Abstract

This short note shows how to plot a field map from an agricultural experiment and why that may be useful.

R setup

library("knitr")
knitr::opts_chunk$set(fig.align="center", fig.width=6, fig.height=6)
options(width=90)

First, a plot of the experimental design of the oats data from Yates.

library(agridat)
## The desplot and gge functions are now in the 'agrifun' package.
library(desplot)
data(yates.oats)
desplot(block ~ x+y, yates.oats, col=nitro, text=gen, cex=1, out1=block,
        out2=gen, out2.gpar=list(col = "gray50", lwd = 1, lty = 1))

This example is from Ryder (1981). Fit an ordinary RCB model with fixed effects for block and genotype. Plot a heatmap of the residuals.

library(agridat)
library(desplot)
data(ryder.groundnut)
gnut <- ryder.groundnut
m1 <- lm(dry ~ block + gen, gnut) # Standard RCB model
gnut$res <- resid(m1)
desplot(res ~ col + row, gnut, text=gen, cex=1,
        main="ryder.groundnut residuals from RCB model")

Note the largest positive/negative residuals are adjacent to each other, perhaps caused by the original data values being swapped. Checking experiment investigators (managers, data collectors, etc.) is recommended.

References

Ryder, K. 1981. “Field Plans: Why the Biometrician Finds Them Useful.” Experimental Agriculture 17: 243–56. doi:10.1017/S0014479700011601.