A Hellinger distance approach to MCMC diagnostics

Boone, Edward L.; Merrick, Jason R. W.; Krachey, Matthew J.

doi:10.1080/00949655.2012.729588

Cited by 34 publications

(28 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Convergence and mixing were checked by visual inspection of the resulting trace files for all estimated parameters, and by calculating the effective sample sizes (i.e., the number of independent samples) and the Hellinger distance (Boone et al, 2014) between sets of 1000 batched recorded samples. Effective sample sizes were generally larger than 1000 for all parameters, and…”

Section: Mcmc Modelmentioning

confidence: 99%

The fitness landscape of the codon space across environments

Fragata

Matuszweski

Schmitz

et al. 2018

Preprint

View full text Add to dashboard Cite

show abstract

Section: Mcmc Modelmentioning

confidence: 99%

The fitness landscape of the codon space across environments

Fragata

Matuszweski

Schmitz

et al. 2018

Preprint

View full text Add to dashboard Cite

show abstract

“…We checked the Hellinger distance statistic (Boone et al. ) to confirm model convergence. Data and the JAGS model code are provided at https://github.com/cbrown5/BenthicLatent.…”

Section: Methodsmentioning

confidence: 99%

“…We used only a single chain to avoid issues with sign switching on the latent variables (Hui 2016). We checked the Hellinger distance statistic (Boone et al 2014) to confirm model convergence. Data and the JAGS model code are provided at https://github.com/cbrown5/BenthicLatent.…”

Section: Case Studymentioning

confidence: 99%

Estimating the footprint of pollution on coral reefs with models of species turnover

Brown

Hamilton

2018

Conservation Biology

View full text Add to dashboard Cite

Ecological communities typically change along gradients of human impact, although it is difficult to estimate the footprint of impacts for diffuse threats such as pollution. We developed a joint model (i.e., one that includes multiple species and their interactions with each other and environmental covariates) of benthic habitats on lagoonal coral reefs and used it to infer change in benthic composition along a gradient of distance from logging operations. The model estimated both changes in abundances of benthic groups and their compositional turnover, a type of beta diversity. We used the model to predict the footprint of turbidity impacts from past and recent logging. Benthic communities far from logging were dominated by branching corals, whereas communities close to logging had higher cover of dead coral, massive corals, and soft sediment. Recent impacts were predicted to be small relative to the extensive impacts of past logging because recent logging has occurred far from lagoonal reefs. Our model can be used more generally to estimate the footprint of human impacts on ecosystems and evaluate the benefits of conservation actions for ecosystems.

show abstract

“…We used only a single chain to avoid issues with parameter switching on the latent variables (Hui 2016). We checked the Hellinger distance statistic (Boone et al 2014) to confirm model convergence. Data are provided in Appendices S1-3, JAGS model code is provided in Appendix S4.…”

Section: Estimating the Footprint Of Pollution On Coral Reefs Using Mmentioning

confidence: 99%

Estimating the footprint of pollution on coral reefs using models of species turn-over

Brown

Hamilton²

2017

Preprint

View full text Add to dashboard Cite

Ecological communities typically change along gradients of human impact, though it is difficult to estimate the footprint of impacts for diffuse threats like pollution. Here we develop a joint model of benthic habitats on lagoonal coral reefs and use it to infer change in benthic composition along a gradient of distance from logging operations. The model estimates both changes in abundances of benthic groups and their compositional turn-over, a type of beta-diversity. We detect compositional turnover across the gradient and use the model to predict the footprint of turbidity impacts from logging. We then apply the model to predict impacts of recent logging activities, finding recent impacts to be small, because recent logging has occurred far from lagoonal reefs. Our model can be used more generally to estimate the footprint of human impacts on ecosystems and evaluate the benefits of conservation actions for ecosystems.

show abstract

A Hellinger distance approach to MCMC diagnostics

Cited by 34 publications

References 29 publications

The fitness landscape of the codon space across environments

The fitness landscape of the codon space across environments

Estimating the footprint of pollution on coral reefs with models of species turnover

Estimating the footprint of pollution on coral reefs using models of species turn-over

Contact Info

Product

Resources

About