Brassica: Brassica Response to Herbicide

In byandell/pda: Practial Data Analysis for Designed Experiments

Description

The scientist examined the effects of rate of application of a certain herbicide (pyridate) on a range of closely related crops. The experiment was conducted over two years, with three blocks each year. The two years were conducted in roughly the same location, but there was separate randomization in each year. Three herbicide rates (pyrrat = 0, 1, 2) were randomly assigned to separate rows within each crop plot. Recorded values consist of yield per row for each crop*pyrrat combination within each block for each year. Since crop 2 was not used in the second year, it is dropped from further consideration.

Usage

data(Brassica)

Format

Brassica data frame with 144 observations on 10 variables.

[,1]	crop	factor	crop identifier (1-7)
[,2]	pyrrat	ordered	pyridate level (0-2)
[,3]	block	factor	block (1-3)
[,4]	inj7dat	numeric	injury after 7 days
[,5]	in14dat	numeric	injury after 14 days
[,6]	yldkga	numeric	yield in kilograms
[,7]	pconyld	numeric	percent control yield
[,8]	pcnblk	numeric	percent control in block yield
[,9]	year	ordered	year - 1990
[,10]	trt	factor	treatment code

Source

Alan Miller, U WI Horticulture (http://www.hort.wisc.edu)

References

Miller A (1993) 'Pyridate tolerance in Brassicaceae crops', PhD Dissertation, Department of Horticulture, University of Wisconsin-Madison.

Examples

# full data set--code not ready yet!
# model fits for bill and leg
data( Brassica )

# drop Crop 2
Brassica <- data.frame(Brassica[ Brassica$crop != 2, ])
Brassica$crop <- factor( as.character( Brassica$crop ))

# Analysis based on lmer()
library(lme4)
Brassica1 <- data.frame( Brassica[ Brassica$year == 1, ] )
Brassica.lme <- lmer( yldkga ~ crop * pyrrat + (1|crop:block),
   data = Brassica1 )
anova( Brassica.lme )
VarCorr( Brassica.lme )

# Analysis based on aov() using Error() specification

Brassica.aov <- aov ( yldkga ~ crop * pyrrat +
   Error( block/crop ), Brassica1 )
summary( Brassica.aov )

# Analysis of full data set: both years

Brassica.full <- aov ( yldkga ~ crop * pyrrat * year +
   Error( year:block + crop:year:block ), Brassica )

Brassica.red <- aov ( yldkga ~ crop * year + pyrrat + crop:pyrrat +
   Error( year:block + crop:year:block ), Brassica)

# Analysis based on lmer()
Brassica1 <- data.frame( Brassica[ Brassica$year == 1, ] )
Brassica.lme <- lmer( yldkga ~ crop * pyrrat + (1|block:crop),
   data = Brassica1 )
Brassica$yrblk = interaction(Brassica[,c("year","block")])
Brassica$cropyrblk = interaction(Brassica[,c("crop","year","block")])
Brassica.fulle = lmer(yldkga ~ crop * pyrrat * year +
  (1|yrblk:cropyrblk), data = Brassica )
anova(Brassica.fulle)
Brassica.rede = lmer(yldkga ~ crop * year + pyrrat + crop:pyrrat +
  (1|yrblk:cropyrblk), data = Brassica )
anova(Brassica.rede)
anova(Brassica.rede,Brassica.fulle)

lsd.plot(Brassica.rede,factor=c("pyrrat","crop"),
   xlab = "pyr rate by crop",
   ylab = "mean crop yield" )

## Not run: 
# The following code is specialized to get plots for the book.
# It is not recommended as standard practice, but presented to
# show one torturous way to get LSMEANS for a nested design.
Brassica.red <- aov (yldkga ~ crop * year + pyrrat + crop:pyrrat +
   Error(year:block + crop:year:block), Brassica, qr = TRUE)
Brassica.proj <- proj ( Brassica.red )
Brassica.ref <- data.frame(Reference=c(Brassica.proj[[1]]))
for (i in names(Brassica.proj)[-1])
   Brassica.ref[[i]] <- apply( Brassica.proj[[i]], 1, sum )

# whole plot
Brassica.wp <- nested( Brassica.proj, Brassica, c("block","year","crop"), "yldkga",
   c(1,3), Brassica.ref )

Brassica.wpaov <- aov( yldkga ~ year*block + crop + crop:year, Brassica.wp,
   weight = Brassica.wp$weight)

# split plot
Brassica.sp <- nested( Brassica.proj, Brassica, c("block","year","crop","pyrrat"),
   "yldkga", c(1,4), Brassica.ref )

Brassica.spaov <- aov( yldkga ~ year*block*crop + pyrrat +
   crop:pyrrat, Brassica.sp )

margin.plot( Brassica.wpaov, factors = c("crop","year"),
   xlab = "(a) crop by year", ylab = "yield in 10000 kg" )

#Figure H:24.1 Brassicaica split split plot anova

lsd.plot( Brassica.spaov, factors = c("pyrrat","crop"),
   xpos = .6, ypos = 3.25,
   xlab = "(b) pyr rate by crop",
   ylab = "deviation from mean crop yield" )

## End(Not run)

byandell/pda documentation built on May 13, 2019, 9:27 a.m.