As a post-doc, I examined forward flight performance in Panamanian orchid bees using a field-portable wind tunnel and found that maximum flight speed was limited not by energetics or kinematic constraints, but by flight instabilities due to turbulent air flow.  I showed that flight instabilities induced by turbulence limit bees’ flight performance, and documented a novel stabilizing behavior in which bees extend their long, bulbous hind legs to reduce rolling instabilities, which increases the cost of flight.  This led my lab, with the support of an NSF CAREER grant on insect flight in turbulence, to begin measuring environmental flows in other habitats, simulating these in our lab wind tunnel, and examining the implications for flight performance.

We have now documented the effects of unsteady, structured vortex flow (e.g., behind a branch in wind) on flight stability in bumblebees, and we built on this study to experimentally test the effects of both wing flexibility and wing damage on flight stability in vortex flow.   We examined the effects of unsteady flow on bumblebee landing performance, and found that bees approaching flowers in the presence of wind are unable to regulate their speed, and thus experience higher, potentially damaging impact forces upon landing.

In another study, we measured outdoor flow conditions while monitoring the timing of bumblebee foraging trips using radio-frequency ID tags, and found that bees continue foraging regularly, regardless of wind conditions.  We then recreated field-realistic, turbulent flow in our wind tunnel and showed that this negatively impacts flight performance, reducing stability and increasing flapping frequency and the cost of flight.  Finally, we have examined the effects of unsteady flow generated not by wind, but by turbulent, thermal convection cells – which form when solar radiation heats the ground below cooler air – on the flight performance of fruit flies.  Taken together, these studies show that unsteady, turbulent flows impair insect flight stability, limiting performance and/or increasing the energetic cost of flight.

Of course, in natural settings insects are free to choose which part of the habitat they fly through, and they may choose to avoid some complex flight environments – but the technology does not yet exist to track where individual insects fly in natural environments with high spatial and temporal accuracy.  To gain insight into the behavioral choices that flying insects make when faced with complex aerial and physical environments – we designed a two-channel flight tunnel, in which bumblebees fly from their hive on one end to a nectar source on the other, through one of two channels.  We can alter the presence and direction of wind and add fields of obstacles they must fly through.  Because their behavior is highly variable, the key to gaining meaningful insight is collecting massive amounts of data.  Thus, we designed a system to automatically collect and compress thousands of video samples per day from multiple, synchronized mid-speed cameras.  This allows us to document which side of the tunnel and in which direction bees are flying, to infer the choices they make between differentflight environments.

Because a typical experiment results in ~10,000 video sequences with multiple cameras per sequence, we developed a method to automatically detect all bees within a video, and a data association algorithm to resolve three-dimensional trajectories of multiple individuals with overlapping flight paths and occasional occlusion by obstacles – a challenging computational problem.  We have used this system to show that bees strongly prefer flying with headwinds rather than tailwinds, and we performed wind tunnel studies at higher speeds to show that tailwinds pose significant challenges in terms of pitch stability.  We are now examining whether bees prefer to fly through open air or arrays of large or small obstacles, and whether headwinds or tailwinds alter this preference, as obstacles could either shield bees from wind or generate destabilizing, unsteady flow.

We have also performed studies examining how bees maneuver through complex, cluttered physical environments, and how wind and obstacle motion (e.g., induced by wind) affect maneuvering performance.  We found that small individuals outperform large ones when maneuvering through clutter, as large bees reduce their speed more to avoid collisions.  When maneuvering through still obstacles, bees typically fly slowly and cautiously, but moving obstacles (e.g., vertical posts on an orbital shaker) elicit sudden maneuvers at maximum acceleration, and this peak acceleration is inhibited by wing damage.  Thus, cumulative wing damage due to collisions reduces bees’ capacity to accelerate to avoid future collisions, leading to further damage.  We have also found that the combination of wind and obstacle motion affects the flight strategies that honeybees adopt when approaching obstacles.  When approaching moving obstacles in still air, bees slow down and take a more circuitous path during the approach; however, when approaching moving obstacles in wind, bees instead speed up and fly straight through the obstacles – perhaps in an attempt to reduce the amount of wing damage (number of wing collisions) they may experience if they come too close to an obstacle under these challenging conditions.  Finally, we have examined how bees’ flight trajectories and their choice of whether fly through vs. over obstacles depends on height of the “canopy” (obstacle field), and whether wind affects these choices.