Statistical and modelling methods for assessing the relationships between ecological and chemical status in lakes. REBECCA Deliverable 12

Abstract

This report is deliverable no. 12 of the EU FP 6 project REBECCA (Relationships between ecological and chemical status of surface waters). It describes the statistical approaches and modelling methods used within Workpackage 3 (Lakes). An important purpose of the project has been to develop indicators for ecological status, and to de-scribe and quantify the relationships between chemical pressure gradients and the biological indica-tors. Within Workpackage 3 (Lakes) we have developed indicators and analysed relationships for four main taxonomic groups: phytoplankton, macrophytes, macroinvertebrates and fish. This report describes the statistical methods used for both indicator development and relationships analyses, divided into three types of approaches. (1) Univariate methods: mainly used for describing relationships, after the response variable has been developed. (2) Multivariate methods: mainly used for data exploration and for deriving univariate response variables from multivariate community data. (3) Bayesian methods: particularly useful for analysing complex ecological systems, because hierarchical models and network models can be more easily constructed and analysed within a Bayesian framework.