Validação de metodologia para o teste de germinação em sementes de nabo forrageiro (Raphanus sativus L. var. oleiferus)

Abstract

For several vegetable species, there is no procedure in the Rules for Seed Testing used in Brazil for tests such as germination, due to lack of standardization and methodological validation. Among these species is oil radish, a promising crop for bio-diesel production, whose germination is assessed by the methodology recommended for Raphanus sativus, although it has varietal specificity (var. oleiferus) with plant and seed characteristics that differ from other varieties of the species. To ensure standardization of methodologies for the international seed trade, The International Seed Testing Association (ISTA), establishes criteria for methodology validation, whose description should be clear and complete with procedures that should be accurate, robust and precise (reproducibility and repeatability). In this the study aims was to evaluate the validation process of two methodologies for the germination test in oil radish seeds, using the ISTA standard procedure, as well as complementary techniques, for statistical analyses. Germination tests were done in eight laboratories with five lots, using sand and paper substrates with alternating temperatures of 20-30 ºC. Statistical techniques were used to verify the homogeneity of lots (H test), identify the presence of discrepant values (Hampel method) and outliers in variances (Levene's test for average); assess the effects of laboratories and lots (analysis of Variance); verify repeatability, reproducibility, accuracy and robustness (critical limits of repeatability, reproducibility, Mandel's h and k statistics); compare different germination methods (F tests); and reevaluate the quality level of lots (Discrimination Analysis). Both in statistical and phytotechnical aspects, the methodologies for the germination test in oil radish seeds with sand and paper substrates and alternating temperatures of 20-30 ºC can be consider validated, as they demonstrated accuracy, robustness, repeatability and reproducibility.