4.1. Behavioral performance
At the present stage, the study of the aggressiveness of fish mainly adopts mirror image stimulation tests (Falsarella et al. 2017; Gerlai et al. 2000; Liu et al. 2020; Marks et al. 2005; Moretz et al. 2007a; Moretz et al. 2007b) and real opponent stimulation tests (Aguiar and Giaquinto 2018; Boscolo et al. 2018; Brandão et al. 2018; Fontana et al. 2016; Zhang et al. 2020; Zhang et al. 2021). The mirror test is a standardized experimental method that can stimulate the same aggressive behavior in the opponent fish and the target fish. However, some studies have pointed out that hormone levels, behavioral responses, and physiological changes caused by the aggressive response stimulated by the mirror test are different from those caused by real opponent stimulation (Hirschenhauser et al. 2008; Oliveira et al. 2016). Compared with the mirror image test, the real opponent stimulation test induces a more complete expression of aggressive behavior in the opponent fish and the target fish, and the behavior sequence is more complex, which can trigger a behavioral response that is closest to reality and form a typical aggressive phenotypic difference individuals (Hubená et al. 2020). Some studies reported that the real opponent stimulation test could accurately divide aggressiveness after about 5 min (Ariyomo et al. 2013; Teles and Oliveira 2016). Therefore, it is more realistic and accurate to adopt stimulating experiments with real opponents (Teles and Oliveira 2016).
Three related behavioral factors were selected from the 10 parameters related to aggressive behavior, namely, the individual's position, motion state, and physical status. All three factors can represent the significant differences between high aggression and low aggression individuals and can also be used as typical behavioral parameters for rapidly screening individuals with diverse aggression (Colléter and Brown 2011; Oliveira et al. 2011; Quadros et al. 2018). From the perspective of individual motion states, there was no significant difference between the attack speed of high aggression and escape speed of low aggression, which may be due to the small experimental area, leaving limited space for individuals to attack or escape (Ma et al. 2021). The distribution position results determined that aggression was significantly associated with fish personality. Individuals with high aggression usually adopt active coping strategies (approach-attack), while individuals with low aggression usually display passive coping strategies (defense-escape) (Oliveira et al. 2011). The results revealed that the latency to the first movement of high aggression individuals is shorter and closer to the center of the experimental system. In contrast, the latency to the first movement of low aggression individuals is longer, closer to the edge of the experimental system, and with a longer residence time. This finding is consistent with that of Colléter (Colléter and Brown 2011). Assuming that there is no shelter, the moving tendency and position in an open water tank can represent the individual's boldness (Dahlbom et al. 2011). Individuals with a strong tendency to move and who are often located in the central area are typically considered bold ones, while individuals with a weak mobility tendency and often located in the marginal area are considered shy (Oliveira et al. 2011). Therefore, individuals with high aggression are often bolder than individuals with low aggression. In terms of individual physical state, it is interesting to note that the activity of individuals with high aggression is lower than individuals with low aggression. This finding is contrary to most previous studies (Quadros et al. 2018; Yuan et al. 2018). The reasons may be as follows: First, the method used to measure fish activity in this study differs from the general approach, which is measured separately according to the area that fish passed through per unit time. The activity in this experiment was measured in real-time according to the percentage change in the number of pixels when the object moved (Hubená et al. 2020; Whittaker et al. 2021). Obviously, the change in measurement, based on the moving object itself, is more reasonable and accurate. Secondly, during short-term interactions, individuals with low aggression may be more inclined to defend and retreat due to limited space for hiding or escaping (Ma et al. 2021; Paull et al. 2010). This results in more struggling actions, such as swinging the tail in the edge area of the test system, leading to a higher rate in pixel change. Thirdly, studies have established that individuals with high aggression are less flexible to daily changes than individuals with low aggression (Vindas et al. 2017a), suggesting that individuals with high aggression may be less active.
Aggressive behavior can be divided into proactive and reactive aggression based on specific behavioral and physiological characteristics (Wrangham 2018; Zhu et al. 2019). Proactive aggression is triggered by scheduled motivations to achieve personal goals or obtain personal gains (Fanning et al. 2020; Ibabe 2020). Therefore, proactive aggression is often scheduled premeditated, and is associated with lower emotional responsiveness. Conversely, reactive aggression is driven by provocation and/or perceived threat and is usually impulsive and unplanned. Moreover, it is usually linked to higher emotional responsiveness (Dorfman et al. 2013; Fanning et al. 2020; Zhu et al. 2019). The present results showed that the high-frequency aggression in individuals with high aggression was manifested mostly as time concentration, short persistent time, and high repeatability. This result is impressive, as it indicates that the aggressive behavior of individuals with high aggression may actually belong to reactive aggression (Cervantes and Delville 2007).
Herein, differences between different types of aggression were not taken into account. The rationale is that the differences are visible by the naked eye, regardless of aggression frequency and behavioral performance. Individuals with low aggression showed obvious escape and freezing behavior, while high aggression individuals mainly exhibited biting and chasing behavior (Oliveira et al. 2011). In addition, aggression tests that involve three or more interacting individuals were not addressed in the real opponent stimulation test. Earlier studies indicated that different experimental methods lead to differences in behavioral performances between high and low aggression individuals (Oliveira et al. 2005; Way et al. 2015). In fact, fishes form a specific and self-organized structure during group formation. Therefore, the aggressive behavior of fish in a group context is more reflecting of the real aggression situation (Xu et al. 2020). Regrettably, this is challenging to achieve in practice at this stage.
4.2 Physiological indicators
Variations in cortisol levels have been widely used as a stress indicator in fishes (Moltesen et al. 2016). The cortisol responses suggested that the individuals exposed to an aggressive state (high and low) were under significant stress (higher cortisol levels at baseline), which was in agreement with Bessa's finding (Bessa et al. 2021). It is somewhat surprising that no significant difference in cortisol was found between the two aggressive groups. This indicated that the cortical responses evoked by short-duration aggression were similar, regardless of individuals with high or low aggression. This finding does not support the previous research (Sherman and Mehta 2019; Sloman et al. 2001). Studies have shown that the decrease in cortisol levels after stress is more important than the cortisol concentration during stress (Koolhaas et al. 2011), that is, individuals with different aggression may have similar cortisol concentrations at the beginning of the aggression. Another possible reason might be that cortisol was persistently elevated after 10 minutes of aggression (Moltesen et al. 2016).
In teleost fishes, differences in brain monoamine activity have been consistently and reliably utilized to reflect the motivation and outcome of fish aggression (Backström and Winberg 2017; Clotfelter et al. 2007). Considering that monoamines and their metabolites require a certain period of time to accumulate, the 5-HIAA/5-HT ratio may be a more reliable indicator of neural activity for a short duration (Øverli et al. 1999). This can be used to justify the lack of significant difference in the 5-HT content, but the 5-HIAA/5-HT ratio was significantly different in individuals with various aggressive behavior. This finding was also reported by Loveland et al. (Loveland et al. 2014). During aggression, the brain's 5-HT-ergic system was activated, and the activity of 5-HT increased rapidly in low aggression individuals.
In this study, there is a decent correlation between aggressive behavior and physiological indicators. Immobile time and the cortisol level were positively correlated. The longer the immobile time, the higher the cortisol level. A longer immobile time implies that fish have a positive stress coping mechanism (Vindas et al. 2017b). The 5-HIAA/5-HT ratio was also positively correlated with the distance to the central point. This is contrary to studies related to the effects of aggressive fish behavior on brain monoaminergic activity (Xu et al. 2020). This result may be explained by the fact that the aggressive behavior indicators in this experiment were more comprehensive and were not limited to the duration and times involved in the mirror test. The distance to the central point is inversely proportional to the aggressive strength, that is, the closer the distance to the central point, the higher the aggressive strength, and the lower the 5-HIAA/5-HT ratio (Colléter and Brown 2011; Dahlbom et al. 2011). Actually, the knowledge on fish behavior and physiologic parameters is still limited. To improve fish welfare and aquaculture practices, future research needs to combine behavioral and physiological parameters to reduce or avoid significant stress during culture.