Optimum sample size and spatial dispersion of red scale, Aonidiella aurantii on an orange orchard in Australia

Two-stage sampling and geostatistical teechniques for cost-effective and precise sampling were examined using red scale data that were collected from a commercial orange orchard in Kulnura, Australia in mid-summer, 2004 and 2005. The distribution pattern of red scale on a twig and a fruit well followed the negative binomial, and the degree of aggregation was higher on a fruit than a twig. The analysis of variance and two-stage sampling were used to obtain the suitable sample unit (a leaf, a twig containing two leaves and a 15cm. branch and a fruit in this study) and optimum sample size. A fruit was the most suitable than any other sample units, and a twig was better than a leaf. The optimum sample size for twigs and fruits per tree was 4 twigs (2 leaves and 15cm. branch) and 4 fruits (2 directions), respectively. There was a non-linear relationship between 2 years for the density on 40 fruits of the same tree, because the Spearman rank correlation coefficient (0.84) was much higher than the Pearson's (0.29). The autocorrelation analysis showed that omnidirectional 10cm. apart from each sampled tree was needed to obtain the independent data.