In this video Francis Hui introduces boral, a new R package he developed for Bayesian analysis of multivariate data in ecology. It uses Bayesian MCMC estimation to fit latent variable models for unconstrained ordination (read the MEE paper, Model-based approaches to unconstrained ordination, for details), and for multi-species inference while accounting for inter-species correlation:

• Download boral from CRAN
• Read the MEE paper Model-based approaches to unconstrained ordination