Spatial Capture-recapture with Memory: A New Hidden Markov Model Perspective
Lead Research Organisation:
University of St Andrews
Department Name: Mathematics and Statistics
Abstract
Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
Organisations
| Description | Within this project we have focused on the estimation of abundance (and density) of difficult to observe species using data collected via an array of motion-sensor camera traps, typically referred to as spatial capture-recapture data. We demonstrated via simulating data that current state-of-the-art models for analysing such data can lead to over-estimates of the population size. This is because they do not account for the fact that it is more likely that an animal is next detected at a trap close to its last detection than at a trap farther away. We developed a new statistical model that is able to incorporate this "spatial memory" to reflect the movement of animals, resulting in more accurate population estimates and better estimates of their associated activity centres. The new approach is applied to pine Martens and demonstrate a substantially better fit to the observed data than the traditional (non-memory) spatial capture model. |
| Exploitation Route | We have developed a new innovative statistical approach for analysing spatial capture-recapture data and are currently preparing a manuscript for submission as well as disseminating the research via seminar/conference presentations. The initial results suggests that substantial gains may be made via the use of spatial time-to-event models. The approach has wide application within the ecological community for the analysis of such data, and opens up additional new and promising avenues for future research that we plan to investigate. The methods and associated software will be highly useful for ecologists and conservationists in their application to a wide range of wildlife populations using camera trap survey data. We anticipate that publishing the associated manuscript of this work will also lead to other researchers further developing the methods and applying the approach to datasets. |
| Sectors | Environment Other |