Apologies for the delay between updates, but it's been a busy time of year!
At the end of August, Darryl embarked on a 4-week occupancy modelling course world tour, after being invited to deliver courses in Boulder City (NA, USA), Ankara (Turkey) and Ulaanbaatar (Mongolia)! While this was a return to Boulder City for Darryl, it was his first visit to Turkey and Mongolia. The above image is of the Mongolian parliament building in Ulaanbaatar, with a statue of Chinggis Khan in the centre, or Genghis Khan as he's known in the West.
As is usually the case, the locals were welcoming and the course participants were enthusiastic to learn about the basics of occupancy modelling. A personal highlight for Darryl was a quick trip to the scene of the Gallipoli campaign (Turkey) from World War 1, which was a defining event in the national identities of New Zealand and Australia, and commemorated in both countries with ANZAC Day. A personal lowlight was teaching in Mongolia with a stomach bug; a first that he's not keen to repeat! Anyway, a few more 'dots on the map' for where Darryl has taught occupancy modelling courses since 2006 (see below).
Darryl has also been putting the final touches to a manuscript on a species cooccurrence dynamics model (SCDM), that uses a log-linear modelling parameterisation to account for interactions between species. The underlying modelling framework is the same as that published in the 1st and 2nd editions of Darryl's occupancy modelling book, and the parameterisation is an extension of one of the ones outlined for the single season situation in the 2nd edition, and also used by Rota et al. (2016). Conceptually, the approach treats the presence of each species in seasons t-1 and t as covariates, and cooccurrence relationships are investigated by estimating the magnitude of the interaction term between these covariates. With this approach, any interactions are 'symmetric' (apply equally for each species), and it's conceptually simple to extend to many species (but harder in practice). The parameterisation can be applied within either a maximum-likelihood or Bayesian estimation framework. Will update you once the paper has advanced through the publication process!