http://dx.doi.org/10.15361/1984-5529.2016v44n3p403-411
403
Performance of maize genotypes of Rio Grande do Sul using
mixed models
Desempenho de genótipos de milho do Rio Grande do Sul utilizando
modelos mistos
Diego BARETTA1; Maicon NARDINO2; Ivan Ricardo CARVALHO1; Antonio Costa de OLIVEIRA3; Velci
Queiróz de SOUZA4; Luciano Carlos da MAIA5
1 Pós-Graduandos em Agronomia – Universidade Federal de Pelotas (UFPel). Área de Fitomelhoramento
barettadiego@gmail.com; carvalho.irc@gmail.com
2 Autor para correspondência. Pós Graduando em Agronomia – Universidade Federal de Pelotas (UFPel). Área de
Fitomelhoramento. nardinomn@gmail.com
3 Professor PhD. Universidade Federal de Pelotas-RS. Departamento de Fitotecnia. acostol@terra.com.br 4 Professor Dr. Universidade Federal de Santa Maria, Campus de Frederico Westphalen. velciq@gmail.com 5 Professor Dr. Universidade Federal de Pelotas-RS. Departamento de Fitotecnia. lucianoc.maia@gmail.com
Recebido em: 19-09-2015; Aceito em: 28-04-2016 Abstract
The initial process for genetic gain in maize is obtained by identifying sources of variability in open pollinated populations. For the evaluation of these populations, statistical procedures as REML/BLUP analysis, has been increasingly employed by maize breeders. This study aimed to use the method of mixed models (REML / BLUP) for the estimation of genetic parameters and evaluation to a set of agronomic characters in hybrid genotypes, synthetic and landraces and verify its potential for breeding purposes and/or indicate their crop for small farmers in the Pelotas region. The characters evaluated were: tassel length (TL), tassel branch number (TBN), ear insertion height (EI), grain yield (GY), ear row number (ERN), kernels per row (KPR) and weight of a hundred grains (WHG). Were estimated by the mixed models of variance components by restricted maximum likelihood (REML) and the components of the prediction of the average, best linear pre-dictor unbiased (BLUP). The best genotypes which presented final genotypic values superior to the general mean were the double hybrid COODETEC 308 for the characters TL, TBN, EI, GY and ERN; the single hybrid AS1551, triple hybrid AS3466 and synthetic BRS Planalto for TBN, EH, GY and ERN. Also, the pop-ulations Caiano Branco for TL, ERN and EI; Caiano Rajado for GY and ERN; Criolão for TBN and WHF; Dente de ouro for TL and ERN; Branco Roxo Indio for TL and WHG and Argentino Branco for TL and TBN. The genotypes BRS Planalto and Caiano Rajado show grain yield above the overall average and show promise for use by small and breeding programs.
Additional keywords: genotypic values; grains yield; landraces; REML/BLUP; Zea mays L. Resumo
O processo inicial para ganho genético em milho é obtido pela identificação de fontes de variabilidade em populações de polinização aberta. Para a avaliação destas populações, procedimentos estatísticos como análises REML/BLUP, têm sido cada vez mais empregados por melhoristas de milho. Este trabalho teve por objetivo a utilização do método de modelos mistos (REML/BLUP) para estimação dos parâmetros genéticos e avaliação para um conjunto de caracteres agronômicos em genótipos híbridos, sintéticos e crioulos de milho e verificar suas potencialidades para fins de melhoramento genético e/ou indicar seu cultivo para pequenos agricultores na região de Pelotas-RS. Os caracteres avaliados foram: comprimento do pendão (CP), número de ramificações do pendão (NRP), altura da inserção da espiga (AE), rendimento de grãos (RG), número de fileiras de grãos da espiga (NFE), número de grãos por fileiras (NGF) e massa de 100 grãos (MCG). Foram estimados via modelos mistos os componentes de variância pela máxima verossimi-lhança restrita (REML) e os componentes da predição de médias pelo, melhor preditor linear não viesado (BLUP). Os melhores genótipos superiores à média geral foram os híbridos CD 308 para os caracteres CP, NRP, AE, RG e NFE; AS 1551, AS 3466 e BRS Planalto para NRP, AE, RG e NFE e as populações crioulas Caiano Branco para CP, NFE e AE; Caiano Rajado para RG e NFE; Criolão para NRP e MCG; Dente de Ouro para CP e NFE; Branco Roxo Índio para CP e MCG e Argentino Branco para CP e NRP. Os genótipos crioulos BRS Planalto e Caiano Rajado apresentam o rendimento de grãos acima da média geral e mos-tram-se promissores para serem utilizados por pequenos e em programas de melhoramento genético. Palavras-chave adicionais: REML/BLUP; rendimento de grãos; valores genotípicos; variedades crioulas;
404
IntroductionThe genetic variability in maize (Zea mays L.) allows its cultivation in various environments. This cereal can be cultivated from 58 °N to 40 °S latitude, developing at sea level until 3,800 m altitude (Hallauer et al., 2010). Maize is one of the species with the greatest number of genetic studies and consequently the inheritance of many characters are already elucidated. The economic importance, the genetic structure, the number of chromosomes, the type of reproduction, the ease in making artificial pollination and the ability to generate different progeny are elements that contribute to making this cereal a model for outcrossing species (Nass & Paterniani, 2000).
Due to the high nutritional value, combined with its adaptability to different climate conditions and highest yields possible to be achieved, maize stands out as one of the most cultivated cereal in the world, with great social and economic role. Its importance should be highlighted also by the great capacity of hand labor employment generation in rural areas, due to the use as raw material for more than 500 industries, also noted in the production of biofuels (Duete et al., 2009). Brazil presented a production of 75.19 million tonnes at 15.32 million hectares, with an average yield of 4,902 kg ha-1 (CONAB, 2014).
Studies on the behavior and productive potential of landraces and maize hybrids cultivated under different agricultural systems, becomes an important tool for decision-making, as well as in the conduct of cultivation by farmers and technicians (Araujo et al., 2013). Moreover, in conditions where low crop technologies are employed, commercial hybrids may present performance close or even lower than the landraces. Moreover, the use of landraces presents several advantages related to economic viability, such as resistance to diseases, pests and seeds may be stored for subsequent crops, reducing production cost (Carpentieri-Pípolo et al., 2010).
Another very important point refers to the breeders demand for more widely knowledge, both qualitative and quantitative, on Brazilian maize germplasm, which becomes increasingly necessary, and can be evidenced by the large existing compet-itiveness in market for the development of new culti-vars. The choice of germplasm is an essential part in plant breeding programs, whether for the development of open pollinated varieties, hybrids or for basic studies, which can significantly influence in the success or failure of the selection (Araújo & Nass, 2002).
The application of more refined genetic sta-tistical procedures such as the standard analysis of estimation of variance components and predicting of components REML/ BLUP average, shows a trend in plant breeding, providing additional parameters important in the selection of superior genotypes (Maia et al., 2011). Resende & Duarte (2007) emphasize that for selection, genotypic averages will better
represent the future averages of genotypes than phenotypic averages. According to these same authors, even if the evaluations are in the same site or trial region, the effects of blocks and plots probably will not repeat, so if the tests are conducted in different regions, the effect on averages will be higher. As such effects are inserted in some propor-tion in phenotypic averages, it shows that such means are not suitable for inference on genotypic values. The choice of the method should provide the most precise and realistic inference as possible, which should be evaluated using appropriate statis-tical parameters.
The REML/BLUP are procedures that allow greater flexibility in modeling, becoming the standard procedure of statistical analysis (Duarte & Vencovsky, 2001). Thus, the use of efficient models to obtain estimates and average predictions are primordial in the search of the most promising genotypes, among a select group of cultivars (hybrids).
The use of mixed models is already performed frequently in perennial plants such as Brazilian Chestnut (Camargo et al., 2010), Aroeira (Carderalli et al., 2013), Eucalyptus (Gouveia et al., 2015; de Mendonça et al., 2015), Coffee (Carias et al., 2014) and fruitful (Della Bruna et al., 2012). However in annual plants, especially in the case of maize crop, publications addressing this theory are more recent, especially in Brazil (Souza et al., 2015).
This work aimed at using the method of mixed models (REML/BLUP) for the measurements of genetic parameters and evaluation of a character set of agronomic importance in commercial hybrids and maize landraces and verify the potential of these for breeding purposes and/or cultivation by small farmers in the region of Pelotas-RS.
Material and methods
The study was conducted in the experimental area on the Genomics and Plant Breeding Center, of the Universidade Federal de Pelotas-UFPel, located in the municipality of Capão do Leão-RS. The geographic coordinates of this city are 31°45'S latitude, longitude 52°29' W longitude, altitude of 13 m and soil was classified as Dystrophic Argisol (Santos et al., 2006).
Three commercial maize hybrids were eval-uated, being one a simple hybrid (AS 1551), one triple hybrid (AS 3466) and one double hybrid (CD 308), in addition to a synthetic variety (BRS Planalto) and nine landrace populations (Argentino Amarelo, Argentino Branco, Amarelão, Branco oito Carreiras, Branco Roxo Índio, Caiano Branco, Caiano Rajado, Criolão and Dente de Ouro) maintained by farmers of UNAIC (União das associações comunitárias do interior de Canguçu), located in the municipality of Canguçu-RS. These varieties were evaluated in the harvest of 2012/2013, in an experiment conducted in a randomized block design, with six replications.
405
The experimental units were consisted of two rows with five meters length, 0.70 meters spaced. Sowing fertilization was consisted of 200 kg ha-1 ofN-P-K, 5-20-20 formulation. The topdressing fertiliza-tion was performed at V3-V5 stage with 150 kg ha-1 of
nitrogen using urea as source. The sowing was carried out according to the agro-climatic zoning of the site. Soil management and cultural practices were performed according to the phenological stages and the culture need, which after emergence and crop establishment, the hand thinning was carried out for the arrangement of 42 plants per experimental unit, equivalent to 60,000 plants ha-¹.
The evaluated characters were: tassel length (TL, in centimeters), tassel branch number (TBN, in units), ear insertion height (EI, in centimeters), in five plants of each experimental unit, grain yield (GY, in kg ha-1 - yield extrapolated plot to ha-1), ear row
number (ERN, in units), kernels per row (KPR, in units) in five ears of each experimental unit and weight of a hundred grains (WHG, in grams).
Estimates of variance components and genetic parameters were obtained by the restricted maximum likelihood method and the best linear unbiased predictor (REML / BLUP), using genetic-statistical SELEGEN software, model 21, consisting of the following model: y=Xr+Zg+e, where y is the data vector, r is the vector of replication effects (assuming as fixed) added to average, g is the vector of genotypic effects (assuming as random), e is the vector of errors or waste. X and Z represent the incidence matrices for those mentioned effects (Resende, 2007).
The following parameters were obtained: genetic variance (σ²̂g); residual variance between plots (σ²̂e); individual phenotypic variance (σ²̂f); individual heritability coefficient in the broad sense (ĥa2); average
heritability of genotype (range) (ĥmg2 ); selective
accuracy (r̂gĝ); genetic variation coefficient
(CVg(%));residual variation coefficient (CVe(%));
rela-tive variation coefficient (CVg/CVe) (CVr(%); prediction
error variance of the genotypic values (PEV); standard deviation of the predicted genotypic value (SEP) and general average (GA).
Results and discussions
The results for the variance components for the seven characters concerned to the 13 maize genotypes are presented in Table 1. Regarding the estimates of the genetic variation coefficient (CVg(%)),
which quantifies, in percentage, the genetic variation available between evaluated genotypes, magnitudes ranging from 6.45% for KPR to 28.25% for GY were observed. According to Oliveira et al. (2011) values above 10% indicate the presence of genetic variability with the possibility of selection. Such results indicate that populations are subject to genetic progress by selection cycles, either aiming at population
improve-ment, itself, or for extracting promising inbred lineages. On the other hand, a small and discreet genetic progress is expected for KPR and GY characters in view of the low variance revealed between analyzed populations (6.45%) and (7,30).
In general, the values obtained from the residual variation coefficient (CVe(%)) have proved
low magnitude ranging from 7.06% for TL to 18.48% for TBN, maintaining at acceptable levels for field testing, according to classification of Pimentel Gomes (1985). High values of this parameter, indicate low accuracy and less experimental precision, which may be associated with the experiment size, differential response of genotypes to various stresses, incidence of pests and diseases, among others (Pinto et al., 2013).
The relative variation coefficient (CVr(%)), a
parameter which indicates the relationship between the genotypic variation coefficient and the residual variation coefficient, showed values in order of magnitude to the GY, WHG, EI, ERN, TBN, TL and KPR characters of 2. 39%, 2.33%, 2.09%, 2.03%, 1.39%, 1.03% and 0.91%, respectively. The higher this value is the greater genetic control of characters and lower environmental factors of influence on the phenotype are (Vencovsky & Barriga, 1992). In view of this, GY, WHG, EI, ERN and TBN characters had a genetic component (CVg(%) higher than the ambient
component (since the relative coefficient (CVr(%) had
values superior to 1% for these characters.
Regarding the heritability in a broad sense (ĥa2), the characters presented amplitude in
0.40 value, and the KPR character has obtained smaller magnitude with 0,45, while the GY character reached 0.85, the highest value. These results proved high magnitude, since the grain yield character generally shows magnitudes lower than 0.30 (Bernardo, 2002; Hallauer et al., 2010). For plant height and ear row number, these values vary from 0.50 to 0.70.
Regarding the average heritability of geno-type (ĥmg2 ), it is observed that the results varied from
0.833 to 0.972. For the GY character, one of the most relevant characters, value was 0,972. The (ĥmg2 ) is the
quotient that interest for projecting the successful breeding, since genotypes were selected, considering its predicted genotypic values, based on averages of several replications (Maia et al., 2009).
Selective accuracy refers to the correlation between the true genotypic value of treatment and that estimated (or predicted) resulting from infor-mation experimentally obtained (Costa et al., 2005). The results of accuracy (r̂gĝ) showed values above 0,
90 for all characters, confirming the quality of the ex-periment. According to Resende and Duarte (2007), this statistic varies from 0 to 1, and may be classified as very high (r̂gĝ≥ 0,90), high (0,70 ≤ r̂gĝ ˂ 0,90),
moderate (0,50 ≤ r̂gĝ˂ 0,70) and low (r̂gĝ ˂ 0,50. The
406
estimates above 0.90 experiments with at least six replicates are necessary, mainly for characters asyield grains which have low heritability, corroborating the results obtained in this study.
Table 1 – Estimates of genetic parameters (REML) for seven characters of agronomic importance in nine landraces and four commercial varieties of maize to the characters, tassel branch number (TBN), tassel length (TL), ear insertion height (EI), grain yield (GY), ear row number (ERN), kernels per row (KPR) and weight of 100 grains (WHG) in season 2012/2013. Pelotas-RS, FAEM/CGF, 2014.
Parameters TBN TL EI GY ERN KPR WHG σ ̂g 12.96 5.75 246.32 4237289.79 3.25 5.39 45.22 σ ̂e 6.68 5.37 56.36 736708.58 0.78 6.50 8.32 σ ̂𝑓 19.64 11.12 302.69 4973998.37 4.04 11.90 53.55 ĥa2 0.66 0.51 0.81 0.85 0.80 0.45 0.84 ĥmg2 0.92 0.86 0.96 0.97 0.96 0.83 0.97 r̂gĝ 0.96 0.93 0.98 0.98 0.98 0.91 0.98 CVg(%) 25.74 7.30 12.53 28.25 15.44 6.45 16.48 CVe(%) 18.48 7.06 5.99 11.78 7.58 7.08 7.07 CVr(%) 1.39 1.03 2.09 2.39 2.03 0.91 2.33 PEV 1.02 0.77 9.04 119327.00 0.12 0.90 1.34 SEP 1.01 0.88 3.00 345.43 0.35 0.95 1.16 GA 13.98 32.81 125.17 7285.19 11.68 35.99 40.79 (1)𝜎̂
𝑔= genetic variance; 𝜎̂𝑒= residual variance between plots; 𝜎̂𝑓= individual phenotypic variance; ℎ̂𝑎2 = individual heritability
coefficient in the broad sense; ℎ̂𝑚𝑔2 = average heritability of genotype (range); 𝑟̂𝑔𝑔̂ = selective accuracy; 𝐶𝑉𝑔(%)= genetic
variation; 𝐶𝑉𝑒(%)=residual variation coefficient; 𝐶𝑉𝑟(%)=relative variation coefficient (𝐶𝑉𝑔/𝐶𝑉𝑒); 𝑃𝐸𝑉= prediction error
variance of the genotypic values; 𝑆𝐸𝑃= standard deviation of the predicted genotypic value and 𝐺𝐴= general average.
The genetic variation coefficients (σ̂g) found
were, in descending order, 45.2, 4237289.7, 246.3, 3.2, 5.7 and 5.3 for WHG, GY, EI, ERN, TBN, TL and KPR, respectively, which resulted in 84.5%, 85.2%, 81.4%, 80.6%, 66%, 51.7% and 45.3% of total phenotypic variance, respectively. With these results it is evident the existence of genetic variation among genotypes, which shows good chances in the selection regarding the characters in question. Estimates of genetic parameters allow obtaining important information that assist in the direction of breeding programs, working as facilitators of the selection process, serving ultimately as a theoretical framework that supports the recommendation of commercial genotypes (Maia et al., 2009).
Table 2 presents predicted genotypic effects, average genotypic or genotypic values, gain and new average of 13 maize genotypes for TL, TBN and EI characters. According to Borges et al. (2010) genotypic values must be the preferable by plant breeders because these are the true values to be predicted, i.e., the environmental effects are mini-mized, remaining only the true value inherent of genotype, which reduces the chances of individuals selection favored by the environment. Values of new average are the predictions performed by BLUP for commercial crops, i.e., in commercial crops these genotypes should produce, such values on average.
In this study, it can be verified that the
geno-typic values (u + g) are very close to the new average, in all analyzed characters. For TL charac-ter, the CD 308, Dente de Ouro, Argentino Branco, Branco Roxo Índio and Caiano Branco genotypes revealed, in that order, the lower genotypic values, allocating below the overall average of 32.81 cm of the experiment by the order of the genotypes, related to TL character. For the TBN character of AS 1551, CD 308, Argentino Amarelo, Argentino Branco, Amarelão, AS 3466, Criolão and BRS Planalto genotypes revealed the lowest genotypic values allocating below the overall average of 13.98 cm, respectively. Plant breeders have observed that genetic gains in hybrids grain yield may be accom-panied by changes in other features, such as the selection for shorter tassel (Nardino et al., 2016).
Characters involved with size and tassel branch number showed considerable decrease over the past decades. The tassel branch number had a reduction of 2.5 branches per decade. Tassel weight had a reduction of 0.5 grams per decade. For the tassel length character, this had great reduction in modern maize hybrids compared to hybrids released in the 1930s, 1940s and 1950s (Duvick, 2005). Tassel may affect grain yield, reducing the light interception in the canopy of plants, as well as through the use of photo-assimilated resources pro-duced by plant (Duncan et al., 1967).
407
Table 2 - Components estimates of average BLUP (g = genotypic effect predicted, u + g = average genotype or genotypic values, gain and new average) to the character tassel length (TL), tassel branch number (TBN) and ear insertion height (EI) on local varieties and hybrids of maize in the harvest of 2012/2013. Pelotas-RS, FAEM/CGF, 2014.
Order Genotypes g u+g Gain New Average
TL
1 Amarelão 2.96 35.78 2.96 35.78
2 Branco Oito Carreiras 2.87 35.69 2.92 35.74
3 Criolão 2.57 35.39 2.80 35.62 4 Caiano Rajado 1.61 34.43 2.50 35.32 5 BRS Planalto 1.30 34.12 2.26 35.08 6 AS 3466 0.43 33.25 1.96 34.78 7 AS 1551 0.27 33.09 1.72 34.54 8 Caiano Branco -0.63 32.18 1.42 34.24 9 Argentino Amarelo -1.11 31.70 1.14 33.96
10 Branco Roxo Índio -1.41 31.40 0.88 33.70
11 Argentino Branco -2.00 30.81 0.62 33.44 12 Dente de Ouro -3.12 29.69 0.31 33.13 13 CD 308 -3.73 29.08 0.00 32.82 TBN 1 Dente de Ouro 6.40 20.39 6.40 20.39 2 Caiano Rajado 3.93 17.92 5.17 19.15 3 Caiano Branco 3.57 17.56 4.64 18.62
4 Branco Roxo Índio 1.36 15.35 3.82 17.80
5 Branco Oito Carreiras 0.16 14.15 3.09 17.07
6 BRS Planalto -0.07 13.91 2.56 16.55 7 Criolão -0.25 13.73 2.16 16.14 8 AS 3466 -0.60 13.38 1.81 15.80 9 Amarelão -0.60 13.38 1.54 15.53 10 Argentino Branco -0.82 13.15 1.31 15.29 11 Argentino Amarelo -1.82 12.15 1.02 15.01 12 CD 308 -4.28 9.70 0.58 14.56 13 AS 1551 -6.98 6.99 0.00 13.98 EI 1 Argentino Amarelo 25.88 151.05 25.88 151.05
2 Branco Roxo Índio 20.65 145.82 23.26 148.44
3 Dente de Ouro 12.73 137.90 19.75 144.92
4 Criolão 5.68 130.85 16.23 141.40
5 Amarelão 4.36 129.54 13.86 139.03
6 Caiano Rajado 4.08 129.25 12.23 137.40
7 Branco Oito Carreiras 3.32 128.50 10.96 136.13
8 Argentino Branco 2.36 127.54 9.88 135.05 9 BRS Planalto -3.39 121.77 8.41 133.58 10 AS 1551 -13.77 111.40 6.19 131.36 11 Caiano Branco -18.63 106.54 3.93 129.10 12 AS 3466 -18.63 106.54 2.05 127.22 13 CD 308 -24.67 100.49 0.00 125.17
For the EI character it was observed that the CD 308, AS 3466, Caiano Branco, AS 1551 and BRS Planalto genotypes, had in this order, the best performances, with genotypic values below the overall average 125.17 cm, showing to be promising as a gene/alleles source for reducing character. The reduction in height of ear insertion was a major change in the maize crop, enabling more efficient use of nitrogen. In addition, it allowed that the plant's center of gravity stay more balanced, reducing lodging and breakage of stems and favoring
translo-cation of nutrients to grain production (Sangoi et al., 2002). The high height of ear insertion can cause genotype presents higher susceptibility to lodging and can sometimes not be suitable for cultivation in areas with high intensity of winds and very fertile soils (Paixão et al., 2008).
The predicted genotypic effects, genotypic average or genotypic values, gains and new aver-age of 13 maize genotypes for GY, ERN and KPR characters are shown in Table 3.
408
Table 3 - Components estimates of average BLUP (g = genotypic effect, u + g = average genotype or genotypic values, genetic gain and new average) for grain yield character (GY), ear row number (ERN), kernels per row number (KPR) and weight of 100 grains, on local varieties and hybrids of maize in the harvest of 2012/2013. Pelotas-RS, FAEM/CGF, 2014.
Order Genotypes g u+g Gain New average
GY 1 BRS Planalto 3219.92 10505.11 3219.92 10505.11 2 AS 1551 2925.43 10210.62 3072.67 10357.87 3 AS 3466 2080.11 9365.30 2741.82 10027.01 4 CD 308 1855.48 9140.67 2520.23 9805.43 5 Caiano Rajado 702.49 7987.68 2156.68 9441.88 6 Amarelão -232.41 7052.78 1758.50 9043.69
7 Branco Roxo Índio -382.81 6902.37 1452.60 8737.79
8 Criolão -638.45 6646.74 1191.22 8476.41
9 Dente de Ouro -1255.15 6030.03 919.40 8204.59
10 Branco Oito Carreiras -1322.25 5962.93 695.23 7980.42
11 Argentino Branco -1646.20 5638.98 482.37 7767.57 12 Argentino Amarelo -1785.05 5500.13 293.42 7578.61 13 Caiano Branco -3521.09 3764.10 0.00 7285.19 ERN 1 CD 308 2.40 14.09 2.40 14.09 2 AS 1551 2.06 13.75 2.23 13.92 3 BRS Planalto 1.90 13.59 2.12 13.81 4 AS 3466 1.42 13.10 1.95 13.63 5 Dente de Ouro 0.94 12.62 1.75 13.43 6 Caiano Branco 0.78 12.46 1.58 13.27 7 Caiano Rajado 0.78 12.46 1.47 13.15 8 Argentino Branco -0.78 10.89 1.19 12.87 9 Argentino Amarelo -1.13 10.54 0.93 12.61 10 Criolão -1.77 9.90 0.66 12.34 11 Amarelão -2.09 9.58 0.41 12.09
12 Branco Oito Carreiras -2.26 9.42 0.18 11.87
13 Branco Roxo Índio -2.26 9.42 0.00 11.68
KPR
1 AS 1551 2.22 38.21 2.22 38.21
2 Criolão 1.67 37.66 1.95 37.94
3 Argentino Amarelo 1.53 37.52 1.81 37.80
4 Branco Oito Carreiras 1.53 37.52 1.74 37.73
5 BRS Planalto 1.53 37.52 1.70 37.69
6 AS 3466 1.43 37.42 1.65 37.64
7 Caiano Rajado 0.43 36.42 1.48 37.47
8 Amarelão -0.14 35.84 1.27 37.27
9 Branco Roxo Índio -0.33 35.65 1.10 37.09
10 Argentino Branco -1.05 34.94 0.88 36.87
11 Dente de Ouro -1.22 34.76 0.69 36.68
12 CD 308 -2.21 33.77 0.45 36.44
13 Caiano Branco -5.40 30.58 0.00 35.99
WHG
1 Branco Roxo Índio 12.48 53.27 12.48 53.27
2 Amarelão 10.76 51.56 11.62 52.41
3 Branco Oito Carreiras 6.73 47.53 9.99 50.78
4 Criolão 5.09 45.88 8.77 49.56 5 Caiano Rajado -0.25 40.53 6.96 47.75 6 Argentino Branco -0.32 40.47 5.75 46.54 7 BRS Planalto -2.83 37.95 4.52 45.31 8 Dente de Ouro -3.98 36.80 3.45 44.25 9 AS 1551 -4.75 36.04 2.54 43.34 10 CD 308 -4.78 36.00 1.81 42.60 11 AS 3466 -5.18 35.61 1.17 41.97 12 Caiano Branco -5.21 35.57 0.64 41.43 13 Argentino Amarelo -7.73 33.05 0.00 40.79
409
It is noted that for the GY character, the BRS Planalto, AS 1551, AS 3466, CD 308 and Caiano Rajado genotypes stood out from the others, revealing high genotypic values, higher than the overall average, 7285.19 kg ha-1 , in the ordering of genotypes. In thissense, it was obtained an ordering of productive performance consisted of synthetic (BRS Planalto), followed by simple hybrid (AS 1551), triple (AS 3466), double (CD 308) and open pollinated genotypes (Caiano Rajado, Amarelão, Branco Roxo Índio, Criolão, Dente de Ouro, Branco oito Carreiras, Argentino Branco, Argentino Amarelo and Caiano Branco).
The synthetic genotypes have a higher number of individuals involved in its composition, providing greater adaptability of production and greater chances of success in development condi-tions with low technology. For genotypes BRS Planalto being recommended only for the state of Rio Grande do Sul (Embrapa, 2014), which provide it a greater plasticity and stability yield in the region, could be the reasons that contributed to give to this genotype most productive performance against the other evaluated varieties. The order concerning to yield by allocating single hybrids, followed by triple, double and open pollinated genotypes is according to reports comparing 23 single hybrids, 11 triple hybrids and 8 double hybrids, on average simple hybrids are superior in productivity to triple and double (Arnhold et al., 2010). Genotypes of open pollinated maize were, on average, less productive than single hybrids, triple and double, however allow seed production by the producers themselves, reducing costs. Variations in productivity and stability can be found between different genotypes, within different classes, allowing independent selection of the genetic class to be managed (Alves et al., 2006).
Considering the ERN character, the CD 308, AS 1551, BRS Planalto, AS 3466, Dente de Ouro, Caiano Branco and Caiano Rajado genotypes presented, in this order, high genotypic values, remaining above the overall average of 11.68 rows, respectively (Table 3). For the KPR character, the positive effect, i.e., above average, 35.99 grains, classified the AS 1551, Criolão, Argentino Amarelo, Branco oito Carreiras, BRS Planalto, AS 3466 and Caiano Rajado genotypes with higher genotypic values. Genotypes with a higher ERN and KPR, result in larger number of grains per ear, which can lead to a higher grain yield, however the weight of 100 grains is also an influencing factor in crop yield (Fernandes et al., 2010).
For WHG character, analyzing the estimates values, it appears that Branco Roxo Índio, Amarelão, Branco oito Carreiras and Criolão genotypes showed the best performance, with genotypic values above the average, 40.79 grams. This higher result for weight grains of these genotypes can be due to lower ear row number, evidenced by them, and consequently, lower ear grain number, contributing to the formation of larger grains.
It is noted that the evaluated genotypes, those who had higher genotypic values above the general average for more than one character was evidenced by: CD 308 for TL, TBN, EI, GY and ERN characters; AS 1551, AS 3466 and BRS Planalto for TBN, EI, GY and ERN; Caiano Branco for TL, ERN and EI; Caiano Rajado for GY and ERN; Criolão for TBN and WHG; Dente de Ouro for TL and ERN; Branco Roxo Índio for TL and WHG and Argentino Branco for TL and TBN. These results show the existence of superior genotypes holders of desirable characteristics, such as ERN, WHG and GY, both in hybrids and in maize landraces. It is also noted that, the evaluation of genotypes used in Brazil have no importance just for producers but also for breeders, who may identify important characteristics of these for the use in breeding programs. In addition, landraces have extremely important because by constituting a source of genetic variability that can be exploited in the search for tolerant and/or resistant genes to biotic and abiotic factors (Araujo & Nass, 2002).
Conclusions
By REML/BLUP methodology, genotypes higher than the overall average were CD 308 hybrids for TL, TBN, EI, GY and ERN characters; AS 1551, AS 3466 and BRS Planalto for TBN, EI, GY and ERN and Caiano Branco landrace populations for TL, ERN and EI; Caiano Rajado for GY and ERN; Criolão for TBN and WHG; Dente de Ouro for TL and ERN; Branco Roxo Índio for TL and WHG and Argentino Branco for TL and TBN.
The BRS Planalto and Caiano Rajado gen-otypes present grain yield above the overall average and show promise for use by small and breeding programs.
The REML/BLUP methodology proved effective in achieving results in genetic evaluation involving maize hybrids and landraces, so its use is recommend maximizing genetic gain with selection of superior genotypes.
Acknowledgments
The authors thank to the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), by the PhD-fellowship concession of the first author. And also thanks to CNPq and FAPERGS. References
Alves SJ, Toledo JFFD, Araújo PMD, Garbuglio DD (2006) Comportamento de diferentes classes gené-ticas de milho com relação à estabilidade e adapta-bilidade. Revista Brasileira de Milho e Sorgo 5(2):291-303.
Araújo PM, Nass LL (2002) Caracterização e avalia-ção de populações de milho crioulo. Scientia Agricola 59(3):589-593.
410
Araujo AV, Junior DDSB, Ferreira ICPV, da Costa CA, Porto BBA (2013) Desempenho agronômico de variedades crioulas e híbridos de milho cultivados em diferentes sistemas de manejo. Revista Ciência Agronômica 44(4):885-892.Arnhold E, Pacheco CAP, Carvalho HWL, Silva R, Junior EAO (2010) Produtividade de híbridos de milho em região de fronteira agrícola no nordeste do Maranhão. Revista Brasileira Ciências Agrárias 5(4):468-473.
Bernardo, R. (2002) Breeding for quantitative traits in plants. Woodbury: Stemma Press. 369p.
Borges V, Ferreira PV, Soares L, Santos GM, Santos AMM (2010) Seleção de clones de batata-doce pelo procedimento REML/BLUP. Acta Scientiarum. Agronomy 32(4):643-649.
Camargo FF, Costa RD, Resende MD, Roa RAR, Rodrigues NB, Santos LD, Freitas AD (2010) Varia-bilidade genética para caracteres morfométricos de matrizes de castanha-do-brasil da Amazônia Mato-grossense. Acta Amazônica 40(4):705-710.
Carderalli A, da Costa RB, de Almeida Azevedo LP, Ribeiro ES, Batista BMF, de Melo RAT (2013) Sele-ção precoce em progênies de meios irmãos de
Myracrodruon urundeuva Fr. All. via metodologia
Reml/Blup. Ambiência 9(3):605-617.
Carias CMDOM, Tomaz MA, Ferrão MAG, da Fonseca AFA, Ferrão RG, Gonçalves LSA (2014) Produtividade de grãos de cafeeiro conilon de diferentes grupos de maturação pelo procedimento REML/BLUP. Semina: Ciências Agrárias 35(2):707-718.
Carpentieri-Pípolo V, Souza A, Silva DA, Barreto TP, Garbuglio DD, Ferreira JM (2010) Avaliação de cultivares de milho crioulo em sistema de baixo nível tecnológico. Acta Scientiarum. Agronomy 32(2): 229-233.
CONAB (2014) Companhia Nacional de Abasteci-mento. Acompanhamento da Safra Brasileira. Disponível
em:<http://www.conab.gov.br/OlalaCMS/uploads/arq uivos/14_05_08_10_11_00_boletim_graos_maio_20 14.pdf> (Acesso em 05 mai. 2014).
Costa RB, de Resende MDV, Contini AZ, Rego FLH, Roa RAR, Martins WJ (2005) Avaliação genética de indivíduos de erva-mate (Ilex paraguariensis St. Hil.) na região de Caarapó, MS, pelo procedimento REML/BLUP. Ciência Florestal, Santa Maria 15(4):371-376.
Della Bruna E, Moreto AL, Dalbó MA (2012) Uso do coeficiente de repetibilidade na seleção de clones de pessegueiro para o litoral sul de Santa Catarina. Revista Brasileira Fruticultura 34(1):206-215.
de Mendonça, A. R., de Pádua Chaves, S. Carvalho, N. C. (2015) Modelos hipsométricos gene-ralizados mistos na predição da altura de Eucalyptus sp. Cerne 21(1):107-115.
Duarte, JB, Vencovsky, R. (2001) Estimação e predição por modelo linear misto com ênfase na ordenação de médias de tratamentos genéticos. Scientia Agricola 58(1):109-117.
Duete RRC, Muraoka T, Silva E, Trevelin P, Ambrosano EJ (2009) Viabilidade econômica de doses e parcelamentos da adubação nitrogenada na cultura do milho em Latossolo Vermelho Eutrófico. Acta Scientiarum Agronomy 31(1):175-181.
Duncan WG, Williams WA, Loomis RS (1967) Tas-sels and the productivity of maize. Crop Science 7(1):37-39.
Duvick DN (2005) The contribution of breeding to yield advances in maize (Zea mays L.). Advances in Agronomy 86:83-145.
Embrapa (2014) Empresa Brasileira de Pesquisa Agropecuária. Disponível em: < http://www.cnpt.embrapa.br/culturas/milho/cultivares /c_miplanalto.htm > (Acesso em 05 mai. 2014). Fernandes, FCS, Arf, SBO, Andrade, JADC. (2010) Doses, eficiência e uso de nitrogênio por seis culti-vares de milho. Revista Brasileira de Milho e Sorgo 4(2):195-204.
Gouveia, J. F., da Silva, J. A. A., Ferreira, R. L. C., Gadelha, F. H. L., de Araújo Lima Filho, L. M. (2015) Modelos volumétricos mistos em clones de eucalyptus no polo gesseiro do araripe, Pernam-buco. Floresta, 45(3): 587-598.
Hallauer AR, Carena MJ, Miranda Filho JB de (2010) Quantitative genetics in maize breeding. Springer. 663p.
Maia MCC, de Resende MDV, de Paiva JR, Cavalcanti JJV, de Moura Barros L (2009) Seleção simultânea para produção, adaptabilidade e estabi-lidade genotípicas em clones de cajueiro, via modelos mistos. Pesquisa Agropecuária Tropical 39(1):43-50.
Maia MCC, Resende MDV, Oliveira LC, Álvares VS, Maciel VT, Lima AC (2011) Seleção de clones expe-rimentais de cupuaçu para características agroin-dustriais via modelos mistos. Revista Agro@mbiente 5(1):35-43.
411
Nardino, M, Souza, VQ, Baretta, D, Konflanz, VA, Follmann, DN, Carvalho, IR, Ferrari, M, Caron, BO, Schmidt, D (2016) Partial diallel analysis among maize lines for characteristics related to the tassel and the productivity African Journal of Agricultural Research 11(11):974-982.Nass LL, Paterniani E (2000) Pre-breeding: a link between genetic resources and maize breeding. Scientia Agricola 57(3):581-587.
Oliveira RA, Daros E, de Resende MDV, Bespalhok-Filho JC, Zambon JLC, Souza TR, Lucius ASF (2011) Procedimento Blupis e seleção massal em cana-de-açúcar. Bragantia 70(4):796-800.
Paixão SL, Cavalcante M, Ferreira PV, da Silva Madalena JA, Pereira RG (2008) Divergência gené-tica e avaliação de populações de milho em dife-rentes ambientes no estado de Alagoas. Revista Caatinga 21(4):191-195.
Pimentel Gomes, F. (1985) Curso de Estatística Experimental. São Paulo: Nobel, 467p.
Pinto FDO, da Luz LN, Pereira MG, Cardoso DL, Ramos HC (2013) Metodologia dos modelos mistos para seleção combinada em progênies segregantes de mamoeiro. Revista Brasileira de Ciências Agrarias 8(2):211-217.
Resende MDV (2007) SELEGEN‑REML/BLUP: Sistema estatístico e seleção genética computadori-zada via modelos lineares mistos. Colombo: Embrapa Florestas. 359p.
Resende MDV, Duarte JB (2007) Precisão e con-trole de qualidade em experimentos de avaliação de cultivares. Pesquisa Agropecuária Tropical 37(3):182-194.
Sangoi L, Almeida MD, Silva PD, Argenta G (2002) Bases morfofisiológicas para maior tolerância dos híbridos modernos de milho a altas densidades de plantas. Bragantia 61(2):101-110.
Santos HG, Jacomine PKT, Anjos LHC, Oliveira VA, Oliveira JB, Coelho MR, Lumbreras JF, Cunha TJF (2006) Sistema brasileiro de classificação de solos. Rio de Janeiro: Embrapa Solos. 306p.
Souza, VQ, Baretta, D, Nardino, M, Carvalho, IR, Follmann, DN, Konflanz, VA, Schmidt, D. (2015) Variance components and association between
maize hybrids morpho-agronomic characters.
Científica 43(3):246-253.
Vencovsky R, Barriga P (1992) Genética biométrica no melhoramento. Ribeirão Preto: SBG. 496p.
