The effect of data transformation on the genetic parameter estimates of resistance to Cercospora Sojina Hara in soybean
Geraldo de Amaral Gravina; Paulo Roberto Cecon; Sebastião Martins FilhO and Carlos Sigueyuki Sediyama
The objective of this work was to study the effect of data transformation on the estimates of genetic parameters of soybean resistance to Cercospora sojina Hara inheritance. The means and variances of populations (P1 and P2, and the F1, F2, RC1 and RC2 generations) derived from a cross between a resistant (‘Paraná’) and a susceptible (‘Bossier’) cultivar were evaluated for five characteristics associated with the disease: infection degree (ID) – visually evaluated; number of lesions per leaflet (NLF); lesion mean diameter (LMD); percentage of lesioned foliar area (PLFA) and disease index (DI). The studied transformations were: square root of (x), (x + 1), (x + 0,5) and (x + 3/8) and logarithm (x+1). Results showed that the transformations did not alter the significance of the estimated genetic parameters. The additive genetic effect was the most important, for all the evaluated traits. The additive-dominant model was sufficient to explain the variations only for PLFA; for the other traits, the non-allelic interactions were significant. The influence of the effects of the epistatic interactions varied from 5.89% for PLFA up to 35.93% for NLF. For transformed or non-transformed data, the additive-dominant genetic model was satisfactory to explain the behavior of the generation means for the PLFA character. However, for ID, NLF, LMD and DI, the additive-dominant-epistatic model was more appropriate. The ID, NLF and DI traits, which presented higher additive genetic variability than dominance variability, can be utilized by breeders without transformation as selection criteria, to obtain superior homozygous genotypes with selection beginning at the F2 generation.