The global meeting place for people interested in all things related to SWEETPOTATO

Share your research and experience, ask and answer questions, meet your peers.

User guide for Multi Environment Trial analysis with CloneSelector

Genotype by environment interaction, GxE, is one of the challenges in plant breeding and other experiments with plants. Testing across multiple environments is therefore a standard procedure in plant breeding as well as other experiments. The outcome of these experiments is data sets from multiple environments, and these need to be analyzed using statistical methods that allows the scientist to draw conclusions for example related to stability of a genotype across target environments or in relation to the genotypic performance with less interference of GxE. These different types of analysis have in CloneSelector been automated as MET analysis or Multi environment trial analysis.

It should be noted that the MET analysis generates many analytical outputs; however, the researcher must select the analytical outputs that are relevant to the particular trial or experiment, and ensure all assumptions are fulfilled. The MET analysis will for example produce different ANOVA tables, but you must choose the one that have the right combination of fixed and random effects. The MET analysis also produces for example plots of the data or residuals, and other types of analytical tools to assess if the assumptions for a particular analysis are fulfilled. CloneSelector aims to automate the task of carrying out the statistical analysis of a multi-environment trial, however, the researcher must interpret the outputs and ensure they are valid in relation to the particular experiment.

Before you can do the MET analysis you must have a series of experiments in CloneSelector where you have tested at least 3 clones in the same three 3 environments. Some of the tests can also be done for only 2 clones in 2 environments, but most of the analysis requires at least 3 clones that have been tested in at least 3 environments, if you have more clones and more environments that obviously also work. If your data is not in CloneSelector you may still be able to use the MET analysis, but it will require a bit of manual copy/paste – and we cannot guarantee that you will not have problems when trying to analyze non-CloneSelector data.

 

The MET analysis tools in CloneSelector helps you:

  1. Copy in an automated way the individual trial results from the Master sheets into one consolidated file that has the data from all the trials properly aligned for analysis.
  2. You can then create for each trait individual sheets before analyzing the data. The purpose of these sheets is to be able to easily resolve issues with data quality, and especially missing data problems. The analysis requires at least one observation from each genotype and environment combination, and no more than 10% missing data. If this is not fulfilled you will have to delete genotypes or environments before running the analysis.
  3. Once the data is prepared you can run the MET analysis and it will produce some 15 different types of statistical analysis and 5 plots for each trait.
  4. You can also generate a MET summary result sheet that presents the mean values of each trait for each genotype, and also the minimum and maximum values for each trait.
  5. Finally, you can generate a selection index for the traits you choose to include. The index is based on Elstons method using k-values.