The effects of social experience on the behavioral response to unexpected touch in crayfish

Dominance hierarchies in crayfish populations, as in other social animals,are characterized by differences in individual behavior that span the range from the aggressive displays of the most dominant to the submissive and avoidance responses of the most subordinate. Dominant crayfish often display an elevated stance, move about freely and compete aggressively for available resources, whereas subordinate crayfish adopt a more flat posture, move to avoid contact with dominant animals and give way to dominants in competitive interactions (Hayes, 1975; Livingstone et al., 1980; Bruski and Dunham, 1987; Krasne et al., 1997; Figler et al., 1999).

Dominance status affects non-social behaviors as well as social behaviors. Herberholz et al. found that burrow digging increased among new social dominants and was inhibited in new social subordinates(Herberholz et al., 2003). Similarly, an unexpected touch of the tailfan elicited an avoidance reaction from low-ranked or small crayfish and an immediate, oriented and aggressive reaction from high-ranked or large crayfish(Bovbjerg, 1953; Nagayama et al., 1986). Under the threat of unexpected attack, dominant and subordinate crayfish rely on different types of escape circuitries that produce the escape response, the tailflip (Krasne et al.,1997).

To study the effect of social context on the crayfish's response to unexpected touch more completely, we examined the behavioral responses of socially inexperienced and experienced crayfish in three different social conditions: while socially isolated for a prolonged period, while briefly isolated from a dominant or subordinate partner and while in the presence of a dominant or subordinate partner. We found that the behavioral responses of a crayfish to an unexpected touch depended on both its social status and on the social context at the time of the touch. Part of this study was previously presented in abstract form (Song et al.,2000).

Adult crayfish (Procambarus clarkii Girard), of wet body mass 4–8 g, were used in this study. Animals were purchased from a local supplier (Atchafalaya Biological Supply, Raceland, LA, USA) and divided into groups of 8–12 in communal tanks (76 l) until use. Pairs of same-sex animals, mismatched by wet body mass (>5%), were transferred to 9 l tanks(20×25×40 cm, W×H×L). An opaque plastic divider placed halfway along the long axis of the aquarium prevented the animals from seeing or touching each other but failed to isolate them chemically. The bottom of each tank was covered with gravel and the tanks were filled with continuously aerated dechlorinated tapwater. Animals were each fed with five shrimp pellets three times a week and maintained on a 12 h:12 h light:dark cycle.

The animals were `semi-separated' between daily interaction sessions by an open plastic square lattice (the plastic strips were 9 mm thick, 20 cm deep,and the square openings were 15×15 mm wide)(Fig. 1C). This open divider allowed the paired animals to see, touch and smell one another but prevented vigorous fighting. For the eight pairs in Social Group, an opaque divider replaced the open divider at the end of the two weeks of pairing to establish isolated conditions for subsequent experiments.

On the next morning, all animals were tested over a 3-h period(Fig. 1D). Isolates and paired animals remained isolated by a closed divider during testing. A minimum of five manual touches was delivered in alternation to each side of the first abdominal segment with a fine brush by an experimenter working under dim red light and ignorant of the social status or experience of the animal undergoing tests. Each touch stimulus was delivered to minimize disturbances of the water, which would alert the animal to the approach of the brush, so that the animal was unexpectedly touched. Responses were recorded on videotape(Panasonic, WVBP500; 30 frames s^–1) for later analysis. The interval between touches was 10 min. Tests were performed when the animals were stationary and not interacting. Following the experiment, the opaque dividers were replaced to separate the dominant and subordinate for approximately 15 h overnight (Fig. 1E).

The opaque dividers between the pairs were removed to allow the dominant and subordinate animals to interact freely(Fig. 1F). After 30 min of interaction, these animals were tested for 2–3 h while together, whereas the Isolates were tested while remaining isolated(Fig. 1G). After the initial tests, open dividers were placed between the dominant and subordinate animals for 30 min to minimize physical contact between them while permitting chemical signaling (Fig. 1H). The animals were then tested with the open dividers still in place(Fig. 1I). After these tests,the open dividers were replaced with opaque dividers to re-isolate the paired animals for 30 min (Fig. 1J). The animals were then tested for a final time while isolated(Fig. 1K). Each set of tests took 2–3 h to perform.

Video images containing the positions of an animal before and after each manual touch (stimulus) were captured using Scion image (Scion Image, NIH)(Fig. 2). Behavioral responses were examined and categorized into `Orienting', `Avoidance' and `No Response'groups. Orienting responses were movements immediately following the touch that reoriented the animal to face the stimulus source(Fig. 2A). The category includes three behaviors: a turn towards the stimulus source, a brief walk backward while turning to face the stimulus source, or a short backward jump with a brief abdominal flexion to face the stimulus source. Avoidance responses were movements away from the stimulus source. These included five behavioral responses: the animal rapidly walked forward away from the site of the touch (Fig. 2B), it walked backward away, it walked sideways away, it rotated the body axis away from the stimulus source without walking or it tailflipped away. The No Response behavior included all those in which the animal remained in the same position. Movements or responses that were initiated two seconds or more after the manual touch were not considered.

A quantitative description of the movements was obtained by marking the position of the stimulus probe at the time of the manual touch (P) and the positions of the animal's head (H) and the center of body mass (C) after the manual touch (Fig. 2C,D). Using these marks, the head to probe distance value (HP distance value=HP distance/body length) and the angle γ (the angle between the body axis and the HP vector; see Fig. 2C,D) were obtained. The HP distance decreased when an animal moved towards the stimulus probe (Fig. 2C), whereas it increased when an animal moved away from the stimulus probe (Fig. 2D). The angle γ tended to decrease either when the animal turned to orient towards the probe (Fig. 2Ci) or moved a body length or more backward to avoid the probe(Fig. 2Di). The angle γwas large when an animal did not move, turned toward the stimulus probe and moved forward, beyond the probe position(Fig. 2Cii), or moved forward(Fig. 2Dii).

The behavioral response patterns of the isolate, dominant and subordinate animals were compared using the Kruskal–Wallis test (nonparametric one-way ANOVA). When an overall significant difference was found in the behavioral response patterns of the three social types of animals, the Mann–Whitney test (nonparametric two-tailed t-test) was used to identify which type of animals differed. The Friedman test was used to examine the overall difference among the response patterns of the dominant and subordinate animals in the four consecutive social conditions; separated,paired, semi-separated and separated. The Wilcoxon signed rank test was used for pair-wise comparison only when there was an overall significant difference throughout the four different behavioral response patterns.

Eight pairs of animals that were allowed to interact for 30 min day^–1 for two weeks formed dominant–subordinate relationships as indicated by their scores in a dominance index (attack, +2;approach, +1; retreat, –1; tailflip, –2). Six out of the eight pairs of animals in Social Group formed stable dominant–subordinate relationships (Fig. 3), while status reversals occurred in two pairs (pairs 2 and 3 in Fig. 3), after one animal in each pair molted. No status reversals took place during the experiments.

Isolates were tested while isolated; members of pairs in Social Group were tested when separated from their partners by an opaque divider(Fig. 1D). Unexpected lateral touch stimuli were delivered to animals in both groups under dim red light so that the animals could not see the approach of the brush. The response patterns of the Isolates and dominant and subordinate members of Social Group were similar. The predominant response was to orient towards the stimulus:average frequencies fell between 68 and 80% for the three social classes(Isolates, dominants and subordinates of Social Group; Fig. 4A). The frequencies of Avoidance and No Responses were approximately equal in each class, and no statistical difference was found between the three classes' response patterns(Kruskal–Wallis test, P=0.4570 for Orienting Responses, P=0.7133 for Avoidance Responses, P=0.6163 for No Responses).

The Isolate animals and the animal pairs of Social Group spent the next 15 h still separated (Fig. 1E). Isolate animals remained isolated, while animal pairs in Social Group were allowed to interact freely for 30 min under dim red light(Fig. 1F), after which they were tested again (Fig. 1G). The average frequencies of the Orienting Responses of the Isolates and the dominants and subordinates of Social Group significantly differed(Kruskal–Wallis test, P=0.0008). The Isolates and the dominants(who were free to interact with the subordinate partners) displayed the Orienting Responses in similar frequencies to each other(Fig. 4B; Mann–Whitney test, P=0.6126) and to those of the previous day(Fig. 4A). More than 70% of their responses were to orient towards the stimulus; most of the remaining stimuli evoked No Response (Fig. 4B). By contrast, the subordinate animals behaved very differently from both dominants and Isolates, displaying Avoidance Responses to more than 76% of the stimuli and Orienting Responses to only 19%(Fig. 4B). The average frequencies of the Orienting Responses displayed by the subordinates were significantly lower than those of the Isolates (Mann–Whitney test, P=0.0012) and dominants (Mann–Whitney test, P=0.0002). Conversely, the average frequencies of the Avoidance Responses of the Isolates, dominants and subordinates of Social Group significantly differed(Kruskal–Wallis test, P=0.0003), because the subordinates displayed the Avoidance Responses significantly more than the Isolates(Mann–Whitney test, P=0.0003) and dominants (Mann–Whitney test, P=0.0002). The behavioral responses of the subordinates were also different from those of the day before, when the subordinates were separated from their dominant partners(Fig. 4E; Friedman test, P=0.0004 for an overall significant difference; Wilcoxon test, P=0.0078 for pair-wise comparison). This difference indicates that the behavioral responses of the subordinate animals were significantly affected by the presence of the dominant animals.

To determine whether the change in the subordinate animals' responses depended on being able to interact freely with the dominants, the animals were tested again after being separated for 30 min by an open divider, again under dim red light. During this period, the subordinate animals usually stayed away from the open divider, whereas the dominants would often approach it. Both animals, especially the subordinates, often moved along the surrounding walls,touching them with their claws. When the subordinates touched the holes of the open divider, the dominants often tried to attack them, but the plastic mesh prevented serious physical interactions from occurring. The responses of the dominants were as before, with nearly two-thirds being the Orienting Responses(Fig. 4C). However, the subordinate animals displayed the Orienting and the Avoidance Responses at similar frequencies (Fig. 4C),a pattern that newly emerged in subordinates after pairing experience. Moreover, the frequencies of the Orienting Responses of the subordinate animals in a semi-separated condition were significantly different from each of the response frequencies recorded earlier(Fig. 4E), first when the subordinates were isolated on Day 1 (Wilcoxon test, P=0.0313), and then when they were paired on Day 2 (Wilcoxon test, P=0.0156). Similarly, the Avoidance Responses of the subordinate animals in a semi-separated condition were significantly reduced compared with those recorded earlier when they were paired on Day 2 (Wilcoxon test, P=0.0156).

To determine whether the subordinate animals' change in behavior depended on the near presence of the dominant, as communicated through the open divider, we tested the animals again when they were completely separated by an opaque divider. The behavior patterns of both the subordinate and dominant animals (Fig. 4D) were little changed from when the animals were separated by the open divider(Fig. 4C). The behavioral responses of the subordinates were neither significantly different from those of dominants (Fig. 4D;Mann–Whitney test, P=0.2345 for Orienting Responses, P=0.3282 for Avoidance Responses) nor different from those recorded earlier when the subordinates were in a semi-separated condition(Fig. 4E; Wilcoxon test, P=0.5625 for Orienting Responses, P=0.8125 for Avoidance Responses). However, the subordinates when isolated on Day 2(Fig. 4D) displayed the Orienting Responses less frequently compared with when they were isolated on Day 1 (Fig. 4E; Wilcoxon test, P=0.0313) but more frequently compared with when they were paired on Day 2 (Fig. 4E; Wilcoxon test, P=0.0156). These results indicate that both the presence of the dominant animal and the lingering effects of their interactions during the test period changed the behavioral response of the subordinate animals to unexpected touch.

There was no difference in the behavioral response of the dominant animals in the four different agonistic conditions tested (Friedman test, P=0.0979 for Orienting Responses, P=0.09042 for Avoidance Responses, P=0.2992 for No Responses), indicating that the behavioral responses of the dominant animals were not dependent on the presence of their subordinate partners. Thus, it is either the opportunity to interact with the dominant or the interactions themselves that led to a shift in the response patterns of the subordinate animals from Orienting to Avoidance.

To visualize the difference between the Orienting and the Avoidance responses of the socially experienced and inexperienced animals, the positions of each animal before and after the stimulus were quantified (see Materials and methods for details). Two related measures were used: the distance between the final head position and the site of the probe's touch on the animal's side, in body lengths (HP), and the angle γ between the animal's body axis and the direction from the body center to the site of the probe contact.

No Responses were characterized by low HP values and γ angles between 90 and 180°, whereas Orienting Responses were associated with similar HP values and a broader range of angles, between 0 and 180°(Fig. 5). Avoidance Responses displayed much larger HP values and γ angles between 90 and 180°.

Isolate, dominant and subordinate animals displayed similar responses on Day 1 when tested while isolated (Fig. 4A). Their patterns of movement were also similar, clustering around low HP values and γ angles between 90 and 180°(Fig. 5A). A few animals of each social type displayed Avoidance or Orienting Responses characterized by low γ values. The movement pattern of the subordinates reflected the change in their responses when paired on Day 2(Fig. 4B). Subordinates displayed Avoidance Responses (red squares) with much larger HP values,extending to more than three body lengths(Fig. 5B). Most Avoidance Responses also had γ angles near 180°, indicating that the animal had moved forward, away from the probe. A smaller number had low γvalues, indicating that the animal moved backward. When the animals were semi-separated by an open divider, the responses of the social dominants did not change significantly (Figs 4C, 5C), whereas the Avoidance responses of the subordinates were again characterized by large HP values and large γ angles, indicating forward movement away from the probe. Separation of the subordinates brought partial recovery of the original movement patterns; the HP distribution of Avoidance Responses was reduced from its highest values but still extended to greater values than when the subordinates were tested initially (Fig. 5D).

Both the nervous system and behavior of crayfish depend on the animal's social status. Serotonergic modulation of the excitability of the command neurons for escape and the threshold for escape behavior depend on the social status of crayfish (Yeh et al.,1996; Yeh et al.,1997; Krasne et al.,1997; Teshiba et al.,2001), and crayfish display agonistic behaviors typical of their social status (Goessmann et al.,2000; Herberholz et al.,2001). Dominant and subordinate animals also differ in their willingness to engage in other non-social behaviors, such as shelter construction, in the presence of a social partner(Herberholz et al., 2003). Subordinate animals are inhibited from digging in the presence of a dominant partner, while the dominant's burrowing activity is increased. The inhibition was conditional upon the near presence of the dominant but also appeared to linger after the subordinate had been isolated.

Previous reports indicated that the response of a crayfish to an unexpected touch depended on the animal's size(Nagayama et al., 1986) or dominance status (Bovbjerg,1953). The types of escape reflex circuitries to be activated were dependent on social status of crayfish under the threat of unexpected attack,while the excitability of the lateral giant escape reflex was independent of social status when the threat was removed(Krasne et al., 1997). Here,we have found that a crayfish's response to an unexpected touch depends on the animal's social context as well as its social status. The response of a social subordinate in the presence of its dominant partner differed from its response when alone. When kept apart from their partner by a closed barrier, the social status or experience of the animal had no effect on its responses, which were identical to those of socially isolated animals. Isolates and separated dominants and subordinates all turned to confront the source of the unexpected touch (Figs 4A, 5A,B). However, when subordinates were tested while in partial contact with their dominant partners(i.e. semi-separated), their patterns of responses differed from when tested alone (Figs 4C,E, 5C). These differences increased dramatically when the subordinates were tested in the presence of their dominant partner: instead of turning to confront the unexpected touch,they moved away (Figs 4B,E, 5B). It is apparent, therefore,that as in the case of burrowing behavior(Herberholz et al., 2003), the social context of being subordinate in the presence of the dominant partner determined the change in the subordinate's response to the stimulus.

Subsequent tests showed that the change in the subordinate's response brought about by being tested while paired persisted to a small degree for some hours (Fig. 4E). Tests while semi-separated (Figs 4C, 5C) and while alone (Figs 4D, 5D) revealed that the subordinates' behavior was not restored to that of the earlier isolated condition (Figs 4A, 5A) five hours after the tests while paired.

These results suggest that the effect of the paired context in which the unexpected touch was received lingers for some hours after the event. This effect may be a state change, similar to fear, or it may be a memory (i.e. a specific association) of having received the earlier unexpected touches in the presence of the dominant. When the subordinates were tested while semi-separated from their dominant partners, this state or memory may have been strengthened by olfactory stimuli from the dominant. When the subordinates were finally tested while isolated from their dominant partners,the response pattern showed some influence of the earlier paired context in which the animals were tested.

The persistence of the change in the subordinate's behavior when the dominant was absent is similar to the lingering inhibitory effect of the dominant on the subordinate crayfish's burrowing activity(Herberholz et al., 2003). In both these cases, it is not clear whether this persistence represents a specific memory of a more generalized change in behavior, like fear. Crayfish,like other crustaceans, can learn specific associations(Krasne, 1973). Several studies have demonstrated the ability of crayfish to learn to recognize crayfish in the process of dominance hierarchy formation and maintenance(Bovbjerg, 1953) – for individual recognition (Lowe,1956), recognition of aggressive state(Copp, 1986) and status recognition (Zulandt-Schneider et al.,1999) – although what is recognized and when the animal learns are not completely understood.

The complete neural mechanisms of the different behavioral responses to unexpected lateral touch are unknown, but some elements have been identified. Bilateral pairs of serotonergic neurons in the abdomen and thorax of crayfish receive both excitatory and inhibitory inputs in response to a lateral touch of the rostral portion of the abdomen, but the mix of excitation and inhibition depends on the social status of the animal(Drummond et al., 2002). The ipsilateral 5-hydroxytryptamine (5-HT) neurons of Isolate and dominant crayfish were excited by a unilateral touch while the contralateral 5-HT neurons were inhibited. The same neurons in subordinate crayfish were symmetrically excited in several crayfish and inhibited in as many others;asymmetric responses did not occur. These 5-HT neurons have been found to modulate walking leg reflexes, which also display differences between dominant and subordinate animals (F. A. Issa, D. H. Cattaert and D. H. Edwards,unpublished observations). It is tempting to link the asymmetric neuronal responses of Isolate and dominant crayfish to their orienting responses described here, and the symmetric excitatory or inhibitory neuronal responses of subordinate crayfish to their symmetric avoidance responses. Future studies will determine whether these correlations between neuronal responses and behavior are indicative of underlying causal mechanisms.

This study was supported by a National Science Foundation grant IBN-9726819 to D.H.E. and by the Department of Biology, Georgia State University. We thank Dr J. Drummond for helpful comments on the designs of experiments, Dr C. D. Derby for help with the statistics, Dr S. Versteeg for help with quantification of data, and Mr F. A. Issa for help with animal preparation. We also thank Mr F. A. Issa, and Ms L. M. Johnstone for critical readings of the manuscript.