EXPLORING STATISTICAL CLASSIFICATION, GIS ANALYSIS AND MAPPING OF HERBIVORE BEHAVIOURS USING ACCELEROMETERS AND HIGH ACCURACY GPS

Electronic sensors equipped with accelerometers have the potential to remotely monitor and record herbivore behaviours. In the UK, sheep are a significant consumer in both managed pasture and upland ecosystems. The ability to automatically collect behavioural data could help inform research into ecosystem functioning and animal welfare. This study evaluated the placement of accelerometers and the ability of data generated to automatically classify four behaviours in sheep; grazing, standing (non grazing), lying head up and lying head down. An application of this method was used to analyse and map data in a GIS and investigate if sheep show a preference for areas with higher fructan levels in grass. Three sheep were fitted with two accelerometers each. One attached to a head halter and one centrally located across the withers by means of a dog harness. Training data were collected and discriminant function analysis was used to develop a model that could predict future unobserved behaviours. Correct classification rates of 95.2%, 91.0% and 91.8% were achieved for each sheep. In the fructan study, although no preference was detected, the study did demonstrate that data from accelerometers can be used to generate behavioural distribution maps. The use of accelerometers is a suitable method for classifying a range of behaviours in sheep.

Name dc:type.qualificationname

mphil

Level dc:type.qualificationlevel

masters

Grantor dc:publisher.institution

Liverpool John Moores University

Year dc:date.issued

2015

Author dc:creator

• Hartley, A

Contributors dc:contributor

• Sneddon, Jenny
• Brown, Richard
• Small, Richard

• HA Statistics
• QL Zoology

DOI dc:identifier.doi

https://doi.org/10.24377/LJMU.t.00004397

OAI identifier oai:identifier

oai:researchonline.ljmu.ac.uk:4397