By Dr. Hazel Rooney

 

Take home message: Accelerometers such as the HOBO Pendant® G Data Logger are often used on domestic horses for automatic behaviour classification (e.g., lying and standing behaviours). The recommendation drawn from this study is that the logger can be placed on either the front or the hindleg to monitor standing or lying events. However, if lying variants are the focus of behaviour monitoring, it’s recommended that the logger is placed on the front leg of the horse.

 

Introduction: Monitoring the behaviour of domesticated horses, such as standing and lying behaviours, is necessary to promote high standards of animal welfare and care. The use of accelerometer loggers to automatically classify the behaviour of horses, provide an accurate monitoring system, with less inter-observer variation and labour required than visual observations. Although automatic behaviour classification using accelerometers is not a novel technique, research between animal species, sampling intervals, the type of accelerometer and where the logger is placed on the animal is highly variable. Therefore, this study used the HOBO Pendant® G Data Logger to determine the most suitable leg position for mounting an accelerometer (laterally on the left metacarpal or left metatarsal) and the most appropriate sampling interval (10, 20 and 60s) for the automatic classification of standing and variants of lying behaviour in domesticated horses. In addition, the results of this study can be used to develop a model for determining standing and lying behaviours in horses.

Experimental design: The study was conducted at Lluest Equestrian Centre, Aberystwyth University, Wales, UK. Twelve horses were put forward for the study and of these, six were selected by convenience to coincide with the number and location of cameras that were available for to record the lying and standing behaviour of the horses. The horses were of varying breeds including, Sports Horse, Welsh Section D, Cleveland Bay X, and Irish Draught, and were between the ages of 8 and 16.

 

Materials and Methods: The focal behaviours observed in this study were lying and standing, with lying (sternal and lateral) further categorised into lying left (LL), recumbent left (RL), lying right (LR) and recumbent right (RR). Each of the six horses had a HOBO Pendant® G Data Logger attached to the lateral side of their forelimb (left metacarpal) and on their hindlimb (left metatarsal) for a 12-hour period, with sampling intervals programmed for either 10, 20 or 60 seconds each night. The accelerometer data was downloaded each morning using HOBOware software and exported into Microsoft Excel.

 

For ground-truth observations, each stabled horse was continuously recorded for 12 hours using a unique black and white (ANNKE CCTV) camera. After each behaviour recording period, the video recordings were then watched back by one individual and each accelerometer data sheet annotated with the behaviours according to the definitions in Table 1. Data was collected from each horse on three separate nights, with one night at each sampling interval (10 s, 20 s and 60 s). The data was run through a statistical code to determine performance after applying one of three filters (to reduce the effect of possible erroneous data). A simple linear regression was applied to the data to test for significance and predictability.

 

Table 1. Behavioural ethogram of focal behaviours used for the current study. Left - the horses’ left or near-side. Right - the horses’ right or off-side.

 

Results: The significant regression slopes and high R2 values, indicate the very good predictive abilities of the logger to determine standing and lying behaviors. When attached to the front leg, the logger accurately predicted the total lying time, bout frequency of standing and lying variants. When attached to the hind leg, the logger accurately predicted the total lying time, bout frequency of standing and lying variants behaviours. The logger performed best at a 10s sampling interval, when single, likely erroneous events were converted into the behaviour preceding it and was located on the front leg.

 

This dissertation was submitted by Sydney Hatto, while studying for their BSc Equine and Veterinary Bioscience Degree at Aberystwyth University, Wales, UK. I would like to wish Sydney the very best of luck and every success with their future career.

 

Dr. Hazel Rooney, Pig Technical Co-Ordinator, Alltech Ireland

Hazel has been a member of the BSAS Early Career Council since 2020. She works to help pig producers, feed mills and vets to improve the health, welfare, and productivity of pigs in the Irish and European marketplaces.