Methodological Innovation in Long-Term Wildlife Population Monitoring: A Hierarchical Modeling Approach for Nigerian Savannas

Long-term wildlife population monitoring is essential for detecting trends, evaluating management effectiveness, and informing adaptive conservation decisions, yet the statistical methods applied in most Nigerian wildlife monitoring programs are inadequate for accounting for detection probability, spatial heterogeneity, and non-linear population dynamics. This dissertation advances methodological innovation in long-term wildlife population monitoring by developing and validating a suite of hierarchical occupancy and abundance models tailored to the ecological and logistical constraints of Nigerian savanna systems. Model development is conducted using 10 years of longitudinal monitoring data from Yankari Game Reserve (North East), Kainji Lake National Park (North Central), and Kamuku National Park (North West), comprising transect surveys, camera trap records, and aerial count data for 12 mammal species. Bayesian hierarchical models incorporating species, site, season, and observer random effects are developed and evaluated against simpler analytical approaches. Simulation studies demonstrate that models failing to account for detection probability generate systematic bias in trend estimation of between 15 and 40 percent depending on species and survey design. An original integrated population model combining aerial count and camera trap data for elephant and large antelope achieves substantially improved precision compared to either data source alone. The dissertation provides a practical modeling toolkit and survey design guidelines for Nigerian wildlife monitoring practitioners, contributing both methodological innovation and applied conservation science. Keywords: wildlife monitoring, hierarchical models, Bayesian analysis, Nigeria, population trend