Investigation of different yellow sticky trap applications based on biotechnical control of pear pest Cacopsylla pyri L. (Hemiptera: Psyllidae) using artificial intelligence

This thesis research was conducted on the pest Cacopsylla pyri L. (Hemiptera: Psyllidae) in pear orchards in Örençay, Elazığ province between 2022-2023. The study was carried out using yellow sticky traps numbered 1023 and conventional traps to monitor this pest. During the first year, different trap hanging models (1, 3 and 4 traps per tree) were tried, the number of adults falling into the traps and the number of adults and nymphs in the tree were recorded daily and these were analyzed graphically according to time and directions. In addition, how these changes were related to abiotic factors (such as temperature, humidity, air pressure, wind speed) was investigated and regression models were applied for pest population estimates. According to the data obtained in the first year, it was found that traps numbered 1023 were more efficient than conventional traps, and traps placed in the south direction caught more adults compared to other directions. In the second year of research, traps numbered 1023 were placed in four different directions (north, south, east, west) based on the findings of the first year. The data obtained from the traps, the number of adults, the changes in nymphs and adults, and the numbers of natural enemies were monitored weekly. The study results of both years show that the number of adults of the pest decreased over time, and this was related to some environmental factors and the cyclical biology of the pest. In addition, a decrease in the number of adults was observed with the increase in temperature in July, which revealed the effect of environmental conditions. It was recorded that the number of nymphs and adults in the tree decreased in parallel with the increase in the number of adults falling into the traps. The data were statistically evaluated with Kruskal Wallis, t-test and ANOVA methods in the SPSS program. Various prediction models such as Elastic Net and Isotonic Regression were applied for pest prediction, and their accuracies were compared. Pace Regression was selected as the most appropriate model, and the Linear Regression model was also added in the second year study. Prediction models were created by considering environmental factors (such as temperature and humidity) and the effects of these models on the pest population were examined. In the second year, natural enemies were also monitored and the effects of these enemies on the pest population were evaluated. The results showed that the number of natural enemies changed according to environmental conditions. Studies revealed that temperature and humidity had significant effects on pest populations, and in studies conducted in 2023, it was observed that the number of adults, nymphs and impact-effect adults in all directions increased significantly compared to 2022. In the comparison of the two years, it was determined that the pest populations were higher in 2023. The most appropriate prediction models, Pace Regression and Linear Regression, show that environmental factors, especially temperature and humidity changes, affect the pest population. As a result of the research, it was determined that the pest population changed according to the directions in the prediction models applied in 2022, and in 2023; the pest population, especially for the south direction; It was found that the most important variables that positively affect the number of adults are the humidity at noon and the humidity at night, while the variables that negatively affect the number of adults are the humidity at morning and the temperature at night. In addition, it was observed that Chrysoperla carnea is the species most caught in yellow sticky traps among natural enemy species and that natural enemy species have a denser population in the south. This research emphasizes the role of environmental factors in estimating Cacopsylla pyri populations and provides important data on the use of artificial intelligence-based prediction models in combating pear psylla. These data are also important for pear IPM (Integrated Pest Management) studies by contributing to future pest population estimates.