How to Create Multifactorial Studies in BMS

Owned by Julian Pietragalla, created
Last updated: 15 Mar, 2023

Introduction

This guideline explains how to create multifactorial studies in BMS. Experimental factor is any experimental parameter (e.g., germplasm entry, irrigation, fertilization). “Levels” of factors (valid entries) are usually controlled by an experimenter, or else they are imposed by a situation (e.g., planting date). Factors and response variables tend to get mixed up when they appear as parallel columns in a spreadsheet, and a factor in one study (e.g., harvest date) might be equivalent to a response (trait) variable in another study.

 

Factors in BMS studies

First we need to clarify the meaning of the following basic terms, which are used throughout this guideline and which are sometimes confused: treatments, factors, factor levels, and response variables

We consider three illustrative examples:

a)       A study that evaluates grain yield of 24 germplasm entries.

b)      A study that evaluates grain yield of 8 germplasm entries under each of 3 different fertilization rates.

c)       A study that evaluates grain yield of 4 germplasm entries at 3 levels of spacing, for 2 planting dates.

These three studies all have 24 treatments.

In the first, there is just a single factor, germplasm, which has 24 levels. Thus, in this simple case, whether we think of the different germplasm entries as the treatments, or the levels of a treatment factor makes no difference.

In the second study there are 2 factors, namely germplasm, with 8 levels, and fertilization rates, with 3 levels. Each treatment consists of the combination of a particular germplasm entry and a particular fertilization level. Thus, there are 24 different combinations, or treatments. This is sometimes known as an 8 by 3 factorial treatment structure.

The third study has 3 factors, germplasm, with 4 levels, spacing, with 3 levels, and planting dates with 2 levels. Thus, the 24 treatments are arranged in a 4 by 3 by 2 factorial treatment structure.

For the 3 studies the evaluated response variable is grain yield.

 

Creating a Treatment Factor Variable in BMS

For using treatment factors in BMS studies a pair of related variables must be created in the ontology manager in advance to study creation. The pair of variables must share the same property, and the variable type as Treatment factor.

Figure 1. A pair of variables, "NFERT_NO" and "NFert_kg_ha", sharing the property "Nitrogen" generated for describing the levels and labels of Nitrogen fertilization rates as a treatment factor.

 

The first variable of the pair describes the number of the levels of the treatment factor; thus, the variable has a numeric scale. The second variable describes/labels each of the treatment factor levels, thus the scale may have numerical, text or categorical type.

Figures 1, 2 and 3 show an example of a pair of variables, "NFERT_NO" and "NFert_kg_ha", sharing the property "Nitrogen" used for describing the levels and levels of Studies with Nitrogen fertilization rates as a treatment factor.

Figure 2. First variable of the pair for Nitrogen fertilization that describes the number of the levels of the treatment factor.

 

 Creating a Multifactorial Study in BMS

For this exercise, the step by step of a 3 by 2 multifactorial study is described, where 3 germplasm entries (Varieties A, B and C) by 2 Nitrogen levels (0 and 150 kg/ha) are evaluated for grain yield response at 2 locations.

  1. Open the Study Manager and click “Start a New Study”.

  2. Enter mandatory information. Name, describe, and choose the study type. Set the file directory by selecting "Change Folder" option.

  3. In the Germplasm and Check tab click “Browse” to select the germplasm list for use in the study.

  4. In the Treatment Factors tab, click Add+ and select the treatment factor variable that defines the levels, NFert_NO for this example. Then display the Label dropdown menu and select the variable that labels each level treatment, NFert_kg_ha for this example. Then specify the value label for each level, 0 and 150 for this example.

     

  5.  In the Environments tab, specified the number of environments for this study and then add Environmental details and Environmental conditions variables as needed for describing each location instance.

  6. In the Experimental Design tab, import or generate the design, choose RCBD and generate it with 2 replications for this example.

  7. After a successfully design generation, the study is ready to print labels, add traits to be collected, export electronic studybooks and import collected data. For this example, a grain yield response (GY_Calc_kgha) variable was added.

 

 

Importing an Externally Generated Multifactorial Study

In some cases, multifactorial studies due the complex nature of applied treatment factors may require an experimental design generated outside BMS, for these cases the user imports the externally generated design using a csv or excel file template containing all required the Experimental Design and Treatment Factors variables. 

The following example shows how to import a study experimental design for 2 Locations, with a split plot design of two factors, 2 levels of Nitrogen fertilization (0 and 150 kg/ha) by 3 varieties (germplasm entries). In this case the Nitrogen fertilization is the factor also used as blocking BLOCK_NO (Main plot), and the germplasm entries correspond as the SUB_BLOCK factor.

 

Where plot numbers are:

The template file containing the experimental design for this study includes following columns:

TRIAL_INSTANCE: Mandatory variable. Used to describe the number of locations, every row of the design template must have the location number. If only one location all the data rows will have value = 1

ENTRY_NO: Mandatory variable. It is the Germplasm entry assigned to each PLOT, based on the germplasm list that need to be created before the Study creation.

PLOT_NO: Mandatory variable. It is the experimental unit that combine all the factors.

BLOCK_NO: Also known as MAIN PLOT or WHOLE PLOT in split plot design, it is the factor that is hard to apply.

SUB_BLOCK: Also known as SUB PLOT in split plot design, it is the factor that is easy to apply.

REP_NO: Replication number

ROW: Not mandatory for this experimental design but describes the field layout. It may be used for spatial correction during the analysis.

COL: Not mandatory for this experimental design but describes the field layout. It may be used for spatial correction during the analysis.

NFERT_NO: This is a treatment factor variable that describes the number of the levels.

NFert_kg_ha: This is a treatment factor variable that labels each treatment level.

GY_Calc_kgha: This is the response variable; it may be added after design importation at the Observation tab.

TRIAL_INSTANCE, ENTRY_NO and PLOT_NO are mandatory variables to any BMS study, the other design specific variables must be in BMS or created in advance using the Ontology Manager.

The study creation process is similar to the previous exercise until the step in the Experimental Design tab where the user may import or generate the design.

  1. In the Experimental Design tab, click on “Or import an experimental design”, specify the file format and browse for the file template containing the experimental design.

  2. Revise mapped and un-mapped variables, remap or map accordingly, then click Next.

  3. Review that the Designation column was filled with the correct germplasm names and click Finish.

     

     

  4. The Observation tab has now the imported experimental design and the study is ready for printing labels, add traits to be collected, export electronic studybooks and import collected data. The germplasm Designation column will be filled with the Preferred names of the entries added in the Germplasm & Checks tab.

 

Example template file for importing a split plot design as multifactorial 3 by 2; 3 germplasm entries by 2 Nitrogen fertilization levels, in 2 Locations:

For more information read in BMS manual the Studies management chapters.