Accelerometry review dataset

The use of animal-borne devices to measure acceleration has yielded insights into animal locomotion, behaviour and energy expenditure. We present the first systematic review of accelerometry on animals, spanning 60 years (1961-2021), across ecology, biomechanics, agriculture, neurosciences, medical and veterinary sciences from over 400 species and 1,520 studies. The dataset covers 1524 studies. Each row represents an individual study. The variables include the publication title, first author, year of publication, country affiliated with the first author, region (continent) affiliated with the first author, number of authors listed on the publication, countries affiliated with all authors listed on the publication, countries where the data were collected, regions where the data were collected, tag model and manufacturer, sampling frequency bracket, number of accelerometer axes, deployment duration, taxon, sample size by taxon and total sample size of the publication (tags successfully recovered and employed in the study), species, status (whether animals were studied in captivity or in the wild), main research topics and subtopic, the use of additional sensors other than accelerometers, the software used for analysis (only the eight most used softwares are included), and what machine learning approach and algorithm were used for behavioural classification.Most work to date is on terrestrial mammals, particularly domesticated species, while aerial and aquatic animals are relatively under-represented owing to device size and attachment constraints, and just one study has been conducted on amphibians. Despite this, accelerometry research has expanded globally across nearly 100 countries, across all groups of animals including birds, fish, invertebrates, mammals, and reptiles, and across wild and captive animals, including domesticated and non-domesticated species. A variety of software has been used, with open-source software and, more recently, machine learning tools being employed. On-board processing of accelerometry data, including real-time behavioural classification with deep learning to improve accuracy, can reduce power consumption and memory use several-fold, increasing deployment duration. Transfer of device types, sensors and analytical routines between fields in accelerometry offers considerable opportunities for advancing the field. For example, precision livestock farming uses real-time welfare monitoring, which could be adapted for aquaculture or even wild animal monitoring in areas with good data connectivity, while medical sector health monitoring could be used for wild animal disease tracking. The rapid growth in accelerometry requires urgent collaborative approaches to address big data challenges. This includes establishing standardised data repositories, developing open-electronics for low-cost device development and bespoke sensors, fostering capacity building and technical collaboration for device and algorithm advancements. This dataset also highlights geographic inequities in research accessibility and authorship, with less than 6% of accelerometry studies originating from authors from Africa, South America and Asia (excluding Japan, which contributed the second most studies globally) combined. The field must prioritise translational and inclusive research practices and cross-disciplinary collaboration to maximise scientific value and foster innovation while maximising animal welfare.