Assessing mosquito activity and wingbeat frequency in wildtype and genetically modified Anopheles coluzzii strains

Abstract:

Anopheles coluzzii mosquitoes are a major malaria vector in Sub-Sahelian Africa, where more than 90% of all global malaria cases occur every year. Despite the widespread application of current methods for malaria reduction, more than 600,000 people died of malaria in 2024, highlighting the need for new approaches. The University of California Malaria Initiative (UCMI) has developed a genetically engineered mosquito (GEM) carrying two genes that prevent  Plasmodium infection in the GEM. This study aims to evaluate whether genetic changes affect other aspects of mosquito biology, such as locomotion and wingbeat frequency. Wingbeat frequency data was collected via a recording device and analyzed using a custom MATLAB script. Locomotion data was collected initially using the Drosophila Activity Monitor system (DAM system) in order to establish general activity trends. Additionally, a more precise system for locomotion analysis, the Flybox, was constructed. Locomotion and wingbeat traits are important for fundamental behaviors, including mating, dispersal, and blood-feeding, and therefore influence mosquito population dynamics in the wild. Experiments compared two strains of An. coluzzii: wild-type and GEM. 

Introduction:

The World Health Organization estimated around 282 million malaria cases occurred in 2024, 95% of which occurred in the Sub-Saharan African region [17]. The primary vectors in this region are members of the Anopheles gambiae species complex [18]. While a variety of methods have been developed in order to combat malaria transmission rates, they have been unable to eradicate malaria in this region. Common methods such as the use of piperonyl butoxide (PBO) treated bednets, have begun to decrease in efficiency as Anopheles populations in Sub-Saharan Africa have begun to evolve Pyrethroid resistance [4]. Other common vector control methods, such as habitat control and environmental management, have been difficult to maintain on a large scale due to widespread urbanization, economic strain, and the need for frequent re-application of chemical control strategies [4,8]. For these reasons, next-generation gene editing technologies are being considered for field applications to aid in malaria elimination. The UCMI GEM mosquito is designed to equip a field population with malaria resistance. Deployment of this GEM may reduce transmission and ultimately eradicate malaria while minimizing impact on the natural balance of the ecosystem in which these mosquitoes reside. UCMI’s GEMs are able to mate with wild-type (WT) mosquitoes, passing down their malaria resistant genes to 99% of their offspring [2]. Other transgenic mosquitoes are designed to control mosquito-borne disease by curbing existing populations [7]. One benefit of the UCMI GEM mosquito is potential for reduced ecological impact, as mosquitoes will not be removed from the environment. However, these GEMs must successfully survive, mate, and persist for multiple generations in order to succeed as a malaria control tool.

Previous studies have shown that mosquito behaviors such as mating, feeding, locomotion, and oviposition have marked circadian patterns [15]. Mosquito locomotion habits impact swarming and host-seeking behavior [15,6]. Systems such as the Drosophila Activity Monitoring (DAM) system and the Flybox, both originally developed to monitor the locomotion activity of Drosophila, have been adapted for use in monitoring and comparing the behavior of GEM and WT An. coluzzi mosquitoes. The DAM system is equipped with an infrared (IR) beam that shines through the middle of a glass test tube. Each time a mosquito in the tube crosses that beam, it is registered as a movement by the system’s corresponding software in intervals of one minute [6,10]. The DAM is the most established of the three methods this experiment used to collect mosquito movement data, but the data it gathers is the least specific. GEM and WT movement were both measured using this method. In order to simulate real-world conditions, the DAM system was placed in a temperature and humidity-controlled chamber, with lighting to stimulate the diurnal cycle. Although the DAM system provided preliminary data on movement differences, a more robust system was needed to understand mosquito activity levels. Therefore, the FlyBox was built in-house following the author’s manual [11]. The FlyBox includes a dark chamber with user-controlled lighting that simulates light-dark conditions and an IR camera that tracks mosquito movement precisely through the experimental enclosure [14]. Data from the DAM and FlyBox can be combined in order to understand 24-hour locomotion activity patterns.

In addition to an examination of locomotion activity, we also monitored mosquito swarms in order to determine differences between GEM and WT mating behavior. During mating, which typically occurs at dusk, male mosquitoes gather and form swarms in a fixed location to act as mates for a handful of female mosquitoes that enter the swarm sporadically [15]. Mosquitoes’ wingbeat frequency impacts their mating success, as a male mosquitoes must differentiate a female frequency amongst the ambient male frequencies in order to mate. The main flight tone in mosquitoes is considered the fundamental frequency, or the lowest frequency in a harmonic series [3]. During mating, male and female wingbeats undergo a harmonic convergence, a phenomenon where the two sexes slightly alter their fundamental frequencies in order to achieve a fixed integer ratio resulting in a stronger, synchronized audio [3]. In order to ensure that female GEM mosquitoes are able to compete with the WT population, it is necessary to ensure that the wingbeat frequency of the GEM and WT females are not distinctly different. Differences in wingbeat frequency could result in unsuccessful mating between GEM and WT individuals during mating swarms. 

Ultimately, understanding of circadian rhythms and wingbeat frequencies of the GEM mosquitoes informs the predicted reproductive success of a GEM population post-release.

Materials & Methods:

1. Wingbeat Frequency

1.1 Mosquito Preparation

Prior to the experiment, a batch of adult An. coluzzii mosquitoes were separated by strain (WT & GEM) and sex (male and female) and aspirated into four 1x1-ft cubical cages with stockinette sleeves. Each cage was initially filled with exactly 100 mosquitoes prior to the day of the experiment. Each cage was then labeled with the strain, sex, and the number of mosquitoes in the cage. A cup filled with 10% sucrose solution was placed in the cage as a food source. Cages were then held in a temperature(28˚C) and humidity(80%) controlled chamber till the day of the experiment.  On the day of the experiment, mosquitoes that had died were counted and the precise number of mosquitoes was noted, ensuring that each cage contained between 50-100 mosquitoes.

1.2 Audio Data Collection

Figure 1. Cage Set Up For Audio Data Collection 

Wingbeat frequency was recorded with an ELEKHOMI 32GB Digital Voice Recorder (ELEKHOMI, China). The audio device was inserted into the cage through the sleeve. A rubber band was clamped around the arm and sleeve (Fig.1) to prevent mosquitoes from escaping through the gap in the stockinette sleeve. The audio device was then carefully maneuvered by hand to collect approximately 5-minute inflight audio data sample, with a 1-minute interval of buffer data collected pre and post trial to account for any extraneous noises caused by the insertion and removal of the recording device. The data was then trimmed down to an exact length of 5 minutes, cutting out the data from the buffer period, and saved as a .wav file.  

1.3 Data Analysis

AI (OpenAI, 2025) was used to aid the development and debugging of a custom MATLAB script in order to isolate the fundamental frequency of the collected wingbeat data. All AI generated code and changes to existing code were checked by the authors for accuracy. The audio file was filtered down into a general broad frequency range of 300-900 Hz, filtering out extraneous noise outside of the known range of wingbeat frequencies [3,12]. A fast fourier transform (FFT; Signal Processing Toolbox; MATLAB R2024B) was then applied to bring the data into the frequency range and the fundamental harmonic was extracted from the data [1]. The extracted fundamental harmonic was then compared against established fundamental frequencies for the A. aegypti species complex [3]. The general observed frequency range for An. gambiae complex lays around 350-450 Hz and the general range for A.aegypti mosquitoes lays around 400-650 Hz [11]. Past literature does not specify ranges for the fundamental harmonics of the two sexes within the An. gambiae complex. However, these fundamental frequencies have been previously observed and recorded for A. aegypti mosquitoes in non-swarming conditions, and have been established to be around the 500 Hz range for females and the 400 Hz range for males [3,13]. It is expected that the fundamental frequency for the An. gambiae complex would lay in the same range, with values on the lower end of the range [11]. 

2. Drosophila Activity Monitor (DAM) Experiment

2.1 Locomotion Activity Monitor Set-Up

Locomotion activity was first monitored using a Trikinetics Locomotion Activity Monitor (LAM 25) system (Fig. 2) [10].

Figure 2. a) Locomotion Activity Monitor (LAM25) from Trikinetics [9]. The system consists of a tube holder with slots for 32 tubes. The blue and yellow circles surrounding the tubes are the locomotion detection mechanism. b) Visual representation of mosquitoes and tubes post-preparation in the DAM experiment. 

An IR beam was sent vertically across the diameter of the tube at its horizontal midpoint (Fig. 3) [5,10]. Disruption of the IR beam was tracked via DAM system software using a binary system, where 1 represents a disruption of the beam, and 0 represents no disruption. 

2.2 Mosquito Preparation & Data Collection

The tubes used for the DAM  experiment were 25 mm in diameter, and open on both ends [5,10]. One end was covered by securing a 3.0’’x 3.5’’ rectangle of mosquito net, and the other was closed off by inserting a cotton ball half-soaked in a 10% sucrose solution. The soaked half of the cotton ball was inserted into the tube to serve as a food source for the mosquitoes over the seven-day experimental period, while the dry side poked slightly out of the tube for easy removal at the end of the experiment (Fig. 3). In some instances, the cotton ball was soaked in sucrose for too long and resulted in the mosquitoes sticking to it. This prevented the mosquitoes from moving around in the tube and was registered as inactivity. The data from these individuals was not included in the final results.

Each of the 4 rows of 8 slots for tubes in the monitor was assigned to one of four different groups (segregated by sex & strain). Using the LAM25 system, a total of three trials, spanning seven days each, were conducted. The placement of the four groups was changed for each trial to ensure that differences in locomotion activity data was not impacted by an individual’s location on the monitor. 

2.3 Data Analysis

Not all mosquitoes used in the experiment survived the seven-day duration of the experiment. Since data was collected for each tube on a minute-by-minute basis, and there was no way to pinpoint the exact time of death for these mosquitoes, their data was removed from the data collected for the entire run. Then, the number of movements across each group were totaled for each minute. The movements across each hour were averaged, then divided by the number of mosquitoes to standardize results across each group. 

3. Flybox

3.1 Assembling the Flybox

Although the DAM experiments revealed periods of activity and inactivity in the mosquitoes, the DAM model falls short in providing details about what kind of activity is being detected. A more thorough analysis of the movement behaviors exhibited by each strain can be conducted with a tool called the Flybox (Fig. 4). 

Figure 4. An image depicting the Flybox built by the Vector Genetics Laboratory using the protocol designed and developed by the Rosbash Lab at Brandeis University [14].

Initially developed by the Rosbash Lab at Brandeis University [14], this machine includes a dark chamber with user-controlled lighting that simulates light-dark conditions. An IR camera paired with the FlyBoxScanner software[9] provides a frame-by-frame picture analysis in combination with tracking changes to enable three-dimensional mapping of mosquito activity. 

The outer frame of the box and 3D printed pieces were created via machines at the Engineering Student Design Center at UC Davis, using templates provided by the Brandeis Lab [14]. Internal wiring and assembly of PCB boards were done at the Vector Genetics Laboratory following the instructions from the Brandeis Lab [14]. As the box was initially designed for Drosophila, minor modifications were made to the box to fit the objectives of the lab’s experiments. Modifications include: (1) excluding red and green LED strips and (2) rewriting the existing Arduino code to control only white lights. While most of the code from the Brandeis’ Lab’s Github [14] was retained, three main files (start_menu.cpp, events.cpp, and lights.cpp) were edited and simplified to ensure smooth troubleshooting of the box by less technically skilled users in the future. The team also coded a feature to manually toggle the white LED lights that mimics the IR toggling feature from the original box code. The toggle feature allowed users to quickly confirm that both the IR lights and white LED lights are functioning without running an event file. All coding modifications can be found in full on GitHub [1]. 

3.2 Custom 24-Well Plate Creation

While the original protocol calls for a standard 24-well plate from Corning Inc., the researchers noticed during a trial that mosquitoes moved between wells as a result of a non-flush fit between the plate lid and the top of the wells. Additionally, this standard configuration did not offer enough vertical movement for the mosquitoes, especially after the sucrose-saturated cotton swab was placed at the bottom. To address these issues, the researchers developed a custom 24-well plate made up of twenty-four 5 mL self-standing screw-cap tubes from Axygen Inc., arranged in a 4X6 (4 rows, 6 columns) configuration. Holes were drilled in the conical bottom of the tubes to allow airflow. The screw caps were replaced with the custom lid shown in Figure 5 (a) below.

Figure 5 a) SolidWorks drawing of the custom cap part.  b) Tubes in the 4X6 configuration, with the base plate and a soaked cotton pad.

Vertical slits along the side of the custom cap were manually added to allow for tolerance fit into the tubes. The top face (∅16.5 mm) of the custom lid was then covered in a mesh to allow mosquitoes to settle on it when feeding from the cotton pad below. The tubes were arranged as shown in the diagram in figure 5(b).

3.3 Flybox Data Collection Protocol

The Flybox contains a 24-well plate with a lid that can house mosquitoes (Fig. 5). A 10% sucrose solution-soaked cotton pad will be placed at the bottom of the well, in the base plate, as a food source for the mosquito (for the preliminary data collection a standard well plate was used as the custom plate had been yet to be developed). Since there are 24 wells and four groups of mosquitoes, six mosquitoes will be obtained from each group for a single run. Mosquitoes will be frozen at -20˚C for 2 minutes to immobilize them. The mosquitoes are then gently placed into each tube with tweezers, and the tube is sealed. All the tubes are flipped to the orientation shown in figure 5(b), and the well plate is placed inside the designated slot in the Flybox. A test run conducted with a few mosquitoes to demonstrate the tracking software (Fig. 6)

Figure 6. An image from the frame-by-frame analysis conducted by the FlyBoxScanner software. It is to be noted that for this test run, well 17 unintentionally contained two mosquitoes.

As with the DAM experiment, the location of the mosquito groups will be changed for each run. Experiments will be conducted over a period of 7 days, and the frame-by-frame analysis along with tracking data will be used to analyze the specific behaviors of each mosquito strain & sex. 

Results:

Figure 7. Isolated Fundamental Frequencies from the first trial of the wingbeat frequency experiment. The x-axis plots frequency in Hz, and the y-axis is the magnitude of the frequency components. The red line represents the isolated fundamental frequency determined by the algorithm.

Figure 8. Isolated Fundamental Frequencies from the second trial of the wingbeat frequency experiment. The x-axis plots frequency in Hz, and the y-axis is the magnitude of the frequency components. The red line represents the isolated fundamental frequency determined by the algorithm.

Figure 9. Isolated Fundamental Frequencies from the third trial of the wingbeat frequency experiment. The x-axis plots frequency in Hz, and the y-axis is the magnitude of the frequency components. The red line represents the isolated fundamental frequency determined by the algorithm.

Figure 10. (a) Visualization of the average movements per hour per mosquito taken across a 24 hour period on the fifth day of trial one. (b) Visualization of the average movements per hour per mosquito taken across a 24-hour period on the fifth day of trial two. (c) Visualization of the average movements per hour per mosquito taken across a 24-hour period on the fifth day of trial three. The data for the fifth day was chosen as a representative time period for data across all three trials, as the fourth day consisted of a mid-experiment check. 

Figure 11. Image Depicting Interpretation Methods for Minute-by-Minute Flybox Scanner Software Data. The first 3 columns are used to provide time and date stamps for the minute a movement occurs. Columns 4-7 are binary initializers; these columns are used to ensure that a movement is registered and counted. If movement occurs, it references the initializer set to true (1). If movement does not occur, it references the initializer set to false (0). Columns 8-10 are set by the software to read as Ct, 0, 0, which adds a space between the initializer columns and the columns containing data. The remaining 24 columns represent the data collected from each respective well, as labeled in Fig.5.

1. Wingbeat Experiment

There were three trials completed for the wingbeat frequency experiment. The data was uploaded into a custom MATLAB script which isolated the fundamental frequencies and generated images (Fig. 6, 7, and 8). From this data, the average measured fundamental frequencies are as follows: WT Male: 410.5 Hz, WT Female: 525.5 Hz, GEM Male: 530.2 Hz, GEM Female: 436.6 Hz. These numbers are within the normal range of fundamental frequencies for non-swarming mosquitoes in the An. gambiae species complex [12]. However, previous studies suggest it is typical for female mosquitoes to have a higher fundamental frequency than male mosquitoes during non-swarming hours [3]. Our results have a notable trend of higher male than female observed fundamental frequency in the audio samples for GEM and WT individuals. 

2. DAM Experiment

Data from the three seven-day DAM experimental runs were downloaded as Excel spreadsheets and processed to isolate data from each 24 hour period (ranging from 00:00 to 23:59). Then, the average number of movements per mosquito per hour was determined for the fifth day of each trial. Figures 9(a), 9(b), and 9(c) above show an inactivity period approximately between 4:00 and 16:00 hours. Based on the plots, the activity between 16:00 and 20:00 appears to be higher for both strains and sexes, with the activity decreasing between 20:00 and 4:00 of the following day. However, between the three trials, the mosquito group with the most activity per mosquito per hour differs. In figures 10(a) and 10(b), the most activity per mosquito per hour occurs for the WT male and GEM male. For figure 10(c), the most activity per mosquito per hour occurs for the GEM female. 

3. Flybox Experiment

As of now, no data collection has been conducted with the Flybox method. A brief five-minute trial with live mosquitoes was conducted to test the flybox’s movement detection software. The scanner software successfully converted the movement data into a readable .txt file where each column represents the corresponding numbered well, and each row corresponds to the minute the movement occurs. Interpretation of the text file can be found in Figure 11.

Discussion (Next Steps/Recap):

This study aimed to characterize the behavior of the UCMI GEM mosquitoes in comparison to their wild type counterparts in terms of circadian rhythm-based movement and mating behavior. To study mating behavior, the fundamental wingbeat frequencies of the two mosquito strains and sexes were analyzed. While the difference in fundamental frequency between WT and GEM individuals may suggest that WT males could experience difficulty in recognizing GEM females entering a swarm, this study alone does not provide adequate evidence of insufficient mating ability in the GEM mosquitoes due to a few confounding factors. 

The data for each run of the wingbeat experiment was collected around the hours of 10:00 and 15:00, but not at a specific set time. Since swarming behavior is more common towards dusk, it is possible that audio samples, specifically of male mosquitoes collected later in the day, closer to observed swarming periods, display a higher fundamental frequency compared to other samples collected earlier in the day. Additionally, while the trials were conducted in a room with minimal noise, the background hum of various machinery within the insectary where the experiment was conducted may have been picked up by the recording device, distorting the audio and ultimately the calculated fundamental frequency. 

The wingbeat frequency experiment cannot independently determine differences in mating behavior between the wild type and GEM strains. However, it establishes a baseline to conduct further research in this area. While successful cross-breeding of the two strains has been displayed in a laboratory setting, a dedicated trial is necessary to determine whether the strains would display preferential mating selection in the field. Future studies may be proposed to track the generational mating choices of the mosquitoes when allowed to mate freely in a natural environment. Such an experiment would allow us to gain a more thorough understanding of whether the GEM mosquitoes are able to compete with the WT swarming and mating behavior.

The DAM experiment informed overall trends in activity for both strains and sexes across a 24 hour period. Variations between the most active groups in figure 10 (a, b, and c) were cross-compared with the location of each group on the activity monitor, but there did not appear to be a correlation between the two. The variations indicate that while the DAM model may be a useful tool in analyzing overall trends in mosquito locomotion activity, it fails to distinguish specific behaviors and activities exhibited by the strains and sexes. Furthermore, the DAM model detects mosquito activity only if a mosquito disrupts a beam in the middle of the tube. If a mosquito moves on either end of the tube without crossing the beam, the movement is not registered. Consequently, further analysis on locomotion activity using a multidimensional movement tracker like the FlyBox is required to truly understand how the activity patterns of the two strains and sexes may differ. 

To visualize mosquito activity in a multidimensional manner, the team plans to replicate similar 24 hour experiments using the Flybox. Data and video footage acquired from the tracking software will be collected and analyzed to isolate specific behavioral locomotion patterns exhibited by each strain. From the test run of the Flybox, we were able to establish a protocol for how to proceed with the experiments using this device. The researchers determined that this initial test is also crucial to run each time the Flybox is used to ensure the scanner software is correctly synchronized with the well-plates. The camera angle in the box results in a misalignment of the top and bottom circles of each well near the edges of the plate (Fig. 6), which could result in inaccurate movement detection per well unless tested prior to a trial. The text file generated when collecting data will only populate if objects (such as mosquitoes) are detected within the well during the experiment. Otherwise, the file may appear empty. The Flybox would not only provide more specific hour to hour locomotion data, but also allow the team to better understand whether there is an alignment in circadian aspects of locomotion data and in the overall amount of activity between the two strains and sexes. This would allow the team to postulate behavioral patterns not only important to mating, but also examine other locomotion related activities such as blood feeding and migration. 

Competing Interest Statement:

The authors declare no competing interests.

Acknowledgments:

We would like to thank our PI Dr. Gregory Lanzaro PhD, our current mentors Claire Mackenzie Egan and Marc Crepeau, our former mentor Dr. Melina Aulino Campos De Lima PhD, and our insectary specialists Danspaid Mabuka, Brian Leetakubuulidde, and Ivan Mugeni. We would also like to thank the team at Brandeis Rosbash Lab for their support during the process of building and troubleshooting the Flybox. All plots were made using MATLAB R2024B. Generative AI (OpenAI 2025) was used to debug and develop custom scripts and make modifications to existing scripts. 

References:

1. Bosky, A., Ravi, N. (2025, Dec). Vector-Genetics-Laboratory-Research-Project 

[Wingbeat Analysis Script and Flybox Modifications].GitHub. https://github.com/nikhi-ravi/Vector-Genetics-Laboratory-Research-Project

2. Carballar-Lejarazú, R., Ogaugwu, C., Tushar, T., Kelsey, A., Pham, T. B., Murphy, J., 

gambiae. Proceedings of the National Academy of Sciences, 117(37), 22805–22814. https://doi.org/10.1073/pnas.2010214117

3. Cator, L. J., Arthur, B. J., Harrington, L. C., & Hoy, R. R. (2009). Harmonic Convergence 

in the Love Songs of the Dengue Vector Mosquito. Science, 323(5917), 

1077–1079. https://doi.org/10.1126/science.1166541

4. Churcher, T. S., Lissenden, N., Griffin, J. T., Worrall, E., & Ranson, H. (2016). The 

impact of pyrethroid resistance on the efficacy and effectiveness of bednets for malaria control in Africa. eLife, 5, e16090. https://doi.org/10.7554/eLife.16090

5. Drosophila Activity Monitoring System. (n.d.). 

https://trikinetics.com/Downloads/DAMSystem%20Overview%202108.pdf

6. Duffield, G. E. (2024). Circadian and Daily Rhythms of Disease Vector Mosquitoes. 

Current Opinion in Insect Science, 101179–101179. 

https://doi.org/10.1016/j.cois.2024.101179

7. Evans, B. R., Kotsakiozi, P., Costa-da-Silva, A. L., Ioshino, R. S., Garziera, L., Pedrosa, 

M. C., Malavasi, A., Virginio, J. F., Capurro, M. L., & Powell, J. R. (2019). 

Transgenic Aedes aegypti Mosquitoes Transfer Genes into a Natural Population. 

Scientific Reports, 9(1). https://doi.org/10.1038/s41598-019-49660-6

8. Ferguson, N.M. Challenges and opportunities in controlling mosquito-borne infections. 

Nature 559, 490–497 (2018). https://doi.org/10.1038/s41586-018-0318-5

9. Jose, A. (2024, Jan) flybox-scanner [Flybox Scanner Software]. 

GitHub.https://github.com/jose-elias-alvarez/flybox-scanner

10. LAM10/16/25 Locomotor Activity Monitor. (n.d.). 

https://trikinetics.com/Downloads/LAM25%20Data%20Sheet.pdf

11. Le, J.Q., Yu, A. D., Dai, X., Allum, C., Bradley, O. J., McAneney, S., Schwarzinger, F., 

Sherman, Z., Alvarez, J.E., Berglund, G., Shin, D., Keefer, E., Neeley, L., Rosbash, M. (2024). Flybox: A Flexible Open-Source Behavior Monitoring System. bioRxiv. https://doi.org/10.1101/2024.05.15.594443

12. Mukundarajan, H., Hol, F. J. H., Castillo, E. A., Newby, C., & Prakash, M. (2017). Using 

mobile phones as acoustic sensors for high-throughput mosquito surveillance. eLife, 6, e27854. https://doi.org/10.7554/eLife.27854

13. OpenAI. (2025). ChatGPT (Dec 22 version) [Large language model]. 

https://chat.openai.com/chat

14. Rosbash-Lab-FlyBox. (2023, May 7). GitHub. https://github.com/Rosbash-Lab-FlyBox
15. Somers, J., Georgiades, M., Su, M. P., Bagi, J., Andrés, M., Alampounti, A., Mills, G., 

Ntabaliba, W., Moore, S. J., Spaccapelo, R., & Albert, J. T. (2022). Hitting the right note at the right time: Circadian control of audibility in Anopheles mosquito mating swarms is mediated by flight tones. Science Advances, 8(2). https://doi.org/10.1126/sciadv.abl4844

16. Traoré, A. S., Porciani, A., Moiroux, N., Dabiré, R. K., Simard, F., Costantini, C., & 

Mouline, K. (2021). Effects of insemination and blood-feeding on locomotor 

activity of wild-derived females of the malaria mosquito Anopheles coluzzii. 

Parasites & Vectors, 14(1). https://doi.org/10.1186/s13071-021-04967-0

17. World Health Organization. (2024, December 11). Malaria. World Health Organization. 

https://www.who.int/news-room/fact-sheets/detail/malaria

18. World Health Organization. World malaria report 2024: addressing inequity in the global 

malaria response. Geneva; 2024. License: CC BY-NC-SA 3.0 IGO. 2024.