SEXUAL SELECTION ACCOUNTS FOR THE GEOGRAPHIC REVERSAL OF SEXUAL SIZE DIMORPHISM IN THE DUNG FLY, SEPSIS PUNCTUM (DIPTERA: SEPSIDAE)

Authors

  • Nalini Puniamoorthy,

    1. Institute of Evolutionary Biology and Environmental Studies, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland
    2. E-mail: nalini@ieu.uzh.ch
    3. Department of Biological Sciences, National University of Singapore, 14 Science Drive 4, Singapore 117543, Singapore
    Search for more papers by this author
  • Martin A. Schäfer,

    1. Institute of Evolutionary Biology and Environmental Studies, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland
    Search for more papers by this author
  • Wolf U. Blanckenhorn

    1. Institute of Evolutionary Biology and Environmental Studies, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland
    Search for more papers by this author

Abstract

Sexual size dimorphism (SSD) varies widely across and within species. The differential equilibrium model of SSD explains dimorphism as the evolutionary outcome of consistent differences in natural and sexual selection between the sexes. Here, we comprehensively examine a unique cross-continental reversal in SSD in the dung fly, Sepsis punctum. Using common garden laboratory experiments, we establish that SSD is male-biased in Europe and female-biased in North America. When estimating sexual (pairing success) and fecundity selection (clutch size of female partner) on males under three operational sex ratios (OSRs), we find that the intensity of sexual selection is significantly stronger in European versus North American populations, increasing with male body size and OSR in the former only. Fecundity selection on female body size also increases strongly with egg number and weakly with egg volume, however, equally on both continents. Finally, viability selection on body size in terms of intrinsic (physiological) adult life span in the laboratory is overall nil and does not vary significantly across all seven populations. Although it is impossible to prove causality, our results confirm the differential equilibrium model of SSD in that differences in sexual selection intensity account for the reversal in SSD in European versus North American populations, presumably mediating the ongoing speciation process in S. punctum.

Evolutionary biologists largely agree that divergence in sexual dimorphism and mating behavior is frequently driven by sexual selection (Andersson 1994; Arnqvist et al. 2000; Gray and Cade 2000; Panhuis et al. 2001; Boake 2005; Gavrilets and Hayashi 2005; Ritchie 2007). Differences in body size between the sexes, or sexual size dimorphism (SSD), are ubiquitous but variable across the animal kingdom. Species and even populations within a species can differ greatly in the direction and extent of SSD, and there are numerous studies exploring the evolutionary mechanisms underlying this variation (Andersen 1994; Fairbairn 1997; Blanckenhorn 2000; Ding and Blanckenhorn 2002; Drovetski et al. 2006; Blanckenhorn et al. 2007a; Fairbairn et al. 2007; Serrano-Meneses et al. 2007; Stillwell and Fox 2007; Teuschl et al. 2007). It is established that body size affects reproductive success via different mechanisms in the sexes, so the optimal size associated with the maximum fitness often varies for males and females. According to the differential equilibrium model of the evolution of SSD, dimorphism in body size arises when the net effects of sexual and natural selection differ between the sexes (Price 1984;
Andersson 1994; Preziosi and Fairbairn 2000; Blanckenhorn 2000). For instance, most mammals and many birds exhibit male-biased SSD, which is primarily attributed to greater mating
success of larger males due to male–male competition (via access to territories and/or females) or female choice (Abouheif and Fairbairn 1997). SSD is typically reversed among invertebrates and most ectothermic vertebrates, where female-biased SSD is driven by strong fecundity selection in terms of increased investment in offspring production associated with larger female size (Abouheif and Fairbairn 1997; Blanckenhorn et al. 2007b; Stillwell et al. 2010). Fecundity and sexual selection for larger females or males is presumably held in check by counteracting forces favoring small size in terms of adult and/or juvenile viability or survival, although empirical evidence for these selective processes is far scarcer and often difficult to come by (Blanckenhorn 2000). Additionally, the degree to which the sexes differ in size is also considerably affected by genetic, developmental, and phylogenetic constraints (Badyaev 2002;
Lindenfors et al. 2002; Ramos et al. 2005; Hu et al. 2010;
Tammaru et al. 2010).

Although the above arguments intuitively explain variation in dimorphism among taxa, they are necessarily simplistic and incomplete because the crucial issue is the relative strength of sex-specific sexual, fecundity, and viability selection in any given species (Price 1984; Arak 1988; Schluter et al. 1991; Andersson 1994; Blanckenhorn 2000). For example, strong sexual selection for large males also regularly occurs in species with smaller males (Andersson 1994; Fairbairn and Preziosi 1994; Fairbairn 1997; Blanckenhorn et al. 1999). In the ideal case, when all the relevant selection pressures are measured, the differential equilibrium model can generate quantitative predictions about the SSD expected of a given population or species (Arak 1988;
Blanckenhorn 2000; Preziosi and Fairbairn 2000; Fairbairn et al. 2007). Therefore, the model has to be tested in a microevolutionary context by comparing populations of a single species exhibiting variation in dimorphism (e.g., Storz et al. 2001; Schauble 2004; Teder and Tammaru 2005; McGarrity and Johnson 2009; Lyapkov et al. 2010; Yu et al. 2010). However, in most cases, intraspecific variation in SSD is slight and quantitative but not qualitatively reversed. We know of only one study documenting albeit minor SSD reversals in some traits but not others in the house finch (Badyaev and Hill 2000). Here, we investigate a unique example of strong qualitative reversal in SSD among cross-continental populations of the dung fly Sepsis punctum (Diptera: Sepsidae).

Sepsidae are a family of flies with approximately 320 described species across 36 known genera. Like most insects, sepsid flies generally display female-biased size dimorphism, although examination of museum specimens indicates that in some species SSD is male-biased (Blanckenhorn et al. 2007b). Sepsis punctum in particular has a widespread distribution ranging from North America to Europe, North Africa, and parts of Asia. It is a generalist that can be found on various types of decaying organic matter, although vertebrate excrements, and cow dung in particular, are its most common breeding substrate (Pont and Meier 2002). Schulz (1999) first noticed that SSD might be reversed between European and North American S. punctum. This situation presents the ideal opportunity to test the differential equilibrium model of SSD across replicate cross-continental S. punctum populations that vary in both the direction and magnitude of SSD. Using laboratory common garden experiments, we first ascertain whether SSD is indeed male-biased in European and female-biased in American populations. Using standardized quantitative measures of selection (Lande and Arnold 1983; Arnold and Wade 1984a,b), we next estimate (1) adult viability selection on body size in terms of intrinsic (physiological) longevity, (2) fecundity selection on female body size in terms of clutch and egg size, and (3) sexual and fecundity selection on male body size in terms of male mating success and the number of eggs of his mate (assortative mating). We estimate sexual selection in population cages at three operational sex ratios (OSRs), as a function of which competition for mates and consequently the intensity of sexual selection is expected to increase (Bonduriansky 2001). According to the equilibrium model of SSD, we expect that in the European populations of S. punctum the intensity of sexual selection on male size should be greater than the intensity of fecundity selection on female size, whereas this should be reversed in North America; in other words, continental differences in sexual selection on male size should be large compared to continental differences in fecundity selection on female size and in viability selection on male and female size, which should be small or nonexistent.

Methods

POPULATION SAMPLING AND FLY CULTURE MAINTENANCE

We sampled four European S. punctum populations from Nyköping, Sweden (SE: 58.67°N, 16.94°E); Berlin, Germany (DE: 52.45°N, 13.28°E); Vienna, Austria (A: 48.20°N, 16.36°E); and Zürich, Switzerland (CH: 47.40°N, 8.55°E), and three North American populations from Davis, California (CA: 38.54°N, −121.75°W); Athens, Georgia (GA: 33.96°N, −83.38°E); and Manhattan, New York (NY: 40.78°N, −73.96°E). Wild-caught females were brought to the laboratory and used to establish stock cultures of multiple (10–20) replicate lines per population that were housed in separate plastic containers and regularly supplied with fresh cow dung, sugar, and water ad libitum.

COMMON GARDEN EXPERIMENTS

We conducted laboratory common garden experiments to ascertain patterns of SSD among the European and North American populations. We allowed mated females, housed in replicate group containers per population, to oviposit in pots of fresh cow dung for 2–3 h. We then reared the offspring in abundant cow dung in a climate chamber at standardized 24°C, 60% humidity, and 14 h light cycle, measured the development time and head width of emergent flies as a standard index of body size. This method of using laboratory lines instead of wild-caught females removes confounding environmental variation influencing phenotypic body size, establishing that the body size differentiation is indeed
heritable.

TESTING THE DIFFERENTIAL EQUILIBRIUM MODEL

Adult viability (i.e,. intrinsic longevity) selection

Viability selection on males and females is affected by multiple extrinsic factors such as parasitism, predation, thermoregulation, food availability, etc. as well as by intrinsic physiological and genetic factors (reflecting ageing). Estimation of juvenile or adult mortality as a function of body size in the wild in small mobile insects is essentially impossible. Instead, we tested whether there are size- and sex-dependent differences in intrinsic adult longevity between European and North American populations as a function of body size under laboratory conditions in population cages (cf. Blanckenhorn et al. 1999). We provided stock cultures with varying amounts of dung to generate a range of phenotypic body sizes, and reared the offspring under the standard conditions mentioned earlier. The emerging flies were individually sexed under a microscope within 12 h of eclosion and set up under two different “housing” treatments (Teuschl et al. 2010): males only and females only (i.e., two treatments per population; five replicate containers per treatment; approximately 18–20 individual flies per container). Each container was provided with fresh dung, sugar, and water ad libitum. We monitored all 70 containers and more than 1300 individuals daily for adult mortality. Dead flies were removed every day, scored for adult life span, and measured for body size (head width).

Fecundity selection

To estimate fecundity selection, we randomly selected 30–60 once mated females of various body sizes from the stock lines, set them up individually in glass vials, provided them with fresh dung and counted their first (and sometimes additionally their second) clutch sizes, which is good proxy for lifetime fecundity in the study species (N. Puniamoorthy, unpubl. data). Because investment in offspring production can also be affected by the amount of resources invested in each egg, we additionally measured the average egg volume of five eggs in each clutch for each female in all seven populations. Every female was frozen afterwards and measured for body size (head width).

Sexual selection: Male mating success

For each population, we supplied stock lines with two pots of fresh dung each. To generate individuals of varying sizes, one dish was removed after 2 h (no larval competition), whereas the other was left overnight (competition). These dung dishes were subsequently placed into larger plastic containers and housed in climate chambers at 24oC. Emerging flies were sexed within 24 h of eclosion and thereafter housed separately in single-sex group containers with dung, sugar, and water. We waited three to four days to ensure sexual maturity and then conducted mating trials with randomly assembled virgin flies in population cages at three OSRs: five males plus five females (OSR = 1), 10 males plus five females (OSR = 2), and 20 males plus five females (OSR = 4). There were four to five replicates per OSR per population. Females always entered the population container first, which was equipped with water and sugar and some fresh dung; the males were added later. We tracked which male copulated with which female by isolating the mating pairs from the singletons. Each group trial lasted for a maximum of 2 h after which all individuals (both mated and unmated) were measured for body size. From these data, male sexual and fecundity selection differentials could be calculated (Supplementary file 1: Raw data).

In this study, because we were only interested in instantaneous pairing success, we did not allow for multiple mating. Early field observations of sepsid flies note that although male densities at a dung pat can rise up to 500 individuals in the first few minutes of the dung dropping, this number decreases drastically within the first 30 min (Hammer 1941). In fact, Parker (1972a,b) additionally showed that in S. cynipsea, the highest female arrival, oviposition, and capture rates occur within 10 min of the dropping and declines sharply after that. Copulation in S. punctum usually lasts approximately 20–30 min (N. Puniamoorthy, pers. obs.), during which time males are not available for remating. Hence, given that dung pats in nature become unattractive as oviposition sites quickly, multiple mating at the same dropping is relatively unlikely, so we believe our experimental setup simulates nature rather well.

STATISTICAL ANALYSIS

We used standardized regression methods to generate univariate linear selection differentials to assess the intensity of adult viability, female fecundity, and male sexual and fecundity selection on (adult) body size (Lande and Arnold 1983; Arnold and Wade 1984a,b). In general, for each population and replicate container, we produced standardized z-scores for body size (head width) by subtracting the sample mean from each value and dividing the difference by the standard deviation: inline image. Relative fitness was calculated as the absolute fitness component (i.e., adult longevity, female clutch and egg size, and male pairing success [1 or 0] or the body size of his female partner) divided by the population or container mean fitness (Arnold and Wade 1984b). We used models of relative fitness on z-scored body size inline image to estimate univariate linear selection
differentials.

To estimate viability selection, we regressed adult longevity on standardized body size, separately for the sexes and the replicate containers within populations. This yielded one viability selection estimate per replicate container. All five estimates per population were then averaged, yielding a corresponding confidence interval.

For female fecundity selection, we regressed relative clutch size or relative egg volume on standardized female body size. Selection coefficients of consecutive selection episodes are additive because fitness components are cumulative and hence multiplicative (Arnold and Wade 1984b). Thus, we can easily compute a female fecundity selection differential subsuming clutch and egg size. This yielded one fecundity selection differential per population with its appropriate standard error (or confidence interval) derived from regression.

A male's reproductive success is affected by both his mating success and the fecundity of his mate, which in turn depends on her body size (as above). We estimated sexual selection differentials based on mating success (males that copulated vs. those that did not) separately for each replicate container. Additionally, we regressed relative female body size (being proportional to her fecundity) on standardized male body size. Adding (i.e., subsuming) both yielded the male fecundity selection differentials, one estimate per replicate container for all populations and OSRs, which were then averaged, yielding a corresponding confidence interval (see, e.g., Blanckenhorn et al. 1999 for further details on these methods).

The above procedure describes calculation of the selection differential estimates. Significance testing, for all fitness components, was performed using the full models including continent, population nested within continent, replicate nested within population within continent (not applicable for female fecundity selection), and OSR (sexual selection only) as fixed or random factors and body size as a continuous covariate, including all relevant interaction terms. Variation in selection in all cases is established by significant factor by body size interactions. All analyses were done using the software SPSS Statistics version 19.0 (SPSS Inc.).

Results

COMMON GARDEN EXPERIMENTS

SSD is clearly reversed comparing the two continents, with populations displaying male-biased SSD in Europe and female-biased SSD in North America (Fig. 1; continent by sex interaction: F1,5 = 27.88, P = 0.003). Furthermore, European flies are on average larger than North American flies and take longer to develop
(Table 1; body size: F1,5 = 12.77, P = 0.016; development time: F1,5 = 5.46, P = 0.067; continent by sex interaction: F1,5 = 10.22, P = 0.023).

Figure 1.

Sexual body size dimorphism in seven cross-continental populations of the dung fly Sepsis punctum (Sample size:
Neurope = 498, Nnamerica = 618).

Table 1.  Population mean (± SE) for body size, development time, adult longevity, female clutch size, egg volume and male pairing success (under different OSRs) (sample
size, N).
PopulationCommon gardenAdult viabilityFemale fecundityMale mating success
SexHead width (mm)Development time (days)N"Housing" treatmentHead width (mm)Lifespan (days)NHead width (mm)First clutchNEgg volumeNOSRHead width (mm)
PairedNUnpairedN
Europe
AustriaMale1.16 ± 0.0714.25 ± 0.6528Male only1.08 ± 0.0769.16 ± 41.8291     11.09 ± 0.16221.09 ± 0.1818
 Female1.12 ± 0.1013.54 ± 0.9523Female only1.06 ± 0.0577.24 ± 35.19931.02 ± 0.1381.03 ± 2.54330.25 ± 0.013321.02 ± 0.17190.93 ± 0.1522
              41.04 ± 0.19220.92 ± 0.1758
 GermanyMale1.26 ± 0.0414.45 ± 0.5580Male only1.05 ± 0.1348.23 ± 35.31100     11.08 ± 0.54201.04 ± 0.0720
 Female1.21 ± 0.0313.54 ± 0.54100Female only0.98 ± 0.1362.15 ± 30.96991.09 ± 0.1159.61 ± 23.27560.24 ± 0.011921.09 ± 0.05201.04 ± 0.0620
              41.07 ± 0.06191.05 ± 0.0761
 SwitzerlandMale1.18 ± 0.1415.40 ± 0.4663Male only0.94 ± 0.1074.55 ± 42.3793     11.18 ± 0.09221.14 ± 0.0718
 Female1.15 ± 0.0914.91 ± 0.86105Female only0.97 ± 0.1071.42 ± 40.73850.98 ± 0.1369.83 ± 25.00570.24 ± 0.012021.18 ± 0.08201.15 ± 0.0720
              41.18 ± 0.09211.14 ± 0.0760
 SwedenMale1.29 ± 0.06 14.6 ± 0.8847Male only1.36 ± 0.1351.97 ± 27.009811.13 ± 0.18211.02 ± 0.1519
 Female1.21 ± 0.0514.27 ± 0.8452Female only1.07 ± 0.1065.42 ± 31.601001.06 ± 0.1489.80 ± 27.90300.26 ± 0.012921.16 ± 0.12200.99 ± 0.1520
41.21 ± 0.10201.00 ± 0.1760
North America
 CaliforniaMale1.10 ± 0.0313.93 ± 0.7283Male only0.97 ± 0.0468.70 ± 36.4391     11.01 ± 0.08210.99 ± 0.0919
 Female1.14 ± 0.0313.72 ± 1.81127Female only1.00 ± 0.0559.66 ± 34.27871.03 ± 0.1068.40 ± 18.00470.24 ± 0.013421.01 ± 0.08180.98 ± 0.0922
              41.00 ± 0.08161.00 ± 0.0959
 GeorgiaMale1.02 ± 0.0412.29 ± 0.8666Male only0.84 ± 0.1055.93 ± 33.4198     10.90 ± 0.06270.82 ± 0.0923
 Female1.05 ± 0.0312.55 ± 1.0067Female only0.87 ± 0.0978.76 ± 34.661010.94 ± 0.0771.79 ± 14.52460.24 ± 0.014020.92 ± 0.07150.92 ± 0.0825
              40.96 ± 0.03180.93 ± 0.0758
 New YorkMale1.05 ± 0.0314.34 ± 0.78112Male only0.98 ± 0.0451.30 ± 24.3779     11.02 ± 0.03191.00 ± 0.0419
 Female1.07 ± 0.0314.69 ± 0.84163Female only1.00 ± 0.0560.22 ± 33.081020.95 ± 0.1364.42 ± 24.94580.24 ± 0.011620.99 ± 0.03140.99 ± 0.0428
              41.00 ± 0.04130.99 ± 0.0466

TESTING THE DIFFERENTIAL EQUILIBRIUM MODEL

Adult viability (i.e., intrinsic longevity) selection

Adult viability was overall slightly positively related with body size (F1,1228 = 3.98, P = 0.046), thus implying no counterselection against large body size (contrary to expectation: cf.
Blanckenhorn 2000). This effect could largely be attributed to the Austrian males and the New York population (both sexes); all other populations showed no effect whatsoever of body size on adult longevity (Table 2; mean level and range indicated in Fig. 2). Standardized adult viability selection coefficients for males range between −0.048 ± 0.105 (95% CI) for the Swedish population and +0.157 ± 0.493 for the Austrian population; for females, the range is from −0.010 ± 0.102 (95% CI) for the Georgian population and +0.079 ± 0.246 for the Austrian population (Table 2). There were strong systematic differences between the sexes in longevity (females living longer on average; F1,1228 = 16.86, P < 0.001), some unsystematic variation among populations (F5,28 = 2.55, P = 0.050), but no significant difference between the continents (F1,28 = 0.04, P = 0.847; corresponding sex by factor interactions also n.s.). Viability selection for body size consequently was largely nil and did not vary systematically between the continents, the sexes, or the populations (all corresponding factor by body size interactions P > 0.1, except the three-way sex by population by body size interaction: F5,1187 = 3.33, P = 0.005).

Table 2.  Univariate selection differentials (mean ± 95% CI) for adult viability selection (βVS), female fecundity selection (βFS), male sexual selection (βSexS), and male fecundity selection (βmFS).
PopulationAdult viabilityFemale fecundity βFSMale reproductive success
“Housing” treatmentβVSOSRβSexSβmFS
Europe
    10.013 ± 0.1770.040 ± 0.177
 AustriaFemale only0.079 ± 0.2460.248 ± 0.12720.324 ± 0.4360.363 ± 0.436
 Male only0.157 ± 0.493 40.438 ± 0.0950.532 ± 0.095
    10.311 ± 0.1650.428 ± 0.165
 GermanyFemale only0.020 ± 0.1480.326 ± 0.05320.427 ± 0.4750.418 ± 0.475
 Male only−0.030 ± 0.105 40.208 ± 0.3560.190 ± 0.356
    10.179 ± 0.2150.302 ± 0.215
 SwitzerlandFemale only0.013 ± 0.2230.291 ± 0.05720.206 ± 0.4220.219 ± 0.422
 Male only0.009 ± 0.163 40.388 ± 0.6980.389 ± 0.698
    10.306 ±0.4740.340 ±0.474
 SwedenFemale only0.026 ± 0.1270.263 ± 0.06320.521 ± 0.7560.557 ± 0.756
 Male only−0.047 ± 0.020 40.876 ± 0.1780.881 ± 0.178
    10.260 ± 0.1980.310 ± 0.198
North America
 CaliforniaFemale only0.016 ± 0.1650.169 ± 0.05720.163 ± 0.3850.154 ± 0.385
 Male only−0.004 ± 0.157 40.117 ± 0.5980.099 ± 0.598
    10.329 ± 0.3930.362 ± 0.393
 GeorgiaFemale only−0.010 ± 0.1020.170 ± 0.03320.015 ± 1.0640.007 ± 1.064
 Male only0.032 ± 0.181 40.342 ± 0.5110.330 ± 0.511
    10.053 ± 0.1200.163 ± 0.120
 New YorkFemale only0.025 ± 0.1310.343 ± 0.04720.155 ± 0.7990.157 ± 0.799
 Male only0.068 ± 0.338 40.229 ± 0.6590.237 ± 0.659
Figure 2.

Mean fecundity (sexual) selection intensity on male body size in seven cross-continental populations of the black scavenger fly Sepsis punctum at three operational sex ratios (OSR). White, gray, and black boxes show selection intensity increases with OSR (i.e., male competition). The (equal) levels of fecundity selection on female body size (light gray bars; confidence limits) and of adult viability selection (dark gray bars; confidence limits) on female and male body size do not differ significantly between the continents.

Fecundity selection

Larger females lay larger clutches in all populations (overall strong main effect of body size: F1,317 = 610.58, P < 0.0001; Table 2). Standardized female fecundity selection coefficients based on clutch size range between 0.169 ± 0.057 (95% CI) for the California population and 0.343 ± 0.047 for the New York
population (mean and range indicated in Fig. 2 and Table 2). Clutch size varied among populations within continents (F5,317 = 15.63, P = 0.001), but not between continents (F1,5 = 0.63, P = 0.427). Crucially, fecundity selection differentials on body size (based on clutch size) did not vary among populations within continents (population by body size interaction: F5,317 = 1.35, P = 0.244) or among continents (continent by body size interaction: F1,317 = 1.17, P = 0.280).

Overall, larger females also laid larger eggs (main effect of body size: F1,175 = 15.85, P < 0.001; Table 2), but the relationship with body size was much weaker. Corresponding standardized female fecundity selection coefficients based on (cube-root-transformed) egg volume range between 0.002 ± 0.010 (95% CI) for the Swedish population and 0.020 ± 0.015 for the New York population. We had egg volume data for about half of the clutches treated above, which varied unsystematically among populations within continents (F5,175 = 6.08, P < 0.001), but not among continents (F1,5 = 0.96, P = 0.443). However, when tested against the global error, eggs were significantly smaller in North America
than in Europe after controlling for body size (F1,175 = 6.01,
P = 0.015). Nevertheless, fecundity selection on body size based on egg volume did not vary among populations within continents (population by body size interaction: F5,175 = 0.49, P = 0.781) or among continents (continent by body size interaction: F1,175 = 2.10, P = 0.148).

Sexual selection

In the European populations, 42 of the 48 replicate sexual selection differentials based on pairing success were positive, indicating strong sexual selection for larger male body size. Furthermore, sexual selection for large males intensified with increasing OSR and with body size, supporting Rensch's rule (Fig. 2). Sexual selection differentials for the American populations were also generally positive (27 out of 36) albeit lower, but there was no clear pattern of increased selection with OSR (Fig. 2 and Table 2). The full (logistic) general linear model consequently indicated overall strong positive effects of body size (head width) on pairing success (F1,930 = 22.23, P < 0.001), a significant interaction of continent and OSR (F2,22 = 3.34, P = 0.044), and, most importantly, a significant OSR-by-continent-by-body size interaction (F2,930 = 3.87, P = 0.021). The latter demonstrates variation in sexual selection on body size among the continents and the three OSR treatments.

Selection differentials reflecting assortative mating by size given pairing and hence the fecundity of the female partner were weak in comparison and did not vary significantly, ranging from −0.018 to 0.123; nevertheless, on average these added to the sexual selection differentials based on pairing success, making the combined male fecundity selection differentials even more positive across all populations and OSRs (73 out of 84)
(Table 2).

Discussion

We have shown here that a unique reversal in SSD between
European and Northern American populations of the black scavenger fly S. punctum is associated with, and presumably mediated by, substantial differences in the strength of positive sexual selection on males. As a result, European flies are larger than North American flies and SSD is male-biased and stronger, in agreement with Rensch's rule (Fairbairn 1997; Blanckenhorn et al. 2007b; Fairbairn et al. 2007). European females are also larger than North American females despite no differences in fecundity selection on female size, but this can be expected due to a genetic correlation in body size between the sexes alone (Fairbairn 1997). In
European (but not North American) populations, sexual selection also increased with the degree of male–male competition for females (i.e., the OSR), as expected by sexual selection theory (Bonduriansky 2001). This outcome confirms the differential equilibrium model of the evolution of SSD (Andersson 1994; Blanckenhorn 2000; Preziosi and Fairbairn 2000).

We emphasize that although we were able to show an association between sexual selection intensity and SSD (and probably mating system) evolution in accordance with the differential equilibrium model, such evidence must remain correlational as we cannot reconstruct the causality of evolutionary events. This is because evolutionary shifts in mating behaviors and the mating system are expected to be rapid and intimately associated with changes in sexual selection intensity, ultimately affecting the evolution of body size and SSD (Ding and Blanckenhorn 2002).

We also emphasize that although we considered three major fitness components (viability, fecundity, and sexual selection), comprehensive treatment of all relevant aspects of selection affecting SSD evolution, let alone in the field, is virtually impossible in any single species (Blanckenhorn 2000, 2004). In particular, we did not assess juvenile viability selection on body size, which in animals with complex life cycles such as insects is unattainable because larval and adult body size traits cannot easily be compared and individuals that die before adulthood cannot be measured (Blanckenhorn et al. 1999). One of the main mechanisms selecting against large body size occurs because individuals often grow for longer time to become larger, which increases cumulative mortality (Blanckenhorn 2000, 2007; Blanckenhorn et al. 2007a). And indeed, European S. punctum have longer development times than North American ones and the sex difference in development time differs between continents (Table 1). However, because the differences in absolute time are small (Table 1), it is doubtful that juvenile viability selection against long development fully compensates the much stronger sexual selection for large male size in European flies (cf. Blanckenhorn 2007). Furthermore, assessment of intrinsic (i.e., physiological) adult viability in the laboratory, as done here, does not necessarily reflect extrinsic adult viability in the field. Moreover, assessing female fecundity selection in the laboratory is a limited approximation of reproductive output in the field (Clutton-Brock et al. 1988). Nevertheless, given no relationship of intrinsic longevity (life span) with body size here, we have confidence in our estimates.

Recent comparative studies have highlighted the rapid divergence in sexual dimorphisms and mating behavior in sepsid flies (Eberhard 2001; Ingram et al. 2008; Puniamoorthy et al. 2008; Puniamoorthy et al. 2009; Tan et al. 2010). There have also been very early reports of interesting courtship behavior in this family (Hammer 1941; Hafez 1948; Parker 1972a,b; Mangan 1976). In S. punctum, the cross-continental differences in SSD documented here are accompanied by stark differences in the mating system (not treated in detail here; Schulz 1999). North American populations display precopulatory courtship behavior in form of vigorous shaking of the male abdomen when approaching the female, a behavior that is absent in the European populations (Puniamoorthy et al., in prep.). In contrast, European males show no distinct precopulatory courtship but instead scramble and/or contest competition among males, as evident by frequent male–male mountings and common “take-overs” where a male displaces another mounted male (Parker 1972b; Zerbe 1993). In fact, our ongoing studies indicate that European females also remate more readily, whereas North American females remate very rarely (Puniamoorthy et al., in prep. cf. Teuschl and Blanckenhorn 2007). More detailed, in-depth behavioral studies of the systematic mating system differences between the continents should further help explain the reversal to male-biased SSD in Europe. Although the genetic distance between North American and European S. punctum is
almost 3% (based on the DNA barcoding gene: R. Meier et al., unpubl. data), European and North American flies readily hybridize and produce viable offspring (Schulz 1999; Puniamoorthy
et al., in prep.).

An increasing number of studies have documented considerable intraspecific variation in SSD, usually in response to environmental, latitudinal, or even altitudinal clines (e.g., Badyaev and Hill 2000; Teder and Tammaru 2005; Fox and Czesak 2006;
Stillwell and Fox 2007; Liu et al. 2010; Hu et al. 2011). Most of these studies treated (quantitative) variation merely in the magnitude of SSD. Our study is a unique exception in that we phenomenologically tested the differential equilibrium model of the evolution of SSD in a species showing strong qualitative variation in dimorphism. We could confirm the model by showing that sexual selection on male body size in S. punctum is consistently stronger in European than in North American populations, whereas fecundity selection acting on female body size and adult viability selection are weaker and not different between the continents. Unpublished molecular data by R. Meier and colleagues in Singapore (cf. Su et al. 2008) suggest that the SSD and mating system of North American S. punctum is the ancestral state as, like many invertebrates, most sepsid species display female-biased SSD. The male-biased SSD in European S. punctum populations is therefore presumably secondarily evolved due to sexual selection in association with a change in the mating system, as predicted by theory (Andersson 1994; Fairbairn 1997; Bonduriansky 2001; Ding and Blanckenhorn 2002).

Associate Editor: P. Lindenfors

ACKNOWLEDGMENTS

We would like to thank U. Briegel and S. Matic for assistance in maintaining the fly cultures, R. Meier, and his laboratory for the preliminary molecular data on S. punctum populations as well as
D. Berger and R. Walters for helpful discussions. All the raw data used in this study have been archived in the supplementary file (Puniamoorthy_
et_al_2012_SSD_ArchivedRawData.xls). All authors read and contributed to this manuscript and there is no conflict of interest. NP was supported by the National University of Singapore-Overseas Graduate Scholarship (NUS-OGS), MAS was supported by a grant from the
German Research Council (DFG), and WUB by a grant from the Swiss National Fund (SNF).

Ancillary