Ecological inference using data from accelerometers needs careful protocols

Machine Learning Methods and Visual Observations to Categorize Behavior of Grazing Cattle Using Accelerometer Signals

Accelerometers worn by animals produce distinct behavioral signatures, which can be classified accurately using machine learning methods such as random forest decision trees. The objective of this study was to identify accelerometer signal separation among parsimonious behaviors. We achieved this objective by (1) describing functional differences in accelerometer signals among discrete behaviors, (2) identifying the optimal window size for signal pre-processing, and (3) demonstrating the number of observations required to achieve the desired level of model accuracy,. Crossbred steers (Bos taurus indicus; n = 10) were fitted with GPS collars containing a video camera and tri-axial accelerometers (read-rate = 40 Hz). Distinct behaviors from accelerometer signals, particularly for grazing, were apparent because of the head-down posture. Increasing the smoothing window size to 10 s improved classification accuracy (p < 0.05), but reducing the number of observations below 50% resulted in a decrease in accuracy for all behaviors (p < 0.05). In-pasture observation increased accuracy and precision (0.05 and 0.08 percent, respectively) compared with animal-borne collar video observations.

Turbulence causes kinematic and behavioural adjustments in a flapping flier

Turbulence is a widespread phenomenon in the natural world, but its influence on flapping fliers remains little studied. We assessed how freestream turbulence affected the kinematics, flight effort and track properties of homing pigeons (Columba livia), using the fine-scale variations in flight height as a proxy for turbulence levels. Birds showed a small increase in their wingbeat amplitude with increasing turbulence (similar to laboratory studies), but this was accompanied by a reduction in mean wingbeat frequency, such that their flapping wing speed remained the same. Mean kinematic responses to turbulence may therefore enable birds to increase their stability without a reduction in propulsive efficiency. Nonetheless, the most marked response to turbulence was an increase in the variability of wingbeat frequency and amplitude. These stroke-to-stroke changes in kinematics provide instantaneous compensation for turbulence. They will also increase flight costs. Yet pigeons only made small adjustments to their flight altitude, likely resulting in little change in exposure to strong convective turbulence. Responses to turbulence were therefore distinct from responses to wind, with the costs of high turbulence being levied through an increase in the variability of their kinematics and airspeed. This highlights the value of investigating the variability in flight parameters in free-living animals.

Daily activity timing in the Anthropocene

Animals are facing novel 'timescapes' in which the stimuli entraining their daily activity patterns no longer match historical conditions due to anthropogenic disturbance. However, the ecological effects (e.g., altered physiology, species interactions) of novel activity timing are virtually unknown. We reviewed 1328 studies and found relatively few focusing on anthropogenic effects on activity timing. We suggest three hypotheses to stimulate future research: (i) activity-timing mismatches determine ecological effects, (ii) duration and timing of timescape modification influence effects, and (iii) consequences of altered activity timing vary biogeographically due to broad-scale variation in factors compressing timescapes. The continued growth of sampling technologies promises to facilitate the study of the consequences of altered activity timing, with emerging applications for biodiversity conservation.

The role of individual variability on the predictive performance of machine learning applied to large bio-logging datasets

Animal-borne tagging (bio-logging) generates large and complex datasets. In particular, accelerometer tags, which provide information on behaviour and energy expenditure of wild animals, produce high-resolution multi-dimensional data, and can be challenging to analyse. We tested the performance of commonly used artificial intelligence tools on datasets of increasing volume and dimensionality. By collecting bio-logging data across several sampling seasons, datasets are inherently characterized by inter-individual variability. Such information should be considered when predicting behaviour. We integrated both unsupervised and supervised machine learning approaches to predict behaviours in two penguin species. The classified behaviours obtained from the unsupervised approach Expectation Maximisation were used to train the supervised approach Random Forest. We assessed agreement between the approaches, the performance of Random Forest on unknown data and the implications for the calculation of energy expenditure. Consideration of behavioural variability resulted in high agreement (> 80%) in behavioural classifications and minimal differences in energy expenditure estimates. However, some outliers with < 70% of agreement, highlighted how behaviours characterized by signal similarity are confused. We advise the broad bio-logging community, approaching these large datasets, to be cautious when upscaling predictions, as this might lead to less accurate estimates of behaviour and energy expenditure.

The role of wingbeat frequency and amplitude in flight power

Body-mounted accelerometers provide a new prospect for estimating power use in flying birds, as the signal varies with the two major kinematic determinants of aerodynamic power: wingbeat frequency and amplitude. Yet wingbeat frequency is sometimes used as a proxy for power output in isolation. There is, therefore, a need to understand which kinematic parameter birds vary and whether this is predicted by flight mode (e.g. accelerating, ascending/descending flight), speed or morphology. We investigate this using high-frequency acceleration data from (i) 14 species flying in the wild, (ii) two species flying in controlled conditions in a wind tunnel and (iii) a review of experimental and field studies. While wingbeat frequency and amplitude were positively correlated, R² values were generally low, supporting the idea that parameters can vary independently. Indeed, birds were more likely to modulate wingbeat amplitude for more energy-demanding flight modes, including climbing and take-off. Nonetheless, the striking variability, even within species and flight types, highlights the complexity of describing the kinematic relationships, which appear sensitive to both the biological and physical context. Notwithstanding this, acceleration metrics that incorporate both kinematic parameters should be more robust proxies for power than wingbeat frequency alone.