A comparison of two certification schemes for dairy cow welfare in relation to resource-based, management-based and animal-based measures

Traditionally risk-based measures have been used to assess animal welfare in legislation and certification schemes. With increased knowledge and research, debate today tends to focus on and advocate the use of animal-based measures. However this is not a panacea for welfare and the most sound conclusion should be that assessment of animal welfare needs both. The aim of this thesis is to identify the points that need to be taken into consideration to find the optimal balance in the use of input and outcome measures according to the aim of the assessment and how these should be applied when designing a certification scheme? To answer this, a comparison of the use of resource-based, management-based and animal-based measures in two certification schemes for dairy cow welfare, Freedom Food (UK) and Svenskt Sigill (SWE), was made. To enable the comparison, six general categories regarding welfare were defined (feed, water, health, environment, management and behaviour) and assessment points of the two schemes divided accordingly. Both schemes predominantly use input measures with the exception of a slightly more balanced use of input measures and outcome measures regarding both behaviour and health. There was a tendency of animal-based measures being non-specific, thereby not being valid or reliable. The conclusion drawn from this thesis is that there are too many parameters that might differ between different certification schemes, e.g. number of assessment visits and assessment time, budget, etc., and therefore a statement of a general approach cannot be made. Instead I describe a number of key points that are important to consider when choosing between and balancing the use of risk-based and animal-based measures: 1) Risk-factors known for impairing welfare should not be allowed. 2) The difficulty to modify some resources, especially those regarding housing, needs to be taken into account. Therefore the use of prior approval is recommended. 3) How long it takes for a change in a risk-based measure to affect the animal-based measure is important, e.g. breeding versus amount of bedding. 4) When the predictive value, i.e. the correlation, to a risk-based measure is high it can replace the animal-based measure. 5) Animal-based measures used must be valid, repeatable and feasible. 6) To consider the use of automated assessment of animal-based measures to make them feasible. 7) That animal-based measures are necessary for benchmarking and the implementation of threshold levels.