ABSTRACT. The study of traits in crops enables breeders to guide strategies for selecting and accelerating the progress of genetic breeding. Although the simultaneous evaluation of characteristics in the plant breeding programme provides large quantities of information, identifying which phenotypic characteristic is the most important is a challenge facing breeders. Thus, this work aims to quantify the best approaches for prediction and establish a network of better predictive power in flood-irrigated rice via methodologies based on regression, artificial intelligence, and machine learning. Multiple regression, computational intelligence, and machine learning were used to predict the importance of the characteristics. Computational intelligence and machine learning were notable for their ability to extract nonlinear information from model inputs. Predicting the relative contribution of auxiliary characteristics in rice through computational intelligence and machine learning proved to be efficient in determining the relative importance of variables in flood-irrigated rice. The characteristics indicated to assist in decision making are flowering, number of grains filled by panicles and length of panicles for this study. The network with only one hidden layer with 15 neurons was observed to be efficient in determining the relative importance of variables in flooded rice.

ABSTRACT. The genotype by environment interaction is the main factor that influences the response of evaluated genotypes in trials of value for cultivation and use. Adaptability and stability analyses are fundamental to understanding the performance of genotypes in a growing region. Some of these methodologies incorporate previous information for recommending an extra group of genotypes denominated as specific ideotypes under certain cultivation conditions. Based on this strategy, the centroid method and its modifications have been widely used due to the simplicity of classification of the evaluated genotypes. However, these methodologies present problems in identifying adaptability patterns of some genotypes. Artificial intelligence techniques, such as fuzzy C-means, can be an alternative to reduce these difficulties, since they use, in addition to distance information between genotypes, memberships (measures quantifying how much an observation belongs to a particular class) to increase discriminatory power. Therefore, our aim was to propose and evaluate the phenotypic adaptability method by fuzzy clustering to assist cultivar recommendations. The adaptation of the fuzzy C-Means method to classify the genotypes was implemented in BioFuzzy software. The grain yield data of black common bean genotypes were used to evaluate the potential of the method. The results obtained by this method were compared with those obtained by the centroid method. The phenotypic adaptability method by fuzzy clustering was effective in identifying the adaptability patterns of common bean genotypes. Moreover, the discriminatory power was higher than that observed with the centroid method. ABSTRACT use region conditions strategy However techniques Cmeans, Cmeans C means, means C-means difficulties measures class Therefore recommendations CMeans Means software Moreover

ABSTRACT. This paper aimed to evaluate the effectiveness of subset selection of markers for genome-enabled prediction of genetic values using radial basis function neural networks (RBFNN). To this end, an F1 population derived from the hybridization of divergent parents with 500 individuals genotyped with 1000 SNP-type markers was simulated. Phenotypic traits were determined by adopting three different gene action models - additive, additive-dominant, and epistatic, representing two dominance situations: partial and complete with quantitative traits having a heritability (h2) of 30 and 60%; traits were controlled by 50 loci, considering two alleles per locus. Twelve different scenarios were represented in the simulation. The stepwise regression was used before the prediction methods. The reliability and the root mean square error were used for estimation using a fivefold cross-validation scheme. Overall, dimensionality reduction improved the reliability values for all scenarios, specifically with h2 =30 the reliability value from 0.03 to 0.59 using RBFNN and from 0.10 to 0.57 with RR-BLUP in the scenario with additive effects. In the additive dominant scenario, the reliability values changed from 0.12 to 0.59 using RBFNN and from 0.12 to 0.58 with RR-BLUP, and in the epistasis scenarios, the reliability values changed from 0.07 to 0.50 using RBFNN and from 0.06 to 0.47 with RR-BLUP. The results showed that the use of stepwise regression before the use of these techniques led to an improvement in the accuracy of prediction of the genetic value and, mainly, to a large reduction of the root mean square error in addition to facilitating processing and analysis time due to a reduction in dimensionality.

ABSTRACT Genomic selection (GS) emphasizes the simultaneous prediction of the genetic effects of thousands of scattered markers over the genome. Several statistical methodologies have been used in GS for the prediction of genetic merit. In general, such methodologies require certain assumptions about the data, such as the normality of the distribution of phenotypic values. To circumvent the non-normality of phenotypic values, the literature suggests the use of Bayesian Generalized Linear Regression (GBLASSO). Another alternative is the models based on machine learning, represented by methodologies such as Artificial Neural Networks (ANN), Decision Trees (DT) and related possible refinements such as Bagging, Random Forest and Boosting. This study aimed to use DT and its refinements for predicting resistance to orange rust in Arabica coffee. Additionally, DT and its refinements were used to identify the importance of markers related to the characteristic of interest. The results were compared with those from GBLASSO and ANN. Data on coffee rust resistance of 245 Arabica coffee plants genotyped for 137 markers were used. The DT refinements presented equal or inferior values of Apparent Error Rate compared to those obtained by DT, GBLASSO, and ANN. Moreover, DT refinements were able to identify important markers for the characteristic of interest. Out of 14 of the most important markers analyzed in each methodology, 9.3 markers on average were in regions of quantitative trait loci (QTLs) related to resistance to disease listed in the literature.

Resumo: O objetivo deste trabalho foi avaliar o método gráfico de rede de similaridade na classificação de genótipos de arroz (Orzya sativa) irrigado quanto a sua adaptabilidade e estabilidade. Duas medidas estatísticas foram utilizadas para representar a proximidade de comportamento (baseada na correlação de Pearson) ou de valores (baseada na distância de Gower) entre pares de genótipos ou entre genótipo e ambiente. Foram avaliados dados de produtividade de 18 genótipos de arroz irrigado em três locais de Minas Gerais, nas safras de 2012/2013, 2013/2014, 2014/2015 e 2015/2016, em delineamento em blocos ao acaso. Os genótipos foram previamente avaliados quanto à adaptabilidade e à estabilidade pelos métodos de Eberhart & Russell e centroide. As representações gráficas fornecidas pelas redes de similaridade facilitaram o reconhecimento do padrão de interação genótipo x ambiente, o que permitiu superar as dificuldades de interpretação devido à discordância entre os resultados obtidos com os métodos de Eberhart & Russell e centroide. As redes de similaridade facilitam os estudos de interação genótipo x ambiente.

Abstract: The objective of this work was to evaluate the similarity network graphic methodology for the classification of flood-irrigated rice (Orzya sativa) genotypes regarding their adaptability and stability. Two statistical measures were used to represent the proximity of the behavior (based on Pearson’s correlation) or values (based on Gower’s distance) between pairs of genotypes or between genotype and environment. Productivity data of 18 genotypes were evaluated in three locations in the state of Minas Gerais, Brazil, in the harvests of 2012/2013, 2013/2014, 2014/2015, and 2015/2016, in a randomized complete block design. The genotypes were previously assessed for adaptability and stability by the Eberhart & Russell and centroid methods. The graphical representations provided by the similarity networks allowed to better identify the pattern of the genotype x environment interaction, overcoming the interpretation difficulties due to the disagreements between the results obtained by the Eberhart & Russell and centroid methods. The similarity networks improve genotype x environment interaction studies.

ABSTRACT. The objective of this study was to compare different graphical dispersion analysis techniques in two- or three-dimensional planes. In this study, the data from different published works were used in order to determine the best methodology for analyzing the genetic diversity of different species. In this study, efficiency is measured by the amount of original distance absorbed by the projection of distances technique, which in the case of major components is equal to the amount of total variation originally available and retained by the principal components used for dispersion purposes. The projection of dissimilarity measurement technique, principal component analysis (PCA), and principal coordinate analysis (PCoA) were used. Considering the analysis by means of three orthogonal axes, the graphical dispersion efficiency was 82.22 for PCA, 87.22 for PCoA, and 85.25 for the projection of distances technique. For the 2D analysis, considering the two main axes, the mean dispersion efficiency was 69.90 for the PCA, 75.06 for the projection technique, and 78.16 for PCoA. Considering the studies carried out with experimental data of six different species, it is concluded that the principal coordinate analysis is superior.

Resumo: O objetivo deste trabalho foi avaliar o uso de redes neurais artificiais em comparação à modelagem por meio de modelos lineares generalizados na predição de resistência à ferrugem em café arábica (Coffea arabica). Foram utilizados 245 indivíduos provenientes de uma população F2, oriundos da autofecundação do híbrido F1 H511-1, resultante do cruzamento da cultivar suscetível Catuaí Amarelo IAC 64 (UFV 2148-57) e do genitor resistente Híbrido de Timor (UFV 443-03). Os 245 indivíduos foram genotipados com 137 marcadores. Realizaram-se análises com redes neurais artificiais e com modelos lineares generalizados sob o enfoque bayesiano. As redes neurais identificaram quatro marcadores importantes pertencentes a grupos de ligação que foram recentemente mapeados, enquanto o modelo generalizado bayesiano identificou somente dois marcadores pertencentes a esses grupos. Foram observadas taxas de erro de predição inferiores (1,60%) para predizer a resistência à ferrugem em café arábica, quando foram utilizadas as redes neurais artificiais em vez de modelos lineares generalizados sob o enfoque bayesiano (2,4%). Os resultados mostraram que as redes neurais artificiais são uma abordagem promissora para predizer a resistência à ferrugem em café arábica.

Abstract: The objective of this work was to evaluate the use of artificial neural networks in comparison with Bayesian generalized linear regression to predict leaf rust resistance in Arabica coffee (Coffea arabica). This study used 245 individuals of a F2 population derived from the self-fertilization of the F1 H511-1 hybrid, resulting from a crossing between the susceptible cultivar Catuaí Amarelo IAC 64 (UFV 2148-57) and the resistant parent Híbrido de Timor (UFV 443-03). The 245 individuals were genotyped with 137 markers. Artificial neural networks and Bayesian generalized linear regression analyses were performed. The artificial neural networks were able to identify four important markers belonging to linkage groups that have been recently mapped, while the Bayesian generalized model identified only two markers belonging to these groups. Lower prediction error rates (1.60%) were observed for predicting leaf rust resistance in Arabica coffee when artificial neural networks were used instead of Bayesian generalized linear regression (2.4%). The results showed that artificial neural networks are a promising approach for predicting leaf rust resistance in Arabica coffee.

Analysis using Artificial Neural Networks has been described as an approach in the decision-making process that, although incipient, has been reported as presenting high potential for use in animal and plant breeding. In this study, we introduce the procedure of using the expanded data set for training the network. Wealso proposed using statistical parameters to estimate the breeding value of genotypes in simulated scenarios, in addition to the mean phenotypic value in a feed-forward back propagation multilayer perceptron network. After evaluating artificial neural network configurations, our results showed its superiority to estimates based on linear models, as well as its applicability in the genetic value prediction process. The results further indicated the good generalization performance of the neural network model in several additional validation experiments.