Quantifying the movement, behaviour and environmental context of group-living animals using drones and computer vision.

Methods for collecting animal behaviour data in natural environments, such as direct observation and biologging, are typically limited in spatiotemporal resolution, the number of animals that can be observed and information about animals' social and physical environments. Video imagery can capture rich information about animals and their environments, but image-based approaches are often impractical due to the challenges of processing large and complex multi-image datasets and transforming resulting data, such as animals' locations, into geographical coordinates. We demonstrate a new system for studying behaviour in the wild that uses drone-recorded videos and computer vision approaches to automatically track the location and body posture of free-roaming animals in georeferenced coordinates with high spatiotemporal resolution embedded in contemporaneous 3D landscape models of the surrounding area. We provide two worked examples in which we apply this approach to videos of gelada monkeys and multiple species of group-living African ungulates. We demonstrate how to track multiple animals simultaneously, classify individuals by species and age-sex class, estimate individuals' body postures (poses) and extract environmental features, including topography of the landscape and animal trails. By quantifying animal movement and posture while reconstructing a detailed 3D model of the landscape, our approach opens the door to studying the sensory ecology and decision-making of animals within their natural physical and social environments.

Vortex phase matching as a strategy for schooling in robots and in fish.

It has long been proposed that flying and swimming animals could exploit neighbour-induced flows. Despite this it is still not clear whether, and if so how, schooling fish coordinate their movement to benefit from the vortices shed by others. To address this we developed bio-mimetic fish-like robots which allow us to measure directly the energy consumption associated with swimming together in pairs (the most common natural configuration in schooling fish). We find that followers, in any relative position to a near-neighbour, could obtain hydrodynamic benefits if they exhibit a tailbeat phase difference that varies linearly with front-back distance, a strategy we term 'vortex phase matching'. Experiments with pairs of freely-swimming fish reveal that followers exhibit this strategy, and that doing so requires neither a functioning visual nor lateral line system. Our results are consistent with the hypothesis that fish typically, but not exclusively, use vortex phase matching to save energy.

DeepPoseKit, a software toolkit for fast and robust animal pose estimation using deep learning.

Quantitative behavioral measurements are important for answering questions across scientific disciplines-from neuroscience to ecology. State-of-the-art deep-learning methods offer major advances in data quality and detail by allowing researchers to automatically estimate locations of an animal's body parts directly from images or videos. However, currently available animal pose estimation methods have limitations in speed and robustness. Here, we introduce a new easy-to-use software toolkit, DeepPoseKit, that addresses these problems using an efficient multi-scale deep-learning model, called Stacked DenseNet, and a fast GPU-based peak-detection algorithm for estimating keypoint locations with subpixel precision. These advances improve processing speed >2x with no loss in accuracy compared to currently available methods. We demonstrate the versatility of our methods with multiple challenging animal pose estimation tasks in laboratory and field settings-including groups of interacting individuals. Our work reduces barriers to using advanced tools for measuring behavior and has broad applicability across the behavioral sciences.

Development of site fidelity in the nocturnal amblypygid, Phrynus marginemaculatus.

Amblypygids are capable of navigation in the complex terrain of rainforests in near complete darkness. Path integration is unnecessary for successful homing, and the alternative mechanisms by which they navigate have yet to be elucidated. Here, our aims were to determine whether the amblypygid Phrynus marginemaculatus could be trained to reliably return to a target shelter in a laboratory arena-indicating goal recognition-and to document changes in behavior associated with the development of fidelity. We recorded nocturnal movements and space use by individuals over five nights in an arena in which subjects were provided with two shelters that differed in quality. The target shelter, unlike the alternative shelter, shielded subjects from light in daylight hours. Individuals consistently exited and returned to a shelter each night and from the third night onward chose the target shelter more often than the alternative shelter. Indeed, on the fifth night, every subject chose the target shelter. This transition was associated with changes in movement and space use in the arena. Notably, the movement features of outbound and inbound paths differed but did not change across nights. Individuals were also characterized by distinct behavioral strategies reflecting candidate homing mechanisms.

Importance of the antenniform legs, but not vision, for homing by the neotropical whip spider Paraphrynus laevifrons.

Amblypygids, or whip spiders, are nocturnal, predatory arthropods that display a robust ability to navigate to their home refuge. Prior field observations and displacement studies in amblypygids demonstrated an ability to home from distances as far away as 10 m. In the current study, micro-transmitters were used to take morning position fixes of individual Paraphrynus laevifrons following an experimental displacement of 10 m from their home refuge. The intention was to assess the relative importance of vision compared with sensory input acquired from the antenniform legs for navigation as well as other aspects of their spatial behavior. Displaced individuals were randomly assigned to three treatment groups: (i) control individuals; (ii) vision-deprived individuals, VD; and (iii) individuals with sensory input from the tips of their antenniform legs compromised, AD. Control and VD subjects were generally successful in returning home, and the direction of their movement on the first night following displacement was homeward oriented. By contrast, AD subjects experienced a complete loss of navigational ability, and movement on the first night indicated no hint of homeward orientation. The data strongly support the hypothesis that sensory input from the tips of the antenniform legs is necessary for successful homing in amblypygids following displacement to an unfamiliar location, and we hypothesize an essential role of olfaction for this navigational ability.

Amblypygids: Model Organisms for the Study of Arthropod Navigation Mechanisms in Complex Environments?

Navigation is an ideal behavioral model for the study of sensory system integration and the neural substrates associated with complex behavior. For this broader purpose, however, it may be profitable to develop new model systems that are both tractable and sufficiently complex to ensure that information derived from a single sensory modality and path integration are inadequate to locate a goal. Here, we discuss some recent discoveries related to navigation by amblypygids, nocturnal arachnids that inhabit the tropics and sub-tropics. Nocturnal displacement experiments under the cover of a tropical rainforest reveal that these animals possess navigational abilities that are reminiscent, albeit on a smaller spatial scale, of true-navigating vertebrates. Specialized legs, called antenniform legs, which possess hundreds of olfactory and tactile sensory hairs, and vision appear to be involved. These animals also have enormous mushroom bodies, higher-order brain regions that, in insects, integrate contextual cues and may be involved in spatial memory. In amblypygids, the complexity of a nocturnal rainforest may impose navigational challenges that favor the integration of information derived from multimodal cues. Moreover, the movement of these animals is easily studied in the laboratory and putative neural integration sites of sensory information can be manipulated. Thus, amblypygids could serve as model organisms for the discovery of neural substrates associated with a unique and potentially sophisticated navigational capability. The diversity of habitats in which amblypygids are found also offers an opportunity for comparative studies of sensory integration and ecological selection pressures on navigation mechanisms.