Dealing with latent discrete parameters in StanHiroki It?
?
Stan can handle latent discrete parameters in population ecology models by marginalizing them out. This allows estimating quantities like population size and transition probabilities. Stan formulations for capture-recapture and multistate models use matrix calculations in hidden Markov models. Stan shows faster computation and better sampling than OpenBUGS for these examples, though the models can be more complex to specify in Stan.
- Coppices under active management can support more plant species diversity on the forest floor compared to abandoned forests. A shifting mosaic of stands under different stages of coppice rotation appears best for maintaining stable plant species abundance over time.
- Mass mortality of oak trees caused by fungi and beetles has expanded in Japan since the 1980s. This is indirectly linked to abandonment of traditional Satoyama woodland management. Regeneration of tree species is limited by dense undergrowth in gaps created by oak mortality.
- Deer browsing threatens regeneration, even within deer-proof fences, and adds costs to coppice woodland management.