Estimation of repeatability and number of evaluations for characterization of mango germplasm.
Costa J. G. da
Author Affiliation: Embrapa Semi-Árido, P.O. Box 23, 56300-970, Petrolina, Pernambuco, Brazil.
: 295-298
Abstract : A study was conducted to estimate the coefficient of repeatability for total fruit number per tree (TFT) fruit production per tree (FPT), fruit weight average (FWA), pulp weight (PW), skin weight (SW), stone weight (STW), fruit longitudinal diameter (FLD), fruit transversal diameter (FTD), FLD:FTD ratio, stone longitudinal diameter (SLD), stone transversal diameter (STD), SLD:STD ratio, and to determine the necessary number of evaluations for mango germplasm characterization. Repeatability estimates were carried out by principal component analysis from the correlation matrix for four years for TFT, FPT and FWA and four evaluations at the same year for the other characters. Repeatability estimates were 0.51, 0.53, 0.81, 0.84, 0.82, 0.72, 0.91, 0.91, 0.91, 0.91, 0.78 and 0.65 for TFT, FPT, FWA, PW, SW, STW, FLD, FTD, FLD:FTD ratio, SLD, STD and SLD:STD ratio, respectively. With four evaluations, it was possible to infer, with accuracy, ~80% for the TFT and FPT, 89% for SLD:STD ratio and over 90% for the other characters. Considering the savings in labour and time, and data accuracy, except for TFT and FPT, all the characters had efficient discrimination of the accessions. To obtain a 95% of genotypic determination coefficient (R2), two evaluations for FLD:FTD ratio, four evaluations for PW, and five evaluations for FWA were necessary. For SLD:STD ratio, five evaluations were necessary to obtain 90% of R2. TFT and FPT were not the most appropriate for mango germplasm characterization since six and five evaluations, respectively, were necessary to obtain 85% of R2.