Comparison of testers in the selection of S 3 families obtained from the UENF-14 variety of popcorn

O uso de topcross vem se mostrando uma opcao interessante para a cultura do milho; entretanto, para o milho-pipoca, ha pouca informacao sobre a escolha do testador adequado. Nesse contexto, este estudo teve como objetivo analisar quatro testadores, sendo duas variedades de polinizacao aberta (BRS Angela e UENF-14), um hibrido topcross (IAC 125) e uma linhagem (P2), visando avaliar a capacidade combinatoria de 50 familias S3 de milho-pipoca, obtidas a partir da variedade UENF-14. Avaliou-se o rendimento de graos (RG) e capacidade de expansao (CE). Obtiveram-se as estimativas das capacidades geral e especifica de combinacao, bem como efetuou-se a discriminacao de cada testador por meio de indice de diferenciacao. Os testadores BRS Angela (para RG) e IAC 125 (para CE) foram os que se destacaram, quando combinados com as melhores familias S3derivadas da variedade UENF-14, para producao de hibridos de milho-pipoca para as Regioes Norte e Noroeste Fluminense.

Any breeding program focusing on successful hybrid combinations must concentrate its efforts to identify superior lines and their ability to transmit these desirable traits to the hybrids (Hallauer et al. 2010).In this context, diallel analysis is considered an important statistical tool for estimating parameters useful in selecting parents (Seifert et al. 2006;Barreto et al. 2012;Souza Neto et al. 2015).
However, a limiting factor of diallel analysis is the number of assessed parents, usually not exceeding ten, because of the effort to obtain the hybrids.In this case, the topcross technique, proposed by Davis (1927) and Jenkins and Brunson (1932), has proven to be a more appropriate option, aiming at overcoming the impossible assessment of progenitor lines in works with hybrids, involving a large number of lines.
The term "topcross" is used to denote crosses of lines with a tester.The tester can be a variety, a hybrid or even a line and receives this name for playing a role in evaluating the combining ability of the lines.Testers can be classified according to their genetic base (broad versus narrow), the degree of relatedness with the material evaluated (related versus unrelated) and their intrinsic genetic value (high pattern versus regular or inferior pattern) (Miranda Filho and Gorgulho 2001;Hallauer et al. 2010).
Despite the wide acceptance of the topcross method, the ideal tester selection process still remains a goal to be achieved by hybrid development programs, since issues of choice, type, number and efficiency of testers are perpetuated amidst theoretical and experimental studies (Ferreira et al. 2009).According to Hallauer et al. (2010), the choice of the tester should be on simplicity in use and generation of information that correctly classifies DOI: http://dx.doi.org/10.1590/1678-4499.215the relative potential of lines in cross, thus maximizing the genetic gain.According to Rodovalho et al. (2012), it is more important to use testers with broad genetic base (synthetic and open pollinated varieties) during the initial phase of breeding, since they test for general combining ability or additivity.
Compared to the common maize, there are few studies regarding the definition of testers for popcorn (Pinto et al. 2004;Viana et al. 2007;Scapim et al. 2008;Arnhold et al. 2009).An important detail that possibly hampers to choose the most suitable tester is the fact that the two main traits of the crop, popping expansion and grain yield, which should be evaluated together, have different mechanisms of inheritance (Pereira and Amaral Júnior 2001).Popping expansion is controlled by a small number of genes and there is predominance of additive gene action (Yongbin et al. 2012;Rodovalho et al. 2014).On the other hand, grain yield is genetically controlled by a large number of genes with predominance of non-additive gene effects.
Given these considerations, this study analyzed and compared four testers (BRS Angela, UENF-14, IAC 125 and P 2 ) in the assessment of 50 partially inbred families (S 3 ), obtained from the UENF-14 variety of popcorn.

MAteriAl And MetHOds
For this study, 50 S 3 families of popcorn were obtained from the variety UENF-14 of Universidade Estadual do Norte Fluminense Darcy Ribeiro (UENF) (Amaral Júnior et al. 2013).Four materials were evaluated as testers: BRS Angela (open pollinated variety), IAC 125 (topcross hybrid), P 2 Line (S 7 generation coming from the CMS-42 composite of the Embrapa Maize and Sorghum) and UENF-14 (open pollinated variety).Hybrids were evaluated in the Experimental Station of PESAGRO-RIO, Itaocara, State of Rio de Janeiro, Northwestern Fluminense region (21°39'S latitude and 42°04'W longitude), in the growing season of 2013/2014.
Five trials were implemented in blocks with two replications, four for assessing the hybrids derived from crosses of families with each tester and the fifth for testing the performance of S 3 families.Each experimental unit consisted of a row of 3 m, with spacing of 0.90 m between rows and five plants per meter.In the trials, the cultural practices, such as fertilization at planting and topdressing, irrigation, pest and weed control, among others, were made as required by the popcorn crop.
The traits evaluated were grain yield (GY) and popping expansion (PE).GY was assessed by measurement of the grain mass produced in each plot after discarding the cob and is expressed in kg•ha −1 .In order to adapt the measurement of the yield, it was carried out the stand correction method, by analysis of covariance between the number of plants per plot and GY, according to the covariance of ideal stand methodology, proposed by Vencovsky and Barriga (1992).PE was determined by means of a seed sample, taken from the basal center of the ears (Granate et al. 2002) in each plot.All samples were sent to cold dry chamber to reach the equilibrium moisture content of 12 to 13%.PE was determined in laboratory with the use of a microwave Panasonic NN-S65B, placing a sample of 30 g seed in a plastic container obtained in the USA, at the power of 1,000 W for 2 minutes and 30 seconds, in two replications per treatment.PE was expressed by the ratio between the volume of the popped popcorn, measured in a 2,000-mL measuring cylinder, and the initial weight of grains (30 g), being the final unit expressed in mL•g −1 .
Statistical analysis of data followed the model of randomized blocks.The mean squares of treatment and the residue were used for the F test.The analysis of general and specific combining ability, based only on topcross crosses, was performed according to the scheme of analysis of variance in partial diallel at the level of average of treatments using the model proposed by Griffing (1956), adapted by Geraldi and Miranda Filho (1988).The efficiency of testers was evaluated by statistical indices of performance (P) and differentiation (D) proposed by Fasoulas (1983).The P index gives the percentage, in relation to the number of means, that a particular cultivar statistically outperforms the others based on the minimum significant difference (MSD), determined by test of means.The T index gives the percentage of pairwise comparisons between cultivars that showed significant differences.Analyses were run using the software GENES (Cruz 2013).

resUlts And discUssiOn
The mean squares of treatments (topcrosses) were significant at 5% probability for the four testers and families per se, indicating genetic variability for GY and PE (Table 1).The coefficients of variation (CVe) for GY and PE were within acceptable limits for agricultural experimentation, except for the experiment assessing the lines per se that showed CVe values of 24.62 and 26.00, respectively, considered high (Scapim et al. 1995;Arnhold and Milani 2011;Fritsche Neto et al. 2012).
In the case of estimates of genetic parameters, the largest genetic variances were observed for the crossing BRS Angela × S 3 (22,830.61)and with the S 3 families per se (0.93) for GY and PE, respectively (Table 1).The greatest increase in variability in topcross hybrids from the tester BRS Angela, for GY, can be explained because it is a broad base and unrelated tester, or because it presents many heterozygous loci.It is noteworthy that the high variability observed in S 3 families per se, for PE, is possibly related to the intrapopulation genetic variance, as they are individuals from the same population (UENF-14) (Amaral Júnior et al. 2013).
Considering the estimates of heritability, in general, their magnitudes remained very close independently of the tester.The exception applies to hybrids from the tester BRS Angela, specifically for PE, noting that this tester was not able to express the genetic variability among S 3 families (Table 1).
There was a significant effect of general and specific combining abilities, estimated by diallel analysis of Griffing (1956), adapted by Geraldi and Miranda Filho (1988).The general and specific combining abilities of S 3 families and testers have statistical significance, revealing differences for both traits between the values of the estimates of general (G i ) and specific (S ij ) combining abilities (Table 2).Regarding the general combining ability, the tester with the best estimate of G i for GY was the P 2 Line, while, for the PE, testers IAC 125 and BRS Angela achieved the best estimates, reflecting the greater allelic complementarity of these testers with S 3 families (Figure 1).Barreto et al. (2012) evaluated the combining ability of S 2 families of popcorn and also obtained a similar result for the tester IAC 125, indicating it as the tester with the best general combining ability (GCA) indices for PE.
S 3 families with higher estimates of GCA were 9, 33, 4, 20 and 16 for GY,and 18,23,42,11,and 25 for PE (Figure 2).Among the 50 S 3 families, only 9, 16 and 23 stood out for concomitantly presenting significantly positive estimates of GCA for GY and PE and should be evaluated more carefully in the breeding program conducted in the Northern and Northwestern Fluminense Regions by UENF.
As for the specific combining ability of S 3 families, for crosses with the tester BRS Angela, families 21, 4, 13, 50, 22 and 45 demonstrated the highest values of specific combining ability (SCA); for the tester IAC 125, the families 9, 20, 11, 35, 3 and 45 stood out.For tester P 2 , families 49, 43, 24, 41, 18 and 50 showed the highest SCA indices and, in relation to the tester UENF-14, families 19, 2, 4, 45, 48 and 47 were those with higher estimates of S ij (Table 3).The two hybrids showing better estimates of the effects of SCA were derived from crosses between tester P 2 and family 49 and between the tester IAC 125 and family 9.These hybrids, respectively, had S ij effects of 2,359 and 1,960, besides having values around 7,400 kg•ha −1 .These two materials are very interesting because they express high genetic non-additive effects in their genotypes, derived from allelic complementation of their parents, and will certainly comprise hybrids with **Significant 1% probability; MS = Mean squares; SV = Sources of variation; DF = Degrees of freedom; GY = Grain yield; PE = Popping expansion; GCA = General combining ability; SCA = Specific combining ability.

Figure 1.
Estimates of the effects of general combining ability (G i ) associated with testers for grain yield (kg•ha -1 ) and popping expansion (mL•g -1 ) according to Griffing (1956).high average GY, which may be useful in interpopulation breeding programs in popcorn.
According  (1983), for GY, were obtained by tester UENF-14 (D = 28.98) and BRS Angela (D = 28.24),indicating these testers as having greater efficiency in discriminating S 3 families (Table 3).Rodovalho et al. (2012) examined different testers in the discrimination of 64 S 2 families of popcorn, for GY, and also indicated the tester BRS Angela using this differentiation index.Considering the trait PE, the highest differentiation indices, 25.22 and 19.92%, were obtained, respectively, for the testers UENF-14 and IAC 125, indicating that these parents discriminated better the trait PE of S 3 families.
Among the testers used, the P 2 Line exhibited only one hybrid of high GY by combining with one of the seven S 3 families of greater values of GCA, the family 2. The BRS Angela was the tester that concomitantly classified most high-yield hybrids with S 3 families of high GCA.Of the seven S 3 families with the highest GCA indices, five that comprised the topcross hybrids of higher yields, Figure 2. Estimates of the effects of general combining ability (G i ) associated with S 3 families for grain yield (kg•ha -1 ) and popping expansion (mL•g -1 ) according to Griffing (1956).

ˆˆˆˆ
among the 50 produced with this tester, in descending order of yield, were the families 4, 9, 16, 23 and 33.Of the nine topcross hybrids from the IAC 125 that showed better average of GY, three were formed with families (9, 20 and 33) which obtained high indices of GCA.In the tester UENF-14, three topcross hybrids with high average GY stood out, combined with families of high GCA effect, namely, families 2, 4 and 20 (Table 5).Thus, it can be said that the tester BRS Angela discriminated more consistently S 3 families, according to their genetic merit for GY.Among the nine hybrids showing better values of PE, obtained with the tester BRS Angela, only one (sixth most expansive) was formed with a family of high index of GCA; the best family for such effect was the family 18.For the tester IAC 125, three hybrids (fourth, seventh and ninth) of greater PE were formed by families that presented high GCA indices, the families 23, 25 and 9.The tester P 2 , as well as the tester BRS Angela, classified only one hybrid with high PE (sixth best hybrid) with the family 18, which had the best GCA index for the trait.Among the hybrids from the UENF-14 tester with the highest values of PE, three were formed by families with significantly high GCA effects, the families 11, 25 and 47 (Table 5).The testers IAC 125 and UENF-14 stood out for having three topcross hybrids, classified with high PE indices within each group (tester), obtained from crossing with S 3 families that held high indices of GCA.However, the hybrids from the tester IAC 125 provided a higher general average for the trait PE in relation to hybrids produced by crossings with the tester UENF-14.Therefore, the best tester for PE is the IAC 125.were the most appropriate, when combined with S 3 families derived from UENF-14 for the production of popcorn hybrids for the Northern and Northwestern Fluminense Regions.
Effect of general combining ability associated with the progenitor i.
g i = Effect of general combining ability associated with the progenitor i.

table 2 .
Partial diallel analysis of means of treatments for grain yield and popping expansion in the study of general and specific combining abilities.

table 3 .
(Fasoulas 1983)s of specific combining ability (S ij ) and discrimination ability of testers, according to the D index and P performance index(Fasoulas 1983), for grain yield (kg•ha -1 ) of topcross hybrids, based on the Student's t-test (0.05) for means comparisons.

table 4 .
(Fasoulas 1983)s of specific combining ability (S ij ) and discrimination ability of testers, according to the D index and P performance index(Fasoulas 1983), for popping expansion (mL•g -1 ) of topcross hybrids, based on the Student's t-test (0.05) for means comparisons.
Classification order of S 3 families with respect to general combining ability and nine topcross hybrids with the highest indices with testers, for grain yield and popping expansion cOnclUsiOnThe testers BRS Angela (for GY) and IAC 125 (for PE)