**4. Discussion**

Our results confirmed that concentrations of cortisol (or its metabolites) in plasma, feces and hair samples taken at the same time are not correlated. These results were expected since each sample matrix provides information about the endocrine state at different times. GC levels in plasma reflect an immediate physiological state [12], while feces provide information about the endocrine state a certain time before the sample collection [11] and hair is supposed to reflect the status of an accumulated period of some weeks to a few months [22]. Therefore, each biological sample matrix reflects distinct time-windows and thus complementary information can be gained.

The levels of cortisol and its metabolites did not differ between sex or age classes. These results were in agreemen<sup>t</sup> with those obtained by Huber et al. [17] who did not find differences in the concentration of FCMs between female and male red deer. However, other studies described sex-specific differences in glucocorticoid levels for some ruminant species [47,48] as well as age variations [49]. Regarding the influence of the reproductive state on stress levels, changes in cortisol levels were reported in red

deer, with older females having higher cortisol levels in late gestation than non-pregnan<sup>t</sup> females [50]. However, for reindeer (*Rangifer tarandus*) no di fferences were found in plasma cortisol concentrations between adult males, barren, and pregnan<sup>t</sup> females [51]. Although the reproductive state of the females was not assessed in the present study, previous long-term data from the same study area has shown that during the sampling period more than 80% of females were usually pregnan<sup>t</sup> (unpublished data), which means that probably more than 80% of the sampled females for this study were also pregnant. This fact, together with the absence of di fferences between males and females or age classes suggests that all animals were exposed to the same levels of stress during the sampling period. In fact, considering the type of hunting process used in our study area, which is not targeting any particular sex or age class, the obtained results are in agreemen<sup>t</sup> with our predictions. In fact, the absence of di fferences between males and females or age classes suggests that all animals were exposed to the same levels of stress during the sampling period.

We found a trend in plasma cortisol concentrations to decrease during the hunting season from November to February. A decrease of GC concentrations after regular and frequent occurrences of a stressor is often an indication of acclimation [2], as observed in some studies with Brahman cattle (*Bos taurus indicus*) and Magellanic penguins (*Spheniscus magellanicus*) [52,53]. However, besides the fact that we are dealing with a major stressor in our study, we need to consider the possible influence that this specific hunting method may have on our results. Since we have no information for how long deer were chased by the dogs before being killed, the cortisol values in plasma could reflect di fferent chasing periods. Furthermore, the observed trend for plasma may also be influenced by other factors, such as food intake or/and any environmental disturbance [54–56], which make an interpretation more di fficult. Although an influence of circadian rhythm in cortisol levels has been reported [55], the hunting events in the study area occurred within the same period of the day, decreasing the possible effect of daily circadian rhythm in our results.

Besides the trend obtained for plasma cortisol concentrations, the results also showed a positive correlation of plasma cortisol levels and the percentage of lymphocytes. We expected the opposite, namely a decrease of the percentage of lymphocytes with the increase of the stress levels, mainly due to the e ffect of stress as an immune suppressor [57]. Instead, our results were more in line with the immune-enhancing character of acute stress, promoting the passage of leukocytes from the blood to other parts of the body, while chronic stress induces immune suppression [57–59]. Although this interpretation requires caution given the existence of some unevaluated health-related factors, it points to the important relation between white blood cells and cortisol levels, and the need for more studies approaching this interaction.

We did not find di fferences in FCM concentrations across the sampled months. However, a seasonal pattern of GC levels in cervids is suggested by some authors who documented higher values in colder months than in warmer months of the year [17,60]. The minor influence of Mediterranean mild winters [34] where the occurrence of snow is uncommon and food availability is not significantly affected may have contributed to the lack of a seasonal pattern in our FCM levels during the hunting season. In fact, some studies in Mediterranean red deer [61] and roe deer (*Capreolus capreolus*) [62] suggested summer as the season with the most energetic constraints due to decreased food quality and quantity due to hydric stress. These results point, although weakly, given the small control sample size in terms of control group, to an influence of hunting in the stress levels. These results are in agreemen<sup>t</sup> with those reporting higher FCM levels in chamois (*Rupicapra rupicapra tatrica*) in areas with high disturbance than at low disturbance locations [16]. Hair cortisol concentrations di ffered significantly across months, with an increase from the beginning (October) to the end (February) of the hunting season. Cortisol levels recorded in February were significantly higher than the ones obtained in October. Taking into account that the molt from the summer to the winter coat is gradual, especially in adults where the development of new hair can occur before shedding the old one [63], our samples from October included new hair, which began to grow in September and October, and hair from the summer months which had not been shed yet. Furthermore, once hair follicle activity is reduced in February, because the end of the winter season is getting close and the winter coat is fully grown [63], cortisol measured in hair that was sampled in this month should largely reflect the conditions from the previous three months [64]. Therefore, the increase in cortisol levels across months may be an indication of a period of prolonged stressful conditions induced by hunting activity, which is supported by the higher FCM levels in individuals from impact areas than from control sites found in our results. Similarly, Caslini et al. [65] found that hair cortisol levels in the same species (*Cervus elaphus*) were higher in greater density areas associated with more di fficult environmental conditions and higher levels of anthropogenic disturbance (such as tourism). Our results are in agreemen<sup>t</sup> with those reported in the mentioned study which suggested that long-term HPA axis activity and allostatic load, as a consequence of higher densities, anthropogenic disturbances and/or environmental conditions in red deer populations, can be evaluated using cortisol hair levels as an index [65]. In addition, Bryan et al. [66] also documented higher hair cortisol levels in heavily hunted wolves (*Canis lupus*) than in wolves with lower hunting pressure. Hair cortisol seems to be a good indicator of long-term stress and has gained importance as a novel method to assess stressful conditions [19,22]. Hair can often be collected without capture and handling of the animals (i.e., hair traps). Collection of hair for hormone analyses may thus be a useful, non-invasive tool to monitor prolonged stressful conditions. Moreover, the fact that cortisol levels in hair may provide a long-term endocrine profile [67,68] can be extremely useful to study chronic stress and animal welfare [69]. On the other hand, taking into account our results regarding plasma cortisol levels and FCM concentrations, hair might not be suitable at capturing short-term stress levels.

Based on our results, and previous work, increased hair cortisol concentrations in our red deer population seem to be a consequence of hunting activities. However, considering that our study was focused on a wild population, there are other factors that may be contributing to the GC levels obtained in hair, like temperature, food availability or season [17,60,70]. As the energetic balance is a crucial factor in the ability of the animal to respond e ffectively to certain stressors [4], the body condition, a measure of the long-term energetic reserves [26,71], can have an influence on cortisol levels. Therefore, considering this bidirectional interaction, not only can stress a ffect body condition, energetic parameters could also be important in dealing with stress. Our results did not show any correlation between cortisol levels and BMF, which is a measure of physical condition, making the influence of season and/or food availability on cortisol levels unlikely. Moreover, the source of cortisol accumulated in hair is unclear, and some possible explanations have emerged. Keckeis et al. [72] reported a local production of GC in the hair follicles of guinea pigs, however, how this mechanism is modulated is still unknown. Recently, experimental evidence was provided in domestic sheep (*Ovis aries*) that mechanical irritation of the skin significantly increased hair cortisol concentrations [73]. Another study suggested the existence of a cutaneous HPA axis, able to synthesize and secrete cortisol, as well as negative feedback regulation by cortisol under corticotropin-releasing hormone (CRH) expression [74]. The uncertainty about the origin of cortisol in hair leads to an additional caution in the interpretation and analysis of GC levels in these types of samples [75,76]. To decrease the influence of confounding factors in our study, hair samples were taken from the dorsal region in all the individuals. However, further investigation would be very important to clarify whether cortisol concentrations are a ffected by the level of hair pigmentation as well as body area, hair type, or if there is any pattern along the hair shaft as suggested by some studies [18,20,21]. In addition, it is also relevant to emphasize the importance of the combined use of di fferent indicators of stress to obtain more complete and precise information about the stress conditions of wild populations [10]. Plasma, feces and hair are complementary tools, which can provide information from di fferent time-windows, allowing a better evaluation of the e ffects of human activities, like hunting, on the physiological stress response.

In terms of physical condition, our results showed that females were in better physical condition than males. This could be due to di fferential costs of reproduction for each sex, with males going through a phase of hypophagia and high activity levels during the rut [26,37]. Young individuals also had lower BMW indices than adults, which may be the result of a greater investment of these

animals into growth [77]. Despite the observed di fferences in physical condition, and contrary to our predictions, the stress parameters we measured were not associated with physical condition. However, Cabezas et al. [78] reported lower values of body condition in animals with high GC levels in wild rabbits (*Oryctolagus cuniculus*). The absence of an association between stress levels and physical condition may indicate that the studied red deer population had enough fat reserves to cope with the stress induced by hunting activities.
