Experimental adaptation to marine conditions by a freshwater alga

Abstract

The marine-freshwater boundary has been suggested as one of the most difficult to cross for organisms. Salt is a major ecological factor and provides an unequalled range of ecological opportunity because marine habitats are much more extensive than freshwater habitats, and because salt strongly affects the structure of microbial communities. We exposed experimental populations of the freshwater alga Chlamydomonas reinhardtii to steadily increasing concentrations of salt. About 98% of the lines went extinct. The ones that survived now thrive in growth medium with 36 g⋅L⁻¹ NaCl, and in seawater. Our results indicate that adaptation to marine conditions proceeded first through genetic assimilation of an inducible response to relatively low salt concentrations that was present in the ancestors, and subsequently by the evolution of an enhanced inducible response to high salt concentrations. These changes appear to have evolved through reversible and irreversible modifications, respectively. The evolution of marine from freshwater lineages is an example that clearly indicates the possibility of studying certain aspects of major ecological transitions in the laboratory.

Chlamydomonas, constitutive and inducible response, evolutionary rescue, phenotypic plasticity, recombination, salt tolerance

From time to time, a lineage may become adapted to conditions that lie far outside those that would be tolerated by its ancestors. In most cases, this need imply no more than the ability to grow in a specific extreme environment, as in the evolution of antibiotic resistance in bacteria (Davies and Davies 2010) or heavy metal resistance in plants (Gregory and Bradshaw 1965). The evolved lineage then flourishes but does not become further modified. In exceptional cases, the novel conditions to which a lineage has become adapted are widespread in nature, and its new ecological attributes then have the potential to lead to an adaptive radiation.

Here, we report the evolution of a marine way of life in the freshwater alga Chlamydomonas reinhardtii. It has been suggested that the marine-freshwater boundary is exceptionally difficult to transgress (Lee and Bell 1999; Vermeij and Dudley 2000). In plants and yeasts, for example, moving between regions of different salt concentrations requires changes in influx, efflux, and containment of ions, as well as changes in the ability to detoxify reactive oxygen species (Brewster et al. 1993; Mendoza et al. 1994; Zhu 2000). The pressures that freshwater and high-salt conditions impose on microbes are so different that salt is more important in governing community composition than temperature, pH, substrate, or other physicochemical variables (Lozupone and Knight 2007). Transitions between the two conditions are consequently infrequent and ancient, as revealed by the large phylogenetic distances between freshwater and marine microorganisms (Logares et al. 2009). High-salt habitats are much more extensive than freshwater habitats, and beside the ocean covering 70% of the surface of the Earth include enclosed seas, inland saline lakes, and coastal saltmarshes. Hence, the transition from freshwater to marine conditions both enforces major physiological changes and provides an unparalleled range of ecological opportunities.

Individuals that encounter novel conditions, such as high salt concentration, may be constitutively able to tolerate them and to continue to grow and reproduce. The constitutive response may evolve if there are alleles segregating in the population that confer different degrees of tolerance. Alternatively, an individual that in its current state is unable to tolerate these novel conditions may be able to modify its state so as to be able to grow and reproduce, a process called phenotypic plasticity. The inducible response may be under genetic control through regulatory elements (e.g., lactase expression in E. coli; Dykhuizen and Hartl 1978; Dykhuizen and Davies 1980) and the capacity to mount an inducible response may itself evolve (Lande 2009). Hence, adaptation to a novel environment may be attributable to the evolution of the constitutive response or the induced response or both. Both processes have been shown to play a role in natural populations adapting to changes in the environment (Reale et al. 2003; Charmantier et al. 2008; Gienapp et al. 2008; van de Pol et al. 2012) as well as in facilitating macroevolutionary events such as the origin of new taxonomic groups and of novel traits (Wund et al. 2008; Rajakumar et al. 2012; Standen et al. 2014).

The extent to which the constitutive and inducible responses will evolve will depend on the availability of beneficial variation. A lack of variants with positive growth rates will limit the ability of natural selection to bring the population's mean phenotype toward the new optimal phenotype (Lynch et al. 1991). Not surprisingly, the most common outcome of changes in ecological conditions is therefore extinction (Burger and Lynch 1995; Bell and Collins 2008). In some cases, however, “evolutionary rescue” may occur (Gomulkiewicz and Holt 1995), with a population evolving to tolerate conditions that would have been lethal to its ancestor. Rescue is more likely in large populations (Bell and Gonzalez 2009; Willi and Hoffmann 2009), in diverse and sexual populations (Agashe et al. 2011; Lachapelle and Bell 2012; Bell 2013a), and when environmental deterioration is slow (Perron et al. 2008; Bell and Gonzalez 2011). Rescue is thought to involve positive genetic correlations of fitness between different levels of stress, such that tolerance of lethal stress is an indirect response to selection at lower levels of stress (Samani and Bell 2010; Gonzalez and Bell 2013).

Preexisting or evolved phenotypic plasticity can also lead to survival. In plastic individuals, the inducible response to changes in environmental conditions can trigger behavioral, physiological, or morphological changes which may decrease the distance between the phenotype of the individual and the phenotype that maximizes fitness. Phenotypic plasticity can lead to greater genetic variation if it reduces the effectiveness of selection (Draghi and Whitlock 2012) and reduce the rate of population decline following environmental change, and thereby provides an opportunity for genetic adaptation to occur (Chevin and Lande 2010; Gomez-Mestre and Jovani 2013; Schaum and Collins 2014).

Plasticity may eventually become constitutively expressed, a process called genetic assimilation (Waddington 1942, 1952, 1953; Schmalhausen 1949; West-Eberhard 2003; Pigliucci et al. 2006; Crispo 2007; Lande 2009; Pfennig et al. 2010). This may occur as the result of selection against plasticity if it is costly to maintain (Snell-Rood et al. 2010), through mutational degradation or drift following long periods of stasis (Masel et al. 2007), or through strong stabilizing selection, which reduces genetic variation and thereby attenuates the genetic correlation between plasticity and the mean breeding value (Lande 2009). The outcome of genetic assimilation is therefore a reduction in plasticity and the constitutive expression of a trait equivalent to that originally produced as a plastic response to the new environment. Genetic assimilation is often difficult to identify because the ancestral reaction norms are not known or because it can occur rapidly (Pigliucci and Murren 2003). Nevertheless, there is some evidence from natural populations that genetic assimilation may contribute to survival and adaptive radiation following environmental change (Gomez-Mestre and Buchholz 2006; Bull-Herenu and Arroyo 2009; Scoville and Pfrender 2010).

We propagated experimental lines of the green alga C. reinhardtii in gradually increasing concentrations of salt until we obtained lines capable of growing in seawater within about 500 generations. Chlamydomonas reinhardtii typically lives in soil and freshwater. The salinity of soil water is expected to vary depending on soil composition and anthropogenic fertilization, but the salinity of rainwater itself, or the overflow from rivers and lakes, is usually lower than 500 parts per million. The strains used to initiate this experiment have been propagated in the laboratory for over 10 years on medium containing 0.025 g⋅L⁻¹ NaCl (0.0004 M). The salinity of seawater on the other hand is about 35 parts per thousand or 35 g⋅L⁻¹ (0.6 M), of which about 90% is sodium (Na⁺) and chloride (Cl⁻). High salinity imposes strong osmotic and oxidative stresses in C. reinhardtii by disrupting the homeostasis of ions (Na⁺, Cl⁻, K⁺, and Ca²⁺) and degrading proteins, and thereby reducing rates of photosynthesis and cell division (Husic and Tolbert 1986; Neelam and Subramanyam 2013). In general, salinities between 5 and 7 g⋅L⁻¹ NaCl (0.085 and 0.120 M) are sufficient to reduce the growth of C. reinhardtii by about 50%, and salinities higher than between 8 and 15 g⋅L⁻¹ NaCl (0.137 and 0.26 M) are sufficient to suppress growth completely (Reynoso and de Gamboa 1982; Moser and Bell 2011; Lachapelle and Bell 2012). The marine way of life is therefore inaccessible to C. reinhardtii. A green alga, identified morphologically as a Chlamydomonas sp. was previously isolated off the coast of Japan and characterized for its high salt tolerance (Miyasaka et al. 1998, 2000; Tanaka et al. 2007). We use this strain as a comparison for the growth of our salt-selected lines in seawater.

To determine the mechanism of adaptation to high salt, we measured the constitutive and the inducible responses to different salt concentrations by manipulating the acclimation environment. We compared the reaction norms of the salt-selected lines to that of their ancestors and found that both types of response had been modified by natural selection. Plasticity for growth in low salt in the ancestors has been genetically assimilated in the salt-selected lines, and plasticity for growth in high salt has been enhanced. Our experiment does not by any means reproduce all of the stages in the colonization of the oceans by terrestrial or freshwater organisms. It does permit some components of this process to be implemented in the laboratory, however, where the mechanism of adaptation can be elucidated by replicated experiments.

Methods

BASE POPULATIONS

We isolated one spore from each of 40 different lines that had been propagated independently for two years in the laboratory, growing in the dark on medium supplemented with acetate. These dark lines, from now on referred to as the ancestors, were derived from a previous experiment (Bell 2005), whose ancestors were derived from a cross among standard laboratory strains (CC-124 × [CC-1952 × (CC-1952 × CC-2343)]). The lines have not experienced salt concentrations higher than 0.025 g⋅L⁻¹ NaCl (4.28 × 10⁻⁴ M) during more than 10 years of culture in our laboratory.

SELECTION EXPERIMENT IN EVER INCREASING SALT CONCENTRATION

Details of the initial stages of the selection experiment can be found in Lachapelle and Bell (2012). Briefly, experimental lines varying by their sexuality (asexual, facultatively sexual, or obligately sexual) and initial diversity (low or high) were propagated in an environment where the concentration of salt increased by 1 g⋅L⁻¹ NaCl every two growth cycles (i.e., every about 10 generations). The lines that survived longest came from high-diversity, sexually derived ancestors. The two lines able to grow in the highest concentration of salt (up to 30 g⋅L⁻¹ NaCl) were used for crosses to continue the selection experiment. It is this continuation of the experiment that we report here. A wild-type strain of opposite mating type to each line (CC-2935 mating type minus) was used to perform the initial cross. The progeny were then mated within and across the F1 families to generate the F2. Gamete fusion and zygote germination followed standard practice (e.g., Lachapelle and Bell 2012). We grew the progeny in 34 g⋅L⁻¹ NaCl for two growth cycles. Only 23 resistant recombinants survived out of about 10⁶ cells. The progeny was therefore clearly incapable of growth in 34 g⋅L⁻¹ NaCl, and these 23 surviving cells were presumably the ones with the least negative growth rates. We isolated them and propagated each individually, once again in gradually increasing concentrations of salt, starting at 24 g⋅L⁻¹ NaCl. The lines were cultured in 48-well plates with 1.4 mL of Bold's medium supplemented with salt, and transferred every week (two weeks when growth was poor) using a 0.2 mL inoculum. The salt concentration was increased every two or three growth cycles up until 36 g⋅L⁻¹, at which point it was maintained constant. From the 23 starting lines, 13 survived up to 36 g⋅L⁻¹, and 10 have subsequently survived repeated transfers in that concentration. At the time of assay, the surviving lines had been propagated for a total of about 500 generations since the beginning of the selection experiment (Fig. 1).

Figure 1.

Schematic of the history of the salt-selected lines. Only 10 lines are now surviving in 36 g⋅L⁻¹ NaCl.

SEAWATER GROWTH ASSAY

To determine whether adaptation to high salt had resulted in a transition from freshwater to marine conditions, we assayed the surviving salt-selected lines, the ancestral lines, the wild-type strain that was used to set up the crossing trial, and a related marine chlorophyte (Chlamydomonas sp. CW-80, isolated off the coast of Japan; Miyasaka et al. 1998) in seawater. The seawater was collected in August 2013 off the coast of Dunbar, UK, and filter-sterilized 2 h after collection. The assay was performed with the same inoculum size and cycle period that the salt-selected lines experienced during the selection experiment. The ancestral lines had been propagated in the dark, using acetate as a carbon source, for the duration of the selection experiment. For the assay, all lines were acclimated in Bold's medium without salt, in the light without acetate, for two cycles before being transferred to the seawater.

Cell density at the end of the first and second cycles in seawater was estimated for two independent replicate cultures using flow cytometry (BD FACSCanto II, BD Biosciences, Oxford, UK). The instrument was calibrated with CS&T beads, and sample acquisition was made using a high-throughput system. Data were acquired and analyzed with the BD FACSDiva version 6 software. Electronic analysis gates were applied to the forward scatter (pulse area FSC-A and width FSC-W) and side scatter (pulse area SSC-A) plots (proxies for cell size and complexity, respectively) to exclude events that are outside expectations for intact C. reinhardtii cells, as well as to sort the single cells from clumps of cells. We excluded clumps because we cannot estimate how many cells they contain. Clumps arise as a physiological response to salt in both ancestral and evolved cultures, and should therefore not bias our estimates of growth. All events that were inside the intact and the single-cell gates in a volume of 30 μl acquired at a rate of 1 μl⋅sec⁻¹ were used to estimate cell density in each culture. Culture samples with cell counts of 10 or fewer were not included in further analyses because of the potential for false positives at very low or zero cell density. Cell density at the end of the first cycle was used to estimate cell density at the start of the second cycle. We calculated the rate of increase per week as the natural logarithm of final cell density divided by initial cell density.

MEASURING THE INDUCIBLE AND CONSTITUTIVE RESPONSES TO SALT

To determine the extent to which the constitutive and the inducible responses to salt were altered in the selection lines, we performed assays comparing the salt-selected lines to the ancestors, and comparing the responses to salt after acclimation in medium lacking salt and in medium containing a stressful but sublethal concentration of NaCl. All assays, unless noted otherwise, were carried out in the light without acetate, as in the extended selection experiment. Note that by “constitutive” we mean that the phenotype is independent of environmental cues. Although constitutive phenotypes are generally associated with genetic changes, it is well recognized that epigenetic changes are mitotically stable (Jablonka and Raz 2009). A constitutive phenotype can therefore arise from genetic and/or epigenetic changes in asexual populations, and this is investigated as described in the following subsection.

The constitutive response was determined in two ways. First, we compared the growth of the salt-selected lines to the ancestral lines after a period of growth in medium lacking salt. The difference between the two selection histories reflects the direct response to selection and the degree of adaptation that is expressed without need for prior acclimation to salt. The assay was initiated by growing all lines in the light, in medium without salt Supplementation, for two growth cycles of one week each. After this period of acclimation, two replicates of each line were transferred to a range of salt concentrations (0, 5, 10, 15, 20, 32, 36, and 40 g⋅L⁻¹ NaCl) and grown for two cycles. Fitness was estimated as in the seawater growth assay described above. The difference in responsiveness (i.e., the change in the rate of increase as a function of salt concentration), as well as the amount of variance in growth that could be explained by the history of the lines (i.e., ancestral or salt-selected) was used to determine the degree of change in the constitutive response.

Second, we compared the contribution of constitutive and inducible responses to salt. Two replicates of each salt-selected line were acclimated in each of 0, 10, and 36 g⋅L⁻¹ NaCl for two growth cycles of one week each before being transferred to a range of salt concentrations (0, 10, 15, 20, 30, 36, and 40 g⋅L⁻¹ NaCl). Fitness was estimated as in the seawater growth assay. The variance of growth among lines estimates differences in the constitutive response, and the variance of growth among acclimation environments estimates differences in the inducible response.

We carried out a further assay to determine whether the inducible response to salt in the salt-selected lines is evolved or ancestral, and whether the response of the ancestral lines to salt is due to the salt itself or to photosynthetic growth. We assayed the ancestral lines in the dark and in the light after acclimation in medium lacking salt and in medium containing 5 g⋅L⁻¹ NaCl (because most ancestral lines cannot sustain growth in higher concentrations). After acclimation, growth was assayed over a range of salt concentrations (0, 5, 10, 15, 20, and 30 g⋅L⁻¹ NaCl).

CHARACTERIZING THE PHENOTYPE OF SEXUAL PROGENY

To examine further the mechanisms responsible for the evolution of the constitutive and the inducible responses to salt, we crossed each of two of the selection lines to an ancestral line to create F1 families, and then crossed within and between these families to create the F2. We chose eight random spores from each generation of each cross and acclimated them either in medium lacking salt or in medium containing 10 g⋅L⁻¹ NaCl. They were then assayed over a range of salt concentrations (0, 28, 36, 44, and 48 g⋅L⁻¹). If genetic changes are responsible for the evolution of the constitutive and/or inducible responses, we expect the sexual progeny to retain tolerance of salt to different extents depending on the number of genes involved and interactions among them. If reversible changes, such as epigenetic changes, are responsible for the evolution of the constitutive and/or inducible responses, we expect tolerance of the salt-selected lines to be annulled by meiosis.

STATISTICAL ANALYSES

Cultures for which estimates of the initial or final cell densities were zero were removed from the analysis to permit model fitting. The removal of some datapoints led to unbalanced designs in most cases, so we calculated type III sum of squares in all analyses of variance using the R package “car” (Fox and Weisberg 2011).

To compare the constitutive response in the high-salt lines to the constitutive response in the ancestors, we fitted a linear mixed-effects model using the lmer function in the R package “lme4” (Bates et al. 2012), with selection history as a fixed factor, line nested within selection history as a random factor, salt assay concentration (between 0 and 20 g⋅L⁻¹ NaCl where the relationship is linear) as a continuous variable, and the interactions as factors. We allowed for random intercepts and random slopes. Type III Wald tests were performed to determine significance of the fixed effects.

To compare the constitutive and inducible response in the ancestors when grown in the dark or in the light, we fitted a linear mixed-effects models using the lmer function, with acclimation regime (with or without salt) and condition (dark or light) as fixed factors, assay salt concentration as a continuous variable, line as a random factor, and all interactions. We allowed for random slopes and intercepts.

To test the hypothesis that plasticity in the ancestors has been genetically assimilated in the salt lines, we fitted a linear mixed-effects model using the lmer function with selection history as a fixed factor, lines nested within selection history as a random factor, assay salt concentration (between 0 and 10 g⋅L⁻¹) as covariate, and all interactions. The data used in this analysis come from the ancestors acclimated with salt (inducible response) and the salt lines acclimated without salt (constitutive response).

To determine the effect of acclimation in different concentrations of salt on the high-salt lines, we fitted an analysis of covariance (ANCOVA), with acclimation regime as fixed factor, line as a random factor, assay salt concentration as a covariate, and all interactions. Variance components were then calculated by equating observed and expected mean squares.

To compare the inducible responses in the ancestors to that in the high-salt lines, we fitted a linear mixed-effect model using the function lmer with selection history as a fixed factor, lines nested within selection history as a random factor, assay salt concentration (between 10 and 20 g⋅L⁻¹, or between 20 and 30 g⋅L⁻¹) as a continuous variable, and all interactions. Note here that to look at the evolution of the inducible response, we used data from the ancestors acclimated with 5 g⋅L⁻¹ NaCl and data from the high-salt lines acclimated with 10 g⋅L⁻¹ NaCl.

Finally, to compare the growth of the salt-selected lines and the ancestor to that of the sexual progeny, we calculated confidence intervals for the difference between means, using the t-distribution for unequal sample sizes.

Results

SALT-SELECTED LINES CAN GROW IN SEAWATER

The marine isolate grew well in seawater and could be propagated successfully. The freshwater isolate and all the ancestral lines were incapable of growth in seawater and could not be propagated. The high-salt selection lines had positive growth on average although they varied widely (Fig. 22 mean r = 0.26, variance among lines = 1.57). About half of the high-salt lines (7/13) have positive growth in seawater, although only 2/6 significantly so (one-tailed t-tests for the difference between an estimate and a parametric value; one line could not be tested for significance because of insufficient replication). Some of these lines grew as well as, or even better than, the marine isolate, at least in laboratory conditions.

Figure 2.

Growth of the ancestral lines, marine green alga strain C.W80, salt-selected lines, and wild-type freshwater strain CC2935 in seawater. Each point is the mean of two assay replicates for a given line. There are 20 ancestral lines, 13 salt-selected lines, and one of each of the marine green alga and the wild type.

SELECTION ALTERED THE CONSTITUTIVE RESPONSE TO SALT

The high-salt lines maintain a high positive rate of increase from 0 g⋅L⁻¹ up to 20 g⋅L⁻¹ (Fig. 3: r = 1.75 + 0.02 [NaCl]), whereas growth of the ancestral lines decreases sharply as the salt concentration increases (r = 1.61 − 0.19 [NaCl]). Some ancestral lines have a negative rate of increase at concentrations as low as 5 g⋅L⁻¹ NaCl, and the mean rate of increase is well below zero by 10 g⋅L⁻¹ NaCl. The difference between the response of the high-salt lines and the ancestral lines to salt is highly significant (effect of interaction history:assay salt concentration: Χ² = 94.65, df = 1, P < 0.001).

Figure 3.

Growth of the salt-selected lines and the ancestral lines in different concentrations of salt. There are 13 salt-selected lines and 20 ancestral lines, each assayed twice. The datapoints for the salt-selected lines are plotted 1 g⋅L⁻¹ NaCl higher than assayed to make it easier to see differences between histories. The trend line was fitted using local polynomial regression (loess), with 95% confidence intervals in shade.

THE ANCESTRAL LINES SHOW AN INDUCIBLE RESPONSE TO SALT

Most of the ancestral lines cannot grow in salt concentrations above 5 g⋅L⁻¹ when acclimated in medium without salt. When acclimated in 5 g⋅L⁻¹ NaCl before assay, however, most ancestral lines are able to grow in salt concentrations as high as 30 g⋅L⁻¹ (Fig. 4). Between 0 and 10 g⋅L⁻¹, where the relationship is linear, the growth of the ancestral lines decreases significantly more rapidly with increases in salt concentrations when they have been acclimated without salt than when they have been acclimated with 5 g⋅L⁻¹ NaCl (Table 1; effect of interaction between acclimation and concentration: Χ² = 32.96, df = 1, P < 0.001).

Figure 4.

The effect of acclimation regime and growing condition (light or dark) on the growth of the ancestral lines in different concentrations of salt. There are 20 ancestral lines, each assayed twice. The datapoints for the lines acclimated in 5 g⋅L⁻¹ NaCl are plotted 1 g⋅L⁻¹ NaCl higher than assayed to make it easier to see differences between acclimation regimes. The trend line was fitted using local polynomial regression (loess), with 95% confidence intervals in shade.

Table 1.

Effect of acclimation and growing condition (i.e., light or dark) on the response of the ancestral lines to salt concentrations between 0 and 10 g⋅L⁻¹

	Acclimation
Growing	environment	Intercept	Slope
condition	(g⋅L−1 NaCl)	(±SE)	(±SE)
Light	0	2.2 (0.096)	−0.22 (0.017)
	5	2.2 (0.11)	−0.12 (0.022)
Dark	0	2.3 (0.11)	−0.13 (0.018)
	5	1.9 (0.091)	−0.066 (0.017)

THE INDUCIBLE RESPONSE OF THE ANCESTRAL LINES IS EXPRESSED IN BOTH LIGHT AND DARK CONDITIONS

Growth decreases more rapidly with salt concentration when the ancestral lines are grown in the light than when grown in the dark (effect of interaction between growth condition and salt concentration: Χ² = 10.36, df = 1, P = 0.0013). This is attributable to the higher growth of lines growing in the light than in the dark in medium without salt supplementation, however, and is not due to differences of growth in salt-supplemented media (Fig. 4). The effect of acclimation without salt or in 5 g⋅L⁻¹ NaCl on the response to salt is independent of growing condition (effect of interaction between acclimation, growth condition and salt concentration: Χ² = 0.46, df = 1, P = 0.50).

PLASTICITY FOR GROWTH IN LOW SALT IN THE ANCESTORS HAS BEEN ASSIMILATED IN THE HIGH-SALT LINES

The constitutive response of the high-salt lines to salt concentrations is indistinguishable from the inducible response of the ancestors between 0 and 10 g⋅L⁻¹ (selection history:assay salt concentration interaction: Χ² = 0.00, df = 1, P = 0.99).

THE HIGH-SALT LINES HAVE EVOLVED AN ENHANCED INDUCIBLE RESPONSE TO HIGH SALT

The high-salt lines have a strong constitutive response to salt at concentrations up to 20 g⋅L⁻¹ (Fig. 3), but these lines do not appear to be capable of growing at concentrations of 30 g⋅L⁻¹ NaCl and higher. Nevertheless, these lines have been propagated in 36 g⋅L⁻¹ NaCl for many months without going extinct. Their ability to grow at salt concentrations in excess of 30 g⋅L⁻¹ is conferred by an inducible response.

In the lower range of salt concentrations between 0 and 20 g⋅L⁻¹, acclimation in medium containing salt significantly increases the overall rate of increase relative to lines that have been acclimated in medium without salt (Fig. 5; Table 2; effect of acclimation: F_2,20 = 5.3, P = 0.006). However, acclimation does not significantly affect the slope, meaning that growth decreases at the same rate with increases in salt concentrations whether the lines have been acclimated with or without salt (effect of acclimation:assay concentration interaction: F_1,220 = 1.7, P = 0.19). Note that although growth appears to be higher in no salt than in 10 g⋅L⁻¹ NaCl after acclimation in salt, this effect is not significant (Χ² = 2.70, df = 1, P = 0.10). Comparison of the inducible response of the salt-selected lines to low salt concentrations to the inducible response of the ancestors reveals that it has evolved. Between salt concentrations of 10 and 20 g⋅L⁻¹, growth decreases significantly more rapidly with increases in salt in the ancestors than in the salt-selected lines (selection history:assay salt concentration interaction effect: Χ² = 8.37, df = 1, P = 0.0038), although the intercepts are not statistically different (effect of selection history: Χ² = 3.14, df = 1, P = 0.076).

Figure 5.

The effect of acclimation regime on the growth of the salt-selected lines in different concentrations of salt. There are 10 salt-selected lines, each assayed twice. The datapoints for the lines acclimated in 10 g⋅L⁻¹ NaCl are plotted 1 g⋅L⁻¹ NaCl higher, and the lines acclimated in 36 g⋅L⁻¹ NaCl are plotted 2 g⋅L⁻¹ NaCl higher than assayed to make it easier to see differences between acclimation regimes. The trend line (dashed for 0 g⋅L⁻¹, solid for 10 g⋅L⁻¹, and dotted for 36 g⋅L⁻¹) was fitted using local polynomial regression (loess), with 95% confidence intervals in shade.

Table 2.

Effect of acclimation on the response of the salt-selected lines to a range of different salt concentrations between 0 and 20 g⋅L⁻¹

	Acclimation
	environment	Estimate
Parameter	(g⋅L−1 NaCl)	(±SE)
Slope	0	−0.0014 (0.018)
	10	−0.014 (0.018)
	36	−0.027 (0.013)
Intercept	0	1.9 (0.24)
	10	3.0 (0.24)
	36	2.8 (0.17)

In the higher range of salt concentrations between 20 and 40 g⋅L⁻¹, acclimation has a significant effect on the slope of the salt-selected lines, meaning that lines acclimated with salt maintain the same growth with increases in salt concentration, whereas lines acclimated without salt show a steep decline in growth with increases in salt concentration (Fig. 5; Table 3; ANCOVA effect of acclimation:assay concentration: F_1,215 = 48.4, P < 0.001). Comparison of the inducible response of the salt-selected lines to high salt concentrations to the inducible response of the ancestors reveals that it also has evolved. Between salt concentrations of 20 and 30 g⋅L⁻¹, growth is significantly greater overall in the salt-selected lines than in the ancestors (selection history effect: Χ² = 6.58, df = 1, P = 0.010), although the slope is not different (selection history:assay salt concentration interaction effect: Χ² = 2.99, df = 1, P = 0.084).

Table 3.

Effect of acclimation on the response of the salt-selected lines to a range of different salt concentrations between 20 and 40 g⋅L⁻¹

	Acclimation
	environment	Estimate
Parameter	(g⋅L−1 NaCl)	(±SE)
Slope	0	−0.18 (0.017)
	10	−0.089 (0.018)
	36	−0.10 (0.012)

CONSTITUTIVE AND INDUCIBLE RESPONSES ARE AFFECTED BY MEIOSIS

Without prior acclimation in salt medium, the F1 and F2 progeny grow at the same rate as the ancestors at all salt concentrations, and are unable to grow at concentrations of 28 g⋅L⁻¹ or higher (Fig. 6). This is in contrast to the salt-selected parents, which remain constitutively able to grow in 28 g⋅L⁻¹. Thus, the constitutive ability to grow at high salt concentrations is entirely lost after meiosis and recombination. After acclimation in medium containing 10 g⋅L⁻¹ NaCl, the F1 progeny grows as well as the salt-selected parents in concentrations up to 36 g⋅L⁻¹ NaCl, and grows better than the salt-selected parent in 48 g⋅L⁻¹ NaCl; the F2 progeny does worse than the salt-selected parents in concentrations up to 36 g⋅L⁻¹ NaCl, and does better than the salt-selected parents in 48 g⋅L⁻¹ NaCl (Table 4). Thus, the sexual progeny are able to grow at very high concentrations of up to 48 g⋅L⁻¹ NaCl, which their salt-selected parents are unable to tolerate.

Figure 6.

Growth of sexual progeny of the salt-selected lines after acclimation without salt or with 10 g⋅L⁻¹ salt. Assay concentrations for which there are fewer than four boxes indicate that none of the spores from the generation missing survived the first cycle in that concentration. The rate of increase during the second cycle could therefore not be calculated. Note that the order of the boxplots on the x-axis is the same as in the legend.

Table 4.

Difference in the mean rate of increase between the ancestor, salt-selected lines, F1 sexual progeny, and F2 sexual progeny when assayed in different salt concentrations

		Assay NaCl		Difference
		concentration		in mean rate
Acclimation	Comparison	(g⋅L−1)	Lower CI	of increase	Upper CI
0 g⋅L−1	Ancestor – F1	0	−2.66	−1.05	0.571
		28	−0.266	0.501	1.27
		36	−0.205	0.642	1.49
		44	−0.494	1.31	3.11
		48	−0.406	0.726	1.86
	Ancestor – F2	0	−2.37	−1.12	0.121
		28	−0.678	0.351	1.38
		36	−0.617	0.467	1.55
		44	−0.172	1.63	3.43
		48	−0.240	0.951	2.14
	Salt parent – F1	0	−0.602	−0.253	0.0963
		28	2.51	3.04	3.57
		36	−0.147	0.271	0.688
		44	−0.849	−0.293	0.263
		48	−1.01	−0.454	0.0988
	Salt parent – F2	0	−0.591	−0.330	−0.0697
		28	2.39	2.89	3.39
		36	−0.315	0.0946	0.504
		44	−0.438	0.0264	0.491
		48	−0.733	−0.228	0.276
	F1 – F2	0	−0.388	−0.0773	0.233
		28	−0.471	−0.151	0.170
		36	−0.533	−0.176	0.181
		44	−0.154	0.319	0.793
		48	−0.221	0.225	0.671
10 g⋅L−1	Salt parent – F1	0	0.131	0.738	1.35
		36	−1.53	1.69	4.92
		48	−2.77	−1.68	−0.589
	Salt parent – F2	0	0.610	1.10	1.59
		28	1.33	2.73	4.12
		36	1.38	3.24	5.10
		44	−8.76	−2.24	4.29
		48	−3.57	−2.26	−0.953
	F1 – F2	0	−0.181	0.364	0.909
		36	−0.845	1.55	3.94
		48	−2.27	−0.581	1.11

Lower and upper confidence intervals were calculated for the difference between means using the t-distribution for unequal sample sizes. The assay concentrations that are missing reflect the fact that none of the spores from one of the generations in the comparison survived the first cycle in that concentration.

CONSTITUTIVE AND INDUCIBLE RESPONSES BOTH CONTRIBUTE TO ADAPTATION

In the lower range of assay salt concentrations, the amount of variance in the rate of increase explained by the different lines (i.e., variance in the constitutive responses) is approximately 10 times greater than the amount of variance explained by the different acclimation regimes (i.e., inducible responses), with estimates of 0.40 and 0.047, respectively. The amount of variance explained by the interaction of line and acclimation regime is approximately three times greater than the amount explained by line alone (estimate of 1.3). In the higher range of assay salt concentrations, the amount of variance in the rate of increase explained by the different lines is approximately zero (estimate of −0.94). Acclimation explains a significant amount of the variance (estimate of 0.019), whereas the interaction of lines and acclimation regime explains about 300 times more of the variance than acclimation alone (estimate of 6.8).

Discussion

ADAPTATION TO MARINE CONDITIONS OF GROWTH

New ways of life evolve when organisms adapt to ecological conditions of growth that were not accessible to their ancestors. We have shown that an important ecological transition can occur within 500 generations. Some of the lines that we selected in gradually increasing concentrations of salt are now capable of growth in 36 g⋅L⁻¹ NaCl, far beyond what their ancestors could tolerate. In principle, these lines are now capable of growing in the sea.

About 98% of the experimental lines went extinct well before marine conditions were reached. Chronic exposure to a continuously deteriorating environment therefore requires far more than ancestral plasticity for growth in concentrations up to about 20 g⋅L⁻¹ NaCl for two growth cycles. The lines that have survived vary substantially in their ability to grow in seawater. Thus, most populations that experience a profound deterioration in the conditions of growth will simply become extinct. The experimental adaptation to marine conditions that occurred in this freshwater alga give an example of how survival to marine conditions can be achieved to different extents and in different ways.

In the yeasts Saccharomyces cerevisiae and S. paradoxus, for which the lethal concentration of salt is about 150 g⋅L⁻¹ NaCl, population size, the rate of increase in salt concentration, and connectivity with neighboring populations all affect the probability of surviving the imposed salt regime as well as the probability of surviving a transfer to the lethal concentration (Bell and Gonzalez 2009, 2011; Samani and Bell 2010; Gonzalez and Bell 2013). In the bacterium Serratia marcescens, tolerance to 90 g⋅L⁻¹ NaCl was improved after constant selection in either 80 or 100 g⋅L⁻¹ NaCl for 300 generations, but not after selection in a fluctuating environment, most likely because of weaker selection pressure (Ketola and Hiltunen 2014). Together, these results suggest that the rarity of transitions between freshwater and marine conditions may be a consequence of small population sizes, fast rates of increase in salt, fluctuating conditions, or low connectivity between natural populations.

GENETIC ASSIMILATION OF SALT TOLERANCE

In our experiment, growth of the evolved lines without acclimation to salt is equal to or greater than the growth of ancestral lines acclimated with salt, at salt concentrations of up to about 20 g⋅L⁻¹. Above this concentration, the evolved lines cannot grow without acclimation. Once acclimated, however, they grow much better than the acclimated ancestral lines in all concentrations above 10 g⋅L⁻¹. These results suggest that the ability to grow at very high salt concentrations evolved in two stages: genetic assimilation at lower concentrations, yielding a constitutive response to conditions lethal to the ancestor, and an enhanced inducible response at higher concentrations that permits growth up to about 40 g⋅L⁻¹ NaCl.

Changes in gene expression following long-term exposure to salt have been reported before in C. reinhardtii (Perrineau et al. 2014). Short-term acclimation to about 12 g⋅L⁻¹ NaCl causes a reduction in photosynthesis, upregulation of glycerophospholipid signaling, and upregulation of the transcription and translation machinery. Long-term culture in high-salt medium causes downregulation of genes involved in the stress response and in transcription and translation. Fatty acid metabolism is also more strongly downregulated following long-term than short-term acclimation, which suggests that long-term salt stress leads neither to lipid accumulation nor to the synthesis of starch. Selection can therefore alter gene expression for growth in salt.

Genetic assimilation can occur through genetic or epigenetic modifications. Unlike genetic modifications, which are changes in nucleotide sequence that are transmitted from parent to offspring in both asexual and sexual lineages, epigenetic modifications may be preserved in asexual lineages, either of free-living cells or of tissues in a developing body, but are generally removed during meiosis and are therefore not transmitted in sexual lineages (Jablonka and Raz 2009).

The constitutive tolerance to low salt concentrations was maintained in asexual cultures, but completely lost in the sexual progeny of the salt-selected lines. Indeed, the F1 and F2 progeny have the same phenotype as the ancestor in low salt concentrations after acclimation without salt. If genetic change was responsible for the assimilation of ancestral plasticity in low salt concentrations, we would have expected some of the progeny to have maintained some constitutive tolerance to salt, albeit possibly to lower extents. However, none of the 24 random sexual progeny that we assayed displayed a level of tolerance greater than ancestral. Therefore, we conclude that the assimilation of ancestral plasticity for growth in low salt concentrations is unlikely to be based on genetic changes. Rather, the assimilation of ancestral plasticity occurred through reversible changes in our asexually propagated selection lines. The loss of tolerance following meiosis is consistent with an epigenetic basis, although genomic studies will be required to explicitly test this hypothesis.

The inducible response to salt concentrations of up to 40 g⋅L⁻¹, on the other hand, was retained in sexual progeny, albeit more weakly expressed. This is consistent with genetic modification. This could be caused by loss-of-function mutations in a regulatory gene that hindered the binding of a repressor protein. This explanation, however, would require the existence of a cryptic inducible system in the ancestor whose function is obscure. It is more plausible to invoke gain-of-function mutations in an inducible structural gene. This gene is imagined to contribute to the inducible response at low salt concentrations expressed by the ancestor. During serial transfer at gradually increasing salt concentrations, alleles that spread through natural selection because they confer the ability to grow in ambient conditions may indirectly confer the ability to grow in more severe conditions. Adaptation to lethal conditions, resulting in evolutionary rescue, has been attributed to this kind of indirect response to selection in other experiments with algae and yeast (Bell and Gonzalez 2009, 2011; Samani and Bell 2010; Lachapelle and Bell 2012; Gonzalez and Bell 2013). The partial loss of fitness in F1 and F2 hybrid progeny is the expected result of recombination with ancestral alleles, and suggests that such gain-of-function mutations have occurred in more than one gene in our salt-selected lines.

THE CONTRIBUTION OF PLASTICITY AND GENETIC RECOMBINATION TO EVOLUTIONARY RESCUE

In a deteriorating environment, stress provides a continual stimulus capable of eliciting an inducible response. Where such a response exists, as it did in our selection lines, it enables the population to persist for longer and thereby prolongs the period during which genetic adaptation can occur through natural selection. The phenotypic plasticity of the ancestor for low stress is eventually lost after chronic exposure to increasing stress in our asexually propagated lines. The reversibility of this constitutive response to low salt in sexual progeny suggests the assimilation of ancestral plasticity could have arisen through the accumulation of neutral loss-of-function epigenetic modifications. The loss of plasticity would be accelerated if the inducible response were metabolically costly to maintain and/or activate. Although we have no way of measuring the cost of maintenance, our data show no evidence of a cost of activation: the growth of the ancestral lines in medium without salt is the same whether or not they have been previously acclimated with salt (Fig. 4). Drift could also have played a role in eliminating plasticity, given that the lines were bottlenecked following the first round of crosses. However, it is unlikely that plasticity would have been assimilated in all lines through chance alone.

In this instance of a deteriorating environment, then, the loss of plasticity at low levels of stress is accompanied by the evolution of enhanced plasticity at high levels of stress through genetic modifications. This is consistent with the evolution of enhanced plasticity in fluctuating environments reported by Schaum and Collins (2014). The breadth of conditions that the salt-selected lines can tolerate is much greater than the ancestors, consistent with the “sidestep niche model” whereby enhanced plasticity contributes in widening the niche after environmental change (Lande 2009; Gallet et al. 2014). However, we have no evidence that the niche has shifted or is now narrowing. To the contrary, the assimilation of ancestral plasticity in low salt concentrations seems to have contributed in maintaining the larger niche breadth.

The fact that sexual lines were better able to keep pace with the changing environment (Lachapelle and Bell 2012) indicates that surviving lines were better able to keep track of the moving fitness optimum because of the increased genetic variation generated by recombination. It is possible that the increase in resistance to salt reported here is mostly attributable to recombined variation from the end of this first selection experiment. However, our data do not allow us to make any inferences about the relative contribution of recombination, epigenetic, and genetic modification to the increase in resistance reported here.

Nonetheless, back-crosses of the high-salt lines to the ancestor, or crosses among these families, show that the F1 and F2 continue to grow at salt concentrations of 48 g⋅L⁻¹ at the same rate as at lower concentrations, whereas the high-salt lines themselves are unable to grow. This demonstrates the importance of recombination. The enhanced resistance of recombinants cannot be attributed to a more resistant protein because the high-salt lines themselves cannot grow at these very high salt concentrations. It is not due to the recombination of improved alleles at different loci because it is expressed in the F1 of crosses between the ancestor and the selection lines. It might be attributable to the release, through recombination, of an improved structural gene from linkage with a strongly deleterious mutation at some other locus. In this case, it would be necessary to assume further that this mutation is strongly deleterious only at very high salt concentrations because the F1 and F2 are inferior to the selection lines at salt concentrations of 40 g⋅L⁻¹ or less. Population sizes were very low during some stages of the experiment when the salt concentration was increasing. A neutral or mildly deleterious mutation could have therefore fixed by chance, if not by hitchhiking with a beneficial mutation. The uniform phenotype of random spores is also unexpected. Hence, we report that the range of conditions that can be tolerated is substantially extended in the sexual progeny of adapted parents, but we have not identified a simple genetic mechanism that would explain their superiority.

Conclusion

Experimental evolution has been extensively used to elucidate the mechanism of selection for particular attributes such as the ability to utilize a novel substrate or resist an antibiotic. The evolution of marine from freshwater lineages, of heterotrophs from autotrophs (Bell 2013a,b,c), and of multicellular from unicellular forms (Ratcliff et al. 2012, 2013) are examples that clearly indicate the possibility of studying certain aspects, at least, of major ecological transitions in the laboratory.

Here, we reported the adaptation of a freshwater alga to marine conditions within a few hundreds of generations in the laboratory. Continued selection pressure, sexually generated genetic variation, and phenotypic plasticity largely contributed to extending the limits of tolerance and facilitating the ecological transition. In short, the evolution of tolerance to salt involved two different mechanisms: reversible and irreversible changes. Tolerance to low salt concentrations of unacclimated selection lines was annulled by meiosis, suggesting reversible changes were responsible for the assimilation of ancestral plasticity and adaptation to the limit of tolerance. Tolerance to high salt concentrations of acclimated selection lines was maintained through meiosis, suggesting irreversible genetic changes were responsible for enhancing phenotypic plasticity in the selection lines and extended the range of tolerance to conditions lethal to the ancestor. Both mechanisms contributed to the transition from freshwater to fully marine conditions.

ACKNOWLEDGMENTS

We thank K. Tallon for maintaining the experimental lines, M. Waterfall for assistance with the flow cytometer, S. Tanaka for providing us with the marine strain CW80, S. Collins and I. Kronholm for discussion, and anonymous reviewers for comments. JL and GB are funded by the Natural Sciences and Engineering Research Council of Canada; JL is also funded through a studentship from the University of Edinburgh.

DATA ARCHIVING

The doi for our data is 10.5061/dryad.rs610.

Associate Editor: A. De Visser

Handling Editor: J. Conner

Author notes

These authors have contributed equally.