Heterosis and genetic distance in intervarietal corn hybrids

Abstract: The objective of this work was to evaluate the performance of intervarietal corn (Zea mays) hybrids with topcross crosses between landrace populations, and to confirm whether genetic dissimilarity between populations is correlated with the heterosis of the intervarietal hybrids in the field. Nine topcross hybrids were evaluated with their tester 'BRS Planalto', and the following landrace populations were used as parents: Argentino Branco, Dente de Ouro, Amarelão, Criolão, Caiano Rajado, Branco Oito Carreiras, Branco Roxo Índio, Cateto Branco, and Argentino Amarelo. The tester 'BRS Planalto' and the topcross hybrids Branco Oito Carreiras x 'BRS Planalto' and Criolão x 'BRS Planalto' showed higher per se potential for grain yield. The topcross hybrid Branco Oito Carreiras x 'BRS Planalto' showed a better performance for number of grains per row, grain weight, and ear diameter, whereas Criolão x 'BRS Planalto' displayed a better performance for the number of grains per row and ear length. Greater estimates of genetic distance did not necessarily result in greater heterosis values and were exclusively correlated with grain ear length. Therefore, it is not possible to predict the effects of high heterosis on grain yield, based on the genetic distance between the populations involved in the crosses.


Introduction
Corn breeding programs would greatly benefit from a more comprehensive knowledge of Brazilian corn germplasm, since there is evidence of a great competitiveness in the market for the development of new cultivars (Araújo & Nass, 2002). Competition assays and germplasm characterization trials represent the majority of studies on corn landrace in Brazil (Paixão et al., 2008;Carpentieri-Pípolo et al., 2010;Coimbra et al., 2010) show great landrace potential, showing similar, or even greater, performance of landraces, in comparison to commercial and hybrid varieties, especially at low-technology cultivation levels.
Heterosis effects on grain yield are commonly sought (Allard, 1971) because they have a great influence on the success of corn hybrids. According to quantitative genetics, the heterosis effects are related to allele frequency in the parents, and the genetic divergence between such parents positively affects them (Falconer & Mackay, 1996).
Multivariate analysis, such as genetic dissimilarity, are often used to classify genotypes into heterotic groups (Baretta et al., 2016). According to Sudré et al. (2005), studies on genetic dissimilarity are of great importance for breeding programs involving hybridizations, as they provide parameters for the identification of parents that would supply the progeny with greater heterotic effects. The relationship between genetic dissimilarity of landrace populations and heterosis estimates is still poorly studied. The absence of this information has made it difficult to form heterotic groups suitable for the formation of intervarietal hybrids or for the selection of inbred lines in breeding programs (Baretta et al., 2017;Carvalho et al., 2017).
The objective of this work was to evaluate the performance of some intervarietal hybrids, from topcross crosses between corn landrace populations and a broad genetic-based tester, as well as to verify whether the genetic dissimilarity between the populations is correlated with heterosis of the intervarietal hybrids in the field.

Materials and Methods
The experiment was carried out in the Centro Agropecuário da Palma (31º45'S, 52º29'W, at 13 m altitude), in an experimental area belonging to the Universidade Federal de Pelotas, in the municipality of Capão do Leão, in the state of Rio Grande do Sul, Brazil, on an Argissolo Vermelho-Amarelo distrófico, according to Santos et al. (2006), i.e., a Paleudalf soil.
In A randomized complete block design, with three replicates, was used. The experimental plots consisted of two 5.0 m rows, with 42 plants per plot (with 0.70 m spacing between rows), corresponding to the density of 60,000 plants ha -1 . Cultural traits were performed in accordance with the recommendations for the culture in the region.
The following traits were evaluated: leaf angle (LA), obtained with the first leaf below the first ear, in three plants; number of tassel branches (NT); height of ear insertion (EH); plant height (PH) from soil surface up to the last leaf node of the plant; ear diameter (ED), measured in the central part of three ears; number of grains per ear (NG); number of grain per row (NGR); grain mass per ear (GM); ear mass (EM); ear length (EL); prolificacy (PROL), which refers to the relation between total of ears per total of plants in the plot; 100-grain mass (HGM); and grain yield (GY) per plot, corrected to 13% moisture.
Data were subjected to the analysis of variance for the genitor populations and their topcross hybrids. Heterosis estimates were performed based on the model (Falconer & Mackay 1996).
The genetic dissimilarity between the genitor populations was estimated using the generalized distance of Mahalanobis (D 2 ), with data from the means of the genotypes and the residual covariance matrix, according to Cruz et al. (2012). The UPGMA grouping method was used to form the dendrogram. In order to verify the fit between the dissimilarity matrix and the dendrogram, the cophenetic correlation coefficient (r) (Sokal & Rohlf, 1962)  with the NTSYS program (Rohlf, 2000). The Pearson's correlation coefficient between genetic dissimilarity and heterosis estimates was evaluated for the different characters, using the SAS version 9.3 software (SAS Institute Inc., Cary, NC, USA).

Results and Discussion
The individual mean squares for populations and topcross hybrids differed significantly for most of the characters, except ear height, for populations, and grains number, ear diameter, and ear length for top cross hybrids. The landrace populations significantly interacted with their topcross hybrids as to NG, ED, NGR, GM, EM, EL, and GY (Table 1). Machado et al. (2008) mention that crosses between landrace varieties and improved cultivars are an important strategy for the conservation of genetic variability in corn, as well as for obtaining new varieties.
The variation amplitude between the highest and lowest values verified in all evaluated characters evidenced a strong dispersion in the data, which is indicative of the existence of a great genetic variability between the populations and in their topcross hybrids. The variation verified in the landrace populations allows of the possibility of identification of parents that, when combined, would show a greater or lower hybrid vigor (Table 1). Greater amplitudes were observed in the topcross hybrids than in the genitor populations, for most of the characters: NG, PH, ED, NGR, GM, EM, and GY. This result indicates the presence of complementary genes distributed among the parents, which are potentially able to maximize responses in the F 1 generation, as reported by Bertan et al. (2009). Genetic diversity studies provides an opportunity for the identification of varieties with high-combining capacity, which can assist breeding programs in benefiting from heterosis (Machado et al., 2008).
The landraces Amarelão and Branco Oito Carreiras and the topcross Branco Roxo Indio x 'BRS Planalto' stood out for HGM standard deviations that were higher than the average for this trait (  -Neto (1999) mention that corn breeding programs should consider these traits, as they are positively related to GM. According Nemati et al. (2009), the increase of ear diameter causes an increase in the number of rows per ear, consequently increasing the NG with a positive reflect on grain yield. The Caiano Rajado population and Argentino Amarelo x 'BRS Planalto' topcross had the highest means for PROL. Prolific plants generally show a greater tolerance to adverse conditions, due to their ability to develop at least one ear, even under stressing conditions, and more than one when the environmental conditions are favorable. PROL is also related to plant capacity in compensating for grain yield loss when plant stands are below the ideal one (Sangoi et al., 2010).
It should be noted that the Cateto Branco population and the topcrosses Amarelão x 'BRS Planalto' and Cateto Branco x 'BRS Planalto' show lower averages for EH and PH. These traits are desirable in maize genotypes because they facilitate harvest and make the plants more resistant to lodging.
As to grain yield, the 'BRS Planalto' tester and the topcrosses Branco Oito Carreiras x 'BRS Planalto' and Criolão x 'BRS Planalto' achieved promising results, showing a favorable gene pool for exploitation, to generate populations with high potential yields.
The estimates of heterosis (H 2 ) were very variable in the present study (Table 3). Crosses with high heterosis show a high potential for use in breeding programs (Bertan et al., 2009). Caiano Rajado x 'BRS Planalto' (H 2 = 55.3%), Cateto Branco x 'BRS Planalto' (H 2 = 36.9%), Amarelão x 'BRS Planalto' (H 2 = 23.6%), Criolão x 'BRS Planalto' (H 2 = 19.3%), and Branco Oito Carreiras x 'BRS Planalto' (H 2 = 6.47%) had the highest estimates of heterosis for grain yield, and are recommended for promoting productivity in studies with landrace maize populations. Moreover, they have distinct genetic backgrounds, which probably resulted in a high frequency of dominant alleles in the hybrid progenies. Paterniani et al. (2008) also showed a high range of heterosis values when studying a set of 36   (1) LA, leaf angle; HGM, mass of one hundred grains; NTB, number of tassel branches; EH, ear height insertion (cm); PH, plant height (cm); ED, ear diameter (mm); NG, number of grains per ear; NGR, number of grains per ear row; GM, grain mass per ear (g); EM, ear mass (g); PROL, prolificacy; EL, ear length (cm); and GY, grain yield per plot (kg). (2) Results compromised by the loss of data on a parental during the trial.
Pesq. agropec. bras., Brasília, v.54, e00265, 2019 DOI: 10.1590/S1678-3921.pab2019.v54.00265 hybrids, confirming the high-genetic variability among the hybrids and the great diversity among parental lines. Hallauer et al. (2010) presented a compilation of heterosis data for grain yield in corn, with of 20.63% heterosis mean values, which is higher than the average estimates observed in the present study. However,  report that heterosis was evenly distributed among crosses of landrace populations, and did not observe a differentiated heterotic contribution of each variety, which is in disagreement with our results. These findings corroborate the fact that the heterosis estimates depend on and are directly related to the studied populations. According to Moreira et al. (2009), the generalized distance of Mahalanobis (D 2 ) can be used as a powerful tool to estimate the genetic dissimilarity, in order to assist the selection of parents for more promising combinations. For Barili et al. (2011), genotype classifications in heterotic groups can be done with information on genetic dissimilarity. We observed that in the formation of two large groups, according to their genotype mean dissimilarity (Figure 1), the first one was formed by the 'BRS Planalto' tester and the population Branco Oito Carreiras; and the second group, by the populations Argentino Branco, Argentino Amarelo, Caiano Rajado, Dente de Ouro, Amarelão, Criolão, and Cateto Branco.
Heterosis levels were significant and positively correlated only with the genetic distance determined for the ear length trait (Figure 3). However, no significant correlations were observed for the remaining characters, in the investigated populations (Figures 2 and 3). According to Paiva et al. (2002), the combination of genetically distant parents does not always result in expressive heterosis. Paterniani et al. (2008) reported that genetic distances were not correlated with heterosis, and that this trait does not allow inferences on promising crosses in breeding programs. Moreover, these authors reported that higher estimates of genetic dissimilarity did not entail higher values of specific combining ability, and do not correlate with hybrid productivity. Guimarães et al. (2007) investigated the correlation between heterosis estimates and genetic distance between hybrids, determined with molecular markers, and they also observed that the genetic distance does not allow of inferences on the crosses, in corn breeding programs.
The presence of significant associations between heterosis and genetic distance could immensely help the formation of breeding programs, since it allows of previous decision on parents to be used in the crosses (Dias & Resende, 2001). However, these associations are intrinsic to the gene pool under   investigation. It is important to mention, however, that the absence of significant association between heterosis and genetic distance, as observed in the present study, especially for grain yield, may also be related to nonlinearity between the analyzed parameters (Simon et al., 2012).

Figure 2.
Pearson's correlation coefficient (r) between heterosis, in topcross hybrids, and the genetic distance of the landrace parents. ns Nonsignificant. *Significant, at 5% probability. Landrace populations and topcross hybrids with greater per se potential for grain yield were: Branco Oito Carreiras x 'BRS Planalto' and Criolão x 'BRS Planalto'. The first topcross also displays a great performance for grain number per row, grain weight per ear, and ear diameter; and the second one, for grain number per row, and ear length.