Classificação do coeficiente de variação do comprimento do epicótilo de soja
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidade Federal de Viçosa
Abstract
O comprimento do epicótilo em soja apresenta potencial como descritor morfológico para avaliação de genótipos. Embora o coeficiente de variação (CV) seja amplamente utilizado como indicador de precisão experimental, sua variação difere entre espécies, da característica em estudo, delineamento, tamanho da amostra, o que restringe a aplicação de classificações genéricas. Desta forma, o objetivo deste estudo é estabelecer intervalos específicos de CV (Baixo, Médio, Alto, Muito Alto) para o comprimento do epicótilo de soja, analisando 101 valores de CV nos estádios V2 e V3, individualmente e conjuntamente, provenientes de artigos, teses, dissertações e dados não publicados, abrangendo o período de 2007 a 2025. Quatro métodos foram aplicados: (1) Garcia (1989), baseado na média e do desvio padrão da distribuição normal; (2) método de Vangel (1996), que utiliza limites de uma distribuição de Qui-Quadrado; (3) Costa et al. (2002), baseado em medidas robustas de Mediana e Pseudo-Sigma; e (4) método dos Quantis Amostrais de Judice et al. (2002). Para a validação e comparação entre os métodos foram construídos Intervalos de Confiança (IC) de 95% do tipo BCa (Bias-Corrected and accelerated). Foi estabelecido intervalos específicos de CV e as discrepâncias observadas entre estas classificações e as genéricas reforçam a necessidade de referenciais específicas para características morfológicas distintas como o epicótilo da soja. O método de Costa et al. (2002) para a amostra conjunta foi o mais robusto para os dados, com as classificações: Baixo (CV 8,11); Médio (8,11 < CV 15,36); Alto (15,36 < CV 18,98); e Muito Alto (CV > 18,98). Palavras-chave: Glycine max; Precisão experimental ; Experimentação; Métodos de classificação.
The epicotyl length in soybean has potential as a morphological descriptor for genotype evaluation. Although the coefficient of variation (CV) is widely used as an indicator of experimental precision, its magnitude varies between species, the trait under study, experimental design, and sample size, which limits the applicability of generic classifications. Therefore, this study aimed to establish trait-specific CV ranges (Low, Medium, High, Very High) for soybean epicotyl length by analyzing 101 CV values at stages V2 and V3, both individually and jointly, obtained from articles, theses, dissertations, and unpublished data spanning 2007–2025. Four methods were applied: (1) Garcia (1989), based on the mean and standard deviation assuming a normal distribution; (2) Vangel (1996), which uses limits from a Chi- square distribution; (3) Costa et al. (2002), based on robust measures of the median and pseudo-sigma; and (4) the sample quantile method of Judice et al. (2002). For validation and comparison among methods, 95% bias-corrected and accelerated (BCa) confidence intervals were constructed (Bias-Corrected and accelerated). Trait- specific CV ranges were established, and the discrepancies observed between these classifications and generic ones underscore the need for specific references for distinct morphological traits such as the soybean epicotyl. The Costa et al. (2002) method for the pooled sample was the most robust for the data, with the classifications: Low (CV 8.11); Medium (8.11 < CV 15.36); High (15.36 < CV 18.98); and Very High (CV > 18.98). Keywords: Glycine max ; Experimental precision ; Experimentation; Classification methods.
The epicotyl length in soybean has potential as a morphological descriptor for genotype evaluation. Although the coefficient of variation (CV) is widely used as an indicator of experimental precision, its magnitude varies between species, the trait under study, experimental design, and sample size, which limits the applicability of generic classifications. Therefore, this study aimed to establish trait-specific CV ranges (Low, Medium, High, Very High) for soybean epicotyl length by analyzing 101 CV values at stages V2 and V3, both individually and jointly, obtained from articles, theses, dissertations, and unpublished data spanning 2007–2025. Four methods were applied: (1) Garcia (1989), based on the mean and standard deviation assuming a normal distribution; (2) Vangel (1996), which uses limits from a Chi- square distribution; (3) Costa et al. (2002), based on robust measures of the median and pseudo-sigma; and (4) the sample quantile method of Judice et al. (2002). For validation and comparison among methods, 95% bias-corrected and accelerated (BCa) confidence intervals were constructed (Bias-Corrected and accelerated). Trait- specific CV ranges were established, and the discrepancies observed between these classifications and generic ones underscore the need for specific references for distinct morphological traits such as the soybean epicotyl. The Costa et al. (2002) method for the pooled sample was the most robust for the data, with the classifications: Low (CV 8.11); Medium (8.11 < CV 15.36); High (15.36 < CV 18.98); and Very High (CV > 18.98). Keywords: Glycine max ; Experimental precision ; Experimentation; Classification methods.
Description
Citation
MARTINHO, Miguel Luiz. Classificação do coeficiente de variação do comprimento do epicótilo de soja. 2026. 47 f. Dissertação (Mestrado em Estatística Aplicada e Biometria) - Universidade Federal de Viçosa, Viçosa. 2026.
