Estudo de correlações simples entre ensaios de variedades de trigo realizados no Rio Grande do Sul

Data from grain yields grown in the experimental network of wheat variety tests in Rio Grande do Sul, Brazil, in 1966 and 1967 were used to calculate correlation coefficients between the yields of varieties from the different testing sites. Thirty varieties were seeded at each of 7 locations in 1966 and 20 varieties at each of 8 locations in 1967. However, only 18 varieties were grown both years. Strain numbers, variety names, and pedigrees for each of the 32 strains are presented. Dates of planting, kind and quantity of fertilizer applied and general observations of conditions of each of the 15 tests are given in tabular form. The average grain yield (kg/ha) of each variety at each location was correlated with the average yield at all other locations. Only the data from the early seeding date (usually in June) were used, in as much as these yields were superior in 11 of 14 comparisons. Grain yields averaged considerably higher in 1966 than in 1967. The highest positive correlation coefficient (.731) was between Sao Luiz Gonzaga and Cruz Alta in 1967. Of the 7 possible correlation coefficients between Sao Luiz Gonzaga and the other 7 locations, only Sertao and Piratini were non-significant. Of the possible 49 correlation coefficients, only 18 were statistically significant and of these only two (between Cruz Alta and Vacaria in 1966 and between Piratini and Sananduva in 1967) were negative. Notwithstanding this low number of significant correlations between locations, certain varieties, illustrated by graphical examples, performed in a superior manner at many locations regardless of the value of the correlation coefficient between the locations. This observation is in agreement with international Wheat Trials reported by CIMMYT from Mexico.