Intelligent Pest Monitoring: a new paradigm for crop protection

Developing new pest control methods has often been the center of attention for many researchers and far less has been done to develop more effective detection and identification techniques. It is well known that several plant species emit a wide array of volatiles when infected with pathogens or attacked by pests. By using well-developed mechanical olfaction and other sensory technologies, we can follow some of these plant cues to locate crop problems before they become visible to the naked eye of human scouts. In this project, we suggest a novel approach to pest monitoring by shifting the attention from pests to plants. If interpreted correctly, plant-driven volatile chemical signals can provide more accurate information about the health of the plant. We focused on greenhouse vegetables and their major pests. We examined qualitative and quantitative differences among volatile chemicals emitted from clean and infested plants. After several screening experiments, we selected few indicator chemicals which were suitable representative of pest damage for each pest-plant group. We then sat up larger experiments inside research greenhouses and monitored the indicator volatile changes over time using a portable ultra-fast GC (zNose). We used the results of these experiments as training data sets to develop predictor functions using supervised learning techniques. The suitability of these functions then was tested by cross validation. We are also monitoring plants and collecting volatile samples and other environmental information from commercial greenhouses every week to verify the accuracy of predictor functions in real environments.