SEARCH

SEARCH BY CITATION

Keywords:

  • Acaridae;
  • alternative reproductive tactics;
  • conditional strategy;
  • quantitative threshold trait model

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Alternative reproductive phenotypes (ARPs) occur across a wide range of taxa. Most ARPs are conditionally expressed in response to a cue, for example body size, that reliably correlates with the status of the environment: individuals below the (body size) threshold then develop into one morph, and individuals above the threshold develop into the alternative morph. The environmental threshold model provides a theoretical framework to understand the evolution and maintenance of such ARPs, yet no study has examined the underlying fitness functions that are necessary to realize this. Here, we empirically examined fitness functions for the two male morphs of the bulb mite (Rhizoglyphus robini). Fitness functions were derived in relation to male size for solitary males and in relation to female size under competition. In both cases, the fitness functions of the two morphs intersected, and the resulting fitness trade-offs may play a role in the maintenance of this male dimorphism. We furthermore found that competition was strongest between males of the same morph, suggesting that fitness trade-off in relation to male size may persist under competition. Our results are a first step towards unravelling fitness functions of ARPs that are environmentally cued threshold traits, which is essential for understanding their maintenance and in explaining the response to selection against alternative morphs.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The occurrence of discrete variation in male phenotype and reproductive behaviour has been shown to exist in populations where males compete for access to females (Gadgil, 1972; Shuster & Wade, 2003). Gross (1996) outlined three possible game theoretical outcomes for the evolution of such alternative reproductive phenotypes (ARPs): (i) pure strategies which form a genetic polymorphism; (ii) a mixed strategy, where an individual stochastically expresses different ARPs during its lifetime with probabilities matching those of genotypes in the polymorphism; or (iii) a single conditional strategy, where the phenotype expression depends on condition. Whereas only a few examples of pure strategies exist (the marine isopod Paracerceis sculpta (Shuster & Wade, 1991); the swordtail Xiphophorus nigrensis (Zimmerer & Kallman, 1989); and the side-blotched lizard Uta stansburiana (Ryan et al., 1992), no empirical evidence exists for mixed strategies (Gross, 1996). Instead, most ARPs fall into Gross’ (1996) third category and are conditionally expressed in response to an individual’s status or a proxy thereof such as body size (Oliveira et al., 2008).

The status-dependent selection (SDS) model (Gross, 1996) has been most influential in understanding the occurrence of condition-dependent ARPs. The SDS model assumes that individuals are genetically monomorphic for the conditional strategy and predicts that the average fitnesses of different ARPs are unequal. Shuster & Wade (2003) criticized the assumption of a genetic monomorphism, as this prevents quantifying the effects of selection on ARP expression, and also argued that the average fitnesses of ARPs within a conditional strategy are equal. The major problem with the SDS model as well as with Shuster & Wade’s (2003) critique, however, is that the arguments for equality or inequality of fitnesses of alternative phenotypes are mathematically unsupported (Tomkins & Hazel, 2007). What is more, Shuster & Wade (2003) modelled ARPs as genetic strategies rather than phenotypes that are expressed conditionally. To address these shortcomings of the SDS model as well as Shuster & Wade’s (2003) model, Tomkins & Hazel (2007) advocate using the environmental threshold (ET) model (Hazel et al., 1990, 2004) (Fig. 1), to understand the evolution and maintenance of condition-dependent ARPs. The ET model assumes that there are genetic differences between individuals in the switch point, or threshold, at which development switches from one morph to the other. The switch point is assumed to be under polygenic control that is itself sensitive to a cue (e.g. body size and hormone levels) that reliably correlates with the status of the environment. Assessing this ‘environmental’ cue against the genetically specified threshold of sensitivity at a specific sensitive period during an individual’s development elicits an all-or-none response resulting in the expression of one ARP or another. In contrast to the SDS model, the equilibrium mean switch point in the ET model may not correspond to the intersection of the fitness functions, and when the mean switch point is at its equilibrium, the average fitnesses of the ARPs can be equal or unequal (Tomkins & Hazel, 2007).

image

Figure 1.  The environmental threshold (ET) model (Hazel et al., 1990; Hazel et al., 2004). (a) In the ET model, genetic variation exists between individuals in the switch point at which development switches from one phenotype, X, to the alternative phenotype, Y (the mean switch point is indicated by the dashed vertical line, and the distribution of the switch point is indicated by the dotted, bell-shaped curve). This generates the solid black, cumulative normal curve of phenotype expression (here phenotype Y) in response to increasing values of an environmental cue such as body size. (b) The alternative phenotypes coexist because their fitness functions intersect: in this example, individuals of a body size smaller than the intersection point of the fitness functions s* benefit from expression phenotype X (grey line), whereas individuals of a body size larger than s* benefit from expression phenotype Y (black line). The horizontal arrows indicate the proportion of individuals of each alternative phenotype (grey for X; black for Y). The arrows gradually merge at s*. The proportion of individuals adopting each alternative phenotype is expected to track the intersection of the fitness functions if fitness functions change in response to selection (Tomkins et al., 2011). The dotted, bell-shaped curve in (b) indicates the distribution of body sizes.

Download figure to PowerPoint

Testing the assumptions and predictions of models of the conditional strategy is a prerequisite to understanding the maintenance and evolution of ARPs in single populations (Oliveira et al., 2008). To our knowledge, only one study, on the male dimorphic beetle Onthophagus taurus, has examined fitness functions of different ARPs within a conditional strategy (Hunt & Simmons, 2001). Male morph expression in O. taurus, however, is predominantly determined by differential parental investment into offspring rather than by genotypic differences between individuals (Moczek & Emlen, 1999) as assumed by the ET model. There still is no empirical investigation into the fitness functions of ARPs that are environmentally cued threshold traits, despite their importance in understanding the maintenance and evolution of condition-dependent ARPs using the ET model.

The goal of this study is to empirically investigate fitness functions of the ARPs in the male dimorphic bulb mite (Rhizoglyphus robini). We derived these functions in relation to male size for solitary males and in relation to female size for males in competition. The bulb mite is a polyandrous species where males are either fighters or scramblers. Fighters have a thickened and sharply terminated third pair of legs which they can use to kill other males (Radwan et al., 2000). Fighters also behave more aggressively towards other fighters than towards scramblers (Radwan et al., 2000). Scramblers have unmodified legs and are defenceless, but they live longer and mature more quickly than fighters (Radwan & Bogacz, 2000; Smallegange, 2011). The mechanism for male morph coexistence in this species still puzzles evolutionary biologists (Radwan, 2009). However, in contrast to other acarid mite species such as R. echinopus and Sancassania berlesei (Radwan, 1995, 2001), male morph expression in the bulb mite is known to be independent of variation in population density. Instead, male morph expression shows a threshold response to food quality: mite body size is positively correlated with food quality, which means that if food quality is high (and food is abundant), more males reach the final instar size threshold during development and emerge as fighters (Radwan, 1995; Smallegange, 2011). The threshold underlying male morph expression in the bulb mite also has an appreciable heritable component of h2 = 0.3–0.4 (Smallegange & Coulson, 2011), which fulfils the critical assumption of the ET model that there are genotypic differences in the threshold at which development switches from one morph to the other. The ET model (unlike the SDS model which assumes male morph expression is genetically monomorphic) is therefore the best candidate model to describe the conditional strategy of male morph expression in the bulb mite. Here, we describe an experiment where we continuously observed the behaviour of a fighter or scrambler mating pair using a set-up where the mating pair was alone or in the presence of a fighter or scrambler competitor. We found a fitness trade-off between male size and male morph for solitary males, and a fitness trade-off between female size and male morph under competition.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Experimental procedure

All individuals used were from the Imperial College London stock cultures which are maintained as detailed in Smallegange (2011). The experiment consisted of two treatments: focal morph (fighter or scrambler) and competitor type (no competitor, fighter competitor or scrambler competitor), resulting in 2 × 3 = 6 treatment combinations. Prior to each trial, mites were photographed using a Lumenera Infinity 3.1 camera connected to a Zenith SRZ-4500 (7–45×) stereomicroscope and their body length (without mouthparts) measured to the nearest 0.1 μm using infinity analyze imaging Software (Lumenera Corporation, Ottawa, ON, Canada). To start a trial, a focal male was placed inside the experimental arena (10-mm-diameter plastic tube with a plaster of Paris and powdered charcoal base) either alone or with a fighter or a scrambler competitor. Five minutes later, a female was added. Females were virgins [obtained by individually isolating quiescent tritonymphs (final instar stage) from stock cultures] that had matured 2 days prior to the start of trials (females do not start laying eggs until 1–2 days after maturation). Adult males were taken from the stock cultures and individually isolated 2 days prior to the start of a trial to increase sperm reserves (Radwan & Siva-Jothy, 1996; Radwan, 1997). During this 2-day isolation period, males had ad lib access to yeast that was coloured with either red or green food dye, which allowed us to distinguish the focal individual from the competitor (as bulb mites are partly translucent so that coloured food in the digestive tract of the mites is visible). Food dye does not affect the behaviour or fitness of acarid mites (Woodring & Cutcher, 1968). Focal males were alternately coloured red or green (also if there was no competitor present). Focal and nonfocal males of the correct morph were randomly assigned to each trial.

Once all mites were placed inside the experimental arena, they were observed continuously and the mating and fighting behaviour of both the focal male and the competitor (if present) was scored and analysed (see the details below). Trials where no matings between either the focal or competitor male and the female had occurred within one hour (= 8) were excluded from the analyses. In bulb mites, mating occurs when a male mounts a female to take the mating position, with the sexes facing in opposite directions (Gerson et al., 1991). Mites may stay in this mating position for up to 6 h with males using their anal and tarsal suckers to remain attached to the female (Radwan & Siva-Jothy, 1996). Fighting behaviour in male bulb mites consists of (i) ‘leg-pushing’ whereby males push and strike at other males, often trying to mount the opponent to secure a grip on the opponent and (ii) ‘clasping’ whereby males use their legs to hold their opponent (Radwan et al., 2000). By clasping their opponent, fighters are sometimes able to puncture the cuticle and kill their opponent. We scored the duration of the first mating between the focal male and the female, the fraction of total trial time that males spent fighting (i.e. leg-pushing and clasping) during each trial (this included the time period prior to the focal mating), and the number of times that the competitor succeeded in displacing the focal male from his mating position during each trial (this included the time period prior to the focal mating). When no competitor was present, mites were observed and their behaviour scored until the first mating between the focal male and the female had ended. When a competitor was present, mites were observed for at least one hour during which multiple matings could occur (Radwan & Siva-Jothy, 1996). This means that the first mating between the focal male and the female was not necessarily the female’s first mating. If, after 1 h, the focal male was mating, the behaviour of the mites was scored until the mating pair had separated. After completion of each trial, the female was put into an individual tube and given ad lib access to yeast. After 2 days, the female was removed and all surviving offspring that successfully matured as adults were counted 10–12 days later.

Virgin females and males that were individually isolated prior to the start of trials, as well as the females and offspring that were kept after completion of each trial, had ad lib access to yeast and were kept in 10-mm-diameter plastic tubes with a plaster of Paris and powdered charcoal base (kept moist to prevent desiccation of the mites). These tubes were put in an unlit incubator at 24 °C and > 70% relative humidity. The tops of tubes were sealed by a circle of filter paper (allowing gaseous diffusion) held in place by the tubes’ standard plastic caps with ventilation holes cut into them.

Statistical analyses

Each of the six treatment combinations of focal morph and competitor type was replicated 25 times, resulting in a total of 150 trials. The experiment was carried out in five replicate blocks, where each block contained five replicates of each treatment combination (5 × 6 = 30 trials per block) and trials were conducted randomly within each block. Four response variables were analysed: the duration of the first mating of the focal male (y1), the fraction of time spent fighting (y2), the number of times that the competitor succeeded in displacing the focal male from his mating position (y3), and the number of offspring each female produced after completion of each trial (y4), which we used as a proxy for reproductive success. To analyse mating duration (y1) (mating duration was log-transformed prior to analysis to conform to the assumptions of normality), we assessed the effects of focal morph (F) and competitor type (C) using a general linear mixed-effects model (GLMM) with normal errors and with block (B) included as a random factor. Size of the focal male (SM) was included as a covariate as well as size of the female (SF) and its interactions with focal morph and competitor type (to test for interactive effects between female size and the treatments). The full statistical model for this analysis was:

  • image(1)

Next, using a GLMM with binomial errors, we analysed treatment effects on the fraction of time spent fighting (y2). Trials where no competitor was present were excluded from this analysis, and instead of including focal male size, we included the size difference between the focal and competitor male (SD) and size difference squared (inline image) as covariates. The reason for this is that game theory predicts that fighting bouts last longest when opponents match in size (Smallegange et al., 2007), potentially resulting in a hump-shaped relationship between size difference and intensity of fighting behaviour:

  • image(2)

We also used eqn 2 (using a GLMM with normal errors) to analyse treatment effects on the number of times that a focal mite was successfully displaced from his mating position by the competitor (y3). To analyse treatment effects on reproductive success (y4), we used a GLMM with normal errors and included focal male size and female size along with its interactions with focal morph and competitor type. Because the total mating time (TM) of each female differed between trials, we included this as a covariate as well. The full model was:

  • image(3)

Subsequently, we investigated the fitness functions. Male morph expression is correlated with final instar size, and this has been used as the environmental cue predicting male morph expression in a related mite, R. echinopus (Tomkins et al., 2011). Here, we related the fitness functions to adult male size. The reason for this was that there were logistic constraints to raise a sufficient number of scramblers: because juveniles cannot be sexed and only 30% of males develop into a scrambler on a rich diet (Smallegange, 2011), we would require seven times the number of scramblers (875 in total), for which we did not have the facilities. As final instar size and male size at maturity are correlated (Smallegange, 2011), however, adult male size is a good proxy to use as the environmental cue to link to male morph expression and fitness. We derived the fitness functions by analysing the relationship between focal male size and reproductive success using a GLMM with normal errors. This analysis differed from the previous one (eqn 3) as we used only those trials where no competitor was present to be certain that only the focal male fertilized the eggs. To test the prediction that the fitness functions of the two morphs intersect, we included the interaction between size of the focal male and focal morph. Size of the female was included in this analysis to correct for the fact that the number of offspring produced may depend on female size. As the previous analysis (eqn 3) revealed that total mating time affected reproductive success (see Results), we included this as well. The full model was:

  • image(4)

Finally, we calculated the mean switch point at which the probability of being a scrambler equalled that of being a fighter (= 0.5), by performing a logit regression of the binary response variable male morph expression of the focal male (‘0’, if the focal male was a fighter, and ‘1’, if the focal male was a scrambler) on size of the focal male. The mean switch point at = 0.5 is then calculated as −b/a, where a is the regression coefficient and b the intercept of the logit regression equation inline image. To assess whether focal male size significantly affected male morph expression, we compared the latter logit equation with a constant-only equation (inline image), using a likelihood ratio test. The likelihood ratio (Λ) is calculated as inline image where LLi is the log-likelihood of the full model (f) or constant-only model (c) and pi is the number of estimable parameters in the full and constant-only model. The likelihood ratio is inline image distributed, where ν is the difference in number of estimable parameters (in this case inline image).

In each analysis, a model simplification procedure was used whereby the full model was fitted, after which the least significant term was removed (starting with the highest order interaction) in case this deletion caused an insignificant increase in deviance (significance was assessed by performing a likelihood ratio test). This procedure was repeated until the model only contained significant terms (< 0.05). The random factor replicate block was never removed during model simplification. In the results section, we give the parameter estimates (inline image) from the minimal models of the explanatory variables that had a significant effect on response variables. These estimates are the coefficients in the linear regression model and represent the relationship between an explanatory variable or interaction and the response variable.

To assess whether nonsignificant effects of the treatments focal morph or competitor type were indeed nonsignificant (supporting the null hypothesis) rather than a result of low statistical power, we calculated the effect size between two means inline image and inline image as Cohen’s d (Cohen, 1988): inline imagewith s as the pooled standard deviation:

  • image

The standard deviation of Cohen’s d is calculated as:

  • image

which is used to calculate the 95% confidence interval around d as d ± 1.96·sd. Because we were particularly interested in behavioural differences between the two male morphs, we only calculated effect sizes between the two focal morphs and between the two competitor morphs.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Mating and fighting

The duration of the first mating of focal males was significantly affected by an interaction between female size and focal morph (F*SF: inline image = 5.99 ± 2.24 SE log(s) per mm; t108 = 2.68; = 0.009). Mating duration decreased with increasing female size for fighters, but not for scramblers (Fig. 2a). Mating duration was not affected by competitor type (C: inline image = −0.10 ± 0.19 SE log(s); t108 = 0.438; = 0.662). The effect sizes of fighter vs. scrambler focal morph and fighter vs. scrambler competitor morph were close to zero (Table 1). This supports the null hypothesis that, on average, the duration of a mating does not differ between (focal) fighters and scramblers or between focal males that are with a fighter competitor and focal males that are with scrambler competitor. The average duration of the first mating was 948 ± 66 SE s (or equivalently: 15.8 ± 1.1 SE min). The fraction of total trial time that males spent fighting was unaffected by any of the treatments (C: inline image = −1.38 ± 1.78 SE; =−0.768; = 0.443; F: inline image = −0.45 ± 1.46 SE; = −0.31; = 0.756) or covariates (SD: inline image = −1.98 ± 7.47 SE; = 0.265; = 0.791; inline image: inline image = −14.76 ± 69.28 SE; = −0.213; = 0.831; SF: inline image = −1.07 ± 10.31 SE; = −0.10; = 0.918). However, the effect size of competitor morph was large and significantly different from zero (Table 1), rejecting the null hypothesis that competitor morph does not affect the fighting behaviour of focal males. The effect size was positive as focal males on average spent relatively more time fighting with a fighter competitor (0.83%) than with a scrambler competitor (0.29%). The effect size of focal morph was close to zero (Table 1), supporting the null hypothesis that focal morph does not influence the average fraction of time spent fighting (fighters, 0.55%; scramblers, 0.46%). Note that males on average only spent a very small fraction of their time fighting.

image

Figure 2.  Mating behaviour and fitness in the bulb mite. (a) The duration of the first mating of the focal male decreased with increasing female size for fighters (black line and symbols) but not for scramblers (grey line and symbols) (= 109). (b) The number of times that a focal male was displaced from his mating position depended on both his morph and the morph of the competitor (black bars, fighter competitor; grey bars, scrambler competitor) (= 68). This number was highest when both the focal and competitor male were of the same morph. (c) The reproductive output of females increased as females mated for longer (= 142). (d) The fitness functions of fighters (black line and symbols) and scramblers (grey line and symbols) in relation to male adult body size intersect at body size 0.695 mm (= 50). Fitted functions are = −15.722 · + 31.385 for fighters and = 47.306 · x−12.346 for scramblers. The dotted line denotes the frequency distribution of male adult size. (e) The probability of being a scrambler (P) increased with increasing male adult size (x). The fitted function is inline image (inline image = 7.863, = 0.008), and the mean switch point occurred at body size 0.688 ± 0.042 SE mm (= 150). The horizontal grey line denotes the 95% confidence interval of the mean switch point. A slight jitter was added to the data in (e). Removing the top-left observation in (a) and the two top-right observations in (d) did not qualitatively change the results.

Download figure to PowerPoint

Table 1.   Effect sizes calculated as Cohen’s d (Cohen, 1988) of response variables with nonsignificant effects of focal morph or competitor type. Cohen’s d is calculated for the difference in the mean between focal males that are a fighter vs. a scrambler, and between focal males which were with a competitor fighter vs. a competitor scrambler. Trials where no competitor was present were excluded. Mating duration was log-transformed to conform to the assumption of normality.
Response variableFighter vs. scrambler focal maleFighter vs. scrambler competitor
Δ Meand95% CI intervalΔ Meand95% CI interval
Mating duration0.0980.115−0.261 to 0.491−0.148−0.164−0.676 to 0.348
Fraction of time spent fighting0.1880.109−0.300 to 0.5181.1040.6330.214 to 1.053
Offspring number−0.028−0.002−0.331 to 0.327−2.687−0.175−0.585 to 0.234

The number of times that a competitor succeeded in displacing the focal individual from his mating position was significantly affected by an interaction between morph of the focal male and morph of the competitor (C*F: inline image = 0.535 ± 0.213 SE; t68 = 2.515; = 0.014). When the focal male was a fighter, he was more often displaced from his mating position by another fighter than by a scrambler competitor (Fig. 2b). In contrast, when the focal male was a scrambler, he was more often displaced from his mating position by another scrambler than by a fighter competitor (Fig. 2b). The number of times that a competitor succeeded in displacing the focal individual from his mating position was unaffected by any of the covariates (SD: inline image = 0.130 ± 0.634 SE, t68 = 0.205, = 0.838; inline image: inline image = −0.271 ± 1.020 SE, t68 = −0.266, = 0.791; SF: inline image = 0.102 ± 0.801 SE, t68 = 0.128, = 0.899).

Reproductive success and fitness functions

Reproductive success was affected by the total time that females had mated for (TM: inline image = 3.32 × 10−3 ± 8.23 × 10−4 SE no. of offspring per second; t141 = 4.04; < 0.001). The longer females had mated the more offspring they produced (Fig. 2c). Neither the treatments (C: inline image = 1.84 ± 0.66 SE offspring, t141 = 0.66, = 0.510; F: inline image = −0.043 ± 2.22 SE offspring, t141 = −0.19, = 0.850) nor any of the other covariates (SF: inline image = 28.21 ± 16.75 SE no. of offspring per μm; t141 = 1.684; = 0.094; SM: inline image = 0.010 ± 0.018 SE no. of offspring per μm; t141 = 0.593; = 0.554; TF: inline image = 4.37 × 10−3 ± 8.63 × 10−3 SE no. of offspring per second; t141 = 0.51; = 0.611) influenced reproductive success. The effect sizes of fighter vs. scrambler focal morph and fighter vs. scrambler competitor morph were close to zero (Table 1), supporting the null hypothesis that, on average, the number of offspring produced does not depend on focal morph or competitor morph. Using the no-competition trials, where the focal male was placed alone with a female, we were able to analyse the fitness functions. The results revealed that there was a significant effect of the interaction between focal morph and size of the focal male on reproductive success (F*SM: inline image = 110.26 ± 46.67 SE offspring; t49 = 2.36; = 0.023) (Fig. 2d). The intersection point of the fitness functions was at an adult body length of 0.695 mm (Fig. 2d). Reproductive success was again affected by the total time that females had been mating (TM: inline image = 3.54 × 10−3 ± 1.71 × 10−3 SE no. of offspring per second; t49 = 2.08; = 0.043). The interaction between focal morph and size of the female was nonsignificant (F*SF: inline image = 60.73 ± 40.94 SE offspring; t49 = 1.48; = 0.145) and neither was there a significant effect of female size (SF: inline image = −18.39 ± 20.80 SE no. of offspring per μm; t49 = −0.88; = 0.381).

The likelihood ratio test of the logit regression equations revealed that the logit equation including male body length provided a significantly better fit than the constant-only model (Λ = 7.863, = 0.005). The mean switch point at which the probability of being a scrambler equalled that of being a fighter was at an adult body length of 0.688 ± 0.042 SE mm (Fig. 2e). The mean switch point was not significantly different from the intersection point of the fitness functions at 0.695 mm [inferred from the fact that the 95% confidence interval of the mean switch point overlaps with the intersection point of the fitness functions (Fig. 2e)].

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

It has long been hypothesized that in the bulb mite, the fitness of fighters is always higher than that of scramblers because fighters can kill rival males whereas scramblers cannot (Radwan, 2009). Fighters, however, are not very effective at killing rivals (Radwan & Klimas, 2001). In our experiment, males spent only a very small fraction of their time fighting and we did not observe a single (attempted) kill. Furthermore, there is no evidence that fighters (or scramblers) have a survival advantage when rare (Radwan & Klimas, 2001) and neither did we find that the mating duration of focal males was reduced by the presence of a fighter. Scramblers can live longer than fighters (Radwan & Bogacz, 2000), but Radwan (2009) suggests that this does not affect the relative fitness of the two morphs as male reproductive potential decreases with age. We investigated the reproductive success of solitary male bulb mites in relation to their morph and their body size and found that the fitness functions of scramblers and fighters crossed. Whether the crossing fitness functions provide a mechanism for the coexistence of the two male morphs in the bulb mite first of all depends on how these functions change under competition. Our results on the agonistic behaviour of the males suggest that, in a competitive setting, the shape of the fitness functions might not be that different from those found for solitary males. First, not a single fighter was observed to kill a competitor and neither was the mating duration of focal males reduced by the presence of a fighter. Second, on average, males spent very little time fighting. Competitors did try to displace focal males from their mating position and succeeded in doing so on average every fourth mating (Fig. 2b). Crucially though, we found that competitors were most successful in displacing the mating focal male from the female if the focal male was of the same morph as the competitor. What is more, the number of times that a displacement was successful was on average the same between two fighters as between two scramblers. This intra-morph competition could reduce the fitness of a mating male by lowering the intercept of the fitness function of each morph. However, as the strength of competition is comparable between the two morphs, this reduction might be the same for both morphs suggesting that the fitness functions will still intersect under competition. Other aspects of the environment, for example the relative occurrence of the two morphs or temporal variability in food quality, may also influence relative morph fitness and thereby the coexistence of the two morphs. As yet, however, there is no evidence that the relative frequency of fighters or scramblers affects their mating success (Radwan & Klimas, 2001). Furthermore, we found in a previous study that the stochastic population growth rates of fighters and scramblers are equivalent across a range of environments that differ in temporal variability in food quality (Smallegange & Coulson, 2011). Both findings suggest that, at least in the case of the bulb mite, stabilizing mechanisms such as density- or frequency dependence that could maintain their coexistence are weak. Incorporating frequency dependence and environmental stochasticity in the ET model has the potential to provide general insight into the role of these factors in the maintenance of conditional strategies.

A closer look at the intersecting fitness functions reveals that the reproductive success of solitarily fighters and scramblers differed because scrambler reproductive success increased with increasing male size, whereas fighter reproductive success did not. Developing fighter legs is associated with high costs as fighters emerge from larger instars, at a smaller size, and after a longer development time than scramblers (Smallegange, 2011). By not investing into fighter legs, scramblers might be able to invest resources elsewhere, such as in sperm production. Sneakers in Onthophagus beetles, for example, generally have greater testes than fighter males that guard females (Simmons et al., 2007). Larger ejaculates from larger testes, in turn, can increase female reproductive output by serving as nuptial gifts (Bonduriansky et al., 2005). Hence, if testis size is correlated with body size in scramblers, this could explain why scrambler fitness increased with body size. The higher physiological costs of developing and the maintenance of fighter legs might prevent fighters from investing similarly into sperm production, as a result of which fighter fitness would not increase with the body size.

The ET model assumes that the switch point at which development switches from one morph to the other is sensitive to a cue that reliably correlates with the status of the environment (Tomkins & Hazel, 2007). The environmental cue that has previously been linked to male morph development in acarid mites is final instar size, with scramblers emerging from smaller instars than fighters (Radwan et al., 2002; Smallegange, 2011; Tomkins et al., 2011). Figure 3a shows this relationship for bulb mites raised on the same food type that was used in this study using data taken from Smallegange (2011). Here, however, we observed that larger adults were more likely to be the scramblers. So why does this cumulative frequency distribution of male morph expression reverse during development? A possible reason is that, although scramblers emerge from smaller instars, they are larger at maturity than fighters (Smallegange, 2011), presumably because they do not invest resources into fighter legs (Radwan et al., 2002). Indeed, once the individuals from Fig. 3a had matured, the cumulative frequency distribution of male morph expression had reversed and the probability of being a scrambler was positively related to size at maturity (Fig. 3b). What is more, in this latter study, the mean switch point at which the probability of being a scrambler equalled that of being a fighter occurred at 0.686 mm size at maturity (Fig. 3b). This is almost exactly equal to the mean switch point found in this study (0.688 mm), which, in turn, was not significantly different from the intersection point of the fitness functions. We furthermore observed that the variance in the switch point distribution was very high. This high variance in switch point distribution suggests that there is high genetic variance in switch point expression and which, according to the ET model, would explain why there is such a large overlap between the two morphs in body size distribution (Fig. 2d).

image

Figure 3.  Reversal of the cumulative frequency distributions of switch points illustrated using data from Smallegange (2011). (a) The probability of being a scrambler (P) decreased with increasing final instar size. The fitted function is inline image (inline image = 9.946, = 0.002), and the mean switch point (dashed line) occurred at body size 0.532 ± 0.020 SE mm (= 107). (b) Once the mites from (a) had matured, the probability of being a scrambler increased with increasing size at maturity (x). The fitted function is inline image (inline image = 3.932, = 0.047), and the mean switch point (dashed line) occurred at body size 0.686 ± 0.212 SE mm (= 123). The horizontal grey lines in each panel denote the 95% confidence interval of the mean switch point. In (b), the confidence interval ranges from 0.262 to 1.110 mm and therefore spans the whole data range. Data are for males raised on ad lib access to yeast and which mothers had also had ad lib access to yeast. A slight jitter was added to the data.

Download figure to PowerPoint

The results also revealed a second candidate mechanism to maintain the male dimorphism. The number of offspring that a female produced was affected by how long she had mated for. The duration of a mating, in turn, was affected by female size, and this effect differed between the two male morphs. Whereas mating duration was hardly influenced by female size for scramblers, fighters showed a strong decline in mating duration with increasing female size. As a result, mating duration and its potential fitness pay-off were highest for fighters if males mated with smaller females, but highest for scramblers if males mated with larger females. The fact that this potential fitness trade-off was not reflected by a significant interactive effect of female size and focal morph on reproductive success under competition might have been due to the fact that we did not use sterile competitors.

In conclusion, our results are a first step towards unravelling the fitness functions of an environmentally cued threshold trait that is expressed within a conditional strategy. In contrast to the intersecting fitness functions in relation to female size, which we assessed under competition, the intersecting fitness functions in relation to male size were observed for solitary males. We are now going to assess whether the latter fitness trade-offs hold under competition and whether the male morphs differ in the probability of securing a mating. We will also explore whether a strategy whereby parents produce a mixture of alternative phenotypes, that is a mixed strategy of male morph production (Dawkins, 1980), could be an evolutionary outcome to explain male morph coexistence. All this is essential for understanding the maintenance and coexistence of ARPs and to understand and explain the response to selection against alternative morphs (Tomkins et al., 2011).

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank Jacques Deere and Maurice Sabelis for their comments and discussion. The work was funded by an ERC Advanced Grant awarded to Tim Coulson.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References