Ghana is the world’s second largest producer of cocoa. However, farmers are not able to maximize yield potential due to pests and diseases that destroy as much as 60% of production annually. Mirids, ants, and hemipterans (mealybugs) are among the most detrimental cocoa pests in West Africa. Cocoa trees colonized by the mirids, ants, and mealybugs have reduced cocoa yields and are more vulnerable to pod damage, Black Pod and Cocoa Swollen Shoot Virus. In order to overcome insect-related cocoa decline, early insect damage detection, accurate pest monitoring, and sustainable management are crucial. However, the current status of cocoa pest monitoring and damage detection involves labor intensive, solutions requiring visual inspections, pheromones, and field traps. In Ghana, human-mediated pod damage detection is error prone and highly dependent on the knowledge of Extension officers. Once pests are detected, management is heavily reliant on toxic chemicals. While effective, pesticides are expensive, require protective equipment and applications are harmful to the environment and beneficial cocoa insects. Although safer pesticide alternatives exist, application information and necessary inputs may not be readily accessible by local farmers.
In order to recover insect-related yield losses and maintain the health of the cocoa ecosystem, it is important to invest in low-input technologies for identifying and suppressing pests populations minus the cost and toxicity of pyrethroids and neonicotinoids.
Can smartphones be fitted with artificial intelligence to automate cocoa insect identification and pest damages in the field?
The answer is yes! Our approach relies on similar algorithms used by Facebook’s face recognition technology to identify individuals based on unique facial features. In this case, a cocoa dataset was created with thousands of diverse images of pest and diseases. Each cocoa pest has a unique appearance and causes specific damage that can be easily identified and characterized by computer software. When presented with a cocoa image, machine learning software is trained to provide an accurate identification of pests, diseases, and pod health in the field. Preliminary results indicate that mirid damage and hemipterans (mealybugs) were detected with accuracy. This technology can potentially be fitted into smartphone apps that work offline. Automated insect identification and damage detection can be useful to Extension agents in the field whereby early and accurate pest identification and insect damage detection are vital for healthy cocoa yields.
Once pests are monitored and damage is detected, it is important to manage pests with sustainable methods.
High pressured soapy water, an affordable yet effective, pesticide alternative
I know what you are thinking, household liquid soap? So can I kill pests with the same bottle of liquid soap used to kill germs and clean dishes? The answer is Yes! Prior to the adoption of synthetic chemical sprays, the effective utility of household liquid soap for pest management, dates back to the 1800’s. When sprayed onto insects, soapy water blocks fatty acid synthesis and dissolves the insects’ outermost protective layer, killing the insects within seconds of contact. In 2016, high-pressure soapy water was applied to over 300 cocoa trees infested with ants, mirids, and mealybugs. Two tablespoons of liquid soap were diluted with one liter of water and applied directly to pest infestations, within seconds, all insects were exterminated. Over the course of two months, trees sprayed with soapy water contained less than 50 pests per tree, in comparison to control trees (that did not receive soapy water treatments) that were colonized by thousands of ants, mirids, and mealybugs which increased over time. High pressured soapy water has proven to be a cost-effective pesticide alternative and reduced pests populations while maintaining the health of the cocoa ecosystem. Our field research has shown that artificial intelligence and high pressured soap water are proven low-input technologies that have the potential to transform the future of cocoa farming.