Discussion
The results of this study supported the predictions of the assessment models better than the non-assessment models in Rivulus marmoratus, as some of the contest behaviors were affected by the size of the bigger contestants. Although this study attempted to minimize the variability of the size of the smaller contestants, the size of the smaller contestants still had significant influences on the probability of the contestants initiating attacks in contests.
I. Testing between the Assessment and the Non-Assessment Models By limiting the variability of the size of the smaller contestants, I aimed to discriminate the assessment and the non-assessment models by examining the effect of the size of the bigger contestants in this study.
The assessment models predict that the size of the bigger contestant has negative effect on contest behaviors such as contest duration. The larger the bigger opponent is, the sooner the smaller contestant should retreat to avoid cost further accumulated. On the contrary, the non-assessment models predict that the size of the bigger contestants has no effect on contest duration. How long the smaller contestants should persist in a contest is based on their own size, but is not affected by the size of their bigger opponents (Table 1).
Summarizing the results of the size effects of the contestants on 4
contest characteristics (contest duration, contest intensity, display frequency
and attack frequency), three out of the 4 contest characteristics had the same trends as predicted in the assessment models (contest duration, contest intensity and attack frequency in the contests); however, only one of the trends was significant. Although the effects of the size of the bigger contestant on different contest characteristics varied a lot, its significant effects on contest intensity supported the assessment models. The influences of the body size of the two contestants on the 4 contest characteristics are further discussed below.
1. Contest Duration
Contest duration was not affected by the body sizes of either contestants. In the SRA model (Table 4), the effect of the size of the bigger contestants became more apparent, although still not significant, and the trend was more consistent with the prediction from the assessment models.
1.1.
Multiple linear regression model
Although the sizes of contestants had little influence on
contest duration, the trend was more consistent with the
assessment models (Table 3). There could be many possible
reasons why the sizes of the contestants did not have significant
effect on contest duration. Firstly, the size asymmetries were not
large enough to detect their influences on contest duration. In this
study, I minimized the size difference among the smaller
contestants, hoping to minimize its influence on contest
behaviors, which might allow the effect of the size of the bigger
opponents to become more detectable. But the largest relative size difference in the experiment was only 40.7%, meaning the biggest fish was only 1.407 times as big as its smaller opponent.
A small size asymmetry (30%) was found to have no influence on contest duration in study of Hofmann & Schildberger (2001) as well. Thus, it is possible that a bigger size difference between opponents is needed to detect size effect on contest duration. For example, in Enquist et al. (1990), the largest size asymmetry between opponents was 400-500 % and the sizes of the opponents was found to have significant influence on contest duration. Secondly, Enquist et al. (1990) showed that size effect would be more likely to be detected on qualitative characters, such as contest intensity, than on quantitative character, such as contest duration. Qualitative characters are more stable, while continuous measurements are more likely to bias with extreme outlying values.
1.2. Survival regression analysis model
Including the data from the 6 un-resolved contests in
survival regression analysis, the effect of the size of the bigger
contestant on the contest duration was approaching significant
level as predicted from the assessment models (Table 4). It is
necessary to state that a contest began once one contestant
displayed to its opponent. In the 6 un-resolved contests, displays
happened in all of them. But the aggressive interaction between
opponents seemed to cease after the displays (no attacks in the 5
out of the 6 contests). No obvious retreat behavior was observed in these contests. SRA included data from these 6 contests and treated them as contests with unknown retreat time (but longer than an hour). The negative relationship between the size of the bigger opponent and contest duration become more observable when these censored data were included for the analysis.
2. Contest Intensity
This contest characteristic showed the most significant relationships with the size of the two contestants as predicted from the assessment models. The results also suggested that the sizes of the two opponents had significant and opposite effects on contest intensity (Table 5).
Owing to the design of this experiment, the effect of the size of the smaller contestant was less influential than the effect of the size of the bigger contestant.
3. Frequency of Specific Contest Behaviors
The sizes of the two contestants had some insignificant influences on the frequency of displays and attacks. They are discussed in detail below.
3.1.
Display frequency
The sizes of the two contestants had no influence on this
contest characteristic. Both the sizes had positive effects on
display frequency as predicted in the non-assessment models, but
the effects were far from being significant (Table 6). In Enquist
et al. (1990), the effect of size asymmetry has the least influence on this contest characteristic among the 4 contest characteristics.
The results of my study were consistent with their findings. The sizes of the two contestants did not show any obvious relationship with display frequency.
Considering another result of this study, the smaller contestant had a higher tendency to initiate a contest when the size of its bigger opponent became larger. Possible reasons for this tendency were discussed in a latter section. If the estimated cost for displaying to its opponent is considerably low, contestant might cheat by repeating displays, which will make display frequency an unreliable assessment characteristic.
3.2.
Attack frequency
Comparing with display frequency, the sizes of the two contestants had much more apparent relationships with attack frequency. The patterns of the effects of the sizes of the two contestants met the predictions from the assessment models (Table 6).
As discussed in a latter section, as its size increased, the
smaller contestant would become more willing to initiate attacks
and engage in contests. The positive effect of body size of the
smaller contestants on the willingness to engage in fight might
play an important role in affecting the attack frequencies in
contests. When a smaller contestant has decided to fight, it might
have to fight more fiercely to overcome its size disadvantage and
win over its larger opponent.
It’s possible that R. marmoratus was in its most active reproductive state of its life cycle when it reached the size range of the smaller contestants used in this study, because the fish in this size range produced most eggs in our lab population (personal observation). As a result, the fish of this size range might be more aggressive as its size increased.
II The Effects of Body Size on Other Contest Behaviors
The contest behaviors of an individual were largely affected by the size of its opponent. In this study, the size of the bigger contestants had significant positive relationships with the probability of it initiating contests, retaliating, and winning escalations. Although this study attempted to minimize the variability of the size of the smaller contestant, the size of the smaller contestant still had highly positive relationships with its probabilities of initiating attacks. The relationships between the sizes of the two contestants and various contest behaviors are discussed in detail in the following sections.
1 Contest Outcome
Size has been found to be strongly correlated with the outcome of fights in many different species, e.g in crustaceans (Hazlett 1986;
Caldwell & Dingle 1979), spider (Dijkstra 1969; Riechert 1978;
Austad 1983), insects (Sigurjonsdottir & Parker 1981; O’Neill 1983;
Otronen 1984), fish (Frey & Miller 1972; Barlow 1983) and anurans
(Davies & Halliday 1978). Larger animals usually defeat smaller ones.
Although the sizes of the two contestants had no significant influence on the contest outcome in R. marmoratus (Table 7), their effects became marginally significant when I excluded clone factors from the model (S_BL：p= 0.06, B_BL：p= 0.05).
Contest outcome is the final result of competition between contestants. It may be influenced by factors other than body size.
Interactions between contestants during contest affect the final outcome as well. Investigating how the size of the two contestants influenced contest dynamics would enhance our understanding regarding contest outcome.
As it appeared in the results of my study, body size (measured as body length of fish) might only be one of the estimators of fighting ability in R. marmoratus. There might be some other unexplored characteristics of an individual that influence its fighting ability.
2 Contest Dynamics
2.1.Initiating displays
The smaller contestant (likely loser, Table 7) was equally
likely to initiate the contest against its much larger opponent
(46:36, not significantly different from 50% : 50%,
χ12=1.22,
p=0.27). Several researchers (Jackson 1991; Brabury & Vehrencamp
1998) have predicted that the likely winners of contests should
initiate the encounter and there are some empirical evidences to
support this hypothesis (Jackson 1988; Jackson 1991; Bekoff &
Scott 1989). Why did the losers (smaller contestants) initiate equal number of confrontations in this study ? Firstly, each individual should try to convey to its opponent that it has superior fighting skills and should refrain from signaling any weakness or its intension to retreat (Maynard Smith & Parker 1976; Kreb & Dawkins 1984). Secondly, the smaller of the two contestants might also be expected to initiate some of the contests if fighting ability assessment is inaccurate (Brabury &
Vehrencamp 1998). Lastly, if the estimated cost for display is low, the weaker contestant is more likely to cheat. In R. marmoratus, there seldom was serious damage from engaging in contests in confined space in this study. In nature, the fish would have even more chances to flee and avoid attacks from the winners.
Therefore, the cost is low for initiating confrontations by displaying first.
2.2.
Initiating attacks
As the size of a smaller contestant increased, its likelihood
of initiating attacks would also increase (Table 9). The result
suggested that the aggressiveness (the willing to fight) of the
smaller contestant increases as its own body size increases and is
not affected by the size of its bigger opponents. Nonetheless, out
of the 77 contests where attacks were observed, the smaller
opponent initiated attacks in only 27 contests (35%, significantly
less than 50%,
χ12=6.87, p=0.009).
Resource value might be a possible reason that underlies the positive relationship between aggression and the size of smaller contestants. Resource value is one of the most important factors that have been observed to affect contest behaviors. For example, Austad (1983) showed that resource value would influence the contest duration in the bowl and doily spider, Frontinella
pyramitela. In this species, the holder of a low-valued resourcedid not escalate against an intruder but retreated after a brief interaction. Resource value can therefore be interpreted as one of the important determinants of the willingness to fight. For instance, for male crickets, Gryllus bimaculatus (Hofmann &
Schildberger 2001), the most important resources are territories and females.
R. marmoratus are internally self-fertilizing hermaphroditic fish, therefore they are not likely to compete for mates. Hsu (1997) stated that the most possible resource they competing for is food. They don’t have obvious cyclic oviparous period, and sexually mature individuals can lay fertilized eggs all year around. The amount of eggs produced was observed to be positively correlated with body size (Harrington 1963). In the laboratory, the growth of this fish was observed to be constrained by the availability of food. As soon as live food was not constantly supplied, growth of fish was stunted (Kristensen 1970). So, they may compete for territory to gain access to food to grow faster and lay more eggs.
Although bigger fish tend to lay more eggs than smaller
ones, the smaller contestants in this study were the fastest growing and the most reproductive group of the fish in our lab during the study period (personal observation). It is possible that the positive size effect on the fertility of the smaller contestants plays an important part of the aggressiveness in the fish, which can be investigated in future studies.
2.3.
Probability of escalation
After an individual attacked its opponent, the probability of the opponent fighting back was influenced by the size of the attacker. The larger the attacker, the less likely for the opponent to fight back. There was also a insignificant tendency for the attacked opponent to have a higher probability of escalating the contest as its size increased (Table 10).
The effect of body size on various movement performances and physiological functions has been studied for decades. Size has found to be positively correlated with muscle strength and movement performances (Aasa et al. 2003). Size may be synonym of strength especially for individuals that are well-prepared to fight. When attacked, an individual may be able to evaluate the strength of the attack and use it as an indicator of the fighting ability of the attacker. As the size of the attacker increased, the strength of the attack should also increase and deter the opponent from fighting back.
Generally speaking, the individual that initiate attacks had
more than 50% chance of winning the contest (if the smaller
opponent attacked first=52%, 14/27; if the bigger opponent attacked first=57%, 28/49). If the attacked opponent retreated, the contest would end and the attacker would win the contest.
The smaller contestants won more contests this way (attacked first and the bigger opponent retreated without fighting back, 61%, 14/23). The smaller contestants that initiated attacks won 1.56 times as many contests as the smaller contestants that did not initiated attacks. High motivation increases the chances of the smaller individuals dominating over relatively larger opponents (Wagner 1989; Neat et al.1998; Kotiaho et al. 1999).
2.4.
