Wildlife Abundance Estimation Via Bayesian Hierarchical Model = 베이지안 계층 모형을 이용한 야생동물 개체 수 추정

국문 초록 (Abstract)

생태학에서는 특정 지역 내 야생동물 개체수를 정확히 추정하는 것이 중요한 과제이다. Spatial capture–recapture(SCR) 모형은 카메라 트랩(Camera-trap) 데이터를 이용하여 개체의 위치 정보를 통합함으로써 이러한 추정에 널리 활용된다. 본 연구에서는 설악산 인근 지역의 산양 데이터를 대상으로, 개선된 Reversible Jump MCMC(RJMCMC) 알고리즘을 적용하여 개체수를 추정하였다. 또한, 모형 적합 과정에서 발생하는 식별성(identifiability) 문제를 분석하였다. 본 연구는 공간적 탐지 모형의 식별성 문제를 탐색적으로 고찰하고, 보다 안정적인 모수 추정에 기여하는 데 목적이 있다.

다국어 초록 (Multilingual Abstract)

Estimating wildlife population abundance within a defined region is a fundamental task in ecological research. Spatial capture–recapture (SCR) models have been widely used for this purpose, as they explicitly incorporate the spatial relationship between individual activity centers and detection locations using camera-trap data. In this study, we applied an improved Reversible Jump Markov chain Monte Carlo (RJMCMC) algorithm to estimate the population size of gorals in a region near Seoraksan National Park. In addition, we systematically investigated identifiability issues that arise during the model-fitting process. The aim of this study is to provide an exploratory examination of identifiability problems in spatial detection models and to offer methodological insights for achieving more stable parameter estimation.

목차 (Table of Contents)

• Abstract i
• Contents ii
• List of Tables iv
• List of Figures v
• INTRODUCTION 1
• METHODS 4
• 2.1 Spatial capture-recapture model (SCR) 4
• 2.2 SCR inference via MCMC (Chandler and Royle, 2013) 6
• 2.3 SCR inference via RJMCMC (Herliansyah et al., 2025) 7
• 2.3.1 Reversible Jump Markov Chain Monte Carlo 8
• 2.3.2 Algorithm Justification 11
• 2.3.3 Implementation Details 12
• 2.4 Gelman-Rubin Statistic 14
• Numerical Simulation 16
• 3.1 Validation : Northern Parula Dataset 16
• 3.2 Simulation study 17
• Real Data Analysis 20
• 4.1 Data Description 20
• 4.2 Priors specification 21
• 4.3 Results 22
• Discussion 24
• 5.1 Comparison with Chandler and Royle (2013) 24
• 5.2 Non-identifiability Problem 26
• 5.2.1 Non-identifiability between λ₀ and N 27
• 5.2.2 Weak-identifiability between λ₀ and σ 27
• 5.2.3 Weak-identifiability between N and σ 31
• Conclusion 34
• Abstract (In Korean) 37