Extending the Latent Multinomial Model with Complex Error Processes and Dynamic Markov Bases

TL;DR

This paper extends the latent multinomial model for mark-recapture data by incorporating complex error processes and dynamic Markov bases, improving modeling flexibility and computational feasibility for studies with error-prone identification methods.

Contribution

It introduces a novel extension of the LMM that separates the capture and error processes and develops a dynamic Markov basis MCMC scheme for complex error models.

Findings

Dynamic Markov bases enable efficient sampling for complex error models.

Simulation shows improved accuracy in survival estimates with the extended model.

Application to snake data demonstrates practical utility.

Abstract

The latent multinomial model (LMM) model of Link et al. (2010) provided a general framework for modelling mark-recapture data with potential errors in identification. Key to this approach was a Markov chain Monte Carlo (MCMC) scheme for sampling possible configurations of the counts true capture histories that could have generated the observed data. This MCMC algorithm used vectors from a basis for the kernel of the linear map between the true and observed counts to move between the possible configurations of the true data. Schofield and Bonner (2015) showed that a strict basis was sufficient for some models of the errors, including the model presented by Link et al. (2010), but a larger set called a Markov basis may be required for more complex models. We address two further challenges with this approach: 1) that models with more complex error mechanisms do not fit easily within the LMM and 2) that the Markov basis can be difficult or impossible to compute for even moderate sized studies. We address these issues by extending the LMM to separately model the capture/demographic process and the error process and by developing a new MCMC sampling scheme using dynamic Markov bases. Our work is motivated by a study of Queen snakes (Regina septemvittata) in Kentucky, USA, and we use simulation to compare the use of PIT tags, with perfect identification, and brands, which are prone to error, when estimating survival rates.