Wilkinson D. A., Lebon C., Wood T., Rosser G., Gouagna Louis-Clément. (2014). Straightforward multi-object video tracking for quantification of mosquito flight activity. Journal of Insect Physiology, 71, p. 114-121. ISSN 0022-1910.
Titre du document
Straightforward multi-object video tracking for quantification of mosquito flight activity
Wilkinson D. A., Lebon C., Wood T., Rosser G., Gouagna Louis-Clément
Source
Journal of Insect Physiology, 2014,
71, p. 114-121 ISSN 0022-1910
Mosquito flight activity has been studied using a variety of different methodologies, and largely concentrates on female mosquito activity as vectors of disease. Video recording using standard commercially available hardware has limited accuracy for the measurement of flight activity due to the lack of depth-perception in two-dimensional images, but multi-camera observation for three dimensional trajectory reconstructions remain challenging and inaccessible to the majority of researchers. Here, in silica simulations were used to quantify the limitations of two-dimensional flight observation. We observed that, under the simulated conditions, two dimensional observation of flight was more than 90% accurate for the determination of population flight speeds and thus that two dimensional imaging can be used to provide accurate estimates of mosquito population flight speeds, and to measure flight activity over long periods of time. We optimized single camera video imaging to study male Aedes albopictus mosquitoes over a 30 h time period, and tested two different multi-object tracking algorithms for their efficiency in flight tracking. A. Albopictus males were observed to be most active at the start of the day period (06h00-08h00) with the longest period of activity in the evening (15h00-18h00) and that a single mosquito will fly more than 600 m over the course of 24 h. No activity was observed during the night period (18h00-06h00). Simplistic tracking methodologies, executable on standard computational hardware, are sufficient to produce reliable data when video imaging is optimized under laboratory conditions. As this methodology does not require overly-expensive equipment, complex calibration of equipment or extensive knowledge of computer programming, the technology should be accessible to the majority of computer-literate researchers.
