Animals rely on sensory feedback to achieve flexible coordinated movements, such as walking on rough terrain or grasping an object with appropriate force. Mechanosensory feedback is especially critical, as it operates with remarkable speed and sensitivity. One challenge in understanding these systems is the fact that mechanosensory inputs are transformed by the body structures in which sensory receptors are embedded. This transformation is particularly complex in the case of insect flight, where applied forces and torques are transformed by wing structural mechanics into complex spatiotemporal patterns of wing bending. Information about these dynamic shape changes is encoded by strain-sensitive neurons embedded in the wings, but it remains unclear which features of wing bending drive neural responses. Here we elucidate this feature space, which is essential to understanding the sensory information available to the animal during flight and to explore hypotheses for processing by downstream circuits. We record from primary mechanosensory neurons as the wing is driven through a range of motions and use simultaneous measurements of wing position to reconstruct the full wing shape over time. We then characterize the temporal features encoded by individual neurons using several analyses, including reverse correlation, covariance analysis, and maximally informative dimensions. Our results show that a diversity of features are encoded by this population and that single neurons are selective for multiple distinct features, suggesting a richer feature space is encoded by these neurons than previously known.