Uniformity trial of barley — piepho.barley.uniformity • agridat

Uniformity trial of barley in Germany

data("piepho.barley.uniformity")

Format

A data frame with 1080 observations on the following 5 variables.

row: row ordinate
col: column ordinate
yield: yield per plot

Details

Uniformity trial of barley at Ihinger Hof farm, conducted by the University of Hohenheim, Germany, in 2007.

Note: The paper by Piepho says "The trial had 30 rows and 36 columns. Plot widths were 1.90 m along rows and 3.73 m along columns." However, the SAS code supplement to the paper, called "PBR_1654_sm_example1.sas", has row=1-36, col=1-30. We cannot determine which dimension is "row" and which is "column", and therefore cannot determine the actual dimensions of the field.

Source

H. P. Piepho & E. R. Williams (2010). Linear variance models for plant breeding trials. Plant Breeding, 129, 1-8. https://doi.org/10.1111/j.1439-0523.2009.01654.x

References

None

Examples

# \dontrun{
  data(piepho.barley.uniformity) 
  dat <- piepho.barley.uniformity
  libs(desplot)
  desplot(yield ~ col*row, dat,
          tick=TRUE, # aspect unknown
          main="piepho.barley.uniformity.csv")
#> Please use desplot(data,form) instead of desplot(form,data)

  libs(asreml,dplyr)
  dat <- mutate(dat, x=factor(col), y=factor(row))
  dat <- arrange(dat, x, y)
  
  # Piepho AR1xAR1 model (in random term, NOT residual)
  m1 <- asreml(data=dat,
               yield ~ 1, 
               random = ~ x + y + ar1(x):ar1(y), 
               residual = ~  units,
               na.action=na.method(x="keep") )
#> Model fitted using the gamma parameterization.
#> ASReml 4.1.0 Fri Dec 11 17:49:49 2020
#>           LogLik        Sigma2     DF     wall    cpu
#>  1     -2115.600       15.6955   1075 17:49:49    0.0 (3 restrained)
#>  2     -2101.978       16.5258   1075 17:49:49    0.0 (3 restrained)
#>  3     -2087.509       15.9315   1075 17:49:49    0.0 (3 restrained)
#>  4     -2076.005       15.2437   1075 17:49:49    0.0 (3 restrained)
#>  5     -2071.390       14.9905   1075 17:49:49    0.0 (3 restrained)
#>  6     -2069.380       14.7343   1075 17:49:49    0.0 (2 restrained)
#>  7     -2057.416       13.4665   1075 17:49:49    0.0 (1 restrained)
#>  8     -2052.089       13.5044   1075 17:49:49    0.0 (1 restrained)
#>  9     -2049.050       13.7143   1075 17:49:49    0.0 (1 restrained)
#> 10     -2037.688       13.5521   1075 17:49:49    0.0
#> 11     -2035.168       13.8721   1075 17:49:49    0.0
#> 12     -2033.901       13.8242   1075 17:49:49    0.0
#> 13     -2033.512       13.6703   1075 17:49:49    0.0
#> Warning: Log-likelihood not converged
#> Warning: Some components changed by more than 1% on the last iteration.
  m1 <- update(m1)
#> Model fitted using the gamma parameterization.
#> ASReml 4.1.0 Fri Dec 11 17:49:49 2020
#>           LogLik        Sigma2     DF     wall    cpu
#>  1     -2033.492       13.6383   1075 17:49:49    0.0
#>  2     -2033.491       13.6335   1075 17:49:49    0.0
#>  3     -2033.489       13.6212   1075 17:49:49    0.0
#>  4     -2033.489       13.6163   1075 17:49:50    0.0
  # Match Piepho table 3, footnote 4: .9671, .9705 for col,row correlation
  # Note these parameters are basically at the boundary of the parameter
  # space. Questionable fit.
  libs(lucid)
  lucid::vc(m1) 
#>     effect component std.error z.ratio bound %ch
#>          x    0.7531   0.5642      1.3     P 0.1
#>          y    1.2      0.7423      1.6     P 0.2
#>        x:y   22.75    15.53        1.5     P 0.1
#>  x:y!x!cor    0.9672   0.02536    38       U 0  
#>  x:y!y!cor    0.9708   0.01645    59       U 0  
#>    units!R   13.62     0.654      21       P 0  
# }