Introduction

Mosquito bites transmit diverse pathogens, including viruses, unicellular organisms, and even multicellular nematodes1. More than half a million people die of malaria each year; most are young children2. Controlling mosquito populations and these diseases remains a global problem3. Mosquito-borne diseases can be epidemic and spread rapidly with changes in agriculture and migration4. Humans who labor outside in tropical climates are at highest risk. Mosquito populations, in part, are controlled by insecticides which promote resistance and are detrimental to the environment5,6. Recent advances in mosquito genetics and biological control can lower mosquito populations without insecticides but still change the earth’s ecology in safe, modest ways by changing mosquito genetics or adding symbiont populations7,8,9,10,11,12,13. Textiles have been shown to be a pragmatic deterrent in reducing or eliminating bites from mosquito vectors in the form of bednets14. Furthermore, recent research reported the construction of mosquito bite blocking textiles15. Importantly, these textile applications are safe16. We found that modern clothing does not stop mosquito proboscises; some clothing options reduce the wearer's ability to perceive mosquito landing events. Popular form-fitting athletic “heat-gear” exacerbates the problem and does not block bites.

Female mosquitos feed with piercing/sucking proboscises (Fig. 1a). The proboscis has an outer labium (Lab). At feeding, the labium retracts, exposing the fascicle, which is a repertoire of six serrated blades and microneedles bound together by liquid surface tension (Fig. 1b)17,18,19. The labrum (Lm) is a beveled needle that pierces and draws blood. Adjacent to the labrum are paired mandibular (Md) and maxillary (Mx) stylets. Maxillary stylets saw skin at a vibrational frequency of 30 Hz to reduce the force needed to puncture20. The flexible fascicle can bend at 90° angles and is innervated and controlled by delicate musculature18,21. The measurements of the proboscis in Aedes aegypti (the yellow fever mosquito) are 2.32 mm long and 60 μm wide. The labrum is 25 µm in diameter17,18,22,23,24.

Fig. 1: Testing common textiles for mosquito bite blocking.
figure 1

a Mosquito head and mouthparts. b Enumeration and dimensions of the mosquito fascicle’s microneedles. c Needles of flatbed knitting machines with order of movement numbered. d Screen of common textile’s ability to block mosquitos, comparing weave and knit sleeves. Each dot represents data from one replicate experiment with a cage of 20 unique females. Number of bites are quantified on the y-axis. Absence of asterisks indicates non-significance by analysis of variance (ANOVA). e Quantification of detected landing events on a sleeved arm. Shown are three biological replicates with four technical replicates each. f Pixel quantification of garment bite risk, defined as where either clothing clings to skin or skin is uncovered (red). Asterisk (*) indicates significance (p < 0.05) by ANOVA in (d) or t-test in (e). Graphs plots mean and standard deviation. Lab labium, Lm labrum, Mx maxillary stylets, Md mandibular stylets, Hp hypopharynx.

Full size image

Modern clothing is manufactured as weaves or knits (Fig. S1). Weaving interlaces multiple weft and warp fibers, whereas knitting constructs recursive loops from a single fiber forming courses (rows) and wales (columns). Unique knit patterns can be described through symbol diagrams that convey the knit geometry, termed knit diagrams (Fig. S1c–e)25,26,27. The written knit code can be translated into robotic machine primitives interpretable by modern flatbed computer numerical control (CNC) knitting machines, which can knit complicated structures with simple up-down needle movements (Fig. 1c). There are nearly infinite knit configurations and fiber inputs. Thus, the search space of possible knit permutations is vast. Our hypothesis was simply that certain textile configurations would block mosquito bites and others would not. We sought to test and define them.

Results and discussion

We performed an initial blocking screen on common clothing. Experiments consisted of placing an arm with the sleeve in a cage of 20 female mosquitos for 15 min. We quantified the number of bites received. Five common weaves tested did not block, but one knit did (Figs. 1d and S2). Notably, modern clothing, including Under Armour compression heat gear, Nike socks, and two garments that advertise insect protection including Rynoskin and a protective horse mesh, did not block. Microscopy revealed that these textiles were full of spaces through which mosquitos could probe (Fig. S2). Under Armour also reduces the perception of mosquito landing events (Fig. 1e); as such, it is worse than exposed bare skin. Mosquitos easily pierce clothes, so we quantified where clothing clings to skin for the most common garments in males and females (red in Fig. 1f). Notably, the protection afforded from long sleeves is not much better than that of short sleeves because both cling to the skin in large areas of the upper back and shoulders where mosquitos are attracted to bite (red regions of garments in Fig. 1f and red dots in Fig. 4e). How clothing fits individual bodies is also a factor relative to mosquito bites (Fig. 1f).

Results from initial experiments indicated that knits were capable of blocking mosquito bites with variable efficacy. We sought to determine which features and parameters created the blocking effect. We then screened eight distinct knit geometries (Figs. 2 and S3). We also simulated these knit geometries to facilitate geometric visualization and clarity (Fig. S3). We observed that heat from drying in a standard wash–dry cycle shrunk polyester knits. In the interlock knit, post-knit heat treatment from drying converted a non-blocker into a blocker by shrinking inter-wale and inter-loop space, resulting in a reduction of surface area by ~3.322 cm2 (16.5%) (Fig. 2i, j). Thereafter, all knits we tested were heat treated via a wash/dry cycle at 327 K. Of eight knits screened, only one was blocked, which was interlock. The interlock knit uniquely positions interlocking loops on top of each other (Fig. 2d).

Fig. 2: Bite blocking data of all knitted textiles developed at Auburn University using the 100% polyester control yarn of 282 microns.
figure 2

a–h Microscopy images and knit diagrams of each knit developed. Scale bar indicates 5 mm distance. i Microscopy images of an unwashed (top) and washed (bottom) interlock knit. Microscopy images are zoomed in at a greater scale. Red Arrows are used to highlight significant points of shrinking which enhance the knits bite blocking abilities. White arrows point out space that remains constant. j Graph of the total surface area of an interlock textile measured before and after washing. This experiment was conducted 5 times, each red dot representing a single interlock piece that was washed, dried, and then measured. k Graph of the number of bites received during a single 15-min experiment. Each red dot corresponds to one cage of 20 females, and each sleeve was tested a minimum of three times. Carrot (^) indicates washed. Asterisk (*) indicates p < 0.05. Graphs plot mean and standard deviation.

Full size image

Because we were able to convert a non-blocking knit into a blocking knit, we sought to define treatments and parameters that improve blocking. We discovered three other parameters capable of enhancing mosquito blocking. Increasing thread thickness converted a single-jersey knit into a blocking knit (Fig. 3a). Increasing spandex content converted jersey-skip knits to blockers (Fig. 3b). Finally, decreasing stitch length enhanced the blocking of the interlock knit (Fig. 3c). When mosquitos land on blocking knits they tend to probe more, though total probing time was less because they fly away when they cannot get bloodmeals (Fig. 3d). We tested our blocking knits against Aedes aegypti and Psorophora howardii (Fig. 3d, e). Psorophora howardii is a large mosquito species native to the Southeastern United States with a proboscis averaging 8–10 mm in size, much larger than Aedes aegypti (Fig. S4). None of the textiles tested were thicker than the length of tested mosquito proboscises (Fig. S5). This shows that our knits block more than one species of mosquito. We conclude that unique recipes and optimizations are often required to generate the blocking effect. Moreover, programmed knit diagrams often come out geometrically different postproduction because compressive forces alter the structures, as observed with interlock. In future studies, we will investigate the material science of fibers and explore fiber coatings and treatments that impact stretchability and proboscis penetration.

Fig. 3: Determining parameters controlling bite blocking of diverse mosquitos.
figure 3

a Increasing fiber diameter enhances bite blocking in single jersey knits. Numbers above textiles indicate fiber diameter in µm. Scale bar is 5 mm for all images. As before, red dots are independent mosquito bite experiments with 20 females. b Increasing spandex content compresses knit conformations and enhances bite blocking in jersey-skip knits. The final panel (3% spandex†) is constructed of 3% spandex, 19% polyester, and 78% cotton. c Decreasing stitch length enhanced bite blocking in interlock knits. d Quantification of Aedes aegypti probing on interlock (stitch length—10) vs. bare arm. Red dots indicate the number of probes from an individual mosquito (left); and time spent probing in seconds (s) for individual mosquitos (right). e Similar probing experiments with Psorophora howardii on interlock (Bare arm probes near axis are 1 not 0) (stitch length—10) vs. bare arm. Asterisk (*) indicates significance where p < 0.05. All graphs show mean and standard deviation.

Full size image

We sought to measure and engineer the comfort of blocking textiles (Figs. 4 and S6). We measured comfort by a combination of experiments, including a 9-factor comfort score including grittiness, fuzziness, thickness, tensile stretch, hand friction, fabric-to-fabric friction, force to compress, stiffness, and noise intensity 28,29. Our three blocking textiles, Acrylic Single Jersey, Jegging, and Interlock, ranged in comfort (Fig. 4a–d). We found that blockers were near or even better than Under Armour control in some comfort measures. Interlock was lower in comfort score because of its stiffness, thickness, and lack of stretch (which, in turn, makes it a better blocker). Further iterations of interlock can incorporate spandex/elastic or alternate fibers to increase comfort. Jersey-skips with spandex were the best blockers and were extremely close in comfort to Under Armour due to its mixture of materials (spandex, polyester, and cotton). Mosquitos have feeding/taste preferences30. Because mosquitos target the lower body parts and backside for biting (Fig. 4e), textiles might be engineered to include blocking knits in regions highly attractive to mosquitos and looser, more comfortable knits in regions unattractive to mosquitos. Garments could also be patterned with colors that are less attractive to mosquitos, like white (Fig. 4f)31,32,33.

Fig. 4: Engineering comfort of bite-blocking textiles.
figure 4

a Mean comfort score of combined 9-factor feel tests on textiles. A higher score indicates higher comfort. Red is a comfortable Under Armour control, and green is blocking textiles. Data represent triplicates. b Average heat gained on the skin beneath textile sleeves. A lower score indicates higher comfort (less heat gained). c Average air permeability of corresponding textile sleeves. Y-axis is the distance of salt particles moved by air passing through the textile at 137.9 kPa (20 psi). A higher score indicates higher comfort (more airflow). Data represents a mean of n = 5. Graphs display min and max values with lines representing mean. d 3-dimensional comfort graph of textiles. Colors are the same as above. Arrow direction indicates increasing comfort. e Heat map of mosquito landing events (red dots). Left is anterior and right is posterior. Data represents an overlay of three independent experiments. f Choice tests of mosquito landing events on black vs white sleeve regions. Graphs plot mean with standard deviation. Asterisk (*) indicates significance where p < 0.05. g SIR Dengue model over time showing different percentages of individuals protected by bite blocking textiles. Red, green, and black lines indicate 0%, 75%, and 100% of individuals protected.

Full size image

While the immediate benefits of our textiles are clear, we also wanted to understand their potential impact in a real-world outbreak scenario34. We simulated a dengue outbreak in a susceptible infected recovered (SIR) model with varying levels of textile protection applied to a sample population of 6 million individuals (Fig. 4g). Limitations of the model include the ability to portion the body with respect to textile coverage, recovered individuals cannot be reinfected, and the model does not account for spatial distribution. In the SIR model, as more individuals wear the protective garment, the daily infection rate decreases, delaying and reducing the outbreak peak.

Overall, we showed that modern comfortable textiles can be engineered to block mosquito bites. Effective blocking does not require a compromise in comfort. Our discoveries arm individuals with the power to protect themselves from vector-borne diseases in hot climates. The manufacturing process of these textile garments reduces human labor and environmental impact. For additional data and analyses, please refer to Supplementary Data.

Methods

Knitting garments

We used M1 Plus (Stoll) to design knit pattern files. Pattern files were loaded into an ADF 530-16 Ki BcW flatbed knitting machine (Stoll, Reutlingen, Baden-Wurttemberg, Germany). Knitted sheets were cut and sewn using a CS7000X sewing machine (Brother, Bridgewater, NJ) into sleeves to fit comfortably on the experimenter’s arm. Next, hems are created, and raw edges are finished by a 1034D serger machine (Brother, Bridgewater, NJ). A standard control yarn was used for all knits in Fig. 2. The control yarn is 100% polyester (UNIFI 2/150/96) (number of plies/denier of each ply/number of filaments in each ply) with a diameter of 282 microns (Unifi, Greensboro, NC). Other yarns used in the development of textiles throughout the course of this research were also acquired from Unifi. Thickness (mm) of each knitted textile was measured with electronic calipers.

Mosquitos

Aedes aegypti (Linnaeus, Rockefeller strain) mosquitos were reared in-house within a pathogen-free insectary. Mosquitos were kept at 28 °C (70% Relative Humidity) with a rotating 12-h light/dark cycle at relative. Mosquito eggs are hatched by submerging egg papers in medium-sized shoebox tubs until pupation. Larvae and pupae are fed ~5 mL of a baker’s yeast and water mixture. Pupae are transferred by hand into a mesh cage for eclosion. As a control for age, pupae are allowed to eclose for 72 h, then removed and placed into a new cage; this ensures all mosquitos within a cage are within 3 days age of each other. Mosquito biting experiments were performed with females aged 4–7 days old. Every mosquito experiment was conducted under the same controlled parameters. The mosquitos were first anesthetized on ice and sorted using a cold block. The females were then placed in a cage with only water, starving them for 8–12 h prior to the experiment. To maintain a normal mosquito circadian rhythm, all experiments were conducted in the afternoon. For each experiment, an experimenter wears a predetermined knitted test sleeve and covers their hand with a latex glove, excluding bare-arm controls. The covered arm was placed in a cage of 20 female mosquitos for 15 min. After experiments, both bites and percent blood-fed females were recorded. Bites are characterized as puffy reddish bumps arising on the skin shortly after a mosquito has punctured the skin. Blood-fed females are counted if the female mosquito indicates engorgement of the abdomen with blood.

For full body tests, 40 females are sorted in the same manner. The experimenter dressed in long white sleeves and pants and stood in a full-body cage for 15 min. The landing events were recorded by two outside observers, one in the front and one in the back. Each time a mosquito landed and attempted to probe; a mark was recorded on a human body outline corresponding to that area. We opted to utilize human observers in lieu of cameras because the experiment is more reproducible if expensive equipment is not required. This experiment performed three separate replications on three separate cages of 40 mosquitos (see Fig. 4e). The replicates were digitalized and overlayed in Adobe Illustrator to produce a comprehensive heat map of landing events. Landing events on arms wearing Under Armour were similarly tracked by two outside observers with a small arm cage of 5 female mosquitos, and a duration of 15 min (see Fig. 1e). Experiments involving bites to humans were conducted under approved AU IRB Protocol #21-278 FB.

To acquire video captures of mosquito feeding and biting, 30 female Aedes aegypti mosquitos were transferred into a cage with two arm openings. The experimenter wore a test sleeve and latex gloves. A Moment Macro Lens V2 iPhone camera attachment was used to acquire microscopic videos of mosquito probing behavior. Each video is recorded, tracking the behavior of a single female for 1 min. For Aedes aegypti 30 videos were filmed for each treatment, each video is representative of an independent biological replicate. For Psorophora howardii, only four videos were filmed representing independent biological replicates due to the difficulty of collecting and rearing wild cannibalistic Psorophora. The videos were analyzed, and two data sets were generated: time to fly away and number of probes. Video analysis concludes at 60 s and any data collected within that time ceases. Color choice landing events were tracked using half-white/half-black sleeves. Twenty female Aedes aegypti mosquitos were allowed to land and probe on either side. Five iterations of this experiment were conducted, rotating the sleeve incrementally to control for lighting and behavioral predispositions.

Psorophora howardii (Coquillett) mosquitos were field collected as larvae from flood water in Alabama and reared to adult stages. Larvae were fed 5 mL of liquified fish food solution and supplemented with a carnivorous diet of Culex spp. larvae. After eclosion, they were sex-sorted and set up in video capture experiments in the same manner as Aedes, with the exception that only four mosquitos were used per experiment, due to the difficulty of collecting/rearing Psorophora.

Comfort testing

The hand feel test is a test of perceived sensory comfort. Within this test, nine factors are measured on a scale from 1 to 1028,29. In each test, two textiles are used as controls that equate to the minimum and maximum values on the scale. The nine factors being tested are gritty, fuzzy, thickness, tensile stretch, hand friction, fabric-to-fabric friction, force to compress, stiffness, and noise intensity (the sound a fabric makes as it rubs on the skin). Each sleeve was tested for all factors three times with different individuals. The tests were blind. All values from all tests can be averaged to generate an average “comfort score” shown in Fig. 4a. Sleeves with higher comfort scores are intuitively more comfortable to humans.

Thermal heat absorbed and released by a textile was measured by the following experiment. An experimenter would place a sleeve on their arm and enter a 28 °C (70% relative humidity) incubator for 15 min. Four total digital thermometer readings were taken, two before (on skin and sleeve) and two after incubation (on skin and sleeve). The difference in temperatures on the skin and sleeve is determined post-experiment, and these values were graphed as heat gained.

Air permeability quantifies how breathable a textile is. To measure air permeability, we pass compressed air at 137.9 kPa through the textile and measure its force on the other side by quantifying its ability to move a substrate (20 g salt) a given distance. Distance of the furthest particle is graphed in Fig. 4c. This experiment is conducted on a black table incrementally marked in inches. Textiles that allow for more air to pass through are more comfortable.

Modeling textile exposure to mosquito bites

To determine regions of clothing where textile closely contacts skin (Fig. 1f), we acquired black and white images of average-type individuals wearing common garments, including shorts, t-shirts, pants, dresses, leggings, long-sleeve shirts, and compression shirts. We faux colored pixels where textiles were touching skin red (leaving other areas white) and counted the colored pixel area in Adobe Photoshop using the Adobe selection tool and measurement log. Using front and back images in triplicate, a two-dimensional surface area approximation could be determined for areas of the body exposed to mosquito bites for each garment type.

Microscopy

All microscopy images were taken using a Nikon SMZ1270 stereomicroscope with a Nikon DS-Fi3 camera. The diameter of each yarn was determined by taking two microscopy images of a knitted fabric with and without a mm ruler at the exact same scale. Adobe Photoshop was used to create a digital micrometer scale. The thickness of each yarn was measured at least three times using this scale. Scale bars used in microscopy figures were created by using a millimeter ruler during imaging. A scale bar was made in Photoshop and added to each figure.

Statistics and reproducibility

All data was recorded and analyzed using GraphPad Prism 9 or Microsoft Excel. Two different statistical analyses were performed. First, for any experiment that yielded more than two sets of results, an ordinary one-way ANOVA and Tukey’s multiple comparisons tests were performed for significance. For experiments with only two data sets, an unpaired, nonparametric, Mann–Whitney t-test was performed for significance. Any comparison yielding a P-value < 0.05 is considered significant.

Modeling dengue transmission

We reconstructed a well-known Dengue SIR model in R34. The model assumes that individuals are infected only once and recover with no chance of reinfection. The bite rate factor has been adjusted to reflect individuals wearing protective clothing, with the percentage of the population determining the change in bite rate. The mosquito’s bite rate is influenced by both the variable ‘b’, representing the average number of daily bites a mosquito delivers, and ‘p_value’, which denotes the fraction of the population still susceptible to mosquito bites despite wearing a protective textile. A ‘p_value’ of 0.2 suggests that only 20% of individuals remain vulnerable, meaning 80% are effectively shielded by the textile. When a mosquito attempts to bite an individual, the combined effect of ‘b’ and ‘p_value’ determines the likelihood of successful transmission. In scenarios where the mosquito targets someone protected, its attempt to spread the disease fails, highlighting the textile’s protective role within the model’s simulated environment.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.