Improving density estimates for elusive carnivores: Accounting for sex-specific detection and movements using spatial capture–recapture models for jaguars in central Brazil

Abstract

Owing to habitat conversion and conflict with humans, many carnivores are of conservation concern. Because of their elusive nature, camera trapping is a standard tool for studying carnivores. In many vertebrates, sex-specific differences in movements – and therefore detection by cameras – are likely. We used camera trapping data and spatially explicit sex-specific capture–recapture models to estimate jaguar density in Emas National Park in the central Brazilian Cerrado grassland, an ecological hotspot of international importance. Our spatially explicit model considered differences in movements and trap encounter rate between genders and the location of camera traps (on/off road). We compared results with estimates from a sex-specific non-spatial capture–recapture model. The spatial model estimated a density of 0.29 jaguars 100 km⁻² and showed that males moved larger distances and had higher trap encounter rates than females. Encounter rates with off-road traps were one tenth of those for on-road traps. In the non-spatial model, males had a higher capture probability than females; density was estimated at 0.62 individuals 100 km⁻². The non-spatial model likely overestimated density because it did not adequately account for animal movements. The spatial model probably underestimated density because it assumed a uniform distribution of jaguars within and outside the reserve. Overall, the spatial model is preferable because it explicitly considers animal movements and allows incorporating site-specific and individual covariates. With both methods, jaguar density was lower than reported from most other study sites. For rare species such as grassland jaguars, spatially explicit capture–recapture models present an important advance for informed conservation planning.

Introduction

Owing to worldwide, large-scale habitat conversion and direct conflict with humans, many large and wide-ranging carnivores are of conservation concern. Although many charismatic species such as the big cats have received considerable attention through research over the past decade (Brodie, 2009), their populations continue to decrease (IUCN, 2010). Abundance and population density are key baseline parameters for conservation planning (Lebreton et al., 1992, Reid et al., 2002). Yet reliable estimates are hard to obtain for these species because of their elusive nature and the spatial and temporal scale of their movements that need to be addressed by conservation-oriented studies (Karanth et al., 2006, Karanth and Chellam, 2009).

Camera traps have considerably advanced our ability to study elusive animals (Kays and Slauson, 2008). They have the advantage of being non-intrusive and applicable over large areas with relatively moderate effort (Silveira et al., 2003). Today, camera trapping is used to study a variety of aspects of wildlife ecology, ranging from species presence (Linkie et al., 2007) and behavior (Harmsen et al., 2009) to relative abundance within a species assembly (O’Brien et al., 2003). Particularly for species with individually identifiable coat patterns, data from camera trapping can be analyzed within the analytically sound framework of capture–recapture models to estimate population abundance and density (Karanth, 1995, Karanth and Nichols, 1998, Otis et al., 1978, White et al., 1982) or population dynamics (Karanth et al., 2006, Gardner et al., 2010a). These models account for the fact that we do not necessarily observe all animals in the study area, i.e., the issue of imperfect detection. This methodology has been applied to several different species such as common genets (Genetta genetta – Sarmento et al., 2010), maned wolves (Chrysocyon brachyurus – Trolle et al., 2007), pumas (Puma concolor – Kelly et al., 2008) and is most commonly used to study individually distinctive large cats (e.g., Karanth and Nichols, 1998, Silver, 2004).

Though widespread, the use of camera trapping in combination with capture–recapture models has an important shortcoming that it shares with other methods applied to estimate abundance: the interpretation of abundance, or specifically the estimation of the area to which this abundance refers (e.g., Karanth and Nichols, 1998, Royle et al., 2009). Whereas most models assume geographic closure of the population, i.e., no movement on and off the sampling grid (White et al., 1982), this assumption is generally and widely violated (e.g., Karanth and Nichols, 1998), especially for large mammals. The standard approach is to buffer the grid with half the mean maximum linear distance moved by individuals captured in more than one trap (MMDM, Karanth and Nichols, 1998). Although this approach performed well in simulation studies (Wilson and Anderson, 1985), it is an ad hoc approach with little theoretical justification (Williams et al., 2002). Other approaches have been used to estimate buffer width, for example the full MMDM (twice the MMDM–fMMDM), or the radius of an average home range, both based on telemetry data (Soisalo and Cavalcanti, 2006) and on information from the literature (Wallace et al., 2003). Since density estimates are directly influenced by the chosen buffer width, comparisons of estimates from different methodologies become difficult.

Spatially explicit capture–recapture (SECR) models are a recent advance in the field of density estimation (Efford, 2004, Royle and Young, 2008). These models make use of the spatial location of captures in order to first determine an individual’s activity center and then to estimate the density of activity centers across a precisely defined polygon containing the trap array (Gardner et al., 2009, Royle et al., 2009). They thereby circumvent the problem of estimating the effective area sampled. SECR models can be implemented within a Bayesian framework and therefore provide valid inferences even with small sample sizes. The ‘pseudo-code’ used by the freely available software WinBUGS (Gilks et al., 1994) provides an easy-to-use and flexible framework for fitting Bayesian SECR models.

The flexibility of these models also allows for the incorporation of other factors of interest, for example sex as an individual covariate (Gardner et al., 2010b). Differences between the sexes in their behavior and space use are typical for vertebrate social organizations, particularly for most felids, for which the home range or territory of a single male or a group of males may partially or completely overlap the generally smaller home ranges or territories of one to several females (Sandell, 1989). The resulting differences in space use and movement between the sexes will be reflected in differences in encounter probability with camera traps and should be taken into account when estimating population density and abundance.

The jaguar Panthera onca (Linnaeus, 1758) is the largest American felid and the third-largest big cat. It occurs from the southwestern United States and Mexico to northern Argentina. Over the last century, the species’ range has contracted to approximately 55% of its original extent (Zeller, 2007) because of loss of natural habitat and persecution (Sanderson et al., 2002). The IUCN classifies the jaguar as Near Threatened (IUCN, 2010). In spite of its wide distribution and conservation concern, the species remains less studied than most other large cats (Brodie, 2009).

Here, we used SECR models to estimate jaguar abundance and density in Emas National Park (ENP), which holds one of the last jaguar populations in the Cerrado savanna of central Brazil. The Cerrado was identified as one of the earth’s 25 ecological hotspots (Myers et al., 2000) and is threatened by rapid and large-scale habitat loss. The biome has been neglected by conservation oriented research and few studies have previously investigated jaguar ecology and conservation status in this biome. We therefore deployed 119 camera trap stations across the entire 1320 km² of ENP – the largest single-site camera trapping study implemented for jaguars. Considering the sex-specific differences in behavior outlined above and the fact that differential space use between sexes has recently been documented in jaguars (Conde et al., 2010), we incorporated sex-specific parameters into the SECR model. We compared its results with those from a sex-specific but non-spatial approach. Our study not only provides novel information about jaguar ecology in the Cerrado, but has general implications for estimating population densities of large carnivores and other elusive species for most sampling designs.