2.5. Statistical analyses
In experiments conducted to determine the effect of different gases on storage of Bb conidia, jars flushed with gases other than O2 in which considerable leakage took place (final O2 concentration >3.5%) were discarded. With the exception of the tests with helium (see Section 3), no more than one or two of the eight replicates of each treatment were discarded. In gas flushing experiments with 20% CO2/80% N2 at different temperatures, all glass jars were considered in statistical analyses independently of O2 ingress, as final conidial viabilities across replicates within treatments were similar, regardless of final O2 concentration. Data in the form of percentages were arcsine square root transformed and examined using a one-way analysis of variance. Means were compared by the Tukey–Kramer HSD test or t-test and considered to be statistically
different at the 5% significance level. Data analyses were performed using the JMP statistical package.