Application of Generalized Linear Model Based Multiple Comparison Procedures to Greenhouse Pest Control Experiments

One of the biggest technical challenges in greenhouse production is the regulation of the greenhouse whitefly (Trialeurodes vaporariorum). Whiteflies are one of the most important pests in greenhouse cultivation worldwide. To achieve control of whiteflies using insecticides is difficult because of increasing resistance by whiteflies to chemicals. Within a project cluster the systemical effects of Neem Pellet (NP) and NeemAzal (TM) U (NU) treated in Gerbera jamesonii on Trialeurodes vaporariorum were investigated. For the statistical analysis, evaluated data comprises often counts. These counts are usually not normally distributed especially with decreasing sample size. A solution is the use of generalized linear models with the assumption of negative binomial distribution. Hypotheses of interest were tested using a multiple contrast test framework. Control of the family-wise error rate was achieved by assuming a multivariate normal distribution for linear combinations of parameters. A significant reduction of the whiteflies by neem treatments was detected. Comparisons of the two neem treatments showed no significant difference in their efficiency.