Assessment of human observer and precision livestock derived technical measures and potential for use in the implementation of the Broiler Directive (2007/43/EC)

Broiler chicken are the most abundant farmed animal in the EU, and a very important component of EU food supply. From July 1st 2010 new welfare rules for meat chickens came into effect across the EU through Directive 2007/43/EC. The Directive permits three stocking densities; Up to 33 kgm2 , up to to 39 kgm2, and up to 42 kgm2. In this paper we describe a study in which manually (human observer) animal outcome measures derived from the Welfare Quality protocols for broiler chickens are collected during farm visits made in the UK, Italy and the Netherlands. These measures are compared with data collected in the same way as that required for the Broiler Directive, and used alongside data including house climate, feed intake and animal growth, collected using camera and sound based automated precision livestock (PLF) farming methods (eYeNamic). Common broiler conditions are recorded by manual observation of birds on the farm or at slaughter - including foot pad dermatitis, hock marks, feather cleanliness, walking ability (gait score), and a score for ‘interaction’ with humans (avoidance distance touch test). Breast lesions, cellulitis, emaciation, joint lesions, scratches and wing fractures were identified as difficult or impossible to assess on farm, and so are collected in the slaughterhouse. Mortality and culling are also recorded, these being derived from routine farm records. Examples of methods - foot pad scoring. Some measures are more amenable to automation, some less so. An example of a manually intensive measure made at the slaughterhouse by visual scoring (using agreed scoring scales as seen in Fig 2a,b) which it has been shown to be possible to automate. A commercial camera system exists to do this, if the cost and technical implications are perceived to be of sufficient value by the production company (Fig 2c). Cleanliness scoring - cameras systems at the slaughterhouse can measure bird ‘colour’ and cleanliness (Fig 3a). It may also be possible to use camera systems on farm to look at bird (and litter) colour changes over time (Fig 3b) as well as to analyse bird distribution on patterns. The data collection required by the Broiler Directive can be time consuming, and can require significant ‘manual’ effort in regard to use of visual scoring of pathologies and uses of rejection at slaughter. It is apparent that there may be the potential to replace manual / human observer recording with automated detection systems, and this is the primary aim of this study. Some measures based on activity and distribution have high potential for camera based automation (Figs 4a,b). There are however measures which, currently do not have technical automated methods - for example, analysis of measures dependent on the behaviour of the birds such as walking ability, feeding behaviour s, or behavioural changes preceding or as a result of disease conditions or due to changes in house conditions. Analysis of the correlations between the manually collected measures and the automated PLF measures permits examination of the variation, and similarities, in the findings derived using PLF and human observer based measures. This experimental ‘joint assessment’ may support progression in automated measurement of environmental and animal parameters, some of which may be potentially fed into the statutory requirement for on-going assessment under the Broiler Directive 2007/43/EC.