The Sabangau catchment forms part of an extensive peat-covered landscape in Central Kalimantan, which stretches from the Seruyan River in the west to the Barito River in the east (Fig. 1). A series of peat domes have formed in the floodplains between major rivers, each forested and largely undeveloped until the mid-1990s, when a combination of macro-scale agricultural projects and forest fires removed much of the forest cover. The Sabangau catchment is the largest (ca. 7,200 km²) and least-developed landscape within this area, with contiguous forest cover of 5,584 km² (Husson et al. 2006); settlements occur only on the banks of the surrounding rivers. The primary research site is the Laboratorium Alam Hutan Gambut (LAHG; Natural Laboratory for the Study of Peat Swamp Forest) a 500-km² area gazetted for research and located in the northeast of the catchment (Fig. 1).

Fig. 1 Map of Borneo showing the location of the Laboratorium Alam Hutan Gambut (LAHG) and the Sabangau catchment. Forest cover is given for land between the Sabangau and Katingan rivers only

---

Domed peatlands typically exhibit a catenary sequence of habitat sub-types, distributed in relation to peat depth and water table and, hence, distance from the watershed (MacKinnon et al. 1997b; Rieley et al. 1997). Three peat-swamp forest sub-types have been identified: a relatively-diverse mixed-swamp forest (MSF) found at the perimeter of the dome, a lower-canopy, species poor low-pole forest (LPF) near the centre of the swamp, and, occurring only in the Sabangau peat-swamp forest, a species-rich tall-interior forest (TIF), at the watershed of the peat dome (Page et al. 1999). This study was carried out in MSF, this habitat sub-type is relatively high in diversity and has a canopy of about 35 m (Morrogh-Bernard et al. 2003).

The LAHG stands on the site of a disused logging concession that practised selective logging for 30 years. When the concession ended in 1996, illegal loggers moved into the area (Husson et al. 2006). The route of the timber extraction railway still remains, and forms the eastern boundary of a grid system. The 9-km² grid system is situated in MSF and contains 13 major transects, which are marked with flagging tape every 25 m.

Surveys were carried out from 28 June to 27 July 2004. With level or undulating terrain, how far groups can be heard cannot easily be determined from a topographic map (Brockelman and Ali 1987). Triangulation was thus required. Three listening posts labelled A, B and C were established at each of five localities, and manned by three teams of two researchers for a period of 4 days. These localities were selected within a 4-km² area: one in the centre and the others in each of the four corners. For the first four localities, spacing of the listening posts exceeded the minimum recommended distance of 300 m (Brockelman and Ali 1987). Due to constraints of the grid system, location five was an exception, with two of the distances between listening posts under 300 m, but this did not alter the effective listening area (Table 2).

Researchers were in position by 0430 hours, and remained at the listening post until 1000 hours or for 30 min after the last group was heard calling. Sightings were recorded when they occurred. Researchers recorded the time, compass direction and estimated distance to all groups heard singing. Synchronised bearings were recorded at intervals of 3 min. Lengths of song bouts were recorded using a stopwatch, and were deemed to have ended if the gibbons were silent for more than 2 min (Gittins 1984).

In order to use loud calls for sampling gibbon populations, the average probability of calling, and the variables that affect frequency, must be studied in a population of known groups (Brockelman and Ali 1987; Brockelman and Srikosamatara 1993). OBrien et al. (2004) found that the probability of agile gibbons calling stabilises by day 4. Thus, each listening post was surveyed for a period of four consecutive days during this study, with the exception of the second sample period, which was surveyed from the 6–7 July and 9–10 July. Heavy rain on the morning of 8 July made data collection impossible. Probability of calling rates was generated from the data collected during this study after the research period, based on the number of groups located and the number of calls recorded for each group.

Triangulation points and sightings were marked daily on maps of the study area, and ultimately overlaid onto a single map, allowing researchers to estimate the number of groups present during the research period (Estrada et al. 2004). Points mapped more than 500 m apart were considered to be different groups (Brockelman and Ali 1987; Brockelman and Srikosamatara 1993; OBrien et al. 2004). This figure is based on the approximate diameter of a groups range and determines the maximum distance that agile gibbons might move between calls (OBrien et al. 2004). A home range of 70 ha would have a radius of 472 m. It was thus felt that 500 m was a conservative separation for calling groups, which incorporated the possibility of larger home ranges for peat-swamp forest gibbons and the ellipsoid nature of many home ranges (OBrien et al. 2004). Points mapped less than 500 m apart were assigned to groups by establishing which groups were singing simultaneously, supplemented with acoustical and directional information. Only the acoustical points and sightings represent when the position of groups was actually determined.

In addition to recording locations of gibbon calls, researchers attempted to locate groups by sight that were close to the listening stations. Group size and composition (number of individuals of each age/sex class) were recorded when a group was sighted, and the groups location noted using GPS and added to the map.

Gibbons seldom sing when it is raining, and windy conditions have also been shown to inhibit singing (Brockelman and Srikosamatara 1993; Leighton 1987). Cloud cover and wind level were thus recorded at 10-min intervals (Brockelman and Srikosamatara 1993; OBrien et al. 2004). Temperature was noted daily; rainfall was recorded once in the morning and evening. To ascertain if weather conditions had an effect on singing frequency or start times, product-moment correlation coefficient, Mann–Whitney U, and chi-square tests were used, with P=0.05.

Agile gibbons at LAHG called most frequently between 0500 and 0700 hours (73% of calls recorded). Start times prior to 0500 hours accounted for 12% of all calls recorded while 9% of calls occurred between 0700 and 0800 hours. Calls recorded after 0800 hours accounted for 4% of calls and calls recorded after 0900 hours accounted for just 2%. The average probability of a single group calling on any given day was 0.62 (n=20 days). Across sample periods, the probability of a group calling on day 1 was 0.59 (±0.14, n=5 days). The probability of a group calling on at least one of the first 2 days was 0.92 (n=10 days), and of calling on at least one of the first 3 days was 1.0 (n=15 days).

Researchers recorded 125 calls over a period of 20 days, 29 of which were solo calls. As it is impossible to determine if these calls were made by paired or unpaired males, thus either over or under-estimating population density, they were not used in the density analysis. The remaining 96 were duets of which 65 produced two-point bearings, three-point bearings or triangulation. These were plotted on a map of the grid system to estimate the number of groups (Fig. 3).

Fig. 3 The location of gibbon (Hylobates agilis albibarbis) groups. The position of all the listening posts are indicated, as is the old railway track that forms the eastern boundary of the grid system. A cross indicates three-point bearings; this was when the compass bearing from each team met at a perfect point. A star indicates two-point bearings; this was when the calling group was only heard by two of the three teams. A triangle indicates where three bearings produced triangulation. This means that the gibbon group in question was calling from somewhere within the triangle. A circle indicates sightings. All bearings and sightings are marked with the time and date. A jagged line marks possible boundaries between groups. The shaded areas indicate which of the acoustical points or sightings were attributed to which group. Those that could not be confidently assigned remain unshaded

---

Nineteen groups were identified and are referred to as G1–G19. The larger appearance of G13 and G17 is due to the ranges of these groups being closer to the listening posts, subsequently yielding more data. Other groups appeared to have only one edge of their range within the effective listening area and, therefore, a smaller sample size was recorded.

Wind conditions were overwhelmingly calm (98%, n=20) and thus had no effect on calling frequency, or start times. Cloud cover was clear on 35% of days, >50% on 10% of days and <50% on 55% of days (n=20). No significant relationship was found between cloud cover and calling frequency or start times. Minimum temperatures averaged 21°C (range 20–24°C) and maximum temperatures averaged 29°C (range 27–31°C), but had no statistical relationship with calling frequency or start times.

A total of 31.4 cm of rain fell during the research period, which took place during the annual dry season. On the mornings of gibbon surveys, it drizzled slightly once. Rainfall tended to occur during the late afternoon and overnight. Rainfall during the night preceding data collection correlated positively with start times of calling (r_p=0.519, n=20, P<0.02) (Fig. 4).

Fig. 4 Figure showing relationship between rainfall the previous night and start times of calling. One can note an increase in start times of calling groups, with a corresponding increase in rainfall

---

For gibbon surveys, researchers more frequently use auditory as opposed to visual cues (but see Alfred and Sati 1990). Haimoff et al. (1986) located 20 groups during a survey of black-crested gibbons (Nomascus concolor) through auditory cues, only six of which were sighted. Similar to our study, the most direct and prolonged sightings took place while the gibbons were calling (Haimoff et al. 1986).

As in other studies of vocal primates (e.g. Gursky 1998; Hanya et al. 2003), our study combined two techniques. Density estimates were obtained through the mapping of calls, and information on groups was obtained by tracking in on calling groups and through opportunistic sightings at the listening posts, where detection was not compromised by the noise of walking on uneven terrain (Ross and Reeve 2003). These techniques were especially successful during our restricted research period. As calls can be heard over greater distances than the animals are detectable by sight, the inclusion of auditory cues increases the number of encounters (Davies 2002). Moreover, sightings are still obtained by including the techniques outlined above, and are more prolonged than brief sightings often obtained during line transects (White and Edwards 2000).

Low density (less than two groups/km²) potentially affects singing frequency, as groups with few neighbours may sing less than groups surrounded by neighbours (Brockelman and Srikosamatara 1993). Our density exceeds this figure, and we witnessed two group encounters involving four groups, indicating that, even if group ranges are not overlapping (Sommer and Reichard 2000), gibbons are coming close to each other. Moreover, research shows that for agile gibbons, singing is spontaneous and not stimulated by other groups (Brockelman and Ali 1987).

Individual density of 7.4 animals/km² was calculated from a mean group size of 3.4 individuals. Previous research on mean group size of H.a. albibarbis in West Kalimantan, Borneo (Mitani 1990), yielded an average group size of 4.1. This figure is higher than the data in this study, but Mitanis research period was longer and the density is higher in that region. It is possible that some individuals of each group sighted were missed during our study, as the gibbons at Setia Alam are not habituated. In addition, agile gibbon group members can be up to 30 m apart at any given time (Gittins and Raemaekers 1980). Because mean sustained sighting length was 21.2 min, mean group size generated by these data were used in the density analysis, as opposed to referring to published data, as no evidence suggests that average group size is consistent across study sites. In fact, there is large variation regarding density, group size and range size between sites, even if only H. agilis is considered (Table 3).

Table 3 Comparison of densities and home range size for Hylobates agilis across study sites

Location of study site	Taxa	Average group density/km2	No. of study groups	Animal density/km2	Mean group size	Home range (ha)	Habitat type	Reference
LAHG, Central Kalimantan, Indonesia	H.a. albibarbis	2.2	19	7.5	3.4	45	Peat-swamp forest	This study
Gunung Palung Reserve, West Kalimantan, Indonesia	H.a. albibarbis	3.6	10	14.9	4.1	28	Mountain forests. Also beach and mangrove forests, peat- and freshwater-swamp forests and lowland rainforests	Mitani (1990)
Tanjung Puting National Park, South Kalimantan, Indonesia	H.a. albibarbis	–	-	8.7 (n=16)	–	–	Primarily heath forest, also Nypa forest, mangrove and peat-swamp forest	Mather (1992)
Barito Ulu Research Area,Central Kalimantan, Indonesia	H. agilis × H. muelleri	2.1	?	8.2	4	43–46	Lowland dipterocarp and heath forest	McConkey et al. (2002)
Bukit Barisan Selatan National Park, Southwest Sumatra, Indonesia	H. agilis agilis	0.7	?	1.4–2.2	2.61	–	70% dipterocarp, forest, also beach forest, freshwater-swamp forest, Nypa forest and hill and mountain forest	OBrien et al. (2004)
Sungai Dal, Gunung Bubu Forest Reserve, West Malaysia	H. agilis unko	4.3	7	18.9	4.4	29	Lowland and hill dipterocarp forest	Gittins and Raemaekers (1980), Gittins (1982)

This study was conducted within MSF, which is the largest habitat sub-type in the Sabangau catchment, covering 2,622 km². Extrapolation of the results across this habitat sub-type yields a gibbon population of 19,500 (±4,200) individuals. Besides MSF, two other major habitat sub-types are recorded in the Sabangau catchment; tall interior forest (TIF: 909 km²) and low pole forest (LPF: 1,782 km²). TIF supports the highest diversity of plants and animals in the catchment (Page et al. 1997) and the highest density of orang-utans (Morrogh-Bernard et al. 2003). Gibbons are frequently heard and seen there (S.J. Husson, personal observation). Thus, it can be expected that gibbon density in this habitat sub-type is at least equivalent to that recorded in MSF (and probably higher). This adds an additional 1,960±400 groups, or 6,760±1,445 individuals to the overall total. LPF, by contrast, has a much lower species diversity and orang-utan density than MSF (Page et al. 1997) and gibbons are very rarely heard calling there. For the purposes of this extrapolation it is safest to assume a density of zero. Thus, we estimate an overall population estimate of about 26,300±5,600 individuals. This is a very broad extrapolation, however, and factors such as tree-species composition and nutrient availability are likely to affect local density. Additionally, logging activities have occurred throughout the catchment causing varying levels of damage.

Some studies show gibbons to be fairly tolerant of logging activities (Chivers 1986), particularly in the short-term (Howell 2003, cited in Meijaard et al. 2005) due to the strong territoriality of these species (Wilson and Johns 1982). Other studies, however, show decreases in density of 20–60% between logged and primary forest (e.g. Johns 1992), and others even greater decreases of over 60% (Nijman 1997, Howell 2003, both cited in Meijaard et al. 2005) owing to reductions in food availability. Logging damage is not consistent across the catchment and some areas of forest have been damaged more heavily than at LAHG. We do not know to what degree the gibbon population has suffered in these areas. For these reasons the extrapolation should be treated with caution. Nevertheless, because we probably underestimate gibbon density in the TIF and LPF habitat types, and as solitary individuals are excluded from the analysis, this population estimate may be considered a minimum and is thus useful for conservation planning for H. agilis.

IUCN stresses the importance of re-evaluating taxa at regular intervals, particularly those that are listed as Near Threatened or Conservation Dependent. IUCN last assessed H.a. albibarbis in 2000 (Eudey 2000). In 1977, Chivers estimated the population of H. agilis to be in the region of 744,000 individuals. He attributed 455,000 of these individuals to H.a. albibarbis. His predictions for the species future were bleak, with a 95% loss of H.a. albibarbis over the following 15 years. In 1984, he updated this status, suggesting H.a. albibarbis was widespread and relatively safe (Brockelman and Chivers 1984, p.4). In 1986, MacKinnon published slightly higher population estimates for H. agilis, in the region of 850,000, but did not differentiate between the subspecies of H. agilis. Both studies warn of the crudeness and possible inaccuracy of the data (Chivers 1977; MacKinnon 1986).

Reliable population estimates are confounded by the fact that, with the exception of North Vietnam, gibbon distribution and abundance in Borneo is the most poorly documented (Brandon-Jones et al. 2004). Moreover, what has been documented shows a coastal bias in terms of information (Marshall and Sugardjito 1986; Brandon-Jones et al. 2004).

The density figures in this study are encouraging and indicate that peat-swamp forest is an important habitat for gibbons. At the time of this study, illegal logging presented the greatest threat. Since this study was conducted, the Sabangau catchment has been designated as a national park. With this status in place, Husson et al. (2006) identify, as an immediate conversation priority, the damming of illegal logging extraction canals in an attempt to restore the hydrology of peat-swamp forest within the Sabangau.

The slow life history of gibbons makes them particularly susceptible to population crashes, which is a concern, as much of the habitat occupied by gibbons continues to be destroyed, which could affect survival rates of dispersing individuals (Mitani 1990). Monitoring over time would assess the stability of the Sabangau gibbon population. Whether the newly protected status of the area will have a positive impact on gibbon populations can be considered in future studies. Research should extend the survey area to establish whether there are differences in gibbon density between areas of differing logging damage and habitat sub-types, in order to produce a more accurate estimate of total population size in the Sabangau catchment.

People are slow to realise that large tracts of forest are essential for human survival (Chivers 1977). The environmental benefits and importance of peat-swamp forest have been well documented (MacKinnon et al. 1997a; Page et al. 1997, Morrogh-Bernard et al. 2003). This area could represent one of the largest continuous populations of H.a. albibarbis in Borneo. It is hoped that these data will encourage longitudinal studies of gibbons in the Sabangau catchment.