While there are many scientifically rigorous methods for monitoring wildlife populations and the threats that they face, they are often difficult to implement in tropical forest environments. In some cases traditional gold standard methodologies can be employed despite the inherent practical and theoretical challenges, but in other situations more novel approaches must be developed. In this thesis we investigate these issues within the context of a large protected area in Eastern Cambodia.
The aims of this study were to; 1. Evaluate the status and trends of wild ungulate populations using distance sampling derived density estimates. 2. Develop and implement an approach to reliably estimate the detectability and abundance of wire snares, which currently represent the greatest threat to mammal populations within the area. 3. Quantify the association between snare abundance and a number of natural and anthropogenic factors hypothesised to influence snare placement. 4. Assess the utility of law enforcement records, and specifically catch-per-unit-effort (CPUE) indices derived from patrol data, as a tool for monitoring threats.
I present rigorous density estimates for several key ungulate species, representing the first such data from the entire lower Mekong region. Whilst smaller ungulate populations appear to be stable, larger species are likely undergoing a decline. A sampling protocol was developed for surveying snares which balanced the requirements of statistical rigour against feasibility and efficiency of implementation in the field. The results of this survey were analysed using N-mixture models to produce detectability-corrected spatially explicit estimates of snare abundance. As predicted, forest type, proximity to settlements, and distance to the Vietnamese border were shown to be important determinants of snare abundance whereas the relationship between snaring levels and both patrol effort and wildlife densities was less clear. This study also demonstrated that while CPUE indices derived from patrol data can adequately reflect true levels of threat, their utility depends greatly on the quality of the patrol data, and on identifying the appropriate spatio-temporal scale at which to undertake the analysis.