Multivariate varying-coefficient models for an ecological system
Article first published online: 20 AUG 2008
Copyright © 2008 John Wiley & Sons, Ltd.
Volume 20, Issue 4, pages 460–476, June 2009
How to Cite
Ferguson, C. A., Bowman, A. W., Scott, E. M. and Carvalho, L. (2009), Multivariate varying-coefficient models for an ecological system. Environmetrics, 20: 460–476. doi: 10.1002/env.945
- Issue published online: 11 MAY 2009
- Article first published online: 20 AUG 2008
- Manuscript Accepted: 6 JUL 2008
- Manuscript Received: 20 AUG 2007
- Department of Statistics, University of Glasgow
- Eurolimpacs Project
- European Union under Thematic Sub-Priority 18.104.22.168 ‘Global Change and Ecosystems’ of the 6th Framework Programme
- Loch Leven;
- varying-coefficient model
The monitoring of Loch Leven at Kinross in Scotland involves many physical, chemical and biological variables, which are interrelated and consequently are potentially both responses and covariates within the system. In order to explore relationships fully and make appropriate inferences about such a system, it is necessary to identify and incorporate all the important forms of dependence. Therefore, multivariate models are required to model multiple responses and covariates. Relationships between variables at Loch Leven also change throughout the year and hence a multivariate varying-coefficient model is presented here to explore changing relationships, whilst accounting for dependence between responses. Copyright © 2008 John Wiley & Sons, Ltd.