- Email: [email protected]
Fighting and assessment in male cichlid fish: the effects of asymmetries in gonadal state and body size
Anim. Behav., 1998, 55, 883–891
Fighting and assessment in male cichlid fish: the effects of asymmetries in gonadal state and body size FRANCIS C. NEAT*, FELICITY A. HUNTINGFORD* & MALCOLM M. C. BEVERIDGE† *Fish Biology Group, Division of Environmental & Evolutionary Biology, I.B.L.S., University of Glasgow and †Institute of Aquaculture, University of Stirling (Received 10 January 1997; initial acceptance 4 April 1997; final acceptance 24 August 1997; MS. number: 5445)
Abstract. In fights between animals over limited resources, the larger contestant often wins. Game theoretical models of animal fighting predict that relative body size is assessed during the fight and thus determines fight duration and intensity. In addition, if the contestants differ in the value they place on the disputed resource, this can also influence the outcome, duration and intensity of the fight. We studied territorial fighting in a cichlid fish, Tilapia zillii, in relation to relative body size and gonad weight. Relative gonad weight was a much stronger predictor of fight outcome than relative body size, even when body weight asymmetries were as large as 30%. This suggested that males with large gonads were fighting harder to defend their territory, perhaps because the value of a territory correlates with the gonadal state of the individual. A detailed analysis of mouth wrestling observed during fighting suggested that relative body size is assessed. However, contestants smaller than their opponent often continued to fight in spite of their size disadvantage. Weight disadvantaged winners appeared to fight more fiercely as suggested by a negative correlation between weight asymmetry and the proportion of bites inflicted by the winner. During escalated fighting, winners and losers differed consistently with regard to a behaviour termed mouth locking. Although neither biting nor persistence in mouth locking was related to gonad weight, we propose that the fish may have been assessing asymmetries unrelated to relative body size and possibly more related 1998 The Association for the Study of Animal Behaviour to levels of cost and the motivation to persist. It is a regular finding in studies of animal aggression that disputes over resources are won by the larger individual. Examples can be found throughout the animal kingdom, for instance, in teleost fish (Koops & Grant 1993), in crustaceans (Pavey & Fielder 1996) and in ungulates (Barrette & Vandal 1990). Contestants with larger bodies than their opponent are usually physically stronger and hence able to inflict greater costs on their rival and incur lesser costs themselves. In game theory terms, a size symmetry leads to differences in ‘resource-holding power’ of the contestants (Parker 1974; Maynard Smith 1982) and Correspondence: F. Huntingford, Fish Biology Group, Division of Environmental & Evolutionary Biology, I.B.L.S., University of Glasgow, Glasgow G12 8QQ, U.K. (email: [email protected]). M. M. C. Beveridge is at the Institute of Aquaculture, University of Stirling, Stirling FK9 4LA, U.K. 0003–3472/98/040883+09 $25.00/0/ar970669
the fight is resolved by assessing relative resourceholding power through informative displays and trials of strength. This information gathering is akin to statistical sampling and when sufficient information has accumulated, it pays the weaker individual to concede the resource and thus avoid the costs of continuing to fight in vain. According to Enquist & Leimar’s (1983) sequential assessment game, fight duration is predicted to be a function of how asymmetrical the contestants are; closely matched opponents need longer to detect the difference in resource-holding power and so fight longer. There are occasions, however, when an effect of body size does not predict the outcome of fights and this is usually when there is an asymmetry in the value that contestants place on the disputed resource, that is to say, the subjective resource value. Individuals that value the disputed resource
1998 The Association for the Study of Animal Behaviour
883
884
Animal Behaviour, 55, 4
more than their opponent fight harder and persist longer in the fight thus increasing their probability of winning (Enquist & Leimar 1987). In nature, asymmetries in the subjective value of a disputed resource can arise under a number of circumstances. For example, when bald eagles, Haliaeetus leucocephalus, gather to feed upon salmon, newly arrived and unfed individuals will often supplant larger opponents that have fed to satiation (Hansen 1986). Alternatively, a territory or mate may become extremely valuable at certain times if the opportunity to reproduce is temporally or physiologically constrained. For example, during the infrequent periods that male elephants, Loxodonta africana, come into breeding condition or ‘musthe’ they become extraordinarily aggressive (Poole & Moss 1981). By their treatment of asymmetries in resourceholding power and resource value, game theory analyses of animal fights have provided a conceptual framework that makes robust and testable predictions of the broad features of animal fights, such as their length, intensity and outcome. However, escalated fighting is a behaviourally complex and protracted affair and the behavioural details by which animals resolve a dispute are not properly understood. This is partly a problem arising from the difficulty of quantifying the contact interactions that comprise a major part of escalated fighting. Contact interactions, such as mouth wrestling in cichlids and antler locking in deer, are where most accurate information for assessment is likely to be transferred. In most studies the occurrence of the interaction is recorded (for example, Enquist & Jakobssen 1986), rather than the outcome of the interaction for each individual. Our purpose in this paper is two-fold. First, we report the surprising result that, despite large weight asymmetries, victory in fights between adult male cichlid fish, Tilapia zillii, defending territories is closely related to asymmetry in gonadal state and not body weight. This result suggested that males with large gonads were fighting harder to defend their territory, perhaps because the value of a territory is a function of the gonadal state of the individual. Second, we address the issue of how fights are resolved when variables other than resource-holding power are important. We ask the following questions. Is there assessment of resource-holding power in circumstances where it does not predict winning?
Is it the case that motivational asymmetries associated with gonadal state (and perhaps resource value) are assessed in T. zillii? Our approach to tackling these questions was to undertake a detailed analysis of behavioural interactions during fighting and relate this to asymmetries in relative body size and gonad weight. Tilapia zillii is a monogamous, substrate spawning cichlid inhabiting lakes and rivers of Northern and Western Africa (Fryer & Iles 1972; Philippart & Ruwet 1982). Mature males establish territories and excavate nests along the lake edge (Bruton & Gophen 1992) and are extremely aggressive towards intruders. Similarly, in captivity, mature males readily establish territories and, if ownership is disputed, a fight rapidly ensues.
METHODS The fish we used were from a 1-year-old cohort of sexually mature males reared at the Institute of Aquaculture, Stirling Unviersity, U.K. Experiments were carried out under a U.K. Home Office project licence (to F.A.H.) and a personal licence (to F.C.N.). Before our experiments, we maintained the fish (between 20 and 50 individuals at any one time) together in a large (150 cm long, 60 cm high and 80 cm wide) glass stock tank at 271C, under a 12:12 h light:dark cycle and fed them commercial fish pellets twice a day. We selected 36 pairs by weight (20–140 g) to give a range of weight differences. We marked individuals by subcutaneous injection of alcian blue dye to aid identification. We then placed the pairs in glass contest tanks (1004030 cm long), one fish on either side of a removable opaque partition that divided the tank equally. To minimize territory variation, each side contained only gravel for nest building, and each fish was fed eight food pellets per day. The water was aerated and quality maintained by an external power filter. We avoided any owner–intruder asymmetries by allowing each pair to ‘own’ their territory for 7 days. After this time we raised the partition and video-filmed the ensuing fight. The fish were separated immediately after one fish had lost the fight (see below) and then killed by an overdose of benzocaine, weighed and later dissected in order to weigh the testes. Gonadal state is represented as the gonadosomatic index, calculated as testes weight as a percentage of body weight, although
Neat et al.: Fighting and assessment in cichlids we performed some analyses using absolute testes weight. Statistical analyses were performed using Statistica and SPSS for Windows. All P-values are two-tailed. Ethical Note We stopped fights as soon as the loser fish had made the decision to quit (a clear behavioural switch: see below), thus preventing further attack from the winner. The degree of escalation observed in these confined laboratory conditions was similar to fights observed under semi-natural conditions where the loser could escape and hide (Neat 1996), suggesting that this study did not impose abnormally stressful conditions on the animals. Scale loss from biting was the most serious form of injury observed during fighting (see Neat & Huntingford 1998), but this can become severe only if the fish are not separated after one has signalled submission (which was prevented).
RESULTS Fight Structure and Behavioural Content The basic structure of fights between male T. zillii was similar to that of many cichlid fish (see Baerends & Baerends-Van Roon 1950), consisting of three broad phases. The first was a period of display, in which the fish raised their fins and inflated their opercular membranes. The contestants often swam in parallel with exaggerated movements, beating at each other with their tails. The fish also showed courtship activities towards an intruder, such as quivering and digging. This, together with the fact that no physical contact was made and all but one fight escalated beyond this stage, suggests that the display phase, rather than being overt fighting, serves to establish gender and social status. This phase lasted 25–10 529 s (median duration and inter-quartile range=209 s and 100–1855 s). Escalated fighting began abruptly with the next phase, termed the ‘mouth-wrestling’ phase, where contestants engaged in multiple bouts of grappling with open mouths. We regarded the first fish to approach its opponent and open its mouth as initiating the bout. An individual was deemed to win the bout if it succeeded in pushing its opponent backwards and releasing contact, thereby
885
ending the bout. The duration of the mouthwrestling phase was defined as the time from initiation of the first bout, to the time at which 75% of the total number of bouts of mouth wrestling had been initiated. We chose this cut-off because, while there is no abrupt transition from mouth wrestling to the next phase, the frequency of initiating falls off sharply around this point. This phase lasted 325–2182 s (median duration and inter-quartile range=582 s and 325–916 s). The final phase, termed the ‘carouselling’ phase, was judged to begin after the 75% cut-off point for the mouth-wrestling phase and was terminated when one of the contestants fled. Typically, the fish rapidly chased each other nose-to-tail, attempting to bite one another. Bites were scored only if they actually landed upon the opponent. Also in this phase there were frequent acts of ‘mouth locking’, where the fish clamped their mouths on each other’s lips (unlike the openmouthed posture of the mouth wrestling) and stayed in this embrace until one fish shook itself free, thereby ‘breaking’ the lock. During a mouth lock, the opercular beat rate increased and, after the lock, the fish often paused and gasped heavily. This phase lasted 64–1904 s (median duration and inter-quartile range=527 s and 305–908 s). Fights ended with one fish suddenly fleeing and making no further aggressive acts towards the opponent. We defined winning and losing by this criterion, although there was one case where both contestants fled simultaneously. We defined the overall duration of escalated fighting as the time from initiation of the first bout of mouth wrestling to the time at which the loser flees, minus the duration of any pauses that occurred. Asymmetries in Relation to Victory Out of the 36 fights staged, 19 were won by the larger individual, 14 by the smaller, one fight remained unresolved, one pair did not fight and one pair were exactly the same weight. Despite size asymmetries in which the larger fish was up to 35% heavier, median weights of winners and losers, 61.2 and 59.7 g, respectively, were not significantly different (Wilcoxon signed-ranks test: Z=1.73, N=34, ). The most extreme case of a smaller fish winning was where the larger fish was 30% heavier. Gonadosomatic index ranged from 0.01 to 0.61% and was not correlated with body weight
Animal Behaviour, 55, 4
886
(Pearson correlation: r=0.12, N=72, ). Twentynine fights were won by fish with a greater gonadosomatic index than their opponent, four by fish with a lesser gonadosomatic index and in one fight the gonadosomatic indexes were equal. Median gonadosomatic indexes of winners and losers were 0.28% and 0.16%, respectively, a highly significant difference (Wilcoxon signed-ranks test: Z=4.48, N=34, P<0.001). Similarly, winners had absolutely heavier testes than losers (Wilcoxon signed-ranks test: Z=4.44, N=34, P<0.001). Condition factor [(weight/length3)100] did not differ significantly between winners and losers (Wilcoxon signed-ranks test: Z=0.35, N=34, ) and there was no correlation (Pearson correlation: r= 0.12, N=72, ) between the condition factor and gonadosomatic index, indicating that the effect of gonadosomatic index is not simply a consequence of winning fish being in better condition. Thus, asymmetry in gonadosomatic index, rather than asymmetry in body weight or condition predicted winning. We investigated the effects of differences in body weight and in testes weight on the probability of victory by logistic regression. The dependent variable was the proportion of fights won by the heavier fish of a pair and the independent variables were the difference in weight and the difference in testes weight. Following Enquist & Leimar (1983), the difference in weight of a pair of fish WD(a,b), where a and b are the winner and loser, respectively, was expressed as: WD(a,b) =ln (wta/wtb) This has the useful property that WD(a,b) = WD(b,a) and, therefore, is negative when a smaller fish won, zero when contestants were equal and positive when the larger fish won. The difference in testis weight was expressed in the same way. The proportion of fights won by the larger fish of the pair was strongly related to the difference in gonadosomatic index, but there was no significant effect of difference in body weight (Table I). The model is best illustrated by a three-dimensional probability surface (see Fig. 1). Asymmetries in Relation to Duration of Phases Multiple regression showed that neither body weight nor testes weight difference was signifi-
Table I. Results of logistic regression: the effect of difference in body and testes weight on the proportion of fights won by the larger fish of a pair Variable
B
P
Testes weight difference Body weight difference Constant
8.69 22.33 1.25
3.65 12.68 1.27
<0.05 <0.1
cantly related to the duration of the display phase or the duration of the mouth-wrestling phase. However, the duration of the carouselling phase was related to both difference in weight and difference in testes weight, in both cases negatively (Table II). The duration of the carouselling phase was not correlated with the gonadosomatic index of the loser (Spearman rank correlation: rs = 0.2, N=33, ). Asymmetries in Relation to Mouth Wrestling Fish that went on to win the fight initiated significantly more bouts of mouth wrestling (median number of bouts initiated were 35 and 27, for winners and losers, respectively; Wilcoxon signed-ranks test: Z=2.25, N=33, P<0.05). However, eventual winners did not win more bouts of mouth wrestling (median number of bouts won by winners and losers were 23 and 17, respectively; Wilcoxon signed-ranks test: Z=1.22, N=33, ). Multiple regression showed that the difference in body weight was strongly related to the proportion of bouts of mouth wrestling initiated and won, whereas the difference in testes weight had no significant effects (Table II). In other words, the greater the weight asymmetry, the more bouts of mouth wrestling the larger fish initiated and won. To examine in more detail the relationship between body weight difference and mouthwrestling behaviour, and to establish if there is a process during this phase that may be termed ‘sequential assessment’, we carried out the following analysis. We divided the sample of 33 fights that yielded data on mouth wrestling into three groups, independent of eventual victory and according to the percentage weight difference of the larger fish (0–2.5%, N=11; 2.6–10%, N=11 and >10%, N=11). We measured the course of this phase in terms of the number of bouts (irrespective of time). This involved dividing the total
Neat et al.: Fighting and assessment in cichlids
887
P (larger fish wins)
proportion of bouts were being initiated and won by the larger fish (Fig. 2). 1.0 0.8 0.6 0.4 0.2
0.30 0.24 W 0.18 0.2 eig 0.1 ht 0.12 0.0 di ffe 0.06 –0.1 nce ffere re 0.00 –0.2 SI di nc G e Figure 1. Relationship between the probability that the larger fish of a pair wins (Y-axis), the difference in gonadosomatic index (GSI), that is, testes weight as a percentage of body weight, of the larger fish (Z-axis) and difference in body weight of the larger fish (X-axis). The surface illustrates how GSI and body weight combine; difference in GSI is a much stronger effect than body weight. The probability of the larger fish winning is maximal when the larger fish has a greater relative GSI but very rapidly becomes minimal as the difference in GSI becomes negative. It is only when the size advantage is very large (>0.2) that there is a chance that a larger fish with a smaller GSI will win the contest. The figure was produced using the Systat statistical software package for Windows.
number of mouth-wrestling bouts initiated for each fight into quarters, and calculating the number of bouts initiated and won by the larger fish in each quarter. For the three groups, in each quarter, the differences between the number of bouts initiated and won by the larger fish and the smaller fish were analysed using the Wilcoxon signed-ranks test. In the early stages of the fight, the large and small contestants in all size difference groups initiated and won a similar proportion of bouts of mouth wrestling and this is not different from the random expectation (0.5; Fig. 2). In the group where contestants were very closely matched (i.e. weight difference of 0–2.5%), this remained the case throughout the fight, indicating that fish that were closely matched for weight were also closely matched in the ability to win at mouth wrestling. In the groups with moderate (2.6–10%) and large (>10%) size disparity, however, after half of all the bouts had occurred, there was an overall impression that a higher
Asymmetries in Relation to Biting and Mouth Locking Pairs of fish tended to match each other bitefor-bite and, overall, there was no significant difference between winners and losers in the total number of bites inflicted (Wilcoxon signed-ranks test: Z=0.21, N=33, ). In 29 fights the total number of bites inflicted was greater than 10 allowing a multiple regression analysis of the proportion of the total bites inflicted by the winning fish; there was a significant effect of the difference in body weight but not of difference in testes weight (Table II). Figure 3 shows the negative relationship between the proportion of bites inflicted by the winner and the difference in weight. As smaller fish did not bite more than larger fish in absolute terms (Wilcoxon signedranks test: Z=1.45, N=33, ), this suggests that winners that are much smaller than their opponents inflict relatively more bites than winners that are larger than their opponents. However, absolute number of bites inflicted by smaller winners (median=37) was not significantly greater than that of larger winners (median=31; Mann–Whitney U-test: U=105, N1 =19, N2 =14, ). The number of bites received by the loser was not correlated with its gonadosomatic index (Spearman rank correlation: rs = 0.03, N=33, ). The mouth-locking behaviour was another component of fighting in which there were consistent differences between winners and losers Out of the fights that yielded sufficient mouthlocking data (>10 mouth locks), winners broke a significantly lower proportion of mouth locks (median=0.3) than losers (median=0.55; Wilcoxon signed-ranks test: Z=3.12, N=29, P<0.01). Further analysis by multiple regression did not suggest that the proportion of mouth locks broken was dependent upon the difference in body or testes weight (see Table II).
DISCUSSION Body size (and with it, resource-holding power) has been shown to be a major factor determining the outcome and structure of many animal fights.
888
Animal Behaviour, 55, 4 Table II. Results of multiple regression analysis: effect of difference in body weight and testes weight on duration of phases and other behavioural measurements Beta
P
Duration of display (F2,30 =0.07, Adj. R2 =0.0, ) Weight difference 0.02 Testes difference 0.06 Duration of mouth wrestling (F2,30 =1.18, Adj. R2 =0.01, ) Weight difference 0.27 Testes difference 0.002 2 Duration of carouselling (F2,30 =5.13, Adj. R =0.21, P<0.01) Weight difference 0.41 <0.05 Testes difference 0.36 <0.05 Proportion of bouts of mouth wrestling initiated by larger fish (F2,30 =11.78, Adj. R2 =0.4, P<0.01) Weight difference 0.66 <0.01 Testes difference 0.08 Proportion of bouts of mouth wrestling won by larger fish (F2,30 =4.69, Adj. R2 =0.19, P<0.05) Weight difference 0.47 <0.05 Testes difference 0.17 Proportion of bites inflicted by winner (F2,25 =5.10, Adj. R2 =0.20, P<0.05) Weight difference 0.49 0.01 Testes difference 0.07 Proportion of mouth locks broken by winner (F2,19 =0.82, Adj. R2 =0.0, ) Weight difference 0.28 Testes difference 0.03 Proportions were arcsine transformed.
In the staged fights reported here between males defending territories, body size evidently did not have the predicted effect on fight outcome. This is in contrast to many studies involving pair-wise contests in fish (Turner & Huntingford 1986; Enquist et al. 1990; Ribowski & Franck 1993). Our results suggest that asymmetry in gonadosomatic index is a better predictor of winning than is asymmetry in body weight. We cannot rule out the possibility that the period of isolation before fighting (7 days) fully negated any effects of previous experience and social status. As we standardized territory quality and found no evidence to suggest that the gonadosomatic index was correlated with condition factor, however, this raises the possibility that the gonadosomatic index relates to the subjective value of a territory. What is the Functional Significance of Variation in Gonadal State? Our study is not alone in finding an effect of gonadal state on aggressive behaviour in cichlids. In Schwank’s (1980) study, gonadal state predicted dominance in male Tilapia mariae,
although gonadal state was measured only indirectly by the length of the genital papilla. In addition, Holder et al. (1991) reported that male Cichlasoma citrinellum become increasingly aggressive the closer they get to spawning (the genital papilla was used as an index of proximity to spawnings). Thus, there is clearly some important functional underpinning of gonadal state in the aggressive behaviour of these cichlids. Tilapia zillii breeds continuously in tropical regions and has a reproductive cycle lasting 4–6 weeks associated with changes in behaviour and gonadal state (Siddiqui 1979). It may be that males at different stages in the cycle act more or less aggressively; gonadosomatic index is likely to increase as the fish get closer to spawning, and the closer to spawning, the more valuable the territory becomes to that individual. Thus, the value of the territory to a particular male may covary with that male’s gonadosomatic index. Under this hypothesis, this study presents an opportunity to evaluate how fights are resolved when a factor other than resource-holding power (possibly resource value, but certainly some motivational factor) is the key to resolving fights.
Proportion won by heavier fish
(a) 0.75 * *
*
*
0.50
0.25
1
2
3
4
Proportion of bites inflicted by the winner
Proportion initiated by heavier fish
Neat et al.: Fighting and assessment in cichlids
889
0.85 0.75 0.65 0.55 0.45 0.35 0.25
–0.35 –0.25 –0.15 –0.05 0.05
0.15
0.25
0.35
Weight difference index
(b) *
0.75 * *
* *
Figure 3. The proportion of bites inflicted by the winning fish plotted against the asymmetry in body weight. When the weight difference is negative, the winner was smaller than its opponents (see text).
0.50
0.25
1 2 3 4 Quarter of mouth wrestling
Figure 2. The proportion of bouts of mouth wrestling (a) initiated and (b) won by the larger fish (p) (the smaller fish won 1p bouts) for each quarter of the total number of bouts initiated. : 0–2.5% (N=11) weight difference group; : 2.6–10% group (N=11); : >10% group (N=11). The dashed line is the 0.5 proportion level that would be expected if there were no effect of weight difference. Those bars marked with an * indicate that the larger fish initiated or won a significantly greater proportion of bouts than the smaller fish (Wilcoxon signed-ranks test: Z=1.99–2.83, N=11, Pc0.02).
Is Relative Body Size Assessed? It is intuitive that contestants should assess body size in situations where it does predict victory, but less clear where it does not. The sequential assessment game (Enquist & Leimar 1983) and its extension to allow for variation in subjective resource value (Enquist & Leimar 1987), assume that the assessment of resource-holding power takes place during the fight, through a sequence of displays and interactions. The fights in the present study did progress in a sequential way, from display through mouth wrestling to
carouselling, and our data demonstrate that success in mouth wrestling was related to difference in body size; fish heavier than their opponent initiated and won more bouts of mouth wrestling. Furthermore, the data suggest that assessment of relative size is a dynamic process (as predicted by Enquist & Leimar 1983). It was only in the latter stages of the fight that the difference in relative strength became apparent. It is interesting that fish that went on to win the fight initiated more bouts of mouth wrestling, suggesting that winners are more highly motivated. These data suggest that body size is coupled with physical strength and this asymmetry is progressively assessed during the mouth-wrestling phase. However, this is evidently not the means by which the fight is resolved, as fish smaller than their opponents often continued to fight in spite of information about their size disadvantage and frequently went on to win the fight. Biting and Mouth Locking in Relation to Relative Size and Gonadosomatic Index According to the sequential assessment game, the effect of increased resource value is to increase the cost of the fight (Enquist & Leimar 1987). It also predicts that fighting begins with the least costly displays, so assuming that gonadosomatic index is correlated with resource value, we might
890
Animal Behaviour, 55, 4
expect to find an effect of relative gonadosomatic index in the later, most costly stages of the fight. Most biting occurs at this stage and often results in injury and provokes retaliation, hence it is likely to be a costly act. Two predictions follow. First, fish that have invested heavily in large testes should take greater risks and bite more. Second, as only the loser of the fight pays the full cost that it is prepared to pay, the cost the loser pays should be correlated with its own gonadosomatic index. There was no evidence, however, to suggest that biting was related to either the contestant’s gonadosomatic index or the difference in gonadosomatic index between contestants. The proportion of bites inflicted by the winner was highest for severely weight-disadvantaged fish, and was lowest for weight-advantaged winners. Although this is only correlational evidence, it does suggest that size-disadvantaged winners are compensating by playing a high-risk strategy. When an individual is disadvantaged, it may be that a high-risk strategy is the only option that offers any chance of winning. High-risk strategies are used by subordinate finches, Serinus serinus, when they overcome dominants in disputes over food (Senar et al. 1992). The data on mouth locking strongly suggest that breaking the mouth lock is related to the likelihood of eventually giving up, but there was no evidence that this behaviour provided information on relative gonadosomatic index. The significance of the mouth lock is not clear, but it may be either a trial of stamina or a means by which the contestants assess the relative costs that each opponent is prepared to incur. As the mouth is clamped shut, oxygen uptake is presumably restricted; thus, a fish that has incurred a greater oxygen debt (and hence higher cost) may be forced to break the mouth lock sooner. How are Fights Resolved? The sequential assessment game is the most recent and detailed model to analyse fight resolution in relation to variation in both relative resource-holding power and resource value (Enquist & Leimar 1987). The evolutionarily stable strategy predicts that the contestants will fight until one of them assesses that its costs of fighting will exceed the benefits it stands to gain from retaining the resource sooner than it will for the opponent. To make such a decision each
contestant must have information on how much cost each individual has incurred and how much cost each opponent is prepared to pay in order to defend the disputed resource. Difference in body size influences the infliction and accruement of costs, which would explain why the fish assess relative body size during mouth wrestling. In addition, motivational factors related to resource value will affect the infliction of cost and level of cost accrued (Leimar et al. 1991). Elwood & Neil (1992) presented evidence that hermit crabs, Pagarus bernhardus, alter their fighting strategies as they acquire information on the value of each other’s shells and Poole (1987) reported data for African elephants that suggest that males assess reproductive state or ‘musthe’ condition. In the present study, the finding that breaking the mouth lock is associated with losing suggests that male T. zillii assess an asymmetry other than body size, possibly each other’s intention to persist, that is to say, how much cost they are prepared to incur during the fight. It is puzzling, however, why there was no relationship between the proportion of mouth locks broken and relative gonadosomatic index. Thus, while it seems likely, we cannot conclude that the fish are making available reliable information about relative levels of cost incurred. Our results lead us to conclude that in addition to the assessment of body size, fighting strategies also serve to obtain information on both how much cost each opponent has paid and how much each is prepared to pay to retain its resource, that is to say, to signal intentions or motivation. ACKNOWLEDGMENTS Many thanks to Paul Walton, Anders Alana¨ra¨, Graham Ruxton, Neil Metcalfe, Olof Leimar, John Lazarus, Angela Turner and two anonymous referees for valuable comments at various stages of the manuscript. We also thank Roddy McDonald, John Laurie and Keith Ranson who gave technical assistance. F.C.N. was funded by a NERC studentship. REFERENCES Baerends, G. P. & Baerends-Van Roon, J. M. 1950. An introduction to the study of the ethology of cichlid fishes. Behaviour (Supplement), 1, 1–200.
Neat et al.: Fighting and assessment in cichlids Barrette, C. & Vandal, D. 1990. Sparring, relative antler size and assessment in male caribou. Behav. Ecol. Sociobiol., 26, 383–387. Bruton, M. N. & Gophen, M. 1992. The effect of environmental factors on the nesting and courtship behaviour of Tilapia zillii in Lake Kinneret (Israel). Hydrobiologia, 239, 171–178. Elwood, R. W. & Neil, S. J. 1992. Assessments and Decisions: a Study of Information Gathering by Hermit Crabs. London: Chapman & Hall. Enquist, M. & Jakobssen, S. 1986. Decision making and assessment in the fighting behaviour of Nannacara anomala (Cichlidae, Pisces). Ethology, 72, 143–153. Enquist, M. & Leimar, O. 1983. Evolution of fighting behaviour: decision rules and assessment of relative strength. J. theor. Biol., 102, 387–410. Enquist, M. & Leimar, O. 1987. Evolution of fighting behaviour: the effect of variation in resource value. J. theor. Biol., 127, 187–205. Enquist, M., Leimar, O., Ljungberg, T., Mallner, Y. & Segerdahl, N. 1990. A test of the sequential assessment game: fighting in the cichlid fish Nannacara anomola. Anim. Behav., 40, 1–14. Fryer, G. & Iles, T. D. 1972. The Cichlid Fishes of the African Great Lakes: their Biology and Evolution. Edinburgh: Oliver & Boyd. Hansen, A. J. 1986. Fighting behavior in bald eagles: a test of game theory. Ecology, 67, 787–797. Holder, J. L., Barlow, G. W. & Francis, R. C. 1991. Differences in aggressiveness in the Midas cichlid fish (Cichlasoma citrinellum) in relation to sex, reproductive state and the individual. Ethology, 88, 297–306. Koops, M. A. & Grant, J. W. A. 1993. Weight asymmetry and sequential assessment in convict cichlid contests. Can. J. Zool., 71, 475–479. Leimar, O., Austad, S. & Enquist, M. 1991. A test of the sequential assessment game: fighting in the bowl and doily spider Frontinella pyramitela. Evolution, 45, 862– 874. Maynard Smith, J. 1982. Evolution and the Theory of Games. Cambridge: Cambridge University Press.
891
Neat, F. C. 1996. Behavioural and physiological studies of fighting in male Tilapia zillii (Cichlidae). Ph.D. thesis, University of Glasgow. Neat, F. C., Taylor, A. C. & Huntingford, F. A. 1998. Proximate costs of fighting in male cichlid fish: the role of injuries and energy metabolism. Anim. Behav. 55, 875–882. Parker, G. A. 1974. Assessment and the evolution of fighting behaviour. J. theor. Biol., 47, 223–243. Pavey, C. R. & Fielder, D. R. 1996. The influence of size differential on agonistic behaviour in the freshwater crayfish, Cherax cuspidatus (Decapoda: Parastacidae). J. Zool., Lond., 238, 445–457. Philippart, J.-Cl. & Ruwet, J.-Cl. 1982. Ecology and distribution of tilapias. In: The Biology and Culture of Tilapias (Ed. by R. S. V. Pullin & R. H. LoweMcConnell), pp. 15–59. Manilla: ICLARM. Poole, J. H. 1987. Announcing intent: the aggressive state of musthe in African elephants. Anim. Behav., 37, 140–152. Poole, J. H. & Moss, C. J. 1981. Musthe in the African elephant, Loxodonta africana. Nature. Lond., 292, 830–831. Ribowski, A. & Franck, D. 1993. Demonstration of strength and concealment of weaknesses in escalated fights of male swordtails (Xiphophorus helleri). Ethology, 93, 265–274. Schwank, E. 1980. The effect of size and hormonal state on the establishment of dominance in young males of Tilapia mariae (Pisces: Cichlidae). Behav. Proc., 5, 45–53. Senar, J. C., Camerino, M. & Metcalfe, N. B. 1992. Fighting as a subordinate in finch flocks: escalation is effective but risky. Anim. Behav., 43, 862–864. Siddiqui, A. Q. 1979. Reproductive biology of Tilapia zillii (Gervais) in Lake Naivasha, Kenya. Environ. Biol. Fish., 4, 257–262. Turner, G. F. & Huntingford, F. A. 1986. A problem for game theory analysis: assessment and intention in male mouthbrooder contests. Anim. Behav., 34, 961–970.
Fighting and assessment in male cichlid fish: the effects of asymmetries in gonadal state and body size FRANCIS C. NEAT*, FELICITY A. HUNTINGFORD* & MALCOLM M. C. BEVERIDGE† *Fish Biology Group, Division of Environmental & Evolutionary Biology, I.B.L.S., University of Glasgow and †Institute of Aquaculture, University of Stirling (Received 10 January 1997; initial acceptance 4 April 1997; final acceptance 24 August 1997; MS. number: 5445)
Abstract. In fights between animals over limited resources, the larger contestant often wins. Game theoretical models of animal fighting predict that relative body size is assessed during the fight and thus determines fight duration and intensity. In addition, if the contestants differ in the value they place on the disputed resource, this can also influence the outcome, duration and intensity of the fight. We studied territorial fighting in a cichlid fish, Tilapia zillii, in relation to relative body size and gonad weight. Relative gonad weight was a much stronger predictor of fight outcome than relative body size, even when body weight asymmetries were as large as 30%. This suggested that males with large gonads were fighting harder to defend their territory, perhaps because the value of a territory correlates with the gonadal state of the individual. A detailed analysis of mouth wrestling observed during fighting suggested that relative body size is assessed. However, contestants smaller than their opponent often continued to fight in spite of their size disadvantage. Weight disadvantaged winners appeared to fight more fiercely as suggested by a negative correlation between weight asymmetry and the proportion of bites inflicted by the winner. During escalated fighting, winners and losers differed consistently with regard to a behaviour termed mouth locking. Although neither biting nor persistence in mouth locking was related to gonad weight, we propose that the fish may have been assessing asymmetries unrelated to relative body size and possibly more related 1998 The Association for the Study of Animal Behaviour to levels of cost and the motivation to persist. It is a regular finding in studies of animal aggression that disputes over resources are won by the larger individual. Examples can be found throughout the animal kingdom, for instance, in teleost fish (Koops & Grant 1993), in crustaceans (Pavey & Fielder 1996) and in ungulates (Barrette & Vandal 1990). Contestants with larger bodies than their opponent are usually physically stronger and hence able to inflict greater costs on their rival and incur lesser costs themselves. In game theory terms, a size symmetry leads to differences in ‘resource-holding power’ of the contestants (Parker 1974; Maynard Smith 1982) and Correspondence: F. Huntingford, Fish Biology Group, Division of Environmental & Evolutionary Biology, I.B.L.S., University of Glasgow, Glasgow G12 8QQ, U.K. (email: [email protected]). M. M. C. Beveridge is at the Institute of Aquaculture, University of Stirling, Stirling FK9 4LA, U.K. 0003–3472/98/040883+09 $25.00/0/ar970669
the fight is resolved by assessing relative resourceholding power through informative displays and trials of strength. This information gathering is akin to statistical sampling and when sufficient information has accumulated, it pays the weaker individual to concede the resource and thus avoid the costs of continuing to fight in vain. According to Enquist & Leimar’s (1983) sequential assessment game, fight duration is predicted to be a function of how asymmetrical the contestants are; closely matched opponents need longer to detect the difference in resource-holding power and so fight longer. There are occasions, however, when an effect of body size does not predict the outcome of fights and this is usually when there is an asymmetry in the value that contestants place on the disputed resource, that is to say, the subjective resource value. Individuals that value the disputed resource
1998 The Association for the Study of Animal Behaviour
883
884
Animal Behaviour, 55, 4
more than their opponent fight harder and persist longer in the fight thus increasing their probability of winning (Enquist & Leimar 1987). In nature, asymmetries in the subjective value of a disputed resource can arise under a number of circumstances. For example, when bald eagles, Haliaeetus leucocephalus, gather to feed upon salmon, newly arrived and unfed individuals will often supplant larger opponents that have fed to satiation (Hansen 1986). Alternatively, a territory or mate may become extremely valuable at certain times if the opportunity to reproduce is temporally or physiologically constrained. For example, during the infrequent periods that male elephants, Loxodonta africana, come into breeding condition or ‘musthe’ they become extraordinarily aggressive (Poole & Moss 1981). By their treatment of asymmetries in resourceholding power and resource value, game theory analyses of animal fights have provided a conceptual framework that makes robust and testable predictions of the broad features of animal fights, such as their length, intensity and outcome. However, escalated fighting is a behaviourally complex and protracted affair and the behavioural details by which animals resolve a dispute are not properly understood. This is partly a problem arising from the difficulty of quantifying the contact interactions that comprise a major part of escalated fighting. Contact interactions, such as mouth wrestling in cichlids and antler locking in deer, are where most accurate information for assessment is likely to be transferred. In most studies the occurrence of the interaction is recorded (for example, Enquist & Jakobssen 1986), rather than the outcome of the interaction for each individual. Our purpose in this paper is two-fold. First, we report the surprising result that, despite large weight asymmetries, victory in fights between adult male cichlid fish, Tilapia zillii, defending territories is closely related to asymmetry in gonadal state and not body weight. This result suggested that males with large gonads were fighting harder to defend their territory, perhaps because the value of a territory is a function of the gonadal state of the individual. Second, we address the issue of how fights are resolved when variables other than resource-holding power are important. We ask the following questions. Is there assessment of resource-holding power in circumstances where it does not predict winning?
Is it the case that motivational asymmetries associated with gonadal state (and perhaps resource value) are assessed in T. zillii? Our approach to tackling these questions was to undertake a detailed analysis of behavioural interactions during fighting and relate this to asymmetries in relative body size and gonad weight. Tilapia zillii is a monogamous, substrate spawning cichlid inhabiting lakes and rivers of Northern and Western Africa (Fryer & Iles 1972; Philippart & Ruwet 1982). Mature males establish territories and excavate nests along the lake edge (Bruton & Gophen 1992) and are extremely aggressive towards intruders. Similarly, in captivity, mature males readily establish territories and, if ownership is disputed, a fight rapidly ensues.
METHODS The fish we used were from a 1-year-old cohort of sexually mature males reared at the Institute of Aquaculture, Stirling Unviersity, U.K. Experiments were carried out under a U.K. Home Office project licence (to F.A.H.) and a personal licence (to F.C.N.). Before our experiments, we maintained the fish (between 20 and 50 individuals at any one time) together in a large (150 cm long, 60 cm high and 80 cm wide) glass stock tank at 271C, under a 12:12 h light:dark cycle and fed them commercial fish pellets twice a day. We selected 36 pairs by weight (20–140 g) to give a range of weight differences. We marked individuals by subcutaneous injection of alcian blue dye to aid identification. We then placed the pairs in glass contest tanks (1004030 cm long), one fish on either side of a removable opaque partition that divided the tank equally. To minimize territory variation, each side contained only gravel for nest building, and each fish was fed eight food pellets per day. The water was aerated and quality maintained by an external power filter. We avoided any owner–intruder asymmetries by allowing each pair to ‘own’ their territory for 7 days. After this time we raised the partition and video-filmed the ensuing fight. The fish were separated immediately after one fish had lost the fight (see below) and then killed by an overdose of benzocaine, weighed and later dissected in order to weigh the testes. Gonadal state is represented as the gonadosomatic index, calculated as testes weight as a percentage of body weight, although
Neat et al.: Fighting and assessment in cichlids we performed some analyses using absolute testes weight. Statistical analyses were performed using Statistica and SPSS for Windows. All P-values are two-tailed. Ethical Note We stopped fights as soon as the loser fish had made the decision to quit (a clear behavioural switch: see below), thus preventing further attack from the winner. The degree of escalation observed in these confined laboratory conditions was similar to fights observed under semi-natural conditions where the loser could escape and hide (Neat 1996), suggesting that this study did not impose abnormally stressful conditions on the animals. Scale loss from biting was the most serious form of injury observed during fighting (see Neat & Huntingford 1998), but this can become severe only if the fish are not separated after one has signalled submission (which was prevented).
RESULTS Fight Structure and Behavioural Content The basic structure of fights between male T. zillii was similar to that of many cichlid fish (see Baerends & Baerends-Van Roon 1950), consisting of three broad phases. The first was a period of display, in which the fish raised their fins and inflated their opercular membranes. The contestants often swam in parallel with exaggerated movements, beating at each other with their tails. The fish also showed courtship activities towards an intruder, such as quivering and digging. This, together with the fact that no physical contact was made and all but one fight escalated beyond this stage, suggests that the display phase, rather than being overt fighting, serves to establish gender and social status. This phase lasted 25–10 529 s (median duration and inter-quartile range=209 s and 100–1855 s). Escalated fighting began abruptly with the next phase, termed the ‘mouth-wrestling’ phase, where contestants engaged in multiple bouts of grappling with open mouths. We regarded the first fish to approach its opponent and open its mouth as initiating the bout. An individual was deemed to win the bout if it succeeded in pushing its opponent backwards and releasing contact, thereby
885
ending the bout. The duration of the mouthwrestling phase was defined as the time from initiation of the first bout, to the time at which 75% of the total number of bouts of mouth wrestling had been initiated. We chose this cut-off because, while there is no abrupt transition from mouth wrestling to the next phase, the frequency of initiating falls off sharply around this point. This phase lasted 325–2182 s (median duration and inter-quartile range=582 s and 325–916 s). The final phase, termed the ‘carouselling’ phase, was judged to begin after the 75% cut-off point for the mouth-wrestling phase and was terminated when one of the contestants fled. Typically, the fish rapidly chased each other nose-to-tail, attempting to bite one another. Bites were scored only if they actually landed upon the opponent. Also in this phase there were frequent acts of ‘mouth locking’, where the fish clamped their mouths on each other’s lips (unlike the openmouthed posture of the mouth wrestling) and stayed in this embrace until one fish shook itself free, thereby ‘breaking’ the lock. During a mouth lock, the opercular beat rate increased and, after the lock, the fish often paused and gasped heavily. This phase lasted 64–1904 s (median duration and inter-quartile range=527 s and 305–908 s). Fights ended with one fish suddenly fleeing and making no further aggressive acts towards the opponent. We defined winning and losing by this criterion, although there was one case where both contestants fled simultaneously. We defined the overall duration of escalated fighting as the time from initiation of the first bout of mouth wrestling to the time at which the loser flees, minus the duration of any pauses that occurred. Asymmetries in Relation to Victory Out of the 36 fights staged, 19 were won by the larger individual, 14 by the smaller, one fight remained unresolved, one pair did not fight and one pair were exactly the same weight. Despite size asymmetries in which the larger fish was up to 35% heavier, median weights of winners and losers, 61.2 and 59.7 g, respectively, were not significantly different (Wilcoxon signed-ranks test: Z=1.73, N=34, ). The most extreme case of a smaller fish winning was where the larger fish was 30% heavier. Gonadosomatic index ranged from 0.01 to 0.61% and was not correlated with body weight
Animal Behaviour, 55, 4
886
(Pearson correlation: r=0.12, N=72, ). Twentynine fights were won by fish with a greater gonadosomatic index than their opponent, four by fish with a lesser gonadosomatic index and in one fight the gonadosomatic indexes were equal. Median gonadosomatic indexes of winners and losers were 0.28% and 0.16%, respectively, a highly significant difference (Wilcoxon signed-ranks test: Z=4.48, N=34, P<0.001). Similarly, winners had absolutely heavier testes than losers (Wilcoxon signed-ranks test: Z=4.44, N=34, P<0.001). Condition factor [(weight/length3)100] did not differ significantly between winners and losers (Wilcoxon signed-ranks test: Z=0.35, N=34, ) and there was no correlation (Pearson correlation: r= 0.12, N=72, ) between the condition factor and gonadosomatic index, indicating that the effect of gonadosomatic index is not simply a consequence of winning fish being in better condition. Thus, asymmetry in gonadosomatic index, rather than asymmetry in body weight or condition predicted winning. We investigated the effects of differences in body weight and in testes weight on the probability of victory by logistic regression. The dependent variable was the proportion of fights won by the heavier fish of a pair and the independent variables were the difference in weight and the difference in testes weight. Following Enquist & Leimar (1983), the difference in weight of a pair of fish WD(a,b), where a and b are the winner and loser, respectively, was expressed as: WD(a,b) =ln (wta/wtb) This has the useful property that WD(a,b) = WD(b,a) and, therefore, is negative when a smaller fish won, zero when contestants were equal and positive when the larger fish won. The difference in testis weight was expressed in the same way. The proportion of fights won by the larger fish of the pair was strongly related to the difference in gonadosomatic index, but there was no significant effect of difference in body weight (Table I). The model is best illustrated by a three-dimensional probability surface (see Fig. 1). Asymmetries in Relation to Duration of Phases Multiple regression showed that neither body weight nor testes weight difference was signifi-
Table I. Results of logistic regression: the effect of difference in body and testes weight on the proportion of fights won by the larger fish of a pair Variable
B
P
Testes weight difference Body weight difference Constant
8.69 22.33 1.25
3.65 12.68 1.27
<0.05 <0.1
cantly related to the duration of the display phase or the duration of the mouth-wrestling phase. However, the duration of the carouselling phase was related to both difference in weight and difference in testes weight, in both cases negatively (Table II). The duration of the carouselling phase was not correlated with the gonadosomatic index of the loser (Spearman rank correlation: rs = 0.2, N=33, ). Asymmetries in Relation to Mouth Wrestling Fish that went on to win the fight initiated significantly more bouts of mouth wrestling (median number of bouts initiated were 35 and 27, for winners and losers, respectively; Wilcoxon signed-ranks test: Z=2.25, N=33, P<0.05). However, eventual winners did not win more bouts of mouth wrestling (median number of bouts won by winners and losers were 23 and 17, respectively; Wilcoxon signed-ranks test: Z=1.22, N=33, ). Multiple regression showed that the difference in body weight was strongly related to the proportion of bouts of mouth wrestling initiated and won, whereas the difference in testes weight had no significant effects (Table II). In other words, the greater the weight asymmetry, the more bouts of mouth wrestling the larger fish initiated and won. To examine in more detail the relationship between body weight difference and mouthwrestling behaviour, and to establish if there is a process during this phase that may be termed ‘sequential assessment’, we carried out the following analysis. We divided the sample of 33 fights that yielded data on mouth wrestling into three groups, independent of eventual victory and according to the percentage weight difference of the larger fish (0–2.5%, N=11; 2.6–10%, N=11 and >10%, N=11). We measured the course of this phase in terms of the number of bouts (irrespective of time). This involved dividing the total
Neat et al.: Fighting and assessment in cichlids
887
P (larger fish wins)
proportion of bouts were being initiated and won by the larger fish (Fig. 2). 1.0 0.8 0.6 0.4 0.2
0.30 0.24 W 0.18 0.2 eig 0.1 ht 0.12 0.0 di ffe 0.06 –0.1 nce ffere re 0.00 –0.2 SI di nc G e Figure 1. Relationship between the probability that the larger fish of a pair wins (Y-axis), the difference in gonadosomatic index (GSI), that is, testes weight as a percentage of body weight, of the larger fish (Z-axis) and difference in body weight of the larger fish (X-axis). The surface illustrates how GSI and body weight combine; difference in GSI is a much stronger effect than body weight. The probability of the larger fish winning is maximal when the larger fish has a greater relative GSI but very rapidly becomes minimal as the difference in GSI becomes negative. It is only when the size advantage is very large (>0.2) that there is a chance that a larger fish with a smaller GSI will win the contest. The figure was produced using the Systat statistical software package for Windows.
number of mouth-wrestling bouts initiated for each fight into quarters, and calculating the number of bouts initiated and won by the larger fish in each quarter. For the three groups, in each quarter, the differences between the number of bouts initiated and won by the larger fish and the smaller fish were analysed using the Wilcoxon signed-ranks test. In the early stages of the fight, the large and small contestants in all size difference groups initiated and won a similar proportion of bouts of mouth wrestling and this is not different from the random expectation (0.5; Fig. 2). In the group where contestants were very closely matched (i.e. weight difference of 0–2.5%), this remained the case throughout the fight, indicating that fish that were closely matched for weight were also closely matched in the ability to win at mouth wrestling. In the groups with moderate (2.6–10%) and large (>10%) size disparity, however, after half of all the bouts had occurred, there was an overall impression that a higher
Asymmetries in Relation to Biting and Mouth Locking Pairs of fish tended to match each other bitefor-bite and, overall, there was no significant difference between winners and losers in the total number of bites inflicted (Wilcoxon signed-ranks test: Z=0.21, N=33, ). In 29 fights the total number of bites inflicted was greater than 10 allowing a multiple regression analysis of the proportion of the total bites inflicted by the winning fish; there was a significant effect of the difference in body weight but not of difference in testes weight (Table II). Figure 3 shows the negative relationship between the proportion of bites inflicted by the winner and the difference in weight. As smaller fish did not bite more than larger fish in absolute terms (Wilcoxon signedranks test: Z=1.45, N=33, ), this suggests that winners that are much smaller than their opponents inflict relatively more bites than winners that are larger than their opponents. However, absolute number of bites inflicted by smaller winners (median=37) was not significantly greater than that of larger winners (median=31; Mann–Whitney U-test: U=105, N1 =19, N2 =14, ). The number of bites received by the loser was not correlated with its gonadosomatic index (Spearman rank correlation: rs = 0.03, N=33, ). The mouth-locking behaviour was another component of fighting in which there were consistent differences between winners and losers Out of the fights that yielded sufficient mouthlocking data (>10 mouth locks), winners broke a significantly lower proportion of mouth locks (median=0.3) than losers (median=0.55; Wilcoxon signed-ranks test: Z=3.12, N=29, P<0.01). Further analysis by multiple regression did not suggest that the proportion of mouth locks broken was dependent upon the difference in body or testes weight (see Table II).
DISCUSSION Body size (and with it, resource-holding power) has been shown to be a major factor determining the outcome and structure of many animal fights.
888
Animal Behaviour, 55, 4 Table II. Results of multiple regression analysis: effect of difference in body weight and testes weight on duration of phases and other behavioural measurements Beta
P
Duration of display (F2,30 =0.07, Adj. R2 =0.0, ) Weight difference 0.02 Testes difference 0.06 Duration of mouth wrestling (F2,30 =1.18, Adj. R2 =0.01, ) Weight difference 0.27 Testes difference 0.002 2 Duration of carouselling (F2,30 =5.13, Adj. R =0.21, P<0.01) Weight difference 0.41 <0.05 Testes difference 0.36 <0.05 Proportion of bouts of mouth wrestling initiated by larger fish (F2,30 =11.78, Adj. R2 =0.4, P<0.01) Weight difference 0.66 <0.01 Testes difference 0.08 Proportion of bouts of mouth wrestling won by larger fish (F2,30 =4.69, Adj. R2 =0.19, P<0.05) Weight difference 0.47 <0.05 Testes difference 0.17 Proportion of bites inflicted by winner (F2,25 =5.10, Adj. R2 =0.20, P<0.05) Weight difference 0.49 0.01 Testes difference 0.07 Proportion of mouth locks broken by winner (F2,19 =0.82, Adj. R2 =0.0, ) Weight difference 0.28 Testes difference 0.03 Proportions were arcsine transformed.
In the staged fights reported here between males defending territories, body size evidently did not have the predicted effect on fight outcome. This is in contrast to many studies involving pair-wise contests in fish (Turner & Huntingford 1986; Enquist et al. 1990; Ribowski & Franck 1993). Our results suggest that asymmetry in gonadosomatic index is a better predictor of winning than is asymmetry in body weight. We cannot rule out the possibility that the period of isolation before fighting (7 days) fully negated any effects of previous experience and social status. As we standardized territory quality and found no evidence to suggest that the gonadosomatic index was correlated with condition factor, however, this raises the possibility that the gonadosomatic index relates to the subjective value of a territory. What is the Functional Significance of Variation in Gonadal State? Our study is not alone in finding an effect of gonadal state on aggressive behaviour in cichlids. In Schwank’s (1980) study, gonadal state predicted dominance in male Tilapia mariae,
although gonadal state was measured only indirectly by the length of the genital papilla. In addition, Holder et al. (1991) reported that male Cichlasoma citrinellum become increasingly aggressive the closer they get to spawning (the genital papilla was used as an index of proximity to spawnings). Thus, there is clearly some important functional underpinning of gonadal state in the aggressive behaviour of these cichlids. Tilapia zillii breeds continuously in tropical regions and has a reproductive cycle lasting 4–6 weeks associated with changes in behaviour and gonadal state (Siddiqui 1979). It may be that males at different stages in the cycle act more or less aggressively; gonadosomatic index is likely to increase as the fish get closer to spawning, and the closer to spawning, the more valuable the territory becomes to that individual. Thus, the value of the territory to a particular male may covary with that male’s gonadosomatic index. Under this hypothesis, this study presents an opportunity to evaluate how fights are resolved when a factor other than resource-holding power (possibly resource value, but certainly some motivational factor) is the key to resolving fights.
Proportion won by heavier fish
(a) 0.75 * *
*
*
0.50
0.25
1
2
3
4
Proportion of bites inflicted by the winner
Proportion initiated by heavier fish
Neat et al.: Fighting and assessment in cichlids
889
0.85 0.75 0.65 0.55 0.45 0.35 0.25
–0.35 –0.25 –0.15 –0.05 0.05
0.15
0.25
0.35
Weight difference index
(b) *
0.75 * *
* *
Figure 3. The proportion of bites inflicted by the winning fish plotted against the asymmetry in body weight. When the weight difference is negative, the winner was smaller than its opponents (see text).
0.50
0.25
1 2 3 4 Quarter of mouth wrestling
Figure 2. The proportion of bouts of mouth wrestling (a) initiated and (b) won by the larger fish (p) (the smaller fish won 1p bouts) for each quarter of the total number of bouts initiated. : 0–2.5% (N=11) weight difference group; : 2.6–10% group (N=11); : >10% group (N=11). The dashed line is the 0.5 proportion level that would be expected if there were no effect of weight difference. Those bars marked with an * indicate that the larger fish initiated or won a significantly greater proportion of bouts than the smaller fish (Wilcoxon signed-ranks test: Z=1.99–2.83, N=11, Pc0.02).
Is Relative Body Size Assessed? It is intuitive that contestants should assess body size in situations where it does predict victory, but less clear where it does not. The sequential assessment game (Enquist & Leimar 1983) and its extension to allow for variation in subjective resource value (Enquist & Leimar 1987), assume that the assessment of resource-holding power takes place during the fight, through a sequence of displays and interactions. The fights in the present study did progress in a sequential way, from display through mouth wrestling to
carouselling, and our data demonstrate that success in mouth wrestling was related to difference in body size; fish heavier than their opponent initiated and won more bouts of mouth wrestling. Furthermore, the data suggest that assessment of relative size is a dynamic process (as predicted by Enquist & Leimar 1983). It was only in the latter stages of the fight that the difference in relative strength became apparent. It is interesting that fish that went on to win the fight initiated more bouts of mouth wrestling, suggesting that winners are more highly motivated. These data suggest that body size is coupled with physical strength and this asymmetry is progressively assessed during the mouth-wrestling phase. However, this is evidently not the means by which the fight is resolved, as fish smaller than their opponents often continued to fight in spite of information about their size disadvantage and frequently went on to win the fight. Biting and Mouth Locking in Relation to Relative Size and Gonadosomatic Index According to the sequential assessment game, the effect of increased resource value is to increase the cost of the fight (Enquist & Leimar 1987). It also predicts that fighting begins with the least costly displays, so assuming that gonadosomatic index is correlated with resource value, we might
890
Animal Behaviour, 55, 4
expect to find an effect of relative gonadosomatic index in the later, most costly stages of the fight. Most biting occurs at this stage and often results in injury and provokes retaliation, hence it is likely to be a costly act. Two predictions follow. First, fish that have invested heavily in large testes should take greater risks and bite more. Second, as only the loser of the fight pays the full cost that it is prepared to pay, the cost the loser pays should be correlated with its own gonadosomatic index. There was no evidence, however, to suggest that biting was related to either the contestant’s gonadosomatic index or the difference in gonadosomatic index between contestants. The proportion of bites inflicted by the winner was highest for severely weight-disadvantaged fish, and was lowest for weight-advantaged winners. Although this is only correlational evidence, it does suggest that size-disadvantaged winners are compensating by playing a high-risk strategy. When an individual is disadvantaged, it may be that a high-risk strategy is the only option that offers any chance of winning. High-risk strategies are used by subordinate finches, Serinus serinus, when they overcome dominants in disputes over food (Senar et al. 1992). The data on mouth locking strongly suggest that breaking the mouth lock is related to the likelihood of eventually giving up, but there was no evidence that this behaviour provided information on relative gonadosomatic index. The significance of the mouth lock is not clear, but it may be either a trial of stamina or a means by which the contestants assess the relative costs that each opponent is prepared to incur. As the mouth is clamped shut, oxygen uptake is presumably restricted; thus, a fish that has incurred a greater oxygen debt (and hence higher cost) may be forced to break the mouth lock sooner. How are Fights Resolved? The sequential assessment game is the most recent and detailed model to analyse fight resolution in relation to variation in both relative resource-holding power and resource value (Enquist & Leimar 1987). The evolutionarily stable strategy predicts that the contestants will fight until one of them assesses that its costs of fighting will exceed the benefits it stands to gain from retaining the resource sooner than it will for the opponent. To make such a decision each
contestant must have information on how much cost each individual has incurred and how much cost each opponent is prepared to pay in order to defend the disputed resource. Difference in body size influences the infliction and accruement of costs, which would explain why the fish assess relative body size during mouth wrestling. In addition, motivational factors related to resource value will affect the infliction of cost and level of cost accrued (Leimar et al. 1991). Elwood & Neil (1992) presented evidence that hermit crabs, Pagarus bernhardus, alter their fighting strategies as they acquire information on the value of each other’s shells and Poole (1987) reported data for African elephants that suggest that males assess reproductive state or ‘musthe’ condition. In the present study, the finding that breaking the mouth lock is associated with losing suggests that male T. zillii assess an asymmetry other than body size, possibly each other’s intention to persist, that is to say, how much cost they are prepared to incur during the fight. It is puzzling, however, why there was no relationship between the proportion of mouth locks broken and relative gonadosomatic index. Thus, while it seems likely, we cannot conclude that the fish are making available reliable information about relative levels of cost incurred. Our results lead us to conclude that in addition to the assessment of body size, fighting strategies also serve to obtain information on both how much cost each opponent has paid and how much each is prepared to pay to retain its resource, that is to say, to signal intentions or motivation. ACKNOWLEDGMENTS Many thanks to Paul Walton, Anders Alana¨ra¨, Graham Ruxton, Neil Metcalfe, Olof Leimar, John Lazarus, Angela Turner and two anonymous referees for valuable comments at various stages of the manuscript. We also thank Roddy McDonald, John Laurie and Keith Ranson who gave technical assistance. F.C.N. was funded by a NERC studentship. REFERENCES Baerends, G. P. & Baerends-Van Roon, J. M. 1950. An introduction to the study of the ethology of cichlid fishes. Behaviour (Supplement), 1, 1–200.
Neat et al.: Fighting and assessment in cichlids Barrette, C. & Vandal, D. 1990. Sparring, relative antler size and assessment in male caribou. Behav. Ecol. Sociobiol., 26, 383–387. Bruton, M. N. & Gophen, M. 1992. The effect of environmental factors on the nesting and courtship behaviour of Tilapia zillii in Lake Kinneret (Israel). Hydrobiologia, 239, 171–178. Elwood, R. W. & Neil, S. J. 1992. Assessments and Decisions: a Study of Information Gathering by Hermit Crabs. London: Chapman & Hall. Enquist, M. & Jakobssen, S. 1986. Decision making and assessment in the fighting behaviour of Nannacara anomala (Cichlidae, Pisces). Ethology, 72, 143–153. Enquist, M. & Leimar, O. 1983. Evolution of fighting behaviour: decision rules and assessment of relative strength. J. theor. Biol., 102, 387–410. Enquist, M. & Leimar, O. 1987. Evolution of fighting behaviour: the effect of variation in resource value. J. theor. Biol., 127, 187–205. Enquist, M., Leimar, O., Ljungberg, T., Mallner, Y. & Segerdahl, N. 1990. A test of the sequential assessment game: fighting in the cichlid fish Nannacara anomola. Anim. Behav., 40, 1–14. Fryer, G. & Iles, T. D. 1972. The Cichlid Fishes of the African Great Lakes: their Biology and Evolution. Edinburgh: Oliver & Boyd. Hansen, A. J. 1986. Fighting behavior in bald eagles: a test of game theory. Ecology, 67, 787–797. Holder, J. L., Barlow, G. W. & Francis, R. C. 1991. Differences in aggressiveness in the Midas cichlid fish (Cichlasoma citrinellum) in relation to sex, reproductive state and the individual. Ethology, 88, 297–306. Koops, M. A. & Grant, J. W. A. 1993. Weight asymmetry and sequential assessment in convict cichlid contests. Can. J. Zool., 71, 475–479. Leimar, O., Austad, S. & Enquist, M. 1991. A test of the sequential assessment game: fighting in the bowl and doily spider Frontinella pyramitela. Evolution, 45, 862– 874. Maynard Smith, J. 1982. Evolution and the Theory of Games. Cambridge: Cambridge University Press.
891
Neat, F. C. 1996. Behavioural and physiological studies of fighting in male Tilapia zillii (Cichlidae). Ph.D. thesis, University of Glasgow. Neat, F. C., Taylor, A. C. & Huntingford, F. A. 1998. Proximate costs of fighting in male cichlid fish: the role of injuries and energy metabolism. Anim. Behav. 55, 875–882. Parker, G. A. 1974. Assessment and the evolution of fighting behaviour. J. theor. Biol., 47, 223–243. Pavey, C. R. & Fielder, D. R. 1996. The influence of size differential on agonistic behaviour in the freshwater crayfish, Cherax cuspidatus (Decapoda: Parastacidae). J. Zool., Lond., 238, 445–457. Philippart, J.-Cl. & Ruwet, J.-Cl. 1982. Ecology and distribution of tilapias. In: The Biology and Culture of Tilapias (Ed. by R. S. V. Pullin & R. H. LoweMcConnell), pp. 15–59. Manilla: ICLARM. Poole, J. H. 1987. Announcing intent: the aggressive state of musthe in African elephants. Anim. Behav., 37, 140–152. Poole, J. H. & Moss, C. J. 1981. Musthe in the African elephant, Loxodonta africana. Nature. Lond., 292, 830–831. Ribowski, A. & Franck, D. 1993. Demonstration of strength and concealment of weaknesses in escalated fights of male swordtails (Xiphophorus helleri). Ethology, 93, 265–274. Schwank, E. 1980. The effect of size and hormonal state on the establishment of dominance in young males of Tilapia mariae (Pisces: Cichlidae). Behav. Proc., 5, 45–53. Senar, J. C., Camerino, M. & Metcalfe, N. B. 1992. Fighting as a subordinate in finch flocks: escalation is effective but risky. Anim. Behav., 43, 862–864. Siddiqui, A. Q. 1979. Reproductive biology of Tilapia zillii (Gervais) in Lake Naivasha, Kenya. Environ. Biol. Fish., 4, 257–262. Turner, G. F. & Huntingford, F. A. 1986. A problem for game theory analysis: assessment and intention in male mouthbrooder contests. Anim. Behav., 34, 961–970.