Imagine designing a fly: you would need wings of course, three body segments, six legs, big red eyes and antennae. Unless you are already a fly enthusiast, you might not realize that you also need to include the halteres: small dumbbell-shaped appendages whose evolutionary ancestor is full hindwings. Without these modified hindwings, your fly would crash about clumsily instead of elegantly dipping through the air to land on your food. Halteres beat in the opposite direction to the wings and they are home to dense fields of sensory dome structures that detect tiny mechanical forces when a fly spins or turns, like a gyroscope. Two decades ago, the observation that more nerves carried information from the eyes to the halteres than to the wings led scientists to hypothesize that flies control their flight by first combining information about what the fly sees with a sense of rotation from the halteres before controlling the wings.
At the time, technological constraints made this hypothesis untestable. Recently, Bradley Dickerson and colleagues from Caltech, USA, tested the theory using modern genetic tools in tethered flies. They observed haltere muscle activity while the flies were shown spinning patterns, which caused the insects to move their wings as if they were spinning during flight too. The haltere muscles were most active when the patterns turned toward the same side as the muscles, whether the pattern simply rotated horizontally or spun in front of the fly. In other words, the little muscles controlling the halteres respond to what the fly sees, not just how the fly turns.
Next, the team tested the prediction that what the fly sees adjusts how the halteres sense spinning, by imaging nerve signals in the brain that were sent by the sensory domes on the halteres. They saw that nerve signals from the halteres changed depending on whether the spinning patterns rotated horizontally or vertically, or spun in front of the fly. The signals from the haltere indicated the involvement of a greater number of sensory domes, not greater activity from the same domes. This distinction means that muscles at the base of the haltere are likely repositioning the appendage to modulate which groups of sensory domes are active depending on which way the fly is turning.
The team then tested the prediction that changing the firing pattern of the nerve signals from the halteres would change how the wings beat. They genetically engineered special flies that produced light-sensitive proteins that cause the haltere motor neurons to fire when light is shone on them. At the same time, the team recorded electrical activity in two motor neurons controlling the wing: one that fires consistently when the wing is at the top of the wingbeat and one that fires sporadically. They found that when they artificially activated haltere muscles, the consistent motor neuron fired earlier than usual, and the sporadic motor neuron fired more often. This observation supports the prediction from 20 years ago that muscles controlling the wings are dependent on haltere muscles during flight.
Dickerson and colleagues have tested the predictions borne out of earlier experiments in larger flies and found that halteres are not just gyroscopic sensors; instead, they are actively controlled based on what the fly sees and they can, in turn, modify how muscles control the wings. From these observations, the team put forth a new evolutionary hypothesis: that halteres are better equipped to sense both rotation and motion because they do not produce aerodynamic force, like the wings do. In this way, flies can use their halteres to orchestrate wing muscles, thereby exerting greater control over their route to your food.