Results and discussion

In study 1, when no apparent environmental perturbation occurred, significant variations of milk cortisol concentrations between- and within-day were not observed (Fig. 1, online Supplementary Table S2). Despite the demonstration that environmental conditions and management can affect the circadian rhythm of cortisol secretion others (Mormede et al. Reference Mormede, Andanson, Auperin, Beerda, Guemene, Mamkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007; Ogino et al. Reference Ogino, Matsuura, Yamazaki, Irimajiri, Suzuki, Kushibiki, Singu, Kasuya, Hasegawa and Hodate2014) did not detect a rhythm in cortisol release in well fed, free ranging cows, while they observed a clear circadian rhythm in cows under tie-stall and restricted feeding conditions. Under our experimental conditions, as in most commercial farms, animals were free ranging and nutritional requirements were adequately met. Likely, these conditions did not favour a circadian rhythm of cortisol release and, considering that milk cortisol reflects plasma free cortisol variations between two consecutive milking (Shutt & Fell, Reference Shutt and Fell1985), this led to the lack of difference in milk cortisol concentrations between morning and afternoon milking. However, these observations cannot be generalised and within-day cortisol release should be also analysed under different environmental and farming situations.

Fig. 1. Effect of breed, day of sampling and time of sampling on cortisol concentration in milk (ln of pg/ml) in Study 1. Milk samples were collected in three consecutive days (D1, D2, D3) at 6:00 AM and 6:00 PM. HF, Holstein Frisian; NR, Norwegian Red; AM – before noon milk sampling; PM – after noon milk sampling; Variations in milk cortisol concentrations within day, between days of sampling and between breeds resulted to be not significant.

The significant increase (P < 0·01) in milk cortisol concentrations on days 3, 4 and 5 (Table 1), observed in Study 2, suggests that relocation is a trigger for the HPA axis in the bovine species. Indeed, Gupta et al. (Reference Gupta, Earley, Ting and Crowe2005) observed a significant increase in plasma cortisol in steers after repeated regrouping and relocation. In lactating cows, the increase in plasma cortisol induced by relocation is the most likely explanation for the subsequent increase in milk cortisol.

Table 1. Effects of breed, group of lactation, day of sampling and parity on milk cortisol concentration and milk yield in Study 2

HF, Holstein Frisian; NR, Norwegian Red; PF: from postpartum to fresh cows; FH: from fresh to high production cows; HL: from high production to low production cows; Primiparous: first calving cows; Pluriparous: second and third calving cows; ns: not significant; values within columns with different letters A, B, C differ significantly, P < 0·01.

The differences in milk cortisol concentrations following the relocations can hardly be explained by differences in milk yield. Indeed, milk yield was affected (P < 0·01) by variables related to the stage of lactation (production group). On the contrary, none of the variables related to the stage of lactation showed any effect on milk cortisol concentrations. Fukasawa & Tsukada (Reference Fukasawa and Tsukada2010) reported that fresh cows showed significantly higher milk cortisol concentrations in comparison to mid and late lactating cows, thus suggesting that the stage of lactation can affect milk cortisol concentrations. However, in the study of Fukasawa & Tsukada (Reference Fukasawa and Tsukada2010) the group of fresh cows included animals between 7 and 90 DIM (days in milking), while in Study 2 of this paper all animals in the PF group were beyond 50 DIM.

In addition, our previous observations (Gabai, unpublished data), indicate that milk cortisol concentrations are significantly higher (P < 0·01) in the first week postpartum (1·52 ± 0·14 ng/ml), decrease by the second week of lactation (0·86 ± 0·12 ng/ml) and then remain stable as far as day 30 after parturition (0·83 ± 0·12 ng/ml). Moreover, potential effects on milk cortisol related to mammary inflammation can be excluded, as SCC in our animals was lower than 200 000 cells/ml and an increase in milk cortisol concentration can be only observed when SCC is higher than 400 000 cells/ml (Sgorlon et al. Reference Sgorlon, Fanzago, Guiatti, Gabai, Stradaioli and Stefanon2015). Thus, relocation was the prominent stimulus for HPA activation and milk cortisol increase.

Milk yield showed a significant (P < 0·01) increase on day 3 followed by a decrease on days 4 and 5 (Table 1). The transient increase in milk yield could be explained by the positive effect of cortisol on glycaemia (Sgorlon et al. Reference Sgorlon, Colitti, Asquini, Ferrarini, Pallavicini and Stefanon2012), possibly related to increased peripheral gluconeogenesis. However, this temporary increase in milk yield lasted only 1 d and then decreased. The effect of cortisol on milk yield is controversial, as most information derives from exogenous ACTH/glucocorticoids administration (van der Kolk, Reference Van der Kolk1990). However, stress and glucocorticoids may affect milk yield by different mechanisms (Silanikove et al. Reference Silanikove, Shamay, Shinder and Moran2000).

Despite the two breeds being present on the same farm under the same environmental conditions, breed differences in milk cortisol were not observed in either study. Neither Parity nor the Breed × Parity interaction influenced milk cortisol concentrations. However, we observed that the extent of the response to the environmental challenge was very variable among cows and, in our opinion, this observation deserves further consideration. Here, the reduced number of animals does not allow us to identify subpopulation of cows with different response intensity. In a previous work (Sgorlon et al. Reference Sgorlon, Fanzago, Guiatti, Gabai, Stradaioli and Stefanon2015), we observed a significant difference in milk cortisol concentrations between Italian Holstein and Italian Simmental cows and hypothesised that the different ability to cope with milk yield could explain, at least in part, that difference. It is important to consider that, in that previous work, milk cortisol was measured in animals belonging to fourteen farms, and that one breed only was present in each farm. Indeed, mean milk cortisol concentrations were different among farms and Holstein herds did not necessarily show the higher cortisol concentrations. This observation suggests that different environmental conditions can affect the ‘basal’ HPA axis activity, and superimpose the genetic background (Mormede et al. Reference Mormede, Andanson, Auperin, Beerda, Guemene, Mamkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007).

Up to now, the limited amount of data available on milk cortisol concentrations does not allow the identification of a threshold to define cows’ wellbeing. In cattle, the function of the HPA axis is not well characterised. Studies on rhythmicity, responses to pharmacological treatments and to chronic stress led to inconsistent results and available data are far from conclusive (Mormede et al. Reference Mormede, Andanson, Auperin, Beerda, Guemene, Mamkvist, Manteca, Manteuffel, Prunet, van Reenen, Richard and Veissier2007). Therefore, more studies are needed for better understanding of both environmental and genetic factors affecting milk cortisol concentrations. However, based on the results of this work, cortisol measurement in milk samples collected during consecutive days can be considered as a suitable biomarker to assess the individual HPA response of dairy cows to the same short/medium-term environmental challenge and the overall conditions of the herd.

Further studies are required to investigate if milk cortisol can be useful for prediction of animals with an increased risk of developing diseases and to aid the selection of more resilient animals.