Ge. Scott et N. Zummo, SIZE OF MAIZE SAMPLE NEEDED TO DETERMINE PERCENT KERNEL INFECTION BY ASPERGILLUS-FLAVUS, Plant disease, 79(8), 1995, pp. 861-864
Aspergillus flavus usually infects a low percentage of maize kernels,
but it is a very important pathogen because it produces aflatoxin. The
object of this study was to determine the needed kernel sample size p
er plot in replicated trials to ascertain the percentage of infection
by A. flavus at a reasonable level of precision. We compared 65-, 130-
, 195-, and 390-kernel sample sizes. Two experiments were simulated to
determine which statistical attribute(s) was the most sensitive to ch
anges in sample size. The changes in error mean squares and F values b
etween 65- and 390-kernel samples were of a magnitude greater than six
, which was larger than for the other statistical attributes. Based on
averages of error mean squares, F values, and Spearman correlation co
efficients over eight experiments with actual data, we concluded that
a 390-kernel sample size is not too large, and a sample size of approx
imately 200 kernels should be considered a minimum. Based on other cal
culations, a minimum sample size of 100 kernels is needed just to have
a reasonable probability of detecting the fungus when it is present.