Instructors: Marc Kéry & Andy Royle, Swiss Ornithological Institute & USGS Patuxent Wildlife Research Center
Date: 9–13 November 2015
Venue: Patuxent Wildlife Research Center, Laurel/MD
Course fee: USD$550 (normal rate), USD$350 (student rate)
The analysis of abundance and of the dynamic rates governing their change lies at the core of ecology and its applications such as conservation and wildlife management. Meta-population designs, where repeated measurements of some quantity such as counts or distance measurements are made at a collection of sites, underlie a vast number of studies in ecology and management. Inference about such data is conveniently based on hierarchical models, where one submodel describes the underlying true state of the process (e.g., abundance at a site) and another submodel describes the observation process that connects the true state to the observations.
In recent years, much progress has been made in the development of methods and computer algorithms to fit hierarchical models. In particular, Bayesian statistical analysis and the general-purpose Bayesian software packages BUGS and JAGS have revolutionized the ways in which ecologists can conduct complex population analyses. On the other hand, the R package unmarked contains a wealth of functions for a frequentist analysis of hierarchical models of abundance.
This course introduces key hierarchical models used in the analysis of abundance and survival and their spatial and temporal patterns, and provides both Bayesian and frequentist methods for their analysis. We use packages unmarked and wiqid and especially WinBUGS, OpenBUGS and JAGS to fit and understand some of the most widely used models for the analysis of animal and plant populations. These include:
- Binomial (Royle 2004) and multinomial N-mixture models (Dorazio et al. 2005,
Royle et al. 2007)
- Conventional distance sampling and Hierarchical distance sampling (e.g., Royle et al. 2004, Sillett et al. 2012)
- Dynamic models of abundance for replicated counts (Dail & Madsen 2011) or distance sampling data (Sollmann et al. 2015)
- Cormack-Jolly-Seber (CJS) models, especially hierarchical CJS models for variation in survival over space (Saracco et al. 2010) or in a community of species
This is an intermediate-level workshop with about 80% spent lecturing and 20% on solving exercises. A working knowledge of modern regression methods (GLMs, mixed models) and of program R is required. Previous experience with the BUGS language is beneficial.
Please bring your own laptops and install a recent version of R, with the latest version of package unmarked, plus JAGS and/or WinBUGS 1.4. OpenBUGS works for most of what we do.
Please apply here by 1 October 2015: