Antisymmetry in male fiddler crabs and the decision to feed or breed

Authors


Summary

  • 1 In male Sand Fiddler Crabs, Uca pugilator, a major cheliped (with claw), used in intersexual displays and intrasexual contests, develops opposite a minor cheliped used for feeding. Cheliped size demonstrates antisymmetry because greater development is equally likely on the right or left side.
  • 2 The side with the major cheliped (major side) also has longer walking legs which may facilitate use of the claw. In contrast, eye stalk asymmetry is equally due to antisymmetry and fluctuating asymmetry. Fluctuating asymmetry is a subtle, non-adaptive departure from the population-level trajectory relating growth on major and minor sides.
  • 3 In a South Carolina (USA) marsh, cheliped and leg antisymmetries are greater and eye stalk asymmetry is less among males able to defer feeding in favour of breeding. However, the composition of up-slope breeding and down-slope feeding subpopulations changes across the lunar cycle.
  • 4 The number of mates sequestered in breeding burrows is positively correlated with cheliped and leg antisymmetry and negatively correlated with eye stalk asymmetry. Male fitness is a function of the product of time spent breeding and the number of mates per unit time while breeding. Both fitness components are predicted by relative cheliped antisymmetry and eye stalk fluctuating asymmetry, which are themselves significantly negatively correlated.

Introduction

A goal of evolutionary ecology is to determine the nature of variation among individuals that is associated with differences in resource use and reproductive success. Recently, fluctuating asymmetry, a subtle departure from bilateral symmetry, has received tremendous interest as a measure of individual quality because it may reveal the extent to which genes fail to control growth and development, especially in harsh environments (Møller & Swaddle 1998; Simmons et al. 1999). Thus, fluctuating asymmetry is a non-adaptive asymmetry. Other bilateral asymmetries may be adaptive, such as directional asymmetry, where greater development of a structure occurs on a particular body side, and antisymmetry, where greater development is equally likely on the right or left side (Graham, Freeman & Emlen 1993).

In male fiddler crabs (genus Uca), one cheliped bears a greatly enlarged as claw (= major cheliped) that is waved to attract receptive females (Crane 1975; Pope 2000) and employed in agonistic contests for control of breeding burrows (Crane 1975; Hyatt & Salmon 1978). The minor cheliped is used to scoop up moist sediment containing the bacteria, algae and decomposition products on which they feed (Miller 1961). Cheliped differentiation represents adaptive antisymmetry (Rowe, Repasky & Palmer 1997) as claw development is random with respect to body side (Vernberg & Costlow 1966). Eye stalks and walking legs of fiddler crabs may also exhibit antisymmetry, being longer on the major side (i.e. side possessing the claw) (Crane 1975).

The asymmetry of chelipeds, eye stalks and walking legs among males of the Eastern Sand Fiddler Crab, Uca pugilator (Bosc) was measured. Uca pugilator is abundant in marshes along the eastern coast of the USA from Massachusetts to Florida (Miller & Vernberg 1968). At our South Carolina study site, adults of U. pugilator feed on down-slope sand flats, which are moistened two times per day. They feed in droves of hundreds to thousands of individuals (Crane 1975). Droves arise as individuals move from marsh margins onto sand flats after the recession of high tides (Salmon & Hyatt 1983). Crabs do not mate while droving (Hernnkind 1968; Crane 1975; Salmon & Hyatt 1983; Salmon 1987). Breeding occurs on up-slope marsh margins. Only unusually high spring tides, which occur near new and full moons, inundate breeding areas. Infrequent inundation means that burrows, where mating and egg brooding occur, are less likely to collapse.

The size and sex of U. pugilator individuals comprising droves or the breeding subpopulation vary across seasons and the lunar cycle (Pratt, McLain & Kirschstein 2002). Males holding more desirable breeding burrows are less likely to abandon them to enter a feeding drove (Hyatt 1977; Christy 1983; Salmon & Hyatt 1983).

Thus, at any given moment the proportion of high-quality males that are breeding should be higher than that of low-quality males. Consequently, phenotypes associated with male quality would vary among breeding and non-breeding subpopulations. In the present study, we ask if the degree of claw, walking leg or eye stalk asymmetry is condition-dependent, varying among the fluid subpopulations of waving (= breeding) and droving crabs. We also ask if the contribution of fluctuating asymmetry to bilateral asymmetry of these antisymmetrical structures varies by male activity and if the degree of antisymmetry of cheliped development is correlated with the degree of fluctuating asymmetry of eye stalks.

Materials and methods

LOCATION AND COLLECTION

Droving crabs were collected in 1997 to assess claw, walking leg and eye stalk antisymmetries as functions of size and sex. With respect to droving vs waving, claw antisymmetry was measured in 1998–2000, eye stalk antisymmetry was measured in 1998 and 2000, and walking leg antisymmetry was measured in 2000. Measurements were taken on crabs collected on 17 and 24 May in 1997, on 6, 18 and 20 September in 1998, on 7, 14, 21, 28 May, 4 June and 11 September in 1999, and on 16, 21, 28 April and 5 July in 2000. On these and 14 other dates, claw and body size were measured to relate breeding synchrony to the lunar phase.

Uca pugilator were collected from the northern section of Cat Island marsh in Beaufort, South Carolina. This 3-ha section consists of open sandy flats where droving occurs. Smooth Cord Grass, Spartina alterniflora Loisel, occurs on the down-slope margin of sand flats. Needle Rush, Juncus roemarianus Scheele, occurs on up-slope margins. The marsh rises steeply at an embankment along its eastern edge, which is where breeding males occur.

In 1997, males and females were hand-picked from droves in an effort to sample evenly across body size (carapace width). In following years, random samples were obtained from droves by retrieving individuals that had fallen into pit traps composed of plastic guttering measuring 2 m long × 15 cm wide × 10 cm deep. Breeding males were sampled by the exhaustive collection of all waving individuals at points on the embankment. Waving males were captured by blocking entrances to breeding burrows with a trowel when crabs emerged to display.

After collection, fiddler crabs were placed in 45·5-litre coolers containing estuarine water (salinity = 30 ppt) and taken to the laboratory where they were maintained on a diet of TetraMin tropical fish flakes (TetraWerke, Melle, Germany) for 1–3 days. Prior to measurement, crabs were cooled on ice to arrest movement. Once measured, live crabs were returned to the marsh.

Some breeding burrows were excavated in 1998 (14 March, 5 and 11 April) and in 2000 (16 and 21 April) to relate antisymmetry to mating success. Each male and any females sequestered in internal chambers were collected. On other dates, females associated with breeding males were collected when present at the burrow entrances.

STRUCTURES MEASURED

Body size was indexed by the width of the carapace between the antero-lateral angles (just posterior to the eye stalks) (see Crane 1975). The length of the propodus was measured to index the size of claws. The propodus is the larger of the two elements of the pincer. The merus (proximal long shaft) of the first pair of legs was measured to index leg length. In 1997, it was determined that the difference in length between major and minor sides of the first pair of walking legs of males was primarily attributable to variation in the length of the merus as opposed to the combined length of all more distal segments (carpus, manus and dactyl) (R2 = 0·662, F = 152·827, df = 1,78, P < 0·001). The length of the eye stalk was measured from the proximal end of the longer (second) segment to the distal tip of the ommatidia.

Propodus lengths of the major cheliped and carapace width were measured to the nearest 0·1 mm using digital callipers. Measurements of the smaller structures were made with an ocular micrometer to the nearest 0·026 mm except in 1997 when they were made to the nearest 0·053 mm. Measurements were made blind with regard to whether crabs were from droves or breeding burrows. Each measurement consisted of taking two to three readings then recording the value that recurred. All measurements were repeated. Average values of measurements were used in statistical analyses.

Asymmetry was quantified as the absolute value of the difference between right and left side measures divided by the mean of right and left side measures, ?right – left?/[(right + left)/2]. The denominator represents correction for size-dependence of the between-sides difference.

COMPONENTS OF ASYMMETRY

The distribution of trait asymmetry may represent contributions from directional asymmetry, fluctuating asymmetry and antisymmetry (Graham et al. 1998), each of which may be modelled as one or more normal distributions that differ in variance or in location of the mean (Van Dongen, Lens & Molenberghs 1999). Thus, fluctuating asymmetry is modelled as a normal distribution centred at 0, directional asymmetry as a normal distribution with a non-zero mean, and antisymmetry as two normal distributions that are identical except that they have non-zero means of opposite sign. The actual distribution is approximated by the weighted sum of component normal distributions (Lens & Van Dongen 2000).

To estimate the contributions to eye stalk asymmetry, the combination was selected of a single normal distribution centred at zero (fluctuating asymmetry component) and paired normal distributions with oppositely signed non-zero means (antisymmetry component) whose weighted sum was the best fit to the observed distribution. If necessary, one directional contribution was employed. Asymmetry was measured as the difference, length of the eye stalk on the major body side minus the length on the minor side. As the data sets were small (N = 276 in 1998 and N = 253 in 2000), the range of values was subdivided into 11 equally wide zones. The number of observations expected in each zone was calculated for normal distributions based on their weighted contribution and SD. The best model was the set of distributions that minimized the sum of the square of the deviation between actual and expected numbers of observations in each zone. The magnitude of deviations of the best fitting set was then compared with a Kruskal–Wallis test to the fit of the best distributions for 100% fluctuating asymmetry and 100% antisymmetry.

A complementary approach was used to determine if the contribution of fluctuating asymmetry varied by male activity. Here, fluctuating asymmetry and measurement error are envisioned as producing variation about the developmental trajectory relating left and right sides of a bilateral structure (Graham et al. 1998). Residual variances about the major axis regression estimate of the developmental trajectory (see Graham et al. 1998) were compared between male groups. Linear major axis regressions were used because residuals did not increase with size, suggesting that an additive, not multiplicative, error model was appropriate (Graham et al. 1998).

MEASUREMENT ERROR AND STATISTICS

Measurement error may lead to overestimates of asymmetry (Palmer 1994). Measurement error was evaluated by testing the null hypothesis that the absolute size difference between body sides was not different from the mean absolute difference between repeated measures (Table 1).

Table 1.  Mean difference between major side and minor sides (mm) of males (= asymm) by year and measurement error for major and minor body sides. Measurement error is the difference between duplicate measures (mm). t= test statistic for the null hypothesis that either (1) the mean of the between-sides difference minus mean measurement error is 0 or (2) the correlation between the body sides difference and mean character size is 0. Null hypothesis is rejected for t > 1·97
 PropodusEye stalkMerus
 MajorMinorAsymmtMajorMinorAsymmtMajorMinorAsymmt
1997N = 203   N = 210   N = 193   
Mean0·0320·028  9·958 30·10·0120·0100·037 5·30·0130·0090·60222·9
SD0·0390·035  4·704 0·0230·0210·042 0·0160·0150·361 
Correlation    99·9    0·5   11·3
1998N = 269   N = 276   N = 0   
Mean0·0660·064 14·678 86·40·0160·0170·14420·1    
SD0·0510·072  2·780 0·0330·0350·117     
Correlation    63·1   –1·7    
1999N = 1057   N = 0   N = 0   
Mean0·0500·043 12·126 97·0        
SD0·0670·041  4·063         
Correlation   289·9        
2000N = 418   N = 253   N = 253   
Mean0·0450·038 13·404 65·50·0150·0160·08316·90·0820·0840·87829·1
SD0·0690·066  4·168 0·0360·0350·064 0·0930·0890·434 
Correlation   138·3    2·6   14·5

Multivariate linear hypotheses were tested using Systat (Wilkinson 1988). Asymmetry values were compared by using activity, droving vs waving, as a categorical or grouping variable and carapace width as a covariate to control for effects of body size.

Results

MEASUREMENT ERROR

Measurement error was significantly smaller than between-sides variation for all sets of measurements (Table 1). The absolute value of the between-sides difference was consistently significantly correlated with mean character size for the propodus and merus (Table 1). The correlation between the eye stalk between-sides difference and mean size was significant in 2000 and marginally significant in 1998. These results justify correction for character size in the estimation of asymmetry.

COMPONENTS OF ASYMMETRY

Asymmetries of the propodus (Fig. 1), merus (Fig. 2) and eye stalk (Fig. 3) were not normally distributed among droving and waving adult males. The contribution of fluctuating asymmetry to between-sides variation in propodus and merus length was small as the bulk of the distributions was displaced far from zero. Eye stalk asymmetry in 1998 was best approximated by multiple normal distributions that suggested 13·8% directional asymmetry, 47·4% fluctuating asymmetry and 38·8% antisymmetry. The mixed asymmetry model provided significantly better fit to the data than did models based exclusively on either fluctuating asymmetry or antisymmetry (Kruskal–Wallis test, χ2 = 6·324, df = 2, P = 0·042). Eye stalk asymmetry in 2000, which was less than in 1998 (Table 1), was best approximated by a model with 43% fluctuating asymmetry and 57% antisymmetry. This mixture model also explained the data better than pure fluctuating asymmetry or pure antisymmetry models (Kruskal–Wallis test, χ2 = 7·969, df = 2, P = 0·019).

Figure 1.

Distribution of size-corrected antisymmetry of the propodus of chelipeds among droving and waving adult male fiddler crabs collected from 1998 to 2000. As the major cheliped (claw) is on the left side approximately 50% of the time, the distribution of signed (±), not absolute value, antisymmetries is bimodal.

Figure 2.

Distribution of size-corrected antisymmetry of the merus of the first pair of walking legs among droving and waving adult male fiddler crabs collected in 2000. As the longer merus is on the left side approximately 50% of the time, the distribution of signed (±), not absolute value, antisymmetries is bimodal.

Figure 3.

Distribution of size-corrected antisymmetry of eye stalks among droving and waving adult male fiddler crabs collected in 1999 and 2000. As the longer eye stalk is on the left side approximately 50% of the time, the distribution of signed (±), not absolute value, antisymmetries is bimodal and significantly non-normal (Lilliefors test, P < 0·001). A similar distribution results for non-size corrected asymmetry (see Materials and methods).

SIZE AND ASYMMETRY vs SEX

In 1997, the length of the propodus, merus and eye stalk were significant allometric functions of carapace width, length = b(carapace width)a (Table 2). Functions were estimated for major and minor sides in males and for the mean of right and left sides in females. The allometric constant, a, was close to 1·000 for all structures except the major propodus of males where length increased almost with the square of carapace width. The constant, a, was greater in males than in females for all structures (Table 2). Consequently, the length of structures was greater in males than females of a given width (Table 3).

Table 2.  Allometric constants estimated for males (major and minor body sides) and females (mean of left and right sides) in 1997. The equation is: Y = b(carapace width [mm])a
 baR2FdfP
Propodus
Male – major side0·0971·9810·95015110·4542,201<0·001
Male – minor side0·3751·0370·95636831·3822,201<0·001
Female – mean0·4150·9740·97524134·1282,90<0·001
Merus
Male – major side0·4171·0930·95230342·6452,191<0·001
Male – minor side0·4111·0680·91015787·2202,191<0·001
Female – mean0·5070·9720·94810937·9842,90<0·001
Eye stalk
Male – major side0·3920·9070·97477484·4502,199<0·001
Male – minor side0·3940·9070·97681734·0522,199<0·001
Female – mean0·3940·8940·98543587·2312,90<0·001
Table 3.  Analysis of sex differences in the mean length and antisymmetry of the propodus, merus and eye stalk employing carapace width as the covariate. The comparison of propodus length is limited to feeding chelipeds. R2 = variation explained by the model incorporating both sex and carapace width
 Source of variationLengthAntisymmetry
 FdfPR2FdfPR2
PropodusWidth 6833·4491,292<0·0010·9623104·2851,292<0·0010·932
 Sex   56·2371,292<0·001  374·6301,292<0·001 
MerusWidth 6035·0801,281<0·0010·959 223·3211,281<0·0010·527
 Sex   81·9021,281<0·001   45·6371,281<0·001 
EyestalkWidth12316·4001,290<0·0010·978  17·1571,290<0·0010·056
 Sex   34·8491,290<0·001    0·4381,290 0·509 

Asymmetry of the propodus and merus, but not the eye stalk, was significantly greater in males than females (Table 3). The larger merus tended to be on the same body side as the larger propodus in males (χ2 = 46·078, df = 2, P < 0·001) but not in females (χ2 = 3·397, df = 4, P = 0·494). There was no tendency for longer eye stalks to be on the same body side as the longer propodus in either males (χ2 = 1·507, df = 2, P = 0·471) or females (χ2 = 3·595, df = 4, P = 0·464).

PROPODUS ANTISYMMETRY AND DROVING vs WAVING

Propodus antisymmetry was greater for waving than droving males in 1998–2000 (Table 4). This result is not attributable to differences in the size of the minor propodus. In 1998 and 1999, the size of the propodus of the minor cheliped did not vary significantly by male activity when body size was statistically controlled (Table 4). In 2000, the minor propodus averaged slightly but significantly larger among waving males, reducing their propodus antisymmetry.

Table 4.  Effect of carapace width and male type on propodus antisymmetry, size of minor propodus or size of major propodus. Male type: waving or droving
  199819992000
  WidthTypeWidthTypeWidthType
  1. Year 2000 statistics includes males of carapace width under and over 17·22 mm. Without males over 17·22 mm, values for super-scripted entries are:a380·33, b12·63, cd1,367, e0·618, f1999·66, g51·80, hi1,367, j0·892. See text for possible justification of separate analysis.

Propodus antisymmetryF130·6751·131178·7120·63413·81a13·73b
 df1,2651,2651,9642,9641,412c1,412d
 P<0·001<0·001<0·001<0·001<0·001<0·001
  R2 = 0·422 R2 = 0·598 R2 = 0·627e 
Minor propodusF357·521·145927·570·154120·245·18
 df1,2661,2661,9641,9641,4151,415
 P<0·0010·287<0·0010·704<0·0010·016
  R2 = 0·669 R2 = 0·870 R2 = 0·934 
Major propodusF1222·4254·7510061·9914·992506·65f57·62g
 df1,2651,2651,9641,9641,412h1,412i
 P<0·001<0·001<0·001<0·001<0·0010·016
  R2 = 0·832 R2 = 0·922 R2 = 0·908j 

Developmental trajectories for the propodus varied significantly by male activity in 1998 (t = 7·243, N = 276, P < 0·001), 1999 (t = 2·881, N = 1055, P = 0·004) and 2000 (t = 5·532, N = 253, P < 0·001). However, there were no significant differences in fluctuating asymmetry of propodus length as residual variances of the linear major axis regression did not vary by male activity in 1998 (F = 0·220, df = 1,266, P = 0·639), 1999 (F = 2·504, df = 2,987, P = 0·082) or 2000 (F = 2·920, df = 1,251, P = 0·089).

The length of the major propodus was significantly longer (when body width was controlled) in wavers than in drovers (Table 4). Allometric curves based on pooled samples of wavers and drovers underestimated claw size in wavers and overestimated claw size in drovers. Thus, residuals were on average positive for breeding males and negative for drovers (Table 5).

Table 5.  Ratio of actual to predicted length of the major side propodus (mm) for waving and droving males and the mean difference (breeding males – droving males) in model residuals. SE in parentheses. R2 and N are explained variation and sample size for allometric relationships
YearEstimated allometric curveRatiosResidual (mm)
WavingDroving
1998Propodus = 0·233 carapace1·644R2 = 0·807, N = 2681·044 (0·006)0·980 (0·005)1·351 (0·159)
 Range for carapace = 11·4–18·3 mm   
1999Propodus = 0·198 carapace1·701R2 = 0·910, N = 9741·009 (0·003)0·983 (0·004)0·326 (0·090)
 Range for carapace = 10·0–19·3 mm   
2000Propodus = 0·142 carapace1·824R2 = 0·918, N = 4181·025 (0·005)0·982 (0·010)1·071 (0·160)
 Range for carapace = 10·0–18·6 mm   

PROPODUS vs. OTHER ANTISYMMETRIES

In 1998 and 2000, eye stalk asymmetry among adult males was (1) significantly negatively correlated with propodus asymmetry, (2) significantly smaller among wavers than drovers, but (3) not significantly correlated with carapace width (Table 6). The same statistical relationships held for the absolute value of the difference in length between right and left eye stalks (i.e. asymmetry not corrected for size dependence; Table 6). The fluctuating asymmetry component of eye stalk asymmetry, measured by residuals of the developmental trajectory, was inversely proportional to propodus asymmetry in 1998 (F = 6·165, df = 1,266, P = 0·014) and 2000 (F = 14·638, df = 1,248, P < 0·001).

Table 6.  Male type (waving or droving), carapace width, and propodus antisymmetry as sources of variation for eye stalk and merus antisymmetry. Below, antisymmetry refers to correction for size dependence (as elsewhere; see Materials and methods) while asymmetry refers simply to the absolute value of the between-sides difference (included due to the inconsistent relationship between eye stalk size and asymmetry; see Table 1). R2 is explained variation for the model: antisymmetry = constant + male type + carapace width + propodus antisymmetry
 Source of variationR2
 Male typeWidthPropodus
 FPFPFP
Eye antisymmetry (1998; df = 1,264) 7·982 0·005 0·488 0·48516·090<0·0010·174
Eye asymmetry (1998; df = 1,264) 7·894 0·005 0·984 0·32223·920<0·0010·166
Eye antisymmetry (2000; df = 1,246)14·314<0·001 0·534 0·46512·609<0·0010·222
Eye asymmetry (2000; df = 1,246)14·632<0·001 4·003 0·047 9·443<0·0010·143
Merus antisymm (2000; df = 1,246) 1·768 0·18529·651<0·001 1·857 0·1740·228

The mean length of eye stalks did not vary between wavers and drovers (1998 –R2 = 0·796; effect for male type: F = 0·417, df = 1,273, P = 0·519; effect for carapace width: F = 1061·325, df = 1,273, P < 0·001. 2000 –R2 = 0·904; effect for male type: F = 0·634, df = 1,250, P = 0·426; effect for carapace width: F = 1605·130, df = 1,250, P < 0·001). Also, developmental trajectories were similar by male activity (1998: t = 0·208, N = 276, P = 0·835; 2000: t = 0·861, N = 253, P = 0·390) but residual variances that reflected fluctuating asymmetry differed significantly by activity (1998 –F = 26·302, df = 1,274, P < 0·001; 2000 –F = 24·699, df = 1,251, P < 0·001).

In 2000, merus asymmetry was not significantly correlated with either propodus antisymmetry or male activity, but was significantly correlated with carapace width (Table 6). The mean length of the merus was greater among waving males than droving males (R2 = 0·924; effect for male type: F = 5·608, df = 1,250, P = 0·019; effect for carapace width: F = 1903·982, df = 1,250, P < 0·001). Waving males were more likely than droving males to have a longer merus on the major body side (χ2 = 12·973, df = 1, P < 0·001). Consequently, the developmental trajectory varied by male activity (t = 4·639, N = 253, P < 0·001). However, residual variances and, thus, fluctuating asymmetry for merus length did not vary by male activity (F = 0·254, df = 1,251, P = 0·615).

ASYMMETRY AND THE LUNAR CYCLE

Asymmetry of the propodus of waving males varied with the proportion of lunar illumination across 29 dates from 1998 to 2000 (Fig. 4). Asymmetry was greater at new and full moons. The model, asymmetry = 1·030 + 0·205(illumination – 0·5)2, explained 51·0% of the variation in mean propodus asymmetry across 29 dates (F = 28·134, df = 1,27, P < 0·001). However, lunar illumination did not explain variation in mean carapace width (Fig. 5; F = 0·104, df = 1,27, P = 0·749).

Figure 4.

Mean antisymmetry of the propodus of chelipeds of waving males as a function of the proportion of the moon illuminated on 29 days from 1998 to 2000.

Figure 5.

Mean carapace width of waving males as a function of the proportion of the moon illuminated on 29 days from 1998 to 2000.

ASYMMETRIES AND MATING SUCCESS

From 1998 to 2000, the proportion of females in droves with eggs (4/734 = 0·005) was much less than for females associated with waving males (278/555 = 0·509; χ2 = 453·903, df = 1, P < 0·001). In 1998, the number of females in excavated breeding burrows was significantly correlated with propodus asymmetry (t = 4·233, P < 0·001) but not carapace width (t = –0·439, P = 0·661; R2 = 0·287, df = 2,82 for the multivariate model). In 2000, significant effects on the number of females occurred for propodus (t = 2·061, P = 0·048), merus (t = 2·398, P = 0·023) and eye stalk asymmetries (t = –3·550, P = 0·001) but not for carapace width (t = –1·332, P = 0·193; R2 = 0·531, df = 4,30 for the multivariate model).

Discussion

SEX DIFFERENCES IN ASYMMETRICAL CHARACTERS

The propodus of chelipeds, the merus of walking legs, and eye stalks of U. pugilator are all significantly longer in males than females for a given carapace width. Sexual selection favours the development of a cheliped with a claw for intersexual signalling and intrasexual fighting (Crane 1975). Sexual selection may also favour longer walking legs on the major side. Uca pugilator males wave the claw in a lateral-circular pattern which is associated with straightening of the major-side walking legs (Crane 1975). Antisymmetry of walking legs has evidently evolved to elevate the major side as fiddler crabs with other waving patterns lack leg antisymmetry (Miller 1973; Takeda & Murai 1993). Thus, antisymmetry of chelipeds and walking legs is adaptive in males.

Longer eye stalks permit visual detection of more distant objects (Crane 1975), such as females. Eye stalk length is not more asymmetrical in males than females. Thus, in U. pugilator, eye stalk symmetry may reveal the precision of developmental mechanisms that target some characters for symmetrical growth and others for antisymmetrical growth.

ANTISYMMETRY AND MALE BEHAVIOUR

The degree of asymmetrical development of all three structures may be condition-dependent. This is suggested by variation in size and asymmetry that is associated with the decision of males to feed or breed. Breeding males at Cat Island marsh occupy embankment burrows at densities of 30–250 m−2. Density is highest (from March through September) when lunar illumination is about 50%. When density is high, propodus antisymmetry is relatively low, reflecting the movement of males with small claws onto breeding grounds. These males abandon droving when the tidal flux is insufficient to adequately moisten sand flats. At new and full moons, the higher spring tides moisten sand flats and many males leave breeding territories to feed in droves. Males must acquire and store energy to compete for mates but time spent on energy acquisition reduces time available for mating (Caravello & Cameron 1987). Males that have had greater access to food or that are more efficient at assimilating energy could allocate more resources to claw growth. The same males may defer feeding decisions in favour of more time engaged in courtship.

Both droves and breeding grounds contain males of variable claw size relative to body size. However, a male experiencing mating success is less likely to abandon a desirable breeding burrow (Hyatt 1977; Christy 1983; Salmon & Hyatt 1983) which can accommodate several ovigerous females (Christy 1982). Within a breeding area, males with larger claws are more likely to attract females via waving (Hyatt 1977). These males also possess burrows more likely to be chosen by females (Christy 1983) because they are better able to acquire and defend prime sites from challengers (Hyatt 1977). Consequently, males with larger claws are more likely to remain on breeding grounds when conditions are favourable for droving (Robertson et al. 1980; Reinsel & Rittschof 1995). Thus, not only does cheliped antisymmetry at the breeding embankment increase at new and full moons with the departure of small-clawed males, but antisymmetry of waving males is greater than that of droving males of any given size at any given moment during the breeding season.

Claw growth is energy-limited (Crane 1975). Additional time spent feeding may permit greater allocation to claw growth as crabs continue to moult (Hartnoll 1974). Lost claws will regenerate and, given time, will approach normal sizes (A. E. Pratt & D. K. McLain, unpublished data). Thus, physiological processes may preferentially direct energy to undersized structures to accelerate their growth. This suggests the potential for males to develop sexually competitive bodies by reducing current reproductive effort. However, the costs of possessing a large claw may reinforce a relationship between claw size and male condition.

Eye stalk length does not vary among waving and droving males, suggesting equal energetic investment. However, eye stalk asymmetry was less among waving males. This indicates differential ability to control development. In brachyuran crabs, subtle asymmetries present in very young individuals persist through subsequent moult cycles (Chippindale & Palmer 1993). Therefore, eye stalk asymmetry may reveal male condition at an early age. Males with more symmetrical eye stalks also have larger claws for their body size. If claw size reveals condition through recent moults, the correlation indicates consistency in relative condition throughout life.

SEXUAL SELECTION ON ANTISYMMETRIES

In most populations of U. pugilator, breeding occurs in synchrony with the amplitude of semidiurnal tides (Morgan 1996). At Cat Island marsh, major breeding efforts are tied to lunar cycles but some females are mating and brooding eggs on any given day in the breeding season. Reduced breeding synchrony intensifies sexual competition between males and favours tenure at breeding grounds (Andersson 1994). Females generally incubate eggs in breeding burrows for approximately 2 weeks (Christy 1982) but then must feed for 2–4 weeks before they can produce another clutch (Salmon 1987). The long recovery period intensifies sexual competition between males by reducing the proportion of mating-receptive females (reviewed in Andersson 1994).

Thus, fitness of males is a multiplicative function of the rate at which females find the burrow suitable and tenure of the male at breeding burrows. Contests between males can result in burrow turnover. Consequently, it is not known if a resident male has mated all females in his burrow. However, contests result in a correlation between burrow quality and male fighting ability (Hyatt 1977; Christy 1983; Jennions & Backwell 1996). Therefore, males should experience mating success relative to the number of females in burrows they possess. The number of females sequestered in burrows is a positive function of the degree of cheliped and walking leg antisymmetry and a negative function of eye stalk asymmetry. Males with these traits are also less likely to forfeit breeding opportunities in order to feed in droves. Thus, greater expression of adaptive, condition-dependent antisymmetries and reduced expression of non-adaptive asymmetry explain variation in male fitness.

STRUCTURES AND COMPONENTS OF ASYMMETRY

Non-adaptive departure from bilateral symmetry may reveal how well development is buffered against environmental influences that could perturb growth (Møller & Swaddle 1998). More symmetrical individuals may have greater mating success (Møller & Thornhill 1998), possess greater resistance to stress (Polak 1997) and parasites (Sasal & Pampoulie 2000), have higher fertilization rates (Otronen 1998), and be capable of greater feats of physical performance (McLachlan & Cant 1995). A similar link between fitness and non-adaptive asymmetry is suggested for Sand Fiddler Crabs because males with more symmetrical eye stalks spend more time competing for mates and mate more frequently than males with less symmetrical eye stalks.

Mixture analysis suggested that the contribution of fluctuating asymmetry to eye stalk asymmetry is approximately 45%. While this is a rough estimate (Van Dongen et al. 1999), it indicates that developmental instability contributes substantially to between-sides differences in length. Directional asymmetry can also reveal failure to buffer development, especially under relatively extreme environmental conditions (Lens & Van Dongen 2000). Eye stalk asymmetry had a directional component in 1998 when both the mean and range in asymmetry were greater than in 2000.

Both greater expression of adaptive cheliped antisymmetry and reduced expression of non-adaptive, fluctuating asymmetry of eye stalks reveal male condition in the decision to feed or breed and correlate with mating success, a component of male fitness. However, fluctuating asymmetry of the propodus of chelipeds was not predictive of male quality. Thus, the degree of adaptive antisymmetry is its own unique predictor of male fitness.

Claw size, the frequency of waving, and population density vary across the range of U. pugilator (Salmon 1967). Where densities are high, competition for quality breeding burrows may be especially intense. Intense mating competition may select for genes that also increase mean population fitness (Kodric-Brown 1997). Perhaps there is a causal relationship between adult fiddler crab densities of 2 000 000 per hectare (Crane 1975) and mating competition that favours endurance, control over development and extravagant sexually selected antisymmetries.

Ancillary