- Top of page
- Materials and Methods
- Conflict of Interest
- Supporting Information
In recent research rotifers have been established as a model system for eco-evolutionary questions (Yoshida et al. 2003; Becks and Agrawal 2010; Becks et al. 2012). For example, they have been successfully used to investigate the role of sex during adaptation (Becks and Agrawal 2012). Demonstrating adaptation in rotifers is difficult, since cryopreservation, that is, the possibility of comparison with a frozen ancestor, is not well established (King et al. 1983; Toledo et al. 1991). If cryopreservation techniques are not available, adaptation in experimental lines may be measured by following the increase of mean population fitness or of offspring over time (Becks and Agrawal 2012). However, it is difficult to estimate whether a single general high fitness clone evolved due to laboratory selection, or if specific adaptation is observed selecting for two or more different clones. This could be estimated by tests of local adaptation (Kawecki and Ebert 2004; Blanquart et al. 2013). So far, for rotifers under laboratory selection local adaptation was not shown in detail. Local adaptation was shown for natural populations (Campillo et al. 2011; Alcántara-Rodríguez et al. 2012), but true local adaptation has not been shown so far for laboratory selected populations. Some authors did tests of local adaptation in laboratory rotifer populations, but the requirements for local adaptation were only partly fulfilled (Becks and Agrawal 2010). As rotifers are used as important model organisms in laboratory natural selection experiments it seems to be important to get further insight into to process of adapation in such populations. Our purpose of this study was to show local adaptation in rotifers and track changes in fitness directly in the populations.
Local adaptation, which is adaptation to specific environental conditions, can be estimated from measurements of performance (fitness) of populations adapted to varying habitats (Kawecki and Ebert 2004; Hereford 2009). Organisms in their “home” habitat should have a higher fitness compared to conditions in an “away” habitat, or more precisely, in the habitat of the other population (Kawecki and Ebert 2004). Furthermore, the organisms of the “local” habitat (their home habitat) should have a higher fitness compared to organisms from a “foreign” habitat (Kawecki and Ebert 2004). Only if both criteria of “home versus away” and “local versus foreign” are fullfilled, local adaptation is confirmed (Kawecki and Ebert 2004).
Selection regimes in the laboratory can be categorized as laboratory natural selection (LNS) or artificial selection (Garland and Rose 2009). While LNS aims at selection of performance (fitness) and allows multiple solutions for optimisation, artificial selection aims at selection of specific traits, for example, bristle number in Drosophila (Garland and Rose 2009). Examples of LNS in the literature include experiments demonstrating adaptation to a harsh and a benign environment in rotifers (Becks and Agrawal 2012), or to two harsh conditions in yeast (Gray and Goddard 2012a).
Local adaptation was investigated for many different organisms and is a highly important pattern observed in natural and laboratory systems (Hereford 2009). In our study we used the monogonont rotifer Brachionus calyciflorus to test for local adaptation after LNS. An open question we wanted to answer is whether rotifers can develop local adaptation under LNS in a minimum of time, where both criteria of “home versus away” and “local versus foreign” are fulfilled. This time was estimated by following the change of fitness directly in the populations. Thus, our study tries to combine both, the process of selection and adaptation over time to estimate divergent selection and the test of local adaptation for two different environments. A similar study was done by Gray and Goddard (2012a) using yeast, but, to our best knowledge, not for metazoans like rotifers. Furthermore, in contrast to previous studies, which had measured the fitness of individuals extracted from the experimental populations at specific time points (Becks and Agrawal 2012; Gray and Goddard 2012a), we wanted to follow the process of adaptation continuously in the experimental rotifer populations. Thus we re-adjusted rotifer density to a fixed value (two individuals per ml) every 3–4 days, instead of exchanging a fixed proportion of the culture medium (which is standard practice in such experiments). This allowed us to obtain an estimate of population growth for each transfer interval. In addition, our experimental design ensured that we always selected for maximum growth rate rather than tolerance of high population density. The latter is an inevitable consequence of dilution regimes that involve an exchange of a fixed proportion of the culture volume (Flegr 1997).
Monogonont rotifers are cyclical parthenogens with haploid dwarf males (Fussmann 2011). Females normally reproduce by ameiotic parthenogenesis, but sporadically undergo sexual reproduction (Nogrady et al. 1993). Induction of sexuality and production of males and diapausing eggs (embryos waiting in a resting stage) is mainly density dependent and triggered by a mixis inducing protein (Gilbert 2003a; Stelzer and Snell 2003; Schröder 2005; Snell et al. 2006).
We tested the hypothesis that these small metazoans can rapidly build up local adaptation when faced with two different laboratory environments: a food-limited environment, and a high-salt environment (oceanic sea salt). As soon as fitness (population growth) of the diverging populations appeared to reflect adaptation to the new conditions, we tested in a separate assay for fitness of “local” populations versus “foreign” populations and compared “home” versus “away” conditions. In this assay, fitness of single females was quantified as the ability to produce the highest number of offspring per unit time, which corresponds to selection for high population growth rates. To ensure that sexual offspring actually did recruit into our populations we additionally tested for spontaneous hatching of the resting eggs.
- Top of page
- Materials and Methods
- Conflict of Interest
- Supporting Information
In this study we provide evidence that experimental populations of the rotifer B. calyciflorus rapidly develop local adaptation to high-salt and low-food environments under a selection regime that explicitly favours fast population growth. These results are comparable to results obtained for yeast (Gray and Goddard 2012a) and show that rotifers can be used as a model to test for local adaptation after laboratory natural selection. The study of Becks and Agrawal (2010) tested for local adaptation in two homogenous environments using rotifers, but could confirm local adaptation only partly as they could show the local versus foreign criterion for one side only. Thus our study fills this gap by providing data that local adaptation due to LNS can be rapidly observed in rotifers.
The level of adaptation seemed different for the two environments: it seemed to be higher in the high-salt than in the food-limited environment (Fig. 1). This observation was supported both by the data of the selection experiment itself (a larger increase of mean population growth over time in the salt-populations) as well as in a subsequent test for local adaptation. Both local populations had a higher fitness than the foreign populations and were more fit at “home” than in the “away” environment (Table 4 and Fig. 3). Populations raised in their local habitat had on average a 38% higher fitness than populations raised in the foreign habitat, which is comparable to the 45% value reported by Hereford (2009). The higher level of adaptation in the high-salt populations during the selection phase, measured relative to the normal-COMBO populations, were accompanied by higher levels of sex (i.e., males per female, diapausing eggs per female; Fig. 2). According to our results of the spontaneous hatching rate of diapausing eggs, we estimated that about 43 asexual and eight sexual generations passed during the selection and adaptation phase.
The decrease of fitness in the control population over time is difficult to explain. One might expect that a population from the field introduced to new conditions (e.g., our control environment) would also undergo adaptation, thereby increasing in fitness, a process called evolutionary domestication (Simões et al. 2007). Potentially, our normal-COMBO populations over-exploited their resources during the times between transfers in the selection experiment and thereafter ran into starvation. In that case, the transfer period might have been too long for the control. However, this time period was necessary for the two harsh conditions, where the population growth rates were reduced. In experimental yeast populations the growth rate r increased, but the yield in biomass declined (Jasmin et al. 2012). In our normal-COMBO populations rotifers might have been selected for starvation tolerance rather than high growth rates, but this is highly speculative. Because of these uncertainties we did not include the normal-COMBO populations in the test for local adaptation (which would have been a better design) and used this “benign” control only to detect a fitness increase for the time of selection in the treatments. Thus, the result of higher adaptation in the high-salt population cannot quantitatively be confirmed, which would be possible if these comparisons would be done. As there was something going on since the normal-COMBO populations probably also evolved during the selection experiment, the data from the normal-COMBO population might be regarded as a third treatments rather than a control. However, we propose that the observation of fitness decrease under normal COMBO conditions does not influence our interpretation of local adaptation in the two harsh conditions, which was the main focus of your study.
Our results suggest that the process of selection and adaptation can be influenced by the type of habitat, as we found a less pronounced increase in fitness in the food-limited population compared to the high-salt population, having a much higher slope in the LME. There are several possible explanations for this difference. First, our experimental population might have been at a different distance from the final adaptive peak in the two environments (Poelwijk et al. 2007). This could be due to a lack of relevant genetic variation in the low-food environment (Barrett and Schluter 2008). We isolated our rotifers from the field, which at least periodically should select for low food conditions and introduce adaptations to low-food conditions into the population. However, similar to that, adaptation to increased salinity should occur in populations that inhabit ephemeral water bodies. However, we detected adaptation only in the high-salt population but not in the food-limited population. The fitness of food-limited populations was 33% lower compared to the normal-COMBO environment, giving some room for adaptation. As there were no more selection pressures than food-limitation, one should expect that the food-limited populations should at least partially increase in fitness due to adaptation of metabolic pathways. Other studies found rapid adaptation to starvation of Drosophila melanogaster within a few generations by metabolic and behavioural changes (Schwasinger-Schmidt et al. 2012), which clearly demonstrates that organisms can increase their fitness under such conditions. A second explanation for the difference in adaptation to the two environments relates to the rate of sexual reproduction. In our experimental populations sexual rates were increased in the high-salt populations compared to the low-food populations. This observation is in agreement with the idea that sex might increase adaptation to new environmental conditions (Colegrave 2002; Kaltz and Bell 2002; Goddard et al. 2005; Becks and Agrawal 2010, 2012; Dudycha et al. 2013).
Our data on male and diapausing egg ratios suggest that rates of sex were elevated in the high-salt environment already from the beginning of the experiment and remained relatively high throughout (Fig. 2). Thus, we suggest that the high-salt environment either acted as a trigger for sex or, alternatively, that the normal-COMBO and low-food environment suppressed sex. Overall sex induction in Brachionus is mediated by a density-dependent protein, which is excreted by the females in a process analogous to quorum sensing (Stelzer and Snell 2003; Snell et al. 2006; Kubanek and Snell 2008). However there are several environmental and genetic factors that modulate the response to this chemical (Gilbert 2003b). For example, extreme conditions near the physiological limits of rotifers tend to suppress sexual reproduction (Snell 1986). Furthermore, the production and viability of diapausing eggs depends on the availability of food (Gilbert 2010). Thus food limitation might have partly repressed sexual reproduction in our experiment. As for the second hypothesis, we are not aware of any study showing that high-salt environments elicit elevated rates of sexual reproduction in B. calyciflorus, but we cannot exclude this possibility.
In addition to these phenotypic responses to environmental factors, rates of sexual reproduction might themselves evolve during selection experiments. For example, Becks and Agrawal (2010, 2012) showed that sex in their rotifer cultures was evolving towards higher levels during adaptation. Other authors (Smith and Snell 2012) recently found a rapid evolutionary increase of sex in rotifers under temporary conditions (i.e., short hydroperiod duration). Fussmann et al. (2003) and Stelzer (2011) demonstrated a rapid evolutionary loss of sex in benign environments in rotifer cultures (Fussmann et al. 2003; Stelzer 2011). Such evolutionary responses can be confirmed by standardized assays that measure the propensity for sex of individual clones isolated from the evolving populations at different times during the selection experiment. Since we did not conduct such assays we cannot rule out the possibility that such evolutionary responses occurred later in our experiment. However two observations suggest that a direct environmental influence was at least more important than an evolutionary response: First, our base population exhibited different sexual rates in high-salt versus low-food environments already at the beginning of the experiment. Second, our indicators of sex in the evolving populations (males per females, diapausing eggs per female) did not suggest that sexual rates changed considerably or directionally during the experiment (see Fig. 2). The data presented in Fig. 2 propose that sex did not evolve over time by selecting clones with high sexual propensity. The only change was observed for males per female, but this was very small and cannot be used to prove an evolutionary change. Thus, most likely food-limitation has suppressed higher sex levels in our food-limited clones.
We did not measure fitness of both treatment populations at the beginning of the experiment, since all populations consisted of the same 83 clones. Therefore, it is highly unlikely to see the observed pattern of local adaptation before the selection took place. Thus, we can assume that there was no trade-off for the populations at the start of the experiment. If the rotifers would have been perfectly adapted to one of the experimental conditions, because of similar selection in the field, we would not expect to see one population performing better in the home habitat after selection compared to the away conditions. It is possible, that the populations performed a little better in one environment than in the other as suggested by Fig. 1 when isolated from the field. But, after selection, populations had a higher fitness in their home habitat than in the away habitat. This can only be explained by local adaptation.
Our results support the idea that two different stressful environments (high-salt, low-food) might produce different levels of adaptation. The intensity of selection can be very important for adaptation (Robertson 1960; Jones et al. 1968; Rumball and Rae 1968; Frankham 1977). The effect of selection is predicted to first increase with intensity, but only decrease at very high intensity (Bell 2008). We used two stressful habitats in our study since we could not take an increase in fitness under benign conditions, such as normal-COMBO medium, for granted (Goddard et al. 2005; Gray and Goddard 2012b). For example, studies on Saccharomyces cerevisiae have shown that fitness increased over time in harsh environments but remained essentially unchanged in benign environments (Goddard et al. 2005; Gray and Goddard 2012b). In this respect it is interesting that the initial fitness of our high-salt populations was significantly lower than that of the low-food populations (see in Table 1), suggesting that the high-salt conditions were harsher than the food-limitation, but the fitness gain during adaptation was higher in the high-salt populations.
In a study by Becks and Agrawal (2012), two lines of B. calyciflorus, which were already adapted to harsh or benign conditions were transferred to the respective “away” environment while control populations stayed in the environment to which they were adapted. The authors found an initial drop in fitness of populations in the “away” environment followed by a recovery within approximately 50 days. Additionally, the level of sex increased during adaptation and sexual offspring had a higher fitness than asexual offspring (Becks and Agrawal 2012). In the study of Becks and Agrawal (2012) changes in fitness during adaptation were measured in separate assays which differed from the conditions experienced by the adapting populations: While the adapting populations were maintained under a serial dilution regime involving replacement of only 10% of the culture volume every other day, the separate assays quantified lifetime reproductive output of individually cultured females. One might argue that the serial dilution regime involves more or less stationary growth (i.e., a growth rate of approximately 0.05 per day, which is an extremely small value compared to the maximum growth rates that can be reached by B. calyciflorus) and very high population densities. By contrast the separate assays, which were conducted under benign conditions (individuals cultured with daily transfers to fresh culture medium), resemble exponential growth conditions in an “empty” habitat. In our study we tried to avoid such incongruity by providing very similar conditions in the adapting populations and assay individuals, respectively (see 'Materials and Methods'). Notwithstanding these methodological differences, the speed of adaptation and fitness gains in our populations adapting to the high-salt environment were quite comparable to those reported by Becks and Agrawal (2012).
Our data support the view that not a general fitness genotype evolved being superior under all laboratory conditions, but that at least two (probably several) different specialized genotypes locally adapted. However this is not an ultimate prove- a statement would require a detailed analysis of the evolved populations using molecular markers. The observation of local adaptation is hard to explain with the idea of a general fitness genotype, because it is not clear why this genotype should have a lower fitness in an away habitat. Even harder to explain is why there should be a foreign depression in case there is a general fitness genotype. Thus, we expect that at least two different genotypes were selected for the high-salt and food-limited environment, respectively. Local adaptation can be estimated for two different habitats (Kawecki and Ebert 2004; Konijnendijk et al. 2013). This can be two harsh, or a harsh and a benign habitat, but ultimately the habitats have to be different. For future research it would be interesting to evaluate local adaptation in rotifers under benign conditions. Finally, we think to test for local adaptation would be a good alternative in case cryopreservation is not suitable.