Estimates of genetic parameters in F4 – F5 soybean populations resistant to Asian soybean rust

The objective of the present study is the evaluation of genetic parameters in F4 and F5 soybean populations from two crosses, which are potentially resistant to Asian soybean rust. The genotypes were developed from the cross between parents, which were resistant and susceptible to the disease, and totaled 137 genotypes in generation F4 and 283 genotypes in F5. The experimental design was augmented blocks with two checks between the treatments. The following agronomic characters were evaluated: plant height at maturity, first pod insertion height, number of nodes and branches, agronomic value, lodging, number of pods and seeds per plant and seed production. After the genotypes were submitted to analyses of variance, heritability, and selection gains, it was observed that Cross 1 had superior progenies, which were potentially resistant to Asian soybean rust and also had good productivity.


INTRODUCTION
The perspective for soybeans is excellent at both the national and world levels.Brazil cultivated a total grain area of around 53.26 million hectares during the 2012/13 crop season, 4.7% higher than for 2011/12, corresponding to an area increase of 2.38 million hectares.The total soybean area [Glycine max (L) Merrill] in Brazil is 27,721.60 million hectares and it increased by 10.7%, or 2.68 million hectares, compared to the previous crop.Productivity was an average 2,938 kg/ha, a 10.8% increase compared to 2012, which was a record, with the estimated production being 81,456.10 million tons, a 22.7% increase over the previous crop (Conab, 2013).This significant development of the soybean crop can be attributed to Genetic Improvement Programs, focusing on the selection of progeny, which have desirable characteristics, including the production of erect plants of a suitable height, facilitating mechanical harvesting; higher grain yield, requiring a smaller area for high production; resistance to diseases and insects, reducing production; resistance to diseases and insects, reducing losses; oil and protein content for foodstuffs and also biodiesel production; as well as early maturation and drought resistance to better adapt to different regions and adverse climates (Sediyama et al., 2009).The selection of superior genotypes in a Genetic Improvement Program is based on the estimate of genetic and phenotypic parameters, together with an experimental design and a suitable selection method.This constitutes a solid basis for determining the best performances of the agronomic characters to be studied, consequently resulting in potentially better selection (Falconer, 1987).
The objective of the current study was to evaluate genetic parameters from selection in F 4 and F 5 soybeans, originating from two crosses and which were potentially resistant to Asian soybean rust.

Experimental site characteristics
The trial was done during the 2009/10 and 2010/11 crop seasons, represented by the F4 and F5 generations respectively, on the Teaching, Research and Extension Farm (FEPE) of the University of Agricultural Science and Veterinary -UNESP, Jaboticabal Campus São Paulo state (latitude 21° 14' 05'' S, longitude 48º 17' 09'' W, altitude 615.01 m).The predominant soil type is a Red Eutroferric Latosoil and the climate is subtropical (Cwa).
The area was plowed and harrowed twice and all the cultural practices recommended for soybeans were applied (Embrapa, 2011).The experimental design was augmented blocks, with 137 genotypes in the F4 and 283 genotypes in F5 generation, and two checks between them.Each family was derived from the threshing of an individual plant.Sowing was manual in 5 m long rows, 0.5 m apart, resulting in a planting density of around 20 seeds per linear meter.
The checks used were the comercial cutlivars: MGBR 46-Conquista and Coodetec-219.Eight plants were evaluated per plot.

Genetic material
The genotypes used were developed from crosses between parents, which were resistant (R) and susceptible (S) to Asian soybean rust: PI 200526 Shiranui (R) x COODETEC 205 (S) and PI 200456 (R) x MG/BR-46 (Conquista) (S) (Costa, 2008).The parents resistant to the pathogen correspond to the introduction of exotic plants (PI's) and the susceptible plants are cultivars already adapted to Brazilian conditions, with a good agronomic performance for grain-producing characters.
The genotypes are in the F4 and F5 generations and were selected according to their performance for Asian rust resistance, previously evaluated in a greenhouse.

Agronomic characters evaluated in the F4 and F5 generations
The genotypes were evaluated and selected for the main soybean agronomic characters in the R8 stage, according to the Fehr and Caviness (1977) scale: Plant height at maturity (PHM)the distance in cm between the soil surface and the main stem tip at maturity; first pod insertion height (IHP)distance in cm between the soil surface and insertion of the first pod; agronomic value (AV) evaluated using a visual scale varying from 1 (poor plants) to 5 (excellent plants), with the ranking being representative of a group of visual characters: architecture, number of full pods, vigor and Charnai et al. 1201 health, premature opening of the pods, lodging and foliar retention at maturity; lodging (L)character evaluated using a visual scale varying from 1 (all plants in the plot erect) to 5 (all plants lodged); number of nodes (NN), the number of nodes on the plants at maturity; number of branches (NB), the number of branches per plant at maturity; number of pods per plant (NPP), the total number of pods on the plant; number of seeds per plant (NSP), the number of seeds produced per plant, and seed production per plant (SP), weight in grams of grain produced per individual plant.
The rankings attributed to the L and AR characters were transformed to 5 .0  x to obtain more normally distributed data.

Statistical analyses
All the analyses were done using the Genes Software, 2008 version (Cruz, 2008).Analyses of variance were done for each characteristic evaluated, for each control and for each check and the segregated population (family) of each crossing in the two generations, with the statistical model adopted according to Cruz (2001).The statistical model for the analysis of variance was: Yij = µ + fi + ei + pij + ij, where Yij is the observation corresponding to the j th plant of the i th famíly; µ is the overall mean of the generation, whether it be the check or the family; with fi being fi ~ NID (0, ) the genetic effect attibuted to the i th famíly, with i = 1,2...n; ei with ei~ NID (0, ) being the environmental effect between rows of a check or families; pij with pij~ NID (0, ) being the genetic effect attributed to the j th plant of the i th family, with j = 1,2...n; ij and the environmental effect between plants within rows of a check or family.
Thus, with the data between and within the plots of the checks and the segregated lines, the phenotypic, genotypic, environmental and additive components were estimated, which also permitted the estimation of the heritability coefficients in the narrow and broad sense, between and within the families, according to the following expressions (Cruz, 2001) Where: : phenotypic variance between plants within the families; : genotypic variance between plants within families; : environmental variance between plants within families; : phenotypic variance between families; : genotypic variance between families; F: coefficient of endogamy, varying from 0.87 and 0.93 for the F4 and F5 generations, respectively; : additive variance.
The calculations for the selection gain or selection response were made as follows (Falconer and Mackay, 1996): GS= h²S where, GS: selection gain or selection response; h²: heritability coefficient; S= (XSi + XOi), with S the selection differential; XSi the mean of the individuals selected for the character i and XOi the original mean of Table 1.Summary of analysis of variance for the characters: plant height at maturity (PHM), first pod insertion height (IHP), number of nodes (NN), number of branches (NB), number of pods per plant (NPP), number of seeds per plant (NSP) and seed production (SP) , in F4 and F5 soybean progenies for Crosses 1 and 2, at Jaboticabal, São Paulo state.

Generation F4
Cross the population, considering a selection intensity of 25%.

RESULTS AND DISCUSSION
Lodging and agronomic value characteristics were not analyzed statistically because the rankings attributed by plant breeders were subjective.The plants were evaluated in the field and only those which showed suitable phenotypes for these characteristics were selected.
The analysis of variance of the F 4 generation, Cross 1 (Table 1), showed that the characters first pod insertion height and seed production per plant showed significant differences between the genotypes (P≤5%), not observed for the remaining characters.The characters number of nodes and number of pods per plant in Cross 2 were significantly different (P≤5%) as also were plant height at maturity, first pod insertion height, number of branches, number of seeds per plant and seed production per plant (P≤1%).This demonstrates a higher variability in the Cross 2 genotypes, making possible greater genetic gains with the selection.
The genetic variation coefficients in the two crosses had values below 25%, including the quantitative characters, which were near the limit of up to 20% for cultivars in advanced generation of inbreeding (Brasil, 1998).This indicated good experimental precision, that is, a narrow band of values for each character, even the quantitative ones, highly influenced by the environment, lying within a normal distribution curve.The CVg/CVe ratios were close to 1 for most characters, favoring the selection of superior genotypes (Cruz and Regazzi, 1997).
There were significant differences for the characters first pod insertion height, number of nodes and number of branches (P≤5%) and also for plant height at maturity, number of pods per plant, number of seeds per plant and seed production per plant (P≤1%) in Cross 1 of the F 5 generation (Table 1).The results of Cross 2 were similar to those of Cross 1 but first pod insertion height and number of branches were not significant.The values of the genetic variation coefficient in Cross 1 were within acceptable limits, except for seed production per plant (33.23%).In Cross 2, the characters number of pods per plant, number of seeds per plant and seed production per plant had values of 39.30, 36.30and 43.31, respectively, higher than ideal due to the fact that they are highly influenced by the environment and, therefore, subject to a wider range of values.These characters also showed a lower experimental precision when compared to the previous generation, which was caused by environmental change.The CVg/Cve ratio was higher than 1 in both crosses, once again demonstrating good selection results.
The genetic variance in both generations of Cross 2 was higher than the environmental variance, demonstrating the predominance of genetic componentes compared to environmental ones, and indicating favorable conditions for improving the characters under evaluation.According to Burton (1952), the genetic variation coefficient (CVg) should be associated with heritability in order to compare the genetic variability of different populations and characters and help in predicting genetic gain.Thus, high values for the genetic variation coefficient generate high values of heritability and are associated with greater genetic variability, greater selective accuracy and more possibilities for successfully selecting soybean lines with better grain yield (Cargnelutti Filho et al., 2003;Storck and Ribeiro, 2011).
When comparing broad heritability values between and within familes in the analysis of the heritability coefficient in Cross 1, in generation F 4 (Table 2), the characters first pod insertion height (0.60), number of nodes (0.39) and number of branches (0.40) showed the highest values for broad heritability between the families, whereas the number of pods per plant (0.57), number of seeds per plant (0.54) and seed production per plant (0.67) were superior for broad heritability within the families.However, in an Improvement Program, the total restricted heritability is considered the most reliable for character heritability since it shows additive genetic variance as a component.This is the main reason for the parental similarity, and shows the genetic properties observed in a population and its response to selection.Its job is to orient the relative quantity of genetic variance used in improvement (Falconer, 1987).The highest estimates of total restricted heritability were for the characters plant height at maturity (0.14), number of nodes (0.19) and number of branches (0.13), whereas the lowest values were for first pod insertion height (0.05), number of pods per plant (0.07), number of seeds per plant (0.08) and seed production per plant (0.07).
The heritability in the broad and narrow senses between families in Cross 2 were superior to the heritability within families, indicating that the genetic fraction is more important in determining the phenotypic differences between genotypes than the differences between individuals of the same family (Backes et al., 2002).The highest values obtained for total restricted heritability between the genotypes were for plant height at maturity (0.38), first pod insertion height (0.31) and number of branches (0.28) and the lowest estimates continued to be for the characters directly linked to production.This was expected since they are quantitative characters, controlled by a large number of genes and, therefore, has a greater environmental influence (Backes, 2002;Mauro et al., 2000;Toledo et al., 2000).
Once again, in generation F 5 (Table 2), heritabilities in the broad and narrow senses were greater between than within families.All the characters in Cross 2 had values higher than 50%, indicating that the genetic contributions are more pronounced than those attributed to environmental factors in the phenotypic expression of the character (Falconer, 1987).The characters with the highest values in total restricted heritability in Cross 1 were plant height at maturity (0.52) and number of nodes (0.27), whereas the lowest estimates were for number of seeds per plant (0.10) and seed production per plant (0.08).The characters plant height at maturity (0.85), number of nodes (0.49) and number of seeds per plant (0.24) in Cross 2 had the highest values for the coefficients of total restricted heritability, with number of branches (0.06) and seed production per plant (0.06) having the lowest values.
When comparing the two crosses and the two generations it can be observed that the values for restricted heritabilities within the families are always less than those for the broad heritability within the families.This is due to the restricted heritability being estimated by the additive variance, which, in turn, represents only a Table 2. Estimates of Heritability Coefficients in the broad and narrow senses, between and within families, and total restricted for the characters: Plant height at maturity (PHM), first pod insertion height (IHP), number of nodes (NN), number of branches (NB), number of pods per plant (NPP), number of seeds per plant (NSP) and seed production (SP), in F4 and F5 soybean progenies for Crosses 1 and 2, at Jaboticabal, São Paulo state.

Generation F4
Cross part of the genotypic variance used in calculating broad heritability.This relationship is extremely valuable because it shows the importance of predicting gain based on restricted heritability, and this is even more important when there is an interest in comparing the potentials of different populations, which is the objective of this study (Backes et al., 2002).The analysis of the selection gain estimates in generation F 4 (Table 3) demonstrates that in Crosses 1 and 2 mass gains are superior when compared to selection gains between and within (GSed%) families for most of the characters, except for number of nodes and number of branches in Cross 1.
The highest values in selection mass gains in Cross 1 were for the characters number of branches (9.74%), number of pods per plant (6.20%) and seed production per plant (7.16%).Considering the primary production characters (number of pods per plant and seed production per plant), selecting better genotypes for this character suggests that greater indirect productivity can be achieved, it was mentioned by Vianna et al. (2013) also.On the other hand, the biggest gains for Cross 2 were for the characters first pod insertion height (11.76%) and number of branches (19.44%).The importance for number of branches is that a plant with more branches means more pods and, consequently, a higher productivity (Navarro Junior and Costa, 2002).
There is no superiority for the mass gain compared to selection gains for generation F 5 between and within families in Cross 1 (Table 3).However, the same superiority in the previous generation can be observed for Cross 2 for the most characters, except for first pod insertion height, which is important for this character since first pod insertion height should not be higher than Table 3. Estimates of selection gain between and within families, selection between and within families and mass selection for the characters: Plant height at maturity (PHM), first pod insertion height (IHV), number of nodes (NN), number of branches (NB), number of pods per plant (NPP), number of seeds per plant (NSP) and seed production (SP), in F4 and F5 soybean progenies for Crosses 1 and 2, at Jaboticabal, São Paulo state.

Generation F4
Cross 12 cm to avoid production losses (Sediyama et al., 1999).The highest values of mass selection gain in Cross 1 were obtained for the characters plant height at maturity (17.51%), first pod insertion height (9.89%) and number of nodes (8.08%).The height of the soybean plant has a significant indirect effect on pod number and that is the reason taller plants are selected and with a suitable first pod insertion height due to the tendency for these plants to produce more pods and, therefore, be more productive (Peluzio et al., 2005).
However, when selecting for this character, mechanical harvesting must be taken into account and soybean plants should be between 60 and 120 cm tall (Rezende and Carvalho, 2007).According to Jiang and Elgli (1993), the character for the number of nodes also contributes indirectly to production since more nodes can result in more flowers per plant and, consequently, more pods.The highest values for the mass selection gain for Cross 2 were for the characters: Plant height at maturity (30.0%), number of nodes (20.38%), number of pods per plant (17.99%) and number of seeds per plant (26.29%).The values of selection gains for plant height at maturity in the second crop stood out and this second cross had high values compared to Cross 1 for most characters, principally production characters.
Negative values were observed for number of branches, which is not desirable and this may be due to some environmental interference since stimulating the development of tall plants means that plants are less branched and have a tendency to lodge.This explains the inversion in values as well as showing that this character had a value considered high in the heritability estimate, thus resulting in low selection gain estimates.
With an analysis of the selection gain estimates it was possible to observe that the use of simple improvement methods, such as mass selection, gives significant results and is a satisfactory selection process, even considering characters governed by many genes and highly influenced by the environment, such as for number of pods per plant, number of seeds per plant and seed production per plant.Also, if the primary production characters are considered, which are those which determine selection success when the objective is a production increase, Cross 1 remained more stable with significant gains and a greater potential for obtaining superior progenies when compared to Cross 2.

Conclusions
1.The genotypes with potential for resistance to Asian soybean rust, which belonged to the two crosses, showed good agronomic field performance; 2. Cross 1 (PI 200526 Shiranui x COODETEC 205) was superior to Cross 2 (PI 200456 x MGBR-46 Conquista), and was the most promising for developing new cultivars.