ArticlePDF Available

The olfactory gating of visual preferences to human skin and visible spectra in mosquitoes

Authors:

Abstract and Figures

Mosquitoes track odors, locate hosts, and find mates visually. The color of a food resource, such as a flower or warm-blooded host, can be dominated by long wavelengths of the visible light spectrum (green to red for humans) and is likely important for object recognition and localization. However, little is known about the hues that attract mosquitoes or how odor affects mosquito visual search behaviors. We use a real-time 3D tracking system and wind tunnel that allows careful control of the olfactory and visual environment to quantify the behavior of more than 1.3 million mosquito trajectories. We find that CO2 induces a strong attraction to specific spectral bands, including those that humans perceive as cyan, orange, and red. Sensitivity to orange and red correlates with mosquitoes’ strong attraction to the color spectrum of human skin, which is dominated by these wavelengths. The attraction is eliminated by filtering the orange and red bands from the skin color spectrum and by introducing mutations targeting specific long-wavelength opsins or CO2 detection. Collectively, our results show that odor is critical for mosquitoes’ wavelength preferences and that the mosquito visual system is a promising target for inhibiting their attraction to human hosts.
Olfactory gating of mosquito color preference a The Ae. aegypti eye. Image courtesy of Alex Wild (with permission). b Spectral reflectance of behaviorally important objects for Ae. aegypti females: human skin (brown line); flower (Platanthera obtusata; green line) and within a small puddle filled with Ae. aegypti larvae (blue line). Lines are mean; shaded area around the mean is the ±sem (n = 6–10). c Wind tunnel system with real-time tracking system, odor and visual stimulation. d Heat map of the CO2 plume in the wind tunnel. e Example of individual trajectories (top: [x-,y-axes], bottom: side view [x-,z-axes]). The arrows represent the start of a trajectory; circles are the visual objects. f Heat (occupancy) maps showing the distribution of female mosquitoes without (left panels) and with CO2 delivery (right panels) while in presence of a white and a red objects (top), or white and green objects (bottom). Plots shows the top and side views of the tunnel working section. g Reflectance of the visual stimuli used in the experiments, as quantified using a spectrophotometer and calibrated with a white Spectralon standard. Different colored traces correspond to the different stimuli. h Relative flight activity between the different phases of the experiments (pre-, CO2 and post-CO2) for black and white circles (n = 12) and color and white circles (n = 51). There were no significant differences when the tested visual object was black vs. objects of different wavelengths (Kruskal–Wallis test, df = 1, Chi-sq = 0.01, P = 0.92), although for both groups CO2 significantly elevated the number of flying mosquitoes compared to the filtered Air treatment (P < 0.002). i Mean preference index for the test object vs. the control (white) object. There was a significant effect of color on the attraction to the tested object (Kruskal–Wallis test, Preference index ~ pair of visual stimuli used: df = 8, Chi-sq = 597.23, P < 0.0001). Several hues were significantly more attractive than the control, white object (one-sample two-tailed t-test: ***: P < 0.001). Boxplots are the mean (line) with 95% confidence interval (shaded area) (n = 25,529; 17,729; 53,786; 23,694; 34,343; 31,037; 32,257; 24,774; 42,595; 20,929; and 48,198 mosquito trajectories for the all-white, all-black, black-and-white, Bv-T2, Bw-, Gw-T1, Gc-, YGc-, Yw-, O- and R-Hue treatments, respectively).
… 
The contribution of orange-red wavelengths in attraction to faux human skin a Spectral reflectance of human skin and faux skin used in behavioral experiments. Lines are the mean and shaded area is the ±sem (n = 6). b Ultra-thin optical filters (450 nm, 600 nm, and 700 nm) attenuated discrete bands in the object’s reflected spectrum. c, d Occupancy maps of the mosquito’s distribution around the visual objects during exposure to CO2. During CO2, mosquitoes were significantly attracted to the faux skin color compared to the white (control) object (c). However, the optical filter attenuating the 550–630 nm band reduced the number of mosquitoes investigating the faux skin color (d). e, f Mosquitoes significantly preferred the faux human skin colors (e), although optical filters in the yellow to red wavelengths significantly decreased the attractiveness of the visual object (f). Boxplots are the mean (line) with 95% confidence interval (shaded area); letters denote statistically significant differences between groups. g Mosquito lines deficient in long-wavelength opsins (opsin1 and opsin2), or unable to detect CO2 (Gr3 mutant), were tested in their attraction of human skin color. h Mean preference indices for the Gr3 mutants (blue) and opsin mutants (orange). All mosquito lines showed similar preferences to the white and skin color visual objects during exposure to filtered air (Kruskal–Wallis test: df = 3, Chi-sq = 1.68, P = 0.64). However, during CO2 the lines were significantly different from one another in their visual preferences (Kruskal–Wallis test with multiple comparisons: df = 3, Chi-sq = 96.01, P < 0.001): only the heterozygote (Gr3−/+) and wild-type (LVP) lines showed significant attraction to the skin color (one-sample t-test: P < 0.001), whereas the opsin double mutant line (op-1/op-2) and the Gr3−/− mutants showed no attraction (one-sample two-tailed t-test: P > 0.31). Boxplots are the mean (line) with 95% confidence interval (shaded area) (n = 13,999; 15,035; 14,016; 48,624; 19,284; 20,713; 40,690; 26,135; 39,649; 16,208; 3,550; 9,277; 13,799; 10,948; and 5,679 mosquito trajectories for the Y02, Vile 45, Y10, R10, IR filter, coverslip, 450 nm filter, 600 nm filter, 700 nm filter, wt, Gr3−/−, Gr3−/+, op1-R, op-2G, and op1-R,op2-G treatments, respectively).
… 
The importance of long wavelengths in attraction to human skin a Cage assay with real-time tracking system, odor, and visual stimulation through two windows on the front of the cage. b Example of individual trajectories (top: skin and control (white glove), bottom: skin+filters (550–700 nm), and control). c Occupancy maps showing the distribution of female mosquitoes during CO2 stimulation while in presence of the skin and control (top), and the skin+filters (550–700 nm) and control (bottom). d Relative flight activity between the different phases of the experiments (pre-, CO2, and post-CO2) for the skin and skin+filters treatments (n = 6 trials/treatment). There was no significant difference in the relative activity during the CO2 phase between the skin and skin+600-nm filter treatments (Kruskal–Wallis test, df = 1, Chi-sq = 0.004, P = 0.96). e The percentage of mosquitoes visiting the windows over the duration of the experiment. Few mosquitoes investigated the windows before the CO2 exposure. However, exposure to CO2 significantly increased the numbers of mosquitoes visiting the windows relative to the pre-CO2 period (Kruskal–Wallis test with multiple comparisons: df = 5, Chi-sq. = 277.85, P < 0.0001), although during CO2 there were no significant differences in the total number of mosquitoes investigating the windows between treatment groups (Kruskal–Wallis test with multiple comparisons: P > 0.57). Lines are the means and shaded areas the ±sem. f Mean preference index for the different treatment groups (white glove vs. white glove, skin vs. white glove, and skin + filter (550–700 nm) vs. white globe). Boxplots area the mean (line) with 95% confidence interval (shaded area). Different letters denote statistically significant differences between groups (Kruskal–Wallis test with multiple comparisons, P < 0.01). (n = 13,597 for the skin treatment group; n = 9502 for the for the skin + filters treatment group; and n = 9368 for the control group).
… 
This content is subject to copyright. Terms and conditions apply.
ARTICLE
The olfactory gating of visual preferences to human
skin and visible spectra in mosquitoes
Diego Alonso San Alberto1,4, Claire Rusch1,4, Yinpeng Zhan2, Andrew D. Straw3, Craig Montell2&
Jeffrey A. Riffell 1
Mosquitoes track odors, locate hosts, and nd mates visually. The color of a food resource,
such as a ower or warm-blooded host, can be dominated by long wavelengths of the visible
light spectrum (green to red for humans) and is likely important for object recognition and
localization. However, little is known about the hues that attract mosquitoes or how odor
affects mosquito visual search behaviors. We use a real-time 3D tracking system and wind
tunnel that allows careful control of the olfactory and visual environment to quantify the
behavior of more than 1.3 million mosquito trajectories. We nd that CO
2
induces a strong
attraction to specic spectral bands, including those that humans perceive as cyan, orange, and
red. Sensitivity to orange and red correlates with mosquitoesstrong attraction to the color
spectrum of human skin, which is dominated by these wavelengths. The attraction is eliminated
by ltering the orange and red bands from the skin color spectrum and by introducing
mutations targeting specic long-wavelength opsins or CO
2
detection. Collectively, our results
show that odor is critical for mosquitoeswavelength preferences and that the mosquito visual
system is a promising target for inhibiting their attraction to human hosts.
https://doi.org/10.1038/s41467-022-28195-x OPEN
1Department of Biology, University of Washington, Seattle, WA 98195, USA. 2University of California, Santa Barbara, Santa Barbara, CA 93106, USA.
3Institute of Biology I & Bernstein Center Freibug, Albert-Ludwigs-Univesität Freiburg, Freiburg im Breisgau, Germany.
4
These authors contributed equally:
Diego Alonso San Alberto, Claire Rusch. email: jriffell@uw.edu
NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications 1
1234567890():,;
Content courtesy of Springer Nature, terms of use apply. Rights reserved
The behavioral preference of insects for certain bands in the
visible light spectrum plays a profound role in structuring
ecological communities by mediating processes such as
plant-insect/predator-prey interactions and disease transmission13.
For biting insects, such as mosquitoes, tsetse ies, and kissing bugs,
vision plays an essential role in various behaviors, including ight
control, object tracking for host- or nectar-nding, and locating
oviposition sites4. The visual stimuli that mediate these behaviors
are integrally tied to other host-related cues, such as scent and heat.
For instance, when combined with an odor lure, tsetse ies are
highly attracted to what humans perceive as blue color5,6,and
kissing bugs prefer visual objects only when also associated with
odors7. Visually guided mosquito behaviors are also thought to play
a role in host attraction810.Ithaslongbeenknownthatmosqui-
toes are attracted to dark, high-contrast objects9,11, which has led to
the development of black traps10.Forhigh-contrastvisualstimuli,
recent work has shown that certain odors stimulate visual search
behaviors in Aedes aegypti mosquitoes. This species is not attracted
to black objects in the absence of CO
2
, but after encountering a CO
2
plume, they become highly attracted to such objects11. Other cues
(heat, water vapor, skin volatiles) mediate behaviors such as landing
and biting1113.
Despite the potential importance of color in mediating mos-
quito biting behaviors, surprisingly, details regarding other
wavelengths that attract mosquitoes or how odors sensitize that
attraction remain unclear. The visual spectra of important
resources can be diverse and dominated by short and medium
wavelengths (e.g., owers or oviposition sites) or long wave-
lengths (e.g., human skin) (Fig. 1a, b). Despite interest in devel-
oping traps and lures that exploit mosquito spectral preferences,
only a few studies have examined these preferences, and the
results of those studies have been contradictory. For instance,
studies of Ae. aegypti have shown no difference in spectral pre-
ference in the 450600 nm wavelength range14,15. By contrast,
other studies have demonstrated specic preferences but for
different wavelength bands: Ae. aegypti mosquitoes were attracted
to blue in one study16 and only to greenyellow in other
studies17,18. Other studies have shown that mosquitoes some-
times prefer red14,17,19, although it is thought that mosquitoes
lack opsin receptors sensitive to these wavelengths. Because many
mosquito species are attracted to dark visual objects, responses to
long wavelengths (red to human observers) may represent
achromatic responses from visual channels that are sensitive to
medium-wavelengths and therefore are perceived as dark gray
or black when presented against a light-colored background.
Nevertheless, these prior studies did not characterize the actual
ight trajectories of the mosquitoes, nor control for the change in
behavioral state associated with the smell of a host. Accurate
control of both a visual objectsreectance and its contrast with
the background is required to determine whether mosquitoes are
attracted to specic wavelengths.
Here in this study, we use a large wind tunnel and a computer
vision system to close knowledge gaps regarding mosquito visual
and olfactory responses by examining Ae. aegypti free-ight
responses to objects of different wavelengths, with and without
thepresenceofCO
2
.Aedes aegypti provides an excellent model for
studies aimed at elucidating spectral preferences and determining
how these preferences are modulated by odor. Aedes aegypti, which
are active during the dawn and dusk periods20, have 10 rhodopsins,
5 of which are expressed in the adult eye21. Little is known about
opsin tuning, although they are orthologs of medium-wavelength
sensitive opsins (green), and previous electroretinogram (ERG)
studies suggested that Ae. aegypti is sensitive to medium-long
wavelengths in the greenyellow spectrum22,23.Weshowthat
when encountering odor, mosquitoes become particularly attracted
to hues that are dominant in human skin. We also demonstrate
that knockout of either the olfactory channel that gates visual
attraction or the opsins that allow detection of objects that reect
long wavelengths eliminates attraction to skin tones.
Results
Olfactory gating of spectral preferences of Ae. aegypti mos-
quitoes. Examining olfactory and visual search behaviors in
mosquitoes often requires simulating conditions in which the
statistics of the stimuli (e.g., intensity, duration) and resulting
mosquito behavior are as natural as possible. We therefore
examined Ae. aegypti behavior in a large wind tunnel spanning
450 mosquito body lengths and equipped with a 16-camera, real-
time tracking system for monitoring and quantifying mosquito
behaviors11,24. A checkerboard pattern was projected on the
bottom (oor) of the wind tunnel, and a low-contrast gray hor-
izon was projected on each side of the tunnel to provide optic
ow (Fig. 1c). Similar to our previous assays, we placed two
identically sized circles (3 cm diameter) on the oor of the tunnel
in the upwind area of the working section, 18 cm apart and 33 cm
from the odor source (Fig. 1c). In each experimental trial, 50
mated Ae. aegypti females were co-released into the tunnel, and
their trajectories were recorded over a 3 h period (1.3 million total
trajectories were recorded, with an average trajectory duration of
3 s). Simultaneous release of the mosquitoes provided an efcient
method to examine their olfactory-visual responses, and was not
statistically different from when the mosquitoes were released
singly (see Materials and MethodsStatistical analyses for
details). The tunnel was lled with ltered air for 1 h, after which
aCO
2
plume (95% ltered air, 5% CO
2
) located 33 cm away and
separate from the visual objects was introduced into the tunnel
and left for 1 h (Fig. 1d). Measurements of the plume in the
working section showed an exponential decay, typical of turbu-
lent diffusion, with a concentration of ~1500 ppm ~30 cm from
the odor source. In the last hour of the experiment, only ltered
air was released into the wind tunnel.
During exposure to ltered air, the mosquitoes exhibited
random behavior to the odor source and visual objects, and they
spent much of their time exploring the ceiling and walls of the
tunnel and rarely investigated the visual objects (Fig. 1e). By
contrast, upon exposure to the CO
2
plume, the number of ying
mosquitoes more than doubled (Fig. 1f, h; Wilcoxon signed-rank
test, number of trajectories in Air vs. CO
2
,P< 0.002). During this
time, the mosquitoes exhibited odor-tracking behavior, spending
most of the time in the working sections central area with
signicantly elevated ight velocities (Fig. 1f, s1; KruskalWallis
test: df =2, Chi-square =597.23, P< 0.0001). The CO
2
also
triggered an attraction to visual objects. Here, we dene attraction
as the amount of time a trajectory spends around an object
relative to the evenly reecting control (white to the human
observer). The Ae. aegypti mosquitoes showed no interest in the
objects during the ltered air treatment (only 14% of mosquitoes
investigated), but during CO
2
release, the percentage and number
of mosquitoes investigating the visual objects increased signi-
cantly (21%; paired Studentsttest: P=0.002) (Supplementary
Fig. S1). After the plume was stopped, the attraction to the visual
objects ceased (Fig. 1e, Supplementary Fig. S1e; Wilcoxon signed-
rank test, CO
2
vs. Post-CO
2
:P< 0.001).
To ensure that Ae. aegypti mosquito visual preference
behaviors were in response to discrete wavelength bands rather
than object contrast, we employed visual stimuli in the range
430660 nm (violet to red to a human observer), with each visual
stimulus having the same approximate contrast with the back-
ground (Fig. 1g). While investigating the visual objects, the
mosquitoes would y upwind and hover immediately downwind
of a visual object, at ~35 cm, while exhibiting brief excursions
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x
2NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
before returning to the objects (Fig. 1e). During CO
2
release, the
same total number of mosquitoes was recruited to the visual
objects of differing dominant wavelengths (Supplementary
Fig. S1e). Relative to the evenly reecting object (white to the
human observer) that served as a control, however, the
mosquitoes preferred certain dominant wavelengths, such as
496 nm and longer wavelengths 590 nm as demonstrated by
their focused clustering around the object (Fig. 1e). By contrast,
other dominant wavelengths (437 nm, 452 nm, 510 nm, and
520 nm; appearing violet, blue, green and greenyellow to the
human observer, respectively) elicited no attraction responses
compared to the evenly reecting control (Fig. 1e, i). Across all
dominant wavelengths, CO
2
had a strong effect on ight velocity
and duration, but there were no signicant differences between
treatments (Supplementary Fig. S1b, d; KruskalWallis test: Chi-
square < 16.82, P> 0.11), demonstrating that the presence of CO
2
c
400 500 600 700
50
100
Reflectance (%)
b
puddle
host
flower
0
Wavelength (nm)
a
0.08
Occupancy (%)
0
2500
CO
2
(ppm)
400
Clean air
d
e
f
top view
side view
CO2
Clean air CO2
top view
side view
Preference Index
***
437 452 496 510 520 590 600 660 __
control:
test:
peak (nm):
Black Wavelengths
-pre CO2
-post
0
1
2
3
Flight Activity
400 500 600
0
Wavelength (nm)
50
100
Reflectance (%)
g
h
i
700
0.5
0
-0.1
***
***
***
*** ***
__ __
***
***
-pre CO2
-post
2 m
0.6 m
0.6 m
3D tracking
cameras
CO source
2
Visual
targets
top view
top
view
side
view
20 cm
white
590nm
600nm
660nm
437nm
520nm
510nm
452nm
496nm
black
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x ARTICLE
NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications 3
Content courtesy of Springer Nature, terms of use apply. Rights reserved
is necessary for attraction to specic bands of the visual spectrum,
and that object attraction did not result from higher ight velocity
increasing the probability of a mosquito randomly encountering a
visual object.
To further investigate Ae. aegypti mosquito spectral prefer-
ences, a preference index value (dened as the time spent
investigating a spectral object minus the time investigating the
evenly reecting control, divided by the sum of the times spent
investigating both objects) was calculated for each mosquito that
investigated a visual object. The dominant wavelengths differed
signicantly in terms of mosquito preference (Fig. 1i,
KruskalWallis test: df =8, Chi-square =597.23, P< 0.0001);
several dominant wavelengths were more attractive to mosquitoes
than the evenly reecting object that served as a non-attracting
control (Fig. 1i, one-sample t-test: P< 0.001). As the dominant
wavelengths transitioned from 510 to 660 nm, the attractiveness
of the object also increased. For instance, 600 nm and 660 nm
visual objects were strongly preferred by female mosquitoes
(multiple comparison KruskalWallis test: P< 0.05), whereas
510 nm, 452 nm, and 437 nm objects (green, blue and violet to the
human observer, respectively) were not more attractive than the
control object (multiple comparisons KruskalWallis test:
P> 0.05). However, mosquitoes were not strictly attracted to
the longest wavelengths, as they were also signicantly attracted
to objects with a dominant wavelength at 496 nm (cyan to the
human observer). As a control to test for the effect of visual object
attraction relative to other regions of the working section, we
examined the preference between the evenly reecting (control)
object and a randomly selected volume in the wind tunnel.
Compared with the randomly selected volume, female mosqui-
toes investigated the evenly reecting control object signicantly
more in presence of CO
2
(preference index =0.53 ± 0.03
(mean ± sem), one-sample t-test: P< 0.001).
Behavioral preferences for orange-red wavelengths reected
from human skin. Across all skin tones and differences in pig-
mentation, human skin is dominated in the long-wavelength
range (590660 nm) (Fig. 2a)25, but it is unclear which bands in
the human skin spectrum are most attractive to Ae. aegypti
mosquitoes. To examine whether Ae. aegypti exhibit different
preferences for certain spectral bands reected by human skin, we
rst utilized color cards designed for cosmetics purposes to match
human skin tones (Pantone SkinTone Guide) (Fig. 2a). Beha-
vioral experiments were performed in the wind tunnel to indi-
vidually test various faux skin tones that ranged from light to dark
(Y02, Y10, R10, and an unpleasant orange shade typical of
individuals using cheap tanning lotion [which we designated vile
45]) using the evenly reecting object (white to the human
observer) as a control. Similar to the previous experiments, only
during CO
2
release did mosquitoes become highly attracted to
skin tones (Fig. 2c, e; KruskalWallis test: df =8, Chi-square =
184.37, P< 0.001), exhibiting no behavior characteristic of
attraction before exposure to CO
2
(paired t-test: P=0.09).
Moreover, the mosquitoes exhibited the same level of attraction
to each skin tone, with the skin tones being not signicantly
different from one another (Fig. 2c, e; KruskalWallis test with
multiple comparisons: P> 0.05).
To determine which region of the human skin visual spectrum
is most attractive to Ae. aegypti, we overlaid the R10 skin color
card with optical lters to attenuate discrete bands (Fig. 2b).
Whereas the 450 nm optical lter had no signicant effect on
behavioral attraction to the skin tone compared with the positive
controls (Fig. 2f; KruskalWallis test with multiple comparisons:
P> 0.58), lters blocking longer wavelengths (550700 nm)
reduced the attractiveness of the visual object (P< 0.05). In
particular, application of the 600 nm lter was associated with a
300% reduction in attraction compared with the positive controls
(Fig. 2cf). Importantly, results for controls consisting of an
overlaid infrared (IR) lter or a clear nylon coverslip did not
differ signicantly from the unmanipulated skin tone
(KruskalWallis test with multiple comparisons: P> 0.33). To
further examine the importance of visual and olfactory integra-
tion in controlling Ae. aegypti visual preferences, we examined
single and double mutants of the long-wavelength photoreceptors
opsin-1 and opsin-2 and a line with a mutation in the Gr3
receptor14, which transduces CO
2
signals (Fig. 2g). Whereas the
Gr3-heterozygote, opsin-1 and opsin-2 single mutants, and wild-
type control mosquitoes were signicantly attracted to the skin
tone during CO
2
exposure (one-sample t-test: P< 0.001), neither
the opsin-1,opsin-2 double mutant nor Gr3 mutant were attracted
to the skin tones (Fig. 2h; one-sample t-test: P> 0.31).
The behavioral preference to the long wavelengths in the skin
color cards may not reect mosquito behaviors to the hues reected
fromhumanskin.Toexaminethisfurther,wetestedAe. aegypti
mosquitoes in a smaller opaque cage (45 cm × 30 cm × 30 cm)
wheretheywereexposedtotwosmallwindows(16cm
2)(Fig.3a, b).
The cage and its environment were constructed to prevent
uncontrolled contamination from thermal or olfactory cues from
the human volunteer, and the windows being made from clear, heat
absorptive glass to block the radiant heat from the skin
(Supplementary Fig. S2). Subsequent testing showed no contamina-
tion during the experimental trials. The back of a hand was
displayed in one window, and the back of a heat-protective white
Fig. 1 Olfactory gating of mosquito color preference. a The Ae. aegypti eye. Image courtesy of Alex Wild (with permission). bSpectral reectance of
behaviorally important objects for Ae. aegypti females: human skin (brown line); ower (Platanthera obtusata; green line) and within a small puddle lled
with Ae. aegypti larvae (blue line). Lines are mean; shaded area around the mean is the ±sem (n=610). cWind tunnel system with real-time tracking
system, odor and visual stimulation. dHeat map of the CO
2
plume in the wind tunnel. eExample of individual trajectories (top: [x-,y-axes], bottom: side
view [x-,z-axes]). The arrows represent the start of a trajectory; circles are the visual objects. fHeat (occupancy) maps showing the distribution of female
mosquitoes without (left panels) and with CO
2
delivery (right panels) while in presence of a white and a red objects (top), or white and green objects
(bottom). Plots shows the top and side views of the tunnel working section. gReectance of the visual stimuli used in the experiments, as quantied using
a spectrophotometer and calibrated with a white Spectralon standard. Different colored traces correspond to the different stimuli. hRelative ight activity
between the different phases of the experiments (pre-, CO
2
and post-CO
2
) for black and white circles (n=12) and color and white circles (n=51). There
were no signicant differences when the tested visual object was black vs. objects of different wavelengths (KruskalWallis test, df =1, Chi-sq =0.01,
P=0.92), although for both groups CO
2
signicantly elevated the number of ying mosquitoes compared to the ltered Air treatment (P< 0.002). iMean
preference index for the test object vs. the control (white) object. There was a signicant effect of color on the attraction to the tested object
(KruskalWallis test, Preference index ~ pair of visual stimuli used: df =8, Chi-sq =597.23, P< 0.0001). Several hues were signicantly more attractive
than the control, white object (one-sample two-tailed t-test: ***: P< 0.001). Boxplots are the mean (line) with 95% condence interval (shaded area)
(n=25,529; 17,729; 53,786; 23,694; 34,343; 31,037; 32,257; 24,774; 42,595; 20,929; and 48,198 mosquito trajectories for the all-white, all-black, black-
and-white, Bv-T2, Bw-, Gw-T1, Gc-, YGc-, Yw-, O- and R-Hue treatments, respectively).
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x
4NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
glove was displayed in the other window (as an internal control).
Similartotheassaysinthewindtunnel,wefoundthatmosquitoes
were highly activated by CO
2
(Fig. 3ce), and this increased their
visual attraction to visual stimuli, including skin (Fig. 3b, c, f;
KruskalWallis test: df =2, Chi-square =84.04, P< 0.001). Mos-
quitoes showed no preference during control experiments with two
white gloves displayed in the window, but signicantly preferred
skin (Fig. 3f; KruskalWallis test with multiple comparisons:
P< 0.001). However, when optical lters were placed over the
window, blocking the longer wavelengths (550700 nm), the
attraction was signicantly reduced (KruskalWallis test with
multiple comparisons: P< 0.001) and not signicantly different
from the negative control (Fig. 3f; KruskalWallis test with multiple
comparisons: P=0.34). Collectively, these results demonstrate that
the long-wavelength band of the visual spectrum plays an important
role in determining mosquito attraction to skin color. In addition,
knockout of either visual or olfactory detection receptors suppresses
mosquito visual attraction to long-wavelength host cues.
Spectral sensitivity of the Aedes eye. The preference of Ae.
aegypti for long wavelengths in the orange-red band motivated us
to examine the sensitivity of the retina by recording ERGs that
extracellularly measure the summed responses of retinal cells to
visual stimuli (Fig. 4a). In the rst series of experiments, a moving
bar of differing dominant wavelengths (blue [peak 451 nm], green
[537 nm], or red [640 nm], all at the same intensity; 18° wide at
30°/s clockwise) was projected on a black background while
conducting the ERG recordings (Fig. 4b). When the moving bar
reached the mosquitos visual eld, the ERG exhibited a negative
response that quickly returned to baseline after the bar moved
past the mosquitoseld of view (Fig. 4b). A signicant difference
in wavelength-evoked responses was observed (KruskalWallis
test: df =3, 40.03, P< 0.001), with the 451 nm and 537 nm bars
eliciting the strongest responses (Fig. 4b). Although responses to
the 640 nm bar were signicantly weaker, this dominant wave-
length still elicited ERG responses that were signicantly higher
than the baseline and those of the no-stimulus controls (Wil-
coxon signed-rank test: P< 0.001).
To further characterize Ae. aegypti spectral sensitivity, we used
a scanning monochromator to examine ERG responses across the
near ultraviolet (UV) to far-red wavelength range (350750 nm).
Ae. aegypti exhibited the highest sensitivity to short (410 nm,
3.2 mV) and medium wavelengths (520 nm, 2.8 mV) (Fig. 4c, d).
Strong ERG responses (0.471.27 mV) were still noted in the
medium-long wavelengths (>590 nm), although at >700 nm, the
responses decreased to approximately 0.25 mV, which was still
signicantly higher than the baseline control (Wilcoxon signed-
rank test, P< 0.03).
The role of visual contrast in determining Ae. aegypti pre-
ferences. Mosquitoes are very sensitive to detecting dark objects
that contrast highly with the background8,11. In the above
experiments, we kept the total object contrast (400700 nm) with
the background approximately the same, but the Ae. aegypti
preference for long-wavelength objects and skin tones motivated
0
20
40
60
80
400 500 600 700 400 500 600 700
Preference Index
0.5
0
Y02
+450nm filter
+600nm filter
+700nm filter
Reflectance (%)
450 nm
filter
600 nm
filter
700 nm
filter
Host skin
Faux skin hue
a,b
ab
c
Wavelength (nm)
a
c
b,c
+IR filter
Vile 45
Y10
R10
ns
top view
side view
Occupancy (%)
CO | R10
2 CO | R10+600nm filter
2
d
ef
a
+coverslip
X
CO2
XGr3
mutant
op-1,2
mutant
g
Preference Index
0.5
(-|-)
Gr3
(-|+)
Gr3
RG
op-1 ,op-2
0
-0.2
R
op-1
G
op-2
wt
h
0.08
0
Fig. 2 The contribution of orange-red wavelengths in attraction to faux
human skin. a Spectral reectance of human skin and faux skin used in
behavioral experiments. Lines are the mean and shaded area is the ±sem
(n=6). bUltra-thin optical lters (450nm, 600nm, and700nm)attenuated
discrete bands in the objectsreected spectrum. c,dOccupancy maps of the
mosquitos distribution around the visual objects during exposure to CO
2
.
During CO
2
, mosquitoes were signicantly attracted to the faux skin color
compared to the white (control) object (c). However, the optical lter
attenuating the 550630 nm band reduced the number of mosquitoes
investigating the faux skin color (d). e,fMosquitoes signicantly preferred the
faux human skin colors (e), although optical lters in the yellow to red
wavelengths signicantly decreased the attractiveness of the visual object (f).
Boxplots are the mean (line) with 95% condence interval (shaded area);
letters denote statistically signicant differences between groups. gMosquito
lines decient in long-wavelength opsins (opsin1 and opsin2), or unable to
detect CO
2
(Gr3 mutant), were tested in their attraction of human skin color.
hMean preference indices for the Gr3 mutants (blue) and opsin mutants
(orange). All mosquito lines showed similar preferences to the white and skin
color visual objects during exposure to ltered air (KruskalWallis test: df =3,
Chi-sq =1.68, P=0.64). However, during CO
2
the lines were signicantly
different from one another in their visual preferences (KruskalWallis test with
multiple comparisons: df =3, Chi-sq =96.01, P<0.001): only the
heterozygote (Gr3/+) and wild-type (LVP) lines showed signicant attraction
to the skin color (one-sample t-test: P< 0.001), whereas the opsin double
mutant line (op-1/op-2)andtheGr 3/mutants showed no attraction (one-
sample two-tailed t-test: P> 0.31). Boxplots are the mean (line) with 95%
condence interval (shaded area) (n=13,999; 15,035; 14,016; 48,624; 19,284;
20,713;40,690;26,135;39,649;16,208;3,550;9,277;13,799;10,948;and
5,679 mosquito trajectories for the Y02, Vile 45, Y10, R10, IR lter, coverslip,
450 nm lter, 600 nm lter, 700 nm lter, wt, Gr3/
,Gr3
/+,op1-R,op-2G,
and op1-R,op2-G treatments, respectively).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x ARTICLE
NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications 5
Content courtesy of Springer Nature, terms of use apply. Rights reserved
us to evaluate whether these responses were due to contrast alone
(calculated as the Weber contrast, or the difference in spectral
energy reected by an object and the background, divided by the
sum of the two) or whether mosquitoes can discriminate long-
wavelength objects >590 nm independently of intensity. As the
rst step in determining how features of a visual stimulus impact
mosquito visual preference, gray objects that contrasted differ-
ently with the background were tested against the evenly
reecting control (Fig. 5, Weber Contrasts: 0.280.02). Similar
to the above results and across all tested stimuli, the presence of
CO
2
increased mosquito ight activity and the number of visits to
the visual objects (Fig. 5c, Wilcoxon signed-rank test, Air vs. CO
2
:
P< 0.001). Female mosquitoes exhibited signicantly greater
attraction to the majority of the gray visual objects than the
control object (Fig. 5d, one-sample t-test: P< 0.001). However,
when exposed to the lightest gray object, which closely approxi-
mated the background and the evenly reecting control, mos-
quitoes showed no preference for either object (Fig. 5d; Students
ttest: P=0.33). Overall, object darkness and contrast with the
lighter background was signicantly related to mosquito pre-
ference, with mosquitoes investigating and preferring darker
objects (Fig. 5d; KruskalWallis test: df =5, Chi-square =634.16,
P< 0.001). Although Ae. aegypti showed a distinct preference for
darker objects, mosquito ight velocity and duration did not
signicantly differ across treatments (KruskalWallis test: Chi-
square > 7.40, df =5, P> 0.055).
Effects of contrast and spectral discrimination on Aedes pre-
ferences. Given the difference between high retinal sensitivity to
medium-long (green) wavelengths, lack of attraction to those
objects, and the relatively low ERG sensitivity to long (orange/
red) wavelengths but strong behavioral attraction to those objects,
we asked the following question: what role does the darkness of
the object (i.e., its contrast) have on behavioral preferences vs. the
objects dominant wavelength? Although we lacked the photo-
receptor tunings that would allow us to normalize the inputs
between spectral channels, we experimentally manipulated object
contrast to equalize the perceptual contrast of the objects to the
mosquitoes and then examined the mosquitos ability to dis-
criminate between different bands of the visual spectrum.
We rst performed behavioral experiments in which we altered
the darkness of the gray objects to determine the contrast that
matched the attraction shown to the 660 nm object (red to human
Skin Skin+filters
-pre CO2
-post
0
2
4
Flight Activity
***
***
-pre CO2
-post 0 6 12 18 24
0
50
100
Time (min.)
0
0.35
0.70 a
b
b
Mosquito Visits (%)
Preference Index
test:
control:
white glove skin skin+filters
3D tracking
cameras
windows
CO | Skin
2
front view
CO | Skin + 600nm filters
2
0.08
Occupancy (%)
0
ac
def
front view
b
skin
filters
control
control
filters
skin
control
Fig. 3 The importance of long wavelengths in attraction to human skin. a Cage assay with real-time tracking system, odor, and visual stimulation
through two windows on the front of the cage. bExample of individual trajectories (top: skin and control (white glove), bottom: skin+lters
(550700 nm), and control). cOccupancy maps showing the distribution of female mosquitoes during CO
2
stimulation while in presence of the skin
and control (top), and the skin+lters (550700 nm) and control (bottom). dRelative ight activity between the different phases of the experiments
(pre-, CO
2,
and post-CO
2
) for the skin and skin+lters treatments (n=6 trials/treatment). There was no signicant difference in the relative activity
during the CO
2
phase between the skin and skin+600-nm lter treatments (KruskalWallis test, df =1, Chi-sq =0.004, P=0.96). eThe percentage of
mosquitoes visiting the windows over the duration of the experiment. Few mosquitoes investigated the windows before the CO
2
exposure. However,
exposure to CO
2
signicantly increased the numbers of mosquitoes visiting the windows relative to the pre-CO
2
period (KruskalWallis test with
multiple comparisons: df =5, Chi-sq. =277.85, P< 0.0001), although during CO
2
there were no signicant differences in the total number of
mosquitoes investigating the windows between treatment groups (KruskalWallis test with multiple comparisons: P> 0.57). Lines are the means and
shaded areas the ±sem. fMean preference index for the different treatment groups (white glove vs. white glove, skin vs. white glove, and skin +lter
(550700 nm) vs. white globe). Boxplots area the mean (line) with 95% condence interval (shaded area). Different letters denote statistically
signicant differences between groups (KruskalWallis test with multiple comparisons, P<0.01).(n=13,597 for the skin treatment group; n=9502 for
the for the skin +lters treatment group; and n=9368 for the control group).
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x
6NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
observer)(Fig. 6a, b). Across a range of gray contrast levels, the
660 nm object was signicantly more attractive to Ae. aegypti
than the gray objects (Fig. 6a, c, e; KruskalWallis test: Chi-
square =149.75, df =6, P< 0.0001). As the gray objects became
darker, however, they became more attractive to the mosquitoes,
and the strong preference for the 660 nm object decreased
(Fig. 6e). The attractiveness of the 660 nm object equaled that of
the darkest gray and black objects (Fig. 6e; one-sample t-test:
P=0.07 and P=0.08 for the darkest gray [gray 1.5] and black
objects, respectively). We then determined which contrast of a
510 nm object (green to human observer) matched the attraction
to the 660 nm object. For this purpose, we used the darkest gray
object that was equally attractive as the 660 nm object and tested
it against objects with dominant wavelengths of 510 nm but
differing in their background contrast (Fig. 6b). The dark gray
object was signicantly more attractive than most 510 nm objects,
but the darkest 510 nm object elicited the same level of attraction
(Fig. 6f; preference index =0.01; one-sample t-test: P=0.66).
To determine whether Ae. aegypti mosquitoes can discriminate
between objects of different dominant wavelengths but similar
levels of apparent contrast, we tested attraction to the 660 nm vs.
the darkest 510 nm object (Fig. 6g, h). During exposure to CO
2
,
mosquitoes were strongly attracted to 660 nm but not the dark
510 nm object (Fig. 6g, h; one-tailed t-test: P< 0.0001). We also
determined whether mosquitoes could discriminate between
objects that overlapped in their spectral bands. For this purpose,
we matched the apparent contrast of the non-attractive 452 nm
object (blue to human observer) to an attractive gray. Next, we
tested mosquito responses to the 497 nm object (another
attractive spectral band) and the dark 452 nm object, as well as
the 497 nm object vs. the dark 510 nm object. Similar to the
results observed with the 660 nm object, mosquitoes signicantly
preferred the 497 nm object over the dark 452 and 510 nm objects
(Fig. 6h; one-tailed t-test: P< 0.0001). Thus, mosquitoes easily
discriminated between objects with overlapping and distinct
spectral bands, even when the object contrasts matched.
By incorporating object contrasts, reectance values, and peak
wavelengths as independent variables into a series of linear
models, the results of the behavioral tests offered a means to
examine the relative contributions of these variables toward
mosquito preferences. In these models, all possible combinations
were tested, and the best model was selected based on its Akaike
Information Criterion (AIC) score, where the AIC estimates the
value of each model and lower scores reect the quality of the
statistical model. Using combinations of the independent
variables, we found the best model (and hence, lowest AIC
score) relied on object peak wavelength and contrast, and
excluded reectance (Supplementary Fig. S3a, b). But which
dominant wavelengths might be critical for mediating these
behaviors? To further explore the relationship between the object
350 400 450 500 550 600 650 700 750
0
20
40
60
80
100
2 sec.
1 mV
350 nm 410 nm 450 nm 520 nm 600 nm 650 nm 700 nm
Wavelength (nm)
Sensitivity (% ± SE)
noise
a
c
projector
3 sec.
1 mV
d
b
ERG
Fig. 4 Retinal sensitivity to visual stimuli. a Experimental setup for the
electroretinogram (ERG) experiments. bERG responses to a blue (410 nm),
green (520 nm) or red (>590 nm) moving bar (mean ± sem; n=7
mosquitoes). Responses to the moving object were signicantly higher than
the baseline for all tested dominant wavelengths (410 nm, 520 nm or
590 nm), although 410 nm and 520 nm bars elicited stronger responses
(KruskalWallis test, Amplitude responses~wavelength moving bar, df =2,
Chi-sq. =40.03, P< 0.001, n=7 mosquitoes). cERG responses to pulses
of light (350750 nm). Traces are the mean responses (shaded area is the
±sem; n=8 mosquitoes) to discrete wavelengths showing the elevated
responses to 410 nm (violet to human observer) and 520 nm bands. d
Retinal sensitivity curve across the tested wavelengths from 350 to
750 nm. Two maxima occurred at the short (420 nm) and medium
(~530 nm) wavelengths, although responses were still signicantly elevated
above the noise at wavelengths more than 650 nm (two-sided t-test, ERG
vs. noise: P< 0.01).
Occupancy (%)
0
Black Greys
**
***
***
***
***
***
*** ***
control:
test:
0.5
0
-0.2
Preference Index
Flight Activity
4
2
0
cd
-pre CO2
-post
-pre CO2
-post
contrast:
b
a
top view
side view
top view
side view
0.02 -0.30
CO | Grey9.5
2 CO | Grey4.0
2 0.08
Fig. 5 The effect of achromatic contrast on mosquito attraction to visual
objects. a Occupancy map of the distribution of female mosquitoes in the
wind tunnel (top and side views) during CO
2
delivery in presence of a white
and a light gray object (Weber Contrast: 0.05). bAs in a, except the gray
object has a higher contrast (0.23). cRelative ight activity between the
different phases of the experiments (pre-, CO
2
and post-CO
2
). Mosquitoes
exhibited similar ight activities across all tested visual objects (KruskalWallis
test, df =1, Chi-sq =3.24, P=0.07). dMean preference indices for the test
(gray, or black) vs. control object (white) with 95% condence interval
(n=12,764; 27,537; 37,085; 28,644; 36,050; 25,896; and 21,514 mosquito
trajectories for the white, grey9.5, grey6.5, grey4.5, grey4.0, grey2.5, and black
treatments, respectively). Object contrast had a signicant effect on the
attraction to the tested object (KruskalWallis test: Preference Index~contrast
tested object, df =5, Chi-sq =634.16, P< 0.001). All gray objects were
signicantly more attractive than the control, white object (one-sample tw-
tailed t-test, **: P< 0.01, ***: P<0.001) except for the lightest gray object
(0.05), which was not more attractive (one-sample t-test: P=0.33).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x ARTICLE
NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications 7
Content courtesy of Springer Nature, terms of use apply. Rights reserved
wavelengths and its attractiveness, a series of linear models were
run using a multivariate analysis (Principal Components
Analysis, PCA) of each objects visual spectrum. The PCA
analysis allowed reduction of highly collinear and dimensional
spectral data into reduced components, which can then be used as
independent variables in the model. The best model explained
~17% of the variance in object attractiveness (Supplementary
Fig. S3c) and indicated that preference is negatively correlated
with the content of the medium wavelength band (500575 nm)
in the objects visual spectrum (Supplementary Fig. S3d, e).
Comparison of wavelength preference between mosquito species.
The strong and specic responses of Ae. aegypti to specicbandsin
the visual spectrum and the similarity in long-wavelength opsin
gene expansion in other mosquito species21 motivated us to
examine the wavelength preferences in Anopheles (An.) stephensi
and Culex (Cx.) quinquefasciatus mosquitoes. To examine the visual
preferences in An. stephensi and Cx. quinquefasciatus,werst
conducted ERG recordings of 7-day-old females and examined their
retinal responses to discrete wavelengths from 350 to 750 nm
(Fig. 7a). Both mosquito species exhibited the strongest response to
short wavelengths (350420 nm; UV-visible violet to the human
observer), and the second strongest response was observed in
the medium wavelengths (500520 nm; cyan-green). The strong
response of An. stephensi and Cx. quinquefasciatus to the short
wavelengths of 370 nm contrasted with that of Ae. aegypti (dashed
line), which exhibited the strongest response at ~400 nm (Figs. 3d
and 7b, c). All three species exhibited similar responses to the long
wavelengths in the orange to red band (620750 nm).
Are mosquito speciesbehavioral preferences for visual objects
correlated with their ERG responses, and are their color preferences
similar? To answer these questions, we tested the responses of An.
stephensi and Cx. quinquefasciatus mosquitoes to objects that
dominate in wavelengths at 452 nm, 510 nm, 660 nm, and black
objects in the wind tunnel using a methodology similar to that used
for Ae. aegypti, except at lower light levels (1.28 µW/cm2). As with
the results for Ae. aegypti,encounteringtheCO
2
plume caused a
doubling in the number of ying mosquitoes and increased
the percentage of mosquitoes that investigated the visual objects
5.58- to 9.15-fold relative to air-only treatment for An. stephensi and
Cx. quinquefasciatus, respectively (Fig. 7d; KruskalWallis test:
df =2, Chi-square > 7, P< 0.01). However, in contrast to the tight
clustering around attractive visual objects by Ae. aegypti (Fig. 1f),
occupancy maps showed that the responses of An. stephensi
and Cx. quinquefasciatus mosquitoes were much more diffuse
(Fig. 7e, f). Nonetheless, Cx. quinquefasciatus formed a clustering
hotspot around the outlet of the odor plume nozzle that was much
stronger than the responses of the other two species (Figs. 1f, 7e, f).
Although we tried to minimize the odor nozzles visual signature by
using clear acrylic and tubing, Cx. quinquefasciatus mosquitoes
might have located the plume source based on the high CO
2
concentration or by seeing the nozzle to some degree.
We next examined the preferences of the mosquito species for
different spectral objects (black, blue, green, or red) relative to the
evenly reecting white object (the non-attractive control). Similar
to Ae. aegypti, when both An. stephensi and Cx. quinquefasciatus
mosquitoes were subjected to ltered air treatment, they showed
no preference for the black object or any of the spectral objects
relative to the white control object (preference indices of -0.04
and 0.08 for An. stephensi and Cx. quinquefasciatus, respectively;
KruskalWallis test: df =3, Chi-square < 3.66, P> 0.30). How-
ever, their spectral preferences changed when exposed to CO
2
.
After encountering the CO
2
plume, An. stephensi preferred the
ab
cd
ef
Reflectance (%)
0
50
100
400 500 600 700
Reflectance (%)
0
50
100
400 500 600 700
0.08
Occupancy (%)
0
top view
side view
top view
side view
-0.6
0
0.2
-0.6
0
0.2
control:
test:
h
-0.1
0
0.5
control:
test:
*** *** ***
*** ***
***
***
*** ***
***
***
***
Preference Index
Preference Index
Preference Index
g
top view
CO | Red & Grey2.5
2 CO | Grey1.5 & Green-S2
2
CO | Red & Green-S2
2 0.08
Occupancy (%)
0
Fig. 6 The role of dominant wavelength vs. contrast in mosquito visual
attraction. a Spectral reectance of 660 nm (R-Hue), black, and gray
objects used in the experiments: Gray 9.5, 6.5, 4.5, 4.0, 2.5, and 1.5, with
Weber contrast values of 0.17, 0.30, 0.05, 0.10, 0.20, 0.24,
0.27, and 0.28, respectively. bSpectral reectance of 496 nm and
510 nm objects used in the experiments: Gw-T1, Gw-T3, Gc-T1, Gc-Hue,
G-S1, and G-S2, with Weber contrast values of 0.18, 0.11, 0.17,
0.20, 0.25, and 0.27, respectively. c,dOccupancy maps showing
the distribution of female mosquitoes in the wind tunnel (top and side
views) during CO
2
delivery in presence of the 660 nm (R-Hue) and
Grey2.5 objects (c), or 510 nm (G-S2) and Grey1.5 objects (d); both the
510 nm and Grey1.5 objects have the same levels of contrast with the
background (Weber Contrasts of 0.27). eMean preference indices for
the 660 nm vs. gray objects with different levels of contrast with the
background. The Grey1.5 object (Weber Contrast value of 0.28) was
not signicantly different from 0 (one-sample two-tailed t-test: P=0.07)
and was subsequently used in experiments in (f) (blue arrow). fThe
Grey1.5 object was signicantly more attractive than most of the 510 nm
objects (one-sample two-tailed t-test: P< 0.001, ***), although the
darkest 510 nm object (G-S2; Weber Contrast =0.27) was not
signicantly different from 0 (one-sample two-tailed t-test: P=0.66))
(n=10,09815,578 trajectories for each tested object). gAs in (c),
occupancy maps showing the distribution of trajectories around the
660 nm and dark 510 nm objects. hMean preference indices for 660 nm
vs. dark 510 nm objects, and 496 nm vs. dark 452 nm or dark 510 nm
objects. Mosquitoes signicantly preferred the 660 nm and 496 nm
objects over the dark chromatic objects (510 nm and 452 nm)(one-
sample two-tailed t-test: P< 0.0001). For (e,fand h), boxplots area the
mean (line) with 95% condence interval (shaded area), and asterisks
denote P< 0.001 (one-sample t-test) (n=719127,717 mosquito
trajectories for eachtestedobject).
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x
8NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
black and 660 nm objects (Fig. 7h; KruskalWallis test: df =3,
Chi-square =38.6, P< 0.001), but they were not signicantly
attracted to the 452 nm or 510 nm objects (Fig. 7h; multiple
comparison KruskalWallis test: P> 0.05). By contrast, Cx.
quinquefasciatus mosquitoes preferred the 452 nm and 660 nm
objects (KruskalWallis test: df =3, Chi-square =13.6, P< 0.01;
with multiple comparisons: P< 0.05) but were not signicantly
attracted to the 510 nm or black objects (Fig. 7i; multiple
comparison KruskalWallis test: P> 0.05). Collectively, these
results show that odor strongly sensitizes attraction to visual
objects across mosquito species; however, spectral preferences can
be species-specic.
Discussion
Free-ight behavioral experiments with Ae. aegypti mosquitoes
have shown that these insects integrate olfactory, visual, skin
volatiles, and thermal cues to function efciently and robustly in
complex environments11,13,26. However, we know very little
about mosquito visual-guided behaviors or how vision is involved
in host selection. In this study, we utilized real-time tracking of
mosquito behaviors in a large wind tunnel. The wind tunnel
system enables control of aerodynamic conditions to structure the
odor plume and allow the plume and visual objects to be
decoupled in time and space. Both are important considerations
when testing olfactory-visual integration. In this study, and
similar to our previous work using achromatic objects11,26, the
presence of CO
2
increased mosquito responses to visual objects in
a wavelength-specic manner. Both chromaticity and contrast
were important components in visual object attraction and could
partly explain Ae. aegypti mosquito preferences for objects that
appear orange and red to human observers. These results were
qualitatively similar to those reported by Smart and Brown
(1957), who examined the landing responses of mosquitoes on
colored cloth in a eld16. The results of their study showed that
CO | An. stephensi
2CO | Cx. quinquefasciatus
2
350 450 550 650 750
0
50
100
350 450 550 650 750
0
50
100
ERG
Ae. aegypti
An. stephensi
Cx. quinque.
An. stephensi Cx. quinque.
1 mV
2 mV
4 mV
1 s
350 nm 400 nm 700 nm Wavelength (nm) Wavelength (nm)
abc
Retinal sensitivity (%)
Retinal sensitivity (%)
def
0
Preference Index
0.5
-0.3
control:
test:
Ae. aegypti An. stephensi Cx. quinquefas.
b
b
a
aa
b
c
c
c
a
b,c
a,b
0
Preference Index
0.5
-0.3
0
Preference Index
0.5
-0.3
ghi
top
view
side
view
top
view
side
view
top view
top view
top view
top view
An. stephensi
Cx. quinquefasciatus 0.08
Occupancy (%)
0
Fig. 7 Species-specic responses to dominant wavelengths. a Electroretinogram (ERG) recordings from Ae. aegypti,An. stephensi and Cx. quinquefasciatus
mosquitoes. Each trace is the mean ± sem of 79 mosquitoes/species. b,cRetinal sensitivity to discrete wavelengths show that both An. stephensi (b) and
Cx. quinquefasctiatus (c) have the strongest responses in the UV (360 nm) and green (520 nm) bands. The dashed line is the retinal sensitivity of Ae.
aegypti.dRepresentative ight trajectories [(x,y) and (x,z)] of Cx. quinquefasciatus (left) and An. stephensi (right) mosquitoes. Cx. quinquefasciatus showed a
mild attraction to the blue object (top), whereas An. stephensi showed an attraction to the red object (bottom). e,fOccupancy maps (x,y) of An. stephensi
(e) and Cx. quinquefasciatus (f) distribution around the 452 nm (top) and 660 nm (bottom) objects during CO
2
exposure. giThe preference indices for Ae.
aegypti (g), An. stephensi (h) and Cx. quinquefasciatus (i) in response to the black, 660 nm (R-Hue), 510 nm (Gc-T1), and 452 nm (Bw-T1) objects. Lines are
the means and shaded bars are the 95% condence intervals, and letters above bars denote statistical comparisons (KruskalWallis test with multiple
comparisons: P< 0.05) (for Ae. aegypti,n=29,254; 13,580; 18,190; and 12,086 trajectories; for An. stephensi,n=5817; 9134; 2153; and 5627 trajectories;
and for Cx. quinquefasciatus,n=3746; 13,553; 4238; and 2835 for black, 660 nm, 510 nm, and 452 nm treatments, respectively).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x ARTICLE
NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications 9
Content courtesy of Springer Nature, terms of use apply. Rights reserved
mosquitoes (Aedes sp.) are attracted to red and black cloths, with
darker colors being more attractive than lighter shades16. Our
ERG results and wind tunnel assay data suggest that mosquitoes
can detect and are attracted to the long-wavelength bands in the
orange and red portions of the human visual spectrum, although
if objects contrast highly with the background and/or are darker,
then they become more attractive. The demonstrated spectral
preferences of mosquitoes and their lack of attraction to spectral
bands in the 450 nm region, and 510550 nm regions (blue and
green to human observer), impacted the fashion industry in the
early twentieth century. For instance, Nuttall and Shipley
(1902)27 suggested wearing khaki pants would be appropriate for
ensembles in a tropical environment, and the US military chan-
ged its dress shirts from dark blue to light blue in part to mitigate
mosquito biting28. Nonetheless, a key question is what drives
these responses in mosquitoes, particularly the attraction to
darker and higher-contrast colors? An important component
could be the visual OFF responses to the visual background29.In
our wind tunnel, we projected a light gray checkerboard pattern
on the bottom of the tunnel to provide optic ow and contrast
with the visual objects. The lack of opsins for the long wave-
lengths could cause an OFF response in downstream neurons that
receive input from the photoreceptors. One complication with
this hypothesis is that mosquitoes still preferred the long wave-
lengths when tested against a dark 510 nm object that matched it
in apparent contrast, suggesting they have an ability to detect
these long wavelengths. This was further demonstrated by the
ERG responses to long wavelengths. Recent work in Drosophila
has shown that photopigments provide a mechanism for ies to
long wavelengths >600 nm30, and it is possible that similar pro-
cesses play a role in mosquitoes.
Although the contrast with a lighter background did impact the
attraction responses of Ae. aegypti mosquitoes, the preferences for
the dominant wavelengths that appear red and cyan to humans
were greater than the apparent contrasts of the competing visual
cues (objects with dominant wavelengths at 510 nm, or 452 nm).
The opsin-2 gene, which is tuned to the green to orange band of
the visual spectrum, is highly expressed in the mosquito retina.
The results of this study and those reported elsewhere24 suggest
that opsin-1 and opsin-2 play important roles in determining
object preference and visual attraction to human skin. Additional
research will be needed to identify the opsins and neural circuits
involved in small-eld object detection and color preference and
determine how odor modulates those responses.
Compared with other insects, such as honeybees or the tsetse
y, we know little about mosquito visual ecology or how visual
cues are integrated with other senses in these insects. Whereas
shallow pools of water can be rich in medium-long wavelengths
and owers dominant in short-medium wavelengths, hostsskin
is dominated by long wavelengths in the >600 nm region of the
visual spectrum (Fig. 1b). Abundant work by the cosmetics
industry has shown that human skinirrespective of skin tone or
pigmentationhas a lower peak in the green wavelength
(530 nm, ~20%) and a dominant reectance in the long wave-
lengths (>600 nm, 2060%). The diurnal Ae. aegypti and noc-
turnal and crepuscular An. stephensi and Cx. quinquefasciatus
mosquitoes are all active during periods in which these longer
wavelengths are dominant. For example, Ae. aegypti exhibit peak
activity in the mornings and late afternoons, and An. stephensi
and Cx. quinquefasciatus mosquitoes are especially active during
moonlit nightsboth environments are long-wavelength
shifted20,31,32. In this study, we show that mosquitoes are espe-
cially sensitive to long wavelengths (590660 nm) for host
detection; blocking these wavelengths can suppress object
attraction. Moreover, we found that mosquitoes can distinguish
between overlapping and discrete spectral bands (Fig. 6h), even
when their apparent contrasts match. Insect photoreceptors
within an ommatidial cartridge transduce light intensity and
spectral information. At the photoreceptor terminals, they also
provide antagonistic inputs to downstream neuron targets, thus
allowing discrimination of spectral inputs (termed color oppo-
nency). In Drosophila melanogaster, color opponency at the
photoreceptor terminals plays an important role in their color
discrimination and preference29, and it could be that similar
processes are at play in mosquitoes. In a variety of insects,
including ies, bees, and butteries, spectrally sensitive photo-
receptors form connections with transmedulla neurons that
project into the lobula, where additional spectral and motion
processingincluding color opponencyoccurs3335.
It is important to note that our current experiments did not
incorporate close-range cues from a host, such as heat, water
vapor, or skin volatiles. These cues play critical roles in control-
ling landing and biting behaviors, and future work could deter-
mine how visual spectra are processed in tandem with these other
stimuli. Nonetheless, previous work has shown that visual cues
can promote mosquito orientation and search behaviors in
combination with odor, heat, or water vapor11,13. The integration
of multimodal stimuli in driving behavioral responses raises
questions regarding how the sensory systems are linked in the
brain. Odor stimulation increases visual responses in the object-
detecting neuropil in the Ae. aegypti lobula26. Neuropil in this
brain region is responsive to moving objects but not wide-eld
motion. Interestingly, whereas olfactory stimulation increases
visual responses in the lobula, visual stimulation does not mod-
ulate glomerular responses in the antennal lobe, the primary site
for processing olfactory information in the mosquito brain. Why
might this occur? Mosquitoes have a relatively poor visual reso-
lution (~10°); thus, vision may not provide ne-scale information
about the identity of an object. Instead, an objects odor may
provide information about its identity, whereas vision can provide
details regarding the location of the object.
Despite the potential importance, few studies have examined
retinal responses to long wavelengths in mosquitoes22,23, and how
peripheral and downstream visual circuits, such as those in the
optic lobes, process this information remains unknown. Evolu-
tionary analyses of long-wavelength opsins in diverse mosquito
species have suggested these genes are functionally important. In
Ae. aegypti,Anopheles coluzzii, and Cx. quinquefasciatus, these
genes have undergone duplication events and may be under
positive selection, explaining the commonalities in ERG respon-
ses. Aedes aegypti has 10 putative opsins, ve of which are
potential medium- to long-wavelength opsins in the adult
(>500 nm)21. One of these genes, opsin-1, is expressed in the
largest group of photoreceptor cells (R1R6) in the Ae. aegypti
eye22. The R1-R6 photoreceptors form a trapezoid-like structure
in the ommatidial cartridge, with the R7 and R8 cells in the
middle. Perhaps similar to D. melanogaster with rhodopsin-1
(Rh1), the Ae. aegypti opsin-1 is likely involved in motion- and
dim-light sensitivity36. The inner R7 and R8 photoreceptors are
involved in color vision, and the Ae. aegypti opsin-2 is expressed
in the R7 photoreceptors located in distinct bands on the dorsal
and ventral surfaces of the eye, and possibly used for navigation
and biting behaviors. For mosquitoes, opsins that differ in spec-
tral tuning can be co-expressed in the R7 photoreceptor in the
female retina (e.g., the long-wavelength opsin-2 and short-
wavelength opsin-9), thereby increasing their range in wave-
length sensitivity37. This may contrast the opsin co-expression in
D. melanogaster R7 cells in the dorsal region of the eye, where two
short (UV)-sensitive opsins are co-expressed38, presumably to
increase the spectral contrast with the green-absorbing opsins in
the R8 layer. The opsin co-expression and broadening of spectral
input in mosquitoes could be advantageous under low-light
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x
10 NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
conditions, as photon capture would be maximized, thus allowing
for detection of suitable hosts or perhaps sources of nectar37,39,
but this advantage would come at the cost of detection of parti-
cular wavelength bands. In our study, however, mosquitoes were
able to discriminate between distinct and overlapping spectral
bands (510 nm and 660 nm; 496 nm and 510 nm; and 496 and
452 nm), all at the same apparent contrasts, suggesting that other
opsins, perhaps in the R8 cells, or downstream circuits, play a role
in increasing the separability of those hues.
Preferences for certain dominant wavelengths plays a critical
role in a diverse array of insect vectors, and it is integrally tied to
olfaction. For instance, traps that incorporate visual and olfactory
cues have proven transformative as a low-cost method for con-
trolling tsetse ies in parts of Africa40. Like Ae. aegypti, the
majority of tsetse ies (e.g., Glossina morsitans morsitans and G.
pallidipes) locate hosts based on smell, and once they are within
close range (<10 m), visual cues cause the insects to investigate
and potentially bite if the object is a host. Beginning with Vale
(1974)41 and continuing into the 1990s, researchers found that
ies are most attracted to a specic blue (~460 nm), followed by
red and black, but they are not attracted to green and white5,6.
Researchers subsequently found that incorporating blue and black
hues in traps was particularly effective at inducing ies to come
into contact with insecticide-treated screens. Triatoma infestans
(i.e., the kissing bug) is another example of an insect that inte-
grates odor and visual cues to mediate attraction to targets.
Exposure to an aggregation pheromone modied the responses of
T. infestans to colored objects, including the red, blue, and black
hues, although the bugs always reject green (~525 nm) and white
hues7. In a similar manner, our results obtained from tests of
different mosquito species demonstrate the importance of olfac-
tion in mediating mosquito spectral preferences. In the absence of
CO
2
, mosquitoes did not demonstrate any preference between
evenly reecting (white to humans) and dominant wavelengths,
but they became attracted to certain dominant wavelengths in the
presence of CO
2
. However, there were species-specic differences
in their wavelength preferences. Whereas Ae. aegypti was equally
attracted to both 660 nm and black objects, An. stephensi was
most attracted to black, followed by 660 nm objects. By contrast,
Cx. quinquefasciatus was attracted to 452 nm, followed by 660 nm
objects; surprisingly, however, this species was not attracted to
small black objects. Collectively, the results of our current study
and those of other studies show that the visual systems of insect
disease vectors and their behaviors constitute attractive targets for
the development of traps incorporating visual features that can be
species-specic in terms of attraction, thus providing incentive to
identify molecular targets that compromise mosquito olfactory-
visual responses.
Methods
Mosquitoes, odor delivery, and wind tunnel. Mosquitoes (Aedes aegypti: Rock-
efeller, Liverpool, Gr3[ECFP]13 and opsin-1,opsin-2 mutant lines; Anopheles ste-
phensi (Indian strain) and Culex quinquefasciatus) were raised at the University of
Washington campus. Mosquito lines were provided from BEI Resources (Mana-
ssas, VA, USA) (Ae. aegypti: Rockefeller, Liverpool, Gr3[ECFP]13;An. stephensi
and Cx. quinquefasciatus). In the case of the Ae. aegypti opsin-1,opsin-2, and opsin-
1,opsin-2 lines, we used existing mosquito lines that were generated as previously
described24. Briey, the op1 and op2 alleles were generated by selecting short-guide
RNAs that targeted the GPROp1 (LOC5568060) and GPROp2 (LOC5567680)
loci24. Lines were homogenized and veried by PCR before testing24 (Supple-
mentary Tables 13). Mosquitoes were raised in groups of 100 individuals and
anesthetized with cold to sort males from females after cohabitating for 7 days. At
this time, more than 90% of the females have been mated, as indicated by their
developing embryos. For each experimental trial in the wind tunnel, we released 50
females into the wind tunnel working section 3 h prior to the mosquitos subjective
sunset (the time period of peak activity for Ae. aegypti). Each visual stimulus was
tested in 412 experimental trials (on average, ~5575 trajectories were quantied
per trial) and 12,300 Ae. aegypti were own in total across all experiments (99.4%
of all mosquitoes ew in the working section, and thus their behaviors were
captured by the real-time tracking system), for a total of 1,305,695 trajectories.
After one hour, the 5% CO
2
plume (or ltered air in control experiments), was
automatically released from a point source at the immediate upwind section of the
tunnel and at a height of 20 cm and in the centerline of the tunnel. The CO
2
remained on for 1 h, before switching off for another hour of ltered air (post
CO
2
). The 1 h time periods were chosen because mosquitoes did not adapt or
habituate to the CO
2
plume due to the diameter of the plume (~1.5 cm) and the
size of the tunnel lowered the encounter probability of the mosquitoes to the
plume. In addition, the 1 h periods provided a baseline of behavior before CO
2
release, and after encountering the CO
2
, mosquitoes would remain activated and
visually sensitized for 510 minthus, the 1 h post-CO
2
period allowed sufcient
time for mosquitoes to return to baseline.
The CO
2
and ltered air were automatically delivered using two mass ow
controllers (MC-200SCCM-D, Alicat Scientic, Tucson, AZ) that were controlled
by a Python script that allowed synchronizing odor and ltered air delivery with
the trajectory behaviors. The CO
2
plume was quantied using a Li-Cor LI-6262
CO2/H2O analyzer (Li-Cor, Lincoln, NE) for a total of 500 locations throughout
the tunnel (Fig. 1). Data yielded an exponential decay similar to a model of
turbulent diffusion at the airow (40 cm/s) and turbulent intensities (5%) of this
tunnel, such that 20 cm from the source and parallel to the wind ow the plume
was ~1700 ppm (Fig. 1c, d), which is in the range of the plume of human breath.
All behavioral experiments took place in a low-speed wind tunnel (ELD Inc.,
Lake City, MN), with a working section of 224 cm long, 61 cm wide, by 61 cm high
with a constant laminar ow of 40 cm/sec (Fig. 1). We used three short-throw
projectors (LG PH450U, Englewood Cliffs, NJ) and rear projection screens
(SpyeDark, Spye, LLC, Minneapolis, MN) to provide a low contrast checkerboard
on the oor of the tunnel and gray horizons on each side of the tunnel. The
intensity of ambient light from the projectors was 96 lux across the 420670 nm
range. A 3D real-time tracking system (the open-source Braid system)11,42 was
used to track the mosquitoestrajectories. Sixteen cameras (Basler AC640gm,
Exton, PA) were mounted on top of the wind tunnel and recorded mosquito
trajectories at 60 frames/sec. All cameras had an opaque IR Optical Wratten Filter
(Kodak 89B, Kodak, Rochester, NY) to mitigate the effect of light in the tracking.
IR backlights (HK-F3528IR30-X, LedLightsWorld, Bellevue, WA) were installed
below and the sides of the wind tunnel to provide constant illumination beyond the
visual sensitivity of the mosquitoes. The temperature within the wind tunnel,
measured using ibuttons and FLIR cameras (FLIR One Pro, FLIR Systems Inc.,
Goleta, CA USA), was 22.5 °C and did not show any variability within the working
section11,24. Ambient CO
2
was constantly measured outside of the tunnel and was
~400 ppm.
Visual stimuli and experimental series in the wind tunnel. To determine the
role of odor in the innate spectral preferences of mosquitoes, and identify the role
of achromatic contrast and wavelength discrimination, a series of different
experiments were conducted. In each experiment, two visual stimuli, separated by
18 cm, were presented to the mosquitoes on the oor in the upwind area of the
tunnel and perpendicular to the direction of airow. Visual stimuli consisted of
paper circles that were 3 cm diameter (Color-aid Corp., Hudson Fall, NY, USA).
Reectance spectra for all the visual stimuli were characterized using an Ocean
Optics USB2000 spectrophotometer with a deuterium tungsten halogen light
source (DH2000) calibrated with a white Spectralon standard (Labsphere, North
Sutton, NH, USA). The projector and working section light intensities were
measured using a cosine-corrected spectrophotometer (HR- +2000, Ocean Optics,
Dunedin, FL, USA) 5 cm from the projector source (63 µW/cm2). The achromatic
contrasts of the visual objects relative to the background were measured using the
calibrated spectrophotometer and the Weber contrasts were calculated by the
intensity (µW/cm2) of the object (I
object
) and background (I
backgroun
), where (I
object
-
I
background
)/I
background
. In the rst experimental series, using wt (ROCK) Ae. aegypti,
we examined the innate preference of individual colors relative to the non-
attractive white object. Tested objects had distinct and peak wavelengths of 437 nm,
452 nm, 496 nm, 510 nm, 520 nm, 590 nm, 600 nm, and 660 nm (Bv-T2, Bw-, Gw-
T1, Gc-, YGc-, Yw-, O- and R-Hue; Color-aid Corp.). These stimuli all had similar
achromatic contrasts (0.12 to 0.18) and peak reectance values, but had distinct
peak wavelengths (Fig. 1g). The position of the respective hue and white object in
each replicate trial was randomized.
In the second experimental series, experiments were performed to examine
which spectral bands of the human skin might attract mosquitoes. To ensure the
replicability of the experiments, and enable the control of visual object humidity,
temperature, and odor, we rst elected to rst use faux skin mimics (Pantone
SkinTone Guide; Pantone LLC, Carlstadt, NJ 07072 USA). Using wt (ROCK) Ae.
aegypti, we tested four different skin tones (R10, Y10, and Y02) and a skin tone that
we named vile 45, which matched the putrid orange tone from individuals who
use cheap tanning lotion (PANTONE 16-1449X, Gold Flame). Similar to our
previous wind tunnel experiments, each individual skin tone was paired with an
evenly reective object (white to human observer) that served as a control. To
attenuate different spectral bands reected from the skin tone, we used ultra-thin
(~200 um thick) plastic lters that were placed over the object. Filters were selected
to attenuate the 450530 nm band, the 550630 nm band, or the 650730 nm band
(36333 [notch lter], 35894 [long-pass lter], and 35896 [long-pass lter],
respectively; Edmund Optics Inc., Barrington, NJ USA). Reectance measurements
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x ARTICLE
NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications 11
Content courtesy of Springer Nature, terms of use apply. Rights reserved
of the object with the lters showed the transmission loss was <5% for bands
outside of the ltered wavelengths (Fig. 2B). As a control to determine if
attenuating a spectral band outside of the visual spectrum effects mosquito
behavior, we used an IR lter that allows transmission of 350750 nm band of the
visual spectrum (14-547 [KG2], Edmund Optics Inc., Barrington, NJ USA). To
control for the physical effect that placing the plastic lter over the object may have
had on mosquito behavior, we used plastic coverslips with the same refractive index
as glass (72261-22; Electron Microscopy Services, Hateld, PA USA). Experiments
were also conducted with Gr3/mutants13 and the opsin-1/,opsin-2/double
mutants24 to examine how the loss of olfactory or visual detection, respectively,
impacted mosquito attraction to the skin color. As controls, we used heterozygote
(Gr3/+), single mutants (opsin-1/, opsin-2/), and the wild-type
(Liverpool) lines.
In the third series, the role of achromatic contrast, and not wavelength, on
mosquito visual preferences were examined. Gray circles (3 cm diameter), differing
in their Weber Contrast (0.28 to 0.05) and ranging from near-black to very
light gray (Grays 1.5, 2.5, 4.0, 4.5, 6.5, and 9.5; Color-aid Corp., Greyset), were run
in combination with a white circle (Weber Contrast of 0.02). Similar to the above
experiments, wt (ROCK) Ae. aegypti were used in the experiments.
In the fourth experimental series, we examined whether the spectral preferences
of the mosquito (Ae. aegypti [ROCK]) while controlling for the apparent contrast
of the different hues. We normalized the perceptual contrasts of the visual objects
by testing a range of grays of different contrasts with the background (Grays 1.5,
2.5, 4.0, 4.5, 6.5, and 9.5 [Weber Contrasts of 0.28 to 0.05]; Color-aid Corp.,
Greyset) in combination with, and against, the 660 nm object (R-Hue; Weber
Contrast =0.17). Once we identied the gray object that was as attractive as the
660 nm object, identied by the PI that not signicantly from 0 (t-test: P> 0.05), we
then tested that gray (Gray 1.5; Weber Contrast =0.28) against a range of
different objects that had the same peak wavelength (510 nm) but different Weber
Contrasts, from light to dark (0.11 to 0.27). Based on the 510 nm object that
elicited the same level of attraction as the dark gray (PI =0), we then tested the
dark 510 nm object vs. the red object. Similar to experiments with the 510 nm
objects, we tested the dark gray (Gray 1.5; Weber Contrast =0.28) against a
range of different objects that had the same peak wavelength (452 nm) but different
Weber Contrasts, from light to dark (0.13 to 0.28), followed by testing the
660 nm object vs. the dark 452 object, and the attractive cyan (495 nm) object vs.
the dark 452 nm or dark 510 nm objects with the same apparent contrast.
Visual attraction to skin in a Cage-assay. To assay the visual attraction to the
skin, acrylic cages (45 × 30 × 30cm; McMaster-Carr; cat. # 8560K171) were con-
structed to allow for video recording and tracking from above. Thermal insulation
and white sheeting were wrapped around the cages exterior to prevent any heat
cues from outside of the cage while providing a uniform visual environment to the
interior. The sides and windows of the cage were sealed to prevent odor con-
tamination from the experimenter from leaking into the cage. Using thermal
imaging (FLIR One Pro, FLIR Systems Inc., Goleta, CA USA), solid-phase
microextraction bers (75 um CAR/PD MS SPME ber, 57344-U; Supelco, Belle-
fonte PA USA) for VOC collection and subsequent analysis using Gas Chroma-
tography with Mass Spectrometric Detection (Agilent Technologies, Palo Alto, CA,
USA), and CO
2
measurements both inside and outside of the cage, allowed for the
testing of odor and heat contamination within the cage. Results showed no odor,
CO
2
or heat contamination (Fig. S2). A small 4 cm vent on the cage side facing
away from the experimenter and underneath the hood ventilation system allowed
ltered air or 5% CO
2
input to the cage. In contrast to the wind tunnel assays,
ambient airow was small (<10 cm/s) within the cage. To conduct the visual
preference assays, two 4 × 4 cm windows, spaced 18 cm apart, were cut into the
acrylic. Windows were sealed with heat absorptive glass (Schott KG2, Edmund
Optics) to prevent both thermal and odor contamination into the cage. Similar to
the visual stimuli used in the wind tunnel experiments, mosquitoes were tested
with a uniform reective controlin one window (white-colored glove to human
observer), and the other window displaying either human skin, or human skin
through long-wavelength optical lters (550730 nm band; 36333 and 35894
lters; Edmund Optics Inc., Barrington, NJ USA). Positions of the visual stimuli
displayed in the windows were randomized between experimental replicates. To
test for any side preference or contamination in the cage, control experiments were
also conducted using two white gloves as the visual stimuli in the windows.
Experiments were performed in a chamber held at ~2022.5 °C, and the cage was
situated underneath a hood ventilation system allowing air exchange. A CO
2
Flypad (Genesee Scientic; cat. # 59-119) was placed immediately adjacent to the
vent on the side of the cage, and similar to the wind tunnel experiments, CO
2
was
controlled by two mass ow controllers (MC-200SCCM-D, Alicat Scientic,
Tucson, AZ) via a Python script that allowed synchronizing odor and ltered air
delivery with the trajectory behaviors. Filtered air was released for the rst 8 min of
each experiment, followed by the release of 5% CO
2
for 8 min, before switching off
for another 8 min (post-CO
2
). Two cameras (Basler AC640gm, Exton, PA) were
mounted above the cage and recorded mosquito trajectories at 100 frames/sec. IR
backlights (HK-F3528IR30-X, LedLightsWorld, Bellevue, WA) were installed above
the cage. Human skin reectance measurements and assays were from three males
and three female individuals on the University of Washington (Seattle, WA USA)
campus (ages 2546 years old), and volunteers were from various ethnic groups.
An ergonomic armstand set was used to position and keep the arms steady over the
24 min. period. Because CO
2
causes behavioral changes in the mosquitoes that can
last 510 min after CO
2
exposure is stopped, the 24 min. experiment was the
minimum time that allowed us to examine the mosquito responses at the different
time periods (Air, CO
2
, post-CO
2
). Protocols were reviewed and approved by the
University of Washington Institutional Review Board, and all human volunteers
gave their informed consent to participate in the research. Similar to the wind
tunnel experiments, we used 68 day-old, non-blood-fed, mated females who were
sucrose deprived for 24 h but had access to water. We released 50 Ae. aegypti
(ROCK) females into the cage for each experiment, and the assays were initiated
3 h before lights off (ZT12). Ambient CO
2
was constantly measured both inside
and outside of the cage.
Visual preferences in Cx. quinquefasciatus and An. stephensi.Anopheles ste-
phensi (Indian strain) and Culex quinquefasciatus mosquitoes were separately
raised in groups of 100 individuals and anesthetized with CO
2
to sort males from
females after cohabitating for 7 days. At the time of their subjective sunset, groups
of 50 female mosquitoes were released into the working section of the wind tunnel.
After 1 h of ltered air, 5% CO
2
was released from a point source at the upwind
section of the tunnel (height of 20 cm, and in the center of the tunnel) for 1 h, after
which ltered air was released from the point source. The intensity of ambient light
from the projectors was ~1.3 µW/cm2. The low-light intensity, relative to that used
with Ae. aegypti, was necessary to recruit females to the visual objects. At higher
light intensities An. stephensi and Cx. quinquefasciatus mosquitoes responded to
the CO
2
plume but they did not respond to the visual objects. We found that these
two species began to investigate the visual objects only at light intensities <5 lux.
Like our experiments with Ae. aegypti, the temperature within the wind tunnel was
~22.5 °C.
Trajectories analysis. As described above, for each experimental trial we release a
group of 50 mosquitoes because the working section of the wind tunnel was large
enough to minimize any interactions between individuals, while allowing for the
efcient capturing of behaviors to the CO
2
plume and visual objects. Our tracking
system is unable to maintain mosquito identities for extended periods of time, but
we considered individual trajectories as independent for the sake of statistical
analysis. To ensure that the release of the 50 mosquitoes reected behaviors of
single mosquitoes in the working section, experiments were performed as described
in Material and MethodsVisual Stimuli and Experimental Series in the Wind
Tunnel but using single mosquitoes. A 660 nm and evenly reective objects (red
and white to the human observer) were used as the visual stimuli in these
experiments. Fifty Ae. aegypti mosquitoes were individually tested and released into
the working section of the tunnel and their behaviors were compared to mosqui-
toes that were co-released. Results of these experiments showed that all 50/50
individual mosquitoes ew in the tunnel, which compares with the 99.4% of
mosquitoes that ew in the tunnel in all the co-released experiments. We also
found that ight velocities, durations, and PIs of the individual mosquitoes to red
and white circles were not statistically different from co-released mosquitoes
(unpaired t-tests, P=0.75, 0.93, and 0.44 for ight duration, ight velocity, and PI,
respectively). During the CO
2
exposure, on average and for those mosquitoes that
investigated the visual objects (38% of the 50 mosquitoes individually-released into
the tunnel), one mosquito produced nine ight trajectories during the 1 h period
(range from 1 to 37 trajectories) and investigated the visual object 3.70 times
(±2.33). This was similar to the estimated number of trajectories per mosquito that
investigated the objects in co-released experiment s (4.76 ± 3.64). Since 38% of the
mosquitoes investigated the visual objects, this percentage provided evidence that
our trials were not biased from insufcient sampling. Therefore, co-release of the
mosquitoes allowed efcient testing of mosquito behavior to the visual objects and
did not differ from individually released mosquitoes.
Analyses were restricted to trajectories that were at least 90 frames (1.5 s) long.
Only trajectories that lasted for more than 1.5 s were analyzed (average length
trajectory: 3.1 s, longest trajectory: 96.4 s, total number of 1,305,695). The ight
activity in the different phases of an experimental trial (pre-, CO
2
and post-CO
2
)
where the mosquitoes encountered ltered air or CO
2
were quantied by the
number of trajectories recorded during one phase divided by the number of
trajectories recorded in the previous phase. To examine the mosquito behaviors
and preferences to the two visual stimuli in the tunnel, a ctive volume was created
around the visual cues (area: 14 × 14 cm, height: 4 cm). The volume was centered
over the object in the crosswind direction, and shifted slightly downwind in the
wind line direction. This volume was chosen as it captures the area of primary
activity of the mosquitoes. A sensitivity analysis was performed by adjusting the
volume size and demonstrated that this volume best captured the mosquitoes
investigating the visual objects while excluding mosquitoes transiting to other areas
of the working section.
Occupancy maps were calculated by dividing the wind tunnel into 0.3 cm2
squares. For each replicate experiment, the number of mosquito occurrences within
each square was summed and divided by the total number of occurrences in all
squares to yield a percentage of residency. We did not quantify landings on the
spots due to limitations of the camera angles needed to identify landings. During
the ltered airtreatment, mosquitoes often investigated certain areas of the
working section, such as the top or corners of the working section, causing hot
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x
12 NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
spots in the occupancy maps. This is typical for mosquito activity without a
stimulus. By contrast, when CO
2
is released, these hot spots are no longer apparent,
and instead, the mosquitoes investigate the visual objects or navigate to the odor
source, as demonstrated by a hotspot in the central area of the working section or
near the odor source. For each mosquito line, the replicate trials were pooled to
create an occupancy heat map for the tested visual stimulus.
To calculate the fractions of trajectories that approached either visual object, for
each trajectory we calculated a preference index by determining the amount of time
a trajectory spent in each volume divided by the total time it spent in both volumes.
If the trajectory spent all of its time in only one volume then it was assigned a
preference index of 1 (test object) or 1 (white neutral object). Approximately
2550% of the trajectories approached either object. From these preference indices,
we calculated the global mean, and bootstrapped the 95% condence interval of the
mean through random resampling of the individual trajectories 500 times. To
determine whether the mosquitoes preferred the visual objects compared to
elsewhere in the tunnel, we calculated the preference index for each trajectory at
each time point as the amount of time the mosquito spent in a particular
4 × 14 × 14 cm volume that was randomly selected in the tunnel and compared
them to the volumes containing the visual objects. Mean ight velocities were
calculated from the 3D tracks of each individual trajectory. To further examine
whether mosquito responses to the visual objects changed throughout the
experiment, the percent of time (per each minute interval) the mosquitoes
investigated the visual objects was calculated. Statistically signicant groups were
estimated using a KruskalWallis, MannWhitney U-test with Bonferroni
correction at a P=0.01 level, or the one-sample t-test. All recorded data were
analyzed using Matlab (Mathworks, 2019a release). To explore the potential impact
of pseudoreplication on the statistical results, we used a multilevel analysis using
the restricted maximum likelihood method and setting the experimental trial as a
random, nested effect. We found that, for the visual stimuli used in the
experiments, there was a signicant impact of the visual spectra of the object on the
preference index (P< 0.001) but the experimental trial did not contribute to the
model results. Thus, potential for pseudoreplication due to biased sampling did not
contribute to our results. The model was created using R 4.0.3 and the lmer
function from the lme4 package43.
Electroretinogram (ERG) recordings. ERG recordings were performed by xing
6 day-old, non-blood-fed female mosquitoes to a coverslip using Bondic glue. Mos-
quitoes were dark-adapted for 1h prior to stimulation. The recording glass electrode
(thin-wall glass capillaries; OD, 1.0mm; length, 76 mm; World Precision Instruments,
cat. # TW100F-3) was pulled using a micropipette puller (Sutter Instrument, p-2000),
and lled with Ringers solution (3 mM CaCl
2
, 182 mM K Cl, 46 mM Na Cl, 10 mM
Tris pH 7.2). The reference electrode, a sharpened tungsten wire, was placed into one
compound eye in a small drop of electrode gel (Parker, cat. # 17-05), and the
recording electrode was placed immediately on the surface of the contralateral eye.
Two different types of visual stimuli were presented to the mosquitoes. In the rst
series, the mosquitoes were placed at the center of a semi-cylindrical visual arena
(frosted mylar, 10 cm diameter, 10 cm high); a video projector (Acer K132 WXGA
DLP LED Projector, 600 Lumens) positioned in front of the arena projected the visual
stimuli. To test the response to moving objects, similar to what the mosquito might
encounter in ight, we tested responses to a 19° wide bar moving from left to right
(Clockwise) (Fig. 4). The mosquitowas randomlytested with blue, green, and red bars
(distinct peaks at 455, 547 nm, and 633 nm, 18 lux), and each colored bar was tested
1030 times per mosquito (n=7 mosquitoes).
The second stimulation method used a digital monochromator to examine
responses to different wavelengths across the mosquito visual spectra
(350750 nm). Mosquitoes were exposed to a 1 s pulses of light (10 lux) from a
light source (35-watt Halogen; ThorLabs) and a ber optic scanning
monochromator (MonoScan 2000, Mikropak GmbH, Ostldern, Germany) that
provided control of the transmitted wavelengths (±2 nm). Light was transmitted via
optical bers (QP600-1-SR-B X, Ocean Optics, FL 32792, USA) and through a
neutral density lter (fused silica, Thorlabs Inc., 0-1 OD). Each mosquito
preparation was tested to wavelengths of 350750 nm in 10 nm increments (n=8
mosquitoes/species). The visual stimuli were calibrated using a cosine-corrected
spectrophotometer (HR- +2000, Ocean Optics, Dunedin, FL, USA) that was
placed immediate to the recording preparation, allowing us to scale the irradiance
of the tested stimuli. The light-induced responses were amplied by using an A-M
Systems amplier (10-100x; A-M Systems, 1800) and digitized using a Digidata
data acquisition system (Digidata 1550B, Molecular Devices, San Jose, CA 95134).
Data were visualized and analyzed using Matlab software (Mathworks).
Linear models. Linear models were created using R 4.0.3 and the lm function with
the default option. Comparison between models were performed using the AIC
function that calculates the Aikaikes Information Criterion (AIC) for each model.
For the rst series of models, the dataset consisted of the mean preference index
per experiment, the contrast value, peak wavelength and brightness value for the
tested object. For the second series of models, the dataset contained the mean
preference index per experiment and the area under the curve (AUC) of the
reectance measurement for the tested object calculated with bins of 25 nm from
350 to 675 nm. A Principal Component Analysis (PCA) was applied to the AUC
vector to remove collinearity in the objects spectrum.
Reporting summary. Further information on research design is available in the Nature
Research Reporting Summary linked to this article.
Data availability
The source data underlying Figs. 1h, 3d, and 5c are provided as a Source Data le. The
wind tunnel data generated in this study have been deposited in the Dryad Data
Repository at https://doi.org/10.5061/dryad.d51c5b04d. Source data are provided with
this paper.
Code availability
Software is available on https://github.com/riffelllab (https://doi.org/10.5281/
zenodo.5579784) and https://github.com/strawlab/strand-braid.
Received: 27 July 2021; Accepted: 5 January 2022;
References
1. Briscoe, A. D. & Chittka, L. The evolution of color vision in insects. Ann. Rev.
Entomol. 46, 471510 (2001).
2. Hannah, L., Dyer, A. G., Garcia, J. E., Dorin, A. & Burd, M. Psychophysics of
the hovery: categorical or continuous color discrimination?. Curr. Zool. 65,
483492 (2019).
3. van der Kooi, C. J., Stavenga, D. G., Arikawa, K., Belušič, G. & Kelber, A.
Evolution of insect color vision: from spectral sensitivity to visual ecology.
Ann. Rev. Entomol. 66, 435461 (2021).
4. Clements, A. N., The biology of mosquitoes. Volume 2: sensory reception and
behaviour. (CABI publishing, 1999).
5. Green, C. Effects of colours and synthetic odours on the attraction of Glossina
pallidipes and G. morsitans morsitans to traps and screens. Physiol. Entomol.
11, 411421 (1986).
6. Torr, S. & Vale, G. Know your foe: lessons from the analysis of tsetse y
behaviour. Trends Parasitol. 31,9599 (2015).
7. Reisenman, C. E., Lorenzo Figueiras, A. N., Giurfa, M. & Lazzari, C. R. Interaction
of visual and olfactory cues in the aggregation behaviour of the haematophagous
bug Triatoma infestans. J. Comparat. Physiol. A 186,961968 (2000).
8. Bidlingmayer, W. L. & Hem, D. G. Mosquito (Dipteral Culicidae) ight
behaviour near conspicuous objects. Bullet. Entomol. Res. 69, 691700 (2009).
9. Kennedy, J. S. The visual responses of ying mosquitoes. Proc. Zool. Soc.
London, A 109, 221242 (1939).
10. WL, B. How mosquitoes see traps: role of visual responses. J. Am. Mosquito
Contr. Assoc. 10, 272 (1994).
11. van Breugel, F., Riffell, J. A., Fairhall, A. & Dickinson, M. H. Mosquitoes use
vision to associate odor plumes with thermal targets. Curr. Biol. 25,
21232129 (2015).
12. Kline, D. L. & Lemire, G. F. Field evaluation of heat as an added attractant to
traps baited with carbon dioxide and octenol for Aedes taeniorhynchus. J. Am.
Mosquito Cont. Assoc-Mosquito News 11, 454456 (1995).
13. McMeniman, C. J., Corfas, R. A., Matthews, B. J., Ritchie, S. A. & Vosshall, L. B.
Multimodal integration of carbon dioxide and other sensory cues drives mosquito
attraction to humans. Cell 156,1060
1071 (2014).
14. Fay, R. & Prince, W. A trap based on visual responses of adult mosquitoes. J.
Am. Mosquito Contr. Assoc-Mosquito News 28,17 (1968).
15. Muir, L. E., Kay, B. H. & Thorne, M. J. Aedes aegypti (Diptera: Culicidae)
vision: response to stimuli from the optical environment. J. Med. Entomol. 29,
445450 (1992).
16. MR, S. & AWA, B. Studies on the responses of the female Aedes mosquito.
Part VIIthe effect of skin temperature, hue and moisture on the
attractiveness of the human hand. Bull. Entomol. Res. 47, 89 (1957).
17. Brett, G. On the relative attractiveness to Aedes aegypti of certain coloured
cloths. Transact. Royal Soc. Trop. Med. Hygiene 32, 113124 (1938).
18. Snow, W. F. The spectral sensitivity of Aedes aegypti (L.) at oviposition. Bull.
Entomol. Res. 60, 683696 (1971).
19. Muir, L. E., Thorne, M. J. & Kay, B. H. Aedes aegypti (Diptera: Culicidae)
vision: spectral sensitivity and other perceptual parameters of the female eye. J.
Med. Entomol. 29, 278281 (1992).
20. Trpis, M., McClelland, G. A., Gillett, J. D., Teesdale, C. & Rao, T. R. Diel
periodicity in the landing of Aedes aegypti on man. Bull. World Health Org.
48, 623629 (1973).
21. Giraldo-Calderón, G. I., Zanis, M. J. & Hill, C. A. Retention of duplicated
long-wavelength opsins in mosquito lineages by positive selection and
differential expression. BMC Evol. Biol. 17, 84 (2017).
22. Hu, X. et al. Patterned rhodopsin expression in R7 photoreceptors of
mosquito retina: Implications for speciesspecic behavior. J. Compar. Neurol.
516, 334342 (2009).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x ARTICLE
NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications 13
Content courtesy of Springer Nature, terms of use apply. Rights reserved
23. Hu, X., Whaley, M. A., Stein, M. M., Mitchell, B. E. & OTousa, J. E.
Coexpression of spectrally distinct rhodopsins in Aedes aegypti R7
photoreceptors. PloS ONE 6, e23121 (2011).
24. Zhan, Y., Alonso San Alberto, D., Rusch, C., Riffell, J. and Montell, C. Aedes
aegypti vision-guided target recognition behavior requires two redundant
rhodopsins. BioRxiv.130 (2020).
25. Angelopoulou, E. The reectance spectrum of human skin. 584 (Technical
Reports (CIS), 1999).
26. Vinauger, C. et al. Visual-olfactory integration in the human disease vector
mosquito Aedes aegypti. Curr. Biol. 29, 25092516 (2019). e5.
27. Nuttall, G. H. & Shipley, A. E. Studies in relation to malaria. II.(Cont.) The
structure and biology of Anopheles (Anopheles maculipennis). Epidemiol.
Infect. 2,5884 (1902).
28. Packard, A. Color-preference in insects. J. NY Entomol. Soc. 11, 132137
(1903).
29. Schnaitmann, C., Pagni, M. & Reiff, D. F. Color vision in insects: insights from
Drosophila. J. Comp. Physiol. A 206, 183198 (2020).
30. Sharkey, C. R., Blanco, J., Leibowitz, M. M., Pinto-Benito, D. & Wardill, T. J.
The spectral sensitivity of Drosophila photoreceptors. Sci. Rep. 10,113.
(2020).
31. Day, J. F. Host-seeking strategies of mosquito disease vectors. J. Am. Mosquito
Control Assoc. 21,1722 (2005).
32. Silver, J. B. Designing a Mosquito Sampling Programme. In Mosquito Ecology:
Field Sampling Methods (ed. Silver, J.) 123 (Springer, 2008).
33. Paulk, A. C., Phillips-Portillo, J., Dacks, A. M., Fellous, J.-M. &
Gronenberg, W. The processing of color, motion, and stimulus timing are
anatomically segregated in the bumblebee brain. J. Neurosci. 28, 63196332
(2008).
34. Swihart, S. The neural basis of colour vision in the buttery, Heliconius erato.
J. Insect Physiol. 18, 10151025 (1972).
35. Yonekura, T., Yamauchi, J., Morimoto, T. & Seki, Y. Spectral response
properties of higher visual neurons in Drosophila melanogaster. J. Comp.
Physiol. A 206, 217232 (2020).
36. Leming, M. T. Light-dark and circadian effects on the visual response of Aedes
aegypti. (University of Notre Dame, 2015).
37. Hu, X., Leming, M. T., Whaley, M. A. & OTousa, J. E. Rhodopsin
coexpression in UV photoreceptors of Aedes aegypti and Anopheles gambiae
mosquitoes. J. Exp. Biol. 217, 10031008 (2014).
38. Mazzoni, E. O. et al. Iroquois complex genes induce co-expression of
rhodopsins in Drosophila. PLoS Biol. 6, e97 (2008).
39. Peach, D. A., Ko, E., Blake, A. J. & Gries, G. Ultraviolet inorescence cues
enhance attractiveness of inorescence odour to Culex pipiens mosquitoes.
PloS ONE 14, e0217484 (2019).
40. Shaw, A. P. et al. Costs of using tiny targetsto control Glossina fuscipes
fuscipes, a vector of gambiense sleeping sickness in Arua District of Uganda.
PLoS Neglected Trop. Dis. 9, e0003624 (2015).
41. Vale, G. The responses of tsetse ies (Diptera, Glossinidae) to mobile and
stationary baits. Bull. Entomolog. Res. 64, 545588 (1974).
42. Stowers, J. R. et al. Virtual reality for freely moving animals. Nat. Methods 14,
9951002 (2017).
43. Bates, D., Maechler, M., Bolker, B. & Walker, S. Fitting Linear Mixed-Effects
Models Using lme4. J. Statis. Software 67,148 (2015).
Acknowledgements
We are grateful for the advice and discussions with C.E. Reisenman and F. Van Breugel.
Support for this project was funded by Air Force Ofce of Scientic Research under
grants FA9550-20-1-0422 (J.A.R.); the National Institutes of Health under grants R01-
AI148300 (J.A.R.), R21-AI137947 (J.A.R.); and an Endowed Professorship for Excellence
in Biology (J.A.R.). C.M. was supported by EY008117 (C.M.) from the NEI, AI165575
(C.M.) from NIAID, DC016278 (C.M.) from NIDCD, and from the U.S. Army Research
Ofce and accomplished under cooperative agreement W911NF-19-2-0026 (C.M.) for
the Institute for Collaborative Biotechnologies. A.D.S. was supported by the Momentum
program of the Volkswagen Foundation.
Author contributions
D.A.S.A., C.R., and J.A.R. designed research; D.A.S.A., C.R., and J.A.R. performed wind
tunnel experiments; C.R. and J.A.R. conducted electroretinogram experiments; A.D.S.
assisted in visual stimulus, wind tunnel software, and experimental designs; Y.Z. and
C.M. generated the opsin mutant lines. All authors wrote and edited the paper.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information The online version contains supplementary material
available at https://doi.org/10.1038/s41467-022-28195-x.
Correspondence and requests for materials should be addressed to Jeffrey A. Riffell.
Peer review information Nature Communications thanks Claudio Lazzari, Marcelo
Lorenzo and Zainulabeuddin Syed for their contribution to the peer review of this work.
Peer reviewer reports are available.
Reprints and permission information is available at http://www.nature.com/reprints
Publishers note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional afliations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the articles Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
articles Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
© The Author(s) 2022
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28195-x
14 NATURE COMMUNICATIONS | (2022) 13:555 |https://doi .org/10.1038/s41467-022-28195-x |www.nature.c om/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
... In fact, blocking the ability to detect CO2 or color filtering has proven to prevent the targeting of the skin, indicating the strict interaction between olfaction and vision in mosquito feeding behavior. This finding has thus been proposed as a promising basis for the set-up of strategies to prevent mosquito attraction to human hosts [95]. ...
Article
Full-text available
Light-based phenomena in insects have long attracted researchers’ attention. Surface color distribution patterns are commonly used for taxonomical purposes, while optically-active structures from Coleoptera cuticle or Lepidoptera wings have inspired technological applications, such as biosensors and energy accumulation devices. In Diptera, besides optically-based phenomena, biomolecules able to fluoresce can act as markers of bio-metabolic, structural and behavioral features. Resilin or chitinous compounds, with their respective blue or green-to-red autofluorescence (AF), are commonly related to biomechanical and structural properties, helpful to clarify the mechanisms underlying substrate adhesion of ectoparasites’ leg appendages, or the antennal abilities in tuning sound detection. Metarhodopsin, a red fluorescing photoproduct of rhodopsin, allows to investigate visual mechanisms, whereas NAD(P)H and flavins, commonly relatable to energy metabolism, favor the investigation of sperm vitality. Lipofuscins are AF biomarkers of aging, as well as pteridines, which, similarly to kynurenines, are also exploited in metabolic investigations. Beside the knowledge available in Drosophila melanogaster, a widely used model to study also human disorder and disease mechanisms, here we review optically-based studies in other dipteran species, including mosquitoes and fruit flies, discussing future perspectives for targeted studies with various practical applications, including pest and vector control.
... Three projectors (LG PH450U, Englewood Cliffs, NJ, United States) provide a low contrast checkerboard on the floor of the tunnel and grey horizons on each side of the tunnel by displaying on rear projection screens (SpyeDark, Spye, LLC, Minneapolis, MN, United States). A 3D real-time tracking system (Rodriguez et al., 2021;Alonso San Alberto et al., 2022) was used to track mosquito trajectories. Sixteen cameras (Basler AC640gm, Exton, PA, United States), each with an opaque Infrared (IR) Optical Wratten Filter (Kodak 89B, Kodak, Rochester, NY, United States), were mounted on top of the wind tunnel and recorded mosquito trajectories at 60 frames/s. ...
Article
Full-text available
Approximately 3.4 billion people are at risk of malaria, a disease caused by infection with Plasmodium spp. parasites, which are transmitted by Anopheles mosquitoes. Individuals with severe falciparum malaria often exhibit changes in circulating blood levels of biogenic amines, including reduced serotonin or 5-hydroxytryptamine (5-HT), and these changes are associated with disease pathology. In insects, 5-HT functions as an important neurotransmitter for many behaviors and biological functions. In Anopheles stephensi, we show that 5-HT is localized to innervation in the head, thorax, and midgut, suggesting a gut-to-brain signaling axis that could support the effects of ingested 5-HT on mosquito biology and behavioral responses. Given the changes in blood levels of 5-HT associated with severe malaria and the key roles that 5-HT plays in insect neurophysiology, we investigated the impact of ingesting blood with healthy levels of 5-HT (1.5 µM) or malaria-associated levels of 5-HT (0.15 µM) on various aspects of A. stephensi biology. In these studies, we provisioned 5-HT and monitored fecundity, lifespan, flight behavior, and blood feeding of A. stephensi. We also assessed the impact of 5-HT ingestion on infection of A. stephensi with the mouse malaria parasite Plasmodium yoelii yoelii 17XNL and the human malaria parasite Plasmodium falciparum. Our data show that ingestion of 5-HT associated with severe malaria increased mosquito flight velocity and investigation of visual objects in response to host odor (CO2). 5-HT ingestion in blood at levels associated with severe malaria also increased the tendency to take a second blood meal 4 days later in uninfected A. stephensi. In mosquitoes infected with P. y. yoelii 17XNL, feeding tendency was decreased when midgut oocysts were present but increased when sporozoites were present. In addition to these effects, treatment of A. stephensi with 5-HT associated with severe malaria increased infection success with P. y. yoelii 17XNL compared to control, while treatment with healthy levels of 5-HT decreased infection success with P. falciparum. These changes in mosquito behavior and infection success could be used as a basis to manipulate 5-HT signaling in vector mosquitoes for improved control of malaria parasite transmission.
Article
Cryptochrome (CRY) is a short-wavelength light-sensitive photoreceptor expressed in a subset of circadian neurons and eyes in Drosophila that regulates light-evoked circadian clock resetting. Acutely, light evokes rapid electrical excitation of the ventral lateral subset of circadian neurons and confers circadian-modulated avoidance behavioral responses to short-wavelength light. Recent work shows dramatically different avoidance versus attraction behavioral responses to short-wavelength light in day-active versus night-active mosquitoes and that these behavioral responses are attenuated by CRY protein degradation by constant light exposure in mosquitoes. To determine whether CRY1s mediate species-specific coding for behavioral and electrophysiological light responses, we used an “empty neuron” approach and transgenically expressed diurnal Aedes aegypti (AeCRY1) versus nocturnal Anopheles gambiae (AgCRY1) in a cry-null Drosophila background. AeCRY1 is much less light sensitive than either AgCRY1 or DmCRY as shown by partial behavioral rhythmicity following constant light exposure. Remarkably, expression of nocturnal AgCRY1 confers low survival to constant white light as does expression of AeCRY1 to a lesser extent. AgCRY1 mediates significantly stronger electrophysiological cell-autonomous responses to 365 nm ultraviolet (UV) light relative to AeCRY1. AgCRY1 expression mediates electrophysiological sensitivity to 635 nm red light, whereas AeCRY1 does not, consistent with species-specific mosquito red light responses. AgCRY1 and DmCRY mediate intensity-dependent avoidance behavior to UV light at different light intensity thresholds, whereas AeCRY1 does not, thus mimicking mosquito and fly behaviors. These findings highlight CRY as a key non-image-forming visual photoreceptor that mediates physiological and behavioral light responses in a species-specific fashion.
Article
In a recent study, Zhao et al. decipher how the olfactory system encodes human versus animal odors in the mosquito Aedes aegypti. By combining genome engineering, in vivo calcium imaging, advanced chemistry, and behavioral analysis, the authors provide compelling evidence that the discriminatory coding of host odors is surprisingly simple – and bridges labeled line with combinatorial coding.
Article
Full-text available
Predatory animals pursue prey in a noisy sensory landscape, deciding when to continue or abandon their chase. The mosquito Aedes aegypti is a micropredator that first detects humans at a distance through sensory cues such as carbon dioxide. As a mosquito nears its target it senses more proximal cues such as body heat that guide it to a meal of blood. How long the search for blood continues after initial detection of a human is not known. Here we show that a 5-second optogenetic pulse of fictive carbon dioxide induced a persistent behavioral state in female mosquitoes that lasted for more than 10 minutes. This state is highly specific to females searching for a blood meal and was not induced in recently blood-fed females or in males, who do not feed on blood. In males that lack the gene fruitless , which controls persistent social behaviors in other insects, fictive carbon dioxide induced a long-lasting behavior response resembling the predatory state of females. Finally, we show that the persistent state triggered by detection of fictive carbon dioxide enabled females to engorge on a blood meal mimic offered up to 14 minutes after the initial 5-second stimulus. Our results demonstrate that a persistent internal state allows female mosquitoes to integrate multiple human sensory cues over long timescales, an ability that is key to their success as an apex micropredator of humans.
Article
Full-text available
Drosophila melanogaster has long been a popular model insect species, due in large part to the availability of genetic tools and is fast becoming the model for insect colour vision. Key to understanding colour reception in Drosophila is in-depth knowledge of spectral inputs and downstream neural processing. While recent studies have sparked renewed interest in colour processing in Drosophila, photoreceptor spectral sensitivity measurements have yet to be carried out in vivo. We have fully characterised the spectral input to the motion and colour vision pathways, and directly measured the effects of spectral modulating factors, screening pigment density and carotenoid-based ocular pigments. All receptor sensitivities had significant shifts in spectral sensitivity compared to previous measurements. Notably, the spectral range of the Rh6 visual pigment is substantially broadened and its peak sensitivity is shifted by 92 nm from 508 to 600 nm. We show that this deviation can be explained by transmission of long wavelengths through the red screening pigment and by the presence of the blue-absorbing filter in the R7y receptors. Further, we tested direct interactions between inner and outer photoreceptors using selective recovery of activity in photoreceptor pairs.
Article
Full-text available
Color vision is widespread among insects but varies among species, depending on the spectral sensitivities and interplay of the participating photoreceptors. The spectral sensitivity of a photoreceptor is principally determined by the absorption spectrum of the expressed visual pigment, but it can be modified by various optical and electrophysiological factors. For example, screening and filtering pigments, rhabdom waveguide properties, retinal structure, and neural processing all influence the perceived color signal. We review the diversity in compound eye structure, visual pigments, photoreceptor physiology, and visual ecology of insects. Based on an overview of the current information about the spectral sensitivities of insect photoreceptors, covering 221 species in 13 insect orders, we discuss the evolution of color vision and highlight present knowledge gaps and promising future research directions in the field. Expected final online publication date for the Annual Review of Entomology, Volume 66 is January 8, 2021. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
Article
Full-text available
Color vision is an important sensory capability that enhances the detection of contrast in retinal images. Monochromatic animals exclusively detect temporal and spatial changes in luminance, whereas two or more types of photoreceptors and neuronal circuitries for the comparison of their responses enable animals to differentiate spectral information independent of intensity. Much of what we know about the cellular and physiological mechanisms underlying color vision comes from research on vertebrates including primates. In insects, many important discoveries have been made, but direct insights into the physiology and circuit implementation of color vision are still limited. Recent advances in Drosophila systems neuroscience suggest that a complete insect color vision circuitry, from photoreceptors to behavior, including all elements and computations, can be revealed in future. Here, we review fundamental concepts in color vision alongside our current understanding of the neuronal basis of color vision in Drosophila, including side views to selected other insects.
Article
Full-text available
The fruit fly Drosophila melanogaster can process chromatic information for true color vision and spectral preference. Spectral information is initially detected by a few distinct photoreceptor channels with different spectral sensitivities and is processed through the visual circuit. The neuroanatomical bases of the circuit are emerging. However, only little information is available in chromatic response properties of higher visual neurons from this important model organism. We used in vivo whole-cell patch-clamp recordings in response to monochromatic light stimuli ranging from 300 to 650 nm with 25-nm steps. We characterized the chromatic response of 33 higher visual neurons, including their general response type and their wavelength tuning. Color-opponent-type responses that had been typically observed in primates and bees were not identified. Instead, the majority of neurons showed excitatory responses to broadband wavelengths. The UV (300–375 nm) and middle wavelength (425–575 nm) ranges could be separated at the population level owing to neurons that preferentially responded to a specific wavelength range. Our results provide a first mapping of chromatic information processing in higher visual neurons of D. melanogaster that is a suitable model for exploring how color-opponent neural mechanisms are implemented in the visual circuits.
Article
Full-text available
Inflorescence patterns of ultraviolet (UV) absorption and UV-reflection are attractive to many insect pollinators. To understand whether UV inflorescence cues affect the attraction of nectar-foraging mosquitoes, we worked with the common house mosquito, Culex pipiens and with two plant species exhibiting floral UV cues: the tansy, Tanacetum vulgare, and the common hawkweed Hieraciumm lachenalii. Electroretinograms revealed that Cx. pipiens eyes can sense UV wavelengths, with peak sensitivity at 335 nm. Behavioural bioassays divulged that UV inflorescence cues enhance the attractiveness of inflorescence odour. In the presence of natural floral odour, female Cx. pipiens were attracted to floral patterns of UV-absorption and UV-reflection but preferred uniformly UV-dark inflorescences. Moreover, Cx. pipiens females preferred UV-dark and black inflorescence models to UV-dark and yellow inflorescence models. With feathers and pelts of many avian and mammalian hosts also being UV-dark and dark-coloured, foraging Cx. pipiens females may respond to analogous visual cues when they seek nectar and vertebrate blood resources.
Article
Full-text available
There is increasing interest in flies as potentially important pollinators. Flies are known to have a complex visual system, including four spectral classes of photoreceptors that contribute to the perception of colour. Our current understanding of how colour signals are perceived by flies is based on data for the blowfly Lucilia sp., which after being conditioned to rewarded monochromatic light stimuli, showed evidence of a categorical colour visual system. The resulting opponent fly colour space has four distinct categories, and has been used to interpret how some fly pollinators may perceive flower colours. However, formal proof that flower flies (Syrphidae) only use a simple, categorical colour process remains outstanding. In free-flying experiments we tested the hoverfly Eristalis tenax, a Batesian mimic of the honeybee, that receives its nutrition by visiting flowers. Using a range of broadband similar-dissimilar colour stimuli previously used to test colour perception in pollinating hymenopteran species, we evaluated if there are steep changes in behavioural choices with continuously increasing colour differences as might be expected by categorical colour processing. Our data revealed that colour choices by the hoverfly are mediated by a continuous monotonic function. Thus, these flies did not use a categorical processing, but showed evidence of a colour discrimination function similar to that observed in several bee species. We therefore empirically provide data for the minimum colour distance that can be discriminated by hoverflies in fly colour space, enabling an improved understanding of plant-pollinator interactions with a non-model insect species.
Article
Full-text available
Background Opsins are light sensitive receptors associated with visual processes. Insects typically possess opsins that are stimulated by ultraviolet, short and long wavelength (LW) radiation. Six putative LW-sensitive opsins predicted in the yellow fever mosquito, Aedes aegypti and malaria mosquito, Anopheles gambiae, and eight in the southern house mosquito, Culex quinquefasciatus, suggest gene expansion in the Family Culicidae (mosquitoes) relative to other insects. Here we report the first detailed molecular and evolutionary analyses of LW opsins in three mosquito vectors, with a goal to understanding the molecular basis of opsin-mediated visual processes that could be exploited for mosquito control. ResultsTime of divergence estimates suggest that the mosquito LW opsins originated from 18 or 19 duplication events between 166.9/197.5 to 1.07/0.94 million years ago (MY) and that these likely occurred following the predicted divergence of the lineages Anophelinae and Culicinae 145–226 MY. Fitmodel analyses identified nine amino acid residues in the LW opsins that may be under positive selection. Of these, eight amino acids occur in the N and C termini and are shared among all three species, and one residue in TMIII was unique to culicine species. Alignment of 5′ non-coding regions revealed potential Conserved Non-coding Sequences (CNS) and transcription factor binding sites (TFBS) in seven pairs of LW opsin paralogs. Conclusions Our analyses suggest opsin gene duplication and residues possibly associated with spectral tuning of LW-sensitive photoreceptors. We explore two mechanisms - positive selection and differential expression mediated by regulatory units in CNS – that may have contributed to the retention of LW opsin genes in Culicinae and Anophelinae. We discuss the evolution of mosquito LW opsins in the context of major Earth events and possible adaptation of mosquitoes to LW-dominated photo environments, and implications for mosquito control strategies based on disrupting vision-mediated behaviors.
Article
Mosquitoes rely on the integration of multiple sensory cues, including olfactory, visual, and thermal stimuli, to detect, identify and locate their hosts [1-4]. Although we increasingly know more about the role of chemosensory behaviours in mediating mosquito-host interactions [1], the role of visual cues remains comparatively less studied [3], and how the combination of olfactory and visual information is integrated in the mosquito brain remains unknown. In the present study, we used a tethered-flight LED arena, which allowed for quantitative control over the stimuli, to show that CO2 exposure affects target-tracking responses, but not responses to large-field visual stimuli. In addition, we show that CO2 modulates behavioural responses to visual objects in a time-dependent manner. To gain insight into the neural basis of this olfactory and visual coupling, we conducted two-photon microscopy experiments in a new GCaMP6s-expressing mosquito line. Imaging revealed that the majority of ROIs in the lobula region of the optic lobe exhibited strong responses to small-field stimuli, but showed little response to a large-field stimulus. Approximately 20% of the neurons we imaged were modulated when an attractive odour preceded the visual stimulus; these same neurons also elicited a small response when the odour was presented alone. By contrast, imaging in the antennal lobe revealed no modulation when visual stimuli were presented before or after the olfactory stimulus. Together, our results are the first to reveal the dynamics of olfactory modulation in visually evoked behaviours of mosquitoes, and suggest that coupling between these sensory systems is asymmetrical and time-dependent.
Article
Standard animal behavior paradigms incompletely mimic nature and thus limit our understanding of behavior and brain function. Virtual reality (VR) can help, but it poses challenges. Typical VR systems require movement restrictions but disrupt sensorimotor experience, causing neuronal and behavioral alterations. We report the development of FreemoVR, a VR system for freely moving animals. We validate immersive VR for mice, flies, and zebrafish. FreemoVR allows instant, disruption-free environmental reconfigurations and interactions between real organisms and computer-controlled agents. Using the FreemoVR platform, we established a height-aversion assay in mice and studied visuomotor effects in Drosophila and zebrafish. Furthermore, by photorealistically mimicking zebrafish we discovered that effective social influence depends on a prospective leader balancing its internally preferred directional choice with social interaction. FreemoVR technology facilitates detailed investigations into neural function and behavior through the precise manipulation of sensorimotor feedback loops in unrestrained animals.
Chapter
Interest in mosquito ecology and the use of appropriate sampling methods commenced at the beginning of the nineteenth century with the discovery that mosquitoes could act as vectors of diseases to humans and domestic animals. Since those early days, the science and technology of mosquito sampling have developed apace, as evidenced by the increase in size and scope of successive editions of this book.