Hall, David  
Childers, Carl  UNIV OF FLORIDA  
Eger, Joe  DOW AGROSCIENCES 
Submitted to: Journal of Economic Entomology
Publication Type: Peer Reviewed Journal Publication Acceptance Date: October 1, 2006 Publication Date: January 1, 2007 Citation: Hall, D.G., Childers, C.C., Eger, J.E. 2007. Binomial sampling to estimate citrus rust mite (acari: eriophyidae) densities on orange fruit. Journal of Economic Entomology. 100: 233240. Interpretive Summary: Citrus rust mites are important economic pests of citrus in Florida. Growers can monitor infestation densities of rust mites in order to make management decisions. Counting mites to estimate the average mite density in a block of trees can be tedious and time consuming. This research project investigated presenceabsence (binomial) sampling as a method for estimating mean densities of citrus rust mites on oranges. Data for the investigation were obtained by counting the number of motile mites within 600 sample units (each unit a 1 cm2 surface area per fruit) across 4 ha blocks of orange trees. A relatively good relationship was found between the proportion of samples infested and mean density per sample. Although the data fit the binomial, confidence limits for mean predictions increased dramatically as the proportion of samples infested increased. Presenceabsence sampling may therefore have less value for estimating mite densities when the proportion of samples infested is large. Binomial sampling for rust mites appeared to be a viable alternative to absolute counts of mites per sample for a grower using a low management threshold such as 2 or 3 motile mites per sample. Count data from sampling plans consisting of as few as 48 samples per 4 ha block of trees were projected to follow the binomial relationship. Technical Abstract: Binomial sampling based on the proportion of samples infested was investigated as a method for estimating mean densities of citrus rust mites, Phyllocoptruta oleivora (Ashmead) and Aculops pelekassi (Keifer), on oranges. Data for the investigation were obtained by counting the number of motile mites within 600 sample units (each unit a 1 cm2 surface area per fruit) per 4 ha block of orange trees: five areas per 4 ha, five trees per area, 12 fruit per tree, and two samples per fruit. A good, linear relationship was found between ln(ln(P0)) and ln(mean), where P0 was the probability of no mites in a sample. Although the data fit the binomial, confidence limits for mean arithmetic predictions increased dramatically as the proportion of samples infested increased. This indicated that binomial sampling using a tally threshold of zero may have less value for estimating mite densities when the proportion of samples infested is large. The effect of reducing sample size on estimating mean densities from proportions of samples infested was investigated using computersimulated data sets for 200, 160, 80, or 48 samples per 4 ha. For these reduced sampling plans, percentages of estimates that fell outside expected 95% confidence limits were generally similar, with 8.9 to 12.0% estimates above and 0.0 to 0.1% below expected limits. A good, linear relationship was found between ln(ln(P2)) and ln(mean), where P2 was the probability of two or fewer motile mites in a sample. However, increasing the tally threshold to 2 motile mites provided little improvement in estimates at large densities. Overall, binomial sampling for rust mites appeared to be a viable alternative to absolute counts of mites per sample for a grower using a low management threshold such as 2 or 3 motile mites per sample.
