- Email: [email protected]
Honest and dishonest displays, motivational state and subsequent decisions in hermit crab shell fights
ANIMAL BEHAVIOUR, 2006, 72, 853e859 doi:10.1016/j.anbehav.2006.01.025
Honest and dishonest displays, motivational state and subsequent decisions in hermit crab shell fights R. W. ELW OOD, R. M. E. POTH AN IK AT & M. B RIF F A
School of Biological Sciences, The Queen’s University of Belfast (Received 22 July 2005; initial acceptance 7 September 2005; final acceptance 24 January 2006; published online 17 August 2006; MS. number: 8625R)
Animal fights are typically preceded by displays and there is debate whether these are always honest. We investigated the prefight period in hermit crabs, Pagurus bernhardus, during which up to four types of display plus other activities that might provide information are performed. We determined how each display influences or predicts various fight decisions, and related these displays to the motivational state of the attacker, as determined by a startle response, and of the motivational state of the defender, as determined by the duration for which it resisted eviction from its shell. Two displays appeared to have consistent but different effects. Cheliped presentation, where the claws were held in a stationary position, often by both crabs but for longer by the larger, seemed to be honest, and allowed for mutual size assessment. This display enhanced the motivation and the success of the larger crab. In contrast, cheliped extension, involving the rapid thrust of the open chelae towards the opponent, did not seem to allow for mutual size assessment and may contain an element of bluff. It was performed more by the smaller crab and enhanced its success. The complexity of displays in this species appears to allow for both honesty and manipulation. 2006 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
Contests in animals commonly involve the use of displays. These may be relatively low-intensity activities at the start of an encounter but may escalate as the contest progresses. In many species, both contestants use the same type of display, but may differ in the vigour or magnitude of performance. This variation may be used in settling the outcome of the contest without recourse to physical combat, presumably because it provides honest and reliable mutual evaluation of fighting ability or resource-holding potential (Parker 1974; Hughes 2000; Maynard Smith & Harper 2003; Hurd & Enquist 2005). Examples are found in the roaring contests of red deer, Cervus elaphus (Clutton-Brock & Albon 1979) and ‘lateral displays’ in cichlid fish, Cichlasoma nigrofasciatum (Keeley & Grant 1993). Other species have several distinct displays and those selected may vary between contestants and, within contestants, between encounters. The function of these displays is less clear but there seems to be scope for deception and manipulation of the opponent (Krebs & Dawkins 1984). Stomatopods, Gonodactylus bredini, for
Correspondence: R. Elwood, School of Biological Sciences, The Queen’s University of Belfast, Belfast, BT9 7BL, U.K. (email: [email protected]. uk). M. Briffa is now at the Marine Biology and Ecology Research Centre, School of Biological Sciences, University of Plymouth, Drake Circus, Plymouth PLA 8AA, U.K. 0003e 3472/06/$30.00/0
example, use various displays including the ‘meral spread’ (Adams & Caldwell 1990), which is typically shown by aggressive individuals and may cause the opponent to retreat. It may, however, be shown by newly moulted individuals, which have a low fighting ability, as a bluff. Also, the ‘open chela display’ of snapping shrimps, Alpheus heterochaelis, is used more often by those individuals that have larger chelae than predicted by body size and the display exaggerates their true fighting ability (Hughes 2000). In male fiddler crabs, Uca annulipes, recently regenerated major chelae are thinner and lighter and hence less energy is required when they are used in displays than in displays of original claws. Males with recently regenerated chelae can bluff fighting ability as it is impossible for the opponent to determine claw mass by means of visual assessment (Backwell et al. 2000). Theory suggested that deceptive displays by the weaker opponent should be rare and that on average displays should be honest (Johnstone 1998), although Backwell et al. (2000) suggested that cheating may be more common than was previously thought on the basis of the evidence from fiddler crabs. In either situation, displays clearly have the potential to change the behaviour of the opponent (Rubenstein & Hazlett 1974; Hyatt & Salmon 1979). In the case of ‘honest displays’ the advantage to the sender is that the displays may settle the contest without costly combat, even if the sender loses the interaction (Hurd 1997; Deag & Scott
853 2006 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
854
ANIMAL BEHAVIOUR, 72, 4
1999). In the case of manipulation or deception, however, the advantage to the sender is that the displays could mask or exaggerate the true fighting ability of the sender and reduce the receiver’s motivation to fight (Hurd & Ydenberg 1996). The receiver’s motivation may easily be detected if the receiver immediately retreats from a weak opponent after that opponent has used a display typically given by an individual of high fighting ability (Hughes 2000). Regardless of the honesty of the signal a fight might still occur. In this case the motivation of the opponent clearly has not been reduced sufficiently to induce retreat. However, causing a slight reduction in the opponent’s motivation to fight, while not preventing a fight, might yet alter the probability of eventual victory. For instance, the time or energy the opponent is prepared to commit may be reduced. Even a slight increase in the probability of victory may make performing the display worthwhile. The benefit to a weak sender of a false signal may thus be great because it might avoid a fight that it would otherwise lose. If the signal does not deter the opponent then the sender is in the same position as if it had not sent the false signal. Thus, if the cost of sending the false signal is low and there is no specific retribution from the receiver then the use of false signals will be selected for (Adams & Mesterton-Gibbons 1995; Johnstone 1998). We investigated the use of various prefight displays in hermit crabs, Pagurus bernhardus, contesting ownership of shells. These displays use the chelipeds and walking legs and also a high posture in which the shell is lifted high off the substrate (Hazlett 1968; Elwood & Neil 1992). However, other activities not normally described as displays may convey important information. These include mutual grappling, approaching the opponent and retreat from the opponent. The prefight phase may be followed by an escalated fight in which one crab, termed the ‘attacker’, initiates the shell fight by grabbing the shell of the ‘defender’, causing the defender to withdraw into its shell (Dowds & Elwood 1983). The attacker may then engage in repeated bouts of vigorous shell rapping in which the attacker hits its shell upon that of the defender, until either the defender is evicted from the shell, enabling the attacker to take that shell, or the attacker gives up. The effect of the power, number of raps per bout, number of bouts and the duration of pauses between bouts on the physiology and resistance of the defender, and the interplay between the physiology and the performance of rapping of the attacker, have been the subject of recent investigation (Briffa & Elwood 2000, 2001a, 2002, 2003, 2005). This energetically demanding rapping is difficult to fake as only animals in good condition (with low lactate) are able to produce high-power vigorous rapping and only these crabs are guaranteed victory (Briffa & Elwood 2001a, 2002, 2005). In contrast, the relatively brief postures seen in prefight displays are unlikely to be energetically costly and thus the condition of the animal is thus less likely to limit their use. Studies on hermit crab prefight displays have used models (e.g. Hazlett 1968) and information theory (Hazlett & Bossert 1965) in which the immediate effects of a display on the behaviour of the receiver were noted. Our aim in the present study, however, was to determine longer-term
consequences of prefight displays. First, we asked whether prefight displays, and other activities, influence or predict which crab takes the role of attacker. This is an important decision as it is only the attacker that, if it evicts the defender, is able to choose which of the two shells to occupy. Second, we examined whether prefight activities influence or predict the winner of the contest. Third, we aimed to determine whether displays influence the motivation of the attackers (but not that of the defenders) by presenting a novel, startling stimulus and recording the duration of the attacker’s startle response (Elwood et al. 1998), which is inversely related to the level of motivation. Attackers with a high potential gain in shell quality and low probable costs have short startle responses (Elwood et al. 1998) and highly motivated attackers are more likely to win (Briffa & Elwood 2001b). Finally, we examined how these prefight activities influence the duration for which the defender resists eviction. Again, this should elucidate relations between the prefight displays and motivation because the duration of resistance is a common measure of motivation in losers of fights (Hack et al. 1997; Bridge et al. 2000). Our key aim was thus to determine how various displays by one crab influence the behaviour or motivation of the receiver and if those displays provide honest information or a means of manipulation.
METHODS Small (0.10e0.64 g) littoral specimens of the common European hermit crab were collected weekly from Ballywalter, Co. Down, Northern Ireland, U.K. between October 2004 and February 2005. They were held in groups of 40e50 in plastic tanks (60 30 cm) filled with aerated seawater at 12 C to a depth of 10 cm, and fed ad libitum on commercial fish food (catfish pellets). We removed the crabs from their shells by cracking the shells open in a bench vice. We used only males for staging encounters. We gave females new shells and returned them to the sea, thus avoiding sex differences in behaviour that have been noted in previous studies (Neil & Elwood 1985). Only male crabs that were free from obvious parasites, loss of appendages and recent moult were used. Use of the vice did not harm the crabs. We allocated male crabs to pairs and pairs to one of three groups. Each crab was weighed and the relative weight difference (RWD) of the pair was calculated by RWD ¼ 1 (small crab weight/large crab weight). The relative weight difference ranged from 0.05 to 0.58 (X SE ¼ 0:22 0:008) and there was no difference in RWD between the three groups. To determine the preferred weight of shell for the larger crab of each pair we used previously calculated regression lines that relate crab weight to preferred shell weight (Jackson 1988). In all cases the smaller crab received a Littorina obtusata shell that was 100% of the preferred weight for the larger crab. In group 1 (L50), the larger crab received an L. obtusata shell that was 50% of its preferred size. In group 2 (G50), the larger crab of each pair received a shell of Gibbula cineraria that was 50% of its preferred shell weight. This species is normally avoided by hermit crabs if given
ELWOOD ET AL.: DISPLAYS AND FIGHT DECISIONS
a choice (Elwood et al. 1979) and those occupying the species have reduced reproductive output compared to those in L. obtusata (Elwood et al. 1995). In group 3 (G100), the larger crab received a G. cineraria shell that was 100% of its preferred shell size. Each crab was isolated with its new shell in a crystallizing dish 10 cm in diameter, containing aerated sea water at 12 C and a fine layer of sand, for approximately 16 h prior to the fight being staged. The fight arena was a glass tank (15 7 cm and 7 cm high) that contained aerated sea water and a layer of sand 1 cm deep to provide adequate grip for the walking legs of the crabs during the fight. We separated the crabs by placing them inside two clear plastic cylinders (3.7 cm in diameter and 5.5 cm high), which were positioned such that the cylinders were touching, enabling visual contact. The arena was placed behind a one-way mirror such that the crabs could not see the observer. To record the behaviour of large and small crabs we used a Psion Workabout hand-held computer configured as a timeeevent recorder with The Observer 3.0 software (Noldus Technology, Wageningen, The Netherlands). The two crabs were placed into the cylinders and allowed 2 min to settle before we removed the cylinders and started the observation. We recorded display and other activities until an ‘attacker’ initiated the fight causing a ‘defender’ to withdraw into its shell. At this point a novel stimulus was applied to elicit a startle response. The stimulus was a weighted (total weight ¼ 180.25 g) rectangle of opaque Perspex (19.5 9.5 cm), which was suspended on a thread 30 cm above the top of the arena. This was released in such a way that it fell on to the top of the arena (the width of the Perspex was greater than that of the arena, so it did not fall into the water but came to rest on the rim) and was then pulled back to its original position. This technique is known to cause a startle response in fighting hermit crabs (Elwood et al. 1998; Briffa & Elwood 2001b), which involves a temporary cessation of activity while the attacker withdraws either partially or completely into its shell. Since defenders withdraw into their shells when the fight starts we could measure the startle response only for attackers. The duration of the startle response was defined as lasting from the application of the stimulus to resumption of activity by the attacker. After the crabs recovered from the startle response, interactions were allowed to continue until the attacker either evicted the defender or gave up without effecting an eviction or, if a fight was not initiated, after 30 min. Prior to a fight being initiated, behavioural recording focused on prefight displays involving the asymmetrically sized chelipeds (the first pair of pereiopods that bear the claws) and the walking legs (second and third pairs of pereiopods; Hazlett 1968). ‘Cheliped presentation’ involves the proximal part of the chelipeds being held forwards, towards the opponent, with the distal part (claw) being perpendicular to the substrate. In ‘cheliped extension’ the claws are held horizontal to the substrate and raised up at least to the level of the head of the displayer, typically with the chelae open. Diagrams of the key displays can be seen in Hazlett (1968) and photographs in Elwood & Neil (1992). ‘Grapple’ involves mutual wrestling
with the chelipeds and walking legs. ‘High posture’ involves the walking legs and chelipeds being used to raise the crab’s shell off the substrate. ‘Ambulatory raise’ involves one or more of the walking legs being raised off and held perpendicular to the substrate. ‘Approach’ involves movement towards the opponent, while ‘retreat’ is movement away from the opponent. During the subsequent shell fight we also recorded the pattern of shell rapping. We examined categorical data either by G tests for contingency analysis or by c2 for goodness-of-fit tests. As the latency from the start of the observation to the start of the shell fight (when the novel stimulus was applied) varied, we used the proportion of that time for which each activity occurred. However, analyses using absolute times did not differ markedly. Furthermore, analysis of prefight duration showed no significant effect on the dependent variables investigated. Principal components analysis (PCA) was used to determine which prefight activities tended to occur together in the same encounters. However, because we wished to examine the effects of particular displays we did not use PCA scores in further analyses. Thus we used the behavioural categories in multivariate logistic regression when the dependent variable was dichotomous. When the dependent variable was continuous we used stepwise regression. Paired t tests were used to compare the large and small crabs to determine which performed each of the displays more than the other. Data were log transformed as appropriate.
RESULTS Of the total of 240 observations there was significant variation between the groups in the type of interaction (fight initiated by large or small crab or no fight; G tests: G4 ¼ 15.14, P ¼ 0.004) because of the large number of fights initiated by the large crab in the G50 group and the low number without fights in this same group. When just observations of fights were considered (N ¼ 148), significant variation between the three experimental groups in the number of fights initiated by the large and small crabs was noted (G2 ¼ 6.51, P ¼ 0.04). More were initiated by the large crab in the G50 group than in the L50 group (G1 ¼ 4.15, P ¼ 0.04) and in the G50 group than in the G100 group (G1 ¼ 5.76, P ¼ 0.02). There was, however, no significant difference between the L50 group and G100 group (G1¼0.042, NS). Of the 148 fights, more were initiated by the large crab (N ¼ 131) than by the small crab (N ¼ 17; goodness-of-fit test: c21 ¼ 140:2, P < 0.0001) (Table 1). When fights were initiated by the small crab there was less of a relative weight difference between the opponents (paired t test: t146 ¼ 2.26, P ¼ 0.025). The PCA on the 13 activities produced six component factors. Here we consider those activities with loadings of >0.5: PC1 comprised high posture by both crabs, ambulatory raise and approach by the large crab and cheliped extension by the small crab; PC2 comprised cheliped presentation by both crabs; PC3 comprised retreat by the large and approach by the small crab; PC4 comprised
855
ANIMAL BEHAVIOUR, 72, 4
Table 1. Numbers of observations where fights were initiated by the large crab or the small crab and where no fights occurred and the number of evictions and nonevictions when the fight was initiated by the large crab Initiation
Group
large crab
L50 (80) 33 G50 (80) 54 G100 (80) 44 Total (240) 131
small crab
No fight
Evictions
Nonevictions
6 2 9 17
41 24 27 92
16 44 36 96
17 10 8 35
Sample sizes are given in parentheses. The larger crab of each pair was in a Littorina obtusata shell that was 50% of its preferred size (L50), a Gibbula cineraria shell that was 50% of its preferred size (G50) or a G. cineraria shell that was 100% of its preferred size (G100).
small crab retreat; PC5 comprised grappling; PC6 comprised cheliped extension by the large crab and ambulatory raise by the small crab. Preliminary analyses indicated that there were no significant group effects on the individual prefight activities, so we explored initiator effects with multivariate logistic regression. This is a useful technique for analysing the effects of behavioural parameters on the likelihood of alternative events occurring during contests (Hardy & Field 1998). The 13 prefight activities, expressed as a percentage of the prefight duration, were the independent variables and the dependent variable was the large crab rather than the small crab taking the attacker role. This indicated that cheliped presentation by the large crab increased the likelihood of the large crab initiating the fight (Wald c21 ¼ 3:92, P ¼ 0.048) whereas extending the cheliped by the small crab had a negative effect on the likelihood of the large crab initiating the fight (Wald c21 ¼ 4:717, P ¼ 0.03). The relation between the 13 prefight activities and RWD was examined by stepwise regression but the regression was not significant. Paired t tests showed that the smaller crabs performed more cheliped extension than did the larger (t147 ¼ 4.685, P < 0.0001) and more retreating (t147 ¼ 2.289, P ¼ 0.024) but spent less time presenting chelipeds (t147 ¼ 1.98, P ¼ 0.05). There were no differences between the crabs in approach, high posture or ambulatory raise. Of the 131 fights initiated by the large crab, 96 resulted in an eviction of the small crab (Table 1). There was no significant difference in the relative weight, small crab weight or large crab weight between those fights in which eviction did and did not occur. There was a significant difference between the three groups in the number that resulted in eviction (G test: G2 ¼ 12.88, P ¼ 0.002). There were more evictions in the G50 group than in the L50 group (G1 ¼ 10.3, P ¼ 0.001) and in the G100 group than in the L50 group (G1 ¼ 9.56, P ¼ 0.002), but there was no significant difference between the G50 and the G100 groups (G1 ¼ 0.002, P ¼ 0.966). For the 131 fights initiated by the large crab, ANOVA showed a significant group effect for startle duration (log), with the G50 group showing the shortest responses (F2,125 ¼ 17.06, P < 0.0001). There was no outcome effect
but there was a significant interaction effect (F2,125 ¼ 4.80, P ¼ 0.01; Fig. 1), because the group effects were more marked in cases of noneviction. There were no significant main or interaction effects for absolute or relative weights or the number of bouts of shell rapping. Further analyses revealed no group effects for the 13 prefight activities and their possible effect on the defender resisting eviction was investigated by multiple logistic regression. This indicated that as the duration of cheliped extension by the small crab before the fight increased, the chance of eventually being evicted declined (Wald c2 ¼ 7.952, P ¼ 0.005) Use of cheliped presentation by the small crab had the opposite effect, increasing the likelihood of an eviction occurring (Wald c21 ¼ 4:063, P ¼ 0.043). We used stepwise regression to assess which of the 13 prefight activities might influence or predict the attacker’s startle duration for the 96 fights that ended with eviction. Three activities entered the regression model (F3,92 ¼ 10.417, P < 0.0001): large crab approach had a positive effect on the startle duration, but large crab presenting chelipeds and ambulatory raise both had a negative effect on the startle duration. A further stepwise regression was used to examine the effect of the same independent variables on the number of bouts of shell rapping required to evict the defender. The duration of the startle response was used as an additional independent variable to assess how the motivation of the attacker might influence the resistance of the defender. Two activities entered the regression model (F2,93 ¼ 8.098, P < 0.001), these being positive effects of grappling and of the large crab showing ambulatory raise but neither startle duration nor cheliped displays entered the model.
DISCUSSION Reliable signals of resource-holding potential (RHP) are expected to evolve if they reduce the need for costly combat to settle disputes (Parker 1974; Johnstone 1998; 1.2
Eviction No eviction
1 Log startle
856
0.8 0.6 0.4 0.2 0
L50
G50
G100
Figure 1. Log duration of startle responses (X SE) for the three experimental groups and for each outcome (eviction or no eviction). The larger crab of each pair was in a Littorina obtusata shell that was 50% of its preferred size (L50), a Gibbula cineraria shall that was 50% of its preferred size (G50) or a G. cineraria shell that was 100% of its preferred size (G100).
ELWOOD ET AL.: DISPLAYS AND FIGHT DECISIONS
Hughes 2000). These signals of RHP are often assumed to be costly, not easily faked and, therefore, essentially ‘honest’ (Zahavi 1975; Grafen 1990; Hughes 2000). The rationale is that easily faked signals would spread in the population if they benefited the sender and thus cease to predict true RHP. However, noncostly signals may also be honest if they act as an index of an animal’s size or condition (Maynard Smith & Harper 2003). Occasionally certain activities may be used to ‘bluff’ or exaggerate actual fighting ability or intent, particularly if the animal that is bluffing is the weaker of the two contestants (Adams & Caldwell 1990; Hughes 2000). Furthermore, the use of a variety of signals may introduce a complexity that confuses the receiver and makes the true assessment of RHP difficult (Riechert 1998). In the escalated shell fights of hermit crabs, repeated bouts of vigorous shell rapping by the attacker are energetically demanding i.e. costly, and the power of the signal is constrained by the physiological state of the sender. Shell rapping thus seems to be an honest signal (Briffa & Elwood 2001a, 2002, 2004, 2005) and an index of condition (Maynard Smith & Harper 2003). However, this may not be the case with the prefight displays of hermit crabs, which may involve relatively brief movements of chelipeds and walking legs. Clearly it is not easy to fake the size of a cheliped and a cheliped that is larger than normal for a particular body size will be expensive to produce. Not only will it require resources to be channelled into that appendage, but the appendage may also subsequently not be efficient in foraging. Furthermore, prolonged waving of the very large claw in fiddler crabs has high energetic costs (Matsumasa & Murai 2005). However, the relatively brief displays of ‘normalsized’ chelae of hermit crabs may not incur significant energetic costs. The general conclusion is that displays of different types in different species may differ with respect to their cost and honesty and thus differ in their effects on tactical decisions during contests (Johnstone 1998; Maynard Smith & Harper 2003). The first key decision that hermit crabs make in a potential shell fight is whether to initiate an attack. This will determine the role, attacker or defender, that each crab adopts during the ensuing shell fight. This is a critical decision as only the attacker might stand a chance of eventually selecting between the two shells, if it evicts the defender. The attacker role is usually taken by the larger crab and when the smaller crab takes that role the relative size difference between the crabs is small. Our results show that prefight displays influence this decision. High levels of cheliped presentation by the large crab predicted that the large crab would be the attacker, whereas high levels of cheliped extension by the small crab appeared to have the opposite effect. These two cheliped displays are very different. When the cheliped is presented it is held stationary in a posture with the main chelate segment (propodus and dactyl) held in such a way that allows the opponent a clear view. The PCA showed that both animals often present the cheliped. This display presumably allows the opponents to assess which has the larger chelipeds and it fits the description of an index (sensu Maynard Smith & Harper 2003). In contrast, cheliped extension involves a rapid thrust of the pointed ends of the chelipeds towards
the opponent and thus seems less likely to afford a clear assessment of cheliped size. Although it can be performed simultaneously by both contestants, the extension display of the larger and smaller crabs loaded on to different PCA factors suggesting that it is not normally a mutual display. This display might be used to bluff or confuse the opponent (sensu Riechert 1998) and, if so, it should be favoured by the smaller opponent. In contrast, if the presentation display is honest it should be favoured by the larger opponent. The results were clear in that the extension display was used more by the smaller crab and the presentation display more by the larger crab. Using models, Hazlett (1968) found little differentiation between cheliped presentation and extension displays but noted that stationary models did not contain the essential movement seen in the latter. Observations of interacting crabs indicated that cheliped extension was more aggressive and more likely to cause the opponent to retreat than was cheliped presentation (Hazlett 1968). The evidence here suggests that cheliped extension does not provide honest information. The next key decision is whether the defender releases its grip on the internal structure of the shell and allows itself to be pulled out by the attacker (Elwood & Neil 1992) and the displays involving the chelipeds early in the encounter are associated with this decision. Again there was a clear benefit to the small crab (defender) from showing the cheliped extension display as high levels of this display were associated with a lower probability of being evicted. Cheliped presentation by the smaller crab, however, had the opposite effect as this was associated with a greater chance of eviction. This is typically a mutual display but is performed for longer by the larger opponent. This display again seems to be an honest signal that enables the larger attacker to assess clearly that it is the larger of the two and thus its fight motivation may be increased. Thus in the two key fight decisions that determine the fitness pay-off from the contest, cheliped presentation favours the larger contestant and cheliped extension the smaller opponent. The motivation of the large crab to fight is predicted to vary with the potential gain from taking the opponent’s shell. Large crabs in the G50 group had more to gain than those in the other two groups, their shells being deficient in both species and size whereas the other groups were deficient in only one of these parameters. This larger potential gain is reflected in the greater number of fights initiated by the large crab, particularly by those in the G50 group. These fights in the G50 group were also more likely to result in eviction than were those in the other two groups. Since there was no difference in RWD between the groups this suggests that rather than being caused by a difference in fighting ability, the difference between groups was due to an elevated motivation of the large crabs in this group. Thus, willingness to enter the escalated fight and willingness to continue until the defender is evicted are both influenced by the gain to the larger crab. The startle response provides an unambiguous measure of the attacker’s motivation (Elwood et al. 1998; Briffa & Elwood 2001b) and this was significantly shorter in the G50 group than in the other groups. Startle responses have been used as motivational probes in
857
858
ANIMAL BEHAVIOUR, 72, 4
a variety of other situations (Culshaw & Broom 1980; Moorehouse et al. 1987; Dill & Gillett 1991). Short startle responses indicate a high motivation to resume the activity and thus the short startle responses in the G50 group clearly reflect the high motivation of that group. If displays by the smaller crab affect the outcome of the fight (as noted above) we might reasonably expect effects of those displays on the attacker’s motivation (startle response), but the only prefight activities that seemed to predict the attacker’s motivation were performed by the attacker. High levels of the large crab presenting its own chelae predicted high motivation and may simply reflect an existing high motivation. There were, however, two other prefight activities of the attacker that predicted the attacker’s motivation, these being approach and ambulatory raise. These both loaded on to PC1 along with high posture by both crabs and extension of the chelipeds by the small crab. One might predict that approach in the prefight stage would indicate a high motivation rather than the low motivation found here. One would also predict more approach by the larger crab but this was not found. It seems that when the larger crab shows high levels of approach this is the result of repeated approaches to which the opponent responds by displays rather than by assuming the defender role. Perhaps highly motivated attackers approach more decisively and thus are more likely to start the fight on the first attempt. Alternatively, the necessity to repeat attempts to initiate the fight proper might reduce the larger crab’s motivation. In contrast to approach, ambulatory raise by the attacker predicted high attacker motivation. This is unexpected as ambulatory raise and approach of the larger crab both loaded positively in the PCA and the function of this display is not clear (see below). The final measure of motivation is for the defender, taken as the duration for which it resists before giving up (Hack et al. 1997; Bridge et al. 2000). However, because fight duration is reduced if the attacker raps with high vigour (Briffa & Elwood 2002) and only defenders receiving weak raps manage to resist eviction (Briffa & Elwood 2002; Briffa et al. 2003) this measure is confounded by the abilities of the attacker. Cheliped displays did not influence contest duration but mutual grappling did. Grappling presumably denotes a reluctance of the smaller crab to assume the defender role. This might occur if the smaller crab assesses that the opponent is weak and, because weak attackers cannot fight with high vigour (Briffa & Elwood 2005), the defender may resist for longer. There was also a positive effect of the ambulatory raise by the attacker on the duration of resistance by the defender but the reason for this is not clear. Hazlett (1968) found that models of ambulatory raise had little or no effect on the receiver. The general conclusion is that cheliped presentation, often by both animals but for longer by the larger crab, affords a clear view of the chelipeds and appears to be an honest signal, the effect on the recipient depending on whether the chelipeds being presented are larger or smaller than those of the recipient. This display is associated with increased chances of the larger crab
becoming the attacker and winning the fight and higher attacker motivation. Cheliped extension, in contrast, seems to contain an element of bluff and is performed more by the smaller crab. When used extensively by the smaller crab it reduces the probability of the larger crab assuming the role of attacker and, if it does take that role, decreases the chances of the attacker winning the contest. We should consider, however, whether cheliped extension by the defender is used as an honest signal that indicates future resistance to eviction. If this is the case, we have a clear prediction that there should be a positive relation between duration of resistance and use of the extension display; however, this was not found. A second problem concerns the apparent effects of this display on the decision of the larger animal to assume the defender role instead of the more normal attacker role. There is no obvious reason why this would be an alternative to having to fight for longer, as it is only in the attacker role that it might take its choice of shells. The results suggest that rather than providing honest information, cheliped extension makes size assessment by the opponent more difficult (Riechert 1998) and is particularly effective in causing the larger animal to take the defender role when the opponents are close in body size. It is congruent with the idea of false signals being used when there is an advantage in avoiding or even winning a fight, which, if the signal does not deter the opponent, the sender will lose. That is, if the cost of the signal is low then the only retribution will be similar to that if the signal was not used (Adams & Mesterton-Gibbons 1995; Johnstone 1998). Thus the prefight period in hermit crabs is complex, involving a number of distinct displays that appear to have different effects and also reflect differing motivational states. Within this complexity there is scope for both honesty and manipulation (Johnstone 1998; Riechert 1998). Acknowledgments We are grateful to the BBSRC for funding this research and thank Mark Laidre for comments on the manuscript.
References Adams, E. S. & Caldwell, R. L. 1990. Deceptive communication in asymmetric fights of the stomatopod crustacean Gonodactylus bredini. Animal Behaviour, 39, 706e716. Adams, E. S. & Mesterton-Gibbons, M. 1995. The cost of threat displays and the stability of deceptive communication. Journal of Theoretical Biology, 175, 405e421. Backwell, P. R. Y., Christy, J. H., Telford, S. R., Jennions, M. D. & Passmore, N. I. 2000. Dishonest signalling in a fiddler crab. Proceedings of the Royal Society of London, Series B, 267, 719e724. Bridge, A. P., Elwood, R. W. & Dick, J. T. A. 2000. Imperfect assessment and limited information preclude optimal strategies in maleemale fights in the orb-weaving spider Metellina mengei. Proceedings of the Royal Society of London, Series B, 267, 273e279. Briffa, M. & Elwood, R. W. 2000. Cumulative or sequential assessment during hermit crab shell fights: effects of oxygen on decision rules. Proceedings of the Royal Society of London, Series B, 267, 2445e2452.
ELWOOD ET AL.: DISPLAYS AND FIGHT DECISIONS
Briffa, M. & Elwood, R. W. 2001a. Decision rules, energy metabolism and vigor of hermit crab fights. Proceedings of the Royal Society of London, Series B, 268, 1841e1848. Briffa, M. & Elwood, R. W. 2001b. Changes in motivational state during hermit crab shell fights. Animal Behaviour, 62, 505e510. Briffa, M. & Elwood, R. W. 2002. Power of shell-rapping signals influences physiological costs and subsequent decisions during hermit crab shell fights. Proceedings of the Royal Society of London, Series B, 269, 2331e2336. Briffa, M. & Elwood, R. W. 2003. Change in the power of shell rapping during shell fights between hermit crabs: does magnitude contribute to repeated signals? Behavioral Ecology, 14, 60e65. Briffa, M. & Elwood, R. W. 2004. Use of energy reserves in fighting hermit crabs. Proceedings of the Royal Society of London, Series B, 271, 373e379. Briffa, M. & Elwood, R. W. 2005. Rapid change in energy status in fighting animals: causes and effects of strategic decisions. Animal Behaviour, 70, 119e124. Briffa, M., Elwood, R. W. & Russ, J. M. 2003. Change in the power of rapping during shell fights between hermit crabs: does magnitude contribute to repeated signals? Behavioral Ecology, 14, 60e65. Clutton-Brock, T. H. & Albon, S. D. 1979. The roaring of red deer and the evolution of honest advertisement. Behaviour, 69, 145e170. Culshaw, A. D. & Broom, D. M. 1980. The imminence of behavioural change and the startle response of chicks. Behaviour, 73, 64e76. Deag, J. M. & Scott, G. W. 1999. ‘Conventional’ signals in avian agonistic displays: integrating theory, data and different levels of analysis. Journal of Theoretical Biology, 196, 155e162. Dill, L. M. & Gillett, J. F. 1991. The economic logic of the barnacle Balanus glandula (Darwin) hiding behaviour. Journal of Experimental Marine Biology and Ecology, 153, 115e127. Dowds, B. M. & Elwood, R. W. 1983. Shell wars: assessment strategies and the timing of decisions in hermit crab shell fights. Behaviour, 85, 1e24. Elwood, R. W. & Neil, S. J. 1992. Assessments and Decisions: a Study of Information Gathering by Hermit Crabs. London: Chapman & Hall. Elwood, R. W., McClean, A. & Webb, L. 1979. The development of shell preferences by the hermit crab Pagurus bernhardus. Animal Behaviour, 27, 940e946. Elwood, R. W., Marks, N. & Dick, J. T. A. 1995. Consequences of shell species preferences for female reproductive success in the hermit crab Pagurus bernhardus. Marine Biology, 123, 431e434. Elwood, R. W., Wood, K. E., Gallagher, M. & Dick, J. T. A. 1998. Probing motivational state during agonistic encounters in animals. Nature, 393, 66e68. Grafen, A. 1990. Biological signals as handicaps. Journal of Theoretical Biology, 144, 517e546. Hack, M. A., Thompson, J. D. & Fernandes, D. M. 1997. Fighting in males of the autumn spider, Metellina segmentata: the effects of relative body size, prior residency and female value on contest outcome and duration. Ethology, 103, 488e498.
Hardy, I. C. W. & Field, S. A. 1998. Logistic regression analysis of animal contests. Animal Behaviour, 56, 787e792. Hazlett, B. A. 1968. Communicatory effect of body position in Pagurus bernhardus (L.) (Decapoda, Anomura). Crustaceana, 14, 210e 214. Hazlett, B. A. & Bossert, W. H. 1965. A statistical analysis of the aggressive communications systems of some hermit crabs. Animal Behaviour, 13, 357e373. Hughes, M. 2000. Deception with honest signals: signal residuals and signal function in snapping shrimp. Behavioral Ecology, 11, 614e623. Hurd, P. L. 1997. Cooperative signalling between opponents in fish fights. Animal Behaviour, 54, 1309e1315. Hurd, P. L. & Enquist, M. 2005. A strategic taxonomy of biological communication. Animal Behaviour, 70, 1155e1170. Hurd, P. L. & Ydenberg, R. C. 1996. Calculating the ESS level of information transfer in aggressive communication. Evolutionary Ecology, 10, 221e232. Hyatt, G. W. & Salmon, M. 1979. Comparative statistical and information analysis of combat in the fiddler crabs, Uca pugilator and U. pugnax. Behaviour, 68, 1e23. Jackson, N. W. 1988. Information gathering and decision making during shell selection by the hermit crab Pagurus bernhardus. Ph.D. thesis, Queen’s University of Belfast. Johnstone, R. A. 1998. Game theory and communication. In: Game Theory and Animal Behaviour (Ed. by L. A. Dugatkin & H. K. Reeve), pp. 94e117. New York: Oxford University Press. Keeley, E. R. & Grant, J. W. A. 1993. Visual information, resource value and sequential assessment in convict cichlid (Cichlasoma nigrofasciatum) contests. Behavioral Ecology, 4, 345e349. Krebs, J. R. & Dawkins, R. 1984. Animal signals: mind reading and manipulation. In: Behavioural Ecology: An Evolutional Approach, 2nd edn. (Ed. by J. R. Krebs & N. B. Davies), pp. 380e402. Oxford: Blackwell Scientific. Matsumasa, M. & Murai, M. 2005. Changes in blood glucose and lactate levels of male fiddler crabs: effects of aggression and claw waving. Animal Behaviour, 69, 569e577. Maynard Smith, J. & Harper, D. 2003. Animal Signals. Oxford: Oxford University Press. Moorehouse, J. E., Fosbrooke, I. H. M. & Ludlow, A. R. 1987. Stopping a walking locust with sound: an analysis of variation in behavioural threshold. Experimental Biology, 46, 193e201. Neil, S. J. & Elwood, R. W. 1985. Behavioural modification during egg-brooding in the hermit crab, Pagurus bernhardus L. Journal of Experimental Marine Biology and Ecology, 94, 99e114. Parker, G. A. 1974. Assessment strategy and the evolution of fighting behaviour. Journal of Theoretical Biology, 47, 223e243. Riechert, S. E. 1998. Game theory and communication. In: Game Theory and Animal Behaviour (Ed. by L. A. Dugatkin & H. K. Reeve), pp. 64e93. New York: Oxford University Press. Rubenstein, D. I. & Hazlett, B. A. 1974. Examination of the agonistic behaviour of the crayfish Orconectes virilis by character analysis. Behaviour, 50, 193e216. Zahavi, A. 1975. Mate selection: a selection for a handicap. Journal of Theoretical Biology, 67, 205e214.
859
Honest and dishonest displays, motivational state and subsequent decisions in hermit crab shell fights R. W. ELW OOD, R. M. E. POTH AN IK AT & M. B RIF F A
School of Biological Sciences, The Queen’s University of Belfast (Received 22 July 2005; initial acceptance 7 September 2005; final acceptance 24 January 2006; published online 17 August 2006; MS. number: 8625R)
Animal fights are typically preceded by displays and there is debate whether these are always honest. We investigated the prefight period in hermit crabs, Pagurus bernhardus, during which up to four types of display plus other activities that might provide information are performed. We determined how each display influences or predicts various fight decisions, and related these displays to the motivational state of the attacker, as determined by a startle response, and of the motivational state of the defender, as determined by the duration for which it resisted eviction from its shell. Two displays appeared to have consistent but different effects. Cheliped presentation, where the claws were held in a stationary position, often by both crabs but for longer by the larger, seemed to be honest, and allowed for mutual size assessment. This display enhanced the motivation and the success of the larger crab. In contrast, cheliped extension, involving the rapid thrust of the open chelae towards the opponent, did not seem to allow for mutual size assessment and may contain an element of bluff. It was performed more by the smaller crab and enhanced its success. The complexity of displays in this species appears to allow for both honesty and manipulation. 2006 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
Contests in animals commonly involve the use of displays. These may be relatively low-intensity activities at the start of an encounter but may escalate as the contest progresses. In many species, both contestants use the same type of display, but may differ in the vigour or magnitude of performance. This variation may be used in settling the outcome of the contest without recourse to physical combat, presumably because it provides honest and reliable mutual evaluation of fighting ability or resource-holding potential (Parker 1974; Hughes 2000; Maynard Smith & Harper 2003; Hurd & Enquist 2005). Examples are found in the roaring contests of red deer, Cervus elaphus (Clutton-Brock & Albon 1979) and ‘lateral displays’ in cichlid fish, Cichlasoma nigrofasciatum (Keeley & Grant 1993). Other species have several distinct displays and those selected may vary between contestants and, within contestants, between encounters. The function of these displays is less clear but there seems to be scope for deception and manipulation of the opponent (Krebs & Dawkins 1984). Stomatopods, Gonodactylus bredini, for
Correspondence: R. Elwood, School of Biological Sciences, The Queen’s University of Belfast, Belfast, BT9 7BL, U.K. (email: [email protected]. uk). M. Briffa is now at the Marine Biology and Ecology Research Centre, School of Biological Sciences, University of Plymouth, Drake Circus, Plymouth PLA 8AA, U.K. 0003e 3472/06/$30.00/0
example, use various displays including the ‘meral spread’ (Adams & Caldwell 1990), which is typically shown by aggressive individuals and may cause the opponent to retreat. It may, however, be shown by newly moulted individuals, which have a low fighting ability, as a bluff. Also, the ‘open chela display’ of snapping shrimps, Alpheus heterochaelis, is used more often by those individuals that have larger chelae than predicted by body size and the display exaggerates their true fighting ability (Hughes 2000). In male fiddler crabs, Uca annulipes, recently regenerated major chelae are thinner and lighter and hence less energy is required when they are used in displays than in displays of original claws. Males with recently regenerated chelae can bluff fighting ability as it is impossible for the opponent to determine claw mass by means of visual assessment (Backwell et al. 2000). Theory suggested that deceptive displays by the weaker opponent should be rare and that on average displays should be honest (Johnstone 1998), although Backwell et al. (2000) suggested that cheating may be more common than was previously thought on the basis of the evidence from fiddler crabs. In either situation, displays clearly have the potential to change the behaviour of the opponent (Rubenstein & Hazlett 1974; Hyatt & Salmon 1979). In the case of ‘honest displays’ the advantage to the sender is that the displays may settle the contest without costly combat, even if the sender loses the interaction (Hurd 1997; Deag & Scott
853 2006 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
854
ANIMAL BEHAVIOUR, 72, 4
1999). In the case of manipulation or deception, however, the advantage to the sender is that the displays could mask or exaggerate the true fighting ability of the sender and reduce the receiver’s motivation to fight (Hurd & Ydenberg 1996). The receiver’s motivation may easily be detected if the receiver immediately retreats from a weak opponent after that opponent has used a display typically given by an individual of high fighting ability (Hughes 2000). Regardless of the honesty of the signal a fight might still occur. In this case the motivation of the opponent clearly has not been reduced sufficiently to induce retreat. However, causing a slight reduction in the opponent’s motivation to fight, while not preventing a fight, might yet alter the probability of eventual victory. For instance, the time or energy the opponent is prepared to commit may be reduced. Even a slight increase in the probability of victory may make performing the display worthwhile. The benefit to a weak sender of a false signal may thus be great because it might avoid a fight that it would otherwise lose. If the signal does not deter the opponent then the sender is in the same position as if it had not sent the false signal. Thus, if the cost of sending the false signal is low and there is no specific retribution from the receiver then the use of false signals will be selected for (Adams & Mesterton-Gibbons 1995; Johnstone 1998). We investigated the use of various prefight displays in hermit crabs, Pagurus bernhardus, contesting ownership of shells. These displays use the chelipeds and walking legs and also a high posture in which the shell is lifted high off the substrate (Hazlett 1968; Elwood & Neil 1992). However, other activities not normally described as displays may convey important information. These include mutual grappling, approaching the opponent and retreat from the opponent. The prefight phase may be followed by an escalated fight in which one crab, termed the ‘attacker’, initiates the shell fight by grabbing the shell of the ‘defender’, causing the defender to withdraw into its shell (Dowds & Elwood 1983). The attacker may then engage in repeated bouts of vigorous shell rapping in which the attacker hits its shell upon that of the defender, until either the defender is evicted from the shell, enabling the attacker to take that shell, or the attacker gives up. The effect of the power, number of raps per bout, number of bouts and the duration of pauses between bouts on the physiology and resistance of the defender, and the interplay between the physiology and the performance of rapping of the attacker, have been the subject of recent investigation (Briffa & Elwood 2000, 2001a, 2002, 2003, 2005). This energetically demanding rapping is difficult to fake as only animals in good condition (with low lactate) are able to produce high-power vigorous rapping and only these crabs are guaranteed victory (Briffa & Elwood 2001a, 2002, 2005). In contrast, the relatively brief postures seen in prefight displays are unlikely to be energetically costly and thus the condition of the animal is thus less likely to limit their use. Studies on hermit crab prefight displays have used models (e.g. Hazlett 1968) and information theory (Hazlett & Bossert 1965) in which the immediate effects of a display on the behaviour of the receiver were noted. Our aim in the present study, however, was to determine longer-term
consequences of prefight displays. First, we asked whether prefight displays, and other activities, influence or predict which crab takes the role of attacker. This is an important decision as it is only the attacker that, if it evicts the defender, is able to choose which of the two shells to occupy. Second, we examined whether prefight activities influence or predict the winner of the contest. Third, we aimed to determine whether displays influence the motivation of the attackers (but not that of the defenders) by presenting a novel, startling stimulus and recording the duration of the attacker’s startle response (Elwood et al. 1998), which is inversely related to the level of motivation. Attackers with a high potential gain in shell quality and low probable costs have short startle responses (Elwood et al. 1998) and highly motivated attackers are more likely to win (Briffa & Elwood 2001b). Finally, we examined how these prefight activities influence the duration for which the defender resists eviction. Again, this should elucidate relations between the prefight displays and motivation because the duration of resistance is a common measure of motivation in losers of fights (Hack et al. 1997; Bridge et al. 2000). Our key aim was thus to determine how various displays by one crab influence the behaviour or motivation of the receiver and if those displays provide honest information or a means of manipulation.
METHODS Small (0.10e0.64 g) littoral specimens of the common European hermit crab were collected weekly from Ballywalter, Co. Down, Northern Ireland, U.K. between October 2004 and February 2005. They were held in groups of 40e50 in plastic tanks (60 30 cm) filled with aerated seawater at 12 C to a depth of 10 cm, and fed ad libitum on commercial fish food (catfish pellets). We removed the crabs from their shells by cracking the shells open in a bench vice. We used only males for staging encounters. We gave females new shells and returned them to the sea, thus avoiding sex differences in behaviour that have been noted in previous studies (Neil & Elwood 1985). Only male crabs that were free from obvious parasites, loss of appendages and recent moult were used. Use of the vice did not harm the crabs. We allocated male crabs to pairs and pairs to one of three groups. Each crab was weighed and the relative weight difference (RWD) of the pair was calculated by RWD ¼ 1 (small crab weight/large crab weight). The relative weight difference ranged from 0.05 to 0.58 (X SE ¼ 0:22 0:008) and there was no difference in RWD between the three groups. To determine the preferred weight of shell for the larger crab of each pair we used previously calculated regression lines that relate crab weight to preferred shell weight (Jackson 1988). In all cases the smaller crab received a Littorina obtusata shell that was 100% of the preferred weight for the larger crab. In group 1 (L50), the larger crab received an L. obtusata shell that was 50% of its preferred size. In group 2 (G50), the larger crab of each pair received a shell of Gibbula cineraria that was 50% of its preferred shell weight. This species is normally avoided by hermit crabs if given
ELWOOD ET AL.: DISPLAYS AND FIGHT DECISIONS
a choice (Elwood et al. 1979) and those occupying the species have reduced reproductive output compared to those in L. obtusata (Elwood et al. 1995). In group 3 (G100), the larger crab received a G. cineraria shell that was 100% of its preferred shell size. Each crab was isolated with its new shell in a crystallizing dish 10 cm in diameter, containing aerated sea water at 12 C and a fine layer of sand, for approximately 16 h prior to the fight being staged. The fight arena was a glass tank (15 7 cm and 7 cm high) that contained aerated sea water and a layer of sand 1 cm deep to provide adequate grip for the walking legs of the crabs during the fight. We separated the crabs by placing them inside two clear plastic cylinders (3.7 cm in diameter and 5.5 cm high), which were positioned such that the cylinders were touching, enabling visual contact. The arena was placed behind a one-way mirror such that the crabs could not see the observer. To record the behaviour of large and small crabs we used a Psion Workabout hand-held computer configured as a timeeevent recorder with The Observer 3.0 software (Noldus Technology, Wageningen, The Netherlands). The two crabs were placed into the cylinders and allowed 2 min to settle before we removed the cylinders and started the observation. We recorded display and other activities until an ‘attacker’ initiated the fight causing a ‘defender’ to withdraw into its shell. At this point a novel stimulus was applied to elicit a startle response. The stimulus was a weighted (total weight ¼ 180.25 g) rectangle of opaque Perspex (19.5 9.5 cm), which was suspended on a thread 30 cm above the top of the arena. This was released in such a way that it fell on to the top of the arena (the width of the Perspex was greater than that of the arena, so it did not fall into the water but came to rest on the rim) and was then pulled back to its original position. This technique is known to cause a startle response in fighting hermit crabs (Elwood et al. 1998; Briffa & Elwood 2001b), which involves a temporary cessation of activity while the attacker withdraws either partially or completely into its shell. Since defenders withdraw into their shells when the fight starts we could measure the startle response only for attackers. The duration of the startle response was defined as lasting from the application of the stimulus to resumption of activity by the attacker. After the crabs recovered from the startle response, interactions were allowed to continue until the attacker either evicted the defender or gave up without effecting an eviction or, if a fight was not initiated, after 30 min. Prior to a fight being initiated, behavioural recording focused on prefight displays involving the asymmetrically sized chelipeds (the first pair of pereiopods that bear the claws) and the walking legs (second and third pairs of pereiopods; Hazlett 1968). ‘Cheliped presentation’ involves the proximal part of the chelipeds being held forwards, towards the opponent, with the distal part (claw) being perpendicular to the substrate. In ‘cheliped extension’ the claws are held horizontal to the substrate and raised up at least to the level of the head of the displayer, typically with the chelae open. Diagrams of the key displays can be seen in Hazlett (1968) and photographs in Elwood & Neil (1992). ‘Grapple’ involves mutual wrestling
with the chelipeds and walking legs. ‘High posture’ involves the walking legs and chelipeds being used to raise the crab’s shell off the substrate. ‘Ambulatory raise’ involves one or more of the walking legs being raised off and held perpendicular to the substrate. ‘Approach’ involves movement towards the opponent, while ‘retreat’ is movement away from the opponent. During the subsequent shell fight we also recorded the pattern of shell rapping. We examined categorical data either by G tests for contingency analysis or by c2 for goodness-of-fit tests. As the latency from the start of the observation to the start of the shell fight (when the novel stimulus was applied) varied, we used the proportion of that time for which each activity occurred. However, analyses using absolute times did not differ markedly. Furthermore, analysis of prefight duration showed no significant effect on the dependent variables investigated. Principal components analysis (PCA) was used to determine which prefight activities tended to occur together in the same encounters. However, because we wished to examine the effects of particular displays we did not use PCA scores in further analyses. Thus we used the behavioural categories in multivariate logistic regression when the dependent variable was dichotomous. When the dependent variable was continuous we used stepwise regression. Paired t tests were used to compare the large and small crabs to determine which performed each of the displays more than the other. Data were log transformed as appropriate.
RESULTS Of the total of 240 observations there was significant variation between the groups in the type of interaction (fight initiated by large or small crab or no fight; G tests: G4 ¼ 15.14, P ¼ 0.004) because of the large number of fights initiated by the large crab in the G50 group and the low number without fights in this same group. When just observations of fights were considered (N ¼ 148), significant variation between the three experimental groups in the number of fights initiated by the large and small crabs was noted (G2 ¼ 6.51, P ¼ 0.04). More were initiated by the large crab in the G50 group than in the L50 group (G1 ¼ 4.15, P ¼ 0.04) and in the G50 group than in the G100 group (G1 ¼ 5.76, P ¼ 0.02). There was, however, no significant difference between the L50 group and G100 group (G1¼0.042, NS). Of the 148 fights, more were initiated by the large crab (N ¼ 131) than by the small crab (N ¼ 17; goodness-of-fit test: c21 ¼ 140:2, P < 0.0001) (Table 1). When fights were initiated by the small crab there was less of a relative weight difference between the opponents (paired t test: t146 ¼ 2.26, P ¼ 0.025). The PCA on the 13 activities produced six component factors. Here we consider those activities with loadings of >0.5: PC1 comprised high posture by both crabs, ambulatory raise and approach by the large crab and cheliped extension by the small crab; PC2 comprised cheliped presentation by both crabs; PC3 comprised retreat by the large and approach by the small crab; PC4 comprised
855
ANIMAL BEHAVIOUR, 72, 4
Table 1. Numbers of observations where fights were initiated by the large crab or the small crab and where no fights occurred and the number of evictions and nonevictions when the fight was initiated by the large crab Initiation
Group
large crab
L50 (80) 33 G50 (80) 54 G100 (80) 44 Total (240) 131
small crab
No fight
Evictions
Nonevictions
6 2 9 17
41 24 27 92
16 44 36 96
17 10 8 35
Sample sizes are given in parentheses. The larger crab of each pair was in a Littorina obtusata shell that was 50% of its preferred size (L50), a Gibbula cineraria shell that was 50% of its preferred size (G50) or a G. cineraria shell that was 100% of its preferred size (G100).
small crab retreat; PC5 comprised grappling; PC6 comprised cheliped extension by the large crab and ambulatory raise by the small crab. Preliminary analyses indicated that there were no significant group effects on the individual prefight activities, so we explored initiator effects with multivariate logistic regression. This is a useful technique for analysing the effects of behavioural parameters on the likelihood of alternative events occurring during contests (Hardy & Field 1998). The 13 prefight activities, expressed as a percentage of the prefight duration, were the independent variables and the dependent variable was the large crab rather than the small crab taking the attacker role. This indicated that cheliped presentation by the large crab increased the likelihood of the large crab initiating the fight (Wald c21 ¼ 3:92, P ¼ 0.048) whereas extending the cheliped by the small crab had a negative effect on the likelihood of the large crab initiating the fight (Wald c21 ¼ 4:717, P ¼ 0.03). The relation between the 13 prefight activities and RWD was examined by stepwise regression but the regression was not significant. Paired t tests showed that the smaller crabs performed more cheliped extension than did the larger (t147 ¼ 4.685, P < 0.0001) and more retreating (t147 ¼ 2.289, P ¼ 0.024) but spent less time presenting chelipeds (t147 ¼ 1.98, P ¼ 0.05). There were no differences between the crabs in approach, high posture or ambulatory raise. Of the 131 fights initiated by the large crab, 96 resulted in an eviction of the small crab (Table 1). There was no significant difference in the relative weight, small crab weight or large crab weight between those fights in which eviction did and did not occur. There was a significant difference between the three groups in the number that resulted in eviction (G test: G2 ¼ 12.88, P ¼ 0.002). There were more evictions in the G50 group than in the L50 group (G1 ¼ 10.3, P ¼ 0.001) and in the G100 group than in the L50 group (G1 ¼ 9.56, P ¼ 0.002), but there was no significant difference between the G50 and the G100 groups (G1 ¼ 0.002, P ¼ 0.966). For the 131 fights initiated by the large crab, ANOVA showed a significant group effect for startle duration (log), with the G50 group showing the shortest responses (F2,125 ¼ 17.06, P < 0.0001). There was no outcome effect
but there was a significant interaction effect (F2,125 ¼ 4.80, P ¼ 0.01; Fig. 1), because the group effects were more marked in cases of noneviction. There were no significant main or interaction effects for absolute or relative weights or the number of bouts of shell rapping. Further analyses revealed no group effects for the 13 prefight activities and their possible effect on the defender resisting eviction was investigated by multiple logistic regression. This indicated that as the duration of cheliped extension by the small crab before the fight increased, the chance of eventually being evicted declined (Wald c2 ¼ 7.952, P ¼ 0.005) Use of cheliped presentation by the small crab had the opposite effect, increasing the likelihood of an eviction occurring (Wald c21 ¼ 4:063, P ¼ 0.043). We used stepwise regression to assess which of the 13 prefight activities might influence or predict the attacker’s startle duration for the 96 fights that ended with eviction. Three activities entered the regression model (F3,92 ¼ 10.417, P < 0.0001): large crab approach had a positive effect on the startle duration, but large crab presenting chelipeds and ambulatory raise both had a negative effect on the startle duration. A further stepwise regression was used to examine the effect of the same independent variables on the number of bouts of shell rapping required to evict the defender. The duration of the startle response was used as an additional independent variable to assess how the motivation of the attacker might influence the resistance of the defender. Two activities entered the regression model (F2,93 ¼ 8.098, P < 0.001), these being positive effects of grappling and of the large crab showing ambulatory raise but neither startle duration nor cheliped displays entered the model.
DISCUSSION Reliable signals of resource-holding potential (RHP) are expected to evolve if they reduce the need for costly combat to settle disputes (Parker 1974; Johnstone 1998; 1.2
Eviction No eviction
1 Log startle
856
0.8 0.6 0.4 0.2 0
L50
G50
G100
Figure 1. Log duration of startle responses (X SE) for the three experimental groups and for each outcome (eviction or no eviction). The larger crab of each pair was in a Littorina obtusata shell that was 50% of its preferred size (L50), a Gibbula cineraria shall that was 50% of its preferred size (G50) or a G. cineraria shell that was 100% of its preferred size (G100).
ELWOOD ET AL.: DISPLAYS AND FIGHT DECISIONS
Hughes 2000). These signals of RHP are often assumed to be costly, not easily faked and, therefore, essentially ‘honest’ (Zahavi 1975; Grafen 1990; Hughes 2000). The rationale is that easily faked signals would spread in the population if they benefited the sender and thus cease to predict true RHP. However, noncostly signals may also be honest if they act as an index of an animal’s size or condition (Maynard Smith & Harper 2003). Occasionally certain activities may be used to ‘bluff’ or exaggerate actual fighting ability or intent, particularly if the animal that is bluffing is the weaker of the two contestants (Adams & Caldwell 1990; Hughes 2000). Furthermore, the use of a variety of signals may introduce a complexity that confuses the receiver and makes the true assessment of RHP difficult (Riechert 1998). In the escalated shell fights of hermit crabs, repeated bouts of vigorous shell rapping by the attacker are energetically demanding i.e. costly, and the power of the signal is constrained by the physiological state of the sender. Shell rapping thus seems to be an honest signal (Briffa & Elwood 2001a, 2002, 2004, 2005) and an index of condition (Maynard Smith & Harper 2003). However, this may not be the case with the prefight displays of hermit crabs, which may involve relatively brief movements of chelipeds and walking legs. Clearly it is not easy to fake the size of a cheliped and a cheliped that is larger than normal for a particular body size will be expensive to produce. Not only will it require resources to be channelled into that appendage, but the appendage may also subsequently not be efficient in foraging. Furthermore, prolonged waving of the very large claw in fiddler crabs has high energetic costs (Matsumasa & Murai 2005). However, the relatively brief displays of ‘normalsized’ chelae of hermit crabs may not incur significant energetic costs. The general conclusion is that displays of different types in different species may differ with respect to their cost and honesty and thus differ in their effects on tactical decisions during contests (Johnstone 1998; Maynard Smith & Harper 2003). The first key decision that hermit crabs make in a potential shell fight is whether to initiate an attack. This will determine the role, attacker or defender, that each crab adopts during the ensuing shell fight. This is a critical decision as only the attacker might stand a chance of eventually selecting between the two shells, if it evicts the defender. The attacker role is usually taken by the larger crab and when the smaller crab takes that role the relative size difference between the crabs is small. Our results show that prefight displays influence this decision. High levels of cheliped presentation by the large crab predicted that the large crab would be the attacker, whereas high levels of cheliped extension by the small crab appeared to have the opposite effect. These two cheliped displays are very different. When the cheliped is presented it is held stationary in a posture with the main chelate segment (propodus and dactyl) held in such a way that allows the opponent a clear view. The PCA showed that both animals often present the cheliped. This display presumably allows the opponents to assess which has the larger chelipeds and it fits the description of an index (sensu Maynard Smith & Harper 2003). In contrast, cheliped extension involves a rapid thrust of the pointed ends of the chelipeds towards
the opponent and thus seems less likely to afford a clear assessment of cheliped size. Although it can be performed simultaneously by both contestants, the extension display of the larger and smaller crabs loaded on to different PCA factors suggesting that it is not normally a mutual display. This display might be used to bluff or confuse the opponent (sensu Riechert 1998) and, if so, it should be favoured by the smaller opponent. In contrast, if the presentation display is honest it should be favoured by the larger opponent. The results were clear in that the extension display was used more by the smaller crab and the presentation display more by the larger crab. Using models, Hazlett (1968) found little differentiation between cheliped presentation and extension displays but noted that stationary models did not contain the essential movement seen in the latter. Observations of interacting crabs indicated that cheliped extension was more aggressive and more likely to cause the opponent to retreat than was cheliped presentation (Hazlett 1968). The evidence here suggests that cheliped extension does not provide honest information. The next key decision is whether the defender releases its grip on the internal structure of the shell and allows itself to be pulled out by the attacker (Elwood & Neil 1992) and the displays involving the chelipeds early in the encounter are associated with this decision. Again there was a clear benefit to the small crab (defender) from showing the cheliped extension display as high levels of this display were associated with a lower probability of being evicted. Cheliped presentation by the smaller crab, however, had the opposite effect as this was associated with a greater chance of eviction. This is typically a mutual display but is performed for longer by the larger opponent. This display again seems to be an honest signal that enables the larger attacker to assess clearly that it is the larger of the two and thus its fight motivation may be increased. Thus in the two key fight decisions that determine the fitness pay-off from the contest, cheliped presentation favours the larger contestant and cheliped extension the smaller opponent. The motivation of the large crab to fight is predicted to vary with the potential gain from taking the opponent’s shell. Large crabs in the G50 group had more to gain than those in the other two groups, their shells being deficient in both species and size whereas the other groups were deficient in only one of these parameters. This larger potential gain is reflected in the greater number of fights initiated by the large crab, particularly by those in the G50 group. These fights in the G50 group were also more likely to result in eviction than were those in the other two groups. Since there was no difference in RWD between the groups this suggests that rather than being caused by a difference in fighting ability, the difference between groups was due to an elevated motivation of the large crabs in this group. Thus, willingness to enter the escalated fight and willingness to continue until the defender is evicted are both influenced by the gain to the larger crab. The startle response provides an unambiguous measure of the attacker’s motivation (Elwood et al. 1998; Briffa & Elwood 2001b) and this was significantly shorter in the G50 group than in the other groups. Startle responses have been used as motivational probes in
857
858
ANIMAL BEHAVIOUR, 72, 4
a variety of other situations (Culshaw & Broom 1980; Moorehouse et al. 1987; Dill & Gillett 1991). Short startle responses indicate a high motivation to resume the activity and thus the short startle responses in the G50 group clearly reflect the high motivation of that group. If displays by the smaller crab affect the outcome of the fight (as noted above) we might reasonably expect effects of those displays on the attacker’s motivation (startle response), but the only prefight activities that seemed to predict the attacker’s motivation were performed by the attacker. High levels of the large crab presenting its own chelae predicted high motivation and may simply reflect an existing high motivation. There were, however, two other prefight activities of the attacker that predicted the attacker’s motivation, these being approach and ambulatory raise. These both loaded on to PC1 along with high posture by both crabs and extension of the chelipeds by the small crab. One might predict that approach in the prefight stage would indicate a high motivation rather than the low motivation found here. One would also predict more approach by the larger crab but this was not found. It seems that when the larger crab shows high levels of approach this is the result of repeated approaches to which the opponent responds by displays rather than by assuming the defender role. Perhaps highly motivated attackers approach more decisively and thus are more likely to start the fight on the first attempt. Alternatively, the necessity to repeat attempts to initiate the fight proper might reduce the larger crab’s motivation. In contrast to approach, ambulatory raise by the attacker predicted high attacker motivation. This is unexpected as ambulatory raise and approach of the larger crab both loaded positively in the PCA and the function of this display is not clear (see below). The final measure of motivation is for the defender, taken as the duration for which it resists before giving up (Hack et al. 1997; Bridge et al. 2000). However, because fight duration is reduced if the attacker raps with high vigour (Briffa & Elwood 2002) and only defenders receiving weak raps manage to resist eviction (Briffa & Elwood 2002; Briffa et al. 2003) this measure is confounded by the abilities of the attacker. Cheliped displays did not influence contest duration but mutual grappling did. Grappling presumably denotes a reluctance of the smaller crab to assume the defender role. This might occur if the smaller crab assesses that the opponent is weak and, because weak attackers cannot fight with high vigour (Briffa & Elwood 2005), the defender may resist for longer. There was also a positive effect of the ambulatory raise by the attacker on the duration of resistance by the defender but the reason for this is not clear. Hazlett (1968) found that models of ambulatory raise had little or no effect on the receiver. The general conclusion is that cheliped presentation, often by both animals but for longer by the larger crab, affords a clear view of the chelipeds and appears to be an honest signal, the effect on the recipient depending on whether the chelipeds being presented are larger or smaller than those of the recipient. This display is associated with increased chances of the larger crab
becoming the attacker and winning the fight and higher attacker motivation. Cheliped extension, in contrast, seems to contain an element of bluff and is performed more by the smaller crab. When used extensively by the smaller crab it reduces the probability of the larger crab assuming the role of attacker and, if it does take that role, decreases the chances of the attacker winning the contest. We should consider, however, whether cheliped extension by the defender is used as an honest signal that indicates future resistance to eviction. If this is the case, we have a clear prediction that there should be a positive relation between duration of resistance and use of the extension display; however, this was not found. A second problem concerns the apparent effects of this display on the decision of the larger animal to assume the defender role instead of the more normal attacker role. There is no obvious reason why this would be an alternative to having to fight for longer, as it is only in the attacker role that it might take its choice of shells. The results suggest that rather than providing honest information, cheliped extension makes size assessment by the opponent more difficult (Riechert 1998) and is particularly effective in causing the larger animal to take the defender role when the opponents are close in body size. It is congruent with the idea of false signals being used when there is an advantage in avoiding or even winning a fight, which, if the signal does not deter the opponent, the sender will lose. That is, if the cost of the signal is low then the only retribution will be similar to that if the signal was not used (Adams & Mesterton-Gibbons 1995; Johnstone 1998). Thus the prefight period in hermit crabs is complex, involving a number of distinct displays that appear to have different effects and also reflect differing motivational states. Within this complexity there is scope for both honesty and manipulation (Johnstone 1998; Riechert 1998). Acknowledgments We are grateful to the BBSRC for funding this research and thank Mark Laidre for comments on the manuscript.
References Adams, E. S. & Caldwell, R. L. 1990. Deceptive communication in asymmetric fights of the stomatopod crustacean Gonodactylus bredini. Animal Behaviour, 39, 706e716. Adams, E. S. & Mesterton-Gibbons, M. 1995. The cost of threat displays and the stability of deceptive communication. Journal of Theoretical Biology, 175, 405e421. Backwell, P. R. Y., Christy, J. H., Telford, S. R., Jennions, M. D. & Passmore, N. I. 2000. Dishonest signalling in a fiddler crab. Proceedings of the Royal Society of London, Series B, 267, 719e724. Bridge, A. P., Elwood, R. W. & Dick, J. T. A. 2000. Imperfect assessment and limited information preclude optimal strategies in maleemale fights in the orb-weaving spider Metellina mengei. Proceedings of the Royal Society of London, Series B, 267, 273e279. Briffa, M. & Elwood, R. W. 2000. Cumulative or sequential assessment during hermit crab shell fights: effects of oxygen on decision rules. Proceedings of the Royal Society of London, Series B, 267, 2445e2452.
ELWOOD ET AL.: DISPLAYS AND FIGHT DECISIONS
Briffa, M. & Elwood, R. W. 2001a. Decision rules, energy metabolism and vigor of hermit crab fights. Proceedings of the Royal Society of London, Series B, 268, 1841e1848. Briffa, M. & Elwood, R. W. 2001b. Changes in motivational state during hermit crab shell fights. Animal Behaviour, 62, 505e510. Briffa, M. & Elwood, R. W. 2002. Power of shell-rapping signals influences physiological costs and subsequent decisions during hermit crab shell fights. Proceedings of the Royal Society of London, Series B, 269, 2331e2336. Briffa, M. & Elwood, R. W. 2003. Change in the power of shell rapping during shell fights between hermit crabs: does magnitude contribute to repeated signals? Behavioral Ecology, 14, 60e65. Briffa, M. & Elwood, R. W. 2004. Use of energy reserves in fighting hermit crabs. Proceedings of the Royal Society of London, Series B, 271, 373e379. Briffa, M. & Elwood, R. W. 2005. Rapid change in energy status in fighting animals: causes and effects of strategic decisions. Animal Behaviour, 70, 119e124. Briffa, M., Elwood, R. W. & Russ, J. M. 2003. Change in the power of rapping during shell fights between hermit crabs: does magnitude contribute to repeated signals? Behavioral Ecology, 14, 60e65. Clutton-Brock, T. H. & Albon, S. D. 1979. The roaring of red deer and the evolution of honest advertisement. Behaviour, 69, 145e170. Culshaw, A. D. & Broom, D. M. 1980. The imminence of behavioural change and the startle response of chicks. Behaviour, 73, 64e76. Deag, J. M. & Scott, G. W. 1999. ‘Conventional’ signals in avian agonistic displays: integrating theory, data and different levels of analysis. Journal of Theoretical Biology, 196, 155e162. Dill, L. M. & Gillett, J. F. 1991. The economic logic of the barnacle Balanus glandula (Darwin) hiding behaviour. Journal of Experimental Marine Biology and Ecology, 153, 115e127. Dowds, B. M. & Elwood, R. W. 1983. Shell wars: assessment strategies and the timing of decisions in hermit crab shell fights. Behaviour, 85, 1e24. Elwood, R. W. & Neil, S. J. 1992. Assessments and Decisions: a Study of Information Gathering by Hermit Crabs. London: Chapman & Hall. Elwood, R. W., McClean, A. & Webb, L. 1979. The development of shell preferences by the hermit crab Pagurus bernhardus. Animal Behaviour, 27, 940e946. Elwood, R. W., Marks, N. & Dick, J. T. A. 1995. Consequences of shell species preferences for female reproductive success in the hermit crab Pagurus bernhardus. Marine Biology, 123, 431e434. Elwood, R. W., Wood, K. E., Gallagher, M. & Dick, J. T. A. 1998. Probing motivational state during agonistic encounters in animals. Nature, 393, 66e68. Grafen, A. 1990. Biological signals as handicaps. Journal of Theoretical Biology, 144, 517e546. Hack, M. A., Thompson, J. D. & Fernandes, D. M. 1997. Fighting in males of the autumn spider, Metellina segmentata: the effects of relative body size, prior residency and female value on contest outcome and duration. Ethology, 103, 488e498.
Hardy, I. C. W. & Field, S. A. 1998. Logistic regression analysis of animal contests. Animal Behaviour, 56, 787e792. Hazlett, B. A. 1968. Communicatory effect of body position in Pagurus bernhardus (L.) (Decapoda, Anomura). Crustaceana, 14, 210e 214. Hazlett, B. A. & Bossert, W. H. 1965. A statistical analysis of the aggressive communications systems of some hermit crabs. Animal Behaviour, 13, 357e373. Hughes, M. 2000. Deception with honest signals: signal residuals and signal function in snapping shrimp. Behavioral Ecology, 11, 614e623. Hurd, P. L. 1997. Cooperative signalling between opponents in fish fights. Animal Behaviour, 54, 1309e1315. Hurd, P. L. & Enquist, M. 2005. A strategic taxonomy of biological communication. Animal Behaviour, 70, 1155e1170. Hurd, P. L. & Ydenberg, R. C. 1996. Calculating the ESS level of information transfer in aggressive communication. Evolutionary Ecology, 10, 221e232. Hyatt, G. W. & Salmon, M. 1979. Comparative statistical and information analysis of combat in the fiddler crabs, Uca pugilator and U. pugnax. Behaviour, 68, 1e23. Jackson, N. W. 1988. Information gathering and decision making during shell selection by the hermit crab Pagurus bernhardus. Ph.D. thesis, Queen’s University of Belfast. Johnstone, R. A. 1998. Game theory and communication. In: Game Theory and Animal Behaviour (Ed. by L. A. Dugatkin & H. K. Reeve), pp. 94e117. New York: Oxford University Press. Keeley, E. R. & Grant, J. W. A. 1993. Visual information, resource value and sequential assessment in convict cichlid (Cichlasoma nigrofasciatum) contests. Behavioral Ecology, 4, 345e349. Krebs, J. R. & Dawkins, R. 1984. Animal signals: mind reading and manipulation. In: Behavioural Ecology: An Evolutional Approach, 2nd edn. (Ed. by J. R. Krebs & N. B. Davies), pp. 380e402. Oxford: Blackwell Scientific. Matsumasa, M. & Murai, M. 2005. Changes in blood glucose and lactate levels of male fiddler crabs: effects of aggression and claw waving. Animal Behaviour, 69, 569e577. Maynard Smith, J. & Harper, D. 2003. Animal Signals. Oxford: Oxford University Press. Moorehouse, J. E., Fosbrooke, I. H. M. & Ludlow, A. R. 1987. Stopping a walking locust with sound: an analysis of variation in behavioural threshold. Experimental Biology, 46, 193e201. Neil, S. J. & Elwood, R. W. 1985. Behavioural modification during egg-brooding in the hermit crab, Pagurus bernhardus L. Journal of Experimental Marine Biology and Ecology, 94, 99e114. Parker, G. A. 1974. Assessment strategy and the evolution of fighting behaviour. Journal of Theoretical Biology, 47, 223e243. Riechert, S. E. 1998. Game theory and communication. In: Game Theory and Animal Behaviour (Ed. by L. A. Dugatkin & H. K. Reeve), pp. 64e93. New York: Oxford University Press. Rubenstein, D. I. & Hazlett, B. A. 1974. Examination of the agonistic behaviour of the crayfish Orconectes virilis by character analysis. Behaviour, 50, 193e216. Zahavi, A. 1975. Mate selection: a selection for a handicap. Journal of Theoretical Biology, 67, 205e214.
859