Inheritance of beta-carotene content in melon

– The objective of this work was to determine the inheritance of beta-carotene content in melon ( Cucumis melo ). The AC-16 accession ( Cucumis melo subsp. melo var. acidulus ) – with a low beta-carotene content and white mesocarp – was crossed with the Vedrantais cultivar ( C. melo subsp. melo var. cantalupensis ) – with a high beta-carotene content and salmon colored mesocarp –, to obtain the F 1 , F 2 , BC 1 , and BC 2 generations. The AC-16 and 'Vedrantais' parents, the F 1 and F 2 generations, and the BC 1 and BC 2 backcrosses of each parent were evaluated. The quantification of beta-carotene was carried out in a high-performance liquid chromatography system. Mean components related to the additive and dominance effects, additive and dominance variances, and heritability were estimated. The beta-carotene content was high in 'Vedrantais' (17.78 µg g -1 ) and low in AC-16 (0.34 µg g -1 ). The following results were observed: additive and dominance effects on the genetic control of the character, incomplete character dominance, estimated number of loci close to two, greater variance for segregating populations (F 2 and backcrosses), and heritability values in the broad (87.75%) and narrow (64.19%) senses. The beta-carotene content in melon is controlled by a major effect gene, with additive and dominance effects associated with polygenes with additive effects.


Introduction
The Northeastern Semiarid Region is the major producer and exporter of melon in Brazil. The national production is concentrated in the state of Ceará, in the Vale do Jaguaribe, and in the state of Rio Grande do Norte, in the Mossoró-Açu Agricultural Pole. These states are responsible for more than 99% of the national production and exports of melon (IBGE, 2022). The edaphoclimatic conditions and the high technology used are the main reasons for the productive success of this species in the Northeast (Figueirêdo et al., 2017).
There is a wide variation in the germplasm of melon (Cucumis melo L., Cucurbitaceae) for its mesocarp color, which is determined by a combination of pigments such as chlorophylls and carotenoids (Abo Sedera et al., 2016;Gur et al., 2017). One of the first works on the inheritance of mesocarp color in melon was developed by Hughes (1948), who verified the presence of the genes named gf and wf which influence the green and white mesocarp colors, respectively. Subsequently, Clayberg (1992) reported that the orange color is dominant over the green and white colors. The main pigment accumulated in the orange or salmon mesocarp in melon is beta-carotene, a precursor substance of vitamin A with antioxidant potential (Chayut et al., 2017). Currently, there is a growing demand for the biofortification of vegetables for human consumption, aiming at improving their quality and helping to prevent and reduce the risk of diseases (Bae et al., 2019;Gómez-García et al., 2020). In this sense, melon breeding programs have sought productive cultivars with excellent fruit quality and higher beta-carotene content.
In order to improve a characteristic, it is important to know its genetic inheritance. This knowledge helps the plant breeder in the decision-making on which strategy is most appropriate to be used to obtain fruit desirable characteristics. In plant inheritance studies, the genetic effects (additive and non-additive), the average degree of dominance, and the number of genes or loci involved are commonly estimated (Cruz et al., 2014).
Some studies have shown a complex inheritance for the accumulation of beta-carotene in melon (Monforte et al., 2004;Cuevas et al., 2010). Other studies identified quantitative trait loci (QTLs) related to that characteristic (Monforte et al., 2004;Cuevas et al., 2009;Diaz et al., 2011;Perpiñá et al., 2016). Nevertheless, studies on the inheritance of beta-carotene content in melon are rare. Thus, research which seek a better understanding of the genetic mechanism involved in the accumulation of beta-carotene in melon mesocarp in Brazilian conditions are still necessary.
The objective of this work was to determine the inheritance of beta-carotene accumulation in melon.

Materials and Methods
The experiment was carried out in 2015, from planting date on May 30 th to harvest date on August 4 th , at the Fazenda Experimental Rafael Fernandes, located in Lagoinha district (5°03'37"S, 37°23'50"W, at 72 m altitude), rural area of the municipality of Mossoró, in the state of Rio Grande do Norte (RN), Brazil. The soil of the experimental area is classified as Latossolo Vermelho-Amarelo, according to the Brazilian Soil Classification System (Santos et al., 2018), i.e., Ferralsols, with a sandy loam texture. The climate of the region is typified as Bswh, that is, semi-arid tropical, very hot and dry, according to the Köppen-Geiger's classification, with a rainy season in the summer, extending to the autumn.
Vedrantais cultivar and the melo accession AC-16 were used as parents. Vedrantais is a French cultivar, belonging to the variety cantalupensis, developed by the Vilmorin Seed Company. This cultivar has round-shaped Charentais-type fruit (IF = 1.0), orange mesocarp color, high levels of sucrose and soluble solids (> 12º Brix). AC-16 accession belongs to the variety acidulus, shows a white mesocarp, a yellow exocarp and low levels of sucrose and soluble solids (< 4º Brix). The crossing these parents resulted in the branch generation F 1 and, subsequently, the generation F 2 and the backcrosses (BC 1 and BC 2 ). One fruit per plant was collected for analysis, discarding the border fruit.
An experimental design in randomized complete blocks was carried out with three replicates. The evaluated generations were the parents (AC-16 and 'Vedrantais'), the F 1 and F 2 , and the backcrosses from each parent (BC 1 and BC 2 ). Due to the genetic variability present in each generation, the plots consisted of the following numbers of plants: 13 ('Vedrantais'), 15 (AC-16 and F 1 ), 172 (F 2 ), 43 (BC 1 ), and 46 (BC 2 ). The preparation of the area performed by plowing and harrowing, with subsequent formation of 0.2 m high and 0.6 m wide ridges. The mineral nutrition of the crop was held by fertigation on the basis of the chemical analysis of the soil. Seeding was performed in expanded polystyrene trays with 200 cells, filled with a commercial substrate (Polifértil Nutrição Importação e Exportação, Uberaba, MG, Brazil). The seedlings were transplanted when they showed the first fully expanded true leaf. The experimental plot consisted of a 3.0 m long row, 2.0 m spacing between rows, and 0.3 m between pits, with one plant per pit. A localized drip irrigation system was used, with 1.5 L h -1 average flow rate, with one emitter per plant. The phytosanitary control and other cultural treatments were conducted in accordance with the technical recommendations adopted in the region for melon tree (Figueirêdo et al., 2017).
For the qualitative analysis of beta-carotene, the company's standard Sigma-Aldrich with 95% purity was used. Methanol, acetonitrile, and ethyl acetate were used to perform the chromatographic analyses. Melon pulp used to quantify beta-carotene was homogenized and stored at -20°C. Βeta-carotene was extracted in a 15 mL Falcon tube, protected from light, by adding 3 g of pulp and 3 mL of isopropyl alcohol. This mixture was homogenized in a vortex and centrifuged (10,000 RCF) for 5 min at 4°C. Supernatant was collected, and the precipitate was reextracted with 6 mL acetone, and then reextracted with 3 mL ethyl acetate. At each reextraction, the mixture was homogenized in a vortex (Shanghai Hannuo Instrument Co. Ltd., XH-B, Shanghai, China) and centrifuged (Shanghai Surgical Instrument Factory, 80-2, Shanghai, China) at 10,000 RCF for 5 min at 4°C.
Supernatants were pooled and concentrated in an evaporator (Sigma Aldrich, Laborota 4003, Saint Louis, MO, USA), at 30°C, 0.09 MPa, protected from light. After the evaporation of the solvents, betacarotene was resuspended in 2 mL ethyl acetate and filtered through 0.45 µm membrane of polyvinylidene fluoride (PVDF). The quantification of betacarotene was performed in a high-performance liquid chromatography system (HP 1090, Hewlett-Packard, Orsay, France). The mobile phase was formed by methanol, ethyl acetate and acetonitrile, at 2:1:1 ratio; with 3 µL injection volume, 0.3 mL min -1 flow rate, 40°C column temperature, and 450 nm maximum wavelength.
To perform the genetic parameter estimations and the adjustment of the additive-dominant model, the mixed model method proposed by Piepho & Möhring (2010) was used. The test implementation of the additive-dominant model was performed using the method proposed by Kearsey & Pooni (1996), according to the following expression: in which: µ i is the value of the i th generation under the additive-dominant model i = 1, G; m is the intercept; [a] is the additive effect; [d] is the dominance effect; x i1 and x i2 are the coefficients of additive and dominance effects, respectively; λ i is the nonadjustment effect.
As the generations were evaluated in a randomized complete block design, the model for plant k, evaluated in plot i, in block j, was the following: in which: the error of the plot p ij and the plant error e ijk have normal distribution with zero average and variances σ 2 p and σ 2 e , respectively. The genetic variation within generation was modeled by the genetic effect g ijk at the plant-specific level. The SAS code used for all analyses was adapted from Piepho & Möhring (2010). The adjustment of the genetic variance structure was performed by the MIXED procedure in the SAS Version 9.2 program (SAS Institute Inc., Cary, NC, USA).
According to the method proposed by Piepho & Möhring (2010), the average components related to the additive [a] and dominance [d] effects were estimated, as well as the variances of addition, dominance, and heritability. The average degree of dominance (ADD) and the minimum number of genes (η) involved in the expression of the character were also estimated.
The genetic models were tested using the maximum likelihood in mixtures of normal density functions as described by Silva (2003). The distributions for each population was evaluated as follows: In order to complement the study of classical inheritance carried out by the components of average and variances, another study was conducted with models that consider the effect of a major gene and polygenes (Table 1). During the construction of the genetic model, the most general model was considered to be the one with the existence of a gene with a greater effect plus polygenes with additive and dominance effects and equal environmental variances, in all generations. Independent genes were also admitted (both polygenes and genes with greater effect). From the likelihood functions for each model, it was possible to compose tests of interest, considering different hypotheses. The tests were performed using the statistical software Monogen v.0.1 (Silva, 2003).

Results and Discussion
Regarding the mesocarp color of the parents, 'Vedrantais' -with orange mesocarp -showed a high content of beta-carotene (17.78 µg g -1 ), while the parent AC-16 -with white mesocarp -showed very low content of beta-carotene (0.34 µg g -1 ), which confirms the contrast between parents, indicating that their choice to obtain segregating generations was adequate (Figure 1). In a study on the inheritance of beta-carotene content and its association with the mesocarp color in melon, Cuevas et al. (2010) used the American line 'Top Mark' (orange mesocarp) and the Chinese line 'Q 3-2-2' (white mesocarp) as parents. The beta-carotene content levels in these parents were 6.56 and 0.15 µg g -1 , respectively, which are lower values than those observed in the present work. It is underlined that beta-carotene content is influenced by the genotype, the stage of fruit development and environmental conditions (Sharma et al., 2020;Singh et al., 2021). Clayberg (1992) checked epistasis between the genes gf and wf for the genotypic combinations (wf + _/gf + _) and (wf + _/gfgf ) to produce fruit with orange mesocarp, as well as for combinations (wfwf / gf + _) to produce fruit with white mesocarp, and the combination (wfwf/gfgf ) which is responsible for the green color. Thus, the parents 'Vedrantais' and AC-16 are (wf + wf + /gfgf ) and (wfwf /gf + gf + ), respectively.
The three parameters ([m], [a], and [d]) which constitute the average components of the additivedominant model were significant, indicating the presence of additive and dominance effects in the genetic control of the character ( Table 2). The negative value of [d] is explained by the fact that the dominance occurs towards the phenotypic manifestation of lower character magnitude, that is, the occurrence of lower beta-carotene content.
The average degree of dominance, obtained from the variance components, was lower than the  (Table 2). Cuevas et al. (2010) estimated four loci in the genetic control of beta-carotene content in the cross between the Chinese lineage 'Q 3-2-2' (white mesocarp) and the line 'Top Mark' (salmon mesocarp). The differences basically occur due to the parents used and the environmental conditions. Regarding the variance components, the estimates are in line with the expectations: greater variances of segregating populations (F 2 and backcrosses), in which the variance of the generation F 2 was greater than the others ( Table 2). The additive variance was greater than the dominance variance. According to Cruz et al. (2014), the additive variance is one of the determining factors of covariance between relatives, and its magnitude is indicative of the relationship between the selection unit and the improved unit.
Heritability values in the broad and narrow senses were 87.75 and 64.19%, respectively ( Table 2). The narrow-sense heritability was, as expected, lower than the broad-sense heritability, since the former contains only the contribution of additive genetic variance. In this case, the greater the additive variance estimate, the greater the heritability in the narrow sense. The estimates obtained in this work are an indication of a favorable condition for selection because most of the phenotypic variation is due to genetic causes.
Differences were found between the model 1 and the others (Table 3), indicating that the model 1 should be used to test the effect of greater gene and polygenes. When comparing models 1 and 5, the presence of the greater gene with additive and dominance effects was tested, to check their presence in the genetic control of beta-carotene content in melon. When comparing models 1 and 7, the presence of polygenes with additive effects and dominance was tested. In this case, the presence of polygenes in the inheritance of beta-carotene content in melon was also checked.
Regarding the comparison of models which consider the effect of a major gene and polygenes (Table 3), it can be inferred that the inheritance of beta-carotene content in melon is complex, as observed by Monforte et al. (2004) and Cuevas et al. (2010). According to Monforte et al. (2004), the production of betacarotene (orange mesocarp) is independent of the genes related to green (gf ) and white (wf ) colors. In a study with microsatellite molecular markers in the cross 'Shongwan Charmi PI 161375' (green mesocarp) and a line type 'Frog Skin' (white mesocarp), the same authors identified three QTLs associated with the orange color. The authors reported that the orange color is associated with the accumulation of beta-carotene. As checked by Cuevas et al. (2010), the quantitative variation of beta-carotene content is strictly associated with the variation of orange mesocarp tones. Thus, there must be genes related to the production of betacarotene and polygenes which affect this character content in melon. Perpiñá et al. (2016) identified a high-effect QTL on chromosome 9 from studies with introgression lines derived from the cross between 'Vedrantais' and 'Ginsen-makuwa', an oriental cultivar. Ramamurthy & Water (2015), when studying the F 2 generation from the cross between melons of the varieties flexuosus and cantalupensis, detected two QTLs influencing the pulp color, both co-located with QTLs already registered for the concentration of betacarotene. By this result, the pulp color can serve as a selection index for the concentration of beta-carotene in melon. Tzuri et al. (2015) identified the gene CmOr, an allele previously described as gf gene, as responsible for the orange color of melon mesocarp. The authors identified salmon CmOr haplotypes capable of inducing beta-carotene accumulation in melon, while the green/white haplotypes do not act this way. Therefore, the accumulation of beta-carotene is closely related to the polymorphism of the CmOr (Monforte et al., 2014;Chayut et al., 2015;Feder et al., 2015;Lian et al., 2021). An identified SNP marker for CmOr may allow the use of new genomic tools to understand the accumulation of beta-carotene and allow to obtain biofortified cultivars.
Due to the complexity of inheritance, the genetic improvement for a high content of beta-carotene in melon can be done by means of backcrossing for the introgression of the gene with greater effect and, also, by methods that seek the gradual accumulation of that pigment, such as recurrent selection. Due  to the presence of additive and dominance effects, the exploration of heterosis is recommended for this character. In fact, currently, melon cultivars around the world are simple hybrids due to their homogeneity and the presence of heterosis in characteristics of interest.

Conclusion
The inheritance of beta-carotene content in melon is complex, with the presence of a greater effect gene, and additive effects and dominance associated with polygenes with additive effects.