betaFeedback

WorksPhysics and Astronomy

Two-dimensional liquid-like order in a confined population of swimming Eudorina colonies

Abstract

Swimming microorganisms accumulate at boundaries, but the spatial statistics of the resulting layers are rarely measured. We report a population of colonial green algae, of order 71 μμm across, that accumulates against the rigid upper wall of a closed 1-2 mm chamber and holds a two-dimensional non-contacting arrangement for about 105 min. The pair correlation function has a first shell at r/Dˉ=1.73r/\bar{D} = 1.73, g=4.8g = 4.8; a hard-disc null at the same area fraction reaches only g=2.1g = 2.1, so the ordering is not excluded volume alone. Nearest-neighbour separation is 1.80±0.131.80 \pm 0.13 diameters. The bond-orientational order parameter is ⟨∣ψ6∣⟩=0.456\langle|ψ_6|\rangle = 0.456 against 0.385 for the null, only 36% of interior colonies are six-coordinated, and the orientational correlation g6(r)g_6(r) decays over 1.5 diameters with η6=1.31η_6 = 1.31 against the hexatic bound of 1/4: the arrangement is a two-dimensional liquid, not a crystal or a hexatic. Colonies execute a persistent random walk with reorientation time 1.4-1.7 s at Reynolds number ∼9×10−4\sim 9\times 10^{-4}. Breakup begins at 127 min. Thermal convection is excluded on Rayleigh-number grounds. Caveats: the length calibration is nominal rather than metrological, and this is a single population observed once. All data and code are deposited.

Introduction

Motile microorganisms accumulate at solid boundaries, a consequence of swimming, steric interaction and hydrodynamic coupling to the wall that is common to bacteria, spermatozoa and algae. In dense suspensions this accumulation drives bioconvection, described first for Euglena by Wager [1] and named by Platt [2], with continuum theory following [3, 4, 5, 6]. In dilute populations near a wall the relevant physics is instead pairwise: Drescher et al. [7] showed that two Volvox colonies swimming against a boundary attract one another into stable orbiting states, analysed subsequently by boundary-element simulation [8].

Between these limits lies a regime that is less well characterised: a population dense enough to interact but too dilute to convect, held against a boundary long enough for its spatial statistics to be measured. Balasubramanian reported that Volvox barberi forms two-dimensional flocks packing close to optimally, essentially in contact [9]. We report here the opposite case, in a smaller volvocine colony: a stable arrangement in which colonies are held apart, with a preferred separation near two diameters and a pair correlation function characteristic of a two-dimensional liquid.

The measurement is straightforward, and we make no claim that the apparatus is sophisticated. Its interest lies in the specific numbers, in the null model against which the ordering is tested, and in a measured lifetime for the structure.

Material and methods

Sample and chamber

The sample was collected from Lake Oroville, a reservoir in northern California, on 29 March 2018. The collection is documented in Ref. [13]. It was not a monoculture. It stood beside a window for two weeks before imaging, receiving natural daylight and no added nutrients, which enriches for motile phototactic flagellates. Nothing here should be read as a limnological statement about the reservoir.

For imaging the sample was held in a closed clear plastic coin capsule (Air-Tite “Direct Fit” H40; internal diameter 40.6 mm, internal depth 3.17 mm) filled to a fluid depth of 1–2 mm. Both bounding surfaces are therefore rigid and no-slip, matching the geometry assumed in Ref. [7]. The nearest sidewall lies about 20 mm from the imaged field, some 280 colony diameters, so lateral confinement is negligible.

Exclusion of thermal convection

In a layer of depth dd the Rayleigh number is

Ra=gαΔTd3νκ,(1)\mathrm{Ra} = \frac{g\alpha\Delta T d^3}{\nu\kappa}, \tag*{(1)}

with gα/νκ≈1.4×1010/(m3 K)g\alpha/\nu\kappa\approx1.4 \times10^{10}/(\mathrm{m}^3\,\mathrm{K}) for water near room temperature. Taking Rac≈1100\mathrm{Ra}_{\mathrm{c}} \approx1100, onset requires 79 K across the layer at d=1d = 1 mm and 10 K at 2 mm. Neither is attainable, and the thermal diffusion time d2/κ≈28d^2/\kappa\approx28 s at 2 mm precludes sustaining such a gradient. The structure reported below is not a convective one; it is also static rather than circulating, which bioconvective plumes are not.

Imaging and its limitations

Imaging used a handheld USB digital microscope (AmScope UTP class): a 2-megapixel CMOS sensor, 1600×1200 effective pixels, continuously variable magnification, eight built-in LEDs. Two sequences were recorded: a continuous 117 s run at native rate for dynamics, and a 20× time lapse spanning 153 min of elapsed time for structure and lifetime; the latter sequence may be viewed directly in Ref. [14].

An instrument of this kind costs a small fraction of a research or teaching compound microscope. That was a deliberate choice rather than a constraint, for the reasons set out in § 4; the price is paid in metrology.

Use of a very low cost imaging system carries significant limitations. The quoted magnification is not an optical quantity: it is defined against an unspecified display and set continuously by working distance, and with no eyepiece there is no field number from which to compute a field of view. We adopt 1.55 µm per pixel with an estimated 20 % uncertainty, but this was not established with a stage micrometer and should be regarded as nominal. Second, the recorded video is 1280×720 while the sensor is 1600×1200, so the frames were cropped or rescaled by an unrecoverable amount; this is an additional error path not covered by that 20 %. Third, no numerical aperture is published, so no depth of field can be computed and the planarity of the layer cannot be established optically. The empirical argument for near-planarity is that objects elsewhere in the same frames appear as clearly defocused rings while the analysed cluster is sharp.

Every dimensionless quantity reported below—g(r)g(r), the nearest-neighbour ratio, the mean-squared displacement exponents, the reorientation time—is independent of the calibration. Quantities in microns are not.

The native-rate clip proved to be a 15 Hz capture padded into a 30 Hz container, with every second frame duplicated; frames were decimated by two before tracking. Analysed without decimation the velocity autocorrelation acquires a spurious period-two oscillation.

Identification

The organism is Eudorina: a coenobium of 16, 32 or 64 biflagellate cells embedded at the periphery of a mucilage matrix, typically 50–200 µm in overall diameter with individual cells 10–24 µm across. It swims by coordinated flagellar beating, is bottom-heavy and therefore gravitactic, and is strongly phototactic.

The identification was made by direct examination under a compound laboratory microscope at magnification sufficient to resolve the individual cells of the coenobium. This is not an observation the low-cost imaging system used for the quantitative work could have supplied: at its resolution and depth of field (§ 2.3) individual cells within a colony are not separable. The morphological measurements that can be made from the video are reported in § 3.6; they are consistent with Eudorina but would not have established it.

Detection, tracking and statistics

Colonies were segmented on a greenness index Ig=G−(R+B)/2I_g = G - (R + B)/2, Gaussian blurred, thresholded at a fixed level, and labelled by connected components; equivalent diameters follow from component area. Trajectories were built by greedy nearest-neighbour linking with a maximum inter-frame displacement of 12 pixels.

The pair correlation function was computed over all pairs, with separations normalised by the mean colony diameter in each frame and the histogram normalised against the two-dimensional ideal-gas expectation, the tail over 3.5<r/Dˉ<5.53.5 < r/\bar{D} < 5.5 used to set the baseline.

For the null model, the two-dimensional area fraction inside the convex hull of each cluster was measured on the native-rate sequence, giving ϕnr=0.17\phi_{\mathrm{nr}} = 0.17. Hard discs were then placed by random sequential adsorption in a square box of the same area, with the same colony-size distribution, frame by frame, and g(r)g(r) computed identically. The corresponding figure for the time-lapse sequence, over which the stability statistics of § 3.1 are computed, is ϕtl=0.104\phi_{\mathrm{tl}} = 0.104; the two differ because the sequences resolve different numbers of small colonies, and each analysis uses the value measured on its own frames.

Bond-orientational order

To distinguish a two-dimensional liquid from a crystal we computed the local bond-orientational order parameter

ψm(i)=1Ni∑j∈Nieimθij,(2)\psi_m(i) = \frac{1}{N_i} \sum_{j \in\mathcal{N}_i} e^{im\theta_{ij}}, \tag*{(2)}

where Ni\mathcal{N}_i are the Delaunay neighbours of colony ii, NiN_i their number, and θij\theta_{ij} the angle of the bond from ii to jj. Particles on the convex hull were excluded, since their neighbour shells are truncated. The same quantity was computed for the hard-disc null configurations. We also recorded Delaunay coordination numbers, and computed ψ4\psi_4 as a square-symmetry control.

Following the Kosterlitz–Thouless–Halperin–Nelson–Young (KTHNY) description of two-dimensional order, we further computed the orientational correlation function

g6(r)=⟨ψ6∗(0)ψ6(r)⟩⟨∣ψ6∣2⟩,(3)g_6(r) = \frac{\langle\psi_6^*(0)\psi_6(r)\rangle}{\langle|\psi_6|^2\rangle}, \tag*{(3)}

binned in units of mean colony diameter and averaged over frames, and fitted it to both an exponential Aexp⁡(−r/ξ)A\exp(-r/\xi) and a power law Ar−η6Ar^{-\eta_6}. The same pair of measures—a translational correlation length together with the decay of long-range orientational order—has been used to characterise order–disorder transitions in other driven two-dimensional patterns, for example the capillary ripple lattices of Ref. [12].

Robustness

The entire pipeline was recomputed with a deliberately different segmentation (a hue-window signal with an Otsu-selected threshold in place of the fixed threshold on G−(R+B)/2G-(R+B)/2), different frame subsets and three minimum-area cuts. Results of that comparison are reported in § 3.7, and the script is included in the deposit.

Results

A stable, non-contacting two-dimensional arrangement

Over the first 105 min the population holds a loose but clearly ordered arrangement (Figure 1). Median nearest-neighbour separation, normalised by mean colony diameter, is 1.80±0.131.80 \pm0.13 across 93 samples of the record, with a median of 30 colonies in the field and area fraction 0.104. Contact corresponds to 1.0. Because this is a projection, foreshortening can only reduce apparent separations, so 1.80 is a lower bound on the true three-dimensional spacing.

Stills from the time-lapse sequence

Figure 1. Stills from the time-lapse sequence (Ref. [14]) at 21 min intervals of elapsed time, same region of the field. Colonies exchange neighbours and the cluster drifts, but characteristic spacing and local coordination persist.

The ordering exceeds excluded volume

The pair correlation function (Figure 2) shows a hard exclusion region below one diameter, a first coordination shell peaking at r/Dˉ=1.73r/\bar{D}=1.73 with g=4.8g=4.8, a minimum near 2.5 and a weak second shell. The hard-disc null at the same area fraction and size distribution reaches only g=2.1g=2.1, at the closer separation r/Dˉ=1.57r/\bar{D}=1.57. The observed shell is thus more than twice as high as excluded volume alone produces, and sits further out.

Pair correlation function

Figure 2. Pair correlation function (blue) against a hard-disc random-sequential-adsorption null at the same area fraction ϕnr=0.17\phi_{\rm nr}=0.17 and size distribution (red). The peak position is robust to segmentation; its height is sensitive to the tail-normalisation window and should not be treated as a measured quantity.

The order is liquid-like, not crystalline

A strong first shell in g(r)g(r) establishes short-range order but does not by itself distinguish a liquid from a crystal; what separates them is whether orientational and translational order persist to large separation. Three measurements say the arrangement is liquid-like (Figure 3).

Delaunay triangulation, frame-averaged orientational order, coordination numbers, and orientational correlation function

Figure 3. (a) Delaunay triangulation of a representative frame, interior colonies coloured by ∣ψ6∣|\psi_6|. (b) Distribution of frame-averaged ⟨∣ψ6∣⟩\langle|\psi_6|\rangle, observed against the hard-disc null, with the conventional hexatic and crystalline ranges shaded. (c) Delaunay coordination numbers of interior colonies; the mode is 5, not 6. (d) Orientational correlation function g6(r)g_6(r) with exponential and power-law fits; the panel at right summarises the KTHNY diagnostic.

First, g(r)g(r) decays to unity by r/Dˉ≈3.8r/\bar{D}\approx3.8. The second shell is present but broad, spanning r/Dˉ≈2.9r/\bar{D}\approx2.9–3.4, rather than resolving into the separate 3\sqrt{3} and 2 peaks a hexagonal lattice produces. We note in passing that the first shell falls at r/Dˉ=1.73r/\bar{D}=1.73 and 3=1.732\sqrt{3}=1.732; this is a coincidence of expressing separations in colony diameters and is not evidence of a hexagonal lattice constant.

Second, the local bond-orientational order parameter of equation (2) averages ⟨∣ψ6∣⟩=0.456±0.057\langle|\psi_6|\rangle=0.456\pm0.057 over 38 frames and 574 interior colonies. For reference, two-dimensional crystals give 0.7–0.95, a hexatic phase 0.5–0.7, and a liquid 0.3–0.45. The measured value sits at the top of the liquid range and short of hexatic. The hard-disc null gives 0.385±0.0540.385\pm0.054, so the orientational order does exceed random packing—consistent with the g(r)g(r) result—but by only about 0.07. The square-symmetry control gives ⟨∣ψ4∣⟩=0.317\langle|\psi_4|\rangle=0.317, not far below ψ6\psi_6, which is what a disordered arrangement rather than a six-fold symmetric one produces.

Third, Delaunay coordination numbers of interior colonies have mean 5.38 and mode 5, with only 36 % six-coordinated and a distribution running from three to nine. A hexagonal crystal has essentially every interior particle six-coordinated.

Fourth, and most directly, the KTHNY criteria are stated in terms of how orientational correlations decay rather than how large they are. Equation (3) gives a g6(r)g_6(r) that falls to the noise level by r/Dˉ≈2.4r/\bar{D}\approx2.4 (Figure 3d). An exponential fit returns a correlation length ξ=1.5\xi=1.5 colony diameters; forcing a power law instead returns η6=1.31\eta_6=1.31. A hexatic phase requires power-law decay with η6≤1/4\eta_6\leq1/4, so the measured value exceeds the bound by a factor of 5.

Short-ranged orientational correlations with a finite ξ\xi are the signature of a liquid.

Two limits on that inference should be stated. The cluster spans only about five colony diameters, so the fit covers r/Dˉ=1.4−−4.5r/\bar{D}=1.4--4.5, 8 bins and roughly half a decade; discriminating exponential from power-law decay on goodness of fit alone is therefore weak, and it is the magnitude of η6\eta_6 relative to 1/41/4, not the choice of functional form, that carries the argument. We also did not attempt the companion translational correlation gG(r)g_G(r): with a median of 22 colonies per frame there is no resolvable structure-factor peak from which to take a reciprocal lattice vector, and imposing one by hand produces an artefact of the imposed wavevector rather than a measurement. Finally, KTHNY is an equilibrium theory for passive two-dimensional systems. We use its criteria as a diagnostic, and do not claim that the theory applies to an active suspension.

We therefore describe the arrangement as a two-dimensional liquid with unusually strong short-range order, and specifically not as a crystal. The result of interest is that this liquid is more ordered than hard discs at the same area fraction, not that it is close to crystalline.

Single-colony dynamics

From 24 continuously tracked colonies the mean-squared displacement (Figure 4b) is superdiffusive at short lag, local logarithmic slope 1.4, crossing to 0.8 beyond about 2 s: a persistent random walk. The short-lag exponent falls below the ballistic value of 2 because additive centroid noise contributes a lag-independent offset.

Frame-difference diagnostic, mean-squared displacement, and normalised velocity autocorrelation

Figure 4. (a) Frame-difference diagnostic showing the period-two alternation that reveals frame duplication in the native-rate clip. (b) Mean-squared displacement, 24 tracks, with ballistic and diffusive guides. (c) Normalised velocity autocorrelation; open circles mark the first lags, where the depression is a measurement-noise artefact.

The velocity autocorrelation (Figure 4c) crosses zero at 1.6 s, consistent with the MSD crossover; under the alternative segmentation the crossing falls at 1.4–1.7 s. The first lag is depressed by the anti-correlation that differencing white centroid noise produces, and is excluded. A weak negative lobe beyond the crossing indicates trajectories that curve back on themselves.

Taking vv from the ballistic regime gives 13–17 µm/s across methods, persistence length vτ≈21v\tau\approx21 µm and Deff≈41D_{\mathrm{eff}} \approx41 µm2^2/s. The Reynolds number is ∼9×10−4\sim9 \times10^{-4}. The speed is an order of magnitude below free-swimming volvocine values [10], as expected for colonies swimming into a rigid wall.

Lifetime and breakup

The arrangement is not indefinitely stable. Taking the stable-phase mean plus three standard deviations as a threshold, the first sustained excursion occurs at 127 min. Beyond 105 min the median nearest-neighbour ratio rises to 2.25, the colony count falls from 30 to 19, and the area fraction falls from 0.104 to 0.043. The breakup is therefore a dilution of the layer rather than a collapse into contact. We did not determine the cause; oxygen depletion, nutrient exhaustion, changing illumination and loss of motility in an ageing sample are all uncontrolled.

Colony morphology

The size distribution is bimodal, with a dominant mode near 71 µm and a smaller mode consistent with released juveniles or free unicells, pooled over 2193 detections. Bimodality is expected for a coenobium reproducing by autocolony formation, and is plausibly sharpened by synchronisation to the light:dark cycle during enrichment.

Azimuthally averaged signal profiles across 63 colonies under four independent optical probes place the intensity maximum off centre, at r/R=0.3r/R = 0.3–0.50.5 (Figure 5), the direction a hollow coenobium predicts. The magnitude is small—the central deficit is between 1 % and 4 % depending on probe—and on images of this quality would not have been sufficient to establish the internal structure on its own. It is consistent with the morphology confirmed in § 2.4 rather than evidence for it, and it illustrates the limit of what the low-cost instrument can resolve.

A single colony and azimuthally averaged signal against normalised radius

Figure 5. (a) A single colony, digitally magnified; individual cells are not resolved. (b) Azimuthally averaged signal against normalised radius for 63 colonies under four optical probes, each normalised to its own maximum. All place the maximum off centre; the central deficit is only 1–4% deep.

The dominant colony diameter of 71 µm sits inside the Eudorina range and well outside that of Volvox, which begins near 300 µm. Because the calibration is nominal rather than metrological (§ 2.3), this agreement is worth stating in the other direction as well: the independently confirmed identification is evidence that the assumed length scale is approximately correct, and it is the only independent check on it that we have.

Robustness to method

Recomputing with the alternative segmentation, three results are stable: the g(r)g(r) first-shell position moves from 1.73 to 1.72; the mean colony diameter moves from 71 µm to 68 µm; the size distribution remains bimodal. Three are not: the nearest-neighbour median rises to 1.91–2.07, because the alternative segmentation detects fewer small colonies and a missed colony lengthens its neighbours’ separations; the swimming speed rises to 17 µm/s; and the excess kurtosis of the log-diameter distribution changes sign, so that statistic should not be used as evidence of bimodality even though the bimodality itself survives.

We therefore treat the g(r)g(r) peak position as the primary structural result and the nearest-neighbour median as secondary, quoted with its method spread.

Discussion

A population of swimming colonies held against a rigid boundary adopts a two-dimensional arrangement with liquid-like short-range order, a preferred separation near two diameters, and a lifetime of order two hours. The hard-disc comparison rules out the trivial explanation: at the measured area fraction, excluded volume produces a first shell less than half as high and at a smaller separation, so something is holding the colonies apart at a specific distance.

The order is liquid-like rather than crystalline (§ 3.3), which is the more interesting result: a liquid that is measurably more ordered than hard discs at the same density implies an interaction, whereas a crystal at this area fraction would more likely indicate jamming.

The obvious candidate is the surface-mediated hydrodynamic interaction identified for Volvox by Drescher et al. [7], in which a spinning bottom-heavy swimmer near a no-slip wall generates a flow that couples to its neighbours. Our chamber has the rigid boundary that analysis assumes. However, our colonies are several times smaller than V. carteri, and we have not demonstrated that they spin, nor measured the flow field, so this remains a hypothesis rather than a conclusion.

Two limitations bound what can be drawn from this work. The length calibration is nominal, so all micron-valued quantities carry an uncertainty we cannot bound rigorously; this does not affect the dimensionless results, which are the substance of the paper. And this is a single population observed once: we have no replication, no density series, and no free-surface control. A density series would distinguish a pair interaction, which should give a density-independent preferred separation, from a collective effect, which should not. A free-surface repeat would test the boundary-condition dependence directly, since the hydrodynamic image system of a spinning swimmer differs between a no-slip wall and a stress-free surface.

QuantityValueNote
Colony diameter (dominant mode)71 µmnominal calibration
g(r)g(r) first-shell positionr/Dˉ=1.73r/\bar{D} = 1.731.72 alt. method
g(r)g(r) first-shell height4.8null: 2.1 at 1.57
Area fractionϕnr=0.17\phi_{\mathrm{nr}} = 0.17, ϕtl=0.104\phi_{\mathrm{tl}} = 0.104convex hull, two sequences
Nearest-neighbour separation1.80 ± 0.13 diameters1.91–2.07 alt.
Bond-orientational order ⟨∣ψ6∣⟩\langle|\psi_6|\rangle0.456 ± 0.057null: 0.385 ± 0.054
Six-coordinated interior colonies36 %crystal: ≈100 %
g6g_6 correlation length ξ\xi1.5 diametersexponential fit
g6g_6 power-law exponent η6\eta_61.31hexatic bound: 1/4
Reorientation time τ\tau1.4–1.7 sMSD and VACF agree
Swimming speed vv13–17 µm/sacross methods
Persistence length vτv\tau21 µmderived
DeffD_{\mathrm{eff}}41 µm²/sdiffusive MSD
Reynolds number∼ 9 × 10−410^{-4}vD/νvD/\nu
Stable phase105 minbreakup onset 127 min
Record length153 minelapsed

Table 1. Summary. Dimensionless quantities are independent of the calibration uncertainty; those in microns are not. Every entry is read from the deposited data by the script that typesets this table.

We report the measurement nonetheless because the spatial statistics of boundary-accumulated algal layers appear not to have been quantified for these smaller colonial genera, and because the null-model comparison and the measured lifetime are both directly reusable by anyone repeating the experiment.

Cost as a feature

The entire apparatus is a handheld USB microscope, a clear plastic coin capsule and a sample of pond water. There is no laser, no high-speed camera, no microfabrication, no temperature control, no optical table and no consumable beyond the capsule itself. Total outlay is a small fraction of a single research-grade objective, and the analysis runs on a laptop using open-source libraries.

Active matter is normally studied with equipment that concentrates the field in well-funded laboratories: microfabricated channels, confocal or high-speed imaging, precision thermostatting. The measurements reported here—a pair correlation function, a bond-orientational order parameter, an orientational correlation length, a mean-squared displacement—are standard soft-matter observables, and none of them required any of that. The cost of using an inexpensive apparatus is metrological: an uncalibrated length scale and an unmeasurable depth of field, limitations we have set out in detail in §2.3. What it does not cost is the physics, because every conclusion we draw is dimensionless and therefore independent of the calibration.

A primary motivation in our development of a low-cost apparatus was to bring it within reach of undergraduate teaching laboratories and of research groups in institutions with limited equipment budgets, including those in low- and middle-income countries where the organisms themselves are abundant.

Ethics

This work involved environmental sampling of freshwater microorganisms and required no ethical approval.

Data accessibility

All raw video, derived data, and Python and MATLAB analysis code, including the independent-verification script used in § 3.7, are available at doi:10.7267/0g354q328, in ScholarsArchive@OSU, the institutional repository of Oregon State University (https://ir.library.oregonstate.edu/concern/datasets/0g354q328). The deposit includes all three video sequences: the two analysed here and the record of sample collection. The collection footage and the time-lapse sequence are additionally available online at Refs. [13, 14].

Declaration of AI use

The author used Claude (Anthropic) to edit the manuscript for clarity and to assist in debugging the analysis code. The author reviewed and validated all AI-assisted output and takes full responsibility for the content.

Conflict of interest declaration

The author declares no competing interests.

Funding

No external funding was received for this work.

References

  1. [1]Wager H. 1911 On the effect of gravity upon the movements and aggregation of Euglena viridis, Ehrb., and other micro-organisms. Phil. Trans. R. Soc. Lond. B 201, 333–390.DOI
  2. [2]Platt JR. 1961 “Bioconvection patterns” in cultures of free-swimming organisms. Science 133, 1766–1767.
  3. [3]Childress S, Levandowsky M, Spiegel EA. 1975 Pattern formation in a suspension of swimming microorganisms. J. Fluid Mech. 69, 591–613.
  4. [4]Pedley TJ, Kessler JO. 1992 Hydrodynamic phenomena in suspensions of swimming microorganisms. Annu. Rev. Fluid Mech. 24, 313–358.DOI
  5. [5]Hill NA, Pedley TJ. 2005 Bioconvection. Fluid Dyn. Res. 37, 1–20.DOI
  6. [6]Bees MA. 2020 Advances in bioconvection. Annu. Rev. Fluid Mech. 52, 449–476.DOI
  7. [7]Drescher K, Leptos KC, Tuval I, Ishikawa T, Pedley TJ, Goldstein RE. 2009 Dancing Volvox: hydrodynamic bound states of swimming algae. Phys. Rev. Lett. 102, 168101.
  8. [8]Ishikawa T, Pedley TJ, Drescher K, Goldstein RE. 2020 Stability of dancing Volvox. J. Fluid Mech. 903, A11.DOI
  9. [9]Balasubramanian R. 2021 Volvox barberi (Chlorophyceae) actively forms two-dimensional flocks in culture. J. Phycol. 57; preprint arXiv:1803.03542.
  10. [10]Solari CA, Michod RE, Goldstein RE. 2008 Volvox barberi, the fastest swimmer of the Volvocales (Chlorophyceae). J. Phycol. 44, 1395–1398.
  11. [12]Tuffillaro NB, Ramshankar R, Gollub JP. 1989 Order-disorder transition in capillary ripples. Phys. Rev. Lett. 62, 422–425. (doi:10.1103/PhysRevLett.62.422)
  12. [13]Tuffillaro N. Micro motility in Lake Oroville: sample collection, 29 March 2018 [video]. https://www.youtube.com/watch?v=n5yYDjIwznw
  13. [14]Tuffillaro N. Self-organized micro motility, 20× speed [video]. https://www.youtube.com/watch?v=nbEtZ3l3ABk

Paper details

vlnc:W_s2kk7yqd
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Claude (Anthropic)

View statement
Looking for related papers…
Contents