Working Group "Biometrical Methods in Agriculture"

 

 

Speaker: 
Dr. Kristina Warnstorff 
AG Biometrie u. Agrarinformatik 
Landw. Fakultät 
Martin-Luther-Universität 
06099 Halle / S 
Tel: +49 (0) 345/5522692 
Fax: +49 (0) 345/5527128 
e-mail: warnstorff@landw.uni-halle.de

 

The working group "Biometrical Methods in Agriculture" co-operates with the German Phytomedical Society (working group Biometry and Experimental Design), German Society of Agronomy (working group Biometry) and the "Gesellschaft für Pflanzenzüchtung" (working group Biometry).

During the last two years three joint meetings were organized: one session during the Biometrical Colloquium in Rostock (2000) and two summer meetings which took place in Bad Lauchstädt (2000) and Gülzow (2001).

At these three meetings the main emphasis was put on the analysis of series of field trials: experiments across several locations and/or years.

In Rostock the analysis of two factorial balanced trials was demonstrated, focusing on several designs and using mixed linear models with and without S -restrictions on the interaction effects (Richter, Guiard, Krüger). At the meeting the problem of variance heterogeneity in such series and the violation of normality in relation to the risk analysis were discussed (Piepho). This discussion was continued at the summer meetings. The pro and cons of whether environmental effects have to be regarded as fixed or random effects and the corresponding implications for the interpretation (Thöni, Weber et al.) were discussed. Several reports were presented on the problem of unbalanced data especially from the point of view of phytomedical experiments and of the tests of varieties by official testing institutions (Moll, Piepho, Michel, Krüger). Most talks gave recommendations for corresponding software solutions. A joint software project of the Federal Government and the States (Länder) for the analysis of field experiments which is still being developed was presented (Michel, Schmidtke, Voit). Piepho gave a talk on the possibilities of the representation of the result of multiple comparisons in experiments with unequal standard errors of the differences.

In Rostock the second main subject was the analysis and interpretation of long-term experiments. Many problems of the data analysis of these experiments result from their poor layout and from changes in experimental conditions (Warnstorff, Dörfel). Models which consider the repeated structure of data were discussed (Kroschewski, Richter) and a talk about the role of long-term experiments in connection with global climate change was given.

The presentations by Thöni (tetrachoric correlation coefficient as a mean to analyse toxin data) and Saur, Bleiholder (layout and realization of phytomedical experiments) dealt with special problems of experiments in phytomedicine. Götte and Büchse, Thomas, Guiard reported on problems with neighbourhood effects in variety trials and their consideration.

We would like to thank the local organizers of the summer meetings: A. Pfefferkorn in Bad Lauchstädt, J. Prochnow in Etzdorf and V. Michel in Gülzow. They reported about the current situation of the experimental sites and gave us the opportunity to visit the field trials in Bad Lauchstädt - especially the famous long-term experiment "The Static Fertilization Experiment" - , the trials in Etzdorf (BASF) and Gülzow. G. Bätz reported on the history of the experimental sites in Bad Lauchstädt.