SEARCH

SEARCH BY CITATION

Keywords:

  • assortative mating;
  • Callosobruchus maculatus;
  • experimental evidence;
  • frequency-dependent selection;
  • host fidelity;
  • resource preference;
  • sympatric speciation

Abstract

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

Recent models support the idea of sympatric speciation as a result of the joint effects of disruptive selection and assortative mating. We present experimental data, testing models of speciation through frequency-dependent selection. We show that under high competition on a mixture of resources/hosts, strains of the Seed beetle, Callosobruchus maculatus, change their host fidelity and evolve a more generalistic behaviour in resource utilization among females. The change in host fidelity did not result in disruptive selection and was not followed by assortative mating. This means that only one of three fundamental prerequisites for sympatric speciation evolved as a result of the frequency-dependent selection. We conclude that for this process to work, a shift to a novel food resource as a result of selection must also lead to a loss of preference for the original resource such that individuals are only able to use either one of the two.


Introduction

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

There are two key challenges to the theory of sympatric speciation: the evolution of disruptive selection and the subsequent development of assortative mating (Dieckmann et al., 2004; Fitzpatrick et al., 2009). A set of models dealing with these issues are models where sympatric speciation is a result of disruptive selection for resource use where extreme phenotypes suffer less from competition, mate assortatively and produce offspring with the same benefits of reduced competition (Doebeli, 1996; Dieckmann & Doebeli, 1999; Doebeli & Dieckmann, 2000; Drossel & McKane, 2000). For sympatric speciation to occur, ecological conditions must induce disruptive selection in a population in such a way that the genetic variability is maintained and extreme phenotypes are favoured (Seehausen & Jacues, 1999). Given such ecological conditions, the mating system must evolve such that reproductive isolation arises between the phenotypes favoured by disruptive selection. Strong frequency dependence is required for disruptive selection to evolve in the first place, and if it does, adaptive diversification becomes selectively advantageous, as do adaptations that result in diminished gene flow between the emerging lineages i.e. assortative mating (Dieckmann et al., 2004). The basic idea when selection is frequency dependent is that fitness landscapes change dynamically during the process, as the phenotypic composition of the population changes. Thus, a population that starts out in a regime of directional selection may evolve into a state in which it experiences disruptive selection. What initially is an advantageous strategy becomes severely deleterious, simply because the same strategy is adopted by an overabundance of competitors resulting in a population of individuals occupying a fitness minimum, or an evolutionary branching point (Metz et al., 1996; Geritz et al., 1998). The scenario just described is referred to as the gold rush metaphor, where the basic cause of diversification is competition. The reason why assortative mating should evolve in such a situation is easy to understand. Under frequency-dependent disruptive selection, segregation and/or recombination trap the population at a fitness minimum for as long as mating remains random. As disruptive selection acts against intermediate phenotypes, individuals that mate assortatively are favoured by natural selection. Once assortative mating based on the trait under disruptive selection has evolved, reproductive isolation automatically evolves as the ecological trait diverges. Evolutionary branching follows and thus assortative mating enables the population to escape from its trap at an evolutionary branching point (Metz et al., 1996; Geritz et al., 1998). The evolution of assortative mating is crucial here, as evolutionary branching cannot come about without it (Doebeli & Dieckmann, 2003).

Despite the seemingly complicated mechanisms required, theory has identified a variety of evolutionary processes that under certain ecological conditions may result in sympatric speciation (Maynard Smith, 1966; Kondrashov & Mina, 1986; Doebeli, 1996; Dieckmann & Doebeli, 1999; Kondrashov & Kondrashov, 1999; Doebeli & Dieckmann, 2000; Via, 2001; Gavrilets, 2003, 2004). Ecological interactions, such as competition within a population, can induce strong frequency-dependent selection, generate disruptive selection and thus create an opportunity for rapid speciation in sympatry (Dieckmann & Doebeli, 1999; Kondrashov & Kondrashov, 1999; Turelli et al., 2001; Bolnick & Fitzpatrick, 2007). Also, several case studies have convincingly shown that sympatric speciation is not only plausible, but under particular conditions, even a likely scenario (Seehausen & Jacues, 1999; Fitzpatrick et al., 2009). Divergent mate choice evolving as a by-product of habitat preference is, for example, an important mechanism for sympatric speciation through the evolution of divergent host races in insects (Via, 2001; Berlocher & Feder, 2002; Dres & Mallet, 2002; Fry, 2003; Gavrilets, 2003, 2004). Such linkage disequilibrium is crucial in most scenarios of speciation with gene flow (Smadja et al., 2008). Many authors have concluded that linkage disequilibrium between mating and resource preference can only be maintained if the disruptive selection is strong and the fidelity of assortative mating is high (Felsenstein, 1976; Udovic, 1980; Kondrashov, 1986 and Kondrashov & Mina, 1986). However, in phytophagous insects, where hosts are a central part of the species life cycle, the evolution of habitat choice should facilitate the maintenance of linkage disequilibrium through partial prezygotic isolation because of host fidelity, as assortative mating comes for free with the resource utilization of organisms (Via, 2001; Berlocher & Feder, 2002; Dres & Mallet, 2002; Gavrilets, 2003, 2004; Fitzpatrick et al., 2009). According to Bush & Butlin (2004), a possible source of diversity among species of insects is in fact specialist species multiplying by shifts from existing to new hosts.

In this article, we test the hypothesis that a shift in host fidelity follows frequency-dependent selection in a highly competitive environment. We also test if assortative mating evolves as a consequence of host shift in sympatry. For this experiment, a laboratory population of the Seed beetle (Callosobruchus maculatus) has been used. By introducing a novel food resource and keeping strains on a mixture of the novel and the original resource for a number of generations under high competition, we can develop two predictions concerning the outcome. First, models of sympatric speciation via disruptive selection (Dieckmann & Doebeli, 1999; Kondrashov & Kondrashov, 1999) predict that ecological competition may induce strong frequency-dependent selection and thus create an opportunity for rapid speciation in sympatry. This will be seen as a shift in host utilization over generations. Secondly, models of sympatric speciation via host shift (Feder, 1998; Menken & Roessingh, 1998; Schluter, 1998; Via, 2001; Berlocher & Feder, 2002; Dres & Mallet, 2002; Gavrilets, 2003, 2004) predict that populations selected on different resources will evolve preferences for each resource and show resource-dependent assortative mating. This will be seen as a preference for mating with individuals utilizing the same type of resource when conducting mating trials between males and females of the same and different resource origin.

Methods

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

Study population

A population of the Seed beetle, C. maculatus, a cosmopolitan pest of stored legumes (Fabaceae), particularly beans of the genus Vigna (Fox et al., 2004) was used as model organism in this study. This species has nonoverlapping generations and a generation time of approximately 28 days. The population was originally collected in Brazil and has been cultured in a laboratory environment for approximately 20 years. Prior to the experiment, strains have been kept at constant large population sizes (several hundreds of individuals) on black-eyed beans, Vigna unguiculata. During the experiment, the populations were maintained in climate chambers with a relative humidity of 55% and a constant temperature of 26 °C. The resources used in this experiment are black-eyed beans and mung beans, Vigna radiata. They differ in size, softness and water content but are otherwise chemically very similar. The black-eyed bean is the original resource used by the model organism, and the mung bean is a novel resource to the species.

Host fidelity

In the beginning of the experiment, 150 individuals were extracted from each of ten replicated strains of C. maculatus cultured solely on black-eyed beans. The extracted individuals were then transferred to ten replicates of jars with a mixture of 100 g of black-eyed beans and mung beans to constitute the first generation. The mixture of black-eyed beans and mung beans was set at a ratio of 9 : 1, chosen to reflect relative natural conditions where a new, invading resource would exist at lower abundances than the original one in the natural habitat. The experiment lasted for twelve generations, and the strains were kept on the same mixture of beans for the duration of the experiment. In every generation, 150 newly hatched individuals in each strain were transferred to the mixture of 9 : 1 of fresh black-eyed- and mung beans to constitute the next generation. Thus, each new generation started with a population size of approximately 75 males and 75 females. When keeping 150 individuals on 100 g of beans, a highly competitive environment is created as there are not enough beans for all females to deposit their eggs on (one female requires 21 g of beans to deposit her entire egg load). The hypothesis behind this experiment is that the strong competition will force less competitive females to relax their fidelity to the preferred host and start using the alternative and novel host to a larger extent. Such a change in host fidelity should be seen as an increase in utilization of the novel resource as oviposition site over generations. To test this, after mating and egg-laying, ten randomly selected black-eyed beans (∼2,28 g) and ten randomly selected mung beans (∼0,61 g) were extracted from each replicate in every generation. The number of eggs laid on the extracted beans and the individuals later hatched from those beans were counted and compared over all generations. From this data, the change in number of eggs laid and number of individuals hatched could be analysed and compared over generations, as could the hatching success on both types of resources. Our prediction was a linear increase in utilization of the novel resource over time.

Female laying preference

To distinguish whether the loss of fidelity was because of a shift in host preference or simply a broadening of host use, an experiment for studying female laying preference over time was conducted. From the lines suffering under high competition, in each generation, two black-eyed females and two mung females were extracted, mated, and each put in a Petri dish on a 40 g mixture of the two bean types (20 g BE and 20 g M). As the females were removed from the crowded populations before egg laying, they did not experience any competition for resources and therefore were not forced to alternate their host preference. The number of eggs laid on the two hosts by each female in the absence of competition was counted for each generation and compared over all generations. The experiment was conducted for all ten replicates separately.

Female mate preference

After nine generations on the sympatric mixture of the two different resources, mating experiments were conducted in two different ways to see if any association had developed between host preference and mating preference. In the first mating experiment, virgin females from the two resources were introduced to virgin males from both types of resources, from now on referred to as mung males and black-eyed males. In treatment one, a mung female was put together with a sterile mung male and a fertile black-eyed male. In the second treatment, a mung female was put together with a fertile mung male and a sterile black-eyed male. In treatment three, a black-eyed female was introduced to a sterile black-eyed male and a fertile mung male, and in treatment four, a black-eyed female was introduced to a fertile black-eyed male and a sterile mung male. Treatments five and six functioned as control treatments where each female was introduced to two fertile males of her own kind. In each treatment, a number of 10 females were mated. All mating trials took place in Petri dishes with 21 g of BE beans. Mating took place within a few minutes after introduction, and both males were discarded after the female had mated with one of them. The female was then allowed to deposit her egg load, and the number of fertilized eggs and adults hatched was counted. Sterile males were used to be able to decide which male each female had chosen to mate with. Females mated to sterile males lay unfertilized transparent eggs whereas females mated to fertile males lay fertilized white eggs. Female preference for males of her own kind, i.e. assortative mating, should be seen as a low number of fertilized eggs laid and adults hatched in treatments one and three where males of the same kind as the female were sterile, and a high number of fertilized eggs laid and individuals hatched in group two and four, where males of the same kind as the females were fertile. The experiment was conducted for all ten replicates separately.

Males were sterilized prior to mating by irradiation with the dosage of 70 Gy for 56 min using the Caesium-137 source at the division of Biomedical Radiation Sciences, Uppsala University. This ‘sterile male’ technique is a well-used method in insect sperm competition studies and constitutes a reliable method of paternity assignment in C. maculatus. The dosage effectively causes 99.3% sterility in this species without noticeably affecting either behaviour or longevity in the irradiated males (Eady, 1991; Dowling et al., 2007).

Female mating propensity

In the second mating trial, female mating propensity was tested as virgin females from each resource were given a male to mate with from either of the two bean resources. In the first treatment, a mung female was introduced to a mung male, and in the second treatment, a mung female was introduced to a black-eyed male. The same reciprocal setup was performed for black-eyed females. A number of 10 females were mated in each of the two treatments. Matings took place in small Petri dishes, and time to mating and duration of mating were recorded and compared between the different treatments. This was done to see whether females were more prone to mate with males of their own kind, i.e. whether they put up less of a struggle before the male mounted and if they were willing to mate for a longer time, facilitating sperm transfer. This would function as a sign of assortative mating. After mating, both females and males were discarded. The experiment was conducted for all ten replicates separately.

Results

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

Host fidelity

The proportion of eggs laid on mung beans increased significantly over time (Spearman rs = 0.58, < 0.05, Fig. 1). The proportion of adults hatched from mung beans also increased over time (Spearman rs = 0.41, < 0.05). Hatching success varied significantly on mung beans over generations (F10, 98 = 2.40, = 0.013, ancova with the number of eggs as the covariate, Fig. 2a). The variation in hatching success on mung beans was random with respect to time as no significant positive or negative trend could be detected over generations (< 0.001, Kolmogorov–Smirnov one-sample test). No change in hatching success could be detected on black-eyed beans over generations (F10, 98 = 1.48, = 0.16, ancova with the number of eggs as the covariate, Fig. 2b).

image

Figure 1.  The proportion of fertilized eggs laid on mung beans over generations under high competition. There is a significant positive trend in the proportion of eggs laid on mung beans over generations. Vertical bars denote 95 confidence intervals.

Download figure to PowerPoint

image

Figure 2.  Hatching success over time. (a) Hatching success varied significantly on mung beans over time. The variation was random as no significant positive or negative trend could be detected over generations (see Results for details). (b) No change in hatching success could be detected on black-eyed beans over generations. Vertical bars denote 95 confidence intervals.

Download figure to PowerPoint

Female laying preference

This part of the experiment was performed with females not experiencing the competitive environment. The proportion of eggs laid on mung beans by black-eyed females decreased over time (Spearman rs = −0.28, < 0.05). Likewise, the proportion of eggs laid on mung beans by mung females decreased over time (Spearman rs = −0.36, < 0.05).

Female mate preference

In the first mating experiment, we found no significant differences in mate preference between females. There was a significant difference in number of fertilized eggs laid by females from different treatments (= 0.037, Kruskal–Wallis test). However, the expected pattern supporting assortative mating, i.e. a low number of fertilized eggs laid in treatment one (M female with sterile M male and fertile BE male) and three (BE female with sterile BE male and a fertile M male) and a high number of fertilized eggs laid in treatment two (M female with fertile M male and sterile BE male) and four (BE female with fertile BE male and sterile M male), could not be observed (see Fig. 3 for observed pattern). In Fig. 3, it is apparent that females of treatment 1 and 2 lay significantly more fertilized eggs than females of treatment 3 and 4. The reason for this is that females of group 3 and 4, simply by chance, have mated more frequently with sterilized males than has females from treatment 1 and 2. No significant differences could be detected in hatching success between the different treatments (F3, 35 = 0.38, = 0.77, ancova, with number of eggs as covariate, mean ± SE: treatment 1: 50.3 ± 9.27; treatment 2: 47.1 ± 9.27; treatment 3: 31.2 ± 9.27; treatment 4: 17.3 ± 9.27). When comparing the two control treatments, there were no significant differences in fertilized eggs laid (= 0.56, Mann–Whitney U-test) or adults hatched (= 0.41, Mann–Whitney U-test) between mung and black-eyed females. Hatching success was the same for both control treatments (F1, 17 = 2.22, = 0.15, ancova, with number of eggs as covariate, mean ± SE: treatment 5: 71.3 ± 2.70; treatment 6: 68.3 ± 2.70).

image

Figure 3.  Mean number of fertilized eggs laid by females from different treatments. In treatment one, a mung female was introduced to a sterile mung male and a fertile black-eyed male (Mf × BEm). In treatment two, a mung female was introduced to a fertile mung male and a sterile black-eyed male (Mf × Mm). In treatment three, a black-eyed female was introduced to a sterile black-eyed male and a fertile mung male (BEf × Mm). In treatment four, a black-eyed female was introduced to a fertile black-eyed male and a sterile mung male (BEf × BEm). In control treatments five (C Mf) and six (C BEf), each female type was introduced to two fertile males of her own kind. The observed differences between treatments does not correspond to the expected pattern in case of assortative mating, i.e. 1 and 3 < 2 and 4.

Download figure to PowerPoint

Female mating propensity

In the second mating experiment, no significant differences were found in time to mating between females mated to different types of males (M females: = 0.91, n = 18, Mann–Whitney U-test; BE females: = 0.84, n = 17, Mann–Whitney U-test). We could not detect any significant differences in duration of mating between females mated to different types of males (M females: = 0.30, n = 18, Mann–Whitney U-test; BE females: = 0.14, n = 17, Mann–Whitney U-test).

Discussion

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

There should by now be no question that divergent selection can counteract the homogenizing effect of gene flow, and thus maintain populations reproductively isolated from one another. Analyses of natural populations suggest that adaptive genetic divergence in only a few key traits, usually associated with resource use or mate choice, can lead to speciation through divergence with gene flow (Via, 2001; Via & West, 2008).

In our study, we found that females in populations experiencing high resource competition will after only a few generations relax their fidelity to a preferred host and start using an alternative and novel host to a larger extent. The change in host fidelity was shown as an increase in utilization of the novel resource as oviposition site over generations. These results implicate that a shift in host fidelity can be rapid, and in some cases, start almost instantaneously. The shift in host fidelity resulted in a decreased utilization of the original resource despite the severe shortage of ovipositional substrates and the fact that females were not able to deposit more than a fraction of their entire egg loads.

The increased preference for the novel resource over generations was not translated into an increased hatching success of the larvae on that same resource, which indicates that the change in host fidelity is a change in preference alone and does not at this stage translate into a change in adaptation to the new resource. However, a shift in host fidelity could lead to fitness consequences despite the lack of adaptation, in the form of decreased competition on the novel substrate compared to the original one. This is the reason why less competitive females should choose to relax their host fidelity in a highly competitive environment to gain in fitness as a consequence of lessened competition. Host shifts appear to occur commonly following the introduction of a new host that is relatively free of competing insect specialists (Bush & Butlin, 2004).

Females extracted from the competitive regime showed the exact opposite trend to the females experiencing high competition during egg laying, i.e. the fidelity to the original host grew stronger over time, and the preference for the novel resource type decreased. This leads to the conclusion that a more generalistic behaviour was employed by females under frequency-dependent selection to escape the shortage of resources and that disruptive selection was not the mechanism behind the shift in host fidelity in this experiment.

The change in host fidelity was not followed by a linkage between mating and host preference, i.e. assortative mating. This has earlier been shown to evolve in secondary contact in strains of C. maculatus selected for different resource preferences in allopatry (E. Rova & M. Björklund, unpublished data). Females tended to choose their mates at random and showed no propensity to let males mount them faster or to mate for longer periods of time, facilitating sperm transfer, with males of their own kind. The reason for this could be the lack of strong selection against intermediate phenotypes, facilitating the development of assortative mating, as strong divergent selection is necessary to initiate divergence in the face of substantial gene flow (Bolnick & Fitzpatrick, 2007; Butlin et al., 2008). Another possible explanation for the lack of assortative mating could be that the time span over which the experiment was conducted was insufficient for assortative mating to develop. In a founder flush experiment using Drosophila pseudoobscura, a considerable time was needed for any of the lines to develop prezygotic isolation (Powell, 1978; Dodd & Powell, 1985), and in two other studies, with a similar design as ours, about two to three times more generations of selection were needed for mating preferences to evolve as a result of environmentally based divergent selection (Rundle et al., 2005; Rundle et al., 2006). The number of generations is much lower in our study and thus could be part of the reason why we did not detect assortative mating in any of our strains after 12 generations.

There are three prerequisites for speciation to occur in sympatry; niche preference, niche adaptation and assortative mating (Maynard Smith, 1966; Coyne & Orr, 2004). The increase in utilization of the novel resource over time shown in this experiment suggests that there is a relaxation in host fidelity occuring by means of frequency-dependent selection because of high levels of resource competition. This indicates that the first challenge to the theory of sympatric speciation can be overcome in a matter of only a few generations. As no change in fitness/hatching success could be detected over generations despite the obvious change in host preference, no niche adaptation because of disruptive selection seem to have evolved. Without adaptation to the discrete niches, the evolution of assortative mating may be hindered as the trade-off between niches is lacking, selection against intermediate offspring is too weak and recombination keeps the two preference genotypes from diverging. Without substantial assortative mating within a group, the second challenge to the theory of sympatric speciation, divergence cannot go to completion (Maynard Smith, 1966; Coyne & Orr, 2004). For assortative mating to develop in sympatry, it is possible that a more substantial amount of time is needed. It could also be that the change in host preference did not translate into fitness differences because of the similarity of the alternative resources. Selection for a more generalistic behaviour in females, extending their resource spectrum, would then be a more likely explanation to our results, as there is no real cost to a generalistic behaviour. Using resources with more extreme differences may cause more prominent trade-offs between resources because of stronger selection and generalism to be to costly. This could in turn lead to fitness differences between individuals utilizing different resources, stronger selection against intermediate offspring and reinforcement facilitating the evolution of assortative mating and behavioural isolation.

In conclusion, we have shown that one of three major obstacles to sympatric speciation, the evolution of frequency-dependent selection, can be overcome through only a few generations of strong competition. However, rather than resulting in disruptive selection, we found that this selection led to a relaxation of host fidelity when an alternative resource was accessible. This means that the second important factor, the evolution of disruptive selection, is not an obvious result of frequency-dependent selection on different food resources. The third major obstacle, the evolution of assortative mating, could not be shown during this experiment.

The reason sympatric speciation has been such a controversial topic for so long is not geography in itself, but because of the fact that diversifying selection and sexual reproduction are two counteracting processes as gene flow tends to constrain divergence (Bolnick & Fitzpatrick, 2007; Fitzpatrick et al., 2008, 2009). The evolution of behavioural isolation is therefore the most critical step in modelling sympatric speciation and consequently further examination of the ecological conditions under which such isolation could evolve is needed. Ecological habitat choice or mating preference is probably what is required for assortative mating to evolve to facilitate sympatric divergence (Mallet et al., 2009). One idea is that, for the process of sympatric speciation to work, a shift to a novel host as a result of selection must at the same time lead to a loss of preference for the original host and that a trade-off evolves such that individuals are only able to use and mate on either one of the two hosts.

Acknowledgments

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

I thank Isobel Olivier for help in conducting the experiment, and Nina Svedin and Sara Bergek for constructive critique on the article. I am also grateful to the division of Biomedical Radiation Sciences, Uppsala University for access to their Caesium-137 source and especially Bo Stenerlöw and Martin Edvardsson for practical help with the irradiation treatments. This research was supported by funds from the Swedish Research Council (to MB).

References

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