Convolutional neural network analysis of Diaphorina citri (Hemiptera: Liviidae) vibrational communication signals for enhanced mating disruption

Authors: MANKIN, RICHARD; MCNEILL, SETH - Embry-Riddle University; MCNEILL, C - Embry-Riddle University

Submitted to: Journal of Economic Entomology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/13/2026
Publication Date: 6/19/2026
Citation: Mankin, R.W., Mcneill, S., Mcneill, C. 2026. Convolutional neural network analysis of Diaphorina citri (Hemiptera: Liviidae) vibrational communication signals for enhanced mating disruption. Journal of Economic Entomology. https://doi.org/10.1093/jee/toag154.
DOI: https://doi.org/10.1093/jee/toag154

Interpretive Summary: There is a need to restrict spraying of pesticides in Florida citrus groves to protect consumers of citrus products. One method disrupts mating of a major pest, the Asian citrus psyllid. Scientists at the Center for Medical, Agricultural, and Veterinary Entomology, Gainesville, FL and Embry Riddle University, Prescott AZ have developed methods that disrupt psyllid mating by interfering with their mating behaviors with transducers which produce vibrational signals that mimic psyllid vibrational communication duets. When multiple male-female pairs are duetting in large populations or when high levels of background vibrations occur, mating disruption is difficult to maintain, The improvement described in this report enhances the capability to distinguish the male-female signals from each other when multiple pairs are present on a tree. Better understanding of the interactions among multiple mating pairs can help improve the reliability of psyllid control methods.

Technical Abstract: Diaphorina citri Kuwayama (Hemiptera: Liviidae) is an invasive vector of bacteria which causes huanglongbing, a disease with economically devastating impacts to citrus production. Management of D. citri infestations is primarily through insecticides, but alternative control methods remain under consideration, including the co-opting and disruption of D. citri mating-duet vibrational communication signals. This study applies deep learning methods to identify important features of D. citri communication signals and filter out background noise so that male and female communications can be separately identified in noisy environments when multiple pairs are courting. The new methods yielded up to 96.7% accuracy in distinguishing male and female calls under standard cross-validation assessment procedures. These and similar deep learning methods have potential to facilitate better understanding of vibrational communication signals produced in various contexts by D. citri and other hemipteran pests, to further assist development of communication signal mimics that interfere with D. citri social behaviors and, in general, support further development of behaviorally based insect pest management.