Genetic parameters and simultaneous selection using traits of ornamental interest in pepper plants

Pepper plants of the genus Capsicum are widely grown worldwide. The Capsicum annuum species shows ornamental potential; nevertheless, in the state of Piauí, there has been a lack of studies on its performance, related to quality traits needed to meet market demand. The aim of this study was to estimate the genetic parameters and associations between traits of ornamental interest in C. annuum in order to indicate, by simultaneous selection of traits, accessions to start an ornamental pepper breeding program in Piauí. We evaluated 16 accessions of C. annuum under plastic house conditions at Universidade Federal do Piauí. Number of days to flowering, number of days to fruit maturation, plant height, number of fruits per plant, fruit persistence, fruit length, fruit width, leaf length and fruit weight were evaluated. A significant difference (p≤0.01) between accessions for all traits was noticed. Coefficient of genetic variation / coefficient of experimental variation [ Cvg/Cve (%)] ratios were greater than 1 and the genotypic determination coefficient ( H² ) values were all greater than 89.95%. The estimates of the genotypic correlations were higher in relation to phenotypic correlations in most of the traits, however, according to path analysis, an effect of residual variable (0.59) was verified. According to the results, the accessions BAGC 98, 199, 207 and 236 can be used as parents to start an ornamental pepper breeding program in Piauí.

The genetic diversity of genus Capsicum, considering a small number of ornamental pepper varieties available in the market and registered at Ministry of Agriculture, Livestock and Supply (MAPA), has stimulated the development of pepper breeding programs. Considering the Northeast part of Brazil, the inferior quality of the marketed ornamental pepper plants turns out to be an aggravating factor, due to the distance from the largest potted plant producing center, located in São Paulo (Ibraflor, 2019). The main damages observed in these plants are related to the effect of ethylene released during transport, with an emphasis on the absence of flowers, fruits and leaves (Lima et al., 2017;Ribeiro et al., 2019). P i a u í h a s s e v e r a l p e c u l i a r edaphoclimatic conditions throughout its extensive territory (Andrade Júnior et al., 2004;Santos et al., 2011). This state is known for its grain production. The example of pepper and other ornamental species breeding programs developed in Paraiba State (Rêgo & Rêgo, 2016), shows favorable conditions to exploit this new market niche in Piaui State.
Universidade Federal do Piauí has been proposing actions to conduct breeding programs, aiming to strengthen and diversify the production of ornamental plants in the region. These actions are based on new bills, such as Instituição da Política Nacional de Incentivo à Cultura de Flores e de Plantas Ornamentais de Qualidade (Institution of the National Policy to Encourage the Crop of Quality Flowers and Ornamental Plants) newly approved at Comissão de Agricultura e Reforma Agrária (PL 6912-2017) (Agriculture and Agrarian Reform Commission). These actions are based on selecting pepper accessions with ornamental potential; these accessions have been collected (Costa et al., 2019) and stored at the Capsicum Germplasm Bank of the Institution (BAGC/UFPI) (Monteiro et al., 2010).
Searching for new genotypes is facilitated when the correlations between the traits used for selection are already known. Knowing these traits is quite relevant when it is related to late-expressed phenotype, such as number of fruits in pepper plants, or for medium-low heritability variables (Moreira et al., 2018). The correlations show the direction and degree of association between a pair of traits and whether the interrelation is due to genetic factors (pleiotropy or genetic link) or environmental factors (Ramalho et al., 2012). Associated with the trail analysis (Wright, 1921) can also indicate the direct and indirect effects of other variables on the correlation between a pair of traits under study.
Associations between traits are not always favorable. The negative correlation between two traits of economic importance can hinder the selection of genotypes of interest to the breeder. In these cases, geneticstatistical procedures can be applied, such as selection indexes. These indexes allow to obtain simultaneous gains even for traits which have adverse effects, enabling the selection of superior genotypes regardless of the existence or not of correlation (Rangel et al., 2011;Cruz et al., 2012Cruz et al., , 2014.
Obtained data will be the base for choosing more appropriate methods to conduct segregating populations, assisting in obtaining and recommending ornamental pepper cultivars for Piauí, as well as to provide information for crop researchers. In this context, the aim of this study was to estimate genetic parameters and associations between traits of ornamental interest in Capsicum annuum in order to suggest, using simultaneous trait selection, accessions to begin an Ornamental Pepper Breeding Program in Piauí.

MATERIAL AND METHODS
The experiment was carried out in a plastic house, installed in the Department of Phytotechnics, at Centro de Ciências Agrárias da Universidade F e d e r a l d o P i a u í , i n Te r e s i n a (05°05'05''S, 42°05'W, 73 m altitude) from February to November, 2017. The local climate is Aw type, tropical climate with dry winter, according to Köppen classification, 1349 mm rainfall distributed from November to May, 27.6 o C average annual temperature.
We evaluated 16 accessions of Capscium annuum from eight Brazilian states (Table 1). The experiment was conducted in a completely randomized design, with four replicates, considering one plant per plot. Under plastic house conditions (50% shading screen). The accessions were sown in expanded polystyrene trays of 128 cells, filled with the commercial substrate Plantmax®. The seedlings showing six to eight definitive leaves were transplanted to polystyrene pots, size 13, containing soil organic matter/humus substrate at a 2:5 ratio. The seedlings were irrigated daily, on mild sunny hours (8 am or 5 pm), and planting fertilization was done with 2 g of ammonium sulfate, 17 g of simple superphosphate and 2.5 g of potassium chloride, according to recommendation of Alcântara & Ribeiro (2008) and fertilization, every 30 days after planting, using Nutriverde NPK 13-13-15 fertilizer with micronutrients.
Morpho-agronomic characterization of the accessions was carried out according to the descriptors established by International Plant Genetic Resources Institute (IPGRI, 1995) for genus Capsicum and Ornamental Pepper Classification Criteria of the Veiling Holambra Cooperative, which are quality standards nationally established by Instituto Brasileiro de Floricultura (Ibraflor, 2020).
The authors evaluated number of days for flowering (NDF), which refers to number of days since sowing, up to 50% of plants showing at least one opened flower; number of days for maturation (NDM), measured since sowing, up to 50% of plants showing at least one ripe fruit; plant height (ADP), in cm, observed when 50% of plants showed the first fuits in the beginning of ripening; number of fruits per plant (NFP), according to the average of total number of fruits per accession; fruit persistence (PSF), considering the number of the days when the first ripe fruits appeared up to the last one; Fruit length (CDF), in mm, referring to average length of 10 ripe fruits, taken at random; fruit width (LDF), in mm, measured at the widest point of 10 ripe AKOS Acevedo et al. fruits, taken at random; leaf length (CDFol), in mm, which corresponds to the average length of 10 leaves, also taken at random; fruit weight (PDF), referring to the average weight of 10 ripe fruits, taken at random, in grams.
Obtained data were submitted to individual variance analysis. Effects of the accessions were considered fixed. Multicollinearity diagnose was based on the number of condition (NC), considering the ratio between the highest and lowest eigenvalue of the matrix. For NC<100, multicollinearity is considered weak and it does not represent a problem for the analysis; 100≤NC≤1000, multicollinearity is considered moderate to strong and NC≥1000 is considered severe (Cruz et al., 2012). When the authors observed moderate to severe multicollinearity, the variables were discarded; this fact contributed to the presence of multicollinearity.
G e n o t y p i c a n d p h e n o t y p i c correlations were estimated for all trait combinations after multicollinearity diagnosis. The significance of these combinations was estimated using t test with n -2 degrees of freedom, in which n corresponds to the number of evaluated genotypes.
Unfolding of correlations into direct and indirect effects was performed using path analysis (Wright, 1921), in which number of fruits per plant ((NFP) was the base variable. This variable was chosen since it is a later variable, and the other traits were considered explanatory variables.
The accessions were classified using genotype-ideotype distance selection indexes (Schwarzach, 1972 cited by Wricke & Weber, 1986) and the sum of ranking (Mulamba & Mock, 1978). An equal economic weight was attributed to each evaluated trait. Spearman's correlation coefficient (1904) was used for the correlation between the results of the indexes. All statistical analyses were performed using Genes software (Cruz, 2013).

RESULTS AND DISCUSSION
A significant difference (p≤0.01) among the accessions for all evaluated traits was verified, showing genetic variability among them. Variability is important for establishing plant breeding programs. An efficient selection of superior genotypes depends on genetic and environmental parameters for traits of interest, though (Blind et al., 2018).
Coefficients of experimental variation ranged from 5.09%, for number of days for fruit maturation (NDM) to 44.87%, for number of fruits per plant (NFP). This coefficient, relatively high, may be associated with the germplasm used in this study, since this genetic material is still in process of pre-breeding, presenting unevenness in the number of fruits, even within the same accession.
All evaluated traits showed higher values of genetic variance when compared with values for environmental variance, showing predominance of genetic effects over environmental effects. The values of coefficient of genetic variation and experimental coefficient of variation ratios [Cvg/ Cve(%)] ranged from 1.59% for NDM to 5.79% for LDF. Values higher than or equal to the unit showed the existence of great genetic variability to be used in breeding (Cruz et al., 2014).
The evaluated traits indicated high values for coefficient to genotypic determination coefficient, which ranged from 89.95% for PSF to 99.26% for LDF. Thus, the phenotypic values of these treatments have high correlation with genotypic values, which means, the studied germplasm showed heritable phenotypic variability. Moreira et al. (2015), working with inbred lines of C. annuum, found similar values for coefficient of genotypic determination (H²) ranging above 82.29 for fruitrelated traits.
Considering the nine evaluated v a r i a b l e s , t h e d i a g n o s i s o f multicollinearity showed NC = 432.80, indicating a moderate to strong collinearity. In order to deal with this strong multicollinearity, first, the authors used the methodology proposed by Carvalho et al. (1999), called Crest Path Analysis, where a constant (k) was added to the diagonal elements of the X'X matrix. However, this methodology was not efficient to reduce the multicollinearity. Hence, the authors excluded the variables. To choose the variables to be discarded, we observed the ones which presented high correlation among themselves. So, we excluded ADP (plant height) and The authors observed positive and significant genotypic correlation (0.5037) between NDF x NDM and negative correlation for NDF x PSF (-0.5263) ( Table 2). These correlations are important in ornamental peppers, as the shorter the period of days for flowering, for maturation and greater persistence of fruits in plants, the more interesting and attractive it will be for the consumer.
We verified negative correlations for NFP and CDFol. The values observed for phenotypic and genotypic correlations were -0.4163 and -0.5039, respectively (Table 2). Thereby, in accessions with shorter leaves, the amount of fruit will be greater. This correlation can be used to shorten the time of breeding programs which aim to work with number of fruits per plant, since the indirect selection can be carried out in vegetative phase.
Using the path analysis, NFP as base variable, the authors observed a high residual effect (0.59) ( Table 3), showing that the set of variables used in this study does not totally explain the number of fruits in ornamental pepper plants, since the values of the residual exceeded, except CDFol. Thus, the range observed in number of fruits occurred probably due to other traits which were not mentioned in this study, being able to be evaluated in further studies, though.
Low values of coefficient of determination (0.37) and high values of residual variable (0.79) were also obtained by Cabral et al. (2011), using path analysis for number of seeds per plant in beans (Phaseolus vulgaris). Salla et al. (2015) also obtained low values of coefficient of determination (0.12) and high values for residual effect (0.93) studying jabuticaba fruit traits (Plinia cauliflora). Blind et al. (2018) verified that secondary traits, such as average length and fruit circumference and maximum fiber thickness were not effective in determining the variation in number of fruits (R² = 0.288) and fiber production (R² = 0.238) in vegetable loofah (Luffa cylindrica).
Even taking these results into consideration, the explanatory variables should not be totally discarded due to the low direct effect on the main variable. In these cases, the use of selection indexes would be a good alternative to be used in breeding programs. These indexes may provide favorable and simultaneous gains in many traits (Cruz et al., 2014).
Selection indexes of Mulamba & Mock (1978) and Genotype-Ideotype (Schwarzach, 1972 cited by Wricke & Weber, 1986) were used in order to identify promising accessions based on vegetative and reproductive traits, simultaneously. Among the studied 16 accessions, the indexes indicated BAGC 236,BAGC 207,BAGC 199,BAGC 98 and BAGC 228 as the most promising materials, changing only the order within each index. This corresponded to a coincidence of 0.92 ordering the accessions according to Spearman's correlation coefficient (Table 4).
The indexes were efficient for indicating earlier accessions, with   (Veilling Holambra, 2018). Therefore, such accessions, which have an average height between 20 and 21.75 cm (Table  1), are suitable for ornamental purposes and can be used to start crosses in the Ornamental Pepper Breeding Program in the state of Piauí.