Genetic parameters in melon sub-accessions from traditional agriculture

ABSTRACT Family farming in Brazil holds a high diversity of melon germplasm, composing an important source of alleles for breeding programs. Thus, the objective of this study was to estimate genetic parameters and select genotypes from a population of melon sub-accessions from different botanical varieties grown by family farmers, based on morphological parameters. Two experiments were conducted, one in 2019 and another in 2020, in a complete randomized block design, with three replications and five plants per plot, using 27 melon sub-accessions (generation S2) from family farmers, and a commercial variety. Nine quantitative descriptors were evaluated. Assumptions of ANOVA were tested, followed by individual and joint analyses of variance. Significant differences were found among sub-accessions for all descriptors evaluated, with heritabilities higher than 83% and significant genotype-environment interactions for 88.8% of the characteristics evaluated. Thus, genetic variability was found among sub-accessions, with predominance of genetic effects over environmental effects, denoting the possibility to obtain genetic gains by the improvement of several characteristics of agronomic interest. The sub-accessions BGMEL66.0, BGMEL111.0, and BGMEL112.0 are recommended for inclusion in breeding programs focused on obtaining good prolificacy and small fruits with high soluble solid contents. BGMEL sub-accessions (108.3 and 108.5) can generate progenies with high prolificacy, and sub-accessions of the variety momordica can be used for generation of progenies focused on shortening the crop cycle and increasing fruit size.


INTRODUCTION
Melon (Cucumis melo L. Cucurbitaceae) is a species that presents centers of diversity in Africa and Asia (PITRAT, 2013).However, it has great economic expression in Brazil, mainly those from the inodorus and cantalupensis groups; the Northeast region is responsible for 96.84% of the Brazilian melon production (IBGE, 2020).However, creole varieties have been grown in small rural properties, where farmers use their own seeds for new crops (QUEIRÓZ; BARBIERI; SILVA, 2015), resulting in a high variability.
Previous studies using melon germplasm from family farming showed the existence of high variability among melon accessions (DANTAS et al., 2012;ARAGÃO et al., 2013;AMORIM et al., 2016;MACÊDO et al., 2017;ANDRADE et al., 2019).However, some important characters for the improvement of melon were not emphasized in these studies, mainly regarding characteristics of some botanical varieties (PITRAT; HANELT; HAMMER, 2000), which are believed to be important factors and, therefore, should be considered.
Information on genetic factors is essential for any breeding program to identify and maintain favorable alleles; thus, obtaining estimates of genetic parameters is essential to identify the action of alleles involved in controlling characters and estimate genetic gains by selection.(CRUZ; CARNEIRO;REGAZI, 2014).
Melons from traditional agriculture are an important source of alleles for breeding programs.Although several studies have shown variability among melon accessions (AMORIM et al., 2016;MACÊDO et al., 2017), there is no study on this germplasm, focused on generating information for selection of genotypes.It denotes the need for studies using melon germplasm from family farmers for morphological characterization of different botanical varieties based on genetic parameters, focused on selecting superior genotypes.
Thus, the objective of this study was to estimate genetic parameters and select genotypes from a population of melon sub-accessions from different botanical varieties grown by family farmers, based on morphological parameters.

MATERIAL AND METHODS
Two experiments were conducted, one in 2019 and another in 2020, at the Experimental Field of the Department of Technology and Social Sciences of the Bahia State University (DTCS/UNEB), in Juazeiro, Bahia, Brazil (09°25'04.92271''Sand 40°29'04.73710''W,and altitude of approximately 352 meters).Twenty-seven melon subaccessions (AMORIM et al., 2016) (S 2 generation) from the botanical varieties momordica, cantalupensis, and makuwa and some accessions not identified were evaluated.These accessions were from the traditional agriculture of the state of Maranhão, Brazil (Table 1) that were stored in the Active Germplasm Bank of Cucurbitaceae from the Northeast Region at the Brazilian Agricultural Research Corporation (Embrapa Semiarid), in Petrolina, Pernambuco.A commercial variety (Melao Amarelo) was used as control.Thirty seeds of each sub-accession were sown in plastic trays filled with a commercial substrate, in a greenhouse covered with a 50% shade screen, and irrigated daily.The seedlings were transplanted to soils previously prepared with plowing and harrowing 15 days after sowing.
The experiments were conducted in a complete randomized block design, with three replications, five plants per plot, and spacings of 2.5 m between rows and 0.8 m between plants, under localized drip irrigation.
The experiments were conducted approximately in the same period (January to April) in 2019 and 2020.Weeding and plant health status monitoring were carried out; the natural soil fertility was adopted, since this system is commonly used under traditional agriculture.
The evaluations were carried out using the following quantitative descriptors (IPGRI, 2003;PITRAT;HANELT;HAMMER, 2000): fruit weight (kg); fruit diameter and length (cm); fruit cavity diameter and length (cm); pulp thickness (cm); soluble solid contents (ºBrix) in pulp composite samples homogenized in a kitchen food processor; earliness (number of days from transplanting to harvest), and prolificacy (number of fruits per plant, counted at the end of the crop cycle).
Regarding the statistical analyses, firstly, the assumptions of ANOVA were tested, transforming the variables when necessary.Individual analysis of variance was then performed for each growing year to assess whether the sub-accessions differed from each other.Subsequently, test of homogeneity of variances was applied (Fmax: ratio between the highest and lowest residual mean square for each descriptor).Joint analysis (A×B simple factorial) was carried out using the model: Y ijk = µ + G i + A j + GA ij + B/A jk + Ɛ ijk , where µ = overall mean; G i = effect of the i-th genotype ; A j = effect of the j-th environment ; GA ij = effect of the interaction of the i-th genotype with the j-th environment; B/A jk = effect of the k-th block inside the j-th environment; and Ɛijk = random error and effects: G (random) and A (fixed) (CRUZ; REGAZI; CARNEIRO, 2012).All genetic and statistical analyses were processed using the program Genes (CRUZ, 2013).

RESULTS AND DISCUSSION
The data of analysis of variance showed significant differences among melon sub-accessions for all characteristics evaluated (Table 2), denoting the existence of genetic variability among sub-accessions.Similar results were found in previous studies on melon accessions from traditional agriculture (AMORIM et al., 2016;MACÊDO et al., 2017;ANDRADE et al., 2019), supporting those found in the present study and, therefore, denoting the possibility of selecting agronomically superior accessions for the characteristics analyzed (CRUZ; REGAZI; CARNEIRO, 2012).Means followed by same lowercase letter in the columns, or uppercase letter in the rows, are not statistically different from each other by the Scott Knott test at 5% significance.SUB = sub-accession; 19 and 20 = evaluation years of 2019 and 2020; FW = fruit weight (Kg); EARL: earliness (number of days from transplanting to harvest); PROL: prolificacy (number of fruits per plants, counted at the end of the crop cycle); FD and FL= fruit diameter and length, respectively (cm); FCD and FCL = fruit cavity diameter and length, respectively (cm); PT = pulp thickness (cm); SS: soluble solid contents (°Brix); Min and Max refer to the individual values and show the variation within the sub-accession; Fc = Snedecor's F distribution.** = significant at 1%; CV(%) = coefficient of variation; Ama = commercial variety Melao Amarelo; and mo, mk, c, and nd = varieties momordica, makuwa, cantalupensis, and not defined, respectively.
The earliness in the different growing years presented variations in for 37% of the sub-accessions: five from the variety cantalupensis (BGMEL78.0,BGMEL82.2,BGMEL 86.1, BGMEL 87.2, and BGMEL97.1),five from botanical varieties not identified (BGMEL68.3,BGMEL77.3,BGMEL83.2,BGMEL98.0, and BGMEL109.2),and the control.According to the means of the melon varieties, the highest means were found in the growing year I (2019), except for momordica.The lowest means were found for momordica and the highest for the commercial variety (Melao Amarelo) (Table 2), denoting the potential of the variety momordica for selection focused on increasing earliness.
Prolificacy and fruit cavity diameter presented only 18.5% variation, highlighting the variety makuwa and botanical varieties not defined (ND), respectively.Regarding the other characteristics, the highest variations were found for sub-accessions of ND varieties (Table 2); these variations are probably because of the high variability among plants within the sub-accessions, which is due to the introgression of alleles among different botanical varieties (MACÊDO et al., 2017;AMORIM et al., 2016).
All characteristics were, in general, favored in the growing year II (2020), except soluble solid contents.Subaccessions from momordica stood out for earliness and characteristics related to fruit size (fruit weight, fruit diameter and length, fruit cavity diameter and length, and pulp thickness).In addition, they presented a good prolificacy (Table 2), denoting that this botanical variety is important for selection processes focused on improving these characteristics.
Sub-accessions from ND varieties presented the highest prolificacy (Table 2), mainly the sub-accessions BGMEL108.3and BGMEL 108.5, however they presented small fruits with low soluble solid contents.Sub-accessions from makuwa also presented prolificacy and progenies with high soluble solid contents, mainly BGMEL66.0,BGMEL111.0,and BGMEL112.0,with small fruits and good prolificacy, which can be a novelty in the market.The control (commercial variety) presented low prolificacy and mediumsized fruits with low solid soluble contents, under the same crop conditions.The commercial variety presented low performance was probably due to the use of the natural soil fertility management, which is common for traditional agricultural crops, since high chemical fertilizer rates are applied to soils for commercial crops.
The joint analysis of variance (Table 3) showed coefficients of variation varying from 5.40 (earliness) to 57.38 (prolificacy).Joint analysis is recommended only for environments with homogeneous residual variances.According to Cruz, Regazi and Carneiro (2012), several tests can be used to evaluate the homogeneity of residual variances, however, they have limitations or restrictions of use; thus, a practical criterion that can be adopted for grouping experiments to proceed joint analysis is to combine trials whose residual mean squares do not exceed the approximate ratio of 7:1 in the same group.In the present study, the ratios between the highest and lowest variances were between 1.03 and 4.01 for all characteristics evaluated (Table 3), which allowed to proceed the joint analysis and assess the genotypeenvironment interaction.The high variability found among sub-accessions is partially due to existing differences among plants within each sub-accession.Studies on melon germplasm showed a high variation among accessions (DANTAS et al., 2012;ARAGÃO et al., 2013;TRIMECH et al., 2013;YILDIZ ;AKGUL;SENSOY, 2014;ANDRADE et al., 2019) and among plants within accessions (AMORIM et al., 2016;MACÊDO et al., 2017). 1 SV = source of variation; DF = degrees of freedom; EARL = earliness; FW = fruit weight; FD = fruit diameter; FL = fruit length; FCD = fruit cavity diameter; FCL = fruit cavity length;; PT = pulp thickness; SS = soluble solid contents (°Brix); PROL = prolificacy; G = genotype; E = environment; G×E = genotype-environment interaction; RES = residue; CV = coefficient of variation; Fmax = ratio between the highest and lowest residual mean square; **, * = significant at 1% and 5% significance, respectively; ns = not significant.
The joint analysis of variance (Table 3) showed that the genotype effect was highly significant (p≤0.01) for all variables, whereas the environmental effect was not significant, except for prolificacy.The genotype-environment interaction was significant for all variables, except prolificacy.
The predominance of estimates of genetic effects over environmental effects indicates that the genetic factors had a greater effect on the observed phenotype.However, the significant interaction for 88.8% of the variables denotes that the relative performance of the sub-accessions (BORÉM; MIRANDA, FRITSCHE-NETO, 2017) varied in the two growing years for all variables evaluated, except prolificacy.However, the temperature data were similar in the two growing years: mean temperatures varied from 27.46 to 28.03 °C (2019) and from 26.56 to 27.28 °C (2020).Rainfall data showed a small difference between growing years: 0.26 to 5.27 mm (2019) and 1.19 to 9.15 mm (2020) (AGRITEMPO, 2021).
High heritability was found for the genetic parameters, with estimates higher than 83% for all characters evaluated (Table 4), mainly for some characteristics related to fruit size (fruit length, cavity length, and fruit weight).These results denote a great potential for successful selection focused on these characters, as the observed phenotype was mostly affected by the genetic factor.The sub-accessions BGMEL 77.1 and BGMEL 87.1 (variety momordica) and BGMEL 87.3 (variety not defined) stood out for these characteristics (Table 2), showing to be promising for selection processes focused on increasing fruit size.
Lower results were found by Valadares et al. ( 2017) for fruit weight (86.00%) and length (93.00%) when evaluating heritability of 23 melon accessions from the momordica group.However, this difference can be attributed to the use of another set of genotypes in the experiment conducted under greenhouse conditions, which allows for more control of environmental effects.Aragão, Nunes and Queiróz (2015) evaluated melon families and found lower heritabilities than those found in the present study, for all characters evaluated.
The estimated genetic, phenotypic, and environmental variations (Table 4) indicated that the genetic variation (σ² G ) was higher than the environmental variation (σ² E ) for all characters evaluated.Thus, it can be said that genetic effects predominate in the expression of the phenotype, indicating greater reliability and greater genetic gains in phenotypic selection.However, more significant variations between growing years were found for earliness, prolificacy, fruit length, and fruit cavity length. 1 σ² F = phenotypic variation; σ² E = environmental variation; σ² G = genetic variation ; h 2 (%) = heritability; CV g (%) = coefficient of genetic variation; CV e (%) = coefficient of environmental variation; (CV g /CV e ) = CV g to CV e ratio; FW = fruit weight (kg); EARL = earliness; PROL = prolificacy; FD = fruit diameter (cm); FL = fruit length (cm); FCD = fruit cavity diameter (cm); FCL = fruit cavity length (cm); PT= pulp thickness; and SS = soluble solid contents (°Brix).
The coefficients of genetic variation (CV g ) found varied from 2.24% to 61.64% (Table 4) and were higher than the coefficients of environmental variation (CV e ) for all characters evaluated.The lowest CV g were found for earliness and the highest for prolificacy, fruit cavity length, fruit length, and soluble solids.However, the highest CV e were also found for prolificacy and soluble solid contents, indicating that these characteristics were highly affected by environmental factors.Valadares et al. (2017) evaluated melon accessions from the variety momordica and found the highest CV g for pistil scar size (72.04) and soluble solid contents (55.34).Ferreira et al.
(2016) evaluated pumpkin accessions and found the highest CV g for fruit weight and prolificacy, which denotes a greater effect of genetic factors on the expression of the phenotype and reinforces the existence of high variability in the germplasm.
The CV g to CV e ratio (CV g /CV e ) presented values >1 for all characters (Table 4).CV g /CV e equal to or higher than 1 and heritability higher than 80% are favorable conditions for selection (CRUZ; REGAZI; CARNEIRO, 2012), which were found for all characteristics evaluated, denoting great potential for a successful selection.

Table 1 .
Passport data of sub-accessions of Cucumis melo from the Active Germplasm Bank of Cucurbitaceae from the Northeast Region at the Brazilian Agricultural Research Corporation (Embrapa Semiarid), evaluated in 2019 and 2020.

Table 2 .
Test of means for nine characters of melon sub-accessions from family farmers of the state of Maranhão, Brazil, evaluated in 2019 and 2020.

Table 2 .
Continuation.Means followed by same lowercase letter in the columns, or uppercase letter in the rows, are not statistically different from each other by the Scott Knott test at 5% significance.
SUB = sub-accession; 19 and 20 = evaluation years of 2019 and 2020; FW = fruit weight (Kg); EARL: earliness (number of days from transplanting to harvest); PROL: prolificacy (number of fruits per plants, counted at the end of the crop cycle); FD and FL= fruit diameter and length, respectively (cm); FCD and FCL = fruit cavity diameter and length, respectively (cm); PT = pulp thickness (cm); SS: soluble solid contents (°Brix); Min and Max refer to the individual values and show the variation within the sub-accession; Fc = Snedecor's F distribution.** = significant at 1%; CV(%) = coefficient of variation; Ama = commercial variety Melao Amarelo; and mo, mk, c, and nd = varieties momordica, makuwa, cantalupensis, and not defined, respectively.

Table 3 .
Joint analysis of variance among melon sub-accessions from family farmers of the state of Maranhão, Brazil, evaluated in 2019 and 2020.

Table 4 .
Genetic parameters for characters of melon sub-accessions from family farmers of the state of Maranhão, Brazil, evaluated in 2019 and 2020.