Browsing by Subject "Bayesian inference"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Publication
    Embargo
    Disentangling generic limits in Chionoloma, Oxystegus, Pachyneuropsis and Pseudosymblepharis (Bryophyta: Pottiaceae): an inquiry into their phylogenetic relationships
    (Wiley Online Library, 2016-05-08) Nylinder, Stephan; Hedenas, Lars; Alonso García, Marta; Cano Bernabé, María Jesús; Jiménez Fernández, Juan Antonio; Biología Vegetal
    Pottiaceae is the largest known moss family, with one of the most complex taxonomies among the bryophytes. The circumscription of genera within the Pottiaceae is challenging. Here, we elucidate the relationships among four related traditional genera of the Pottiaceae: Chionoloma, Oxystegus, Pachyneuropsis and Pseudosymblepharis, all sharing a complex taxonomic history. In order to resolve phylogenetic relationships among the four genera, a phylogeny derived from nuclear ITS and the plastid markers atpB-rbcL, trnG and trnL-F is inferred. Putative monophyly of these four genera is investigated using maximum likelihood and Bayesian inference analyses. Ancestral state reconstruction shows high levels of homoplasy in the characters historically used for the generic division of Chionoloma s.1. Based on our results, we suggest that Chionoloma, Oxystegus and Pseudosymblepharis should be merged into a single genus, for which the oldest name Chionoloma has priority. Additional analyses are needed to clarify the taxonomic status of Pachyneuropsis. New combinations are provided for those species where required, and lectotypes are designated for five names.
  • Thumbnail Image
    Publication
    Open Access
    Reducing Monte Carlo error in the Bayesian estimation of risk ratios using log-binomial regression models
    (Wiley, 2015) Salmerón, D.; Cano, J.A.; Chirlaque López, María Dolores; Ciencias Sociosanitarias
    In cohort studies binary outcomes are very often analyzed by logistic regression. However, it is well-known that when the goal is to estimate a risk ratio, the logistic regression is inappropriate if the outcome is common. In these cases, a log-binomial regression model is preferable. On the other hand, the estimation of the regression coefficients of the log-binomial model is difficult due to the constraints that must be imposed on these coefficients. Bayesian methods allow a straightforward approach for log-binomial regression models, produce smaller mean squared errors in the estimation of risk ratios than the frequentist methods, and the posterior inferences can be obtained using the softwareWinBUGS. However,Markov chainMonte Carlo (MCMC) methods implemented inWinBUGS can lead to largeMonte Carlo errors in the approximations to the posterior inferences since they produce correlated simulations and the accuracy of the approximations are inversely related to this correlation. To reduce correlation and to improve accuracy, we propose a reparameterization based on a Poisson model and a sampling algorithm coded in R.