Newsletters

Ovine Observer

Video footage captured in a walk-over-weigh (WoW) system can be used to assess sheep welfare

Emily Grant, A. Brown, S. Wickham, F. Anderson, A. Barnes, P. Fleming and D. Miller, Murdoch University WA

Corresponding author: E.Grant@murdoch.edu.au

Introduction

The general health and wellbeing of sheep is subjected to many challenges during production. These come from many different sources and along with routine husbandry practices, challenges also stem from changes in management or environmental conditions. Failure of sheep to adapt to these challenges can result in reductions in production performance and, therefore, economic losses.

The perception of animal welfare in sheep production affects the image of the sheep industry in both the global and domestic markets, and therefore affects the profitability and viability of the industry. However, for a welfare assessment tool to be useful it needs to be robust, yet also able to be applied in a practical manner for on-farm usage.

At present current assessment methodologies are cumbersome and difficult to implement, particularly with common on-farm management strategies and labour constraints. New animal farming technologies, such as electronic animal identification (Eid) and automated data capture may provide the opportunity for behaviour, health and welfare to be monitored in a practical and cost-effective manner in both intensive and extensive sheep management systems.

Qualitative Behavioural Assessment (QBA) is a method for remotely evaluating the welfare of sheep and has been proposed as way to utilise Eid and automated data capture for practical, on-farm welfare and health assessment. It is naturally suited for on-farm application, being quick, easy to implement and non-invasive. Furthermore, it is suggested that QBA can integrate with existing farm management systems, such as walk-over-weighing (WoW), and guide welfare assessments to provide a clear and meaningful picture of animal welfare.

In this application, QBA captures the body language of animals, describing how they are behaving using descriptive terminology such as assured, tense and wary. In doing so, QBA captures information concerning how the animal perceives and responds to its environment. Such assessments of body language or behavioural expression can provide insights into the physical and emotional or psychological health, which are relevant to welfare assessment.

This study investigates whether behavioural assessment could be used as an on-farm welfare tool in the sheep industry. It was hypothesised that QBA could be applied in a mock WoW system to identify individual sheep that were; (1) habituated to human presence; (2) lame; or (3) inappetent.

A ‘habituated’ wether voluntarily traveling through the mock walk-over-weigh (WoW) system. Insert shows one of the cameras set up in the WoW system to remotely capture video footage.
A ‘habituated’ wether voluntarily traveling through the mock walk-over-weigh (WoW) system. Insert shows one of the cameras set up in the WoW system to remotely capture video footage

Materials and methods

Video footage was remotely collected from thirty-six Merino wethers within four treatment groups; control (n = 12), habituated (n = 8), lame (n = 8) and inappetent (n = 8) as they traversed, under their own volition, through a mock WoW system. The habituated sheep had been exposed to a low-stress handling regime for six consecutive days prior to filming. Animals were considered to be inappetent when their average feed intake was in the bottom 2.5% of the group over the six days prior to filming. A six-point lameness scoring system was employed to identify lame individuals (0 = not lame; 6 = will not stand or move; average 2.2 ± 0.3). The control animals were not habituated, lame, or inappetent.

The footage of these 36 sheep as they moved through the WoW system was compiled into a series of assessment clips, with one clip per animal. Eighteen observers evaluated the behaviour of the sheep in each of the 36 clips using a free choice profiling approach. This approach allows the observers to first generate their own unique list of descriptive terms to describe sheep behaviour, for example, assured, tense or wary. After watching each clip the observers then scored each animal for each of their own behavioural terms by making a mark on a line which represented minimum to maximum expression (i.e. 0-100).

The relationships between the behaviours scored by the observers were identified using Generalised Procrustes and Principal Component statistical analyses in GENSTAT. The behavioural expression of the four groups of sheep was compared using ANOVA (GenStat 2008, VSN International, UK).

Table 1 Comparison of behaviour expression scores between different treatment groups. Bold values denote significant differences between compared treatment groups

 

Treatment groups

 

 

 

QBA p-values

 

 

Control vs. Habituated

 

 

0.0136

 

Control vs. Lame

 

 

0.0096

 

Control vs. Inappetent

 

 

 

0.0693

 

 

Figure 1 Average (± S.E.) observer behavioural expression scores for each of the four treatment groups. Different letters indicate treatment groups that were significantly different (P < 0.05).
Figure 1 Average (± S.E.) observer behavioural expression scores for each of the four treatment groups. Different letters indicate treatment groups that were significantly different (P < 0.05)

 

Results and take home messages

The observers had similar assessments of the behavioural expression of the sheep in this study (P<0.001). In support of our hypothesis, the assessment of the control and habituated sheep, and the control and the lame sheep, were significantly different (Table 1; P < 0.05). No other significant differences were identified.

Points to note

  1. Despite being blind to treatment groups, observers could differentiate between the control sheep and the habituated or lame sheep
  2. QBA can be used on remotely captured video footage from a WoW system

The habituated and lame sheep consistently received lower scores than the other groups, being described by the observers as more focused/collected/assured than the control animals (Figure 1.).

In contrast, there was no significant difference between the observer scores given to the inappetent animals compared to the control. This suggests that further investigation into the sensitivity of QBA to certain disease states is necessary given the inability to distinguish the inappetent sheep from the control.

Conclusion

Our findings demonstrate that observers can use the behaviour of sheep to differentiate between sheep that were lame, or acclimated to their immediate environment (habituated), and those that were not (control animals), from film captured as they walked through a WoW system. These results suggest that differences in the way animals interact with mild stimuli in the environment (e.g. the WoW system in this experiment) can be identified using QBA. Thus, QBA represents a simple tool that could potentially be applied to video footage taken by an automatic data capture system (on farm) to provide meaningful information concerning sheep welfare.

Acknowledgements

This research was funded by the Cooperative Research Centre for Sheep Industry Innovation (Sheep CRC). The Wellard La Bergerie Feedlot is thanked for the wethers and facilities.