Avoidance distance in dairy cattle as a behavioural indicator of fear responses: the confounding role of clinical health status in two Algerian dairy farms.

Fear responses toward humans are a central determinant of dairy cattle welfare and productivity. Avoidance distance (AD), the minimum approach distance tolerated by a stationary animal at the feeding rack, is a behavioural indicator of human-animal relationship (HAR) quality used in the Welfare Quality® protocol. This preliminary, hypothesis-generating study examined predictors of AD ≥ 50 cm in 72 dairy bovines (Montbéliarde and Holstein Friesian) from two commercial farms in the wilaya of Aïn Defla, Algeria (Sidi Omer, n = 36; El-Hamoul, n = 36), using univariate chi-square/Fisher's exact tests and a generalised estimating equations (GEE) logistic regression with an exchangeable working correlation structure. Mastitis was by far the strongest correlate of avoidance: all 15 affected animals at Sidi Omer were avoidant (Fisher's exact p < 0.0001), consistent with pain-induced hypervigilance. Age was significant at Sidi Omer (χ² = 17.05; p = 0.0002) but not at El-Hamoul, where tolerance was near-universal. Sex was not significant at Sidi Omer alone (p = 0.149), and the nominally significant combined-sample result (χ² = 7.14; p = 0.0075) did not survive Bonferroni correction. A large between-farm difference was observed (χ² = 24.27; p < 0.0001): 97.2% of El-Hamoul animals were tolerant versus 44.4% at Sidi Omer. Farm membership was almost completely confounded with health status, so this contrast cannot be attributed to stockperson behaviour specifically. These findings support recording health status at AD testing and generate hypotheses on pain-related hypervigilance and farm-level management warranting confirmation in larger, multi-farm studies with direct assessment of stockperson behaviour.