Mosquitoes Use Vision to Associate Odor Plumes with Thermal Targets

Abstract

All moving animals, including flies, sharks, and humans, experience a dynamic sensory landscape that is a function of both their trajectory through space and the distribution of stimuli in the environment. This is particularly apparent for mosquitoes, which use a combination of olfactory, visual, and thermal cues to locate hosts. Mosquitoes are thought to detect suitable hosts by the presence of a sparse CO₂ plume, which they track by surging upwind and casting crosswind. Upon approach, local cues such as heat and skin volatiles help them identify a landing site. Recent evidence suggests that thermal attraction is gated by the presence of CO₂, although this conclusion was based experiments in which the actual flight trajectories of the animals were unknown and visual cues were not studied. Using a three-dimensional tracking system, we show that rather than gating heat sensing, the detection of CO₂ actually activates a strong attraction to visual features. This visual reflex guides the mosquitoes to potential hosts where they are close enough to detect thermal cues. By experimentally decoupling the olfactory, visual, and thermal cues, we show that the motor reactions to these stimuli are independently controlled. Given that humans become visible to mosquitoes at a distance of 5-15 m, visual cues play a critical intermediate role in host localization by coupling long-range plume tracking to behaviors that require short-range cues. Rather than direct neural coupling, the separate sensory-motor reflexes are linked as a result of the interaction between the animal's reactions and the spatial structure of the stimuli in the environment.

Figure 1. Wind tunnel, CO₂ plume, and example flight trajectories

(A) Wind tunnel used in our experiments. Color borders indicate top, side, and upwind views used in subsequent panels. (B) Heat map of a turbulent flow, particle diffusion model of the CO₂ plume based on 65 measurements in the wind tunnel, see Methods and Figure S1 for details. The white dot indicates a mosquito, drawn to scale. (C) Example flight trajectory in clean air. The two colored arrowheads show synchronized points across side and top views. The spacing between the points (33 Hz intervals) indicates the animal’s speed. (D) Example flight trajectory in the presence of a CO₂ plume, showing the mosquitoes’ stereotypical behavior exploring the high contrast object after sensing CO₂. In both c and d, an estimate of the animal’s instantaneous CO₂ (or control) experience is plotted below and color-coded within the trajectories using the scale in b. See Figure S1 for additional trajectories, and Supplemental Movie for animations.

Figure 2. CO₂ triggers mosquitoes to explore high contrast dark objects

(A) Heat map showing where female mosquitoes spent their time over a three hour period. The top panel shows a side view of the data. The bottom panel shows a top down view of the data over the altitude range indicated by the vertical pink line in the top panel. The right panel shows a photograph of the wind tunnel. (B) Same as a, but in the presence of a CO₂ plume. (C) Same as a, but in the presence of a CO₂ plume and with a black and a white visual object on the floor of the tunnel. (D) Same as a, but with male mosquitoes in clean air. We did not find any qualitative differences in male mosquitoes’ behavior in the presence of a CO₂ plume (not shown). (E) Relative flight activity, measured as the ratio of time mosquitoes spent flying in the presence of a CO₂ or clean air plume compared to their prior activity. (F) The ratio of the total time mosquitoes spent near the object divided by the total time they spent elsewhere for CO₂ and clean air conditions. Shading shows bootstrapped 95% confidence intervals of the mean. (G) Time elapsed between when mosquitoes left the plume (conservatively defined here as 401 ppm) and when they approached to within 3 cm of the object. (H) Example trajectories (top row, side view; bottom row, top down view) that contributed to the histogram shown in g, demonstrating the circuitous path many mosquitoes took from the plume to the object. Only the trajectory segments between plume exit (pink arrow) and object approach are shown. See Figure S2 for a description of plume tracking behavior.

Figure 3. Visual stimuli provide an intermediate cue, linking long range olfactory cues and short range heat sensing

(A) Photograph of the ITO coated glass pad. (B) Measurements of the thermal plume created by the heated glass pads at 0.5, 2.5, and 6.5 cm altitude, colored orange, purple, and black, respectively. (C) Photographs and thermal images of the stimuli in the wind tunnel. (D) Mean fraction of trajectories that entered an 8×8×4 cm³ volume above and downwind of either the left or right object (see f). Shading indicates 95% confidence intervals. The letters at the top indicate significantly different groups (Mann-Whitney u-test with Bonferoni correction, p=0.01). (E) Mean preference index for the test object vs. control object with 95% confidence intervals, statistics calculated as in d. (F) Sample trajectory entering one of the test volumes (green) used in d-e. The trajectory is color-coded red for the 2 seconds prior to when it entered the volume. (G) Spatial representation of preference index prior to when mosquitoes entered either test volume shown in f. For each trajectory, we selected the segments 2 seconds prior to when they entered either volume, in addition to the portions spent inside the volumes (red region of the trajectory shown in f). We then calculated the preference index for each 2×2 cm² rectangular region as the amount of time spent on the side of the wind tunnel of the test object compared to the control object, divided by their sum. We then calculated the mean preference index for each 2×2 cm² region across all trajectories, and its 95% confidence interval. Colors indicate preference index for regions where the 95% confidence interval was smaller than 0.5 (out of the total range of −1 to +1); the regions with higher uncertainty are shown in black. A blue or pink color that is more saturated than the arrows on the scale bar represent regions where the mosquitoes showed a statistically significant preference for one side or the other. The average approach trajectory for all the mosquitoes in each trial is shown as a magenta line. Because the average approach trajectories to the two objects were indistinguishable, this line shows the average approach of all trajectories for simplicity. The light green box shows a side view of the volumes shown in f. The colored arrows indicate the altitudes at which the temperature of the air was measured in b. The number of trajectories that approached the test (orange), control (blue), or both (black) objects is indicated in the top right of each panel.

Figure 4

Temperature downwind (0.4 m/s) of a human arm measured at two different ambient temperatures (orange and purple). Horizontal lines indicate the ambient temperatures.