Of all the processes used in foliar evaluation, sampling is currently still the most subject to error. To corroborate this, the present study aimed to determine the size of foliar samples and error variation of samples collected from mango tree leaves. The study used a completely randomized design with six repetitions and four different treatments which consisted of a leaf from each of the cardinal points from 5, 10, 20 and 40 different trees. Mean variation, mean standard error, mean confidence interval and error percentage compared to mean values were calculated based on level of nutrients. Results were calculated by using the semi-amplitude of the interval of confidence expressed as mean percentages. Obtained results led to the conclusion that the chemical evaluation of macro-nutrients from 10 mango plants would be sufficient for assessment. However, for assessment of micro-nutrients, the sampling of 20 plants would be necessary, and, when considering Fe, a total of 30 plants should be used.
leaf diagnosis; Mangifera indica; mineral nutrition
