Is cows' qualitatively assessed behaviour towards humans related to their general stress level?

Summary

The stress level is considered as one important aspect of dairy cow welfare. Beside aspects of housing, management and social herd stressors, stress might also be related to the human-animal relationship (HAR). An established and non-invasive physiological method to assess medium-term stress is the measurement of fecal cortisol metabolites (FCM), reflecting the adrenocortical activity over several hours with a delay of 8-10 hours due to gut passage time. For the assessment of the HAR, behavioural tests recording the cows´ responses towards humans can be used. Beside quantitative methods, e.g. recording the cows´ avoidance distances towards humans, also the qualitative behaviour assessment (QBA) assessing the cows´ body language have been shown to be reliably applicable. Thereby, QBA might reflect the cows´ responses in a more differentiated way than quantitative measures. Using data from 316 dairy cows on 25 German organic dairy farms regarding QBA during a standardised tactile human-animal interaction and FCM recorded on the same day, but with time differences varying between 0-10 h (during winter 2015/16 and 2016/17), we asked whether the cows’ qualitatively assessed behaviour towards humans is related to their general medium-term stress level. For QBA we used a fixed list of 20 descriptors, which had specifically been developed for this purpose. A principal component analysis (PCA) resulted in two components: PC1 explained 67% of variance and appeared to reflect ‘positive’ (pos) and ‘negative’ (neg) valence (characteristic descriptors pos: e.g. trustful, relaxed; neg: e.g. fearful, distressed). PC2 explained 7% of variance and appeared to reflect the level of activation (‘high’ activation: e.g. contact-seeking, aggressive; ‘low’ activation: e.g. patient, insecure). Dividing the sample into four groups (pos_low, pos_high, neg_low, neg_high) using the medians of PC1 and PC2 as cut-points, we compared FCM levels (11,17 dioxoandrostanes, enzyme immunoassay method) by Kruskal-Wallis and post-hoc Wilcoxon rank-sum tests. Groups of cows classified by QBA differed regarding FCM (P=0.010, Kruskal-Wallis test). Cows reacting pos_high during the human-animal interaction (n=65), pos_low (n=93) and neg_low (n=99) had similar FCM levels (medians of 12.0, 10.3 and 10.4; 25-75% quartiles: 6.8-22.8, 4.9-22.0, and 4.2-18.5 ng/g FCM; P=0.225-0.699, Wilcoxon tests). In contrast, neg_high cows had significantly lower FCM levels (6.5 ng/g, 3.7-12.5, n=59) than all other groups (P=0.001-0.029). Neg_high cows presumably had higher fear levels towards humans than pos cows, and expressed this more actively than neg_low cows. Unexpectedly, however, the neg reaction was not related to a generally higher stress level. In this context, it must be taken into account that FCM medians and variation in the investigated sample were generally on a low level. Furthermore, other factors such as social rank or health status might have affected results more profoundly, deserving a more complex analysis including individual factors as a next step.