Environmental tolerance curves, representing absolute fitness against the environment, are an empirical assessment of the fundamental niche, and emerge from the phenotypic plasticity of underlying phenotypic traits. Dynamic plastic responses of these traits can lead to acclimation effects, whereby recent past environments impact current fitness. Theory predicts that higher levels of phenotypic plasticity should evolve in environments that fluctuate more predictably, but there have been few experimental tests of these predictions. Specifically, will still lack experimental evidence for evolution of acclimation effects in response to environmental predictability. Here, we exposed 25 genetically diverse populations of the halotolerant microalgae Dunaliella salina to different constant salinities, or to randomly fluctuating salinities, for over 200 generations. The fluctuating treatments differed in their autocorrelation, which determines the similarity of subsequent values, and thus environmental predictability. We then measured acclimated tolerance surfaces, mapping population growth rate against past (acclimation) and current (assay) environments. We found that experimental mean and variance in salinity caused the evolution of niche position (optimal salinity) and breadth, with respect to not only current but also past (acclimation) salinity. We also detected weak but significant evidence for evolutionary changes in response to environmental predictability, with higher predictability leading notably to an upwards shift in optimal salinities and stronger acclimation effect of past environment on current fitness. We further showed that these responses are related to the evolution of plasticity for intracellular glycerol, the major osmoregulatory mechanism in this species. However the direction of plasticity evolution did not match simple theoretical predictions. Our results underline the need for a more explicit consideration of the dynamics of environmental tolerance and its underlying plastic traits to reach a better understanding of ecology and evolution in fluctuating environments.

Experimental Evolution

25 populations of a mix of two strain of the green microalgae Dunaliella salina were exposed to 25 independent, constant (6 populations) or randomly fluctuating (19 populations) time-series of salinities, with same targeted mean (2.4M), variance (1) and four autocorrelation treatments (-0.5, 0, 0.5, 0.9). Populations were transferred twice a week for 8 to 12 months. 

Acclimated tolerance surface assay and analysis

For each population, we measured an acclimated tolerance surface, with population growth rate parametrized as a bivariate function of present (assay) and past (acclimation) salinity. Cultures form the evolution experiment were placed for 10 days in 2.4M NaCl, then acclimated during 7 days in 9 salinities (0.1, 0.5, 1.1, 1.8, 2.4, 3.2, 3.7, 4.3, 4.7) in culture flask (15mL), then transferred to the same 9 salinities in culture plates (0.8 mL). This led to 75 cross salinity-treatments, eliminating 6 transfers from very low to very high / very high to very low salinities, which could no be achieved due to the dilution level used to reach initial population size of 5000 cells/mL. Growth was inferred from changes in population sizes measured by flow cytometry. Flow cytometry allowed us to detect cells that were alive versus recently dead or dying. Initial population sizes (after dilution but before any growth and mortality) were measured just after dilution in 0.36M (salinity known to not generate mortality) and final population sizes were measured after 3 days.

We partitioned population growth rate into an apparent growth rate and an instantaneous apparent mortality ratio, and used the R package TMB (Template Model Builder) to estimate apparent growth rates and apparent death ratio in a negative binomial and a beta binomial model, respectively. Apparent growth was modelled by a bivariate tolerance surface parametrized by the assay and acclimation optima, the assay and acclimation niche width, and the acclimation strength (quantifying the impact of previous salinity on current salinity optimum). Apparent death, occurring only in transfers from low to high salinities, was a linear function of acclimation and assay salinities, and their interaction. We analyzed the evolution of the acclimated tolerance surface by quantifying the effect of the mean, variance, autocorrelation, and predictability of environments during evolution on each parameter of the apparent growth (ex. assay salinity optimum) and apparent death functions.

Dynamics of glycerol content assay

We analyzed the dynamics of intracellular and total glycerol content after a salinity transfer in 8 population that evolved in fluctuating salinities. After 10 days at 2.4M, populations were acclimated for 7 days in 0.5 or 3.5M NaCl. They were then transferred to 0.5, 2 and 3.5M NaCl and glycerol content was assessed 1h, 8h and 3 days after this transfer. 800 µl Free Glycerol Reagent (Sigma Aldrich) was mixed with 200-µl culture (total glycerol concentration) or 200µl of culture where cells had been removed by centrifugation (5min 2000rpm on 0.2µm filtration plates), to measure the extracellular glycerol concentration. We measured optical density at 540 nm after 5 min incubation at 37°C. Dunaliella densities were then estimated by flow cytometry to yield measurements per unit cell, and we made two independent replicates for each measure.

Glycerol concentration was interpolated from a linear standard curve (Appendix), and intracellular glycerol concentrations (mean and standard errors) were computed by subtracting extracellular to total amount of glycerol. We assessed the evolution of plasticity by measuring the effect of environmental autocorrelation on the environmental variance in glycerol content. 

More details on the protocols and analyses can be found in the paper.

See README file.