Divergência genética, capacidade de combinação e heterose em melancia

Abstract

Watermelon (Citrullus lanatus) belongs to the cucurbit family, which originates in the hot regions of tropical Africa and is widespread worldwide due to the refreshing taste, diuretic properties and its facility to be consumed in natura, especially as a refreshing dessert. The watermelon is considered one of the main cucurbits cultivated in Brazil, in the 2015 harvest the production was 2,119,599t, coming from 97.910 hectares cultivated. However, a large part of the Brazilian watermelon production comes from the planting of a few closely related cultivars, what leads to causing a narrowing of the genetic basis of the species. This factor makes it difficult to obtain gains by using breeding programs for the development of new commercial cultivars. On the other hand, the genetic resources conserved in Active Germplasm Banks (BGAs) plays an important role for genetic variability. It is necessary to increase the knowledge of these genetic resources, in order to provide effective information to help a good selection of genotypes to be used in breeding programs. The objectives of the present work were: to study the genetic divergence; to estimate the general (CGC) and specific (CEC) of combinations and the reciprocal effects (RE) in watermelon genotypes; and heterosis of the hybrids in relation to characteristics of fruit, seed and productivity. Twenty characteristics were evaluated: main branch length; days for the appearance of the first male and female flower; average fruit weight; soluble solids content; fruit length and width; fruit length and width ratio; mean rind thickness; pulp consistence; pulp color; seed length and width; seed width and length ratio; seed mass and thickness; average number of fruits per plant; and yield. The study of genetic diversity was carried out using multivariate analyzes with canonical variables and grouping by UPGMA method. For the reciprocal effects and combining ability estimative, crosses and statistical analysis of the data were performed according to Griffing's model I (1956), while the analysis of diallel crosses and heterosis was calculated in relation to the average of the parents and the superior parent 'JNY (1)', 'ORA (2)', 'KOD (3)', 'SOL (4)', 'CHG (5)' and 'PEA (6)'. The appraised genotypes presented great genetic variability and, in addition, the descriptors used to discriminate the genotypes were efficient. As for the multivariate analysis methods, they were concordant among themselves, demonstrating to be efficient for the selection of divergent genotypes. As concerns, the estimates of the general combining ability, the 'JNY (1)' and 'KOD (3)' genotypes were the most promising to obtain hybrids with reduced fruit size and small seeds. In addition, reciprocal effects indicated that genotypes 'JNY (1)' should be used as donor and 'KOD (3)' as pollen recipient. While the genotypes 'ORA (2)' and 'CHG (5)' can be used to give rise to hybrids with greater fruit mass. Concerning estimates of specific combining ability 'ORA (2)' x 'PEA (6)' combinations; 'ORA (2)' x 'JNY (1)'; 'CHG (5)' x 'KOD (3)'; 'PEA (6)' x 'KOD (3)' and 'CHG (5)' x 'SOL (4)' as being promising genotypes for the smallest fruit size and small seeds. In relation to heterosis, the combinations 'CHG (5)' x 'ORA (2)', 'JNY (1)' x 'PEA (6)' and 'ORA (2)' x 'CHG (5)' revealed potential for the market of early hybrids, with greater fruit mass, long, firm pulp; while the 'KOD (3)' x 'JNY (1)' and 'JNY (1)' x 'KOD (3)' combinations displays potential for smaller fruits and seeds.