MGIDI: towards an effective multivariate selection in biological experiments

Abstract

The efficiency in genotype selection in plant breeding programs could be greater if based on multiple traits, but identifying genotypes that combine high performance across many traits has been a challenger task for breeders.Here, we introduced the theoretical foundations, validated the potential, provided the needed tools for future implementations, and showed how breeders can use the novel multi-trait genotype-ideotype distance index (MGIDI) for genotype selection in plant breeding programs. We illustrated the application of the method using data from a preliminary wheat trial with 14 traits assessed in 44 genotypes.The MGIDI provided negative gains (−2.29% ≤ gains ≤ −0.07%, a total of −5.15%) for all the four traits that wanted to decrease and positive gains (0.46% ≤ gains ≤ 14.28%, a total of 49.64%) for nine of the 10 traits that wanted to increase. The MGIDI should be useful for breeders who aim at the selection of genotypes based on multiple traits because it provides a unique and easy-to-interpret selection process that incorporates the underlying correlation structure of the data and doesn’t require any arbitrary weighting coefficient.The use of the MGIDI will save tremendous time and resources in plant breeding trials of any crop, guaranteeing long-term gains in grain yield without jeopardizing genetic gains in other important traits, thus helping breed of better varieties fast enough to satisfy the growing global population needs for food. A step-by-step guide and an R package containing all the needed functions are provided to facilitate the implementation of the MGIDI index in future studies.

Understanding drought response mechanisms in wheat and multi-trait selection

“…The mixed model analysis was performed in the R 4.0.1 software system, using the functions gamem() and get_model_data() of the package metan [ 38 ].The principal component analysis was realized for the environment jointly with the package factoextra . The inputs for the analyses were the data of Best linear unbiased prediction—BLUP’s genotypes and environment.…”

“…The multi-trait genotype-ideotype distance index (MGIDI) was used to rank the genotypes based on information of multiple traits, as proposed by [ 38 ]. The first step to compute the MGIDI was to rescale the matrix X so that all the values have a 0–100 range.…”

Multi‐trait selection of bread wheat ideotypes for adaptation to early sown condition

“…is robust and provides an easy-to-handle selection process. Idiotypes can be designed based on a blend of desirable and undesirable features (Olivoto & Nardino, 2020). Once idiotype design is complete, it can be assessed across environments, particularly under early planting conditions for performance study and genotype selection.…”

“…It has the potential to overcome the multicollinearity effects, particularly when we use multiple traits for idiotype design. Consequently, it results in improved conditioned matrices and unbiased index coefficients, thus easy to estimate genetic gain (Olivoto & Nardino, 2020). The effective graphical and straightforward approach of the strength and weakness view of genotypes and traits allowed us to select genotypes and traits for further use in the ongoing breeding program.…”

“…The single‐environment multi‐trait genotype–idiotype distance index (MGIDI) for simultaneous selection, considering mean performance and stability in the analysis of multi‐environments using both fixed and mixed‐effect models proposed by researchers (Olivoto, Lúcio, da Silva, Marchioro et al., 2019; Olivoto, Lúcio, da Silva, Sari et al., 2019), is robust and provides an easy‐to‐handle selection process. Idiotypes can be designed based on a blend of desirable and undesirable features (Olivoto & Nardino, 2020). Once idiotype design is complete, it can be assessed across environments, particularly under early planting conditions for performance study and genotype selection.…”