Final Report Summary - BRAINVISUALECOLOGY (Linking neuronal activity to ecology: How the sensory environment of a species shapes the neural representation of the external world in the insect brain.) The project entitled ‘Linking neuronal activity to ecology: How the sensory environment of a species shapes the neural representation of the external world in the insect brain’ was carried out over two years (mid 2013 to mid 2015) in the Lund Vision Group at Lund University, Sweden. It focused on the problem of how brains integrate sensory information to guide animals during complex orientation behavior. First, we aimed at identifying the conserved core circuits underlying navigation behavior and, second, we tackled the question of how those circuits have adapted to match the ecology of individual species. As insects perform very sophisticated navigation behaviors, e.g. long distance navigation or central-place foraging in complex environments, and their brains are easily accessible for physiological studies, they are ideal model organisms for our approach. From earlier work in desert locusts and monarch butterflies, the central complex, a highly conserved region in the center of the insect brain, has emerged as a potential navigation center of the insect brain. In these long-range navigating insects, skylight compass cues (e.g. polarized skylight) are mapped onto the regular neuroarchitecture of this brain region. The population of neurons comprising this ‘internal compass’ can be viewed as an ordered array of head direction cells that encode body orientation with respect to the position of the sun. Whether this system is restricted to long-range navigating insects was the first question of this work. As bees are non-migratory insects, but show highly sophisticated orientation behaviors, we chose several species of bees as models to investigate this question. If compass cues are indeed represented in the central complex of bees, the second question was how this representation differs from the well studied one found in long-range navigating species and whether differences between the chosen bee species can be found that correlate with their sensory environment. The solitary sweat bee Megalopta genalis was used as our primary model. It inhabits dense tropical rainforests and is active only at night. Diurnal sweat bees and diurnal bumblebees were investigated for comparison, as they inhabit a vastly different environment - brightly lit, open fields in Sweden. All species exhibit a behavior called central-place foraging, i.e. they forage in an area around a centrally placed nest to which they return in a straight line even after a convoluted foraging trip. One strategy that is used by bees to achieve this is path integration. For path integration, all angular changes during a foraging trip have to be monitored and combined with the information of how much distance was covered in each compass direction. Both pieces of information are combined to compute a homing vector, which continuously points into the direction of the nest at each moment during a foraging trip. The neural basis of this behavior had been completely unknown. As compass information (i.e. a sense of direction) is required for long-range navigation, as well as for path integration, we hypothesized that the central complex also serves as an internal compass in bees and that it has been adapted to match the requirements of the specific sensory environment of each species and the behavioral demands imposed by central-place foraging. To investigate these questions, we aimed at performing intracellular recordings in our bee species, combined with detailed neuroanatomical studies. To efficiently carry out this work on a tropical bee species, we established a state of the art electrophysiological setup at the Smithsonian Tropical Research Institute in Panama, close to the natural habitat of Megalopta genalis. This recording setup now provides us with the unique opportunity to perform in vivo neurophysiological studies on site in Panama, in direct proximity to one of the most diverse and species-rich habitats on the planet. Additionally, we aimed at expanding the set of stimuli previously used to characterize compass neurons. We therefore developed a novel, LED-based virtual reality arena that combines an artificial ultraviolet sky with a 360 degree LED arena consisting of thousands of interspersed UV and green LEDs, which can be used to display a wide range of motion stimuli, spectral stimuli, and natural scenes. This setup became fully operational during the second year of the project and two similar versions now exist for parallel work, one in Lund, Sweden, and the other in Panama. Finally, to enable more transparent and efficient data storage, as well as wide accessibility of data, we developed an online database for storing, accessing, and displaying 3D and 2D morphological data of individual bee neurons in conjunction with physiological data. We have now expanded this project by establishing an international work group involving multiple laboratories across Europe and have developed the database into a multi-species platform for storing, referencing, accessing and displaying neurophysiological and anatomical data. Emphasizing easy user interaction for up- and downloading datasets and establishing a distributed curator system, this new repository will serve to increase transparency, comparability, durability, and accessibility of data resulting from neuroethological research across an increasing range of model species. The three points mentioned thus far are all major technical achievements that will facilitate research in our field in multiple ways, firstly by making tropical species accessible to functional neurophysiology, secondly by developing a sophisticated, modular, easy to reproduce stimulus display system, and thirdly by providing a shared platform to access anatomical and physiological data from many insect species. Using our newly established infrastructure, we have indeed identified neurons in the bee central complex that strongly respond to polarized light in a way resembling that of migratory species. Anatomically, all components of the compass network present in locusts and butterflies were shown to have counterparts in Megalopta genalis. We were able to functionally examine neurons at the input and output stages of the central complex and found physiological responses reflecting the anatomical data. These data suggest that the central complex in bees serves a similar role in navigation as the homologous circuits in migratory insects, despite their vastly differing behavioral strategies. Characteristic differences in the details of responses to visual compass cues between homologous neurons, between locusts/monarch butterflies on the one hand and Megalopta genalis on the other, have revealed that the only stimuli that are encoded are those that are reliably present in the habitat during the bee’s activity period. This suggests that the detailed features of the bee’s compass circuit indeed reflect the ecological specializations of this species. Given that the compass circuit appears to be highly conserved independently of the lifestyle of insect species, the differences in insect behavior must be based on the information that is combined with directional information to produce species-specific behavior. To reveal what this information is, and how it is combined with the internal compass, we expanded the stimuli presented to central-complex neurons to include visual cues that do not represent compass information. Using rotational and translational optic flow, as well as small-field stimuli to map receptive fields, we identified and characterized an entire new network of cells that encodes large-field motion, i.e. represent self-generated motion cues resulting from forward movement or rotational movements of the bee. Surprisingly, this novel pathway converged with the compass pathway in the central complex. The neurons involved specifically innervate a small compartment of the central complex, the noduli, which had not previously been assigned a function in any species. Not only have we found elements of this potential path-integration circuit in the tropical sweat bee Megalopta genalis, we also identified major parts of the pathway in the distantly related, diurnal bumblebee, indicating a high degree of conservation at least across bees. As bees use translational optic flow to estimate the distance covered during foraging flights, this convergence of a visual compass with an optic-flow representation suggests that the bee central complex might be the long-sought neural substrate of path integration. Moreover, the anatomical outline of the circuit resembles some technical solutions to the problem of path integration remarkably well. Expanding on this unexpected finding, we have begun to develop biologically realistic, computational models of central complex neural circuits in close collaboration with Prof. B. Webb (University of Edinburgh, UK). These models are aimed at verifying the viability of our hypothesis and to generate testable predictions for the behavior of the circuit in novel stimulus situations. Overall, our findings have confirmed the fundamental importance of a conserved compass system in the insect central complex and, maybe more importantly, have pointed us towards an understanding of how this compass circuit can be extended to serve as the neural substrate of path integration, one of the most common and most robust navigation strategies amongst all animals. These simple, highly conserved algorithms for animal navigation could potentially have a fundamental impact on the reliability and efficiency of autonomously navigating robots used across a wide range of applications.
