Social hierarchy effects on stress responses of subordinate specimens in sea bream ( Sparus aurata )

: Social stress can affect the ability of the fish to respond to various stressors, such as pathogens or environmental variations. In this paper, the effects of social stress on gilt-head bream ( Sparus aurata ) were investigated. To study the effects of physiological stress, we evaluated biochemical and cellular parameters as cortisol, glucose, lactate, osmolarity and phagocytosis 24 hours after the establishment of social hierarchy. Social hierarchy was determined and characterised by behavioural observation (aggressive acts and feeding order) of the specimens (dominant “α”, subordinate “β” and “γ”). After the establishment of the social hierarchy, we observed that the levels of plasma cortisol and other biochemical stress markers (glucose and lactate) were higher in subordinate individuals than in dominant ones. In addition, the modulation of phagocytic activity of the peritoneal cavity cells (PEC) demonstrated that social stress appeared to affect the immune response. At last, principal component analysis clearly separated the subordinate fish groups from the dominant groups based on stress markers and phagocytic activity of the peritoneal exudates cells.

3 of 17 [37]. Moreover, contact between conspecific fish does not promote habituation [47]. In fish, chemical and visual cues, are of major importance for inter-individual communication and underlie the communication of stress status between the animals [48]. In fish, social defeat is a consequence of losing a confrontation among animals, and this represents a powerful stressor that can lead to changes in animal behaviour and physiology [37][38][39]44]; in cichilid species, Astatotilapia burtoni and Amatitlania siquia, visual signals of conspecifics contribute to the regulation of social behaviour [49,50]. It is known that the fish can infer social rank by observation alone [51]. In fact, teleost fish often live in an optimal environment for visual signalling [52] and have an excellent visual system with high resolution vision [53].
Physiological consequences of social interaction had been observed in the subordinate fishes of gregary species.
Plasma cortisol level is known to be correlate with the establishment of the hierarchy [54], furthermore, the increases of plasma cortisol level under stress conditions typically causes increases in plasma glucose and lactate levels.
Elevations in plasma glucose are generated initially by catecholamine-mediated glycogenolysis and at later stages, cortisol-mediated gluconeogenesis, and lactate concentrations rise as muscle lactate formed during anaerobiosis is released to the plasma [55][56][57][58]. Also in literature it is reported that when under stress fish also increase the osmolarity values in plasma [29,59].
We have previously demonstred [41], using a paired fish of gilt-head bream (Sparus aurata), that the established hierarchy between the two specimen, determin a change of the principal biochemical parameters (cortisol, glucose and osmolarity) and cellular (phagocytic activity) in subordinate individuals almost immediately after pairing (at 24 hours). The biochemical and cellular stress markers considered were higher in subordinate individuals than in dominant [41]. In this study, we examined the more complex interaction scheme established among three individuals 4 of 17 have demonstred that the social hierarchy was established after 1 hour of cohabitation and remained unchanged after 24 hours of cohabiting. In addition, the onset of hierarchy induced a stress in gilt-head bream, shown by changes in biochemical and cellular parameters of the stress response (cortisol, glucose, lactate) and immunity (phagocytic activity of PEC).

Animals
Twenty-seven specimens (125 ± 25 g body weight) of the seawater teleost gilt-head seabream were obtained from a commercial fish farm (Ittica San Giorgio, Licata, Sicily). After an acclimatising period of one week, the animals were subdivided into the following groups: fishes were placed in tanks by group of three forming three experimental groups. The sampling and analysis were done for each group at 24 h after hierarchical establishment. The experiment was performed in triplicates, and the hierarchy have been clear distinguished after one hour for each experiment.

Experimental conditions and behaviour observation
The aquaria seawater was monitored daily and maintained at an average temperature of 18 ± 1 °C, at a salinity of 38 ± 1 %, oxygen and nitrite concentration respectively of >6 mg L -1 and <0.2 mg L -1 under a photoperiod of 12 h dark and 12 h light. The fish were fed with commercial pellet diet (Trouvit of the Hendrix SpA) once a day ad libitum. To be able to identify the three individuals in an objective manner the fish have been marked as follows: a fish has not been marked, to another have practiced a cut at the level of the dorsal fin and the other at the level of the caudal fin [60].
The behavioural changes were recorded to assign a hierarchical position to each individual; all the three experimental groups were observed to detect changes in the social status until the social positions were established, every group was observed for 24 h using a digital camera and a digital multifunctional system for the data acquisition (DR41). The hierarchy have been maintained for all the observation time. The individual behaviour was examined by the 5 of 17 continuous check of the different behaviour categories [61], and individuals from each group were distinguished as dominant α, subordinate β and subordinate γ. High social status has been correlated with increased aggressiveness and preferential access to the food [62,63]. To define this social distinction, the number of aggressive acts (A+) were observed and defined as a bite or a rapid approach without biting that resulted in the displacement of the subordinate [64], and the feeding order (FO) of each group was determined. according to McCarthy et al. [63].

Blood sampling and peritoneal cell preparation
After 24h of cohabitation, the fishes were anaesthetised with 0.05 % w/v MS222 (3-aminobenzoicacid ethyl ester, Sigma-Aldrich, Italy) in seawater; blood samples were collected via caudal venepuncture into heparin-coated syringes (2500 IU mL −1 heparin sodium salt, Sigma-Aldrich, Italy) and centrifuged (10,000 g for 2 min). Plasma was extracted and stored at −80 °C for later analysis of cortisol, glucose, lactate and osmolarity levels. The peritoneal exudates cells (PECs) were obtained as follows: the fish were anesthetized and after disinfection of the body ventral surface with 70 % ethyl alcohol, in the peritoneal cavity was injected with 15 ml of isotonic (370 mOsm kg -1 ) medium (Leibovitz L15 medium containing 2 % foetal calf serum, 100 units penicillin ml -1 , 100 units streptomycin ml -1 and 10 units heparin ml -1 , Sigma, Italy). After massaging the ventral surface for 10 min, the medium containing the PECs was collected, and the PECs were isolated by centrifugation at 400 g for 10 min at 4 °C. The dead cells were determined by light microscopy after addition of 0.01 % trypan blue to the medium.

Haematological parameters
The concentrations of total cortisol were measured in the plasma sample using a commercially available kit (Intermedical Diagnostics srl, Italy) according to the manufacturer's instructions and confirmed by radioimmunoassay (RIA) [65]. The glucose and lactate plasma levels were determined using the Accutrend Plus Kit (Roche, Italy) according to the manufacturer's instructions. The osmolarity of the plasma samples was measured using a freezing-point depression osmometer type 4b (Roebling, United Kingdom).

Statistical analyses
All experiments were conducted in triplicate. The data are expressed as mean ± standard error (S.E.M.). Data were statistically analysed by on e-way analysis of variance (ANOVA) to determine difference between groups (i.e., dominant α, subordinate β and subordinate γ). Normality of the data was previously assessed using a Shapiro-Wilk test and homogeneity of variance was also verified using a Levene test. Non-normally distributed data were log-transformed prior to analysis and a non-parametric Krustal-Wallis test followed by a multiple comparison test was used when data did not meet parametric assumptions. Statistical analyses were conducted using "Statistica" software (Statsoft Europe, Germany) and a probability level of p <0.05 was considered significant. In addition, to examine the interrelation between two the sets of variables (stress physiological markers and phagocytic activity of the PEC), principal component&correlation analisys has been performed for a multiple group of principal component analysis.

of 17
The experimental design involved the observation of three experimental groups of three specimens of gilt-head bream in tanks in which the three fish were placed simultaneously in the experimental tank. As shown in Table 1, the percent of aggressive acts for each hour (A+) and preferential access to the food (FO) distinguished (p < 0.001) the fish as either dominant α or β and γ subordinates in each group. As shown in Figure 1 between dominant α, and subordinates β and γ were observed ( Fig. 1(a)). The plasma glucose levels were higher in subordinate γ individuals compared to the subordinate β, and dominant α fish ( Fig.1(b)). Also the lactate concentrations were higher in the γ fish while in the subordinate β fish the plasma lactate values remained roughly unchanged compared with the dominant α as shown in Figure 1(c). The plasma osmolarity values in the γ subordinates were higher than that of the dominant α fish whereas in β subordinates, the plasma osmolarity value at

In vitro phagocytosis assay
In Figure 1   Subordinate gamma (Sub-γ). Cortisol, glucose and lactate contribute explaining the variability of the data with the 92.07% along the X axis between the subordinate and dominant fish groups during the experimental time. In the Y-axis, osmolarity and phagocytosis contribute contribute explaining the variability of the data with the 6,44%.

Discussion
Previously, we have demonstrated the establishment of the social hierarchy in pairs of gilt-head bream showing a physiological change in the subordinate individuals. Therefore, the main objective of this paper was to verify if it was possible to study establish social hierarchy in groups of three fish of gilt-head bream and subsequently to evaluate the physiological effects of social stress on individuals.
In this study, fish are moved at the same time to the tank, after one hour hierarchy was established and at 24 h after, significant increases in blood plasma levels of cortisol, glucose, lactate and osmolarity were observed in the subordinate fish (β and γ). In particular, cortisol, glucose, lactate and osmolarity values were significantly higher in subordinate fish γ compared to fish β. This data is consistent with previous works in rainbow trout [41,47,54], where the agonistic interactions between conspecifics constituted a chronic social stress, inducing increase of plasmatic cortisol concentration in subordinate fish. Pottinger and Carrick [66] reported, using the classical paired model, that fish position within a tank, locomotor activity, agonistic behaviour, feeding, and plasma cortisol levels are useful criteria for the determination of social dominance in the rainbow trout.
In gilt-head bream, the cortisol values are correlated with the social status, values being higher in subordinates than those the dominants [67,68]. The glucose levels measured in this study were increased in subordinate fish. Glucose release in blood is generally associated with the secondary stress response being modulated by the action of cortisol that influences glucidic metabolism in fish [8,56].
Mommsen et al. [56] reported that plasmatic, hepatic and muscular glucose levels in teleosts might not be univocally correlated with the stress condition (i.e. cortisol level). Also, in this paper we have observed an increase in lactate levels in the blood of subordinate individuals compared to dominant ones. Increase of blood lactate level is generally reported as a secondary response to stress [69,70].
It is known that stress-induced hormonal responses, lead to osmotic imbalances in fish [71,72]. Thus, stress causes elevation of plasma cortisol and electrolyte loss in freshwater fish [73][74][75]. In agreement with previous studies, in this work we also have observed a significant increase in the levels of osmolarity in individuals subordinate compared to dominant α. Also, we have evaluated the effects of social stress on the phagocytic activity of peritoneal cavity cells. . This receptor was localised to the head kidney, spleen, gills, intestine, heart and liver tissues [30,76], highlighting the crucial role of cortisol in the regulation of homeostasis.
To establish the relationship between haematological parameters (cortisol, glucose, lactate and osmolarity) and immune response (phagocytic activity), a statistical evaluation was performed using a principal component analysis of the data from all dominant α, and subordinate β and γ fish. All the groups are clearly separated supporting the relation between physiology and social stress condition. In particular, as we can observe in Fig.3 along the X axis, cortisol, lactate and glucose values, and along the Y axis osmolarity and phagocytosis coontribute explaining the variability of the data and the distribution the experimental fish groups. These differences between the groups could be attributed to different allostatic load and adaptation time in the responses, indicating that these could be used as allostatic load biomarkers of social stress responses and impact on fish health. Our results show that stress can also be determined from social interactions and from the territorial disputes activating the stress response through cortisol release in gilt-head bream, as occurs in response to other stressors and in vertebrates [77].
Interestingly, the boldness is already observed to have an effect on the growth of sea bream at different density [78] and these results may be correlate directly with the behaviour of dominant sea bream to eat first and more, here demonstrated. The feeding behaviour, moreover, seems to have an effect on the stress response. Indeed, Gesto et al. [79] found in the rainbow trout with a different ability to compete for food showed a different behavioral responses to hypoxia and ammonia. The behavioural indicators of boldness (e.g. hierarchy order) in sea bream could be a consistent proxy of the physiological state [80].

Conclusions
In conclusion, in this study, the hierarchic relationship in gilt-head bream was elucidated through behavioural and serological indicators given the first insights on the time of establishment and physiological traits of dominant and subordinate fish for the first time in a group of three fish. [78]. Thus, the integration between physiological indicators, and experimental behavioral including the hierarchy establishment, could help to elucidate physiological state both in wild and captivity environment. Also, we have observed that individuals who access food first were those who became dominant and thus they had the advantage when compared with other individuals and were winners throughout the experimental period. Social defeat or win might be a stressor that can lead strong effects in behaviour and fish physiology [49].
subordinates. This effect is, however, modulated by many factors such as group size, habitat temperature, fish size sexual maturity. All this factors, in larger groups, contribute to the complexity with which social hierarchies can elicit stress. Indeed, individuals in large groups may face more intense food competition [82], cannibalism [83], susceptibility to capture [84] and sex competition [85], for example.
Further research is needed to study the hierarchic relationship in larger and sex-mixed groups, also in light of the sea bream shoaling behaviour.