metan (version 1.5.0)

waasb: Weighted Average of Absolute Scores

Description

Compute the Weighted Average of Absolute Scores (Olivoto et al., 2019) for quantifying the stability of g genotypes conducted in e environments using linear mixed-effect models.

Usage

waasb(
  .data,
  env,
  gen,
  rep,
  resp,
  block = NULL,
  mresp = NULL,
  wresp = NULL,
  random = "gen",
  prob = 0.05,
  ind_anova = TRUE,
  verbose = TRUE,
  ...
)

Arguments

.data

The dataset containing the columns related to Environments, Genotypes, replication/block and response variable(s).

env

The name of the column that contains the levels of the environments.

gen

The name of the column that contains the levels of the genotypes.

rep

The name of the column that contains the levels of the replications/blocks.

resp

The response variable(s). To analyze multiple variables in a single procedure a vector of variables may be used. For example resp = c(var1, var2, var3).

block

Defaults to NULL. In this case, a randomized complete block design is considered. If block is informed, then an alpha-lattice design is employed considering block as random to make use of inter-block information, whereas the complete replicate effect is always taken as fixed, as no inter-replicate information was to be recovered (Mohring et al., 2015).

mresp

The new maximum value after rescaling the response variable. By default, all variables in resp are rescaled so that de maximum value is 100 and the minimum value is 0 (i.e., mresp = 100). It must be a numeric vector of the same length of resp if rescaling is assumed to be different across variables, e.g., if for the first variable smaller values are better and for the second one, higher values are better, then mresp = c(0, 100) must be used. Numeric value of length 1 will be recycled with a warning message.

wresp

The weight for the response variable(s) for computing the WAASBY index. By default, all variables in resp have equal weights for mean performance and stability (i.e., wresp = 50). It must be a numeric vector of the same length of resp to assign different weights across variables, e.g., if for the first variable equal weights for mean performance and stability are assumed and for the second one, a higher weight for mean performance (e.g. 65) is assumed, then wresp = c(50, 65) must be used. Numeric value of length 1 will be recycled with a warning message.

random

The effects of the model assumed to be random. Defaults to random = "gen". See Details to see the random effects assumed depending on the experimental design of the trials.

prob

The probability for estimating confidence interval for BLUP's prediction.

ind_anova

Logical argument set to TRUE. If FALSE the within-environment ANOVA is not performed.

verbose

Logical argument. If verbose = FALSE the code will run silently.

...

Arguments passed to the function impute_missing_val() for imputation of missing values in the matrix of BLUPs for genotype-environment interaction, thus allowing the computation of the WAASB index.

Value

An object of class waasb with the following items for each variable:

• individual A within-environments ANOVA considering a fixed-effect model.

• fixed Test for fixed effects.

• random Variance components for random effects.

• LRT The Likelihood Ratio Test for the random effects.

• model A tibble with the response variable, the scores of all IPCAs, the estimates of Weighted Average of Absolute Scores, and WAASBY (the index that considers the weights for stability and mean performance in the genotype ranking), and their respective ranks.

• BLUPgen The random effects and estimated BLUPS for genotypes (If random = "gen" or random = "all")

• BLUPenv The random effects and estimated BLUPS for environments, (If random = "env" or random = "all").

• BLUPint The random effects and estimated BLUPS of all genotypes in all environments.

• PCA The results of Principal Component Analysis with the eigenvalues and explained variance of the matrix of genotype-environment effects estimated by the linear fixed-effect model.

• MeansGxE The phenotypic means of genotypes in the environments.

• Details A list summarizing the results. The following information are shown: Nenv, the number of environments in the analysis; Ngen the number of genotypes in the analysis; mresp The value attributed to the highest value of the response variable after rescaling it; wresp The weight of the response variable for estimating the WAASBY index. Mean the grand mean; SE the standard error of the mean; SD the standard deviation. CV the coefficient of variation of the phenotypic means, estimating WAASB, Min the minimum value observed (returning the genotype and environment), Max the maximum value observed (returning the genotype and environment); MinENV the environment with the lower mean, MaxENV the environment with the larger mean observed, MinGEN the genotype with the lower mean, MaxGEN the genotype with the larger.

• ESTIMATES A tibble with the genetic parameters (if random = "gen" or random = "all") with the following columns: Phenotypic variance the phenotypic variance; Heritability the broad-sense heritability; GEr2 the coefficient of determination of the interaction effects; Heribatility of means the heritability on the mean basis; Accuracy the selective accuracy; rge the genotype-environment correlation; CVg the genotypic coefficient of variation; CVr the residual coefficient of variation; CV ratio the ratio between genotypic and residual coefficient of variation.

• residuals The residuals of the model.

Details

The weighted average of absolute scores is computed considering all Interaction Principal Component Axis (IPCA) from the Singular Value Decomposition (SVD) of the matrix of genotype-environment interaction (GEI) effects generated by a linear mixed-effect model, as follows: $$WAASB_i = \sum_{k = 1}^{p} |IPCA_{ik} \times EP_k|/ \sum_{k = 1}^{p}EP_k$$

where \(WAASB_i\) is the weighted average of absolute scores of the ith genotype; \(IPCA_{ik}\) is the score of the ith genotype in the kth Interaction Principal Component Axis (IPCA); and \(EP_k\) is the explained variance of the kth IPCA for k = 1,2,..,p, considering \(p = min(g - 1; e - 1)\).

The nature of the effects in the model is chosen with the argument random. By default, the experimental design considered in each environment is a randomized complete block design. If block is informed, a resolvable alpha-lattice design (Patterson and Williams, 1976) is implemented. The following six models can be fitted depending on the values of random and block arguments.

• Model 1: block = NULL and random = "gen" (The default option). This model considers a Randomized Complete Block Design in each environment assuming genotype and genotype-environment interaction as random effects. Environments and blocks nested within environments are assumed to fixed factors.

• Model 2: block = NULL and random = "env". This model considers a Randomized Complete Block Design in each environment treating environment, genotype-environment interaction, and blocks nested within environments as random factors. Genotypes are assumed to be fixed factors.

• Model 3: block = NULL and random = "all". This model considers a Randomized Complete Block Design in each environment assuming a random-effect model, i.e., all effects (genotypes, environments, genotype-vs-environment interaction and blocks nested within environments) are assumed to be random factors.

• Model 4: block is not NULL and random = "gen". This model considers an alpha-lattice design in each environment assuming genotype, genotype-environment interaction, and incomplete blocks nested within complete replicates as random to make use of inter-block information (Mohring et al., 2015). Complete replicates nested within environments and environments are assumed to be fixed factors.

• Model 5: block is not NULL and random = "env". This model considers an alpha-lattice design in each environment assuming genotype as fixed. All other sources of variation (environment, genotype-environment interaction, complete replicates nested within environments, and incomplete blocks nested within replicates) are assumed to be random factors.

• Model 6: block is not NULL and random = "all". This model considers an alpha-lattice design in each environment assuming all effects, except the intercept, as random factors.

References

Olivoto, T., A.D.C. L\'ucio, J.A.G. da silva, V.S. Marchioro, V.Q. de Souza, and E. Jost. 2019. Mean performance and stability in multi-environment trials I: Combining features of AMMI and BLUP techniques. Agron. J. 111:2949-2960. doi:10.2134/agronj2019.03.0220

Mohring, J., E. Williams, and H.-P. Piepho. 2015. Inter-block information: to recover or not to recover it? TAG. Theor. Appl. Genet. 128:1541-54. doi:10.1007/s00122-015-2530-0

Patterson, H.D., and E.R. Williams. 1976. A new class of resolvable incomplete block designs. Biometrika 63:83-92. doi:10.1093/biomet/63.1.83

See Also

mtsi waas get_model_data plot_scores

Examples

# NOT RUN {
library(metan)
#===============================================================#
# Example 1: Analyzing all numeric variables assuming genotypes #
# as random effects with equal weights for mean performance and #
# stability                                                     #
#===============================================================#
model <- waasb(data_ge,
              env = ENV,
              gen = GEN,
              rep = REP,
              resp = everything())
# Distribution of random effects (first variable)
plot(model, type = "re")

# Genetic parameters
get_model_data(model, "genpar")



#===============================================================#
# Example 2: Analyzing variables that starts with "N"           #
# assuming environment as random effects with higher weight for #
# response variable (65) for the three traits.                  #
#===============================================================#

model2 <- waasb(data_ge2,
               env = ENV,
               gen = GEN,
               rep = REP,
               random = "env",
               resp = starts_with("N"),
               wresp = 65)


# Get the index WAASBY
get_model_data(model2, what = "WAASBY")

# Plot the scores (response x WAASB)
plot_scores(model2, type = 3)

#===============================================================#
# Example 3: Analyzing GY and HM assuming a random-effect model.#
# Smaller values for HM and higher values for GY are better.    #
# To estimate WAASBY, higher weight for the GY (60%) and lower  #
# weight for HM (40%) are considered for mean performance.      #
#===============================================================#

model3 <- waasb(data_ge,
                env = ENV,
                gen = GEN,
                rep = REP,
                resp = c(GY, HM),
                random = "all",
                mresp = c(100, 0),
                wresp = c(60, 40))

# Get P-values for Likelihood-ratio test
get_model_data(model3, "pval_lrt")

# Get the random effects
get_model_data(model3, what = "ranef")

# Get the ranks for the WAASB index
get_model_data(model3, what = "OrWAASB")
# }
# NOT RUN {
# }