Improved methods for estimating abundance and related demographic parameters from mark-resight data

Area of Science:

• Ecology and Evolutionary Biology
• Wildlife Population Monitoring
• Quantitative Biology

Background:

• Mark-resight methods are valuable for estimating wildlife abundance, offering a less invasive alternative to mark-recapture.
• Existing models struggle with imperfect detection and unknown marked sample sizes, leading to biased estimates.
• Individual detection heterogeneity is a common challenge in mark-resight studies.

Purpose of the Study:

• To evaluate deficiencies in current mark-resight models within Program MARK.
• To propose and validate a new composite likelihood estimator for improved abundance estimation.
• To extend mark-resight models to incorporate individual random effects under Pollock's robust design.

Main Methods:

• Developed a zero-inflated Poisson log-normal composite likelihood model.
• Implemented and tested the new estimator against existing models in Program MARK.
• Extended models to include individual-level random effects for sampling occasions.

Main Results:

• The proposed zero-inflated Poisson log-normal estimator demonstrated superior performance, reducing bias and improving confidence interval coverage.
• The new models provide a more realistic approach by accommodating individual random effects.
• Analysis of New Zealand Robin data confirmed the practical utility of the proposed estimators.

Conclusions:

• The zero-inflated Poisson log-normal mark-resight estimator is recommended for low encounter rates, non-negligible detection heterogeneity, and unknown marked sample sizes.
• The developed methods enhance the accuracy and reliability of population abundance estimation in ecological research.
• Practitioners should consider implementing the new estimator in Program MARK for robust mark-resight analyses.