Markov models of mite movements
Continuous time Markov models were used to analyse data from two bioassays to investigate the influence of beta-fraction, a by-product of hop processing, on the two-spotted spider mite. The models were fitted to aggregate counts of the numbers of live and dead mites on treated and untreated halves of discs cut from leaves of hop and French bean plants. Some of the rate parameters were time dependent. Although not all parameters could be estimated precisely, the analysis enabled the quantitative effects of treatment over time to be estimated with reasonable precision. The estimated treatment effects were largely insensitive to the assumed values of other parameters. The first bioassay showed a progressive initial response to increasing concentration of beta-fraction, although data at the intermediate concentration appeared anomalous. The second bioassay showed similar responses on hop and French bean leaves, with a stronger repellent effect on the lower leaf surface than on the upper surface.