PRIMER / PERMANOVA Essentials

Week 1 will cover the core non-parametric methods implemented in PRIMER software. Week 2 will cover the semi-parametric methods encapsulated in PERMANOVA+ and other advanced topics. Sessions will run 5 days per week, Monday to Friday, from 8:30 am to 5:30 pm each day. Participants may register for week 1 only, week 2 only, or both weeks. Online participants will join the course live in the Australian Central Daylight Time (ACDT) time zone (UTC +10.30 hours). Each day will include a mixture of lectures and computer lab sessions to implement the new methods and techniques learned on example datasets. Participants will have the opportunity to discuss and analyse their own data in consultation with the presenter. Those who register to attend in person are expected to bring their own laptop to the venue. A free fully functional (but time-limited) licence of the software will be made available to registered participants for trial use during the course. At the completion of the course, participants will receive an offer for a special discount on PRIMER software. Note that PRIMER is a Windows-only product, so Macs need to run in Windows emulation. This course will cater both to those who are new to PRIMER and to those who are familiar with PRIMER methods but would like a refresher regarding the latest techniques and approaches.

VENUE

This course 

WEEK 1 – PRIMER

Week 1 will provide an extensive overview of statistical methods in non-parametric analysis of multivariate data implemented in PRIMER software. Non-parametric statistics and permutation tests make the methods intuitively simple to understand, so no prior background in statistics is required.

The topics covered in this course shall include:

• Properties of multivariate data (summary stats, shade plots, histograms, draftsman plots, etc.);
• Pre-treatment of data (transformations, normalisations, standardisations, cumulate, aggregate, dispersion/variable weighting);
• Resemblance measures: distance, similarity/dissimilarity (Euclidean, Bray-Curtis, Jaccard, etc.);
• Cluster analysis (hierarchical agglomerative, divisive and k-R cluster methods using CLUSTER), including tests for significant structure within clusters (SIMPROF) to permit non-arbitrary classifications of samples/species;
• Ordination via Euclidean projection using principal component analysis (PCA);
• Ordination to preserve sample relationships via non-metric, metric or threshold-metric multi-dimensional scaling (nMDS, mMDS, tmMDS);
• Non-parametric permutation tests for differences among a priori groups of samples (analysis of similarities; ANOSIM) for one-, two- and three-way multi-factor designs and for ordered factors;
• Visualise differences among groups and variation of group averages using bootstrap averages;
• Relate biotic to abiotic data, including tests of association between resemblance matrices (RELATE), and find optimal subsets of environmental (or other) variables that generate a ‘best’ match to patterns among samples based on species variables (BEST);
• Measures of biodiversity, including richness, even-ness/dominance, abundance-biomass curves, and taxonomic and/or functional diversity and distinctness (DIVERSE);
• Identify important subsets of species and their roles in generating overall multivariate patterns (BIOENV, coherence curves) or group/cluster differences (SIMPER);
• Identify roles of environmental (or other) variables in generating biotic patterns or distinctive splits using non-parametric classification and regression trees (LINKTREE);
• Graphical tools for effective presentation of results, including matrix displays and a variety of plot types (bar, box, dot, violin, means, line, histogram, scatter, surface, shade), animations of ordinations captured to video files, bubble plots and multi-variable segmented bubble plots.