Comparative resilience and precision of digitized optical counting using ImageJ during routine mosquito (Diptera: Culicidae) sample processing.

Abstract

Surveillance is integral for the targeted and effective function of integrated vector management. However, the scale of surveillance efforts can be prohibitive on manpower, given the large number of traps set, collected, processed, and enumerated. For many public health agencies, the sheer effort of weekly trapping, combined with the processing of numerous traps, is a major capacity challenge. To reduce employee fatigue and increase throughput, estimation methods are used in a diagnostic capacity to determine threshold numbers of mosquitoes (Diptera: Culicidae) for operational decision-making. Historically, volume and mass measures correlated to a known number of mosquitoes are the oldest and most widely used within mosquito control programs. Image processing methods using digital counting software, such as ImageJ, have not been tested rigorously in the context of high throughput usage experienced in mosquito operations. We stress-tested volume, mass, and image processing methods using sample calibrations from early in the year and applied them throughout a mosquito active season. We additionally tested resilience with samples that had been frozen, desiccated, old, or from an excessively large trap collection. Furthermore, we compared magnitudes of error after intentionally deviating from best practices. In all cases, mass and volume encountered significant errors. In contrast, the digitized-optical counting method was resilient to going long periods of use without recalibrating, handling different species compositions, and processing aged or damaged samples. If a program has limited logistical power, the aforementioned image-processing method confers the best balance of accuracy and expediency for time-sensitive workloads and efficient operational decision making.