# Trial Design and Analysis for Breeding and Registration

## Introduction:

Statistics is the science of collecting and interpreting data subject to variability. Within modern agronomy there is variability associated with every aspect of our work. To make informed decisions we must understand the nature of this variability, and allow or account for it to make the best use of the data we collect. Statistics tells us how to deal with variability, and how to collect and use data so that we can make good decisions.

A good understanding and appreciation of sound experimental design (including choice of treatments, definition of within-plot sampling strategies, and identification of variables to be measured) and of the analysis and interpretation of designed experiments is essential for the modern agronomist/biological scientist. Whilst there are a number of user-friendly statistical packages providing easy access to statistical analysis tools, the modern agronomist/biological scientist needs to be aware of appropriate tools to be used to produce the results required for registration purposes and other publications. An understanding of how to interpret and present the results produced by these common statistical techniques is also essential.

This module is designed to introduce statistical ideas for experimental design, data collection and analysis that are suitable for the modern agronomist/biological scientist, with a strong focus on the application of these statistical approaches to address real practical scenarios in breeding and registration studies. The aims of the module are to revise basic statistical ideas for data summary, provide students with an overview of the statistical tools necessary to design efficient and effective experiments for breeding and registration work, to introduce the statistical techniques needed for the appropriate analysis of the data generated, and provide appropriate insights into the interpretation of these analyses. Real data examples, focussed on experiments concerned with breeding and registration, will be used to illustrate the application of the statistical methods and the interpretation of the output produced by the analyses.

## Objectives:

On completion of this module participants will be able to:

• Have the knowledge to design efficient and effective experiments for breeding and registration studies.
• Identify and apply appropriate statistical techniques for the analysis of the data collected in these experiments.
• Interpret the output produced by appropriate statistical analyses of experimental data.

## Contents:

• Illustrative examples of study types for crop agronomy, crop breeding and plant protection products (including an introduction to GEP/GLP/EPPO guidelines). These examples will be considered throughout the module.
• Data: types of data, summary statistics, exploratory data analysis, confidence intervals, graphical tools, commonly assumed statistical distributions.
• Testing hypotheses: the language of hypothesis testing, hypothesis testing and the scientific method, hypothesis testing for GEP/GLP/EPPO studies, the t-test for comparing twotreatments.
• Designing experiments: the principles behind good experimental design (replication, blocking, randomisation); the choice of treatment structure (for qualitative and quantitative treatmentfactors); identification of appropriate variables to be measured for a range of trial aims; definition of within-plot sampling plans for different variables and aims; practical designs for real experiments (field and protected crops, annual and perennial crops, screening of germplasm and chemicals, product/breeding line selection).
• The analysis of designed experiments: analysis of variance (ANOVA) as a basic tool for comparing multiple treatments, extracting information based on the treatment structure, testing assumptions and data transformations, and interpretation of the (computer-generated) output. Alternatives where ANOVA cannot be used because of the choice of design.
• The analysis of dose-response studies: simple linear regression; comparison of regression lines; common non-linear models (with example applications including plant growth in response to nutrients); analyses for binary responses against dose (bioassays); interpretation of (computer-generated) output.
• GEP/GLP/EPPO guidelines and the practical application of statistical methods in crop agronomy, crop breeding.