Analysing multivariate ecological data with Generalized Linear Latent Variable Models: 2-part workshop

Bert van der Veen, Norwegian Institute of Bioeconomy Research
Robert B. O’Hara, Norwegian University of Science and Technology
Sam Perrin, Norwegian University of Science and Technology
Jenni Niku, University of Jyväskylä

Participants in this workshop will learn how to explore co-occurrence patterns in species distributions, using a modeling framework that combines the best of ordination methods and Joint Species Distribution Models. This framework, Generalized Linear Latent Variable Models (GLLVMs), unifies classical community and species distribution modelling. On both days, the workshop will cover basic theory around GLLVMs, and provide hands-on experience with examples and exercises using the R-package “gllvm”.

Part 1: We will focus on performing model-based ordination with GLLVMs, and outline the connection with classical ordination methods such as PCA and CA. We will demonstrate how to use model-based ordination to make inferences about species communities.

Part 2: We will focus on using GLLVMs as Joint Species Distribution Models. We will demonstrate how to produce maps from predictions made using GLLVMs, and how to infer species associations from these models. We will shortly outline how the methods we present can be extended in the R-package “HMSC”.

The material is available on github: https://github.com/BertvanderVeen/BES2020GLLVMworkshop.