Comparing spatial capture–recapture modeling and nest count methods to estimate orangutan densities in the Wehea Forest, East Kalimantan, Indonesia
Graphical abstract
Introduction
Accurate information on the density and abundance of animal populations is essential to answering central questions in ecology and conservation biology. Such information allows us to test hypotheses about the relationship between environmental variables and abundance, expanding our understanding of the ecological factors that limit populations. It is also crucial for effective conservation planning, as such information can be used to assess threats to populations and species, set conservation priorities, and monitor populations (Seber, 1982, Williams et al., 2002, Borchers et al., 2003). However, obtaining accurate density and abundance estimates is challenging, especially for animals that are elusive, range widely, and live at low densities (Garshelis, 1992, Karanth, 1995, Thompson, 2004).
This is clearly illustrated in the case of the orangutan. Orangutans, the only Asian great ape, exhibit considerable geographic variation in ecology, behavior, and morphology (Wich et al., 2009). Their population densities also vary widely across their range, with Sumatran orangutans (Pongo abelii) generally exhibiting higher densities than Bornean orangutans (represented by the Northwest subspecies, Pongo pygmaeus pygmaeus; Central subspecies, Pongo pygmaeus wurmbii; and Northeast subspecies, Pongo pygmaeus morio) (Husson et al., 2009, Marshall et al., 2009a). Accurate information on orangutan densities across their geographic range, especially for the little-known Northeast Bornean orangutan (P.p. morio), is necessary if we are to fully understand the ecological factors that limit orangutan populations (Marshall et al., 2009a, Marshall et al., 2009b, Wich et al., 2011a). Information on orangutan abundance and density is also crucial for orangutan conservation. Both orangutan species are classified by the IUCN as endangered; the population of the Bornean orangutan has declined over 50% in the last 60 years and the Sumatran orangutan population has declined an estimated 80% over the last 75 years (Ancrenaz et al., 2008, Singleton et al., 2008). The causes of this decline are extensive habitat loss and fragmentation due to logging, mining, the expansion of oil palm and acacia plantations, and fire (Marshall et al., 2006, Wich et al., 2011b, Meijaard et al., 2012), and forest conversion continues at a rapid rate on both Borneo and Sumatra (Sodhi et al., 2004, Margono et al., 2014). Hunting, and, increasingly, human–orangutan conflict are also major contributors to this decline (Meijaard et al., 2011, Davis et al., 2013). Conservation action is urgently needed to prevent further population declines, and knowledge of densities and abundance are important for implementing effective conservation policy.
However, it is notoriously difficult to obtain accurate abundance or density estimates for orangutans. They are cryptic, solitary, and generally live at low densities, making direct counts impractical for most studies. Because of these difficulties, researchers generally rely on counts of indirect sign to census their populations (Kühl et al., 2008). To-date the most popular survey method for orangutans are nest count methods, in which the sleeping platforms (nests) that orangutans build each night are used to calculate a density of individuals in an area. In the most popular version of these methods, all nests visible from a line transect or in a plot are counted; nest counts are then converted into nest densities by dividing the number of nests counted by the area surveyed, which is either known (plot surveys; van Schaik et al., 2005) or estimated using a detection function (line transects surveyed using distance sampling methods; Buckland et al., 2001, Thomas et al., 2010). Nest densities are then converted into orangutan density estimates using the following formula:in which Dind = density of individuals, Dnest = density of nests, p = proportion of nest builders in the population, r = number of nests built per individual per day, and t = nest decay time (Hashimoto, 1995, van Schaik et al., 1995).
Nest count methods have been used extensively to assess or monitor orangutan populations (Husson et al., 2009). However, these methods have limitations (Mathewson et al., 2008, Marshall and Meijaard, 2009, Spehar et al., 2010). First, these methods rely on the assumption of perfect detection (in the case of line transects, that all nests above the line are counted; in the case of plot surveys, that all nests in the plot are counted) although studies demonstrate that even teams of experienced observers miss nests (van Schaik et al., 1995, van Schaik et al., 2005, Johnson et al., 2005). Another major issue lies in the parameters used to convert nest density into orangutan density (p, r, and t). The proportion of nest builders in the population (p) and the rate at which nests are produced (r) must be based on observed values from known populations, and nest decay rate (t) must also be based on observations of nest longevity in an area, although mathematical modeling (Markov chain analysis) can be used to calculate nest decay from shorter-term observations (Buij et al., 2003, Johnson et al., 2005, Mathewson et al., 2008). Obtaining accurate information for these parameters requires substantial time and effort, so values calculated from a few long-term study sites are often applied in studies across the orangutan range. This can be a concern as some parameters (in particular nest decay, t) exhibit very high variability between sites (Mathewson et al., 2008). As any changes in parameters produce directly proportional changes in the resulting orangutan density estimate, this means that density estimates that do not use precise or locally calculated parameters could be unreliable (Mathewson et al., 2008). Such issues clearly have major implications for our understanding of orangutan ecology and for conservation planning, and finding an alternative to nest surveys should be a high priority. However most studies still calculate densities based on nest surveys, and many of these continue to employ non-local parameters due to limited time and money (Spehar et al., 2010, Meijaard et al., 2012).
A possible alternative for estimating abundance and density are camera trap methods. Camera trapping is becoming a preferred method for studying rare and elusive species (e.g., O'Connell et al., 2010). Recent advances in statistical techniques, namely spatial capture–recapture modeling or SCR (Borchers and Efford, 2008, Royle and Young, 2008, Efford, 2011, Royle et al., 2013, Royle et al., 2015), allow the calculation of population density from ‘captures’ of individual animals obtained using camera traps. SCR models have an advantage over conventional capture–recapture (CR) models in that they allow for flexible trap arrangement (e.g., grid vs. linear arrangements that do not require even spacing across the study area; Efford and Fewster, 2013, Tobler and Powell, 2013) and can incorporate both individual-level covariates (e.g., sex or age class) as well as station level covariates (e.g., road vs. trail or habitat; Sollmann et al., 2011). This type of flexibility is especially important in Borneo and Sumatra, where field conditions like difficult terrain can make research design a challenge.
SCR modeling relates the encounter history of individuals (when and where they are captured) to activity centers of individuals during the trapping period (calculated as the spatial relationship between individuals and camera traps). Density is estimated as number of individuals occurring within some delineated area (the “state-space”), usually defined by the camera trapping array plus a buffer area (Royle and Gardner, 2011). SCR modeling has now been used to estimate densities for many mammals that are elusive, occur at low densities, and occupy large home ranges (Royle et al., 2009a, Royle et al., 2009b, Royle et al., 2011, Gardner et al., 2010a, Gardner et al., 2010b). These methods count the animals themselves and thus do not present problems related to converting indirect sign into animal densities. In addition, if deployed properly camera traps can also provide additional information about habitat use, behavior, and even demography (e.g., Galvis et al., 2014).
Despite its promise and wide application in wildlife studies, camera trapping has only recently been embraced by primatologists (Head et al., 2012, Olson et al., 2012, Tan et al., 2013, Loken et al., 2013, Loken et al., 2015, Galvis et al., 2014, Gregory et al., 2014). Most notably, a recent study simultaneously used nest surveys and camera trapping to estimate the relative abundance and distribution of chimpanzees (Pan troglodytes troglodytes) and gorillas (Gorilla gorilla gorilla) across different habitat types in West Africa, and found that the two methods produced roughly comparable results (Nakashima et al., 2013). However, this study was only able to use mean camera trap capture rate to calculate a relative abundance index for each species. The “next step” that would allow the calculation of absolute abundance and density is the use of new statistical techniques like SCR modeling, which have not yet been applied to primate populations. For camera traps to be used to estimate abundance and population density using SCR modeling or similar techniques, animals must be individually identifiable from photographs and individuals need to be captured and recaptured by camera traps, which are most easily placed on the ground. These criteria may be difficult to meet for some primates, but recent research suggests that this method may be appropriate for use with orangutans. Orangutans do not have unique stripe patterns or markings, but individuals are identifiable based on facial characteristics and other features that can be recognized from photographs. Secondly, recent studies indicate that Bornean orangutans may move on the ground more than previously thought, although Sumatran orangutans seem to engage in terrestrial behavior less often, perhaps because of the presence of a potential terrestrial predator, the tiger (Loken et al., 2013, Loken et al., 2015, Ancrenaz et al., 2014). This increases the likelihood of capture by camera traps for at least Bornean orangutans.
Given their extensive use to estimate densities for other elusive animals, we were interested in examining the applicability of camera trapping and SCR modeling to orangutan populations. The purpose of this study was three-fold: 1) to evaluate the feasibility of using of camera traps and SCR modeling to estimate orangutan densities, using a population of Northeast Bornean orangutans (P.p. morio) as a case study; 2) to compare results obtained using camera trapping and SCR modeling to those obtained using an established method (nest surveys); and 3) to assess the advantages and disadvantages of both methods to make general recommendations for researchers wishing to estimate population parameters for orangutans and other elusive animals.
Section snippets
Study site
This study was carried out in the Wehea Forest in East Kutai District, East Kalimantan, Indonesia. Wehea (01°32′46″N, 116°46′43″E) contains 38,000 ha of mostly undisturbed forest bordered by large tracts of primary and secondary forests currently classified as logging concessions. Logging ceased in the mid-1990s and Wehea has been protected by an agreement between a local community and the local government since 2004. Wehea Forest contains lowland Dipterocarp, sub-montane and montane forests,
Camera trapping and SCR analysis
From March 21 to October 18, 2012, we obtained a total of 112 distinct camera trap records of orangutans. Photos were high-quality enough to allow us to clearly identify individuals in 67 of these 112 records (60%) (Fig. 2); the remaining records in which photos were not of sufficient quality or composition to facilitate identification of an individual (n = 45) were discarded. Adult males accounted for n = 23 or 51% of discarded records; adult females for n = 9 or 20% of discarded records; and
Discussion
This study is the first to use camera traps and SCR modeling to estimate the densities of orangutans, or indeed, any primate. Below, we compare camera trapping and SCR modeling to nest counts and discuss the implications for decision making when choosing a method to assess the populations of orangutans or other elusive primates.
Role of funding sources
Financial support for this research was provided by the following institutions and funding agencies: Vanier Canada Graduate Scholarship, Pierre Elliot Trudeau Foundation, LUSH Cosmetics (CPCT026), Disney Worldwide Conservation Fund, Integrated Conservation, the University of Wisconsin Oshkosh (FDR779), the Rufford Small Grants Foundation (11266-B), and the Orangutan Land Trust. These funders only provided financial support and were not involved in study design; in the collection, analysis and
Acknowledgments
We are very grateful to the Wehea Management Body and the Lembaga Adat of Nehas Liah Bing for allowing us to conduct research in Wehea Forest, and to the State Ministry of Research and Technology of Indonesia for granting us permission to conduct research in Indonesia. We are thankful to Lee Qi for providing maps of Wehea Forest. We are also indebted to the individuals and organizations who provided logistical and organizational support for conducting our research: the Wehea Rangers, ECOSITROP
References (66)
- et al.
Seasonal movements in the Sumatran orangutan (Pongo pygmaeus abelii) and consequences for conservation
Biol. Conserv.
(2002) - et al.
Rapid assessment of orangutan density
Biol. Conserv.
(2003) - et al.
A survey of the orangutan (Pongo pygmaeus wurmbii) population in and around Gunung Palung National Park, West Kalimantan, Indonesia based on nest counts
Biol. Conserv.
(2005) Estimating tiger Panthera tigris populations from camera-trap data using capture–recapture models
Biol. Conserv.
(1995)The behaviour and ecology of wild orangutans (Pongo pygmaeus)
Anim. Behav.
(1974)- et al.
The blowgun is mightier than the chainsaw in determining population density of Bornean orangutans (Pongo pygmaeus)
Biol. Conserv.
(2006) - et al.
Southeast Asian biodiversity: an impending disaster
Trends Ecol. Evol.
(2004) - et al.
Improving density estimates for elusive carnivores: accounting for sex-specific detection and movements using spatial capture–recapture models for jaguars in central Brazil
Biol. Conserv.
(2011) - et al.
Estimating jaguar densities with camera traps: problems with current designs and recommendations for future studies
Biol. Conserv.
(2013) - et al.
Orangutan nesting behavior in disturbed forest of Sabah, Malaysia: implications for nest census
Int. J. Primatol.
(2004)