Modelling Biological Communities: Multi-Species Occupancy Modelling with R and JAGS
Instructors: Marc Kéry, Swiss Ornithological Institute
Mike Meredith, Wildlife Conservation Society
Ngumbang Juat, Biodiversity Conservation Society Sarawak
Date: 12–18 January 2017 (6 days with a break on Sunday)
Venue: Kuala Lumpur, Malaysia: Akademi Hevea, part of the Rubber Research Institute
complex, Sungai Buloh, close to Subang airport
For other information, including costs, registration etc. see http://bcss.org.my/index.htm
Single-species occupancy modelling is a well established method to estimate presence-absence when the species may be present but not detected. It allows the actual presence-absence process and the detection process to be modelled and estimated separately. Both can be modelled as functions of covariates, and it is possible to model separate effects of one covariate on presence-absence and on detection; for instance, a species may be more common in open habitats and also more detectable there.
This approach has now been extended to simultaneously model occupancy and detection for multiple species across multiple sites. MSOMs (or community occupancy models) allow the number of species at each site to be estimated, taking account of imperfect detection, and the number of species not detected at any site can also be estimated.
Modelling is done in a Bayesian framework, resulting in posterior distributions for all the parameters estimated, including species richness and distance metrics (eg, Jaccard or Sørensen indices). These can be summarized as point estimates with credible intervals.