Sampling techniques for measuring selected fruit characters in 'Carabao' mango (Mangifera indica L.).
Gauran N. L., Carpena A. L.
Author Affiliation: Palawan Nat. Agric. Coll., Aborlan, Palawan 2918, Philippines.
Philippine Journal of Crop Science 12 : 93-99
Abstract : Two-stage sampling was used to gather measurements for determining the combination of number of trees and fruits/tree required to meet a stated precision with 95% confidence for evaluating each of 11 fruit characters for breeding and agronomic studies. Results showed that several combinations of trees and fruits/tree could be used for a certain level of precision and degree of confidence. The combination that required the minimum total number of fruits and, therefore, the least cost (if cost is proportional to number of fruits) was 8 fruits/tree for all characters. A single tree was found sufficient as the source of fruit samples when using 0.10y? (10% of the grand mean) and 0.20y? precision levels. By looking at the relationship between the number of fruits and the corresponding CV of the mean, sampling 6 to 10 mature fruits from 2 to 46 trees at the 0.05y? precision level was considered optimum for a given character. At the least stringent precision level of 0.20y?, at least one tree and 8 fruits/tree would provide precise estimates of the means for any characters studied. Stratification of trees was not necessary when sampling for most of the characters, except skin weight and total soluble solids. However, for characters where prior stratification is not recommended, the use of simple random sampling resulted in a slight increase in the variance of the mean. Therefore, a slightly larger sample size is required to meet the desired precision, especially when using the 0.125y? level. An additional 1-3 trees should be included in the sample aside from the recommended values to check the anticipated increase in variance.