In order to elucidate the relationships among behaviors (boldness, aggressiveness, activity, and respiration) and physiology (metabolism and hormone) in the context of fish personality, three experiments were conducted using a typical rockfish species S. schlegelii during a period of 3 years. Generally, the first experiment aimed to explore the relationships among boldness, aggressiveness, basal locomotor activity, and basal cortisol level. In this experiment, fish boldness was determined by a “box-emerging test.” The second experiment aimed to test the relationships among boldness, stressed locomotor activity, stressed opercular beat rate, and standard metabolic rate. In this experiment, fish boldness was measured based on an “open-field test.” The third experiment was conducted to examine the validation of the two behavioral apparatuses used for determining fish boldness in this study. Through these three experiments, we could broadly construct a behavioral-physiological spectrum for black rockfish.

The first experiment was conducted from October 2018 to December 2018. Black rockfish juveniles were obtained from a hatchery in Qingdao, Shandong, China. Fifty-three fish (mean body length 7.53 ± 0.11 cm, mean body weight 9.75 ± 0.27 g) were individually acclimated to the experimental condition (i.e., glass tank, 40 cm × 30 cm × 30 cm) for 2 weeks before the formal experiment. The formal experiment included four parts according to priority (Figure 1A): (I) observing fish basal locomotor activity under a normal condition, (II) evaluating fish aggressiveness using the “mirror-image test,” (III) evaluating fish boldness using the “box-emerging test,” and (IV) sampling fish to measure basal cortisol concentration.

Based on our prior experience and careful observations, the 2-week acclimation period was sufficient for fish to resume normal behaviors and physiology (Zhang et al., 2021). Therefore, the fish locomotor activity in glass tanks during the formal experiment could be assumed to be under a normal condition, and we called this parameter basal locomotor activity. In detail, a camera (HDR-AS100V, Sony Corporation, Tokyo, Japan) was placed in front of each glass tank, and the videos were recorded from 8:00 a.m. to 10:00 a.m., during which no human interference occurred (Figure 1A). Each tank was filmed for 10 min, and then only the middle 5 min of each video was used to determine the basal locomotor activity to completely eliminate the possible disturbance from experimenters (Zhang et al., 2020a, b, 2021). The basal locomotor activity in this study was defined as time spent moving in five minutes, which has been proven to accurately and effectively reflect the basal locomotor activity of fish (Øverli et al., 2002a, b, 2006).

After observing fish basal locomotor activity, a mirror-image test, which has been verified to be valid in several fish species (Woodward et al., 2019; Da Silva-Pinto et al., 2020; Thore et al., 2020; Xu et al., 2021), was adopted to determine fish aggressiveness. In brief, a mirror (27 cm × 30 cm) was gently placed perpendicular to the bottom of the tank on a narrow side of each glass tank (Figure 1A). Videos were recorded for 12 min from the moment the mirror was introduced, and then the middle 10 min of each video was selected to measure the aggressive behavior of the fish. Fish aggressiveness was estimated by counting the number of instances of chasing, nipping, and biting of the target fish for the mirror (Woodward et al., 2019; Zhang et al., 2020b; Xu et al., 2021).

We used the “box-emerging test” to evaluate fish boldness, and the behavioral apparatus and experimental procedure followed the descriptions of previous studies (Brown and Braithwaite, 2004; Brown et al., 2005, 2007; Roberts et al., 2011; Zhang et al., 2019). In short, the apparatus (Figure 1A) mainly consisted of an opaque start box (10 cm × 10 cm × 20 cm) that was placed in one side of a plastic tank (50 cm × 30 cm × 30 cm). The start box was equipped with a vertically sliding trapdoor that could be opened and closed via a fine monofilament, and the top lid of the start box could be removed if required. When starting the trial, a fish was gently transferred from a glass tank to the start box using a hand-netting. After a 2 min acclimatization period, the trapdoor was slowly lifted, and the fish was free to emerge. The time taken by the fish to completely emerge from the start box was recorded. If the fish did not leave the shelter after 10 min, we removed the lid of the box. If the fish still did not leave the shelter after 15 min, we terminated the trial, and they were allocated the maximum time score of 900 s. In this test, the more time duration the fish needed to emerge from the start box, the lower boldness the fish had. One fish was tested at a time, and the seawater in the apparatus was refreshed after each trial.

To measure the basal cortisol levels of fish, all of the fish that had completed the box-emerging test were put back in the original glass tanks for 1 week. This further acclimation period allowed fish to recover to their normal behavioral and physiological status from possible handling stress. Then these fish were euthanized with an overdose of anesthetic (Tricaine methanesulfonate, MS-222, 100 mg/L), and the body weight (precision 0.01 mg) and body length (precision 0.01 mm) were recorded. Subsequently, fish were rapidly dissected, and the whole visceral masses (this was performed because the juveniles were too small to draw blood; Figure 1A) (Guo et al., 2017; Zhang et al., 2020a) were removed and frozen in liquid nitrogen and then stored at −80∘C until analyzing the cortisol concentrations. All sampling procedures were completed within 1 min.

The visceral mass samples were homogenized with normal saline, diluted 10 times, and centrifuged at 4∘C in a refrigerated centrifuge for 20 min at 3000 rpm. The supernatants were collected and used for analysis. The cortisol concentrations were measured using commercial ELISA Assay Kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China), according to the manufacturer’s guidelines. The ELISA assay kit has been validated previously (Zhang et al., 2020a, 2021).

The second experiment was conducted from September 2019 to October 2019. Black rockfish juveniles were obtained from a hatchery in Wendeng, Shandong, China. Twenty-nine fish (mean body length 8.05 ± 0.19 cm, mean body weight 10.41 ± 0.34 g) were individually acclimated to the experimental condition (same to experiment I) for 2 weeks before the formal experiment. The formal experiment consisted of three parts (Figure 1B): (I) evaluating fish boldness using the “open-field test,” (II) observing fish stressed locomotor activity and opercular beat rate under a stressed condition (see below for further explanation), and (III) measuring the standard metabolic rate of fish.

Based on the observations of experiment I, we found that the fish were more willing to hide in the starting box rather than emerge from the shelter during the given testing period, and this phenomenon may lower the discrimination degree for boldness among the experimental fish. Therefore, in experiment II, another classical method, open-field test (Archard et al., 2012; Bergendahl et al., 2016; Salvanes, 2017; Thore et al., 2020), was adopted to evaluate fish boldness. The behavioral apparatus mainly included a quadrate tank (30 cm × 30 cm × 30 cm), which had a circle (diameter 10 cm) in the center of the tank bottom (Figure 1B). Each fish was taken out of the glass tank with a hand-netting and placed into the center circle of the test tank. A camera was fixed above the tank. Then a 10-min video was recorded from the time point when the fish were put in. The fish boldness was estimated by the time duration that fish head was in the center circle during the 10-min video. The more time duration the fish head was in the center circle, the higher boldness the fish had.

Based on the videos filmed in the open-field test, we also determined fish stressed locomotor activity and opercular beat rate. Notably, when conducting the open-field test, the fish were in a state of stress, given that each fish was transferred from a home tank to a relatively unfamiliar experimental tank using a hand-netting. Therefore, we defined the relevant behaviors as stressed behaviors (i.e., stressed locomotor activity and stressed opercular beat rate). Similarly, only the middle 5 min of each 10-min video was analyzed to determine the interested behaviors. The stressed locomotor activity was defined as time spent moving in 5 min, which was the same as experiment I. The opercular beat rate was measured as the number of opercular beats per minute (Braithwaite and Salvanes, 2005; Zhang et al., 2020a).

After completing the open-field test, all fish were recovered to their home tank and were fasted for 1 day. Then, each fish was introduced into a continuous-flow respirometer (Figure 1B) to measure the standard metabolic rate. The experimental apparatus and procedure followed the classical method, which was repeatedly evidenced to be valid in several studies (Xie and Sun, 1990; Fu et al., 2005). In brief, the apparatus consisted of 11 parallel respiratory chambers, one of which did not contain fish and was used as a control. The whole apparatus was covered by opaque sheets to completely eliminate the disturbances from the surrounding environment. An oxygen meter (S9-Field kit, METTLER TOLEDO, China) was used to measure the oxygen consumption rate of the experimental fish. The fish were introduced into the respiratory chambers at 20:00 on the last day and were given one night to recover from the possible handling stress. Then, the dissolved oxygen contents of water outlets were measured at 10:00, 12:00, 14:00, 16:00, and 20:00 on the next day, and the average value of the five measurements was used to calculate the standard metabolic rate of the experimental fish. The calculation formula of the standard metabolic rate (SMR) was SMR = △O₂⋅V/m. In the formula, SMR [mg/(kg⋅h)] was the standard metabolic rate, △O₂ (mg/L) was the difference between the dissolved oxygen content in the respiration chamber and the control respiration chamber, V (L/h) was the flow velocity in the respiration chamber, which was obtained by measuring the flow volume at the outlet of the respiration chamber for 1 min, and m (kg) was the weight of the experimental fish.

The third experiment was conducted in September 2020. Black rockfish juveniles were obtained from a hatchery in Wendeng, Shandong, China. Fifty-four fish (mean body length 9.10 ± 0.61 cm, mean body weight 12.46 ± 1.86 g) were individually acclimated to the experimental condition (same as experiment I) for 2 weeks before the formal experiment. In the formal experiment, the boldness of each fish was first evaluated by the box-emerging test (Figure 1C) and then recovered to their home tank for 1 day to eliminate the possible handling stress. After this acclimation period, the boldness of the same fish was subsequently evaluated by the open-field test (Figure 1C). The detailed procedures were the same as experiments I and II.

First, we classified each fish into the “bold fish” category or “shy fish” category using the hierarchical clustering analysis (Castanheira et al., 2013), based on the boldness value that was determined by box-emerging test in experiment I or open-field test in experiment II. The principle of the algorithm is: at the beginning, all the data points themselves are used as clusters, and then a specific algorithm is used to find the two closest clusters and merge them into one, and the above steps are repeated until the preset number of clusters is reached (Murtagh and Contreras, 2011a, b). Then, an independent-samples t-test was used to examine the difference of above-mentioned behavioral and physiological parameters (i.e., aggressiveness, locomotor activity, opercular beat rate, standard metabolic rate, and cortisol level) between the generated clusters (i.e., “bold fish” vs. “shy fish”). To further clarify the relationships among the boldness, aggressiveness, activity, respiration, metabolism, and cortisol, linear fitting analysis and Pearson’s correlation test were performed to pooled data. The same analytical method was also applied to clarify the relationships between the boldness values determined by the two boldness test apparatuses (experiment III). Normality of the data was tested with Kolmogorov–Smirnov’s test, and the homogeneity of variance in the data was tested using Levene’s test prior to independent-samples t-test. When data failed to meet independent-samples t-test assumptions (i.e., normality and homogeneity), transformations were applied [i.e., logarithmic transformation log10(1+x) for number of aggressive behavior, time spent moving (experiment I:box-emerging test)]. Statistical analyses were performed using SPSS 22.0 for windows. Differences were considered significant at a probability level of 0.05 (P < 0.05). The results were expressed as mean ± SE.

We generated two groups (bold fish, n = 9; shy fish, n = 44) using the hierarchical cluster analysis based on the latency to emerge from the start box in experiment I. Bold fish showed significantly more aggressive behavior compared to shy fish (t₅₁ = 3.358, P = 0.001; Figure 2A). However, there were no significant differences in basal locomotor activity (t₅₁ = 1.344, P = 0.185; Figure 3A) and basal cortisol levels (t₅₁ = 0.546, P = 0.587; Figure 4A) between the two clusters. Fitting analysis showed that the emerging latency was significantly, negatively correlated with the number of aggressive behavior (Figure 2B). However, Pearson’s correlations between emerging latency and time spent moving under the normal condition (Figure 3B) or between emerging latency and basal cortisol level (Figure 4B) were not statistically significant (P > 0.05).

Similarly, we first generated two groups (bold fish, n = 9; shy fish, n = 20) using the hierarchical cluster analysis based on the time duration in the center circle in experiment II. Bold fish had significantly higher standard metabolic rate (t₂₇ = 2.96, P = 0.006; Figure 5A) and stressed locomotor activity (t₂₇ = 2.222, P = 0.035; Figure 6A), while had significantly lower stressed opercular beat rate (t₂₇ = −2.078, P = 0.047; Figure 7A) compared to shy fish. Fitting analysis showed that the duration in center circle was significantly, positively correlated with standard metabolic rate (Figure 5B), and positively correlated with time spent moving under the stressed condition (Figure 6B), but negatively correlated with stressed opercular beat rate (Figure 7B).

A significantly negative correlation between the duration in center circle in the open-field test and the emerging latency in the box-emerging test was detected (r = −0.275, P = 0.044, y = −4.650 x + 804.3; Figure 8A), which indicated that both boldness determination apparatuses were sensitive and valid to estimate the boldness of black rockfish.

Since we controlled the size of the experimental fish within a small gap, we did not find a significant relationship between the size and boldness of the experimental fish in the two boldness experiments (P > 0.05; Figures 8B,C).