Research Seminar - Hawkesbury Institute for the Environment

Seminar abstract: Many ecologists collect multiple response variables and wish to analyse them simultaneously (a “multivariate analysis”). The most common example, and the focus of this talk, is when collecting presence/absence or abundance information on multiple species, in order to make general statements about an assemblage (or community or ecosystem), and its response to environmental variables (or environmental impacts or experimental treatments). Standard practice in ecology until recently has been to analyse data using computational “shortcuts”, such as the distance-based analyses of PRIMER or the algorithms of CANOCO, to analyse multivariate data. This is a bit like Rick Astley though – sounded great in the ‘80’s, but now there’s some much better things we could be doing with our time! There has been a recent movement towards the use of statistical models for multivariate analysis in ecology, harnessing the power of modern computers to analyse data in a more flexible and interpretable way, and providing better statistical power than competitors in the ecology literature.

In this talk we will meet some model-based approaches to multivariate analysis, developed by the UNSW Eco-Stats group.

For more information on Associate Professor Warton, please visit: http://web.maths.unsw.edu.au/~dwarton/, or for the Eco-Stats Research Group, see: http://www.eco-stats.unsw.edu.au/

The seminar will be followed by drinks and nibbles available in the foyer of L9. All are welcome, no RSVP is required.