GENETIC VARIABILITY AND EXPECTED GAIN IN THREE MAIZE POPULATIONS

The study of genetic variability in populations undergoing recurrent selection is important to quantify their real potential for breeding purposes. In the present work, three semiexotic maize populations were evaluated on their yield potential and variability for continuing the recurrent selection program. Half-sib families representing three semiexotic populations CRE-01, CRE-02, CRE-03 were evaluated for yield and agronomic traits male flowering (MF), plant height (PH), ear height (EH), ear placement (EP) ratio = (EH/PH), tassel length (TL), tassel branches (TN), final stand (FS), ears per plot (NE), prolificacy (PR), ear length (EL), ear diameter (ED), ear yield (EY), grain yield (GY). The observed means for grain yield (GY), in percent of the hybrid check, were 93.8%, 89.4%, and 94.6%, showing a relatively good yield potential. Means (cm) for plant height (PH) and ear height (EH) were 221, 237, 241 and 117, 134, 137, respectively, with reductions in relation to the first cycle. Parameters estimates for GY were: additive genetic variance 582.28, 406.70 and 238.11 g2.plant-1; heritability 59.17%, 49.82% and 38.53%; progeny mean basis, and expected gain for progeny selection 8.16%, 6.86% and 3.88%; selection intensity of 20%.

The importance of the maize species as food and especially for animal feed, as well as an expressive economic component, is unquestionable.
Its wide distribution in Brazil and all over the world also contributes for the enlargement of the agricultural activities, besides the introduction of new technologies that will result in higher efficiency of the productive system (Hallauer at al., 2010;Santos et al., 2002).
The activity of plant breeding has been practiced since the primordial times of human civilization and was greatly benefited by the advent of the science of genetics in the last century.The genetic advances, as expressed by the increase in productivity and/or quality in the food production system, has been continuously achieved with success but under the basic requirement that is dependent of existence of genetic variability.In addition, the advances in technology for the control of the environmental effect on the expression of genetic variability also contributed to the efficiency of selection methods.The technology of hybrid seed production, introduced in the beginning of the last century also contributed for widening the horizon of maize breeding.The available methodologies for maize breeding are presented by Hallauer at al. (2010);Hallauer et al. (1988); Miranda Filho and Viégas (1987); Paterniani and Miranda Filho (1987), among others.
In Brazil, the introduction of exotic germplasm has largely contributed to increase the variability and the efficiency of breeding programs for the development of cultivars (Miranda Filho, 1992;Regitano Neto et al., 1997).Old varieties such as Cateto and Dente Paulista were the base for the development of the first hybrid seeds.In the sequence, an expressive advance was brought about with the introduction of exotic germplasm, mainly Tuxpeño and related races from Mexico and Central America, which largely contributed for the development of high yielding semident hybrids (Miranda Filho & Viégas, 1987).Several other introductions have contributed to the maize breeding programs along the time.The importance of incorporation of exotic germplasm in maize breeding programs have long been emphasized (Nass et al., 2001), but the pattern of adaptation in general is below an acceptable level, thus making difficult to attain the desirable objectives.Nevertheless, some experiences have been accumulated and surely will contribute to overcome inherent obstacles (Goodman, 2005;Paterniani, 1990).In Brazil, the introduced germplasm has been mostly from tropical origin, thus making to be less difficult to handle as compared to the introduction of temperate germplasm.In spite of the importance of previous actions for the introduction of exotic germplasm in Brazil (Miranda Filho & Viégas, 1987), recently there has been reined interest of maize breeders for new sources, mainly in relation to tolerance to biotic and non-biotic stresses (Basso & Miranda Filho, 2001;Oliveira et al., 2015a).
Population improvement by using the method of recurrent selection (intra or interpopulation) has been extensively used and several mating schemes are available to be used in breeding programs (Hallauer et al., 2010;Paterniani & Miranda Filho, 1987).All

Material and Methods
A sample of 682 inbred (S 4 ) lines with potential for resistance to corn stunt complex was introduced from CIMMYT (Centro Internacional de Mejoramiento de Maíz y Trigo, Colombia).Fiftyone lines were preliminary selected for incorporation  and expressed in t ha -1 .Two commercial hybrids (P30K75 and SHS 4080) were used as checks, intercalated between sets of ten plots of families.
In the analyses of variance, the following model For the purposes of this work, the following expectations are of interest: E(fi) = E(eij) = 0, E() = (genetic variance among half-sib families), and E() = s2 (plot-to-plot error variance).In the analysis of variance the mean squares for Families and Error have the following expected values: E(Mf) = s2 + 3 and E(Me) = s2.The estimation of the components of variance and other related parameters were obtained according to methods acording Andrade and Miranda Filho (2008) and Vencovsky and Barriga (1992).

Results and Discussion
Observed means of fourteen traits are shown in Table 2 for half-sib families and checks in the analysis of three semiexotic populations (CRE-01, CRE-02, CRE-03); all the results are shown in that sequence.
Means for MF (male flowering, days) were 60.9, 63.1 and 61.7, all above the check mean (4% to 6% or 2.4 to 3.5 days).Recently, earliness has been an important trait in most maize breeding programs because, despite the positive correlation with grain yield (0.14 on the average of 13 estimates (Hallauer et al., 2010)), it increases the feasibility of the second crop (usually maize after soybean), mainly in the Central-West region of Brazil.Therefore, the decrease in days to flower would be desirable in the three semiexotic populations if they are to be used for those conditions.The tassel characteristics (TL -length and TN -tassel branch number), that essentially exhibit the tassel size, showed means around 0.38 m and Table 1 -Observed means of fourteen traits in half-sib families (m 0 ) from three semiexotic maize populations (CRE-01, CRE-02, CRE-03) and hybrid checks (m c ).
In the present study, the tassel characteristics were ST is a measure of prolificacy that is an important trait to be considered in breeding programs.The level of prolificacy was considered low in all three populations, with means of 1.22, 1.04 and 1.00 in the sequence.
Ear length of family means were 17.5, 17.7 and 16.7 cm on the average for populations; such values are equivalent to 96.5%, 103.5% and 95.4% of the check mean.However, the range of variation was relatively wide, with limits of 15.5 to 19.9 in CRE-01 and CRE-02 and of 13.9 to 19.0 in CRE-03.
For ear diameter (cm) all three populations showed 14 branches, in the set of three populations.Such results showed to be somewhat lower than the means for three populations (Dent Composite, Flint Composite, Centralmex) that were around 65 cm and 20 branches, respectively, reported by Geraldi et al. (1985); they were also lower but closer to 40.9 and 18.4, respectively, reported by Andrade and Miranda Filho (2008) for the population ESALQ-PB1.Mock and Pearce (1975) discussed the importance of some traits in the formulation of an ideotype of maize, which included the tassel size for its negative correlation with grain yield.Paterniani (1981) also emphasized the effect of tassel size on the maize plant development and the importance of this trait in breeding programs.
The author also reported on the positive correlation between TN and EP (ear placement), varying from 0.24 to 0.85 in sets of races representing ten countries.
The additive genetic correlation between GY and TN has been reported as negative by Souza Júnior et al.     4).On the average of the three populations, the estimates ( 2 f h > 60%) were for MF, PH, EH, TL and TN; estimates (40% < were found in PR, EL, ED, EY and GY; and 2 f h < 40% was exhibited only by EP.Andrade and   Miranda Filho (2008)  The ratio q = CVg/CVe was presented by Vencovsky and Barriga (1992) as an indicator of a circumstance favoring artificial selection, based on the expression of genetic variability; the authors hypothesized that values of q near or higher than 1 are favorable for selection.Estimates of q > 1 were detected only for TN in CRE-01 and CRE-02, but 82% of the estimates were higher than 0.5.The yield traits (EY and GY) also showed q < 0.5 in CRE-03.Andrade and Miranda Filho (2008) also reported q < 0.5 for yield traits but 0.5 < q < 0.9 for other ten traits in the same study.
The results as described above are quite common in experiments with open pollinated maize populations and it is well known that when dealing with quantitative traits the success of selection depends not only on the genetic variability, but also on experimental techniques and on the effectiveness in controlling the effect of environment.For that reason, if high selection intensity is to be used in a short-term program, care must be taken to assure a higher level of heritability; for examples, larger plot sizes and/or higher number of replications should be used for that purpose.GY, GS, in percent of the population mean, were 7. 24, 6.34, 3.02 and 8.16, 6.86, 3.88 in the order CRE-01, CRE-02 e CRE-0, respectively.On the other hand, higher selection intensity (5.6% in CRE-01; 5.0% in CRE-02 and CRE-03) indicated gains of the order 9. 96, 9.30, 4.83 and 10.81, 10.47, 5.32  pressure, the changes were 3.26% and 5.10%, respectively.For all other traits, changes expected in the mean of half-sib families were below 2%.

Conclusions
The semiexotic populations revealed good potential for grain yield and acceptable patterns for other traits to be useful in breeding programs.
Grain yield for the three semiexotic populations showed relative values of CRE-02 < CRE-01 < CRE-03 which are equivalent to 89.4%, 93.8% and 94.6% of check means, respectively.
The quantification of genetic variability showed

A
work developed by Miranda Filho and Reis (2016), with several levels of exotic germplasm involving five subsets of populations, three of tropical origin and two of temperate origin, indicated good performance for yield (weight of spikes), the average of three sets of tropical origin of 79.5; 74.4 and 76.3% of the control and an average of the two sets of temperate origin of 72.7 and 71.3% of the control.
methods are quite feasible and in general have shown high efficiency not only for population improvement but also for the evaluation of the potential of germplasm to be used in breeding programs(Borém & Miranda, 2009).Each cycle of recurrentselection is completed after closing the three steps designated as (a) development of genotypes or progenies, (b) evaluation and selection of progenies, and (c) recombination of the selected progenies.However, the success of the breeding program depends essentially on the amount of genetic variability of important quantitative traits that will be submitted to selection (Hallauer et al., 2010).Under this view, the present work pursued the following objectives: (i) Quantification of the genetic variability of important agronomic traits in three semiexotic populations the maize.(ii) Evaluation of the expression of grain yield and other traits of the three populations in relation to their potential to be used as basic germplasm for the development of cultivars (improved varieties and hybrids from inbred lines) with acceptable pattern in the maize crop production.
into three local and adapted populations [P-3041 ≡ F 2 generation of a commercial hybrid; CMS-14C ≡ population derived from Pool-25 (CIMMYT, Mexico); and ESALQ-PB23 ≡ broad base (composite) population].Crosses of the three base populations with the introduced lines were used for the synthesis of the three semiexotic populations designated as CRE-01, CRE-02 and CRE-03, after recombination of testcrosses represented by 30, 39 and 32 exotic lines, respectively.Oliveira et al. (2015a) and Oliveira et al. (2015b) give more details about the origin of the three semiexotic populations.Oliveira et al. (2015b) reported on the evaluation of half-sib families from the three semiexotic populations, represented by samples of size 200, 180 and in which Y ij is the experimental unit referring to the i th treatment in the j th block, represented by plot total (EY and GY), sample means (PH, EH, EL, ED) or the specific unity for the other traits.In that model, m is the general mean, and the effect t i with I-1 degrees of freedom is partitioned to represent the variation among half-sib families (f i effects), the contrast between check means and the contrast families vs. checks with n-1, 1 and 1 degrees of freedom, respectively; n is the number of half-sib families.The random effects b j and e ij refer to the variation among blocks and the experimental error, respectively, with J-1 and (I-1)(J-1) degrees of freedom.
Troyer (1994) presented a detailed discussion on the importance of breeding for early maize for the northern part of the American Corn Belt, mainly to develop germplasm adapted to the short crop duration at higher latitudes.Early maize are then desirable to attend some needs such as: (a) to avoid the risk of drought during the crop development, as occurs in the Central-West region in Brazil; (b) to escape from frost risk in the autumn season, as occurs frequently in the South region in Brazil; (c) to escape from snowfall at the maize harvest time in high latitude regions; (d) simply to shorten the crop cycle to make possible a second crop or to take advantage in the market prices.For PH (plant height) means were 221.4,237.0, 241.5 cm and EH (ear height) were 116.7, 134.6, 137.3 cm; for both, PH and EH, estimates were above the check means, in the range of 10% to 14% and 9% to 17%.For both traits the relative values were CRE-01 < CRE-02 < CRE-03, which are the same as those reported by Oliveira et al. (2015a) and Oliveira et al. (2015b) for the same populations; the authors also found higher means for families as compared to check means in both experiments.The importance of plant height and ear height, and indirectly ear placement, in typically tropical germplasm, has long deserved attention of maize breeders for the development of modern cultivars with a better plant architecture toward the control of plant lodging and mechanical harvest.However, there is a positive genetic correlation between yield with PH (r = 0.26) and EH (r = 0.31) as reported by Hallauer et al. (2010) on the average of 23 studies.Andrade and Miranda Filho (2008) also found positive correlation between ear weight and grain weight with PH (~0.60) and EH (~0.50), respectively, in the maize population ESALQ-PB1.Farias Neto and Miranda Filho (2001) reported genetic correlation of PH x EY and EH x EY in the order of 0.19 and 0.29 in subpopulations selected to increase tassel size and of 0.23 and 0.39 in subpopulations selected to increase ear height.The corresponding estimates within subpopulations selected independently to decrease tassel size and ear height were 0.087, 0.176, 0.342 and 0.503, respectively.Therefore, the trend of positive correlation as mentioned above indicates that care must be taken in selection to reduce PH or EH, in the sense to avoid undesirable change in yield potential.An important parameter that deserves attention in maize breeding is the ear placement (EP = EH/PH), which is essentially a measure of the relative position of the ear in the stalk.It is well known that maize germplasms of tropical origin in general exhibit EP > 0.50, and values near or above 0.60 are not uncommon.Andrade and Miranda Filho (2008) reported EP = 0.53, varying from 0.47 to 0.59 in half-sib families of the population ESALQ-PB1.
considered intermediate in magnitude and within the acceptable pattern of modern varieties or hybrids.Plant lodging and stalk breakage, usually evaluated together as one trait, is very inconsistent because it depends on the environment effects such as strong wings and incidence of stalk rot caused by fungi like Diploidia and Fusarium.This trait was practically inexistent in the experiments, with average occurrence of less than one affected plant per plot.For that reason, the trait was not included in the analysis of variance and estimation of parameters.The observed final stand (ST -number of plants per plot) and NE (number of ears per plot) were both in the sequence CRE-01 < CRE-03 < CRE-02, within the range between 11 and 17.The traits were included in the analysis of variance but not in the estimation of parameters, which would not be of interest for breeding purposes.However, the ratio PR = NE /

( 1985 )
in Suwan-DMR (-0.441);Geraldi et al. (1985) in Dent Composite (-0.501),Flint Composite(-0.642)and Centralmex (-0.709), averaging 0.650.However, the correlation between traits is not fully consistent because it is affected by the genetic structure of the base population, so that the estimated parameter can change from one population to another.In fact,Andrade and Miranda Filho (2008) found a positive correlation (0.207) between GY and TN in the original population ESALQ-PB1.After six cycles of divergent selection for tassel size in the same population (ESALQ-PB1), the means for TN (average of two locations) increased to 39.0 in the positive selection (T + ) and decreased to 8.2 in the negative selection (T -); the genetic correlation between GY and TN was negative (-0.022) in T + and positive (0.456) in T -(Farias Neto & Miranda Filho, 2001).Other authors also found positive correlation between GY and TN(Farias Neto   Table 2 -Mean squares ƒ for Families (M 1 ) and Error (M 2 ) and coefficients of variation (CV) in the analysis of variance for fourteen traits in half-sib families from three semiexotic maize populations (CRE-01, CRE-02, CRE-03).
significant F-test (P < 0.01) was detected for all traits in the three populations, except EP (P < 0.05 in CRE-01 and non-significant in CRE-03), thus indicating the existence of wide variability at the population level.For the additive genetic variance ( MF, PH, EH, TL, and ED; 1 < 2 < 3 for EP;2 < 1= 3 for EL; and 3 < 2 < 1 for PR, TN, EY and GY (Table3).The relative values of 2 F σ (phenotypic variance among family means) were very similar, with EY exceptions to the ranges shown to 2 A σ .Particularly interesting is the relative values of variance for the yield traits, which were not very different among the three populations.For GY, the estimates of 2 A σ (582.28 g 2 plant -1 , 406.70 g 2 plant -1 , 238.11 g 2 plant -1 ) were in the same order as the estimates reported by Oliveira et al. (2015b) for the same populations in the first cycle.The coefficient of heritability (family mean basis) for the 11 traits in the three populations varied from 27.5% (EP in CRE-01) to 86.2% (TN in CRE-01) (Table shown by PH, EH, TL, EL, TW (tassel weight) and RN (kernel row number), ED and KR (kernels per row).In general, most of the traits in the three populations under study in the present project exhibited genetic variability within the pattern that has been observed by other authors.The genetic coefficient of variability is a relative measure of genetic variance and does not follow the same pattern for different traits.However, it is clearly seen that the highest values were for CRE-03 in most traits (MF, PH, EH, TL, EL, ED), while the less variable was CRE-03 for TN, EY and GY.

Finally, another important
information to complete the knowledge about the genetic structure of the semi exotic populations refers to their potential to have an expressive response to artificial selection.
expression of additive genetic variance for the three populations, thus assuring positive perspectives for the population improvement through recurrent selection toward the development of outstanding cultivars.The selection for grain weight in the three populations, under the intensity of 20%, that allows to maintain greater genetic variability, indicated considerable gains for continuity in the recurrent selection program.The selection with intensity of 5.56% and 5% indicates a considerable gain, allowing the populations to be used as sources for lineage extraction.

Table 4 -
Estimates of the coefficient of heritability for family means ( 2 f h ), genetic coefficient of variation (CVg) and index of variation (q) in three semiexotic maize populations (CRE-01, CRE-02, CRE-03).Symbology and unities of measurement; LB was excluded for its low expression in the 3experiment].

Table 5 -
Estimates of the differential of selection (ds) and expected gain from selection (Gs) under two selection intensities in three semiexotic maize populations (CRE-01, CRE-02, CRE-03).
t [Symbology and unities of measurement].G s within brackets: in percent of the population mean.a , b , c : Selection intensities of 5.56 %, 5% and 20%, respectively (Oliveira et al., 2015b)ely.In the first cycle of selection, under 10% selection intensity in CRE-01 and 11% in CRE-02 and CRE-03, the expected gains were in the range of 11.54% (CRE-01) to 8.87% (CRE-03) in EY and 11.81% (CRE-01) to 9.27% (CRE-03) in GY(Oliveira et al., 2015b).A comparison between the two cycles of selection in respect to the expected gain is not fully adequate, because they refer to different crop season, i.e., second crop (safrinha) in the first cycle and first crop (normal season) in the second cycle.Nevertheless, it was considered that the expected gains were in acceptable level for the recurrent selection program.In fact, an expected gain of about 10% would assure a yield level practically equivalent to the hybrid check in the three populations.More exactly, the expected yield for CRE-01, CRE-02 and CRE-03 would be 103.1%,98.3% and 104.1% in relation to the check yield.The changes of 4.50% in CRE-01 and 6.27% in CRE-02 under higher selection intensity; for lower selection