Multivariate Analysis of Ecological Data
22 January - 2 February 2013
České Budějovice, Czech Republic
This course, 11 days long (with 9 working days), focuses on the modern approaches to analysing multivariate data and is designed for researchers and students in all fields of biology and nature protection. Course participants are expected to bring with them the results of their own projects and the course schedule offers sufficient time to apply the methods mastered during the course to their own datasets, including consultations offered by both lecturers.
We provide in-depth lectures and practicals on the following topics:
- Classical ordination methods: principal component analysis (PCA), correspondence analysis (CA), detrended correspondence analysis (DCA), metric and non-metric multidimensional scaling (PCO and NMDS)
- Constrained ordination methods: canonical correspondence analysis (CCA), redundancy analysis (RDA), including partial analyses (with covariates), stepwise selection of explanatory variables, variation partitioning, and Monte Carlo permutation tests of multivariate hypotheses
- Additional methods of interest to course participants are briefly discussed, namely the principal response curves (PRC) method, discriminant analysis (LDA), co-correspondence analysis (CoCA), and the approaches to inclusions of spatial and phylogenetic information or species functional traits in the analyses.
- Methods of visualizing the results of ordination methods using ordination diagrams, including thorough explanation of how the ordination diagrams should be interpreted
- In 2013 we start using the upcoming Canoco5 in the course, as this new version will be released to distribution about that time. You will find the new version integrates the work with data, statistical analyses and visualization of results into a single program, very easy to use even for advanced types of analysis. As part of the course, you will receive a trial version to use on your laptop during the course and for a limited time after returning to your institution.
- In addition, this course provides brief overviews of classification methods (cluster analysis and TWINSPAN), of experimental design, and of modern regression methods (GLM, GAM, regression trees).