MAR Workshop ESA 2007, San Jose, CA


5 children total

MAR Workshop ESA 2007, San Jose, CA



Random Pursuits home
iugo-cafe home
Version 0.4

MAR Workshop ESA 2007, San Jose, CA

An introduction to the analysis of community time series using Multivariate Autoregressive (MAR) models
Workshop #6 2007 Ecological Society Annual Meeting, San Jose, CA

Ecologists commonly quantify abundances of multiple interacting species and associated abiotic variables in study sites over time. Monitoring programs, shorter term field studies, and experimental studies all may generate such multispecies and multivariate time series. In spite of the prevalence of this type of data collection, most ecologists have not been trained in time series analyses appropriate to these community time series data. Temporal autocorrelation violates the independence assumptions of common statistical tools, and direct and indirect effects complicate interpretations. To cope, many researchers aggregate their data to a level that minimizes such “problems” but also discards a great deal of systemic detail that could be useful to researchers and managers. Multispecies (or Multivariate) Autoregressive (MAR) modeling is a versatile time series modeling approach that allows users to address many common questions in ecology, such as: 1) Which abiotic and biotic factors best predict the abundance of various taxa?, 2) What are the relative strengths of abiotic and biotic interactions between community members?, 3) How stable is the community structure? In this workshop, we will give participants a basic introduction to autoregressive time series modeling in general and to multivariate autoregressive modeling in particular. Specific case studies where MARs have been used to analyze species interactions, environmental drivers and community stability in aquatic ecosystems will be presented to give concrete examples of how MARs can be used to address ecological questions. In the afternoon, participants will be introduced to LAMBDA a user-friendly MAR analysis toolbox and we will walk through the analysis of sample datasets. By the end of the workshop, participants should have a working knowledge of multivariate autoregressive models, understand the steps required during application of these methods, and be able to do a basic analysis using LAMBDA. The organizers assume no programming knowledge for participants. A working knowledge of multiple regression will be helpful but not necessary. Participants should bring laptops, and software will be loaded on their machines in the first half hour of the workshop.

Created on Apr 25, 2007 at 02:37:05 PM by eli

No comments