Rising CO2 levels will intensify phytoplankton blooms in eutrophic and hypertrophic lakes

Abstract

Harmful algal blooms threaten the water quality of many eutrophic and hypertrophic lakes and cause severe ecological and economic damage worldwide. Dense blooms often deplete the dissolved CO2 concentration and raise pH. Yet, quantitative prediction of the feedbacks between phytoplankton growth, CO2 drawdown and the inorganic carbon chemistry of aquatic ecosystems has received surprisingly little attention. Here, we develop a mathematical model to predict dynamic changes in dissolved inorganic carbon (DIC), pH and alkalinity during phytoplankton bloom development. We tested the model in chemostat experiments with the freshwater cyanobacterium Microcystis aeruginosa at different CO2 levels. The experiments showed that dense blooms sequestered large amounts of atmospheric CO2, not only by their own biomass production but also by inducing a high pH and alkalinity that enhanced the capacity for DIC storage in the system. We used the model to explore how phytoplankton blooms of eutrophic waters will respond to rising CO2 levels. The model predicts that (1) dense phytoplankton blooms in low- and moderately alkaline waters can deplete the dissolved CO2 concentration to limiting levels and raise the pH over a relatively wide range of atmospheric CO2 conditions, (2) rising atmospheric CO2 levels will enhance phytoplankton blooms in low- and moderately alkaline waters with high nutrient loads, and (3) above some threshold, rising atmospheric CO2 will alleviate phytoplankton blooms from carbon limitation, resulting in less intense CO2 depletion and a lesser increase in pH. Sensitivity analysis indicated that the model predictions were qualitatively robust. Quantitatively, the predictions were sensitive to variation in lake depth, DIC input and CO2 gas transfer across the air-water interface, but relatively robust to variation in the carbon uptake mechanisms of phytoplankton. In total, these findings warn that rising CO2 levels may result in a marked intensification of phytoplankton blooms in eutrophic and hypertrophic waters.

Figure 1. Seasonal dynamics of phytoplankton blooms in Lake Volkerak.

(A) Changes in phytoplankton population density (strongly dominated by the cyanobacterium Microcystis) and measured dissolved CO₂ concentration ([CO₂]) during two consecutive years. The dashed line is the expected dissolved CO₂ concentration ([CO₂*]) when assuming equilibrium with atmospheric pCO₂. Dark shading indicates that the lake is supersaturated with CO₂, while light shading indicates undersaturation. (B) Changes in pH, bicarbonate and total DIC concentration. Sampling details are described in Text S1.

Figure 2. Changes in inorganic carbon chemistry during phytoplankton growth in two chemostat experiments.

Left panels: Chemostat experiment with low pCO₂ of 200 ppm in the gas flow and 500 µmol L⁻¹ bicarbonate in the mineral medium. Right panels: Chemostat experiment with high pCO₂ of 1,200 ppm in the gas flow and 2,000 µmol L⁻¹ bicarbonate in the mineral medium. Both chemostats were inoculated with Microcystis CYA140. (A, B) Population density (expressed as biovolume) and light intensity penetrating through the chemostat (I_OUT), (C, D) dissolved CO₂, bicarbonate and carbonate concentrations, (E, F) total DIC concentration and pH, and (G, H) alkalinity (ALK) and concentrations of dissolved inorganic nitrogen (DIN) and phosphorus (DIP). Symbols represent measurements, lines show the model fits. The model and its parameter values are detailed in Text S2.

Figure 3. Trajectories of dissolved CO₂ and population density.

Trajectories predicted by the model for chemostats with (A) low pCO₂ of 200 ppm in the gas flow and 500 µmol L⁻¹ bicarbonate in the mineral medium, and (B) high pCO₂ of 1,200 ppm in the gas flow and 2,000 µmol L⁻¹ bicarbonate in the mineral medium. The trajectories start from a series of different initial conditions, and all converge to the same equilibrium point. Arrows indicate the direction of the trajectories. The model assumes species parameters specific for Microcystis CYA140, and is detailed in Text S2.

Figure 4. Steady-state patterns of phytoplankton population density and inorganic carbon chemistry in chemostat experiments.

Steady-state results are shown for 6 chemostats with Microcystis HUB5-2-4 exposed to different pCO₂ levels in the gas flow and two different bicarbonate concentrations in the mineral medium (0.5 or 2.0 mmol L⁻¹). (A) Phytoplankton population density (expressed as biovolume), (B) light intensity penetrating through the chemostat (I_OUT), (C) dissolved CO₂ concentration, (D) bicarbonate concentration, (E) pH, (F) alkalinity, (G) DIC concentration, and (H) carbon sequestration rate. Symbols show the mean (± s.d.) of 5 measurements in each steady-state chemostat, lines show the model fits. For comparison, dashed lines show steady-state patterns predicted for chemostats without phytoplankton. Shading indicates the level of carbon limitation (L_C) predicted by the model. The model and its parameter values are detailed in Text S2.

Figure 5. Steady-state patterns predicted for phytoplankton blooms in low-alkaline lakes.

Steady-state predictions of the model evaluated across a wide range of atmospheric pCO₂ levels. (A) Phytoplankton population density (expressed as biovolume), (B) light intensity reaching the lake sediment (I_OUT), (C) dissolved CO₂ concentration, (D) bicarbonate concentration, (E) pH, (F) alkalinity, (G) DIC concentration, and (H) carbon sequestration rate. Shading indicates the level of carbon limitation (L_C). For comparison, dashed lines show steady-state patterns predicted for low-alkaline waters without phytoplankton. The model parameters are representative for eutrophic low-alkaline lakes (ALK_IN = 0.5 mEq L⁻¹) dominated by the cyanobacterium Microcystis HUB5-2-4. The model and its parameter values are detailed in Text S2 and Text S3.

Figure 6. Contour plots of phytoplankton blooms predicted for different pCO₂ levels and alkalinities.

Model predictions of (A) the level of carbon limitation, and (B) phytoplankton population density (expressed as biovolume, in mm³ L⁻¹). The vertical solid line represents the present-day atmospheric CO₂ level of ∼400 ppm, while the vertical dashed line shows the atmospheric CO₂ level of 750 ppm predicted for the year 2150 by the RCP6 scenario of the Fifth Assessment Report of the IPCC. The model predictions are based on steady-state solutions across a grid of 40×50 = 2,000 simulations, using the model and parameter values detailed in Text S2 and Text S3.

Figure 7. Sensitivity of the model predictions to variation in phytoplankton traits.

Contour plots of the level of carbon limitation (left panels) and steady-state phytoplankton population density (right panels, expressed as biovolume, in mm³ L⁻¹) predicted for different atmospheric pCO₂ levels and phytoplankton traits. The phytoplankton traits are (A, B) the half-saturation constant for CO₂ uptake (H_CO2), (C, D) the half-saturation constant for bicarbonate uptake (H_HCO3), (E, F) the maximum CO₂ uptake rate (u_{MAX, CO2}), and (G, H) the cellular N:C ratio (c_N). The model considers a low-alkaline lake (ALK_IN = 0.5 mEq L⁻¹). Vertical lines represent atmospheric CO₂ levels of 400 ppm (present-day) and 750 ppm (predicted for the year 2150 by the RCP6 scenario of the IPCC). Horizontal dotted lines represent our default parameter values. The contour plots are based on steady-state solutions across a grid of 40×50 = 2,000 simulations.

Figure 8. Sensitivity of the model predictions to variation in lake properties.

Contour plots of the level of carbon limitation (left panels) and steady-state phytoplankton population density (right panels, expressed as biovolume, in mm³ L⁻¹) predicted for different atmospheric pCO₂ levels and lake properties. The lake properties are (A, B) lake depth (z_MAX), (C, D) CO₂ gas transfer velocity (v), (E, F) DIC concentration of the influx ([DIC]_IN), and (G, H) salinity (Sal). The model considers a low-alkaline lake (ALK_IN = 0.5 mEq L⁻¹). Vertical lines represent atmospheric CO₂ levels of 400 ppm (present-day) and 750 ppm (predicted for the year 2150 by the RCP6 scenario of the IPCC). Horizontal dotted lines represent our default parameter values. In (E, F), the dotted line indicates equilibrium with the atmospheric CO₂ pressure. The contour plots are based on steady-state solutions across a grid of 40×50 = 2,000 simulations.