The potential underlying mechanisms during learning flights

Abstract

Hymenopterans, such as bees and wasps, have long fascinated researchers with their sinuous movements at novel locations. These movements, such as loops, arcs, or zigzags, serve to help insects learn their surroundings at important locations. They also allow the insects to explore and orient themselves in their environment. After they gained experience with their environment, the insects fly along optimized paths guided by several guidance strategies, such as path integration, local homing, and route-following, forming a navigational toolkit. Whereas the experienced insects combine these strategies efficiently, the naive insects need to learn about their surroundings and tune the navigational toolkit. We will see that the structure of the movements performed during the learning flights leverages the robustness of certain strategies within a given scale to tune other strategies which are more efficient at a larger scale. Thus, an insect can explore its environment incrementally without risking not finding back essential locations.

Keywords: Homing; Hymenopteran; Learning; Navigation.

Fig. 1

Cockpit view of a multisensory navigation toolkit. Directional cues, such as the sun’s location, the pattern of polarized light, the magnetic compass, or the wind, provide a sense of global orientation for insects. On their outbound journey, path integration (PI) keeps track of their starting location and provides a straight line to that location. A multisensory scenery, represented here by visual information, might be stored at multiple locations along the journey and around the nest, enabling insects to return home

Fig. 2

Usage of guidance strategies and their interplays, from naive insects to experienced foragers. The insects’ navigational toolkit can use a variety of sensory information (left) and guidance mechanisms (middle). The interplay between guidance mechanisms varies with experience and scale. One guidance mechanisms might help to tune another one (shown by arrows between guidance mechanisms)

Fig. 3

Simulation of directional error given by path integration. The error in the path integrator direction increases with the traveled distance (Vickerstaff and Cheung ; Cheung and Vickerstaff 2010). We simulated an agent guided by a directed random walk and derived the error according to Wystrach et al. (2015). Left: three random walks of different lengths are shown. Red and purple are long journeys, whereas the gray is a short journey. Red is close to the starting location. The gray shaded area below the arrow shows the uncertainty of the simulated path integration. The weakly tortuous paths (gray and purple) have been simulated with von Mises with $\kappa =60$, whereas the strongly tortuous path (red) has been simulated with a von Mises with $\kappa =11$. Right: the lines indicate the error in degrees of the simulated path integrator as a function of the distance traveled and distance from the nest. We used 4000 paths for each tortuosity intensity simulated with von Mises distributions of varying concentration (from $\kappa =1$—extremely tortuous, to $\kappa =100$, almost a direct path). We see that the error increases as a function of the distance traveled and decreases with the distance to the nest. The dots indicate the uncertainty of the path integrator of the three example trajectories in the left of the figure

Fig. 4

Example of learning flights of bumblebees, Bombus terrestris. The first flights of bumblebees was recorded in two environments (Lobecke et al. 2018). The example trajectories are shown in 3D (A1) and in 2D (A2) from a view from above (i.e., looking at the ground) in the first row, where the dots represent the position of the bees’ thorax positions and the lines represent the bees’ orientation. The position of the nest is indicated by the black arrow, and the color indicates the time (yellow for the start and dark purple for the end). The time course of the distance to the nest (shown in blue) and the flying altitude (shown in orange) are shown in the last row. In the flight without objects, the positions of the bees are distributed regularly from exiting the nest to slowly increasing distance and height to the nest (A3). In contrast, the flights where the nest was surrounded by objects (B1 in 3D, B2 in 2D) show an aggregation of positions close to the nest before expanding into wider loops and arcs (B3). The distance to the nest increases over time in all trajectories, but the bees regularly return close to the nest, resulting in the distance getting regularly very small. A similar structure can be seen for the flight altitude, where the bees increase their distance from the floor over time

Fig. 5

Simulation of directional error along a learning flight. The simulated error of the path integration (as depicted in Fig. 3) along the flight is represented in orange. Additionally, the error of a path integrator that is reset around the vicinity of the nest (indicated by the open gray circle) is shown in green. The distance to the nest is indicated by the dashed gray line

Fig. 6

Large-scale learning flight with elongated loops. The nest position is marked with a white dot and the colors indicate the time (light yellow the start and black the end of the flight). The flight contains several elongated loops with straight return to the nest (e.g., red loop with an arrow pointing to the tip of the loop). Arrows indicate the direction traveled of the bumblebee along the flight. Adapted from Woodgate et al. (2016).

Fig. 7

Duration of consecutive learning flights of bumblebees was compared in two environments, as shown in Fig. 4. The nest was not surrounded by objects (in blue) or by two objects (in orange). The data used to generate the figures were obtained from Lobecke (2018) and Bertrand et al. (2020)