Applying Hidden Markov Models to Batch Marked Bee Data
Date
2024
Authors
Johns, Simon
Journal Title
Journal ISSN
Volume Title
Publisher
University of Victoria
Abstract
Batch marking is a specific style of capture-recapture study, where researchers tag all animals captured at the same time with same mark. While less informative, this is an efficient and cost effective alternative to standard capture-recapture methods where unique individual tags are used. This project applied hidden Markov models (HMMs) developed to analyze this style of data onto an Orchid bee dataset, in order to estimate total population size and survival rates. The weekly survival rate was estimated to be 0.63, and the population was estimated to grow from 267 to 710 individuals over an 18 week period. This aligned well with the expectations of the biologists who collected the data. Computational issues arose around choices for tuning parameters and convergence to good solutions, however these were addressed in a methodical way, providing a solid framework for the future use of HMMs on batch marked datasets.
Description
Keywords
batch marking, mark-recapture, Euglossa imperialis, orchid bees, abundance, hidden Markov model