Mosquitoes escape looming threats by actively steering into the bow-wave induced by the attacker

Cribellier, Antoine1

Published Feb 29, 2024 on Dryad. https://doi.org/10.5061/dryad.nvx0k6dvz

Data files

Feb 29, 2024 version files 3.71 GB

Abstract

To detect and escape a threat, night flying insects must rely on other senses than vision alone. Here we study how anthropophilic malaria mosquitoes can escape a swatting hand in the dark using high-speed videography and numerical simulations. We show that these night flying mosquitoes escape looming objects by using the object-induced airflow in two ways. They first actively steer into the bow-wave produced by the attacker, and then passively travel with this bow-wave away from the attacker; these two aspects explain two-thirds and one-third of their escape accelerations, respectively. Thus, flying mosquitoes being attacked in the dark rely both on airflow-sensing to trigger their escape, and on attacker-induced airflow to maximize their escape performance. Similar escape strategies are probably common among small lightweight insects.

README

Data Files on Dryad

Data_S1_-_Experiment 1
Database of the experiment #1. Flight tracks of all Anopheles coluzzii mosquitoes attacked by the (real and virtual) mechanical swatter, and flying in various light conditions. A Matlab (all_data.mat) file containing three-dimensional tracks of all flying mosquitoes, described as the time t (s) and the spatial {x, y, z} (m) coordinates of the mosquito at each video frame. The coordinates are in meters, and in the world reference frame as defined in figure 1, with z oriented vertically up, and the origin of the coordinate frame at the center of the flight arena. The trajectories were determined as described in the materials and methods. Metadata for each track includes age, swatter type (hollow or solid) and mode (on or off), light condition, as well as temperature and humidity over time in the arena during each experiment. In addition, the file contains the kinematic of the mechanical swatter, as well as the geometry of the flight arena and the swatter.

Data_S2_-_Experiment2
Database of the experiment #2. Flight kinematics of escaping Anopheles coluzzii mosquitoes attacked by the mechanical swatter (opaque or transparent disk) in twilight light condition. A Matlab (all_data.mat) file containing three-dimensional tracks of all escaping mosquitoes as well as all body and wing kinematics parameters (positions and angles) over time. These kinematics parameters were determined using the custom-made three-dimensional tracker (Code S3). Metadata for each track includes age, swatter type (opaque or transparent), as well as temperature and humidity over time in the arena during each experiment. In addition, the file contains the kinematic of the mechanical swatter, as well as the geometry of the flight arena and the swatter.

Data S3 - CFD
Computation fluid dynamic (CFD) results. Means (mean folder) and standard deviation (std folder) of the airflow velocities induced by the swatter movement can be found in .csv files (one for each time step (1 ms) and in polar coordinates (r,theta,y)). The piston kinematic and the probe positions can also be found in separated .csv files (piston_kinematic.corrected.csv, probe_pos.csv, prob_pos_cart.csv, prove_pos_polar.csv).

Code Files on Zenodo

Code S1. (separate file)
Python code used to track mosquitoes body and wings kinematics. Recordings of mosquitoes were preprocessed (cropped and stitched) to be analyzed with Deeplabcut. Then a three-dimensional mosquito skeleton was fitted to the two-dimensional tracking results from Deeplabcut to estimate the body and wing kinematics. Scripts to generate stroboscopic images and example videos are also included.

Code S2. (separate file)
Analysis codes. Contains all original Matlab codes that were written to perform the analysis of the article and to generate the panels of Fig. 1-4. Also contains JAGS modelling codes used for doing the Bayesian estimations in Fig. 1, 3-4. Instructions to run the full analysis can be found in the readme.md.

- Move the Experiment1 and Experiment2 datasets in _Data

- Run the following scripts, figures will be saved in the _Figures folder.

Experiment #1:

- Run main_analysis_experiment1.m

One subset of the full dataset will be analysed (lines 116):

subset_name = '_Subset-disk_types-transparent'; % The dataset analysed in the article

(both solid and perforated transparent disk with a clear mesh + control with swatter turned off)

- Run get_groups_estimayed_means to estimate the means of the colision probabilities for the various light conditions and disk types (Fig. 1E)

Experiment #2:

- Run main_analysis_experiment2.m