Physics of Living Systems

Fragmentation and aggregation of cyanobacterial colonies

Yuri Z Sinzato author has email address
Robert Uittenbogaard
Petra M Visser
Jef Huisman
Maziyar Jalaal author has email address

Van der Waals-Zeeman Institute, Institute of Physics, University of Amsterdam, Amsterdam, Netherlands
Hydro-Key Ltd, Haelen, Netherlands
Department of Freshwater and Marine Ecology, Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam, Netherlands

https://doi.org/10.7554/eLife.103503.2

Open access
Copyright information

Figures and data

Methodology used to observe effects of fluid flow on Microcystis colonies.
(A) Experimental setup consisting of a (cone-and-plate) controlled flow setup combined with inverted microscopy. The conical upper surface was rotated by a rheometer head, while the stationary glass slide below the sample allowed optical access for the microscope. (B) Examples of microscopy images. Colony size distributions were calculated after image processing of the captured frames. (C) Changes in size distributions (and other complementary measurements) over time were used to identify aggregation and fragmentation of cyanobacterial colonies. (D) The majority of the measurements were conducted using a laboratory culture of Microcystis strain V163. Colonies collected from Lake Gaasperplas (Netherlands), dominated by the morphospecies M. aeruginosa were also used.

Kinetics of fragmentation of Microcystis strain V163 colonies under cone-and-plate shear flow. The laboratory culture was filtered to select mainly large colonies, and the total biovolume fraction was adjusted to ϕ = 10⁻⁴. Suspensions were subjected to an intense dissipation rate () in panels A-E.
(A) The initial size distribution of colonies, expressed as biovolume fraction of the relative colony diameter (normalized by single cell diameter). The size distribution had a bimodal shape, with large colonies (in yellow) and small colonies (in green, composed mostly of single cells, dimers, and some trimers, as depicted by the inset). (B) Median diameter of small and large colonies as a function of time. Symbols indicate the experimental data, while the lines indicate the predictions from the population model given by Eq. (1) with an erosion (dashed) or equal fragments (dot-dash) hypothesis for large division-formed colonies. Small aggregates colonies follow the equal fragments hypothesis. Bars and shaded region indicate limits of 25th and 75th percentiles. (C) Most large colonies have been fragmented after 1 hour of shear flow, but the bimodal shape of the colony size distribution remained. (D) Biovolume fraction of small colonies (i.e., biovolume of small colonies over the total biovolume) as a function of time. A shift is observed from large to small colonies, captured well by the erosion model, but not by the equal fragments model. (E) The rate of change in biovolume distribution at t = 0h. Negative values indicate loss of colonies by fragmentation, while positive values indicate newly created fragments. The distribution suggests an erosion mechanism, as depicted by the cartoon inside the plot. (F) Fragmentation frequency as a function of the relative diameter of colonies for three values of dissipation rate. The inset shows details for moderate dissipation rates. Error bars indicate the standard deviation and the lines indicate the predictions by Eq. (5). Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5; erosion: S₂ = 31, q₂ = 4.1; equal fragments: S₂ = 33, q₂ = 6.3.

Kinetics of fragmentation of Microcystis strain V163 colonies under cone-and-plate shear flow. The laboratory culture was filtered to select mainly large colonies, and the total biovolume fraction was adjusted to ϕ = 10⁻⁴. Suspensions were subjected to an intense dissipation rate () in panels A-E.
(A) The initial size distribution of colonies, expressed as biovolume fraction of the relative colony diameter (normalized by single cell diameter). The size distribution had a bimodal shape, with large colonies (in yellow) and small colonies (in green, composed mostly of single cells, dimers, and some trimers, as depicted by the inset). (B) Median diameter of small and large colonies as a function of time. Symbols indicate the experimental data, while the lines indicate the predictions from the population model given by Eq. (1) with an erosion (dashed) or equal fragments (dot-dash) hypothesis for large division-formed colonies. Small aggregates colonies follow the equal fragments hypothesis. Bars and shaded region indicate limits of 25th and 75th percentiles. (C) Most large colonies have been fragmented after 1 hour of shear flow, but the bimodal shape of the colony size distribution remained. (D) Biovolume fraction of small colonies (i.e., biovolume of small colonies over the total biovolume) as a function of time. A shift is observed from large to small colonies, captured well by the erosion model, but not by the equal fragments model. (E) The rate of change in biovolume distribution at t = 0h. Negative values indicate loss of colonies by fragmentation, while positive values indicate newly created fragments. The distribution suggests an erosion mechanism, as depicted by the cartoon inside the plot. (F) Fragmentation frequency as a function of the relative diameter of colonies for three values of dissipation rate. The inset shows details for moderate dissipation rates. Error bars indicate the standard deviation and the lines indicate the predictions by Eq. (5). Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5; erosion: S₂ = 31, q₂ = 4.1; equal fragments: S₂ = 33, q₂ = 6.3.

Kinetics of aggregation for a single-cell suspension of Microcystis strain V163 at a moderate dissipation rate of and a total biovolume fraction of ϕ = 10⁻⁴.
(A) Initial size distribution of the suspension as a function of the relative diameter, composed mostly of single cells. (B) Median diameter of colonies formed by aggregation of single cells as a function of time. Symbols indicate the experimental data, while the lines indicate the predictions from the population model given by Eq. (1) with an equal fragment hypothesis for small aggregates colonies. Bars and shaded region indicate limits of 25th and 75th percentiles. (C) After 1 hour of shear flow, the size distribution has shifted towards slightly larger diameters. (D) Time behavior of a suspension of large division-formed colonies under the same moderate dissipation rate and total biovolume fraction. The bimodal size distribution is separated into large (yellow) and small (green) colonies. Dashed lines and shaded regions indicate predictions from the population model given by Eq. (1). Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Kinetics of aggregation for a single-cell suspension of Microcystis strain V163 at a moderate dissipation rate of and a total biovolume fraction of ϕ = 10⁻⁴.
(A) Initial size distribution of the suspension as a function of the relative diameter, composed mostly of single cells. (B) Median diameter of colonies formed by aggregation of single cells as a function of time. Symbols indicate the experimental data, while the lines indicate the predictions from the population model given by Eq. (1) with an equal fragment hypothesis for small aggregates colonies. Bars and shaded region indicate limits of 25th and 75th percentiles. (C) After 1 hour of shear flow, the size distribution has shifted towards slightly larger diameters. (D) Time behavior of a suspension of large division-formed colonies under the same moderate dissipation rate and total biovolume fraction. The bimodal size distribution is separated into large (yellow) and small (green) colonies. Dashed lines and shaded regions indicate predictions from the population model given by Eq. (1). Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Kinetics of the fragmentation of colonies in field samples of Microcystis spp. at an intense dissipation rate () and total biovolume fraction of ϕ = 10⁻⁴.
(A) Comparison of the fragmentation frequency as a function of colony size for the laboratory culture (Microcystis strain V163) and the field samples (Microcystis spp.). Error bars indicate the standard deviation. (B) Brightfield microscopy images of colonies in a Nigrosin-dyed medium (dark region) show evidence of a thick EPS layer (bright region) surrounding a field colony. (C) Initial size distribution of colonies in field samples as a function of the relative diameter. The size distribution had a bimodal shape, with small colonies (green) and large colonies (yellow). (D) The median diameter of the colonies in each subpopulation as a function of time. Bars indicate 25th and 75th percentiles. (E) After 1 hour of shear flow, the small colonies have aggregated slightly, while the large colonies keept their size distribution. (F) The biovolume fraction of small colonies remained nearly constant during the experiment.

Kinetics of the fragmentation of colonies in field samples of Microcystis spp. at an intense dissipation rate () and total biovolume fraction of ϕ = 10⁻⁴.
(A) Comparison of the fragmentation frequency as a function of colony size for the laboratory culture (Microcystis strain V163) and the field samples (Microcystis spp.). Error bars indicate the standard deviation. (B) Brightfield microscopy images of colonies in a Nigrosin-dyed medium (dark region) show evidence of a thick EPS layer (bright region) surrounding a field colony. (C) Initial size distribution of colonies in field samples as a function of the relative diameter. The size distribution had a bimodal shape, with small colonies (green) and large colonies (yellow). (D) The median diameter of the colonies in each subpopulation as a function of time. Bars indicate 25th and 75th percentiles. (E) After 1 hour of shear flow, the small colonies have aggregated slightly, while the large colonies keept their size distribution. (F) The biovolume fraction of small colonies remained nearly constant during the experiment.

Phase diagram indicating the dominant colony formation mechanism as a function of the dissipation rate and the cyanobacterial abundance (expressed by the total biovolume fraction ϕ).
(I) Colonies grow only by cell division at low dissipation rates and total biovolume fractions. (II) As the biovolume fraction increases, aggregation enhances colony growth. (III) For moderate dissipation rates, aggregated colonies are fragmented, and only cell division can increase colony size. (IV) Fragmentation of colonies dominates at intense dissipation rates, irrespective of whether these colonies were formed by aggregation or cell division. Bars on the right side indicate typical values of dissipation rate observed for natural wind-mixing, bubble plumes in artificially mixed lakes and laboratory-scale setups such as cone-and-plate systems and stirred tanks. The dashed arrowed line indicates the transition from an early bloom (left bullet - WHO alert level 1 [59]) to a dense surface scum (right bullet - typical scum biovolume fraction [61]) under typical wind mixing.

Phase diagram indicating the dominant colony formation mechanism as a function of the dissipation rate and the cyanobacterial abundance (expressed by the total biovolume fraction ϕ).
(I) Colonies grow only by cell division at low dissipation rates and total biovolume fractions. (II) As the biovolume fraction increases, aggregation enhances colony growth. (III) For moderate dissipation rates, aggregated colonies are fragmented, and only cell division can increase colony size. (IV) Fragmentation of colonies dominates at intense dissipation rates, irrespective of whether these colonies were formed by aggregation or cell division. Bars on the right side indicate typical values of dissipation rate observed for natural wind-mixing, bubble plumes in artificially mixed lakes and laboratory-scale setups such as cone-and-plate systems and stirred tanks. The dashed arrowed line indicates the transition from an early bloom (left bullet - WHO alert level 1 [59]) to a dense surface scum (right bullet - typical scum biovolume fraction [61]) under typical wind mixing.

(A-D) Micrographic images of phytoplankton field samples collected from the surface layer of Lake Gaasperplas. Microcystis spp. are dominant in the sample.

Kinetics of the fragmentation of Microcystis strain V163 colonies under a moderate dissipation rate () and various values of total biovolume fraction. The laboratory culture was filtered to select mainly large colonies, and total biovolume fraction was adjusted. Plots in the left column depict the median diameter of each size population as a function of time, where bars and shaded regions indicate limits of 25 th and 75 th percentiles. Plots in the right column depict the biovolume fraction of small colonies as a function of time. The total biovolume fraction is (A-B) ϕ = 10⁻⁴, (C-D) ϕ = 2⋅10⁻⁴, (E-D) ϕ = 5⋅10⁻⁴. Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Kinetics of the fragmentation of Microcystis strain V163 colonies under a moderate dissipation rate () and various values of total biovolume fraction. The laboratory culture was filtered to select mainly large colonies, and total biovolume fraction was adjusted. Plots in the left column depict the median diameter of each size population as a function of time, where bars and shaded regions indicate limits of 25 th and 75 th percentiles. Plots in the right column depict the biovolume fraction of small colonies as a function of time. The total biovolume fraction is (A-B) ϕ = 10⁻⁴, (C-D) ϕ = 2⋅10⁻⁴, (E-D) ϕ = 5⋅10⁻⁴. Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Kinetics of the fragmentation of strain V163 colonies under a moderate total biovolume fraction (ϕ = 10⁻⁴) and various values of dissipation rate. The laboratory culture was filtered to select mainly large colonies, and total biovolume fraction was adjusted. Plots in the left column depict the median diameter of each size population as a function of time, where bars and shaded regions indicate limits of 25 th and 75 th percentiles for the experimental data and model predictions, respectively. Plots in the right column depict the biovolume fraction of small colonies as a function of time. The dissipation rate is (A-B) , (C-D) , (E-D) . Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Kinetics of the fragmentation of strain V163 colonies under a moderate total biovolume fraction (ϕ = 10⁻⁴) and various values of dissipation rate. The laboratory culture was filtered to select mainly large colonies, and total biovolume fraction was adjusted. Plots in the left column depict the median diameter of each size population as a function of time, where bars and shaded regions indicate limits of 25 th and 75 th percentiles for the experimental data and model predictions, respectively. Plots in the right column depict the biovolume fraction of small colonies as a function of time. The dissipation rate is (A-B) , (C-D) , (E-D) . Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Influence of varying dissipation rate and total biovolume fraction on the median diameter of colonies formed by aggregation of single cells of Microcystis strain V163. The laboratory culture was filtered to select mostly single cells, and the total biovolume fraction was adjusted before the experiment. (A) ϕ = 10⁻⁴ and , (B) ϕ = 2⋅10⁻⁴ and , (C) ϕ = 5⋅10⁻⁴ and , (D) ϕ = 10⁻⁴ and . Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Influence of varying dissipation rate and total biovolume fraction on the median diameter of colonies formed by aggregation of single cells of Microcystis strain V163. The laboratory culture was filtered to select mostly single cells, and the total biovolume fraction was adjusted before the experiment. (A) ϕ = 10⁻⁴ and , (B) ϕ = 2⋅10⁻⁴ and , (C) ϕ = 5⋅10⁻⁴ and , (D) ϕ = 10⁻⁴ and . Best fit parameters: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5, S₂ = 31, q₂ = 4.1.

Rheology of concentrated colonies of Microcystis strain V163. (A) Shear stress as a function of the shear rate obtained for a steady shear test. Solid line indicates a Herschel-Bulkley fit, , where (SD from best fit) is the dynamical yield stress. Other fit parameters are k = 0.55 ± 0.13 and n = 0.85 ± 0.05 (B) Storage G′ and loss G″ moduli as a function of the deformation strain amplitude γ for a oscillatory shear with angular frequency of 1 rad/s.

Rheology of concentrated colonies of Microcystis strain V163. (A) Shear stress as a function of the shear rate obtained for a steady shear test. Solid line indicates a Herschel-Bulkley fit, , where (SD from best fit) is the dynamical yield stress. Other fit parameters are k = 0.55 ± 0.13 and n = 0.85 ± 0.05 (B) Storage G′ and loss G″ moduli as a function of the deformation strain amplitude γ for a oscillatory shear with angular frequency of 1 rad/s.

Comparison of hypothesis for the fragment distribution of category C₁ colonies. Bars show the experimental results for normalized biovolume distribution of filtered single cells of Microcystis strain V163 after 1 hour under a moderate dissipation rate () and a total biovolume fraction of ϕ = 10⁻⁴. Lines depict the model predictions using an equal fragment hypothesis (dashed) and an erosion hypothesis (solid).

Comparison of hypothesis for the fragment distribution of category C₁ colonies. Bars show the experimental results for normalized biovolume distribution of filtered single cells of Microcystis strain V163 after 1 hour under a moderate dissipation rate () and a total biovolume fraction of ϕ = 10⁻⁴. Lines depict the model predictions using an equal fragment hypothesis (dashed) and an erosion hypothesis (solid).

Cone-and-plate setup used to generate a turbulent shear in a suspension of Microcystis colonies. (A) Schematics of the components with the main dimensions indicated in millimeters. (B) Average energy dissipation rate as a function of the angular velocity of the conical probe. Bullets indicate the measured data, and the solid lines indicate the best fit of the laminar regime (∼ω²) and the inertial regime (∼ω^5/2).

Calibration steps for computing the correction function for the biovolume distribution. (A-B) Uncorrected concentration distribution of colonies measured in the cone-and-plate shear setup. Total number of counted colonies, N = 10776. (C-D) Reference distribution measured in the inverted microscope after sedimentation of the colonies. Total number of counted colonies in panels C and D: N = 3066 and 1455, respectively. (E-F) Ratio between the reference and uncorrected concentration distributions Dashed lines indicate the fitted calibration parameters. (G-H) Normalized biovolume distribution measured in the cone-and-plate shear setup after the correction function. Error bars indicate the root mean squared of counting uncertainty and the size measurement uncertainty. Panels A,C,E,G display the distributions for small colonies (l < 5), while panels B,D,F,H display the distributions for large colonies (l > 5).

Comparison between the erosion and equal fragments hypothesis for the fragmentation distribution of a suspension of large division-formed colonies. Small aggregates colonies follow the equal fragments hypothesis. (A) Median diameter of small and large colonies as a function of time. Symbols indicate the experimental data, while the lines indicate the predictions from the population model given by Eq. (1). Both the erosion (dashed) and the equal fragments (dot-dash) hypothesis recover well the time behavior of the median diameter. Bars and shaded region indicate limits of 25th and 75th percentiles. (B) Biovolume fraction of small colonies (i.e., biovolume of small colonies over the total biovolume) as a function of time. Only the erosion model recovers the transfer from large to small colonies. (C) Initial size distribution of colonies expressed as biovolume fraction as a function of the relative colony diameter (normalized by single cell diameter). Bars indicate the experimental data (large colonies in yellow and small colonies in green), while the lines indicate the model predictions for large colonies. Both experiments and model have a bin width of Δl = 0.5 The model is initialized with the experimental distribution and advanced in time using two fragment distribution functions for the large colonies: (D) erosion, in the left column and (E) equal fragments, in the right column. Time progresses from top to bottom. Insets in panels D and E display the median diameter of each category, following the symbol notation of panel A . The total biovolume fraction is ϕ = 10⁻⁴ and the energy dissipation rate is . Best fit parameters for both models: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5; only the erosion model: S₂ = 31, q₂ = 4.1; only the equal fragments model: S₂ = 33, q₂ = 6.3.

Comparison between the erosion and equal fragments hypothesis for the fragmentation distribution of a suspension of large division-formed colonies. Small aggregates colonies follow the equal fragments hypothesis. (A) Median diameter of small and large colonies as a function of time. Symbols indicate the experimental data, while the lines indicate the predictions from the population model given by Eq. (1). Both the erosion (dashed) and the equal fragments (dot-dash) hypothesis recover well the time behavior of the median diameter. Bars and shaded region indicate limits of 25th and 75th percentiles. (B) Biovolume fraction of small colonies (i.e., biovolume of small colonies over the total biovolume) as a function of time. Only the erosion model recovers the transfer from large to small colonies. (C) Initial size distribution of colonies expressed as biovolume fraction as a function of the relative colony diameter (normalized by single cell diameter). Bars indicate the experimental data (large colonies in yellow and small colonies in green), while the lines indicate the model predictions for large colonies. Both experiments and model have a bin width of Δl = 0.5 The model is initialized with the experimental distribution and advanced in time using two fragment distribution functions for the large colonies: (D) erosion, in the left column and (E) equal fragments, in the right column. Time progresses from top to bottom. Insets in panels D and E display the median diameter of each category, following the symbol notation of panel A . The total biovolume fraction is ϕ = 10⁻⁴ and the energy dissipation rate is . Best fit parameters for both models: α₁ = 0.023, S₁ = 0.034, q₁ = 4.5; only the erosion model: S₂ = 31, q₂ = 4.1; only the equal fragments model: S₂ = 33, q₂ = 6.3.

Sign up for email alerts