| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Behavioral Ecology Vol. 11 No. 6: 614-623
© 2000 International Society for Behavioral Ecology
Deception with honest signals: signal residuals and signal function in snapping shrimp
Department of Zoology, Duke University, Box 90325, Durham, NC 27708-0325, USA
Address correspondence to M. Hughes at the Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544-1003, USA. E-mail: mhughes{at}Princeton.edu .
Received 15 January 1999; revised 1 March 2000; accepted 13 March 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Animals in competitive interactions often assess the competitive ability of opponents using signals. Signals used in competitive interactions are generally predicted to be honest, but open to low levels of deceit. Such "incomplete honesty" in signals can be studied by using signal residuals, the residuals from the regression of a measure of signal structure on competitive ability. Specifically, individuals with positive signal residuals produce signals that exaggerate their competitive ability; deceptive use of these signals may occur if signalers for whom the signal exaggerates their apparent competitive ability use the signal more frequently. I used this framework to examine the use of the open chela display by big-clawed snapping shrimp (Alpheus heterochaelis). Competitive interactions between snapping shrimp are resolved primarily on the basis of body size, and the open chela display is used by males to assess body size. I found that the production of the open chela display by males responding to superior competitors depends on chela residuals, such that individuals for whom the display exaggerates their apparent size produce the display more often. This effect can be seen both in response to isolated chelae and in staged competitive interactions. Interactions involving shrimp with larger chela residuals are long and highly escalated, suggesting that chela residuals affect assessment of competitive ability. Thus, the increased use of the open chela display by males for which the display exaggerates apparent body size is an example of deceptive use of an otherwise honest signal.
Key words: Alpheus heterochaelis, deception, honesty, reliability, signals, snapping shrimp.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Aggressive interactions between animals are potentially costly, both in terms of energetic costs and in terms of the risk of serious injury or death. Thus, animals are expected to evolve mechanisms for assessing their opponents without direct combat, through reliable signals either of aggressive intent (Enquist, 1985
;
van Rhijn and Vodegel, 1980)
or of competitive ability (Enquist and
Leimar, 1983; Maynard Smith
and Harper 1988; Parker,
1974; Parker and Rubenstein,
1981). Signals used for assessment of competitive ability are
expected to be, on average, honest or uncheatable; that is, they are expected
to be physically or physiologically linked to competitive ability, so that
animals of low competitive ability cannot produce the signal of an animal with
high competitive ability (Maynard Smith
and Harper, 1988; Wiley,
1983; Zahavi,
1977). Nonetheless, it has been argued that honest signaling is
vulnerable to corruption in the sense that low levels of deception may be
evolutionarily stable within otherwise honest signaling systems
(Adams and Mesterton-Gibbons,
1995; Bond, 1989;
Dawkins and Guilford, 1991;
Gardner and Morris, 1989;
Getty, 1997). Such deceptive
signaling may occur as a result of imperfect relationships between signal
structure and competitive ability (Dawkins
and Guilford, 1991), high costs to receivers for assessing signals
(Dawkins and Guilford, 1991;
Getty, 1997), low costs to
signalers for producing deceptive signals when deceptive signalers are
relatively rare in the population (Gardner
and Morris, 1989), and high benefits to some signalers when
deceptive displays are successful (Adams
and Mesterton-Gibbons, 1995). There is considerable theoretical
work, then, that predicts signals will evolve to be generally but not
completely honest.
Probably the best known example of dishonesty in intraspecific
communication is the meral spread display of stomatopods, an aggressive
display given both by newly molted and intermolt individuals
(Adams and Caldwell, 1990;
Steger and Caldwell, 1983).
Although the newly molted individuals are incapable of following through with
aggressive behavior, this deceptive use of the signal is often successful
because newly molted individuals are rare in the population; most animals
performing the display are fully capable of following through with an attack
(Steger and Caldwell, 1983).
This example clearly demonstrates intraspecific deception with an otherwise
honest signal and supports the prediction that the deceptive use of otherwise
honest signals can occur at low frequencies. This example does not lend itself
readily to generalization, however. Newly molted stomatopods are clearly
bluffing when they perform an aggressive display; that is, it is clear to the
experimenter that the aggressive signal is a bluff because a newly molted
stomatopod cannot engage in combat. Furthermore, this deception involves two
discrete classes of individuals: those that are intermolt (and thus honest)
and those which are newly molted (and thus bluffing). In other signaling
systems, it can be much more difficult for the experimenter to distinguish
between a successful bluff and an honest signal, especially when the signaled
information is not discrete but continuous (such as size, parenting ability,
mate quality, etc.). For the majority of animal signals, there is little
agreement over how one might test for "incomplete honesty." Here I
propose a general means by which we may test for deceptive use of honest
signals, using as an example a signal used in competitive interactions between
big-clawed snapping shrimp (Alpheus heterochaelis).
Framework: signal residuals
When we say that a signal is honest, we imply that there is a relationship
between variation in the structure or performance of the signal and variation
in some aspect of competitive ability.
Figure 1 illustrates a
hypothetical relationship between a measure of competitive ability (such as
body size, condition, etc.) and a measure of signal structure (such as the
dominant frequency of an acoustic call, the size of an ornament, etc.). The
relationship between signal structure and competitive ability, here expressed
as a linear regression, provides an opportunity for receivers to use the
signal to estimate the competitive ability of the signaler. This signal can be
considered honest in that, on average, receivers will be able to distinguish
between individuals of different competitive abilities based on the structure
of the signal.
|
The shaded area of Figure 1
represents the range of values for these measures for different individuals
within the population. Even when this relationship is causal or constrained by
physical or physiological factors, there will always be some variation in
signal structure that is not explained by variation in competitive ability.
Such variation may be due to genetic differences between individuals,
differences in developmental history, correlations with additional variables
that may or may not also be related to competitive ability, or random effects.
As noted by Wagner (1992),
such variation around a relationship between competitive ability and signal
structure is common, but its functional significance is not well understood.
Indeed, while relationships between signal structure and some measure of
competitive ability are well known and have been used as evidence for honesty
in signaling, signal variation that is unexplained by competitive ability is
typically ignored in analyses of signaling behavior.
The dashed line in Figure 1 connects two individuals with the same competitive ability. These individuals differ in terms of signal structure, such that one individual (A) has a higher value for signal structure than is predicted by the population-based relationship between the signal and competitive ability, and the other individual (B) has a lower value for signal structure than is predicted by this relationship. The vertical deviations of individual points from a regression line can be quantified using the standardized residuals from the regression; residuals reflect that portion of the variation in the dependent variable that is not explained by the independent variable. In other words, residuals quantify the variation in the signal that is independent of competitive ability. In this example, A has a positive value for "signal residual," and B has a negative value.
There are, in general, three possibilities regarding the functional
significance of signal residuals. First, signal residuals may play no role in
the function of a signal. The variation measured by signal residuals may not
be resolved by the sensory system of the receiver, for example, or may be
masked by noise or signal degradation under normal signaling conditions.
Second, signal residuals may themselves be correlated with another measure of
competitive ability; that is, they may function as another source of reliable
information to the receiver. Wagner
(1992) found that in
Blanchard's cricket frogs (Acris crepitans blanchardi), dominant
frequency is a function of body size, but the degree to which individuals
lowered the dominant frequency of their call could be used as a measure of
size-independent fighting ability. In this case, variation in the signal that
is independent of one measure of competitive ability (body size) is not
deceptive but rather provides additional information regarding competitive
ability. Finally, when signal residuals are not correlated with another
component of competitive ability, they may present an opportunity for limited
deception by individuals with large signal residuals. It is this last
possibility that is of interest here.
If receivers assess competitive ability by reference to the population-based relationship between signal structure and competitive ability, and if signal residuals do not themselves correlate with competitive ability, then signal residuals provide some signalers with an opportunity to exaggerate their apparent or signaled competitive ability (Figure 1, arrows). In Figure 1, although A and B have equal competitive ability, the apparent or signaled value for A's competitive ability (A') is considerably higher than the apparent or signaled value for B (B'). Thus, although this signal is, on average, an honest signal of competitive ability, it overestimates competitive ability for individual A and understimates competitive ability for individual B.
We can test for deceptive use of the signal by determining whether the use of the signal depends on signal residuals. In competitive interactions involving equally able competitors, the signal produced by A will be more likely to deter an opponent than the signal produced by B; thus, use of the signal would be more advantageous for individual A than for B. In general, individuals with greater signal residuals may receive a greater benefit from the use of the signal than individuals with lower signal residuals. Thus, if signal exaggeration as reflected by positive signal residuals is exploited by some signalers to enhance their apparent competitive ability, then individuals with larger signal residuals are predicted to perform the signal more often than individuals with smaller signal residuals.
If the signal is produced only by signalers with positive signal residuals, the potential for receivers to be deceived is small because they are likely to assess the relationship between the signal and competitive ability using only this subset of the population. In Figure 1, then, the line used by the receiver to determine competitive ability from the signal would simply be shifted up. If all individuals produce the signal, however, and if receivers are not be able to reliably distinguish between signalers with positive and negative signal residuals by other means, receivers may be deceived, in that they may respond to the signaler as having a greater competitive ability than the signaler actually has.
If receivers respond to signalers with positive signal residuals as having
a greater competitive ability, individuals with positive signal residuals may
win interactions more often than would be predicted by competitive ability
alone, especially under conditions in which competitive ability is difficult
to assess by other means. Interactions involving individuals with positive
signal residuals may also be longer or more highly escalated; escalation to
more costly fighting and prolongation of the interaction are predicted when
assessment of relative competitive ability is difficult
(Parker, 1974;
Parker and Rubenstein, 1981).
When one or more of the individuals in a contest can exaggerate their
competitive ability using a signal, they may initially appear (to each other)
to be more closely matched than they are, leading to a higher level of
escalation or longer duration for the interaction than would otherwise be
expected.
Example: snapping shrimp
In the big-clawed snapping shrimp, aggressive interactions are usually won
by the larger animal (Hughes,
1996; Nolan and Salmon,
1970; Schein,
1977), and the likelihood that the smaller animal wins is greater
when the difference in body size is small
(Hughes, 1996). Although in
other decapod crustaceans chela size is often a significant component of
competitive ability (e.g., fiddler crabs, Uca annulipes:
Jennions and Blackwell, 1996;
shore crabs, Carcinus maenas:
Sneddon et al., 1997), chela
size does not affect the outcome of interactions between snapping shrimp once
the effects of body size are removed
(Hughes, 1996). Although rapid
closure of the major chela on the body of an opponent can inflict serious harm
(Hazlett and Winn, 1962), this
act is almost always preceded by considerable wrestling between opponents,
including grasping with the minor chela and rapid contractions of the abdomen,
as in tail flips (Hughes, personal observation); such fighting behaviors are
more likely to be limited by body size than by chela size, and may explain why
body size rather than chela size determines competitive ability in this
species. To confirm that chela size, independent of body size, does not
significantly contribute to competitive ability, the relationships between
body and chela size and competitive ability will be further tested here.
A visual display performed with the major chela—the open chela (or
cocked claw) display—is one of the most common signals in snapping
shrimp interactions; in this display, the signaler holds the major chela in a
fully open position, oriented toward the receiver
(Hughes, 1996;
Nolan and Salmon, 1970;
Schein, 1977). Chela size is a
function of body size (Hughes,
1996; Nolan and Salmon,
1970; Schein,
1975). Males respond to the open chela display as a function of
its relative size; that is, males respond more to an isolated open chela
display when this chela is similar in size to their own
(Hughes, 1996). Male response
to a closed chela, in contrast, was not a function of its relative size
(Hughes, 1996). Given that
body size is the primary component of competitive ability, these responses of
males to isolated chelae suggest that chela size in the open chela display is
used by males to obtain information regarding the body size of an
opponent.
The relationship between body size and chela size is strong
(r2 =.82; Hughes,
1996); chela size in the open chela display thus provides
receivers with a reliable estimate of competitive ability. Nonetheless, there
is variation in chela size that is independent of body size. If the
population-based relationship between body size and chela size is used by
receivers of the open chela display as a means of assessing body size, then
individuals with larger chelae than predicted by this relationship (i.e.,
larger "chela residuals") signal an overestimate of their body
size when they perform the display, while individuals with smaller chelae than
predicted (i.e., smaller "chela residuals") signal an
underestimate of body size. If the open chela display is an honest signal that
is open to low levels of deceptive use, individuals with large chela residuals
are predicted to use the open chela display more frequently than individuals
with smaller chela residuals. Note that because all individuals produce the
signal, regardless of chela residual, receivers may not be able to distinguish
between individuals with larger and smaller chela residuals, and so there is
the potential for receivers to be deceived.
In this study, I tested for deceptive use of an honest signal in snapping
shrimp. First, I confirmed that body size is the primary component of
competitive ability and that chela residuals do not affect the outcome of
competitive interactions. Second, I determined whether differences between
interactions in terms of duration and level of escalation are related to
differences in the chela residuals of the opponents. Finally, I determined
whether males with large chela residuals perform more open chela displays than
males with small chela residuals. In addition to examining the signal use of
males in the staged competitive interactions performed here, I also analyzed
the responses of shrimp presented with isolated chelae in a previous
experiment (Hughes, 1996). In
this experiment, male shrimp were presented with isolated open chelae, and the
number of open chela displays they produced was a function of the relative
size of the chela. I reanalyzed these data to determine whether the number of
open chela displays produced by these males was influenced by their chela
residuals.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
I collected snapping shrimp at low tide from oyster rubble near Beaufort, North Carolina, USA. Shrimp were collected during two successive spring tides in March and April, 1994, from the same location. Shrimp were housed individually with artifical seawater, gravel, and a piece of oyster rubble for shelter. Details of shrimp care and maintenance are given in Hughes (1994
,
1996). I measured body size (rostrum to telson) and chela size (dactyl to propodus) on all individuals. Relative body size was calculated as the ratio of the larger animal/smaller animal within an interaction. Relative chela size was calculated as the ratio of the chela size of the larger animal/chela size of smaller animal. Chela residuals were calculated as the residuals from the linear regression of chela size on body size, with a data set of 362 males collected over a period of 5 years from three collection sites in the Beaufort Inlet.
The shrimp used in the staged competitive interactions ranged in body size
from 2.2 to 4.0 cm and ranged in chela size from 1.0 to 2.3 cm. I selected
pairs of male shrimp at random, with the caveat that I did not allow pairs
which differed greatly in size. Relative body sizes (larger/smaller within
each interaction) ranged from 1.0 to 1.4, with a mean of 1.07. Relative chela
sizes ranged from 0.8 to 1.4, with a mean of 1.09. In most of the trials, I
could identify individuals by handedness (i.e., side with major chela), size,
or color differences; in a few trials, I marked one individual on the thorax
with a waterproof marker to facilitate identification during the interactions.
Within each interaction, one individual was classified as smaller and the
other as larger based on body size, or in cases where body sizes were equal,
on chela size. Given the range in sizes of animals used, a given body size may
be larger in one interaction but smaller in another; thus, although there are
trends toward larger body and chela sizes among animals classified as larger,
the ranges of these size measurements overlap extensively between animals
classified as larger and smaller within interactions
(Table 1). As a group, the
larger animals did not differ significantly in overall size from the smaller
animals (multivariate ANOVA of body size, chela size, and chela residuals,
Wilks'
= 0.923, F3,72 = 1.988, p
=.123).
|
I constructed four test chambers by modifying aquarium fish-breeding chambers (Hagen brand "Multi-chambers"). These chambers were 9 x 20 cm and had a removable, opaque divider in the center. Before each trial, I filled the test chambers with 700 ml artificial seawater and placed the divider in the center. Approximately 12 h before a trial, shrimp were placed in opposite sides of the chambers. All trials were performed between 0800 and 1100 h. At the beginning of a trial, I removed the divider and placed a piece of PVC pipe (6.5 cm long, 2 cm diam) in the middle of the chamber. The first few trials were videotaped for 1 h; examination of these trials revealed that shrimp usually completed several interactions within the first 15 min, so all subsequent trials were videotaped for that time. In 6 of the 44 trials I completed, the animals did not encounter one another within 15 min; thus the number of trials analyzed was 38. All behavioral data were collected from videotape.
In most of the trials, animals interacted several times within the 15-min period. I consider only the first interaction of a trial here in order to limit the analysis to the behavior of animals establishing a dominance relationship. I defined the winner of the interaction as the individual who did not retreat at the end of the first interaction.
Outcome of interactions
To determine whether chela residuals affect the outcome of interactions, I
first tested for differences in chela residuals between winners and losers
using a paired t test. Given that animals of larger body size usually
win competitive interactions (Hughes,
1996; Nolan and Salmon,
1970; Schein,
1977), it is possible that chela residuals only contribute to
competitive ability when the difference between body sizes is small. For this
reason, I also used a paired t test to test for differences in chela
residual between winners and losers in contests where the size difference
between the individuals was small (relative size 
1.045) using a paired
t test.
Finally, it is possible that differences in chela residuals play a role in determining when the smaller animal is able to win a competitive interaction. I tested for differences between smaller animals that won interactions and smaller animals that lost interactions using a t test. I performed this analysis twice—once including all interactions and once including only interactions between individuals of similar size, as above.
Escalation and duration of interactions
I classified the first interaction of each trial by three levels of
escalation, using the following criteria. (1) The low escalation category
included interactions resolved solely on the basis of signals, and in which
the animals never came into physical contact (except for antennae). (2) The
next level, or intermediate escalation, included interactions in which the
animals did come into contact, usually by laying their chelae on each other,
sometimes pushing against each other. (3) High escalation included those
interactions which were resolved with combat involving an immediate threat of
injury. For an interaction to be classified as highly escalated, at least one
of the animals needed to perform one of the following two behaviors: snapping
the major chela directly onto the body of the opponent or grabbing the
opponent with the minor chela and holding as the opponent struggled to get
away. These criteria for levels of escalation are not dependent on the
signaling behavior of the interacting animals; that is, the number and types
of signals produced by each animal do not affect classification by level of
escalation. All interactions were easily and unambiguously classified as to
level of escalation by these criteria.
I also measured the durations of the interactions. I used the frequency distribution of durations to assign interactions to two categories: short and long.
For each interaction there were eight size variables: the body size, chela
size, and chela residual for both the larger and smaller animal, and relative
body and chela size (larger/smaller within interactions). Because these
variables are likely to be highly correlated, I subjected them to a principal
components (PC) analysis, and used a subset of the resulting PC scores in
subsequent multivariate analyses of variance. I tested for normality and
equality of variances for each test. All statistics were performed using
Systat (Wilkinson, 1989).
Signaling behavior
Signaling in interactions
I tested whether the number of open chela displays produced was related to
the chela residual of the signaling animal using Spearman nonparametric
correlations. I performed this analysis first including all interactions, and
then including only the interactions in which the competitors were of similar
size (relative body size 1.045). For these analyses, the number of open
chela displays was log-transformed. Animals that did not perform any open
chela displays and or snaps throughout the entire trial (including the first
interaction and all subsequent interactions) were excluded from these
analyses; this criterion excluded one smaller animal and two larger
animals.
Signaling in response to isolated chelae
Hughes (1996) presented
shrimp with isolated chelae and measured their response as the number of open
chela displays produced within a 3-min period. The chelae presented were
recently molted from shrimp not used in the experiment and were coated with
shellac to provide a uniform chemical stimulus.
Here I analyzed the responses of 27 males to the isolated open chela, including only those individuals that produced at least one open chela display. I categorized these males by whether they had positive or negative chela residuals and, using a t test, tested for differences in the number of open chela displays performed by males with positive and negative chela residuals.
Finally, I tested whether males with positive chela residuals produced more
open chela displays than males with negative chela residuals when the effects
of the relative size of the presented chela were removed. That is, do
individuals with positive residual chela sizes produce more open chela
displays than would be predicted given the relative size of the chela they
encountered? Responding more or less than expected given relative chela size
can be quantified using response residuals—deviations from the
regression of response on relative chela size
(Hughes, 1996). Using a
t test, I determined whether males with positive chela residuals had
larger response residuals than males with negative chela residuals.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Principal component analysis
The first two components of the PC analysis accounted for more than 77% of the variance in the eight size variables. Table 2 shows the rotated loadings of each of the variables on these components. The first component accounted for 51.54% of the variance and has an eigenvalue of 4.1. This component was dominated by the four measures of absolute size (larger body size, smaller body size, larger chela size, smaller chela size). The second component accounted for 25.51% of the variance and has an eigenvalue of 2.0. This second component was dominated by relative body size and relative chela size. The next two components, which accounted for 12.55% and 9.85% of the variance, were heavily loaded by the smaller animals' chela residual and the larger animals' chela residual, but these components have very low eigenvalues (1.0 and 0.79, respectively). Because these components have low eigenvalues, and because other variables also load onto these components, they are not useful for determining the effects of chela residuals on interactions. Therefore, the four size variables used in subsequent analyses were PC1 (absolute size of both individuals), PC2 (relative size), chela residual of the smaller animal, and chela residual of the larger animal. By definition, the two principal components are orthogonal; neither the chela residual of the smaller animal nor the chela residual of the larger animal are correlated with either component (chela residual of smaller with PC1: p =.11; with PC2: p =.66; chela residual of larger with PC1: p =.14; with PC2: p =.95).
|
Classification of interactions
Of the 38 interactions, 13 were resolved by signals alone (low), 13 were
resolved with some form of physical contact but without combat (intermediate),
and 12 were resolved with behaviors that involve a high risk of injury
(high).
I found a normal distribution of durations less than 30 s (n = 16), with the remaining interactions scattered between 34 and 682 s (n = 22). For analyses by duration, interactions were classified as short (< 30 s) or long (> 30 s).
Outcome of interactions
There was no evidence that chela residuals affected which animal won a
competitive interaction (Figure
2). First, winners and losers did not differ in chela residuals
(t =.682, df = 37, p =.500,
Figure 2A). Second, winners and
losers did not differ in chela residuals when the difference in body sizes was
small (t = 0.633, df = 22, p =.533,
Figure 2B). Third, there was no
significant difference in chela residual between smaller animals that won
interactions and smaller animals that lost interactions (t = -0.116,
df = 33, p =.909, Figure
2C). Finally, there was no significant difference in chela
residuals between smaller animals that won and smaller animals that lost in
interactions between animals of similar size (t = -0.483, df = 21,
p =.634, Figure
2D).
|
There were 13 interactions in which one animal had the larger body size and the other animal had the larger chela size. Nine were won by the animal with the larger body, and four were won by the animal with the larger chela (exact binomial probability =.087), a result consistent with the interpretation that it is body size, rather than chela size, that is important to competitive ability.
Escalation and duration of interactions
Interactions of different levels of escalation involved individuals of
significantly different sizes (Wilks' = 0.548,
F8,64 = 2.808, p =.010). Relative size (PC2,
log-transformed for equality of variances) differed between the levels of
escalation (F2,35 = 3.91, p =.029), and post-hoc
comparisons revealed this difference to be between interactions of low and
high escalation (mean difference = 0.871, p =.022). In other words,
individuals in highly escalated interactions tended to be more similar in size
than individuals in interactions resolved by signals alone.
Chela residuals of the smaller individuals in interactions also differed between interactions of different levels of escalation (F2,35 = 4.758, p =.015, Figure 3). Interactions with high escalation involved smaller animals with significantly greater chela residuals than both intermediate escalation (mean difference = 0.082, p =.040) and low escalation (mean difference = 0.091, p =.021).
|
Neither absolute size (PC1, F2,35 = 0.962, p =.392) nor chela residual of the larger individual (F2,35 = 1.906, p =.164) differed between interactions of different levels of escalation.
Short and long interactions also involved individuals that differ
significantly in size (Wilks' = 0.631, F4,33 =
4.819, p =.004). The chela residuals of smaller individuals were
significantly greater in long interactions than in short interactions
(F1,36 = 16.196, p <.001,
Figure 4). The chela residuals
of larger individuals followed this trend (F1,36 = 3.913,
p =.056, Figure
4).
|
There was also a trend for the absolute size of the individuals (PC1) to be greater in long interactions (F1,36 = 3.879, p =.057). Relative size (PC2) did not differ between long and short interactions (F1,36 = 0.58, p =.452).
To summarize, highly escalated interactions involved individuals of similar size, and the smaller individuals in escalated interactions had greater chela residuals than the smaller individuals in less escalated interactions. Long interactions involved large individuals (in terms of the absolute size of both shrimp) and individuals (both larger and smaller) with larger chela residuals.
Signaling behavior
Signaling in interactions
When all interactions were considered, the number of open chela displays
produced by smaller animals was significantly related to chela residuals
(n = 37, rs =.30, p <.05,
Figure 5A). This relationship
was still evident if the two individuals with the highest responses were
omitted (n = 35, rs =.26, p =.069). In
contrast, the number of open chela displays produced by larger animals was not
related to chela residuals (n = 36, rs =.14,.10
< p <.25, Figure
5B).
|
When only interactions between animals of similar size (relative body size
1.045) were considered, the same results were obtained: the number of
open chela displays produced by smaller animals was significantly related to
chela residuals (n = 23, rs =.45, p
<.025, Figure 6A). This
relationship was still significant if the two individuals with the highest
responses were omitted (n = 21, rs =.40,
p <.05). The number of open chela displays produced by larger
animals was not related to chela residuals (n = 21,
rs =.15, p >.25,
Figure 6B).
|
Signaling in response to isolated chelae
When responding to an isolated open chela display
(Hughes, 1996), the number of
open chela displays produced was greater for males with positive chela
residuals than for males with negative chela residuals (t = 3.415, df
= 25, p =.002, Figure
7A). Males with positive and negative chela residuals did not
differ in terms of the relative size of the chela they encountered in this
experiment (t = 1.577, df = 25, p =.127). Thus, the
difference in number of open chela displays produced by males with positive
and negative chela residuals was not the result of a spurious correlation
between chela residuals and relative size in this experiment. Likewise, there
was no correlation between absolute size and chela residuals for the shrimp
used in this experiment: rs =.17, p =.41). In
fact, males with positive chela residuals had significantly higher response
residuals than males with negative chela residuals (t = -2.802, df =
25, p =.010, Figure
7B). Males with positive chela residuals performed more open chela
displays than would be predicted given the relative size of the chela they
encountered, whereas males with negative chela residuals performed fewer open
chela displays than would be predicted given relative chela size.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In snapping shrimp, from the perspective of receivers, the open chela display is an honest signal of competitive ability: body size is the primary component of competitive ability (Hughes, 1996
; Nolan and Salmon,
1970; Schein,
1977) (Figure 2);
chela size is a function of body size
(Hughes, 1996;
Nolan and Salmon, 1970;
Schein, 1975), and receivers
respond to the signal as a function of relative size (Hughes,
1994,
1996). Nonetheless, as
signalers, the use of the open chela display depends in part on the degree to
which the signal exaggerates competitive ability, as measured by positive
chela residuals. When responding to an isolated, open chela larger than their
own, the number of open chela displays produced by males depends not only on
the relative size of the presented chela
(Hughes, 1996), but also on
the chela residual of the responding male
(Figure 7A). In fact, males
with positive chela residuals produce more open chela displays than predicted
by the relative size of the presented chela, whereas males with negative chela
residuals produce fewer open chela displays than predicted by relative chela
size (Figure 7B). In
competitive interactions, where additional sources of information (including
direct visual observation of body size) are likely to be available, the effect
of chela residual on the signaling behavior of males is weaker but still
evident: smaller individuals in competitive interactions tend to produce more
open chela displays when they have larger chela residuals (Figures
5A,
6A). Thus, whereas receivers
take advantage of the strong relationship between chela size and body size,
using this relationship to assess the body size of competitors from the size
of the chela in the open chela display, signalers take advantage of variation
around this relationship such that individuals with a signal that exaggerates
their apparent competitive ability use the signal more often in competitive
interactions. Because chela residuals do not differ between winners and losers
of competitive interactions (Figure
2), differences between individuals in chela residuals do not
reflect differences in competitive ability. Thus the increased use of the open
chela display by individuals for which the display exaggerates apparent size
represents deceptive use of an honest signal. This increased use of the open chela display by individuals for whom the display exaggerates apparent size only occurs when that individual is the smaller individual in the competitive interaction (Figures 5, 6). The label "smaller animals" applies only within each interaction: as a group, animals that were larger or smaller within interactions were almost entirely overlapping in measures of size (Table 1). Thus it is not the case that all animals with positive chela residuals always use the open chela display more than animals with negative chela residuals, or that all animals with positive chela residuals who are below some size threshold always use the signal more. Chela residuals provide an opportunity for a subset of the population to use a signal deceptively, but whether they do so depends on whether they are at a disadvantage in the interaction—that is, whether they are smaller. In other words, deceptive use of the open chela display depends not only on signal residual, but also on context. That only the smaller animals in interactions use the open chela display deceptively may reflect the disadvantages of being smaller; an animal that is larger does not need to bluff or exaggerate its size.
A shrimp's assessment of whether it is larger or smaller than its
competitor is likely to be based, at least in part, on the relative chela size
of the competitor (Hughes
1994,
1996), as assessed from the
opponent's initial open chela display. When the competitors are closely
matched and the opponent has a large chela residual, this initial assessment
of relative size is likely to be inaccurate. Nonetheless, even when
competitors are closely matched for size (and thus the likelihood that a
larger animal would mistakenly assess its competitor as being larger is
highest), the production of open chela displays by larger individuals is not
influenced by chela residual (Figure
6B). Shrimp are likely to obtain additional information about body
size as the competition progresses; in particular, in these experiments,
shrimp are likely to be able to directly observe the body size of their
opponents. As the shrimp's assessment of the relative size of its opponent
changes, the degree to which the production of open chela displays is
influenced by chela residuals may also change.
If increased use of the open chela display by animals for whom the display
exaggerates apparent competitive ability represents deceptive use of the
signal, it is logical to ask if receivers are deceived. Proving successful
deception (and distinguishing successful deception on the part of the signaler
from successful assessment of independent measures of competitive ability on
the part of the receiver) is not a trivial matter. It is clear from the
experiments presented here, however, that chela residuals do affect the course
of competitive interactions in a manner consistent with receivers being
deceived at the outset of the interactions. Highly escalated interactions are
characterized by larger chela residuals for the smaller individual
(Figure 3); long interactions
are characterized by larger chela residuals for both the larger and smaller
individuals (Figure 4). It is
not the case that interactions involving competitors with large, positive
chela residuals are longer and more escalated as a result of a spurious
correlation between chela residuals and relative size, as neither the chela
residuals of the larger or smaller individual were correlated with relative
size. Nor is it the case that individuals with large positive chela residuals
are in some way more generally aggressive: chela residual has no effect on the
production of snaps (another common signal in competitive interactions, and a
potentially damaging behavior if performed against the body of the competitor;
Hughes, 1994), and there is no
effect of chela residual on the ultimate outcome of the interaction
(Figure 2), as would be
predicted if positive chela residuals were associated with greater aggressive
behavior in general. In short, the effect of chela residual on the behavior of
snapping shrimp is specific to the production of the open chela display, not
to other aspects of aggressive behavior. Given that the size of the chela in
the open chela display is used to assess body size
(Hughes, 1996) and that
smaller animals with large chela residuals perform more open chela displays
(Figures 5A,
6A,
7), interactions involving a
smaller individual with a large chela residual may be longer or more escalated
because the smaller animal appears (at least initially) to the larger animal
as more similar to its own size.
If receivers are deceived, this deception was limited to the early stages
of these interaction; chela residual does not effect who wins
(Figure 2). These interactions,
however, were designed to maximize the likelihood that the individual with
true higher competitive ability would win; animals were confined to a small
space, and interactions occurred in clear water under bright light. These
conditions are unlikely to characterize the conditions under which the
big-clawed snapping shrimp frequently interact with conspecifics; this species
inhabits burrows around oyster shells in the muddy inter- and subtidal zones,
and although it is active throughout the day, it is most active at dawn and
dusk (Nolan and Salmon, 1970).
In low light and/or cloudy water, the bright white spot on the plunger of the
chela, which is visible only in the open chela display, may be the most
visible feature on the otherwise dark-colored shrimp. Under such conditions,
other sources of information regarding competitive ability (such as direct
visual observation of body size) may be highly limited; it is possible, then,
that under these conditions, increased use of the open chela display by
individuals for whom the display exaggerates apparent size may result in
greater likelihood of winning interactions. In other words, the degree to
which the deceptive use of the signal results in successful deception may
depend on the degree to which alternative sources of information regarding
competitive ability are available to the receiver. Further experimental work
is necessary to explore this possibility. At present, the degree to which
deceptive use of the open chela display results in successful deception of
receivers, and the conditions necessary for successful deception, are not
clear.
Not surprisingly, relative and absolute size influence the escalation and
duration of competitive interactions between snapping shrimp. As predicted by
theoretical models (Enquist and Leimar,
1983; Parker,
1974; Parker and Rubenstein,
1981), interactions between animals of similar size are
significantly more likely to escalate and involve risk of injury. Although
snapping shrimp of similar size escalate in interactions, relative size does
not affect the duration of interactions. In contrast, long and short
interactions tend to differ in the absolute size of the competitors: long
interactions involve larger shrimp than short interactions. Longer competitive
interactions are predicted to occur when the value of the resource is high,
and the costs of continuing the interaction are less than the costs of
locating a new resource (Enquist and Leimar,
1987,
1990;
Parker and Rubenstein, 1981).
Adult snapping shrimp range in body size from < 2.0 cm to > 4.0 cm and
in chela size from < 1.0 cm to > 2.0 cm; given this substantial range of
size, larger shrimp may be significantly better able to pay the costs of
prolonged interactions and also may have fewer options for adequate shelter or
mates.
The data presented here demonstrate that signal residuals can be useful for understanding variation between individuals in signaling behavior and for detecting deceptive use of honest signals. I focused on a signal used to assess competitive ability, but this approach would be equally useful for understanding the function of other kinds of signals. For example, if a signal is used by females to assess a particular characteristic of males as potential mates, we can then ask whether signal residuals affect the signaling behavior of males. In other words, do males exploit variation in signals that is independent of information assessed by females? In general, if receivers take advantage of a relationship between variation in a signal and variation in a characteristic of the signaler (Figure 1), signal residuals may provide an opportunity for limited deception by some signalers.
Although it is, in the study of animal communication, convenient to refer to signals as being either honest or deceptive, such a dichotomy may be too simplistic. In snapping shrimp, the strong relationship between chela size and body size benefits receivers because it allows them to assess competitive ability without direct combat. Nonetheless, signaling behavior appears to have evolved to take advantage of individual variation in the signal that is independent of competitive ability, such that individuals for whom the open chela display exaggerates apparent size use the display more often when they are at a disadvantage in an interaction. From the perspective of the receiver, the open chela display is an honest signal of body size; from the perspective of the signaler, there still exists a narrow range of opportunity for deceptive signaling. To state that the open chela display is either honest or deceptive, then, would be incomplete; rather, both signaler and receiver have evolved to use variation in signal structure to their own advantage. Signal residuals provide a useful mechanism for understanding the relationship between signal structure and function from the perspectives of both signaler and receiver. Our understanding of the function of animal signals may be best served by leaving this false dichotomy of honesty versus deception behind.
|
ACKNOWLEDGEMENTS |
|---|
Steve Nowicki, Susan Peters, and the rest of the Nowicki lab group provided much support and feedback during the development of the work presented here. I also thank Martin Beebee, Steve Nowicki, Dan Rubenstein, and an anonymous reviewer for their comments on this manuscript. This work was supported by a Sigma Xi Grant-in-Aid of Research.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Adams ES, Caldwell RL, 1990. Deceptive communication in asymmetric fights of the stomatopod crustacean Gonodactylus bredini. Anim Behav 39: 706-716.[Web of Science]
Adams ES, Mesterton-Gibbons M, 1995. The cost of threat displays and the stability of deceptive communication. J Theor Biol 175: 405-421.
Bond AB, 1989. Toward a resolution of the paradox of aggressive displays: I. Optimal deceit in the communication of fighting ability. Ethology 81: 29-46.
Dawkins MS, Guilford T, 1991. The corruption of honest signalling. Anim Behav 41: 865-873.
Enquist M, 1985. Communication during aggressive interactions with particular reference to variation in choice of behaviour. Anim Behav 33: 1152-1161.
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, 1990. The evolution of fatal fighting. Anim Behav 39: 1-9.
Getty, T, 1997. Deception: the correct path to enlightenment? Trends Ecol Evol 12: 159-160.
Gardner R, Morris MR, 1989. The evolution of bluffing in animal contests: an ESS approach. J Theor Biol 137: 235-243.
Hazlett BA, Winn HE, 1962. Sound production and associated behavior of Bermuda crustaceans (Panulirus, Gonodactylus, Alpheus, and Synalpheus). Crustaceana 4: 25-38.
Hughes M, 1994. Communication in snapping shrimp (Alpheus heterochaelis): signals, signaling behavior, and the responses of receivers (PhD dissertation). Durham, North Carolina: Duke University.
Hughes M, 1996. Size assessment via a visual signal in snapping shrimp. Behav Ecol Sociobiol 38: 51-57.
Jennions MD, Backwell PRY, 1996. Residency and size affect fight duration and outcome in the fiddler crab Uca annulipes. Biol J Linn Soc 57: 293-306.
Maynard Smith J, Harper DGC, 1988. The evolution of aggression: can selection generate variability? Phil Trans R Soc Lond B 319: 557-570.
Nolan BA, Salmon M, 1970. The behavior and ecology of snapping shrimp (Crustacea: Alpheus heterochelis and Alpheus normanni). Form Funct 2: 289-335.
Parker GA, 1974. Assessment strategy and the evolution of fighting behaviour. J Theor Biol 47: 223-243.[Web of Science][Medline]
Parker GA, Rubenstein DJ, 1981. Role assessment, reserve strategy, and acquisition of information in asymmetric animal conflicts. Anim Behav 29: 221-240.
Schein H, 1975. Aspects of the aggressive and sexual behaviour of Alpheus heterochaelis Say. Mar Behav Physiol 3: 83-96.
Schein H, 1977. The role of snapping in Alpheus heterochaelis Say, 1818, the big-clawed snapping shrimp. Crustaceana 33: 182-188.
Sneddon LU, Huntingford FA, Taylor AC, 1997. Weapon size versus body size as a predictor of winning in fights between shore crabs, Carcinus maenas (L.). Behav Ecol Sociobiol 41: 237-242.
Steger R, Caldwell RL, 1983. Intraspecific deception
by bluffing: a defense strategy of newly molted stomatopods (Arthropoda:
Crustacea). Science 221:
558-560.
van Rhijn JG, Vodegel R, 1980. Being honest about one's intentions: an evolutionary stable strategy for animal conflicts. J Theor Biol 85: 623-641.[Web of Science][Medline]
Wagner Jr. WE, 1992. Deceptive or honest signalling of fighting ability? A test of alternative hypotheses for the function of changes in call dominant frequency by male cricket frogs. Anim Behav 44: 449-462.
Wiley RH, 1983. The evolution of communication: information and manipulation. In: Animal behaviour, vol. 2, Communication (Halliday TR, Slater PJB, eds). Oxford: Blackwell; 156-189.
Wilkinson L, 1989. Systat: the system for statistics. Evanston, Illinois: Systat, Inc.
Zahavi A, 1977. The cost of honesty (further remarks on the handicap principle). J Theor Biol 67: 603-605.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. J Whiting, J. K Webb, and J. S. Keogh Flat lizard female mimics use sexual deception in visual but not chemical signals Proc R Soc B, May 7, 2009; 276(1662): 1585 - 1591. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. S. Wilson, R. S. James, C. Bywater, and F. Seebacher Costs and benefits of increased weapon size differ between sexes of the slender crayfish, Cherax dispar J. Exp. Biol., March 15, 2009; 212(6): 853 - 858. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Arnott and R. W Elwood Fighting for shells: how private information about resource value changes hermit crab pre-fight displays and escalated fight behaviour Proc R Soc B, December 7, 2007; 274(1628): 3011 - 3017. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Briffa Signal residuals during shell fighting in hermit crabs: can costly signals be used deceptively? Behav. Ecol., May 1, 2006; 17(3): 510 - 514. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||