in the two peat swamp sites, although frequen-cies were similarly divided between symmetrical gaitwalk and pronograde scramble in each of the sites.Climbing was slightly higher in the Sumatran sitesthan it was in Sabangau, and vertical scramble oc-curred much more often in Ketambe than in eitherpeat swamp site.

DISCUSSIONWe quantified the structural features of the

arboreal environment that were likely to impacton orangutan locomotor behavior. As expected,

Am. J. Primatol.


TABLE V. One-Way Analysis of Variance and Means Separation Tests for “Travel Trees” by Age-Sex Class in PeatSwamp Forest

Age-Sex Class (Means) Tukey post hoca

Flanged Sexually Unflanged NonsexuallyMale active female male active female 1 1 1 2 2 3

Variable F df P (1) (2) (3) (4) vs. 2 vs. 3 vs. 4 vs. 3 vs. 4 vs. 4

Suaq Balimbing (n = 78) (n = 190) (n = 157) (n = 119)DBH (cm) 0.622 3 0.601 30.60 26.80 27.50 28.70 ns ns ns ns ns nsHeight (m) 1.413 3 0.238 16.80 16.40 15.90 17.40 ns ns ns ns ns nsCrown width (m) 6.747 3 0.429 5.59 5.36 5.38 5.84 ns ns ns ns ns nsTrunk-to-trunkdistance (m)

0.438 3 0.726 4.49 4.13 4.00 4.28 ns ns ns ns ns ns

Gap size (m) 1.389 3 0.246 2.36 2.19 2.22 2.59 ns ns ns ns ns nsCrownoverlap (m)

0.014 3 0.998 1.47 1.45 1.45 1.41 ns ns ns ns ns ns

Sabangau (n = 46) (n = 61) (n = 89) (n = 122)DBH (cm) 12.368 3 0.000 11.60 15.60 13.10 10.90 *** ns ns * *** *Height (m) 7.45 3 0.000 12.40 15.70 13.95 13.10 *** ns ns * *** nsCrown width (m) 4.419 3 0.005 3.53 4.48 4.00 3.94 ** ns ns ns ns nsTrunk-to-trunkdistance (m)

6.859 3 0.000 3.89 3.29 2.80 2.57 ns * *** ns ns ns

Gap size (m) 2.231 3 0.095 1.60 1.19 0.96 0.99 ns ns ns ns ns nsCrownoverlap (m)

1.687 3 0.172 0.62 1.41 1.21 1.45 ns ns ns ns ns ns

aTukey post hoc: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ns: not significant.

TABLE VI. Comparison of Attributes of Travel Trees and a Random Sample in Two Peat Swamp Forests

Mean

Site Variable Used Randoma t df Significance

Suaq n = 544 n = 160DBH (cm) 27.9 12.8 8.636 702 ***Crown width (m) 5.9 4.3 6.590 702 ***Height (m) 18.1 11.7 10.818 702 ***Crown volume 104.95 44.04 3.115 702 **

Sabangau n = 308 n = 160DBH (cm) 12.7 11.4 1.912 476 nsCrown width (m) 4.0 4.3 0.908 476 nsHeight (m) 13.6 11.9 0.18 476 ***Crown volume 36.5 49.9 1.196 476 ns

t-test: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.aRandom sample based on the nearest tree >4 cm DBH using the PCQM.

significant variations in habitat structure and theavailability of supports were discovered betweensites, which is reflected in observed differences inorangutan support use in different habitat types, anddifferences in the trees used during travel betweenthe two peat swamp sites. Interestingly, however,we also found that orangutans across the three dis-tinctly different study sites had an essentially simi-lar profile of preferred supports.

Sabangau had a much larger total tree densitythan was found in either Suaq Balimbing or Ke-tambe, yet, as a likely consequence of past distur-bance and low peat nutrient levels in Sabangau, hadonly a low density of large trees (>40 cm DBH). In

terms of tree density, Ketambe and Sabangau wereat two opposite extremes of a gradient, with SuaqBalimbing lying between the two. Suaq Balimbingand Sabangau were found to have a similar densityof medium-sized trees (20–40 cm); whereas for thelargest trees (>40 cm DBH), Suaq Balimbing andKetambe were more similar. Mean tree DBH wassignificantly higher in Ketambe than in Suaq Bal-imbing, however, indicating that the “largest trees”are smaller in Suaq Balimbing.

We anticipated that Sabangau would have thehighest density of small lianas because of past distur-bance; however, Sabangau and Ketambe were foundto have a similar density of small lianas (<2 cm

Am. J. Primatol.

10 / Manduell et al.

TABLE VII. Comparison of Support Attributes between Travel Trees and a Random Samplea

Suaqb Sabangauc

Mean Mean

Variable Travel trees Random t P Travel trees Random t P

Boughs<2 cm 11.0 14.5 3.899 *** 13.9 12.3 1.819 ns2–4 cm 4.6 2.5 3.454 *** 5.4 2.5 6.090 ***4–10 cm 2.9 0.8 6.824 *** 0.9 0.8 0.538 ns>10 cm 1.4 0.2 5.012 *** 0.1 0.3 1.915 ns

Branches<2 cm (median) 841.4 488.3 5.106 *** 852.5 718.6 1.690 ns2–4 cm 15.0 3.4 2.094 * 2.6 2.3 0.580 ns4–10 cm 3.8 0.7 2.526 * 0.3 0.6 1.458 ns>10 cm 0.9 0.2 2.015 * 0.01 0.01 0.427 ns

Lianas<2 cm 3.6 2.7 1.978 * 1.0 0.7 1.850 ns2–4 cm 0.3 0.3 0.900 ns 0.3 0.2 0.907 ns>4 cm 0.1 0.1 0.719 ns 0.01 0.04 1.942 ns

t-test: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.aRandom sample based on the nearest tree >4cm DBH using the PCQM.bSuaq: Travel trees n = 544; Random sample: n = 160 (df = 702).cSabangau: Travel trees: n = 290; Random sample: n = 160 (df = 448).

TABLE VIII. Percentages of Commonly Observed Locomotor Modes in Three Orangutan Study Sites

Mode Submode Ketambea Suaq Balimbing Sabangau

Quadrupedal and tripedal walk 17.6 10.8 8.5Walk 8.0 5.2 4.2Pronograde Scramble 9.4 5.6 4.3

Torso-orthograde suspension 35.0 40.4 47.9Brachiation 6.2 7.6 4.0Forelimb Swing 8.4 6.2 2.9Orthograde clamber 14.4 21.3 35.9Orthograde transfer 6.1 4.8 5.0

Torso-pronograde suspension 3.6 3.4 1.3Inverted pronograde walk 2.3 2.8 0.4Inverted pronograde scramble 1.3 0.6 0.7

Forelimb–hindlimb swing 0.3 2.0 1.0Bipedal walk 7.3 5.4 3.2

Bipedal walk 1.6 0.6 0.1Assisted bipedal walk 5.6 4.8 3.2

Bridge 2.8 1.9 1.9Vertical climb 16.0 13.3 9.8

Flexed-elbow 5.6 9.0 6.4Extended-elbow 1.2 0.5 0.8Vertical scramble 7.1 2.0 0.8

Vertical descent 9.4 6.6 5.2Drop 1.8 0.8 1.1Ride 0.5 0.6 0.8Sway 5.6 14.9 19.0aData from Thorpe and Crompton [2006].

DBH) whereas Suaq Balimbing had the highest den-sity of small lianas. Suaq Balimbing and Ketambehad higher densities of medium-sized lianas (2–4cm DBH), which is likely a result of the more fer-tile Sumatran soils and lack of logging. Contrary toexpectations based on past logging history, both

Suaq Balimbing and Sabangau had a similar densityof large lianas (>4 cm). Suaq Balimbing had a moreopen canopy, which may have provided a good en-vironment for liana establishment, since most lianaspecies need light to germinate and establish [Putz& Appanah, 1987]. Crown overlap was much higher

Am. J. Primatol.


in Sabangau than the other two sites, reflecting thehomogeneous size of the trees and high stem densityin Sabangau, which is likely to have impeded lianaestablishment.

In Ketambe, orangutans used lianas much morefrequently than in Sabangau (18.1% vs. 3.8%, re-spectively). Orangutans in Suaq Balimbing also usedlianas at a reasonably high frequency (12.8%), al-though they used a mixture of tree and liana sup-ports much more than observed elsewhere (Fig. 4).While in all three sites orangutans tended to avoidusing small lianas (<2 cm diameter), it was onlyin the two Sumatran sites that orangutans usedlarger lianas much more than their abundance inthe environment. Lianas often link tree crowns to-gether bridging gaps and providing arboreal path-ways for animals [Emmons & Gentry, 1983; Grand,1983]. However, in forest that has low liana den-sity, orangutans may be more likely to cross gapsby tree swaying using vertical trunks [Cant, 1992].This would appear to be the strategy employed byorangutans in Sabangau, where it is probably facil-itated by the high total stem density, particularlywith regard to smaller and more compliant trees,rather than the lack of lianas. The frequency of treesway by Sabangau orangutans is higher than ob-served elsewhere (Table VIII). Tree sway is knownto be a very efficient travel mode for orangutans[Thorpe et al., 2007], so this strategy may help re-duce energy expenditure during travel in this pop-ulation, which is known to be energetically stressedas a result of the low productivity of the Sabangauforest causing orangutans there to experience longperiods of negative energy balance [Harrison et al.,2010]. Such a strategy may even help mitigate someof the impacts of habitat disturbance on orangutanpopulations in this and other areas.

There was a slight preference for tree trunks>20 cm in all three sites, although they were onlyused more frequently than their abundance wouldpredict in the two peat swamp forests. It must benoted that the majority of trees >20 cm diameterused during locomotion were less than 40 cm diame-ter (79% Suaq Balimbing and 95% Sabangau). Thisis important because orangutans must use extended-elbow climbing techniques to climb trunks of large di-ameter. Extended-elbow climbing has a higher dutyfactor and is therefore likely to be more demandingthan the flexed-elbow techniques they use to climbsmaller diameter supports [Isler & Thorpe, 2003].In both Suaq Balimbing and Sabangau, there wereonly a handful of obervations of vertical climbing us-ing extended-elbow “bear climb” (Table VIII), themajority of which were associated with enteringlarge feeding trees, or travel within feeding trees.This suggests that orangutans avoided this behav-ior where possible, but that where it was essen-tial, any increased energetic cost was outweighedby the reward of immediately accessing a valuable

food resource. Lianas have been highlighted as animportant support for orangutans in Ketambe, en-abling them to access large feeding trees withouthaving to employ the more demanding bear climbrequired to ascend large tree trunks [Isler & Thorpe,2003; Thorpe & Crompton, 2005]. Despite this, bearclimb was observed at slightly higher levels in Ke-tambe [Table VIII; Thorpe & Crompton, 2006] thanin the two peat swamp forests, indicating that thesmaller girthed trees in these peat swamps canbe climbed using flexed-elbow climb. Climbing wasslightly lower in Sabangau than in Suaq Balimbingand Ketambe (Table VIII), this is likely due to pastdisturbance resulting in a more stunted canopy thusreducing the incidence of climbing behavior. In thetwo peat swamp sites, orangutans showed a slighttendency toward using tree trunks for travel (<20cm diameter), whereas this relationship was slightlynegative in Ketambe. This is most likely due to thetaller canopy and larger tree size in Ketambe, caus-ing orangutans to travel at higher levels, and thehigher density of lianas that can be used as alterna-tive supports.

The use of tree trunks as a support was higherin both of the peat swamp forest sites than was ob-served in dry forest, although it was highest in Sa-bangau. However, trunks of 4–10 cm in diameteraccounted for almost half of all locomotion involvingsingle trunks in both peat swamp forest sites (46%in Suaq Balimbing; 48% in Sabangau). Tree trunksof this size are flexible and therefore easily oscillatedabout the trunk, and around 70% of all locomotionon trunks of this size involved tree swaying in bothof the peat swamp forest sites. It would seem thatthe tendency for orangutans in peat swamp forest touse vertical trunks reflects both the higher densityof smaller sized trunks compared with dry forest, aswell as differences between the sites with regard tothe most continuous stratum for travel.

When support availability was compared be-tween the three sites, the two peat swamp forestswere more similar to each other than to the dry for-est. Ketambe had a more varied range of supports,which supports the hypothesis that orangutans inKetambe are more able to use particular locomo-tor/support combinations as a consequence of theirmore heterogeneous arboreal environment. Interest-ingly, orangutans in Suaq Balimbing had a muchhigher frequency of locomotion on multiple supportsof the smallest size (<4 cm diameter) than was ob-served in the other sites. This does not reflect ahigher availability of supports of this type, whichwas the same as for Sabangau. Rather, it reflectsthe fact that orangutans used different strata in thetwo peat swamp sites. Orangutans in Suaq Balimb-ing crossed trees via small peripheral branches inthe crown whereas in the Sabangau they crossed atlower levels using closely spaced trunks, and in Ke-tambe, they benefited from increased access to larger

Am. J. Primatol.


branches in the crowns of trees. Orangutans in Ke-tambe had a stronger relationship with single largesupports, which is not suprising, as we would ex-pect orangutans to use compressive locomotion onlarger, stiffer supports wherever possible becausethis is likely to reduce the energetic cost and risksof arboreal locomotion [Rosenberger & Strier, 1989;Strier, 1992; Warren & Crompton, 1998]. Indeed,orangutans in Ketambe do exhibit higher frequen-cies of quadrupedal walk than were observed in peatswamp (Table VIII). It is therefore likely that theincreased frequency of larger horizontal supports inKetambe facilitates both energetically advantageouslocomotor behavior and increased safety.

Orangutans at all three sites exhibited simi-lar frequencies of locomotion on single supports (4–10 cm), although locomotion on multiple supportsof this size class was more frequent in Sabangauthan at the other sites, reflecting the frequency ofclambering across multiple trunks [Manduell et al.,in preparation]. Orangutans in all three sites usedbranches and boughs more than their abundance inall but the smallest size categories, where they sim-ilarly avoided the smallest branches (<2 cm). Thisdoes not mean that orangutans do not use the small-est supports, however, but rather that they do notuse them in proportion to their abundance. Only inKetambe did orangutans avoid the smallest boughs,whereas in Suaq Balimbing, they were a preferredsupport and in Sabangau, they were used in accor-dance to availability (i.e. neutral selection). Theseresults suggest that orangutans in all forests do se-lect for preferred support/locomotion combinations,but, in more homogeneous forests with a more lim-ited number of support size variation (e.g. Saban-gau), orangutans are restricted to a more limitedrange of preferred locomotion/support combinations(hypotheses 1 and 2). Indeed, that orangutans in Sa-bangau exhibit higher frequencies of a small num-ber of locomotor behaviors compared to orangutansin both Suaq Balimbing and Ketambe (Table VIII)indicates that orangutan locomotion is more limitedin homogenous forest structure.

Our hypothesis that there would be little varia-tion in the trees used among age-sex classes was up-held in Suaq Balimbing but not Sabangau (hypothe-sis 3). The lack of difference in the structural featuresof travel trees in Suaq Balimbing between the vari-ous age-sex categories is likely to be a consequenceof the use of arboreal pathways, which individualsof all age-sex categories were thought to follow inKetambe [Thorpe & Crompton, 2005]. This commonuse of arboreal pathways was also observed in SuaqBalimbing, where individuals traveling togetherwould use the same route when traveling to distantfeeding trees [Manduell, pers. obs.]. Focal individ-uals were also observed to use the same sequenceof trees that had been marked from a previous fol-low of a different focal orangutan [Manduell, pers.

obs.]. However, in Borneo, the presence of arborealpathways was less obvious given that orangutans aremore solitary. This was observed in both Sabangau[Manduell et al., 2011] and the geographically closeorangutan study site of Tuanan, which is also peatswamp forest [Phillips, 2011]. In Sabangau, the sameindividual was observed to use the same sequence oftrees on different occasions [Manduell et al., 2011],but it is thought that the homogeneous nature ofthe forest may mean that selecting certain trees isless important, as their greater homogeneity reducedthe risk of increased energy expenditure through in-creased path lengths resulting from deviations fromstraight line travel [Temerin & Cant, 1983]. Never-theless, sexually active females in Sabangau usedlarger trees for travel than the other age-sex cat-egories. Previous studies have indicated that sexu-ally active females tended toward safer forms of loco-motion [Manduell et al., 2011; Thorpe & Crompton,2005] and this result supports that suggestion.

Our prediction that there would be differences inthe trees used for travel between the two peat swampforest sites as a result of differences in forest struc-ture was upheld (hypothesis 4). Overall, orangutansselected larger trees for travel in Suaq Balimbingthan in Sabangau and, while we expected the aver-age distance between consecutive travel trees to besmaller in Sabangau given the much higher stemdensity, the lack of difference in gap sizes betweenthe two peat swamp forests was surprising. We alsoexpected that orangutans in Sabangau would en-counter larger gaps between trees given the pastlogging disturbance, but rather the results from therandom sample highlighted the openness of the for-est canopy in Suaq Balimbing compared to Saban-gau. The similarity in mean gap size between crownsof adjacent travel trees in Suaq Balimbing and Sa-bangau may indicate a maximum threshold for gapcrossing by orangutans, although further testingwould be required in order to verify this wasn’t sim-ply the maximum distance observed in this study.

It is possible that locomotor strategies could al-ter in response to food availability. Orangutans in allthree sites included in this study employ a “searchand find” foraging strategy, as fruit availability isrelatively regular with less pronounced peaks andtroughs, but is of a typically relatively poor quality[Morrogh-Bernard et al., 2009]. However, in Borneoorangutans are more dependent on lower quality fall-back foods, and particularly bark, during periods offood scarcity than in Sumatra [Harrison & Marshall,2011; Morrogh-Bernard et al., 2009; Wich et al.,2006]. For much of the duration of the data collectionin Sabangau, the orangutans were heavily reliant onfallback foods, such as leaves and bark. Thus, it ispossible that, during periods of higher fruit consump-tion, Sabangau orangutans use arboreal pathwaysfor traveling between preferred known food sourcesin order to minimize path length, but that during

Am. J. Primatol.


periods of fruit scarcity, travel paths are more ran-dom, potentially leading to a higher success ratein finding valuable yet previously unknown foodsources [Morrogh-Bernard et al., 2009].

As predicted, the attributes of trees used fortravel in Sabangau did not differ greatly from therandom tree sample, as a result of the more homoge-neous forest structure in this site (hypothesis 5). Fur-thermore, our prediction that orangutans in SuaqBalimbing would select for larger trees for travelthan were typically available in the forest was alsoupheld (hypothesis 6). In order to minimize devi-ations from direct line travel and increase safety,orangutans in Suaq Balimbing may select for largertrees and travel through the canopy on branches andboughs. Because of the higher availability of fruit atthis site [Marshall et al., 2009], orangutans may re-ceive less additional benefit in terms of finding un-known food sources from nonstraight line travel thanthat hypothesized above for Sabangau orangutans.The high use of multiple small supports observed inSuaq Balimbing further reflects the supports usedwhen crossing between trees via the smallest termi-nal branches of tree crowns. Using larger trees, withassociated larger tree crowns, may reduce the gapsize between adjacent tree crowns, thereby reduc-ing the need for vertical displacement during travel.In contrast, Sabangau orangutans traveled lower inthe canopy, using trunks to cross from tree to treeeither by clambering across closely spaced trees orusing their weight to sway across to the next tree[Manduell et al., in preparation].

Aside from instances of very fast travel duringmating or fighting pursuits, or play, the locomotorbehavior of orangutans is in all likelihood a bal-ance between increasing safety and decreasing en-ergy expenditure. Orangutans that are more ener-getically stressed (e.g. Sabangau) are likely to haveto find a compromise between energetic cost and loco-motor/support combinations that provide increasedsafety, whereas in forests where fruit productivityis higher (e.g. Sumatra), orangutans may be able toplace greater emphasis on safety. Orangutans in Sa-bangau certainly use very high levels of tree sway,whereas orangutans in Ketambe use more compres-sive locomotion on large supports, which is both en-ergetically efficient and safe. However, orangutansin Suaq Balimbing use larger trees than typicallyavailable, presumably to increase safety, but alsouse multiple small supports at much higher levelsthan observed elsewhere. While small supports arelikely to be less efficient energetically than large sup-ports, they may provide more direct routes to knownfood sources and therefore increase efficiency by re-ducing path length. Orangutans have adapted to thestochastic environments in which they live by be-coming low-energy specialists, decreasing their en-ergy needs when food is scarce [Pontzer et al., 2010],however, it appears that orangutans also adapt their

locomotor strategies to reduce the energetic cost oftravel more frequently in forests where food avail-ability is lower. These approaches are complemen-tary and could increase the ability of orangutans tosurvive in habitats where food-energy availability islimited, due to either naturally low nutrient avail-ability and/or anthropogenic disturbance.

The response of primates to habitat structurevariables and their ability to either adapt to, or main-tain consistency through the selection of preferredsupports is both interesting and important, espe-cially in light of increasing impacts of human distur-bance on forest structure. Not all primate species re-spond in the same way to alterations in habitat struc-ture. The positional behavior of red colobus mon-keys (Colobus badius) showed greater differences inthe context of forest type than in seasonal or an-nual comparisons [Gebo & Chapman, 1995]. Acrossthree species of lemur, positional behavior and sup-port use were also found to differ between two for-est habitats, but although all three lemur speciesstudied altered in a similar direction, the degree ofchange was different between species [Dagosto & Ya-mashita, 1998]. In contrast, the locomotor profilesof mustached tamarin monkeys (Saguinus mystax)and five cercopithecid species remained consistentin structurally different forests [Garber & Pruetz,1995; McGraw, 1996]. The level of contrast betweendifferent habitat types will undoubtedly affect theamount of influence on positional behavior, as par-ticular habitat features may matter to a greater orlesser extent in different species.

The results of this study indicate that, whileorangutans in degraded forest (Sabangau) appear tohave retained their behavioral repertoire from moreoptimal habitats (e.g. the two Sumatran sites), theyalso have adapted to the more homogeneous environ-ment by exploiting the high density of small trees tolower the energetic cost of locomotion, further high-lighting the value of logged forests for orangutan con-servation efforts.

CONCLUSIONThe three sites used in this study showed a

large degree of difference in terms of tree andliana density. As predicted, Ketambe and Saban-gau showed the greatest degree of variation, andSuaq Balimbing was more similar to Sabangau interms of structural features, particularly with regardto support availability. Contrary to our prediction,orangutans across all three study sites had an essen-tially similar profile of preferred supports, with themost notable exception being Sumatran orangutans’stronger propensity for using lianas, which was notobserved in the Borneo site. Orangutans in Sa-bangau had a more limited repertoire with highfrequencies of a few behaviors, compared to the two

Am. J. Primatol.


Sumatran sites, whereas the wider range of supportsin Ketambe appears to have facilitated a more var-ied locomotor repertoire. In Sumatra, orangutansclearly used arboreal pathways for travel, as indi-cated by the lack of difference between the age-sexclasses and the selection of larger trees than typi-cally available. This was less apparent in Borneo,where sexually active females selected larger trees,presumably for increased safety, and where, in gen-eral, trees used were similar to those present in thesite, reflecting the more structurally homogeneousnature of disturbed peat swamp forest.

The results of this study demonstrate that for-est structure and support availability have impor-tant effects on orangutan locomotion. This influenceis likely to become increasingly important as foreststructure continues to be altered through human dis-turbance in many areas. The travel pattern observedin Sabangau probably helps reduce energy expendi-ture through travel, which might be expected to helporangutans cope with the changes in habitat struc-ture and reduced availability of food resources, andtherefore energy intake, that accompanies habitatdisturbance.

ACKNOWLEDGMENTSWe would like to thank RISTEK and PHKA

for research permissions, Dr. Suwido Limin fromthe Centre for the International Cooperation inSustainable Management of Tropical Peatlands(CIMTROP) for research permissions in Sabangau,and CIMTROP’s staff for their assistance. Data col-lection in Sabangau was conducted under the aus-pices of the Orangutan Tropical Peatland Project,in collaboration with CIMTROP. We thank TatangMitra Setia (UNAS); SOCP and all of their stafffor allowing us to conduct research in Ketambe andSuaq Balimbing. We are also grateful to PBKEL andTNGL for permissions to conduct research in the Gu-nung Leuser National Park. Particular thanks goto Dr. Helen Morrogh-Bernard and not least to theIndonesian field assistants who helped with datacollection in all three field sites. The research ad-hered to the legal requirements of Indonesia, wherethe research was conducted, and did not violatethe American Society of Primatologists Principlesfor the ethical treatment of nonhuman primates.This manuscript was improved by two anonymousreviewers.

REFERENCES

Appanah S, Gentry AH, LaFrankie JV. 1992. Liana diversityand species richness of Malaysian rain forests. J Trop ForestSci 6:116–123.

Balko EA, Underwood HB. 2005. Effects of forest structureand composition on food availability for Varecia variegata

at Ranomafana National Park, Madagascar. Am J Primatol66:45–70.

Cannon CH, Leighton M. 1994. Comparative locomotor ecologyof Gibbons and Macaques: selection of canopy elements forcrossing gaps. Am J Phys Anthropol 93:505–524.

Cant JGH. 1987. Positional behaviour of female Borneanorangutans (Pongo pygmaeus). Am J Primatol 12:71–90.

Cant JGH. 1992. Positional behaviour and body size of ar-boreal primates: a theoretical framework for field studiesand an illustration of its application. Am J Phys Anthropol88:273–283.

Cant JGH, Youlatos D, Rose MD. 2001. Locomotor behaviourof Lagothrix lagothricha and Ateles belzebuth in Yasuni Na-tional Park, Ecuador: general patterns and nonsuspensorymodes. J Hum Evol 41:141–166.

Cant JGH, Youlatos D, Rose MD. 2003. Suspensory locomotionof Lagothrix lagothricha and Ateles belzebuth in Yasuni Na-tional Park, Ecuador. J Hum Evol 44:685–699.

Cartmill M. 1985. Climbing. In: Hildebrand M, Bramble DM,Liem KF, Wake DB, editors. Functional vertebrate morphol-ogy. Cambridge: Belknap Press. p 73–88.

Coder KD. 2000. Crown shape factors and volumes. TreeBiomechanics 672 Series; Athens, GA: University of Geor-gia, pp 1–5.

Cottam G, Curtis JT. 1956. The use of distance measures inphytosociological sampling. Ecology 37:451–460.

Dagosto M, Yamashita N. 1998. Effect of habitat structureon positional behavior and support use in three species oflemur. Primates 39:459–472.

Emmons LH, Gentry AH. 1983. Tropical forest structure andthe distribution of gliding and prehensile-tailed vertebrates.Am Nat 121:513–524.

Field A. 2005. Discovering statistics using SPSS. London: SagePublications.

Garber PA, Pruetz JD. 1995. Positional behaviour in mous-tached tamarins: effects of habitat on locomotor variabilityand locomotor stability. J Hum Evol 28:411–426.

Gebo DL, Chapman CA. 1995. Positional behaviour in fivesympatric old world monkeys. Am J Phys Anthropol 97:49–56.

Gentry AH. 1991. The distribution and evolution of climbingplants. In: Putz FE, Mooney HA, editors. The biology ofvines. Cambridge: Cambridge University Press. p 3–49.

Grand TI. 1983. Motion economy within the canopy. Fourstrategies for mobility. In: Rodman PS, Cant JGH, editors.Adaptations for foraging in nonhuman primates. Contribu-tions to an organismal biology of prosimians, monkeys andapes. New York: Columbia University Press.

Harrison ME, Morrogh-Bernard HC, Chivers DJ. 2010.Orangutan energy intake and the influence of fruit avail-ability in the non-masting peat-swamp forest of Sabangau,Indonesian Borneo. Int J Primatol 31:585–607.

Harrison ME, Marshall AJ. 2011. Strategies for the use offallback foods in apes. Int J Primatol 32:531–565.

Hunt KD. 1992. Positional behaviour of Pan troglodytes inthe Mahale Mountains and Gombe Stream National Parks,Tanzania. Am J Phys Anthropol 87:83–107.

Husson SJ, Wich SA, Marshall AJ, Dennis RD, Ancre-naz M, Brassey R, Gumal M, Hearn AJ, Meijaard E,Simorangkir T, Singleton I. 2009. Orangutan distribution,density, abundance and impacts of disturbance. In: WichSA, Utami Atmoko SS, Mitra Setia T, van Schaik CP, ed-itors. Orangutans: geographic variation in behavioral ecol-ogy and conservation. Oxford: Oxford University Press. p77–96.

Isler K, Thorpe SKS. 2003. Gait parameters in vertical climb-ing of captive, rehabilitant and wild Sumatran orangutans(Pongo pygmaeus abelii). J Exp Biol 206:4081–4096.

Jacobs J. 1974. Quantitative measurement of food selection.Oecologia 14:413–417.

Knott CD. 1998. Changes in orangutan caloric intake, energy

Am. J. Primatol.


balance, and ketones in response to fluctuating fruit avail-ability. Int J Primatol 19:1061–1079.

Lechowicz MJ. 1982. The sampling characteristics of electivityindices. Oecologia 52:22–30.

Machairas I, Camperio Ciani A, Sgardelis S. 2003. Interpop-ulation differences in activity patterns of Macaca sylvanusin the Moroccan Middle Atlas. Hum Evol 18:85–202.

Mackinnon JR. 1974. The behaviour and ecology of wildorangutans (Pongo pygmaeus). Anim Behav 22:3–74.

Mackinnon K, Hatta G, Halim H, Mangalik A. 1996. The ecol-ogy of Kalimantan. Singapore: Periplus Editions.

Manduell KL, Morrogh-Bernard HC, Thorpe SKS. 2011. Loco-motor behavior of wild orangutans (Pongo pygmaeus wurm-bii) in disturbed peat-swamp forest, Sabangau, CentralKalimantan, Indonesia. Am J Phys Anthropol 145:348–359.

Marsh LK, Loiselle BA. 2003. Recruitment of black howlerfruit trees in fragmented forests of Northern Belize. Int JPrimatol 24:65–86.

Marshall AJ, Ancrenaz M, Brearley FQ, Fredriksson GM,Ghaffar N, Heydon M, Husson SJ, Leighton M, McConkeyKR, Morrogh-Bernard HC, Proctor J, van Schaik CP, YeagerCP, Wich SA. 2009. The effect of forest phenology and floris-tics on populations of Bornean and Sumatran orangutans.In: Wich SA, Utami Atmoko SS, Mitra Setia T, van SchaikCP, editors. Orangutans: geographic variation in behavioralecology and conservation. Oxford: Oxford University Press.p 97–117.

McGraw WS. 1996. Cercopithecid locomotion, support use, andsupport availability in the Tai Forest, Ivory Coast. Am JPhys Anthropol 100:507–522.

Morrogh-Bernard HC, Husson SJ, Knott CD, Wich SA, vanSchaik CP, van Noordwijk MA, Lackman-Ancrenaz I, Mar-shall AJ, Kanamori T, Kuze N, bin Sakong R. 2009.Orangutan activity budgets and diet. A comparison be-tween species, populations and habitats. In: Wich SA,Utami Atmoko SS, Mitra Setia T, van Schaik CP, editors.Orangutans: geographic variation in behavioral ecologyand conservation. Oxford: Oxford University Press. p 119–133.

Morrogh-Bernard HC, Husson SJ, Page SE, Rieley JO. 2003.Population status of the Bornean orang-utan (Pongo pyg-maeus) in the Sebangau peat-swamp forest, Central Kali-mantan, Indonesia. Biol Conserv 110:141–152.

Page SE, Rieley JO, Shotyk ØW, Weiss D. 1999. Interdepen-dence of peat and vegetation in a tropical peat swamp forest.Philos Trans R Soc Lond B 354:185–1807.

Phillips AC. 2011. Locomotor development and gap crossingbehaviour in Bornean orangutans (Pongo pygmaeus) [Ph.D.thesis]. Birmingham, UK: University of Birmingham.

Pontzer H, Raichlen DA, Shumaker RW, Ocobock C, Wich SA.2010. Metabolic adaptation for low energy throughput inorangutans. Proc Natl Acad Sci 107:14048–14052.

Proctor J, Anderson JM, Chai P, Vallack HW. 1983. Ecologi-cal studies in four contrasting lowland rain forests in Gu-nung Mulu National Park, Sarawak. I. Forest environment,structure and floristics. J Ecol 71:237–260.

Putz FE, Appanah S. 1987. Buried seeds, newly dispersedseeds and the dynamics of a lowland rainforest, Malaysia.Biotropica 19:326–333.

Rodman PS, Mitani JC. 1987. Orangutans: sexual dimor-phism in a solitary species. In: Smuts BB, Cheney DL,Seyfarth RM, Wrangham RW, Struhsaker TT, editors. Pri-mate societies. Chicago: University of Chicago Press. p 146–154.

Rosenberger AL, Strier KB. 1989. Adaptive radiation of theateline primates. J Hum Evol 18:717–750

Strier KB. 1992. Atelinae adaptations: behavioral strategiesand ecological constraints. Am J Phys Anthropol 88:515–524.

Teelen S. 2007. Primate abundance along five transect linesat Ngogo, Kibale National Park, Uganda. Am J Primatol69:1030–1044.

Temerin LA, Cant JGH. 1983. The evolutionary divergence ofold world monkeys and apes. Am Nat 122:335–351.

Thorpe SKS, Crompton RH. 2005. Locomotor ecology of wildorangutans (Pongo pygmaeus abelii) in the Gunung Leuserecosystem, Sumatra, Indonesia: a multivariate analysis us-ing log-linear modelling. Am J Phys Anthropol 127:58–78

Thorpe SKS, Crompton RH. 2006. Orangutan positional be-haviour and the nature of arboreal locomotion in Homi-noidea. Am J Phys Anthropol 131:384–401.

Thorpe SKS, Crompton RH. 2009. Orangutan positional be-haviour: inter-specific variation and ecological correlates.In: Wich SA, Utami Atmoko SS, Mitra Setia T, van SchaikCP, editors. Orangutans: geographic variation in behavioralecology and conservation. Oxford: Oxford University Press.p 33–47.

Thorpe SKS, Crompton RH, Alexander RM. 2007. Orangutansuse compliant branches to lower the energetic cost of loco-motion. Biol Lett 3:253–256.

Thorpe SKS, Holder RL, Crompton RH. 2009. Orangutans em-ploy unique strategies to control branch flexibility. Proc NatlAcad Sci 106:12646–12651.

van Schaik CP. 1999. The socioecology of fission-fusion societyin orangutans. Primates 40:69–86.

Villard MA, Merriam G, Maurer BA. 1995. Dynamics in sub-divided populations of neotropical migratory birds in a frag-mented temperate forest. Ecology 76:27–40.

Vogel ER, Haag L, Mitra-Setia T, van Schaik CP, DominyNJ. 2009. Foraging and ranging behavior during a fallbackepisode: Hylobates albibarbis and Pongo pygmaeus wurmbiicompared. Am J Phys Anthropol 140:716–726.

Warde W, Petranka JW. 1981. A correction factor table formissing point-center quarter data. Ecology 62:491–494.

Warren RD. 1997. Habitat use and support preference oftwo free-ranging saltatory lemurs (Lepilemur edwardsi andAvahi occidentalis). Zool Soc Lond 241:325–341.

Warren RD, Crompton RH. 1998. Diet, body size and the en-ergy costs of locomotion in satlatory primates. Folia Prima-tol 69:86–100.

Whitmore TC. 1984. Tropical rain forests of the far east.2nd ed. Oxford: Clarendon Press.

Whitten AJ. 1982. A numerical analysis of tropical rain forest,using floristic and structural data, and its application toan analysis of gibbon ranging behaviour. J Ecol 70:249–271.

Whitten AJ, Damanik SJ, Anwar J, Hisam N. 1987. The ecol-ogy of Sumatra. Yogyakata, Indonesia: Gadjah Mada Uni-versity Press.

Wich SA, Utami Atmoko SS, Mitra Setia T, DjoyosudgarmoS, Guerts ML. 2006. Dietary and energetic responses ofPongo abelii to fruit availability fluctuations. Int J Primatol27:1535–1550.

Wich SA, Utami Atmoko SS, Mitra Setia T, van Schaik CP.2009. In: Wich SA, Utami Atmoko SS, Mitra Setia T, vanSchaik CP, editors. Orangutans: geographic variation in be-havioral ecology and conservation. Oxford: Oxford Univer-sity Press.

Wich SA, Vogel ER, Larsen MD, Fredriksson G, Leighton M,Yeager CP, Brearley FQ, van Schaik CP, Marshall AJ. 2011.Forest fruit production is higher on Sumatra than on Bor-neo. PLoS ONE 6:e21278.

Youlatos D. 2008. Hallucal grasping behavior in Caluromys(Didelphimorpha: Didelphidae): implications for primatepedal grasping. J Hum Evol 55:1096–1101.

Youlatos D, Michael DE, Tokalaki K. 2008. Positional behav-ior of Siberian chipmunks (Tamas sibiricus) in captivity. JEthol 26:51–60.

Am. J. Primatol.

orangutan locomotion

Documents