Travel adaptations of Bornean Agile Gibbons Hylobates albibarbis ( Primates : Hylobatidae ) in a degraded secondary forest , Indonesia

Data are presented on the locomotion of Bornean Agile Gibbons (Hylobates albibarbis) in a disturbed peat-swamp forest. Our results indicate that gibbons favour continuous-canopy forest, higher canopy heights and trees with a larger diameter at breast height. Gibbons select these trees despite the study site being dominated by broken-canopy forest and small trees. Gibbons also change frequently between brachiation, climbing, clambering and bipedal walking in this disturbed forest depending on the size of gap to be crossed. Gibbons are shown to be capable of adapting to some human-induced disturbances in forest continuity and canopy height, and to the presence of smaller trees, e.g., after selective logging. Despite this, gibbons are still limited to crossing gaps of ≤12m in a single movement, and more research is needed to quantify levels of disturbance gibbons can tolerate.

Habitat disturbance presents a major problem for arboreal primates that travel exclusively in the canopy (Cannon & Leighton 1994;Cannon et al. 1994), where efficient travel requires the animal to take the most direct route available.Natural (e.g., tree-falls) and unnatural canopy gaps (e.g., from logging, clearing for hunting fruit bats and fire damage) may pose a problem if the canopy becomes highly uneven, producing less direct travel paths than were originally available (Estrada & CoatesEstrada 1996;Kakati 2000;Onderdonk & Chapman 2000;Estrada et al. 2002;Baranga 2004;Anderson et al. 2007;Chapman et al. 2007;Cristobal-Azkarate & Arroyo-Rodriguez 2007).Indonesia retains about 52.1%, or about 94,432,000ha, of original forest according to FAO (Food and Agriculture Organization of the United Nations http://www.fao.org/forestry/en/).Of this 3.8% (3,549,000ha) is classified as primary forest, the most biodiverse and carbon-dense form of forest.These alarming data on type of forest type remaining indicate that the majority of remaining forest, and thus remaining gibbon habitat, will have experienced some form of disturbance.Gibbons are obligate canopy dwellers, who require intact canopy structure for all aspects of their behavioural ecology (Carpenter 1972;Andrews & Groves 1976;Gittins 1979;Bleisch & Chen 1990;Chivers 1990;Feeroz & Islam 1992;Asquith 1995;Cannon & Leighton 1996;Campbell et al. 2008;Cheyne 2010;Hamard et al. 2010;Kakati 2000;Marshall 2010;Oka et al. 2000).Highly territorial animals such as gibbons may remain within their former ranges even following intensive forest clearance or fires which destroy a high proportion of trees (Marsh & Wilson 1981;Marsh et al. 1987).
Logging (legal and illegal) creates large patches of fragmented forest through (1) fragmentation at ground level (roads and skid trails) and (2) creating gaps in the canopy, making movement more difficult for arboreal species (Meijaard et al. 2005).Selectivelogging has been seen as the long-term 'compromise' for both humans and animals, but areas that are set to be logged selectively are often over-exploited by the timber industry.In this study we provide insight into the actual level of disturbance which gibbons can tolerate by documenting gibbon preferences in a disturbed forest and the long-term implications of forest degradation on gibbon behaviour.The future of gibbon conservation and management depends on understanding how well they can adapt to these altered canopy conditions.

Materials and Methods
The study was carried out in the Natural Laboratory for Peat-swamp Forest in the northeastern corner of the Sabangau Forest (2 0 19'S & 113 0 54'E, Fig. 1) over a period of nine months from September 2005 to June 2006.The area is operated by the Centre for International Cooperation in Management of Tropical Peatlands (CIMTROP).Comparisons between wet (flooded) and dry seasons can provide insight into felid movements in response to a potentially spatially mobile prey base.Tropical peatlands are one of the largest near-surface reserves of terrestrial organic carbon, and hence their stability has important implications for climate change (Page et al. 2002).Burning peatland in Indonesia may release 13-40 % of the mean annual global carbon emissions from fossil fuels (Page et al. 2002;Aldhous 2004;Rieley et al. 2004).It is the largest area of contiguous lowland rainforest remaining in Kalimantan and is recognised as one of the most important conservation areas in Borneo for a variety of reasons including carbon storage, regulation of water supplies and conservation of flora and fauna (Aldhous 2004).The area has been subjected to long-term legal logging, illegal logging, fire and drainage from logging canals, but is now the focus of concerted protection and restoration efforts (Morrogh-Bernard et al. 2003;Cheyne 2010).

Study Gibbons
Gibbons in this study site are breeding on average every 2.5 years (Cheyne in prep.) and their density is estimated at 3.92 groups/km 2 (Cheyne et al. 2007;Hamard et al. 2010).
Data were collected on 24 individuals from six groups and a total of 1,212 data points were recorded by means of continuous, consecutive behavioural sampling on focal individuals for the duration of one follow from morning to evening sleeping trees.The sampling unit was a complete segment of locomotion when movement was initiated from a resting position and ended when the focal individual returned to a resting state, following the definitions of (Cannon & Leighton 1994).Only full locomotion sequences, where both start and end positions could be observed, were included in the analysis.All age/sex classes were included to look for differences between the effects of body size and the presence of infants ventrally.Data were collected for climbing, leaping, brachiation and bipedal walking and merged into two main locomotor modes-brachiating and leaping for comparison.It was decided to focus only on the two most recognisable and distinguishable forms of locomotion to reduce inter-observer error: leaping, defined as discontinuous progression where the hindlimbs provide all the propulsion, and brachiation (arm swinging), defined as discontinuous progression in which the forearms are used in a suspended posture (Fleagle 1976;Cant 1986;Cannon & Leighton 1994).The authors recognise the importance of studying several forms of locomotion and we recommend that in such cases data are collected by only one or two researchers to ensure that there is no confusion between locomotion types.
The majority of observations were recorded during follows typically lasting up to six hours .Training consisted of independently testing individuals' ability to estimate heights tested against known standards (where tree heights were measured using clinometers and range finders).Observers were sampled by SMC to measure inter-individual variability in estimating heights, canopy condition and locomotion as part of the long-term data collection (Cheyne 2010).
Tree variables recorded were (1) height of tree in which the gibbon started and finished the locomotion bout; (2) average surrounding canopy height measured at the location where the locomotion started; (3) distance of the travel bout in metres (between the two trees) was estimated using methods already in place for estimating the distance a gibbon moves by extrapolating from the distance on the ground.These methods are also subjected to rigorous training and regular evaluation (Cheyne 2010); ( 4) forest type at the start location was estimated as follows; 'continuous canopy'-trees of roughly the same heights, not much undergrowth; 'continuous with emergents'-similar to 'continuous canopy' but with more tall, emergent trees and slightly more undergrowth; 'broken canopy'-the commonest type found in the study area; uneven canopy and thick undergrowth and 'gaps'-areas that had been subjected to some disturbance following the descriptions in

Habitat Data Collection
Following, and adapting, methods from (Cannon & Leighton 1994), four plots were set up randomly in each of the six groups territories using random number generation based on GPS coordinates.Twenty-four 50m transects (four/group) were constructed within the territories of each of the six groups.This is half the length of transects used by (Cannon & Leighton 1994), due to the frequent changes in habitat type at the study area.Cannon & Leighton (1994) have described the small dbh size of trees to be a good indicator of poor-quality forest.Habitat data were collected in order to compare the frequency of use of habitat structure to its frequency of availability.At 25m intervals along the transect (3 points/transect, 72 points in total) the distance to the nearest tree with a diameter at breast height (dbh) larger than 6cm was measured by means of the pointcentered quarter method (Mueller-Dombois & Ellenberg 1977) and tree height and average canopy height were measured using methods already described.
Additionally, a 20m line was laid at right angles to the transect on alternating sides.At 5m intervals along this line (5 points/line, 360 total) the forest type and canopy height were sampled.The transects were always moved at least 25m away from any original tracks used for walking, so as to give a fair representation of the majority of actual forest structure.The location of each transect was determined in a stratified random fashion with the proviso that transects had to remain entirely within the same habitat type and within the home ranges of the study groups.Preference for different structural features was measured using Jacob's D value (Jacobs 1974): D = (r-p)/(r+p-2rp) where r is the relative frequency of use and p is the relative value of availability.Jacob's D value is delimited between -1 and 1, and is symmetrical around 0 which indicates neutrality i.e. neither disproportionate avoidance nor selection.Statistical tests were performed with SPSS v16.0.
Habitat type: Broken canopy was by far the most available forest type in the study area (59%) with gaps representing 24%, continuous forest with emergents 14% and continuous canopy only 3%.The Jacob's D value of forest type could only be calculated if the expected values were ≥1.0, so 'continuous canopy' and 'continuous with emergents' were combined for analysis, as the frequency of availability for 'continuous canopy' was constantly low (hereafter known as continuous canopy).Jacob's D values for canopy type are continuous and continuous emergent combined D=0.4; broken canopy D + 0 and gap D=-0.5.There is a significant preference for continuous canopy (χ 2 =13.9, d.f=1, P>0.001) and a significant avoidance of gaps and broken canopy (χ 2 =10.1, d.f=2, P>0.005).

Discussion and Conclusions
The frequency of observed brachiation exceeded that recorded by Fleagle (1976) for Siamang (Symphalangus syndactylus 50%) and by Cannon & Leighton (1994) for Bornean Agile Gibbons (H.albibarbis, 48%).Data reported in this study (66%) are closer to those reported by (Andrews & Groves 1976a) for Lar Gibbons (Hylobates lar, 80%).Despite this, gibbons are limited in the distances they can cross with each locomotion mode, with the maximum distance seen crossed by brachiation being 12m and 6m by leaping.
Gibbons are actively selecting bigger, taller trees with a more uniform canopy than is predominantly available.The amount of time spent in 'broken canopy' far outweighs the others.It must be noted that the selection of larger and taller trees by the gibbons in Sabangau is probably due to the extensive damage and lack of continuity in much of the canopy but could also be a behavioural adaptation to increased food availability in larger trees (Cheyne 2008) or predator avoidance (Cheyne et al. 2012).This demonstrates that selective logging has affected the gibbons' ability to move through the canopy though gibbons in highly disturbed areas have been known to travel on the ground (S.M. Cheyne pers. obs. 2003Cheyne pers. obs. & 2006)).
Uneven canopy and canopy gaps pose a crucial problem for arboreal primates, as they either present a very large break in the canopy or a succession of smaller breaks (uneven canopy).Efficient, cost-effective travel through the canopy, in terms of reducing distance (and time) of direct travel between two points, is heavily constrained by the presence of gaps (Cannon & Leighton 1994).Gibbons may be hypothesised to select continuous forest types over discontinuous types and higher canopies over low.During travel, gibbons tend to follow established routes through the trees, referred to as 'arboreal highways' (Chivers 1974).These routes minimise their chance of encountering gaps and also provides support for the theory that they appear to be selecting actively certain structures for travel.
The key findings of this study are: (1) gibbons can adapt in their locomotor ecology to the effects of selective logging, i.e., reduce the level of travel by brachiation and increase other modes of travel; (2) the gibbons are choosing a 'limited' resource, the continuous tall canopy, but there is evidence of a level of disturbance to which they cannot adapt.Loss of trees between 6-15 m and/or 7-17 cm dbh would be severely detrimental to this gibbon population given the limited availability of larger trees following the selective logging.The exact percentage loss of these trees which gibbons could tolerate needs more work; (3) gibbons clearly prefer continuous canopy.We did not observe crossings of gaps larger than 12m, which may be a constraint of the gibbons' physical abilities rather than a direct response to the presence of gaps, i.e., gibbons cannot cross gaps >12m in one movement.There were no parts of the forest which were completely avoided.Daily path length for Sabangau gibbons ranges from 1-5 km depending on season (Cheyne 2010), considerably more than that reported for lar gibbons in Khao Yai (Bartlett 2009), thus, because the gibbons must be more selective in their use of the habitat they may be having to travel much further though this requires more testing.

Figure 1 .
Figure 1.Map of study site with camera locations and all major villages.

Figure 2 .Figure 3 .
Figure 2. Diagrams of forest type were used as a visual aid in the field.3)Broken canopy 4) Gaps