Abstract
Effective conservation and management of cryptic carnivores require reliable estimates of population density, robust genetic tools for monitoring, and accurate characterization of predator diets. In this dissertation, I evaluated population ecology and trophic interactions of bobcats (Lynx rufus) and cougars (Puma concolor) on Washington’s north Olympic Peninsula using noninvasive genetic sampling, camera trapping, and molecular diet analysis. Together, these chapters provide a comprehensive assessment of methodological performance as well as novel ecological insights for two sympatric felids of management concern.In Chapter 1, I compared three spatial modeling frameworks – spatial capture-recapture (SCR), genotype spatial partial identity models (gSPIM), and the space-to-event (STE) camera model – to estimate densities of bobcats and cougars across a ~600 km² working-forest landscape. Using >600 scat samples collected by detection dogs and a systematic grid of 74 unbaited cameras, I found that SCR and STE produced comparable density estimates when camera viewsheds and temporal windows were well characterized, whereas gSPIM yielded higher but less precise estimates due to incorporation of partial genotypes. For cougars, STE and SCR credible intervals overlapped across years, with strongest agreement at a 3-second snapshot window; bobcat model convergence improved substantially in the second year. These results demonstrate that multiple noninvasive spatial frameworks can deliver consistent density estimates when sampling designs are carefully implemented, and they highlight opportunities to integrate genetic and camera data to improve precision and reduce field costs.
Chapter 2 addresses a major logistical barrier to noninvasive genetics: the degradation of fecal DNA. I conducted a factorial experiment testing two preservation buffers (DETs and DNA/RNA Shield), three DNA extraction kits, and two storage durations. Species-level mtDNA identification success was uniformly high (98%), but individual identification success varied widely. DNA/RNA Shield outperformed DETs for nuclear DNA amplification, producing higher PCR success probabilities (up to 0.99), lower allelic dropout, and fewer false alleles. The Quick-DNA Fecal/Soil Microbe Miniprep Kit consistently produced the highest-quality genotypes, particularly after extended storage. These findings provide practical, evidence-based recommendations for optimizing fecal DNA workflows, improving genotyping reliability, and reducing costs for long-term carnivore monitoring programs.
In Chapter 3, I paired DNA metabarcoding of 89 genetically identified scats with 75 cougar GPS-cluster investigations to evaluate carnivore diet composition and compare how field-based and molecular methods detect prey. Both methods identified Columbian black-tailed deer (Odocoileus hemionus columbianus) as the dominant cougar prey, and dietary overlap between cougar metabarcoding and field clusters was high (Schoener’s D = 0.81). Metabarcoding detected small-bodied prey (e.g., grouse, squirrels) not observed at kill sites, whereas cluster investigations detected large-bodied prey such as Roosevelt elk (Cervus canadensis roosevelti) and American beaver (Castor canadensis) underrepresented in scats. Bobcat diets, assessed solely via metabarcoding, were dominated by small mammals and exhibited low overlap with cougar diets (D = 0.28). Both predators consumed mountain beaver (Aplodontia rufa) – an ecologically and economically important species – at low frequencies. These results demonstrate that metabarcoding and GPS clusters offer complementary strengths and together provide a more complete depiction of carnivore diet composition and prey diversity than either method alone.
Collectively, this dissertation advances noninvasive genetic and camera-based approaches for wildlife monitoring, provides rigorous evaluations of methodological performance, and offers new insights into density, diet, and trophic ecology of sympatric carnivores in working forest landscapes.