Characterization of lab-based swarms of Anopheles gambiae mosquitoes using 3D-video tracking

Abstract

Mosquito copulation is a crucial determinant of its capacity to transmit malaria-causing Plasmodium parasites as well as underpinning several highly-anticipated vector control methodologies such as gene drive and sterile insect technique. For the anopheline mosquitoes responsible for African malaria transmission, mating takes place within crepuscular male swarms which females enter solely to mate. However, the mechanisms that regulate swarm structure or that govern mate choice remain opaque. We used 3D-video tracking approaches and computer vision algorithms developed for the study of other complex biological systems to document swarming behavior of a lab-adapted Anopheles gambiae line in a lab-based setting. By reconstructing trajectories of individual mosquitoes lasting up to 15.88 s, in swarms containing upwards of 200 participants, we documented swarm-like behavior in both males and females. In single sex swarms, encounters between individuals were fleeting (< 0.75 s). By contrast, in mixed swarms, we were able to detect 79 'brief encounters' (> 0.75 s; < 2.5 s) and 17 longer-lived encounters (> 2.5 s). We also documented several examples of apparent male-male mating competition. These findings represent the first steps towards a more detailed and quantitative description of swarming and courtship behavior in one of the most important vectors of malaria.

Figure 1

Characteristics of lab-based An. gambiae swarms. (a) and (b) the number, NR, of mosquitoes released into the cage, for each generation of males (left) and females (right). NR is in the same range, between approx. 800 and 2000, in both cases, but is less variable for females. (c) and (d) The number, N, of mosquitoes participating in 28 acquisitions of single-sex male swarms (left column) and 34 single-sex female swarms (right column) are shown. Different colours refer to different generations. Male swarms are on average larger (mean ± standard deviation, 138 ± 84.6) than female swarms (48 ± 34.4). (e) and (f) N is averaged over all acquisitions of the same generation, left column for male swarms and right column for females. Data represents mean ± standard deviation. (g) and (h) Nmax, the maximum number of swarming mosquitoes across all acquisitions of the same generation, as a function of NR. In males Nmax depends linearly on NR (Pearson coefficient 0.88 – p value 0.005). The slope of the linear fit, equal to 0.19, represents the typical participation ratio in male swarms. In females the linearity between Nmax and NR is not evident.

Figure 2

Swarm structure in single-sex male and female swarms. The superimposition of 2700 images from one of the cameras, clearly shows a gross difference in swarm structure. Male swarms (left panel) are stable and have a roughly cylindrical form, while non-participating males remain stationary on the cage walls. In contrast, female swarms (right panel) comprise two distinct zones, a dense, relatively ordered aggregation near the ground marker, and a second zone, at some distance from the ground marker, characterized by much more disordered female motion.

Figure 3

Characteristic circular movements in the horizontal plane. (a) The 3D reconstructed trajectories of a single-sex male swarm. Different colours represent different mosquitoes. (b) The trajectory of a single individual in the horizontal plane, as it would be seen from above the swarm. Mosquitoes perform consecutive pseudo-circular motions, highlighted in red, superimposed on the complete trajectory of the same individual, shown in light grey. Numbers in the top right corner of each sub-panel represent the temporal order of the ring-like movements.

Figure 4

Differences in 3D velocities in male and female swarms. Graphs represent the probability distribution of the three components of the velocity and of the speed, for 19 male swarms and 11 female swarms. In both males (top row) and females (bottom row), we found mean values of v_x, v_y and v_z close to 0, confirming the stable position in time of swarms at a group level. Top row: males. The narrow distribution of v_y, i.e. the velocity component parallel to the direction of gravity, together with the double-peaked distributions of v_x and v_z, reveals that motion mainly occurs on the xz-plane. The distributions of v_x and v_z, with the two peaks at ± 0.5 m/s, further indicate the lack of any favoured direction of flight on the xz-plane, but rather, in the swarm all directions on the horizontal plane are explored with the same probability, i.e. a change in the v_x, v_z reference frame would display the same double-peaked distributions. Bottom row: females. Female motion is qualitatively distinct, with the distributions of the three components of the velocity and of the overall speed wider than in males. Females are more prone to move in the direction of gravity, with a v_y distribution larger than males (std = 0.38 m/s against the std = 0.18 m/s that we found in males), and with the double peak of the distributions of v_x and v_z being barely visible.

Figure 5

Group velocity probability distributions in the horizontal plane. (a) males. On the left, the top view of the probability distribution of the pairs (v_x, v_z) shown on the right. The radial symmetry displayed by the distribution confirms the absence of preferred directions of flight. The steep (v_x, v_z) distribution, corresponding to the thin yellow circle on the left plot, proves the high correlation of the two velocity components, which also shows that males tend to move at a constant speed in the horizontal plane, i.e. ${\text{v}}_{\text{x}}^2{\text{+ v}}_{\text{z}}^2\text{= constant}$ (b) females. On the left, the top view of the probability distribution of the pairs (v_x, v_z) shown on the right. The distribution is much wider than in males. The well-defined thin circular ridge of the distribution found in males (panel a) is replaced by a thick blurred circle surrounded by a quite large region where v_x and v_z may lie, suggesting higher fluctuation of the speed in the horizontal plane.

Figure 6

Individual velocity probability distributions and autocorrelation. Left box: males. Right box: females. First and fourth column: the evolution in time of the z component of the velocity for three males and three females. Time is indicated on the bottom, coloured horizontal axes of the graphs. The black lines filled with light colours represent the probability distribution (relative to the scale on the top, black horizontal axes). Second and fifth columns: individual mosquito velocity in the v_xv_z-plane. Third and sixth columns: individual velocity autocorrelation functions for three males and three females.

Figure 7

Male–female encounters. (a) From 19 reconstructed acquisitions of single-sex male swarms we extracted the probability distribution of the duration of male-male encounters, defined as the time spent by two mosquitoes at a mutual distance shorter than a threshold, chosen (for each swarm separately) as half the mean nearest neighbor distance, namely in the range between 5.9 cm and 12.6 cm (see Table SI1). Maximum duration of male-male encounters is 0.75 s, indicated with a vertical dashed line. In the inset: for 8 mixed-sex swarms, encounters lasting more than 0.75 s, which are unlikely to be found in single-sex male swarms, are labelled with a unique ID and their duration is shown. Of these encounters 80% (79 out of a total of 96), highlighted with yellow circles, last less than 2.5 s, the horizontal black line. The other 20%, highlighted with orange circles, last more than 2.5 s. They may correspond to male–female mating events. (b–g) the z-component of the trajectories of mosquitoes involved in some of the long-lasting encounters are shown as a function of time. Video footage of these encounters is available online. (b) E1: the longest encounter, lasting 15.88 s (c and d) E2 and E3: the two encounters last 9.3 s and 6.2 s, starting before the beginning of the acquisition and ending when mosquitoes exit the common field of view of the cameras. (e) E4: the encounter starts at the very end of the acquisition. The trajectory highlighted in green is presumably a female joining the swarm a few seconds after the acquisition started. (f) E5: three mosquitoes engaged in a mating competition. A first pair, orange and blue trajectories, is joined by a third mosquito (green trajectory), which is able to replace the blue individual in the established pair. (g) E6: four mosquitoes engaged in a mating competition. One pair, blue and magenta, is joined by two other mosquitoes, denoted by orange and green. The four briefly fly together. The competition ends with the orange-green pair landing on the ground, and the first formed pair disrupted, with the blue and magenta mosquitoes separately returning to the swarm.

Figure 8

Schematic representation and pictures of semi-field insectary chamber. (a) Schematic view of the semi-field setup. The swarming stimuli, feeding and resting stations comprise: the two sunsets; visual ground marker (M, dimensions: internal square 20 cm × 20 cm, central square 40 cm × 40 cm, external square 55 cm × 55 cm); sugar sources (S); mosquito resting shelters (R) and terracotta brick resting shelters (B). The infrared lights (IR) are placed inside the large cage (also depicted in panel d). Cameras (C1, C2, C3), located outside the cage, are directed towards the visual ground marker. (b) 3D rendering of the swarming chamber. The image shows the dimensions of the chamber containing the large cage in which swarming takes place; dimensions and location of the infrared-absorbing black cloth wall coverings; position of the warm light above the marker; internal and external sunsets; the six infrared LED lamps supported by an iron scaffold; the 3 video cameras (C1, C2, C3); contrasting ground marker with the three white-black-white plastic sheets. (c) A picture of the setup from within the cage, showing mosquito resting shelters. In the foreground are the terracotta brick resting shelters, in the background two of the black resting shelters and the swarming ground marker. The interior walls of the cage are covered with black cloth to produce a uniform background in the images. (d) A picture of the six IR lamps from inside the cage, with cameras in the background, outside the cage. The three squares on the net are the holes where camera lenses are positioned. (e) A picture of the three cameras from outside the cage.