Identification and development of measures suitable as potential breeding traits regarding dairy cows´ reactivity towards humans

Behavioural indicators of the human-animal relationship (HAR) are predominantly used in animal welfare science. However, the reactivity of dairy cows – as part of the HAR - is also of interest in the context of dairy breeding, due to its estimated moderate heritability. The avoidance distance (AD) towards an unfamiliar experimenter in a standardized test is regarded an established behavioural indicator of the HAR, but for breeding purposes more feasible measures would be advantageous. The aim of the present pilot study was to identify and develop potential measures of the cow’s reactivity towards humans, which are promising as breeding traits with regard to feasibility, reliability and criterion validity. On three German dairy farms with loose housing and herd sizes of 45 to 195 cows of the Holstein-Friesian and German Black Pied Cattle breed, AD at the feeding place and AD in the barn as well as four alternative HAR measures were recorded and tested for inter-observer reliability (IOR) and inter-test associations for the assessment of criterion validity. Alternative measures were (1) tolerance to standardised tactile interaction (TTI), (2) release behaviour after restraint (RB), (3) qualitative behaviour assessment (QBA) of the cow’s during the TTI and RB Test, and (4) facial hair whorl position and form (HW). TTI, RB and QBA were additionally tested for intra-observer reliability using video recordings of 31 cows. 2 IOR was assessed based on Spearman rank or Kendall W correlation coefficients (in case of QBA with three observers) for metric and ordinal data and based on PABAK coefficients in case of nominal data (HW). Intra-observer reliability was assessed based on Spearman rank correlation coefficients. Inter-test associations between AD at the feeding place and HW were analysed using a General Linear Model and between all other measures using Spearman rank correlation. IOR was good to very good for all measures: AD feeding place rs = 0.79 (n = 84, p < 0.01); AD barn rs = 0.83 (n = 36, p < 0.01); TTI rs = 0.93 (n = 55, p < 0.01); RB rs = 0.90 (n = 54, p < 0.01); QBA W = 0.95 (n = 32, N = 3, p < 0.01); HW PABAK = 0.77 – 0.83 (n = 58). Intra-observer reliability of the alternative behavioural measures was also very good: TTI rs = 0.94 (n = 31, p < 0.01), RB rs = 0.89 (n = 31, p < 0.01); QBA rs = 0.93 (n = 31, p < 0.01). High inter-test correlations were found between AD feeding place and AD barn (rs = 0.77, n = 44, p < 0.01), between TTI and RB (rs = 0.78, n = 52, p < 0.01) as well as between QBA and RB (rs = 0.76, n = 18, p < 0.01). Moderately correlated were QBA and TTI (rs = 0.68, n = 18, p < 0.01), AD feeding place and TTI (rs = 0.50, n = 44, p < 0.01), and AD feeding place and RB (rs = 0.45, n = 43, p < 0,01). No significant associations were found between HW and AD. The present results suggest that TTI, RB and QBA alongside the established AD measures are suitable reactivity measures. They partly reflect similar and partly different aspects of the HAR, with an apparent clustering into distance and handling measures.