Machine learning reveals the control mechanics of an insect wing hinge.

Insects constitute the most species-rich radiation of metazoa, a success that is due to the evolution of active flight. Unlike pterosaurs, birds and bats, the wings of insects did not evolve from legs1, but are novel structures that are attached to the body via a biomechanically complex hinge that transforms tiny, high-frequency oscillations of specialized power muscles into the sweeping back-and-forth motion of the wings2. The hinge consists of a system of tiny, hardened structures called sclerites that are interconnected to one another via flexible joints and regulated by the activity of specialized control muscles. Here we imaged the activity of these muscles in a fly using a genetically encoded calcium indicator, while simultaneously tracking the three-dimensional motion of the wings with high-speed cameras. Using machine learning, we created a convolutional neural network3 that accurately predicts wing motion from the activity of the steering muscles, and an encoder-decoder4 that predicts the role of the individual sclerites on wing motion. By replaying patterns of wing motion on a dynamically scaled robotic fly, we quantified the effects of steering muscle activity on aerodynamic forces. A physics-based simulation incorporating our hinge model generates flight manoeuvres that are remarkably similar to those of free-flying flies. This integrative, multi-disciplinary approach reveals the mechanical control logic of the insect wing hinge, arguably among the most sophisticated and evolutionarily important skeletal structures in the natural world.

Transforming representations of movement from body- to world-centric space.

When an animal moves through the world, its brain receives a stream of information about the body's translational velocity from motor commands and sensory feedback signals. These incoming signals are referenced to the body, but ultimately, they must be transformed into world-centric coordinates for navigation1,2. Here we show that this computation occurs in the fan-shaped body in the brain of Drosophila melanogaster. We identify two cell types, PFNd and PFNv3-5, that conjunctively encode translational velocity and heading as a fly walks. In these cells, velocity signals are acquired from locomotor brain regions6 and are multiplied with heading signals from the compass system. PFNd neurons prefer forward-ipsilateral movement, whereas PFNv neurons prefer backward-contralateral movement, and perturbing PFNd neurons disrupts idiothetic path integration in walking flies7. Downstream, PFNd and PFNv neurons converge onto hΔB neurons, with a connectivity pattern that pools together heading and translation direction combinations corresponding to the same movement in world-centric space. This network motif effectively performs a rotation of the brain's representation of body-centric translational velocity according to the current heading direction. Consistent with our predictions, we observe that hΔB neurons form a representation of translational velocity in world-centric coordinates. By integrating this representation over time, it should be possible for the brain to form a working memory of the path travelled through the environment8-10.

Distinct activity-gated pathways mediate attraction and aversion to CO2 in Drosophila.

Carbon dioxide is produced by many organic processes and is a convenient volatile cue for insects1 that are searching for blood hosts2, flowers3, communal nests4, fruit5 and wildfires6. Although Drosophila melanogaster feed on yeast that produce CO2 and ethanol during fermentation, laboratory experiments7-12 suggest that walking flies avoid CO2. Here we resolve this paradox by showing that both flying and walking Drosophila find CO2 attractive, but only when they are in an active state associated with foraging. Their aversion to CO2 at low-activity levels may be an adaptation to avoid parasites that seek CO2, or to avoid succumbing to respiratory acidosis in the presence of high concentrations of CO2 that exist in nature13,14. In contrast to CO2, flies are attracted to ethanol in all behavioural states, and invest twice the time searching near ethanol compared to CO2. These behavioural differences reflect the fact that ethanol is a unique signature of yeast fermentation, whereas CO2 is generated by many natural processes. Using genetic tools, we determined that the evolutionarily conserved ionotropic co-receptor IR25a is required for CO2 attraction, and that the receptors necessary for CO2 avoidance are not involved in this attraction. Our study lays the foundation for future research to determine the neural circuits that underlie both state- and odorant-dependent decision-making in Drosophila.

The long-distance flight behavior of Drosophila supports an agent-based model for wind-assisted dispersal in insects.

Despite the ecological importance of long-distance dispersal in insects, its mechanistic basis is poorly understood in genetic model species, in which advanced molecular tools are readily available. One critical question is how insects interact with the wind to detect attractive odor plumes and increase their travel distance as they disperse. To gain insight into dispersal, we conducted release-and-recapture experiments in the Mojave Desert using the fruit fly, Drosophila melanogaster We deployed chemically baited traps in a 1 km radius ring around the release site, equipped with cameras that captured the arrival times of flies as they landed. In each experiment, we released between 30,000 and 200,000 flies. By repeating the experiments under a variety of conditions, we were able to quantify the influence of wind on flies' dispersal behavior. Our results confirm that even tiny fruit flies could disperse ∼12 km in a single flight in still air and might travel many times that distance in a moderate wind. The dispersal behavior of the flies is well explained by an agent-based model in which animals maintain a fixed body orientation relative to celestial cues, actively regulate groundspeed along their body axis, and allow the wind to advect them sideways. The model accounts for the observation that flies actively fan out in all directions in still air but are increasingly advected downwind as winds intensify. Our results suggest that dispersing insects may strike a balance between the need to cover large distances while still maintaining the chance of intercepting odor plumes from upwind sources.

Superhydrophobic diving flies (Ephydra hians) and the hypersaline waters of Mono Lake.

The remarkable alkali fly, Ephydra hians, deliberately crawls into the alkaline waters of Mono Lake to feed and lay eggs. These diving flies are protected by an air bubble that forms around their superhydrophobic cuticle upon entering the lake. To study the physical mechanisms underlying this process we measured the work required for flies to enter and leave various aqueous solutions. Our measurements show that it is more difficult for the flies to escape from Mono Lake water than from fresh water, due to the high concentration of Na2CO3 which causes water to penetrate and thus wet their setose cuticle. Other less kosmotropic salts do not have this effect, suggesting that the phenomenon is governed by Hofmeister effects as well as specific interactions between ion pairs. These effects likely create a small negative charge at the air-water interface, generating an electric double layer that facilitates wetting. Compared with six other species of flies, alkali flies are better able to resist wetting in a 0.5 M Na2CO3 solution. This trait arises from a combination of factors including a denser layer of setae on their cuticle and the prevalence of smaller cuticular hydrocarbons compared with other species. Although superbly adapted to resisting wetting, alkali flies are vulnerable to getting stuck in natural and artificial oils, including dimethicone, a common ingredient in sunscreen and other cosmetics. Mono Lake's alkali flies are a compelling example of how the evolution of picoscale physical and chemical changes can allow an animal to occupy an entirely new ecological niche.

The effects of target contrast on Drosophila courtship.

Many animals use visual cues such as object shape, color and motion to detect and pursue conspecific mates. Contrast is another possibly informative visual cue, but has not been studied in great detail. In this study, we presented male Drosophila melanogaster with small, fly-sized, moving objects painted either black, white or gray to test whether they use contrast cues to identify mates. We found that males frequently chased gray objects and rarely chased white or black objects. Although males started chasing black objects as often as gray objects, the resulting chases were much shorter. To test whether the attraction to gray objects was mediated via contrast, we fabricated black and gray behavioral chambers. However, wild-type males almost never chased any objects in these darkly colored chambers. To circumvent this limitation, we increased baseline levels of chasing by thermogenetically activating P1 neurons to promote courtship. Males with thermogenetically activated P1 neurons maintained a similar preference for gray objects despite elevated levels of courtship behavior. When placed in a black chamber, males with activated P1 neurons switched their preference and chased black objects more than gray objects. We also tested whether males use contrast cues to orient to particular parts of the female's body during courtship. When presented with moving objects painted two colors, males positioned themselves next to the gray half regardless of whether the other half was painted black or white. These results suggest that males can use contrast to recognize potential mates and to position themselves during courtship.

Celestial navigation in Drosophila.

Many casual observers typecast Drosophila melanogaster as a stationary pest that lurks around fruit and wine. However, the omnipresent fruit fly, which thrives even in desert habitats, likely established and maintained its cosmopolitan status via migration over large spatial scales. To perform long-distance dispersal, flies must actively maintain a straight compass heading through the use of external orientation cues, such as those derived from the sky. In this Review, we address how D. melanogaster accomplishes long-distance navigation using celestial cues. We focus on behavioral and physiological studies indicating that fruit flies can navigate both to a pattern of linearly polarized light and to the position of the sun - the same cues utilized by more heralded insect navigators such as monarch butterflies and desert ants. In both cases, fruit flies perform menotaxis, selecting seemingly arbitrary headings that they then maintain over time. We discuss how the fly's nervous system detects and processes this sensory information to direct the steering maneuvers that underlie navigation. In particular, we highlight recent findings that compass neurons in the central complex, a set of midline neuropils, are essential for navigation. Taken together, these results suggest that fruit flies share an ancient, latent capacity for celestial navigation with other insects. Furthermore, they illustrate the potential of D. melanogaster to help us to elucidate both the cellular basis of navigation and mechanisms of directed dispersal on a landscape scale.

Flying Drosophilamelanogaster maintain arbitrary but stable headings relative to the angle of polarized light.

Animals must use external cues to maintain a straight course over long distances. In this study, we investigated how the fruit fly Drosophila melanogaster selects and maintains a flight heading relative to the axis of linearly polarized light, a visual cue produced by the atmospheric scattering of sunlight. To track flies' headings over extended periods, we used a flight simulator that coupled the angular velocity of dorsally presented polarized light to the stroke amplitude difference of the animals' wings. In the simulator, most flies actively maintained a stable heading relative to the axis of polarized light for the duration of 15 min flights. We found that individuals selected arbitrary, unpredictable headings relative to the polarization axis, which demonstrates that D. melanogaster can perform proportional navigation using a polarized light pattern. When flies flew in two consecutive bouts separated by a 5 min gap, the two flight headings were correlated, suggesting individuals retain a memory of their chosen heading. We found that adding a polarized light pattern to a light intensity gradient enhanced flies' orientation ability, suggesting D. melanogaster use a combination of cues to navigate. For both polarized light and intensity cues, flies' capacity to maintain a stable heading gradually increased over several minutes from the onset of flight. Our findings are consistent with a model in which each individual initially orients haphazardly but then settles on a heading which is maintained via a self-reinforcing process. This may be a general dispersal strategy for animals with no target destination.

Functional divisions for visual processing in the central brain of flying Drosophila.

Although anatomy is often the first step in assigning functions to neural structures, it is not always clear whether architecturally distinct regions of the brain correspond to operational units. Whereas neuroarchitecture remains relatively static, functional connectivity may change almost instantaneously according to behavioral context. We imaged panneuronal responses to visual stimuli in a highly conserved central brain region in the fruit fly, Drosophila, during flight. In one substructure, the fan-shaped body, automated analysis revealed three layers that were unresponsive in quiescent flies but became responsive to visual stimuli when the animal was flying. The responses of these regions to a broad suite of visual stimuli suggest that they are involved in the regulation of flight heading. To identify the cell types that underlie these responses, we imaged activity in sets of genetically defined neurons with arborizations in the targeted layers. The responses of this collection during flight also segregated into three sets, confirming the existence of three layers, and they collectively accounted for the panneuronal activity. Our results provide an atlas of flight-gated visual responses in a central brain circuit.

An Array of Descending Visual Interneurons Encoding Self-Motion in Drosophila.

The means by which brains transform sensory information into coherent motor actions is poorly understood. In flies, a relatively small set of descending interneurons are responsible for conveying sensory information and higher-order commands from the brain to motor circuits in the ventral nerve cord. Here, we describe three pairs of genetically identified descending interneurons that integrate information from wide-field visual interneurons and project directly to motor centers controlling flight behavior. We measured the physiological responses of these three cells during flight and found that they respond maximally to visual movement corresponding to rotation around three distinct body axes. After characterizing the tuning properties of an array of nine putative upstream visual interneurons, we show that simple linear combinations of their outputs can predict the responses of the three descending cells. Last, we developed a machine vision-tracking system that allows us to monitor multiple motor systems simultaneously and found that each visual descending interneuron class is correlated with a discrete set of motor programs. SIGNIFICANCE STATEMENT: Most animals possess specialized sensory systems for encoding body rotation, which they use for stabilizing posture and regulating motor actions. In flies and other insects, the visual system contains an array of specialized neurons that integrate local optic flow to estimate body rotation during locomotion. However, the manner in which the output of these cells is transformed by the downstream neurons that innervate motor centers is poorly understood. We have identified a set of three visual descending neurons that integrate the output of nine large-field visual interneurons and project directly to flight motor centers. Our results provide new insight into how the sensory information that encodes body motion is transformed into a code that is appropriate for motor actions.

Anatomical Reconstruction and Functional Imaging Reveal an Ordered Array of Skylight Polarization Detectors in Drosophila.

UNLABELLED: Many insects exploit skylight polarization as a compass cue for orientation and navigation. In the fruit fly, Drosophila melanogaster, photoreceptors R7 and R8 in the dorsal rim area (DRA) of the compound eye are specialized to detect the electric vector (e-vector) of linearly polarized light. These photoreceptors are arranged in stacked pairs with identical fields of view and spectral sensitivities, but mutually orthogonal microvillar orientations. As in larger flies, we found that the microvillar orientation of the distal photoreceptor R7 changes in a fan-like fashion along the DRA. This anatomical arrangement suggests that the DRA constitutes a detector for skylight polarization, in which different e-vectors maximally excite different positions in the array. To test our hypothesis, we measured responses to polarized light of varying e-vector angles in the terminals of R7/8 cells using genetically encoded calcium indicators. Our data confirm a progression of preferred e-vector angles from anterior to posterior in the DRA, and a strict orthogonality between the e-vector preferences of paired R7/8 cells. We observed decreased activity in photoreceptors in response to flashes of light polarized orthogonally to their preferred e-vector angle, suggesting reciprocal inhibition between photoreceptors in the same medullar column, which may serve to increase polarization contrast. Together, our results indicate that the polarization-vision system relies on a spatial map of preferred e-vector angles at the earliest stage of sensory processing. SIGNIFICANCE STATEMENT: The fly's visual system is an influential model system for studying neural computation, and much is known about its anatomy, physiology, and development. The circuits underlying motion processing have received the most attention, but researchers are increasingly investigating other functions, such as color perception and object recognition. In this work, we investigate the early neural processing of a somewhat exotic sense, called polarization vision. Because skylight is polarized in an orientation that is rigidly determined by the position of the sun, this cue provides compass information. Behavioral experiments have shown that many species use the polarization pattern in the sky to direct locomotion. Here we describe the input stage of the fly's polarization-vision system.

Cellular mechanisms for integral feedback in visually guided behavior.

Sensory feedback is a ubiquitous feature of guidance systems in both animals and engineered vehicles. For example, a common strategy for moving along a straight path is to turn such that the measured rate of rotation is zero. This task can be accomplished by using a feedback signal that is proportional to the instantaneous value of the measured sensory signal. In such a system, the addition of an integral term depending on past values of the sensory input is needed to eliminate steady-state error [proportional-integral (PI) control]. However, the means by which nervous systems implement such a computation are poorly understood. Here, we show that the optomotor responses of flying Drosophila follow a time course consistent with temporal integration of horizontal motion input. To investigate the cellular basis of this effect, we performed whole-cell patch-clamp recordings from the set of identified visual interneurons [horizontal system (HS) cells] thought to control this reflex during tethered flight. At high stimulus speeds, HS cells exhibit steady-state responses during flight that are absent during quiescence, a state-dependent difference in physiology that is explained by changes in their presynaptic inputs. However, even during flight, the membrane potential of the large-field interneurons exhibits no evidence for integration that could explain the behavioral responses. However, using a genetically encoded indicator, we found that calcium accumulates in the terminals of the interneurons along a time course consistent with the behavior and propose that this accumulation provides a mechanism for temporal integration of sensory feedback consistent with PI control.

Flying Drosophila stabilize their vision-based velocity controller by sensing wind with their antennae.

Flies and other insects use vision to regulate their groundspeed in flight, enabling them to fly in varying wind conditions. Compared with mechanosensory modalities, however, vision requires a long processing delay (~100 ms) that might introduce instability if operated at high gain. Flies also sense air motion with their antennae, but how this is used in flight control is unknown. We manipulated the antennal function of fruit flies by ablating their aristae, forcing them to rely on vision alone to regulate groundspeed. Arista-ablated flies in flight exhibited significantly greater groundspeed variability than intact flies. We then subjected them to a series of controlled impulsive wind gusts delivered by an air piston and experimentally manipulated antennae and visual feedback. The results show that an antenna-mediated response alters wing motion to cause flies to accelerate in the same direction as the gust. This response opposes flying into a headwind, but flies regularly fly upwind. To resolve this discrepancy, we obtained a dynamic model of the fly's velocity regulator by fitting parameters of candidate models to our experimental data. The model suggests that the groundspeed variability of arista-ablated flies is the result of unstable feedback oscillations caused by the delay and high gain of visual feedback. The antenna response drives active damping with a shorter delay (~20 ms) to stabilize this regulator, in exchange for increasing the effect of rapid wind disturbances. This provides insight into flies' multimodal sensory feedback architecture and constitutes a previously unknown role for the antennae.

Antennal mechanosensory neurons mediate wing motor reflexes in flying Drosophila.

Although many behavioral studies have shown the importance of antennal mechanosensation in various aspects of insect flight control, the identities of the mechanosensory neurons responsible for these functions are still unknown. One candidate is the Johnston's organ (JO) neurons that are located in the second antennal segment and detect phasic and tonic rotations of the third antennal segment relative to the second segment. To investigate how different classes of JO neurons respond to different types of antennal movement during flight, we combined 2-photon calcium imaging with a machine vision system to simultaneously record JO neuron activity and the antennal movement from tethered flying fruit flies (Drosophila melanogaster). We found that most classes of JO neurons respond strongly to antennal oscillation at the wing beat frequency, but not to the tonic deflections of the antennae. To study how flies use input from the JO neurons during flight, we genetically ablated specific classes of JO neurons and examined their effect on the wing motion. Tethered flies flying in the dark require JO neurons to generate slow antiphasic oscillation of the left and right wing stroke amplitudes. However, JO neurons are not necessary for this antiphasic oscillation when visual feedback is available, indicating that there are multiple pathways for generating antiphasic movement of the wings. Collectively, our results are consistent with a model in which flying flies use JO neurons to detect increases in the wing-induced airflow and that JO neurons are involved in a response that decreases contralateral wing stoke amplitude.

Body saccades of Drosophila consist of stereotyped banked turns.

The flight pattern of many fly species consists of straight flight segments interspersed with rapid turns called body saccades, a strategy that is thought to minimize motion blur. We analyzed the body saccades of fruit flies (Drosophila hydei), using high-speed 3D videography to track body and wing kinematics and a dynamically scaled robot to study the production of aerodynamic forces and moments. Although the size, degree and speed of the saccades vary, the dynamics of the maneuver are remarkably stereotypic. In executing a body saccade, flies perform a quick roll and counter-roll, combined with a slower unidirectional rotation around their yaw axis. Flies regulate the size of the turn by adjusting the magnitude of torque that they produce about these control axes, while maintaining the orientation of the rotational axes in the body frame constant. In this way, body saccades are different from escape responses in the same species, in which the roll and pitch component of banking is varied to adjust turn angle. Our analysis of the wing kinematics and aerodynamics showed that flies control aerodynamic torques during the saccade primarily by adjusting the timing and amount of span-wise wing rotation.

Social structures depend on innate determinants and chemosensory processing in Drosophila.

Flies display transient social interactions in groups. However, whether fly-fly interactions are stochastic or structured remains unknown. We hypothesized that groups of flies exhibit patterns of social dynamics that would manifest as nonrandom social interaction networks. To test this, we applied a machine vision system to track the position and orientation of flies in an arena and designed a classifier to detect interactions between pairs of flies. We show that the vinegar fly, Drosophila melanogaster, forms nonrandom social interaction networks, distinct from virtual network controls (constructed from the intersections of individual locomotor trajectories). In addition, the formation of interaction networks depends on chemosensory cues. Gustatory mutants form networks that cannot be distinguished from their virtual network controls. Olfactory mutants form networks that are greatly disrupted compared with control flies. Different wild-type strains form social interaction networks with quantitatively different properties, suggesting that the genes that influence this network phenotype vary across and within wild-type populations. We have established a paradigm for studying social behaviors at a group level in Drosophila and expect that a genetic dissection of this phenomenon will identify conserved molecular mechanisms of social organization in other species.

Death Valley, Drosophila, and the Devonian toolkit.

Most experiments on the flight behavior of Drosophila melanogaster have been performed within confined laboratory chambers, yet the natural history of these animals involves dispersal that takes place on a much larger spatial scale. Thirty years ago, a group of population geneticists performed a series of mark-and-recapture experiments on Drosophila flies, which demonstrated that even cosmopolitan species are capable of covering 10 km of open desert, probably in just a few hours and without the possibility of feeding along the way. In this review I revisit these fascinating and informative experiments and attempt to explain how-from takeoff to landing-the flies might have made these journeys based on our current knowledge of flight behavior. This exercise provides insight into how animals generate long behavioral sequences using sensory-motor modules that may have an ancient evolutionary origin.

Central complex neurons exhibit behaviorally gated responses to visual motion in Drosophila.

Sensory systems provide abundant information about the environment surrounding an animal, but only a small fraction of that information is relevant for any given task. One example of this requirement for context-dependent filtering of a sensory stream is the role that optic flow plays in guiding locomotion. Flying animals, which do not have access to a direct measure of ground speed, rely on optic flow to regulate their forward velocity. This observation suggests that progressive optic flow, the pattern of front-to-back motion on the retina created by forward motion, should be especially salient to an animal while it is in flight, but less important while it is standing still. We recorded the activity of cells in the central complex of Drosophila melanogaster during quiescence and tethered flight using both calcium imaging and whole cell patch-clamp techniques. We observed a genetically identified set of neurons in the fan-shaped body that are unresponsive to visual motion while the animal is quiescent. During flight their baseline activity increases, and they respond to front-to-back motion with changes relative to this baseline. The results provide an example of how nervous systems selectively respond to complex sensory stimuli depending on the current behavioral state of the animal.