SEARCH

SEARCH BY CITATION

Keywords:

  • condition dependence;
  • Cyrtodiopsis dalmanni;
  • eyespan;
  • handicap;
  • ornament;
  • sexual selection;
  • Sphyracephala beccarri;
  • stalk-eyed flies

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

Stalk-eyed flies are exemplars of sexual selection leading to the evolution of exaggerated male ornaments (eyespan). In Sphyracephala beccarri, there is no evidence for female mate choice for exaggerated male eyespan and only minor sex differences in eyespan. We used S. beccarri to test whether heightened condition dependence only evolves when male eyespan becomes sexually exaggerated. Male eyespan showed heightened condition dependence under food stress compared with a control trait (wing length). However, female eyespan displayed a similar pattern and there was no sex difference in the degree of increased eyespan sensitivity. The finding that eyespan is a sensitive indicator of food stress, even in an unexaggerated state, suggests that this may have acted as a pre-adaptation to the role of eyespan in sexual signalling in other Diopsid species. These results are consistent with handicap theory and Fisher's view of how sexual selection is initiated.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

Handicap models of sexual selection propose that females use male sexual ornaments to assess male genetic quality, in order to increase the fitness of offspring (Andersson, 1986; Pomiankowski, 1987, 1988; Grafen, 1990; Iwasa et al., 1991; Iwasa & Pomiankowski, 1994). Exaggerated sexual traits are assumed to reduce viability, so only males in good condition can bear the cost of an extravagant ornament. Consequently, a key prediction of the handicap hypothesis is that exaggerated male ornaments should have evolved heightened condition-dependent expression.

A number of reviews have concluded that condition dependence of male ornaments is widespread (Andersson, 1994; Johnstone, 1995). However, this view is less well founded than suggested, because it is based largely on correlational data. There are relatively few experimental studies, and many of these have not employed appropriate controls (reviewed in Cotton et al., 2004a). For instance, most experiments failed to compare the sexual trait with nonsexually selected control traits and it is not possible to assess whether traits possess enhanced condition dependence as required by the handicap hypothesis. In addition, many experiments have only investigated the effect of extreme differences in environmental condition (i.e. no vs. extreme stress), neither of which may represent conditions typically experienced in nature. There remains a pressing need for more appropriately designed and controlled experiments to test the generality of heightened condition dependence in sexually selected characters.

Male eyespan is well known as a trait in stalk-eyed flies that is subject to sexual selection primarily through female choice (Wilkinson & Dodson, 1997), and has become repeatedly exaggerated in several Diopsid lineages (Baker & Wilkinson, 2001). Our group has previously worked on Cyrtodiopsis dalmanni (Wiedemann), a stalk-eyed fly species that is highly sexually dimorphic for eyespan. This species forms nocturnal mating aggregations that are controlled by large eyespan males (Burkhardt & De la Motte, 1985; Wilkinson & Dodson, 1997). Females prefer to roost and mate with males with larger eyespan (Wilkinson & Reillo, 1994; Hingle et al., 2001). We have shown that male eyespan in C. dalmanni exhibits heightened levels of condition dependence than seen in female eyespan or wing length in either sex, both before and after controlling for body size (David et al., 1998; Cotton et al., 2004b). In addition, there is evidence of a genetic basis in the response of male eyespan to environmental variation in condition; some genotypes produced large male eyespan in all environments, whilst others became progressively smaller as stress increased (David et al., 2000).

In this paper we take a new approach by considering condition dependence of male eyespan in the Diopsid stalk-eyed fly Sphyracephala beccarri (Rondani). In contrast to C. dalmanni, it is a species with only slight sexual dimorphism for eyespan (Baker & Wilkinson, 2001). There are no field observations for this species, but in the laboratory S. beccarri does not form nocturnal mating aggregations. Individuals mate opportunistically and males exhibit post-copulatory passive mate guarding (S. Cotton, pers. obs.), which has also been recorded in a closely related species, S. brevicornis (Hochberg-Stasny, 1985). There is no evidence of female mate choice for male eyespan in this or related sexually monomorphic species (Wilkinson & Dodson, 1997; Wilkinson et al., 1998).

Our objective was to use S. beccarri to assess the condition dependence of male eyespan in a species subject to little or greatly reduced sexual selection. Larvae were subjected to five levels of food treatment, varying from low (abundant food) to high (minimal food) larval stress, using the same protocol developed to study the sexually dimorphic species C. dalmanni. We compared male eyespan with two nonsexual control traits: male wing length and female eyespan. We also made a control comparison between female eyespan and female wing length. The analysis was carried out on absolute trait size and on measurements controlled for body size variation. This allowed us to test the assumption that in the absence of an evolved sexual signalling function, male eyespan responds to stressful conditions in the same way as other nonsexual traits. That is, we expect that traits free from sexual selection to show no heightened condition dependence. We complete our analysis of S. beccarri by comparing its response with those of C. dalmanni flies that were subjected to the same food treatments in a previous study (Cotton et al., 2004b).

Fly rearing

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

The laboratory-adapted population of S. beccarri used in this experiment was transferred to our laboratory in 1999, after being collected from South Africa in 1993 by Jerry Wilkinson (University of Maryland). It has since been maintained in cage culture at 25 °C on a 12 h-light : 12 h-dark photoperiod, and population sizes have been kept high (>200 individuals) to minimize inbreeding.

Condition was manipulated by rearing larvae under varying degrees of food stress. Eggs were collected from stock populations over 24 h periods, and batches of 13 were assigned to one of five food treatments: 0.015, 0.03, 0.06, 0.09, and 0.12 g homogenized corn per egg. These food levels were chosen as pilot work showed they caused significant phenotypic changes in trait size (S. Cotton, unpublished).

Measurements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

Adult flies were collected and frozen every 24 h. All individuals (237 male and 271 female) were measured later to an accuracy of 0.01 mm using a monocular microscope and the image analysis program NIH Image (Version 1.55; National Institutes of Health, Bethesda, MD, USA). Measurements were taken of eyespan (between the outermost lateral edges of the eye-bulbs), thorax (middle of the anterior-most part of the head to the posterior edge of the thorax) and wing length [the branch point of the MA and r-m veins to the terminus of the RP4 vein, p.45 in Gullan & Cranston (1994), measurement ‘x’ in David et al. (1998)]. Measurements from damaged traits were not recorded, so sample sizes differ. All flies were measured ‘blind’ by a single person (SC).

Statistical analysis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

Differences between the absolute sizes of traits in S. beccarri were determined with t-tests by comparing flies reared under benign environments (0.12 g per egg). Food treatment was classified as an ordinal variable (food) and one-way anovas were used to test for significance of food effects on traits within each sex. Comparisons between adjacent treatments were performed to identify those responsible for significant effects on trait size. Eyespan (ES) and wing length were compared within each sex in a General Linear Model (GLM) (with factors food, trait and food × trait) to detect differences between traits (trait = ES or wing length) in their response to food treatment. The significance of the interactions was determined with F-tests on the change in explained variance upon removal of each term from the full model (Crawley, 1993, p. 196). Treatments responsible for significant interactions were identified using pair-wise ordinal food × trait terms derived from the model. Similar analyses were performed to compare the response of traits across sexes (GLM with factors food, sex and food × sex). To investigate whether differences between the response of ES and wing length to food stress varied across sexes we tested the significance of the sex × food × trait interaction in a GLM containing all lower order interactions and main effects.

A significant proportion of the response to stress is likely to result from body size scaling (David et al., 1998, 2000; Cotton et al., 2004b). The measurement of thorax was therefore taken a priori as a general indicator of body size and included as a covariate in further analyses. Sexual dimorphism in trait size after controlling for body size in the benign food treatment (0.12 g per egg) was determined using a GLM with factors thorax, sex and thorax × sex.

To assess the effect of food treatment we constructed GLMs using three main effects (food, trait, thorax) and their interactions (if significant or required). This model was based on previous work with C. dalmanni (David et al., 1998; Cotton et al., 2004b). The second-order food × trait interaction was used to detect differences between male ES and wing length in their response to food treatment after removing the effects of body size. This analysis was repeated in females and for separate comparisons of ES and wing length across sexes by replacing trait effects with sex effects. Allometric slopes differ between traits and sexes, and this makes Least Squares Mean estimates (LSMs) generated from covariance models containing both food and trait or sex terms difficult to compare. So, in order to visualize the nature of significant food × trait or food × sex interactions (i.e. the differences between traits and sexes in their response to food treatment) we plotted the LSMs estimated from within-trait GLMs (with factors food, thorax and their interaction). Ordinal between-treatment contrasts from within-trait GLMs were used to identify those responsible for changes in body size-adjusted trait size. As with absolute trait size, we tested the null model that the difference between ES and wing length responses was the same in males and females after controlling for body size through the sex × food × trait interaction in a GLM containing sex, food and trait main effects, thorax as a covariate and all (significant or required) interactions.

Comparison with C. dalmanni

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

We compared the response of S. beccarri to food stress with that observed in C. dalmanni, a highly sexually dimorphic species. Eggs were collected from a laboratory-adapted population of C. dalmanni that originated from flies collected in Malaysia by AP in 1993. Larvae were exposed to the same five food treatments. Only brief results are given (sample size 267 male and 266 female); for a more detailed analysis of C. dalmanni see Cotton et al. (2004b). Differences between the two species in trait size sexual dimorphism were examined using the interaction between sex and species effects in two-factor anovas.

Adjustment for multiple comparisons

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

As multiple tests of the effect of food treatment were performed on each trait, we adjusted the significance level using the sequential Bonferroni method (Rice, 1989; Sokal & Rohlf, 1995). To avoid being overly conservative we treated analyses of the response to food treatment of absolute trait size and body size-controlled trait size separately within each species.

Trait size

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

In the environment amenable to maximal growth (0.12 g per egg), female body size in S. beccarri was larger than male body size, using thorax length as an estimate of body size (Fig. 1; Table 1). Wing length was also larger in females, and so to a minor extent was eyespan (Fig. 1; Table 1). To assess male and female differences in wing length and eyespan independent of body size, we entered thorax length as a covariate in GLMs, and looked for sex differences. After taking account of body size differences between the sexes, females still showed larger wing length, but eyespan was greater in males (Table 1).

image

Figure 1. Silhouettes of male and female (a) Sphyracephala beccarri, and (b) Cyrtodiopsis dalmanni. Scale bars: 2 mm (vertically).

Download figure to PowerPoint

Table 1.  Absolute [mean ± SD (n)] and least squares mean estimates (LSM ± SE) of trait size (mm) of flies grown in benign environments (0.12 g per egg).
Species: traitAbsolute valuesLSM estimates
MalesFemalesDifferenceMalesFemalesDifference
Sphyracephala beccarri
 Eyespan2.20 ± 0.05 (37)2.22 ± 0.05 (44)t79 = 2.25, P < 0.052.24 ± 0.012.15 ± 0.01F1,77 = 22.32, P < 0.001
 Wing1.71 ± 0.04 (37)1.91 ± 0.05 (44)t80 = 14.72, P < 0.0011.75 ± 0.011.87 ± 0.01F1,78 = 54.65, P < 0.001
 Thorax2.06 ± 0.09 (38)2.33 ± 0.07 (44)t79 = 19.91, P < 0.001   
Cyrtodiopsis dalmanni
 Eyespan8.60 ± 0.31 (53)5.92 ± 0.20 (61)t112 = 55.82, P < 0.0018.46 ± 0.035.98 ± 0.03F1,110 = 3363.86, P < 0.001
 Wing2.50 ± 0.07 (57)2.36 ± 0.06 (62)t117 = 11.83, P < 0.0012.47 ± 0.012.38 ± 0.01F1,114 = 61.52, P < 0.001
 Thorax3.03 ± 0.09 (56)2.91 ± 0.10 (62)t116 = 6.97, P < 0.001   

Condition dependence

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

Food treatment had significant effects on the absolute size of all traits; flies became smaller as food availability declined (Fig. 2; F4,218–252 ≥ 68.45, all P < 0.001). Changes in trait size became greater as food stress increased and were particularly marked in the two most stressful treatments (0.03 and 0.015 g per egg, Fig. 2).

image

Figure 2. Changes in mean eyespan (ES) and wing length of Sphyracephala beccarri in response to food treatment. Trait means were standardized to unity in the 0.12 g treatment group to ease comparisons between different sized traits. Trait sizes from other treatments are expressed as proportions of the standardized 0.12 g groups. Error bars are omitted for clarity. Asterisks denote significance of within-trait, between-adjacent treatment comparisons after sequential Bonferroni correction: ***P < 0.001.

Download figure to PowerPoint

We tested whether male eyespan was more sensitive to stress than other traits. Absolute male eyespan was more sensitive to food stress than male wing length (Fig. 2; food × trait F4,436 = 20.30, P < 0.001). This difference occurred in the two most stressful treatments (Fig. 2; 0.06–0.03 g × trait, t = 3.19, P < 0.05; 0.03–0.015 g × trait, t = 3.51, P < 0.001). However, male and female eyespan did not differ in their response to food treatment (Fig. 2; food × sex, F4,470 = 1.70, n.s.).

We then tested the response of absolute female eyespan and found it was more sensitive than female wing length (Fig. 2; food × trait, F4,502 = 17.25, P < 0.001). This difference was also limited to the two most stressful groups (Fig. 2; 0.06–0.03 g × trait, t = 3.14, P < 0.05; 0.03–0.015 g × trait t = 3.80, P < 0.001). An explicit comparison of the sexes showed that the different response of absolute eyespan and wing length was similar in both males and females (sex × food × trait, F4,938 = 0.31, n.s.). So it appears from absolute measures that eyespan is a more sensitive trait than wing length, but there is no evidence of heightened sensitivity attributable to male eyespan.

As body size differed between the sexes and responded to food stress, the patterns reported above could be due to changes in body size. Female thorax size responded more to stress than male thorax size (food × sex, F4,470 = 8.00, P < 0.001, data not shown). So we repeated the analyses with thorax length as a covariate in GLMs. After adjusting for body size, food treatment still had significant effects on all traits (food, F4,212–257 ≥ 4.37, all P < 0.05). Male eyespan remained more sensitive to food treatment than male wing length after controlling for body size variation (Fig. 3; food × trait, F4,424 = 5.13, P < 0.001). In addition, male eyespan was now more sensitive than female eyespan after controlling for body size (Fig. 3; food × sex, F4,459 = 8.63, P < 0.001).

image

Figure 3. Comparisons between eyespan (ES) and wing length of Sphyracephala beccarri in their response to food treatment after controlling for body size. Least squares means estimates were standardized to unity in the 0.12 g treatment group to ease comparisons between different sized traits. Least squares means from other treatments are expressed as proportions of the standardized 0.12 g groups. Error bars are omitted for clarity. Asterisks indicate significance of within-trait, between-adjacent treatment comparisons after sequential Bonferroni correction: ***P < 0.001, *P < 0.05.

Download figure to PowerPoint

We found a similar trend towards female eyespan being more sensitive than female wing length after controlling for body size variation (Fig. 3; food × trait, F4,496 = 2.14, P = 0.075). Comparison of the sexes showed that the different response of eyespan and wing length was similar in males and females when body size differences were taken into account (thorax in model: sex × food × trait, F4,920 = 1.22, n.s.). This confirms the result found with absolute trait size, that the heightened response of eyespan to food treatment was a general rather than sex-specific feature.

We next investigated the possibility that general sex differences could account for these patterns. After controlling for body size, male wing length was more sensitive to food stress than female wing length (Fig. 3; food × sex F4,462 = 12.51, P < 0.001) and male eyespan was more sensitive than female eyespan (see above). These relationships hint at a greater sensitivity of male traits in general. However, such a male effect was not seen with absolute trait measures. The reverse pattern occurred as absolute female wing length was more sensitive to food stress than male wing length (Fig. 2; food × sex, F4,468 = 7.08, P < 0.001), and there was no difference between the sexes in eyespan sensitivity (see above). The lack of a consistent pattern in absolute and body size-controlled comparisons leads us to exclude the hypothesis of generally heightened male sensitivity to food stress.

Comparison with C. dalmanni

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

In the most benign environment (0.12 g per egg), all male C. dalmanni traits were larger than in females (Fig. 1; Table 1). After controlling for body size, both male eyespan and wing length remained larger (Table 1). Inter-specific comparisons of trait size in the 0.12 g treatment group revealed that absolute trait size sexual dimorphism was greater in C. dalmanni than S. beccarri for all characters but most markedly for eyespan, both before (species × sex; thorax, F1,196 = 229.26; eyespan F1,191 = 2181.74; wing length F1,196 = 427.95, all P < 0.001) and after adjusting for body size (species × sex; eyespan, F1,187 = 95.13; wing length F1,191 = 12.31, all P < 0.001).

As food availability declined, the absolute size of eyespan and wing length decreased in C. dalmanni (F4,247–258 ≥ 213.37, all P < 0.001). As has been reported elsewhere, traits differed in the magnitude of their response to food stress (Cotton et al., 2004b). Male absolute eyespan declined more than male wing length (Fig. 4; food × trait, F4,505 = 257.28, P < 0.001) and female eyespan (Fig. 4; food × sex, F4,494 = 93.89, P < 0.001). Absolute female eyespan also responded more than female wing length (Fig. 4; F4,496 = 112.74, P < 0.001). However, unlike S. beccarri, there was a sex difference in the degree of increased eyespan sensitivity when compared with wing length (sex × food × trait, F4,1001 = 76.83, P < 0.001), demonstrating that absolute male eyespan exhibits heightened condition-dependence.

image

Figure 4. Changes in mean eyespan (ES) and wing length of Cyrtodiopsis dalmanni in response to food treatment. Trait means were standardized to unity in the 0.12 g treatment group to ease comparisons between different sized traits. Trait sizes from other treatments are expressed as proportions of the standardized 0.12 g groups. Error bars are omitted for clarity. Asterisks denote significance of within-trait, between-adjacent treatment comparisons after sequential Bonferroni correction: ***P < 0.001, **P < 0.01, *P < 0.05.

Download figure to PowerPoint

Similar patterns were observed when thorax length was included in GLMs (Cotton et al., 2004b). After controlling for body size, the exaggerated male ornament in C. dalmanni was more sensitive to larval stress than male wing length (Fig. 5; food × trait, F4,493 = 5.16, P < 0.001) or female eyespan (Fig. 5; food × sex, F4,488 = 3.62, P < 0.001). A similar trend was found in females, with body size-controlled eyespan responding more than female wing length (Fig. 5; food × trait, F4,486 = 2.20, P = 0.068). However, again unlike S. beccarri, the heightened condition dependence of eyespan compared with wing length in males remained after controlling for body size (sex × food × trait, F4,987 = 2.81, P < 0.05).

image

Figure 5. Comparisons between eyespan (ES) and wing length of Cyrtodiopsis dalmanni in their response to food treatment after controlling for body size. Least squares means estimates were standardized to unity in the 0.12 g treatment group to ease comparisons between different sized traits. Least squares means from other treatments are expressed as proportions of the standardized 0.12 g groups. Error bars are omitted for clarity. Asterisks indicate significance of within-trait, between-adjacent treatment comparisons after sequential Bonferroni correction: **P < 0.01, *P < 0.05.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

The co-evolution of female preference for condition-dependent male sexual traits is central to handicap models of sexual selection (Iwasa et al., 1991; Iwasa & Pomiankowski, 1994, 1999). Theory predicts that as male ornaments become exaggerated (and therefore costly), their expression becomes more closely dependent on the quality of the bearer as only males in good condition can afford to pay the cost of a large ornament. This leads to the prediction that the condition dependence of male ornaments will covary positively with the strength of sexual selection and their degree of exaggeration.

We previously tested this prediction in C. dalmanni, a stalk-eyed fly species with greatly exaggerated male eyespan. In C. dalmanni, male eyespan shows heightened condition dependence relative to female eyespan and male wing length, both before and after controlling for body size variation (David et al., 1998, 2000; Cotton et al., 2004b).

In this paper we adopted the alternative approach and tested the overlooked assumption that male eyespan in species subject to little or no sexual selection should not be strongly condition-dependent. Using the weakly dimorphic stalk-eyed fly species S. beccarri, we found that male eyespan was more sensitive than male wing length to changes in food conditions, using both absolute and body size-controlled measurements. At first sight this argues for heightened sensitivity in male eyespan. However, female eyespan showed a similar pattern of greater sensitivity to changes in condition compared with female wing length, again using absolute and body size-controlled measurements. There was no difference between the sexes in the degree of increased sensitivity of eyespan compared with wing length. So we uncovered a trait difference, not a sex-specific trait difference.

These results were unexpected but they are in line with the hypothesis that the lack of sexual selection in S. beccarri, and the absence of sexual eyespan exaggeration, has not resulted in heightened condition dependence of male eyespan. They suggest why mate choice based on eyespan exaggeration may have originally evolved in the Diopsidae. Fisher (1915, 1930) proposed that sexual selection would be initiated if female preference arose for male traits that conferred a natural selection advantage. Absolute and body size-controlled eyespan fulfil this criterion for sexual selection targets, as they are more sensitive indicators of larval food stress even in their nonexaggerated state. So a female preferring males with larger absolute and/or body-size controlled eyespan would on average mate with males in better condition. Assuming that in nature there is a genetic component to condition and the response to environmental stresses (like food limitation), such female preference would result in inherited fitness benefits. Our work suggests that these benefits would be greater for mate choice based on eyespan than on other traits.

This hypothesis needs to be investigated further. We only compared eyespan to one other trait (wing length). More contrasts are needed to more firmly establish that eyespan is a more sensitive trait; it remains possible that eyespan reacts like other traits whilst wing length is an insensitive trait. It might also be revealing to analyse more closely related species of S. beccarri. Within the Sphyracephela group, S. bipunctipennis has evolved greater exaggeration of male eyespan and marked sexual dimorphism, whereas S. brevicornis has evolved sexual monomorphism for eyespan allometry as well as for absolute eyespan (Baker & Wilkinson, 2001).

Another finding in this study was that body size-adjusted male eyespan was more sensitive to food stress than the homologous female trait. The same was true for body size-adjusted male wing length compared with the homologous female trait. This implies that male traits are more sensitive to stress than those of females. However this hypothesis is not supported by the comparison of absolute trait values, which are not sexually different for eyespan sensitivity and show greater female sensitivity for wing length. This disparity between absolute and body size-adjusted trait measures does not support the hypothesis of generally heightened male sensitivity to food stress.

We compared the responses of S. beccarri with those of C. dalmanni to test the hypothesis that the condition dependence of male ornaments covaries positively with the strength of sexual selection and the degree of exaggeration. Cyrtodiopsis dalmanni is a highly sexually dimorphic stalk-eyed fly species with greatly exaggerated male eyespan, a trait subject to strong sexual selection. We found that male eyespan in C. dalmanni showed heightened condition dependence relative to male wing length and female eyespan, both before and after controlling for body size variation (this paper; Cotton et al., 2004b). As in S. beccarri, female eyespan also declined with stress in C. dalmanni. But unlike S. beccarri, the male response in C. dalmanni was disproportionately greater in eyespan relative to wing length both for absolute and body size-controlled measures. Thus we found a positive association between the degree of exaggeration and the degree of condition dependence.

In conclusion, our experiments add weight to the prediction that heightened condition dependence is associated specifically with costly exaggeration of male sexual traits. This finding is consistent with the prediction made by handicap models of sexual selection. In addition, we raise the possibility that eyespan, even in its unexaggerated state, is a more sensitive indicator of condition than other traits, which may have acted as a pre-adaptation to its role in sexual signalling in other Diopsid species. This latter finding accords well with Fisher's (1915, 1930) original discussion of how sexual selection is initiated, but needs to be substantiated by further investigation.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References

We thank Jerry Wilkinson for providing S. beccarri flies, and Mark Pagel and Dave Rogers for useful discussion. We are grateful to Janne Kotiaho, Mike Ritchie and an anonymous reviewer who provided helpful comments on an earlier version of this manuscript. S.C. was supported by a NERC Research Studentship, with additional funding from the Department of Biology, University College London.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Fly rearing
  6. Measurements
  7. Statistical analysis
  8. Comparison with C. dalmanni
  9. Adjustment for multiple comparisons
  10. Results
  11. Trait size
  12. Condition dependence
  13. Comparison with C. dalmanni
  14. Discussion
  15. Acknowledgments
  16. References
  • Andersson, M. 1986. Evolution of condition-dependent sex ornaments and mating preferences; sexual selection based on viability differences. Evolution 0: 804816.
  • Andersson, M. 1994. Sexual Selection. Princeton University Press, Princeton, NJ.
  • Baker, R.H. & Wilkinson, G.S. 2001. Phylogenetic analysis of sexual dimorphism and eyespan allometry in stalk-eyed flies (Diopsidae). Evolution 55: 13731385.
  • Burkhardt, D. & De la Motte, I. 1985. Selective pressures, variability, and sexual dimorphism in stalk-eyed flies (Diopsidae). Naturwissenschaften. 72: 204206.
  • Cotton, S., Fowler, K & Pomiankowski, A. 2004a. Do sexual ornaments demonstrate heightened condition-dependent expression as predicted by the handicap hypothesis. Proc. R. Soc. Lond. B. 271: 771783.
  • Cotton, S., Fowler, K. & Pomiankowski, A. 2004b. Condition dependence of sexual ornament size and variation in the stalk-eyed fly Cyrtodiopsis dalmanni (Diptera: Diopsidae). Evolution 58: 10381046.
  • Crawley, M.J. 1993. GLIM for Ecologists. Blackwell Scientific Publications, Oxford, UK.
  • David, P., Hingle, A., Greig, D., Rutherford, A., Pomiankowski, A. & Fowler, K. 1998. Male sexual ornament size but not asymmetry reflects condition in stalk-eyed flies. Proc. R. Soc. Lond. B. 265: 22112216.
  • David, P., Bjorksten, T., Fowler, K. & Pomiankowski, A. 2000. Condition-dependent signalling of genetic variation in stalk-eyed flies. Nature 406: 186188.
  • Fisher, R.A. 1915. The evolution of sexual preference. Eugenics Rev. 7: 184192.
  • Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford.
  • Grafen, A. 1990. Sexual selection unhandicapped by the Fisher process. J. Theor. Biol. 144: 473516.
  • Gullan, P.J. & Cranston, P.S. 1994. The Insects: an Outline of Entomology. Chapman and Hall, London.
  • Hingle, A., Fowler, K & Pomiankowski, A. 2001. Size-dependent mate preference in the stalk-eyed fly Cyrtodiopsis dalmanni. Anim. Behav. 61: 589595.
  • Hochberg-Stasny, T.A. 1985. Biology, Behaviour and Life Cycle of Sphyracephala brevicornis Say (Diptera: Diopsidae). MS thesis. University of West Virginia, Virginia.
  • Iwasa, Y. & Pomiankowski, A. 1994. The evolution of mate preferences for multiple handicaps. Evolution 48: 853867.
  • Iwasa, Y. & Pomiankowski, A. 1999. Good parent and good genes models of handicap evolution. J. Theor. Biol. 200: 97109.
  • Iwasa, Y., Pomiankowski, A. & Nee, S. 1991. The evolution of costly mate preferences. II. The ‘handicap’ principle. Evolution 45: 14311442.
  • Johnstone, R.A. 1995. Sexual selection, honest advertisement and the handicap principle: reviewing the evidence. Biol. Rev. 70: 165.
  • Pomiankowski, A. 1987. Sexual selection: the handicap principle does work – sometimes. Proc. R. Soc. Lond. B. 231: 123145.
  • Pomiankowski, A. 1988. The evolution of female mate preferences for male genetic quality. Oxford Surv. Evol. Biol. 5: 136184.
  • Rice, W.R. 1989. Analysing tables of statistical tests. Evolution 43: 223225.
  • Sokal, R.R. & Rohlf, F.J. 1995. Biometry, 3rd edn. Freeman, New York.
  • Wilkinson, G.S. & Dodson, G.N. 1997. Function and evolution of antlers and eye stalks in flies. In: The Evolution of Mating Systems in Insects and Arachnids (J.Choe & B.Crespi, eds), pp. 310328. Cambridge University Press, Cambridge.
  • Wilkinson, G.S. & Reillo, P.R. 1994. Female preference response to artificial selection on an exaggerated male trait in a stalk-eyed fly. Proc. R. Soc. Lond. B 255: 16.
  • Wilkinson, G.S., Kahler, H. & Baker, R.H. 1998. Evolution of female mating preferences in stalk-eyed flies. Behav. Ecol. 9: 525533.