Paul, E. S., Browne, W., Mendl, M. T. et al. 2022. Assessing animal welfare: A triangulation of preference, judgement bias and other candidate welfare indicators. Animal Behaviour 186, 151-177.

To assess the welfare of captive animals, validated measures, so-called ‘welfare indicators’, are required. We used a triangulation approach to investigate the extent to which different measures converged to provide corroborating evidence of welfare. Laying hens were exposed to living conditions designed to be generally preferred (GP) or generally nonpreferred (GNP), using previous studies of chickens’ majority preferences for resources and environments. The hens were also tested at the end of the study to identify their individual preferences for these living conditions, assigned to groups that showed an individual preference, or nonpreference, for their own experimental housing, regardless of whether it was generally preferred or not (IP and INP). Both GP and IP birds showed more ground-foraging behaviour, and lower pulse rates during handling, than GNP and INP birds. Individual preference was associated with more optimistic-like judgement biases when birds were tested after 6 weeks of exposure to these living conditions, but not after 24 weeks. Serum blood glucose levels were also lower in hens showing individual preferences for their experimental living conditions. General preferences were associated with a number of measures, including higher rates of ground-foraging behaviour and lower faecal water content (after both 6 and 24 weeks), lower pulse rate during handling and greater tibia strength and stiffness post mortem. There were no associations between judgement bias and other candidate welfare indicators, but it is not clear whether this represents evidence of absence or merely absence of evidence. Overall, the different approaches did not converge to identify a precise state of animal welfare, although some measures (preference, stress indicators: pulse rate, faecal water and blood glucose, foraging behaviour) were aligned consistently across timescales. We conclude that further work is needed to establish which alternative measures of affective state might be more appropriate indicators of animal welfare.