R packages for Analysis of Design of Experiments

Design of experiments (DoE) involves choice of treatments, allotment of treatments to experimental units, laying out of experimental units as per the conditions, analyzing the collected data and evaluate the factors that influence the variable of interest mostly yield in the agriculture study. All the agriculture experiments are required to use different designs as per the need of the experiment.

 Analysis of these experimental designs needs in-depth knowledge of the subject matter as well as analytical tools like SPSS/ Python/ SAS/ R/ STATA. Among the different analytical tools available R and Python are the open-source tools available in public domain. R has more than 18,000 packages which provides wide range of data analysis. This article aims to provide details of R packages useful for analysis of design of experiments used in agriculture research. The popular R packages useful for analysis of design of experiments are agricolae, doebioresearch, augmentedRCBD and ibd. All these R packages are further discussed in detail.

1. agricoale (Mendiburu, 2021)

This is the most popular R package available for analysis of different designs and multiple comparison test. The functions useful for analysis of different designs are shown in Table 1 and that of multiple comparison test are shown in Table 2.

Table 1: Functions of agricolae package for analysis of different designs

Experimental Design	Function
Split plot analysis	sp.plot( )
Strip plot analysis	Strip.plot( )
Split-split plot analysis	ssp.plot( )
Balanced Incomplete Block Design (BIBD) analysis	BIB.test( )

Table 2: Functions of agricolae for different multiple comparison test

Multiple comparison test	Function
Least Significant Difference test	LSD.test()
Tukey's test	HSD.test()
Student-Newman-Keuls test	SNK.test()
Friedman test	friedman()
Duncan's new multiple range test (DNMRT)	duncan.test()

Apart from these agricolae contains functions for randomization procedure of different designs and different analysis of plant breeding research.

2. doebioresearch (Popat and Banakara, 2020)

The doebioresearch package provides analysis of different designs along with multiple comparison test. It provides three options of multiple comparison test viz., least significant difference test, Duncan’s new multiples range test and Tukey’s test. The package has functions for transformation and retransformation of data for logarithmic, square root and arc sin transformation. The functions for analysis of different designs are shown in Table 3.

Table 3: Functions of doebioresearch R package for analysis of different experimental designs

Experimental Design	Function
Completely randomized design analysis	crd( )
Randomized complete block design analysis	rcbd ( )
Factorial completely randomized design analysis (2 factor)	fcrd2fact( )
Factorial completely randomized design analysis (3 factor)	fcrd3fact( )
Factorial randomized block design analysis (2 factor)	frbd2fact( )
Factorial randomized block design analysis (3 factor)	frbd3fact( )
Split plot analysis	splitplot( )
Strip plot analysis	stripplot( )
Latin square design analysis	lsd( )

3. ibd (Mandal, 2019)

The function aov.ibd( ) provides analysis of variance and adjusted means for the incomplete block design. Apart from ANOVA the package facilitates calculation of A-efficiency, D-efficiency and Information matrix. The package also has functions for generation of balanced incomplete block design for specific parameter values.

4. augmentedRCBD (Aravind et al., 2021)

The package has function named augmentedRCBD( ) which gives ANOVA, adjusted means and mean comparison for augmented design. Augmented design is used when the number of treatments are large and resources like land is scarce. The package also supports calculation of descriptive statistics, plotting of frequency distribution of output and genetic variability analysis of traits (useful in plant breeding research).

Apart from the above-mentioned R packages, a software named META-R which is based on R provides analysis of multiple environment data for randomized complete block design and alpha lattice design (Alvarado et al., 2020). The software also facilities different analysis related to plant breeding research.

By using all these packages different designs can be analysed in R which is easy to use and available for free in public domain. Moreover, with good skills in R a customized script can also be written to facilitate complex analysis.

References

Gregorio Alvarado, Francisco M. Rodríguez, Angela Pacheco, Juan Burgueño, José Crossa, Mateo Vargas, Paulino Pérez-Rodríguez, Marco A. Lopez-Cruz, (2020). META-R: A software to analyze data from multi-environment plant breeding trials. The Crop Journal. 8(5): 745-756.

Aravind, J., Mukesh Sankar, S., Wankhede, D. P., and Kaur, V. (2021). augmentedRCBD: Analysis of Augmented Randomised Complete Block Designs. R package version 0.1.5, https://cran.r-project.org/package=augmentedRCBD.

B N Mandal (2019). ibd: Incomplete Block Designs. R package version 1.5. https://CRAN.R-project.org/package=ibd.

Felipe de Mendiburu (2021). agricolae: Statistical Procedures for Agricultural Research. R package version 1.3-5. https://CRAN.R-project.org/package=agricolae.

Raj Popat and Kanthesh Banakara (2020). doebioresearch: Analysis of Design of Experiments for Biological Research. R package version 0.1.0. https://CRAN.R-project.org/package=doebioresearch.