Agonistic behaviour and energy metabolism of bold and shy swimming crabs Portunus trituberculatus

Similarly to other phenotypic characteristics, behavioural traits show considerable inter-individual variation. However, increasing evidence indicates that individual behaviour can be consistent over time and across contextually different situations (Barber and Dingemanse, 2010). These consistent and repeatable behaviours are often referred to as animal personality traits (Sih et al., 2004; Dingemanse et al., 2010; Budaev et al., 2015). Researchers have developed a generalized framework for testing animal personality traits, with six defined behavioural axes: boldness, exploration, activity, aggressiveness, sociability and hesitancy (Brown et al., 2005; Reale et al., 2007). Both empirical and theoretical studies have suggested that personality has profound effects on individual fitness, growth, reproduction, metabolism and survival (Smith and Blumstein, 2008; Dingemanse and Wolf, 2010). These findings have been confirmed in many aquatic animals, such as echinoderms (Ru et al., 2017), fish (Adriaenssens and Johnsson, 2011; Liu and Fu, 2017) and amphibians (Kelleher et al., 2018). However, to date, there are relatively few studies in crustaceans, and most studies have focused on a few species, for instance, the hermit crab Pagurus bernhardus (Mowles et al., 2012), the Chinese mitten crab, Eriocheir sinensis (Brodin and Drotz, 2014), and the mud crab Panopeus herbstii (Belgrad et al., 2017). Moreover, the correlation and plasticity of personality traits has attracted relatively more attention, whereas information on the relationship between individual physiological conditions and personality remains limited.

The shy–bold continuum is an important dimension of personality variation (Carere and Oers, 2004). Animals can be classified into bold and shy individuals on the basis of their behavioural output (Ariyomo and Watt, 2012; Sneddon, 2003). Numerous studies have examined behavioural differences between bold and shy individuals. For instance, in the lizard Lacerta monticola, bold individuals exhibit greater exploration behaviour than shy individuals (López et al., 2005), and shy three-spine sticklebacks (Gasterosteus aculeatus) have a stronger social relationship than bold individuals (Pike et al., 2008). Previous studies have also shown that aggressiveness is correlated with boldness (Biro and Stamps, 2008; Rudin and Briffa, 2012), as widely confirmed in fish (Rupia et al., 2016). In crustaceans, Courtene-Jones and Briffa (2014) have found that shy P. bernhardus hermit crabs are more likely to win in contests to defend their inhabited shells.

Energy is essential for the physiological processes that generate behaviour (e.g. muscle contraction) (Biro and Stamps, 2010), and the variations in individuals' personality traits essentially reflect differences in metabolism (Nespolo and Franco, 2007). For example, the concentrations of blood glucose and lactate in the carp Cyprinus carpio are lower in bold than in shy individuals (Huntingford et al., 2010); and in the great tit, Parus major, bold individuals have higher body temperature and breath rate than shy individuals (Carere and Oers, 2004). Recent studies have shown that both personality traits and fight performance in animals are related to physiological state (Sneddon et al., 1998; Careau and Garland, 2012). In crustaceans, fighting has been found to increase glucose levels in the haemolymph of the hermit crab P. bernhardus (Briffa and Elwood, 2002). Lactate is a by-product of anaerobic respiration, which quickly depletes energy reserves. It is also an adequate index of metabolic cost during fighting of animals. Accordingly, lactate has been used to estimate the energetic consequences of activities such as fighting in the cichlid fish Tilapia zillii (Neat et al., 1998) and the shore crab, Carcinus maenas (Sneddon et al., 1998). However, further research on the relationships among personality, agonistic behaviour and energy metabolism in crustaceans is still needed.

The swimming crab Portunus trituberculatus, which is naturally distributed in the coastal waters of Asia-Pacific countries, is a suitable species for studying the agonistic behaviour of crustaceans. To explore the relationships among behavioural type, agonistic behaviour and energy metabolism, we established a system to study personality and agonistic behaviour in laboratory conditions; documented the variations in, and correlations among, the personality traits boldness, activity and hesitancy; and quantified agonistic behaviour and energy status (glycogen, lactate and glucose). Our results provide information that should aid in our understanding of crab personality traits and provide a reference for further research on the relationships among personality, agonistic behaviour and energy metabolism in crustaceans. Moreover, these results also have implications for the breeding of crustaceans.

The experiment was conducted in August 2017 at the Key Laboratory of Mariculture, Ministry of Education, Ocean University of China, Qingdao, China. Swimming crabs P. trituberculatus (Miers 1876) were collected from the aquaculture facility in Jiaonan, and were housed in individual aquaria (40.5 l, 45×30×30 cm) with filtered seawater at 24±1°C and salinity 30, and were allowed to acclimate for 2 weeks before the experiments. The photoperiod was 12 h:12 h light:dark. The crabs were fed ad libitum at 08:00 h every morning with Manila clams (Ruditapes philippinarum) purchased from the local seafood market, and the seawater was exchanged once daily. The aquaria were continuously aerated.

A 141 l aquarium (diameter=60 cm, height=50 cm) was used as the experimental setting to measure personality behaviours. The video capture system used to record personality behaviours (Fig. 1) consisted of a camera (Hikvision, DS-2CD864, China, infrared wavelength=850 nm), a monitor (Philips, 233i, China) and a shelter (length×width×height: 30×20×20 cm) with a trapdoor (length×width: 20×20 cm) facing towards the aquarium centre. The infrared cameras were fixed 0.7 m above the experimental aquarium, and the filtered seawater (24±1°C and salinity 30) depth was maintained at 40 cm during the experiment. The photoperiod was 12 h:12 h light:dark. We turned on the light at 08:00 h and turned off the light at 20:00 h. In this experiment, eight sets of observation systems were constructed for collecting personality behaviour data, and the experiments were conducted in a quiet, undisturbed room.

Only healthy male crabs (carapace width=105.52±6.52 mm, mean±s.d., n=88) in the intermoult stage with all their appendages were chosen for the experiments. At the beginning of the experiment, the crabs were placed into the shelters, and then the trapdoors were closed. The crabs were allowed to acclimate for 10 min in the shelter, and then the trapdoor was remotely opened, thus allowing the crabs to access the experimental arena while the infrared cameras recorded their behaviours. We placed one swimming crab in each shelter. The trials were started at 08:00 h and were terminated after 24 h. After the experiment, the crabs were returned to the holding aquaria. The behaviour data were stored on a video recorder (Hikvision, DS-7604N) for further analysis.

Boldness was estimated as the proportion of time in which crabs were not hidden in the shelter (h h⁻¹) (Belgrad et al., 2017); a larger proportion of time indicates greater boldness. Activity was estimated as the number of movements during 10 min after an individual exited the hidden area (Brodin and Drotz, 2014). Thus, if an individual did not leave the shelter, we could not monitor its activity, and the number of movements was zero. Hesitancy was defined as the time during which a crab completely left the hidden area minus the time at which it emerged from the hidden area (Brown et al., 2005). In the few instances in which the crab had not emerged from the shelter after 24 h, no hesitancy was calculated.

A 141-litre aquarium (diameter=60 cm, height=50 cm) was used as the experimental setting to measure agonistic behaviour. The video capture system used to record agonistic behaviour (Fig. 2) consisted of a camera (Hikvision, DS-2CD864, infrared wavelength=850 nm), a monitor (Philips, 233i5) and a separator board (60×50 cm). The filtered seawater conditions were the same as those in the personality trait experiments. Six sets of observation systems were constructed for collecting agonistic behaviour data. In the present study, crabs were classified into bold (proportion of time outside shelter >0.640; n=44) and shy individuals (proportion of time outside shelter <0.64; n=44) according to their boldness output. A bold and a shy individual with similar carapace widths were selected to conduct a fight. The two matched crabs were placed into the aquarium separated by the board and were allowed a settling time of 10 min in continued isolation. To promote fights, the same amount of food extract (clam muscle homogenized in seawater) was slowly injected into the middle of each arena as the separating board was raised while infrared cameras recorded the behaviours. All experiments were conducted during the period 08:00–12:00 h and ended after 1 h. Twenty-two groups of fights were conducted in this experiment. The behaviour data were stored on a video recorder (Hikvision, DS-7604N) for further analysis.

The agonistic behaviour of crabs was investigated in each fight. The fights included a series of discrete agonistic behaviour defined in Table 1 (Sneddon et al., 1997). The initiator of a fight was defined as the first crab to move towards its opponent and make physical contact. The winner of a fight was the crab that elicited repeated retreats from its opponent or successfully climbed on top of the other contestant (Sneddon et al., 1997). The agonistic behaviour frequency during the fights between bold and shy individuals was counted by video analysis. The initiator and winner of each fight were recorded. We also analysed the number of wins in fights initiated by bold or shy individuals (expressed as initiations/wins), and the number of losses in fights initiated by bold or shy individuals (expressed as initiations/losses).

At each sampling time after a fight had ended, the crabs were removed immediately with minimal disturbance and quickly placed in cold water to produce general anaesthesia (Hajek et al., 2009). Then haemolymph samples were taken by piercing the arthrodial membrane at the base of the last pereiopod with a hypodermic needle attached to a syringe (1 ml). A 2 ml haemolymph sample was collected from each individual. The haemolymph samples were placed in a refrigerator (4°C) for 3 h, then centrifuged for 10 min at 4024.8 g and 4°C; the supernatants were collected and stored at −80°C until the final determination. After haemolymph sampling, claw muscle tissue was then dissected, frozen rapidly in liquid nitrogen and stored at −80°C until later analysis. All procedures were performed on an ice plate. The same method was used to collect haemolymph and claw muscle tissues for testing from the individuals that had not been subjected to the fight experiment.

The concentrations of glucose in the haemolymph (mmol l⁻¹) were measured with the glucose oxidase–peroxidase method according to the protocol of commercial assay kits (Nanjing Jiancheng Bioengineering Institute, China) and final optical density (OD) values were measured at 505 nm. The concentrations of glycogen in the claw muscle (mg g⁻¹) were measured with the anthrone–sulfuric acid colorimetric method according to the protocol of commercial assay kits (Nanjing Jiancheng Bioengineering Institute, China) and final OD values were measured at 620 nm. The concentrations of lactate in the haemolymph (mmol l⁻¹) and in the muscle tissues (mmol g⁻¹ protein) were measured with the NBT colorimetric method according to the protocol of commercial assay kits (Nanjing Jiancheng Bioengineering Institute, China) and final OD values were measured at 530 nm. To measure the concentration of lactate (mmol g⁻¹ protein) in the claw muscle tissues, muscle tissues were weighed and homogenized in nine volumes of cold saline (0.75% NaCl, pH=7.0) for 10 min with a glass homogenizer driven by a motor (Scientz, DY89-II, China). The homogenizer was immersed in an ice-water bath during the process. Homogenates were immediately centrifuged for 10 min at 4°C and 698.75 g. The supernatants were then collected and used for measuring lactate concentration. Total protein in the supernatant was determined with the Coomassie Brilliant Blue G250 dye binding method and final OD values were measured at 595 nm (Bradford, 1976), with bovine serum albumin (Sigma-Aldrich A7030) as a standard. The concentrations of lactate in the claw muscle were normalized by the concentration of protein level and expressed as mmol g⁻¹ protein. All absorbance values were read with an automatic microplate reader (Synergy 2, BioTek, USA).

Data are reported as means±s.d. Statistical analysis was performed using SPSS 22.0 statistical software. Spearman's rank correlation was used to analyse the correlations between personality traits. The frequencies of agonistic behaviour between bold and shy individuals were compared with paired t-tests. The concentrations of glycogen and lactate in claw muscle tissues and the concentrations of glucose and lactate in haemolymph were compared with two-way ANOVAs with fighting and behavioural type as fixed factors. Multiple comparisons of the means were performed with Bonferroni's test. The assumption of homogeneity of variance was tested with Levene's test. When heteroscedasticity was significant, the dependent variable was square-root transformed, and then the assumption of homogeneity was tested. The effects of behavioural type (bold or shy) on contest initiation and outcome were tested with a chi-square test. For all tests, P-values <0.05 were considered to be statistically significant.

All procedures were performed under the Regulations of the Administration of Affairs Concerning Experimental Animals of China, as well as the Regulations of the Administration of Affairs Concerning Experimental Animals of Shandong Province.

Large inter-individual differences were clearly observed in all three personality traits (boldness, activity and hesitancy) of P. trituberculatus. The activity results ranged from 0 to 30, the hesitancy results mainly ranged from 0 to 6 min, and the boldness results ranged from 0 to 1. Correlation analysis indicated significant correlations between any two of the three behavioural variables. Boldness was positively correlated with activity (r_s=0.652, n=88, P=0.00; Fig. 3A), activity was negatively correlated with hesitancy (r_s=−0.680, n=88, P=0.00; Fig. 3B), and boldness was negatively correlated with hesitancy (r_s=−0.604, n=88, P=0.02; Fig. 3C). Therefore, in this experiment, the bold individuals were more active and less hesitant than shy individuals.

The number of fights initiated by bold individuals was significantly higher than that initiated by shy individuals (χ²=4.545, P=0.033), and the number of wins for bold individuals was significantly higher than that for shy individuals as the initiators (χ²=4.765, P=0.029); however, the number of losses was not significant (χ²=0.200, P=0.655; Table 2).

Inter-individual variations in fight performance were linked to the behavioural type of the crabs. Bold individuals performed more aggressive behaviours than shy individuals, which tended to exhibit more ‘move away’ behaviour. The major difference between bold and shy individuals was in ‘move to’, ‘cheliped display’, ‘grasp’ and ‘contact’, which were mainly performed by bold individuals in fights (Table 3).

The concentrations of glucose in the haemolymph of crabs were significantly affected by behavioural type (F_1,84=3.036, P=0.040), fighting (F_1,84=6.870, P=0.011) and their interaction (F_2,84=5.992, P=0.016; Table 4). The glucose concentrations in the haemolymph of both bold and shy individuals after fighting were significantly higher than those before fighting (Fig. 4A). Before fighting, the concentrations of glucose in the haemolymph of bold individuals were significantly lower than those of shy individuals, whereas after fighting, the concentrations of glucose in the haemolymph of bold individuals were significantly higher than those of shy individuals (Fig. 4A). Bold individuals showed a more profound increase in glucose in the haemolymph but less of an increase in lactate after fighting than shy individuals (Fig. 4A,B). The concentrations of lactate in the haemolymph of crabs were significantly affected by fighting (F_1,84=25.584, P=0.001) and the interaction between behavioural type and fighting (F_2,84=10.992, P=0.043; Table 4). After fighting, the concentrations of lactate in the haemolymph of both bold and shy individuals were significantly higher than those before fighting (Fig. 4B). Before fighting, the concentrations of lactate in the haemolymph of bold individuals were significantly higher than those of shy individuals; however, there was no significant difference between the two types of individuals after fighting (Fig. 4B).

The concentrations of glycogen in the claw muscle were significantly affected only by fighting (F_1,84=5.755, P=0.019), and the concentrations of lactate in the claw muscle were not significantly affected by behavioural type or fighting (Table 3). After fighting, the concentrations of glycogen in the claw muscle tissues of both bold and shy individuals were significantly lower than those before fighting, and the concentrations of glycogen in claw muscle tissues were not significantly different between bold and shy individuals (Fig. 4C). After fighting, there were no significant changes in the concentrations of lactate in the claw muscle in both bold and shy individuals; similarly, there were also no significant differences in the lactate concentrations in the claw muscle between the two behavioural types of individuals (Fig. 4D).

Animal personality is often a crucial determinant of important fitness correlates, such as growth, reproduction and survival (Smith and Blumstein, 2008; Sih et al., 2004), and it can have important ecological and evolutionary implications (Careau and Garland, 2012; Zavorka et al., 2015). In the present study, our results were in line with findings from earlier studies of personality traits in crustaceans, for example, the hermit crab P. bernhardus (Briffa et al., 2008) and the crayfish Pacifastacus leniusculus (Pintor et al., 2008). Large inter-individual variation in boldness, activity and hesitancy were observed in the swimming crab P. trituberculatus. Recent studies have suggested that adaptive variation in personality can arise as a result of life-history trade-offs (Le Galliard et al., 2013) or as a consequence of variations in energy reserves across individuals (Dall et al., 2004; Dingemanse and Wolf, 2010; Mittelbachgary et al., 2014). Therefore, the differences in personality traits of the swimming crab may be related to the energy status of individuals. Glycogen and glucose, as an important energy pool, represent an individual's physiological state and are closely related to behaviour performance (Briffa and Elwood, 2004). In our experiment, the initial glucose concentrations in the haemolymph were lower in bold crabs than shy crabs. This result was parallel to earlier findings in other species. For example, Huntingford et al. (2010) have shown that the blood glucose concentrations of the common carp (C. carpio) are lower in bold than in shy individuals. Higher glucose concentrations in the haemolymph provide an energy basis for higher activity. In the present study, there were no significant differences in the glycogen concentrations in claw muscle between the bold and shy crabs. One explanation for the result may be the feeding regime during the maintenance period. Glycogen concentration is related to the food sources of the environment (Belgrad et al., 2017), and crabs were fed ad libitum at 08:00 h every morning with enough clams during the maintenance period, thus possibly explaining the insignificant differences in glycogen concentration in the claw muscle between the two groups. Lactate accumulation in the haemolymph and muscle tissues is usually used as an effective indicator to compare the differences in anaerobic respiration among crustaceans (Sánchez et al., 2001). Before fighting, there were no significant differences in the lactate concentrations in claw muscle between bold and shy individuals. However, the initial lactate concentrations in the haemolymph were higher in bold crabs than in shy crabs. These results indicated the presence of an anaerobic respiration process in both types of individuals before fighting, and there was a difference in the anaerobic respiration intensity between the bold and shy crabs.

As predicted, we found significant correlations among personality traits (boldness, activity and hesitancy) in P. trituberculatus. Our results were consistent with previous findings of personality traits in other crustaceans. For example, Mowles et al. (2012) found a stable correlation between boldness and exploration behaviour in the hermit crab P. bernhardus; moreover, in the Chinese mitten crab, E. sinensis (Brodin and Drotz, 2014), and the mud crab P. herbstii (Belgrad et al., 2017), bold individuals are more active than shy individuals. In addition, several studies have also found that boldness is correlated with aggressiveness, and bold individuals are more likely to initiate and win a fight than shy individuals (Chang et al., 2012; Gherardi et al., 2012). In the present study, the number of fights initiated by bold individuals was significantly higher than that initiated by shy individuals, and the bold crabs that initiated fights were more likely to win than the shy crabs. This result is consistent with findings from an earlier study showing that in the fiddler crab Uca mjoebergi, bold individuals are more willing to initiate a fight to gain other individuals' caves (Reaney and Backwell, 2007). During fighting, agonistic behaviour reflects the individual's aggressiveness (Courtene-Jones and Briffa, 2014). In the present study, the frequencies of ‘move to’, ‘cheliped display’, ‘grasp’ and ‘contact’ of the bold crabs were significantly higher than those of the shy crabs, whereas the frequency of ‘move away’ was lower in the bold individuals than the shy individuals. This result indicated that the bold crabs were more aggressive than the shy crabs.

The energy reserves of crustaceans play an essential role in determining fight outcomes and behaviours (Gherardi et al., 2012). During fights, glycogen reserves are mobilized to release glucose to meet energy demands. In this experiment, the concentrations of glycogen in the claw muscles of both bold and shy individuals decreased after fighting, and the concentrations of glucose in the haemolymph showed the opposite trend. These results indicated that both bold individuals and shy individuals had elevated levels of glucose, mobilized from glycogen, to meet energy needs during the fight. After fighting, the concentrations of glucose in the haemolymph were higher in the bold individuals than in the shy individuals. This result suggested a difference in energy metabolism between the bold and shy crabs, and more glucose in the haemolymph provides a basis for bold crabs to initiate and win fights (Briffa and Elwood, 2004). After fighting, the concentrations of lactate in the haemolymph and claw muscles of crabs were higher than those before fighting. These results indicated the presence of anaerobic respiration processes in both types of individuals during fighting. After fighting, the lactate concentrations in the claw muscle of both bold and shy individuals were not significantly different from that before fighting. The concentrations of glucose in the haemolymph of both bold and shy individuals significantly increased after fighting. This may be related to the concentration of hyperglycaemic hormone, which drives glucose concentrations and lower lactate in the muscles (Chung and Webster, 2005). In addition, bold individuals showed a greater increase in glucose in the haemolymph but less of an increase in lactate than shy individuals. However, there were no significant differences in the concentrations of lactate in claw muscles and haemolymph between the two types of individuals after fighting. These results indicated that there was no significant differences in the anaerobic respiration process between bold and shy individuals during fighting.

In conclusion, our study documented the distribution of, and individual variations in, boldness, activity and hesitancy in P. trituberculatus, and demonstrated the links among personality, agonistic behaviour and energy metabolism. Bold crabs were more aggressive than shy crabs, and the differences in energy metabolism affected their performance in terms of personality traits and agonistic behaviour. However, animal behaviour is not likely to be driven by any single factor, but is probably regulated by multiple factors acting simultaneously (Beekman and Jordan, 2017), for example, genetics (Purcell et al., 2014), neuroendocrinology (Chung and Webster, 2005) and the environment (Toscano et al., 2014). Therefore, further determining the influence of different factors on the behaviour of crabs remains a long-term challenge.

We thank International Science Editing (http://www.internationalscienceediting.com) for editing an earlier version of this manuscript.