Determining the inheritance of agronomic traits in chili pepper

The Capsicum genus is widely studied in many countries but most of research focuses on C. annuum. Although its importance as potential genetic resource for breeding to disease resistance and fruit quality, studies on genetics and breeding in C. baccatum var. pendulum are still scarce and it still remains as an underexploited crop. This paper aimed to study the inheritance of agronomic characteristics, to estimate the genetic parameters for seven agronomic traits in C. baccatum var. pendulum, using generation mean analysis and to verify the existence of major genes and polygenes with additive and non-additive effects, using the method of maximum likelihood with the aid of Monogen software. The agronomic traits evaluated were: number of fruits per plant (NFP); average fruit mass per plant (AFM); fruit length (FRL); fruit diameter (FRD); fruit pulp thickness (FPT); soluble solids content (SSC) and fruit dry mass (FDM). There was a significant difference among generations (P1, P2, F1, F2, BC1 and BC2) for all traits evaluated. Narrow sense heritability estimates ranged from 45% for NFP to 75% for FRD. By estimating the average degree of dominance (ADD), we found that partial dominance is the gene action involved in the genetic control for FRL and FRD, while additive effect was observed for NFP, FPT and SSC, complete dominance for FDM and overdominance for AFM. The additive-dominant model was sufficient to explain the results obtained for all evaluated traits. Maximum likelihood method identified the existence of polygenes with additive and dominance effects in SSC, FPT and FDM. Major gene effect was identified for FRL and NFP characteristics. The results indicated the complexity of the inheritance of these traits with the presence of major gene involved.

The domestication of C. baccatum is relatively recent compared with many other cultivated species, occurring

Palavras-chave:
In Brazil, the C. baccatum var.pendulum is widely produced by small farmers, and red chili pepper (or "deer horn" chili pepper) and cambuci (or friar's hat chili pepper) are the most cultivated chili peppers for fresh consumption and for making sauces, potted chili peppers and dehydrated flakes (Moreira et al., 2006;Rêgo et al., 2009;Rodrigues et al., 2012).However, there is a lack of research on genetics and breeding, and consequently on the development of cultivars of this species.Only 15 cultivars were registered in the National Registry of Cultivars (Registro Nacional de Cultivares -RNC) of the Ministry of Agriculture, Livestock and Food Supply of Brazil (MAPA, 2013), with the majority belonging to private seed companies that lack of Capsicum breeding programs in the country (Rodrigues et al., 2012).
Characteristics such as productivity, resistance and/or tolerance to biotic and abiotic stresses, plant architecture, earliness, ripe fruit persistence, as well as characteristics related to fruit quality are the targets of Capsicum breeding programs (Rêgo et al., 2011;Rodrigues et al., 2012).Obtaining adequate information on the nature and magnitude of the gene effects of these characteristics is of major importance for the identification of the most suitable method to be applied in the selection and prediction of responses of segregating generations for the development of cultivars of C. baccatum var.pendulum.
The potential of a base population and the selection efficiency for the development of cultivars can be investigated by means of the relative importance of additive, dominance and epistatic effects for each trait (Hallauer et al., 2010;Gonçalves et al., 2011).The existence of differences between the additive genetic values of individuals in a population is a necessary condition for intrapopulation improvement.However, the feasibility of a breeding program aimed at developing hybrids depends on the existence of dominance effects, which correspond to the interaction between allelic genes.If the goal of the breeding program is to develop superior inbred lines, the magnitude and sign of additive effects and additive x additive effects need to be evaluated.When the goal is to develop a hybrid, it is necessary to perform analysis of contributions of genetic effects of dominance, dominance x dominance, additive x dominance, selecting for heterosis in the desired direction, with greater heterosis expected when such effects are predominantly directional (Mather & Jinks, 1982;Marame et al., 2009a).
Information about the genetic control of agronomic traits in C. baccatum var.pendulum is still scarce in the literature (Rêgo et al., 2009;Gonçalves et al., 2011;Rodrigues et al., 2012) when compared with C. annuum.The knowledge of that information is important for the decision making about the breeding methods to be used or about the size of population to be conducted during the implementation of the improvement program.The aim of this paper was to study the inheritance of agronomic traits in C. baccatum var.pendulum.

MATERIAL AND METHODS
To estimate the nature and magnitude of the genetic variability available and gene effects for agronomic traits of C. baccatum var.pendulum, the segregating generations (P 1 , P 2 , F 1 , F 2 , BC 1 and BC 2 ) of the crosses between parents UENF 1616 x UENF 1732 were evaluated.These parents were chosen based on preliminary results of the C. baccatum var.pendulum breeding program of the Plant Breeding Laboratory of the Universidade Estadual do Norte Fluminense Darcy Ribeiro (UENF), with the objective of obtaining resistant cultivars to Pepper yellow mosaic virus (PepYMV).The parent UENF 1616 comes from the genebank of collection held in Campos dos Goytacazes, Rio de Janeiro State, Brazil, and it is described as susceptible to PepYMV and has yellow fruits at the immature stage and orange at maturity.Its fruits are elongated with length around of cm and diameter of 2.7 cm.The parent UENF 1732 was also collected in a local market in Campos dos Goytacazes, and it is characterized as resistant to PepYMV (Bento et al., 2009), showing fruits of orange color in the immature stage and red at maturity, with campanulate shape, mean length of 4,4 cm and diameter of 4,5 cm.
The field experiment was carried out in Unidade de Apoio à Pesquisa located in the UENF, in Campos dos Goytacazes, from August 2010 to April 2011.The study adopted a randomized block experimental design with three replications and spacing of 1.10 m between rows and 0.60 m between plants.In each block, seven plants of each parent (P 1 and P 2 ), 12 plants of the F 1 generation, 83 plants of F 2 generation and 28 plants of each backcrossing (BC and BC 2 ), totaling 495 plants, were cultivated.Regarding the F 2 generation, plants were arranged in blocks, but replications were not considered, since it was a segregating generation.
During the experiment, the cultural treatments usually recommended for the culture of chili pepper (Filgueira, 2012) such as weeding, staking, fertilization and irrigation by sprinkler system were performed.Seven harvests were performed and the following agronomic traits were evaluated: number of fruits per plant (NFP) (sum of the number of fruits obtained in the seven harvests); average fruit mass per plant (AFM) (average mass in grams, using the ratio of the total mass of fruits per plant and the number of fruits per plant); fruit dry mass (FDM) (average mass in grams of five dried fruits per plant, using forced circulation oven at 65°C and 72 h); fruit length (FRL) (measured in mm, considering ten ripe fruits per plant); fruit diameter (FRD) (measured in mm, considering ten ripe fruits per plant); fruits pulp thickness (FPT) (measured in mm, considering ten ripe fruits per plant) and total soluble solids content (SSC) (measured in ten ripe fruits per plant, using a digital refractometer).
Data were subjected to analysis of variance of the generations and, subsequently, the following estimates and parameters were calculated, according to Cruz et al. (2004).
Phenotypic variance (σ f 2 ): Va r i a n c e a t t r i b u t e d t o t h e environment in F 2 population (σ2 we ): Genotypic variance in the F 2 population (σ2 g ): Additive variance (σ 2 a ): The analysis of parental and segregating generations' means was performed by complete and additivedominant models.In the complete model (m, a, d, aa, ad, dd), the parameters were estimated by the method of ordinary least squares.In the additive-dominant model (m, a, d), the parameters were estimated by the weighted least squares method proposed by Mather & Jinks (1982).The adequacy of complete and additive-dominant model was evaluated using the coefficient of determination (R 2 ), which expresses the degree of similarities of the estimates between the estimated values and the observed ones.The analysis of variance and the analyses of the means of generations were performed by using the GENES software (Cruz, 2013).
In order to identify the existence of major genes and polygenes with additive effects and non-additive for all traits we used the maximum likelihood method by the statistical program Monogen v. 0.1 (Silva, 2003;Gonçalves et al., 2007).

RESULTS AND DISCUSSION
There was a significant effect of generation by the F test for all traits, indicating the existence of variability among the generations evaluated.The coefficients of experimental variation ranged from 13.65% to 45.72%, values which are considered from medium to very high for chili pepper (Silva et al., 2011); however, they are considered satisfactory since variance was highly significant for most of the traits (FRL, FRD, SSC and FDM) and significant for the others traits (NFP, AFM and FPT).The coefficients considered high and very high for the traits under study can be attributed to the great variability existing in the fruits, resulting from the extensive segregation observed in the F 2 generation.
From the phenotypic variances of generations, genetic parameters were estimated including the broad sense heritability (h b 2 ), the narrow sense heritability (h n 2 ), the average degree of dominance (ADD) and the minimum number of genes (σ) (Table 1).For NFP, the h b 2 and h n 2 were 45% with predominant additive gene effects, because the dominance variance was negative, being, in this case, considered equal to zero.The ADD indicated additive gene action and the number of genes controlling this trait was 16, featuring a polygenic inheritance.Rêgo For AFM, it can be observed that the dominance variance (2.68) was greater than the additive variance (0.42), in accordance with the values obtained for narrow sense heritability (5%).In the broad sense, h b 2 was 35% for this trait.Marame et al. (2009b) studied the heritability and heterosis in C. annuum var.annuum genotypes and observed low values of narrow sense heritability (14%) and high values in the broad sense heritability (72.6%) for yield per plant and they concluded that the components of dominance were important, while the additive components contributed very little to achieve high yields.Considering this, hybrids exploitation could be recommended for commercial purposes.Regarding ADD, there was prevalence of overdominance for this characteristic.This result is discordant to those observed by Juhász et al. (2009), who also estimated the gene effects involved in the inheritance of agronomic traits in C. annuum and found only additive effects for this trait.These authors determined that AFM was controlled by a large number of genes (15), and eventually strongly influenced by environmental conditions.Silva et al. (2013) estimated the genetic parameters for AFM in C. annuum and observed absence of dominance for this trait.These observations demonstrate the complexity of the AFM in Capsicum plants.
The FRL and FRD traits showed predominance of additive variance over the variance of dominance, with h 2 b of 75% and 79%, respectively, while h 2 n was 64% and 75%, respectively.For both traits, a partial dominance (0.59 and 0.31, respectively) was indicated and five genes were calculated for each of these traits.These results regarding the predominance of additive effects are also corroborated by those obtained for C. baccatum var.pendulum by Rêgo et al. (2009), Rodrigues et al. (2012) and Medeiros et al. (2014).Rêgo et al. (2011) studied the phenotypic diversity in fruits of C. baccatum var.pendulum and also observed high values for h 2 b , 99% and 98% for FRL and FRD, respectively.Sood & Kumar (2011) studied the genetic estimates of fruit traits in C. annuum and observed partial dominance for FRL and FRD (ADD = 0.62 and 0.84, respectively), results similar to those obtained in the current study.
Table 2. Coefficient of determination R 2 (%) for non-orthogonal decomposition of the sum of squares of adjusted parameters for seven agronomic traits in the complete model (m,a,d,aa,ad,dd), evaluated in plants of P 1 , P 2 , F 1 , F 2 , BC 1 and BC 2 generations, from cross between UENF 1616 x UENF 1732 Capsicum baccatum var.pendulum {coeficiente de determinação (R 2 %) pela decomposição não-ortogonal da soma de quadrados de parâmetros ajustados para sete caracteres agronômicos, no modelo completo (m,a,d,aa,ad,dd), avaliados em plantas das gerações P 1 , P 2 , F 1 , F 2 , BC 1 e BC 2 , a partir do cruzamento entre UENF1616 x UENF 1732 de Capsicum baccatum var.pendulum}.Campos dos Goytacazes, UENF, 2011.Santa Clara, and they also identified dominance variance equal to zero for the number of eggs per leaf and for number of large mines.According to the authors, the negative values of dominance variance may be related to the low accuracy of the estimates of environmental variance.Thus, h 2 b and h 2 n were equal (56%) and ADD indicated additive gene action.The minimum number of genes was seven in the control of this trait.With respect to the SSC, the reverse was found, i.e., the dominance variance was similar to the genotypic one, with h 2 b of 38%.By presenting a negative additive variance (considered equal to zero) the ADD would not be a good parameter to estimate this variable, because by its formula , its value would be 0.00, not consistent with the values observed in additive and dominance variances.The predominance of additive variance for FPT is in accordance with the study by Rêgo et al. (2009) and Rodrigues et al. (2012).For SSC, the predominance of dominance variance was different to that observed by Rodrigues et al. (2012), but it was consistent with that observed by Rêgo et al. (2009) andMedeiros et al. (2014).

Genetic parameters
Regarding the FDM, there was a genotypic, additive and dominance variance of 0.17, 0.11 and 0.06, respectively, and therefore, an h 2 b and h 2 n of 55% and 35%, respectively.This predominance of additive effects over the dominance ones is also consistent with the one found by Rêgo et al. (2009) and Rodrigues et al. (2012).A complete dominance gene action was observed in the control of this trait with a minimum number of genes equal to 11 (Table 1).
From the generation mean analysis, the complete model and the additivedominant model were tested.In the additive-dominant model, the mean showed greater magnitude for all studied traits, by the non-orthogonal decomposition of the sum of squares of parameters, explaining respectively 94. 01, 98.98, 85.07, 89.42, 99.51, 98.21 and 94.94% of the variability available to NFP, AFM, FRL, FRD, FPT, SSC and FDM.Juhász et al. (2009) studied the detection of epistatic gene action in the genetic control of agronomic traits in pepper and found that the mean was also the main parameter in the inheritance of the traits studied in the additivedominant model.
The additive-dominant model was sufficient to explain all the data in studies in which the correlation coefficients reached values of 0.96, 0.82, 0.99, 0.99, 0.95, 0.99 and 0.96 for NFP, AFM, FRL, FRD, FPT, SSC and FDM, respectively (Figure 1).Sood & Kumar (2011) also confirmed that the additive-dominant model is suitable for explaining the characteristics of fruit production in C. annuum.
For NFP and FDM traits, it was noted that plants of the F 1 generation had higher means than those of their parents (Figure 1) characterizing the presence of heterosis, although for NFP, there was the predominance of additive effect and, for FDM there was complete dominance.This probably occurred because the traits are complex due to the large number of genes that govern them (16 and 11, respectively), which had epistatic action of 12.99% for NFP and of 35% for FDM, resulting in a gene complementation for these traits.Another hypothesis could be the underestimation of the variance of dominance in relation to the additive one.According to Bernardo (2002), the method of least squares tends to reduce the dominance variance, underestimating these values.Juhász et al. (2009) identified epistatic action of 62.72% for NFP trait in C. annuum, corroborating the data obtained in this study.
Regarding the trait of fruit dry mass, the BCs obtained higher mean than the parents and the individuals of the F 2 generation.Regarding NFP, BC 2 produced more fruits than the parents.However, plants of the F 1 generation had fruits with lower average mass than the fruits produced by P 1 and BC 1 , indicating that the F 1 generation plants produced more fruits, but they were lighter.Similar results were found by Rêgo et al. (2011) who identified a negative correlation for the NFP and AFM traits in accessions of C. baccatum var.pendulum.
In the complete model, the gene effect attributed to epistatic effects was significant for AFM and FDM at the level of 1% probability by t test (data not shown).Both characteristics contributed respectively with 30.91 and 35.00% of the variability available for these traits (Table 2).Epistatic effects for AFM trait were also obtained by Juhász et al. (2009) in C. annuum.However, Gonçalves et al. (2011) studied the inheritance of traits related to fruit production in C. baccatum var.pendulum based on Hayman diallel analysis and they identified the absence of epistasis for FDM and AFM traits.Neverthless, these authors evaluated only crossings of five parents, this number being considered insufficient for the accurate estimation of epistatic effects by Hayman analysis.
By means of the likelihood function was possible to detect the models of inheritance to determine the presence of major genes.For the SSC, FPT and FDM characteristics occur polygenic control with additive and dominance effects.However, for characteristic NFP there is action of a major gene and polygenes with only additive effect.Moreover, for FRL characteristic, the presence of a major gene with additive and dominance effects was observed, in agreement with the hypothesis of these characteristics have complex inheritance.
This study allows concluding that the inheritance of most agronomic traits in chili pepper (C.baccatum var.pendulum), was polygenic with additive and dominance effects and also that for the FRL and NFP traits major genes play an important role regarding genetic control.Epistatic gene effects were more important for number of fruits per plant, average fruit mass and dry mass of fruit.this research.Also Dr. Eduardo Bearzoti and Dr. Wilson Roberto Maluf for their valuable contribution in the use of statistical software Monogen.
h a 2 ): Narrow sense heritability (h ƞ 2 ): Average degree of dominance (ADD) based on variances: Minimum number of genes involved in determining a character (ƞ): where: σ 2 P1 = variance of P 1 generation (UENF 1616); σ P 2 2 = variance of P 2 generation (UENF 1732); = variance of F 1 generation (UENF 1616 x UENF 1732); σ B 2 et al. (2009) and Rodrigues et al. (2012) evaluated diallel crosses in C. baccatum var.pendulum and found both additive and non-additive effects in genetic control for NFP, and dominance effects were predominant with different results from the ones obtained in the current study.Nevertheless, Medeiros et al. (2014) also evaluating the combining ability in diallel crosses of C. baccatum var.pendulum, found the predominance of additive effects, indicating the possibility of satisfactory gains with selection for this trait in segregating generations.