Occupancy models in BUGS/JAGS and unmarked
Instructors: Marc Kéry & Andy Royle, Swiss Ornithological Institute & USGS Patuxent Wildlife Research Center, Laurel/MD
Date: 12–14 November 2013
Venue: USGS Patuxent Wildlife Research Center, Laurel/MD, USA
Costs: 300 US$ (graduate student rate), 400 US$ (all others), 2 free seats available, see (1).
This workshop covers several key models used in the spatio-temporal analysis of distribution (occupancy modeling) at large spatial scales (e.g., for monitoring data), both in a likelihood and in a Bayesian analysis framework. As a unique feature of this workshop, modern community models (=multi-species occupancy models or Dorazio-Royle models; see below) and spatial models (CAR, autologistic) are covered. We use programs R (especially package unmarked) and WinBUGS and JAGS for model fitting. The course includes:
• Introduction to distribution and abundance in ecology and applied fields (e.g., conservation, biodiversity and disease monitoring)
• Introduction to hierarchical models
• Frequentist and Bayesian analysis of statistical models
• Single-season occupancy models (MacKenzie et al., Ecology, 2002) for statistical mapping and modeling of static distributions
• Dynamic occupancy models (MacKenzie et al., Ecology, 2003) for analysis and mapping of species range dynamics or changing distributions
• Static community models (multi-species occupancy models; Dorazio & Royle, JASA, 2005;
Dorazio et al., Ecology, 2006)
• Dynamic community models (Dorazio et al., Ecology, 2010)
• Spatial occupancy models: conditional autoregressive (CAR) spatial smoothing and autologistic modeling (e.g., Bled et al., Ecol.Appl., 2011)
This is an intermediate-level workshop, though previous experience with programs BUGS or JAGS, or Bayesian statistics, is not assumed (but it would be helpful). However, a good working knowledge of modern regression methods (i.e., of linear and generalised linear models) and of program R is required.
Please bring your own laptops and install a recent version of R plus WinBUGS 1.4., with upgrade patch and immorality key decoded (in this order). Alternatively, JAGS and OpenBUGS work fine for almost everything. Also install the R packages unmarked, arm, raster, spdep, gstat and rgdal (and their dependencies).
Send your application to Andy Royle (email@example.com), describing your background and knowledge in statistical modeling, R and WinBUGS/OpenBUGS/JAGS, by 31 October 2013 at the latest. Workshop invitations will be sent out on a first-come-first-serve basis.
(1) You can apply for your course fee to be waived. Please describe in a short email your situation and how the course would benefit your work and the state of the universe.