Bayesian integrated population modelling (IPM) using JAGS
Res Altwegg, University of Capetown, South Africa
Date: 20 – 24 June 2022
Venue: University of Capetown, Capetown, South Africa
Computers: Bring your own laptop with latest R and JAGS
Registration: 600 US$ (normal rate), 300 US$ (student rate)
Integrated population models (IPMs) represent the powerful combination, in a single Leslie-type of model, of multiple data sources that are informative about the dynamics of an animal population (Besbeas et al. 2002). Typical IPMs combine one or more time-series of counts with other data sets that are directly informative about survival probabilities, such as ring-recovery or capture-recapture, or productivity, such as nest survey data. However, many other sources of demographic information may be envisioned instead or in addition, including age-at-death data, radio tracking data, occupancy or replicated point count data. For non-statisticians the only practical manner to develop and fit IPMs is by using BUGS software (JAGS and Nimble).
This course is a practical and hands-on introduction to developing and fitting integrated population models using BUGS software. It is based on the brand-new book by Schaub & Kéry, Integrated Population Models (Academic Press, 2022), which will be handed out as part of the course.
Beyond IPMs, the course also provides an in-depth introduction for ecologists and wildlife managers to a wide variety of models fit using BUGS software and as documented in the IPM book.
Contents include the following topics:
- Basic introduction:
- Hierarchical models as an overarching theme of population modelling, including IPMs
- Bayesian analysis of hierarchical models
- Introduction to BUGS software in the context of generalised linear models (GLM) and traditional random-effects models
- Ingredients of Integrated Population Models:
- State-space models
- Cormack-Jolly-Seber models for estimating survival probabilities
- Multistate capture-recapture models for estimating survival and transition probabilities
- Integrated Population Models (IPMs)
- Introduction to matrix population models and their analysis with BUGS
- Theory of integrated population models
- Various case studies which differ in complexity and in the data types that are combined
In this intermediate-level workshop about 80% of the time is spent on lecturing and 20% on solving exercises. No previous experience with BUGS software, or Bayesian statistics, is assumed. However, a good working knowledge of modern regression methods (linear models, GLMs) and of program R is required. Moreover, a basic understanding of capture-recapture and/or occupancy models is desirable.