Ice gliding diatoms establish record-low temperature limits for motility in a eukaryotic cell

Significance

In the harsh Arctic environments, ice-dwelling diatoms remarkably survive and thrive, but their behavioral adaptations are poorly understood. Our study provides direct cellular observations of these microorganisms in ice. We find that many Arctic diatoms possess a unique ice-gliding ability. This ability, absent in temperate diatoms, arises from specific interactions with icy substrates and resilience to extreme cold, allowing ice diatoms to navigate the ice matrix, accessing light and nutrients. Through thermo-hydrodynamic modeling, we provide physical explanations for cold-adapted motility involving the strategies of enhanced energy efficiency and optimized mucilage properties. Our findings, illuminating the behavioral adaptability of phytoplankton in extreme environments, deepen our understanding of ecological resilience and aid in predicting how polar ecosystems respond to climate change.

Abstract

Despite periods of permanent darkness and extensive ice coverage in polar environments, photosynthetic ice diatoms display a remarkable capability of living inside the ice matrix. How these organisms navigate such hostile conditions with limited light and extreme cold remains unknown. Using a custom subzero temperature microscope during an Arctic expedition, we present the finding of motility at record-low temperatures in a Eukaryotic cell. By characterizing the gliding motility of several ice diatom species, collected from ice cores in the Chukchi Sea, we record that they retain motility at temperatures as low as $-$15 ^°C. Remarkably, ice diatoms can glide on ice substrates, a capability absent in temperate diatoms of the same genus. This unique ability arises from adaptations in extracellular mucilage that allow ice diatoms to adhere to ice, essential for gliding. Even on glass substrates where both cell types retain motility at freezing temperatures, ice diatoms move an order of magnitude faster, with their optimal motility shifting toward colder temperatures. Combining field and laboratory experiments with thermo-hydrodynamic modeling, we reveal adaptive strategies that enable gliding motility in cold environments. These strategies involve increasing internal energy efficiency with minimal changes in heat capacity and activation enthalpy, and reducing external dissipation by minimizing the temperature sensitivity of mucilage viscosity. The finding of diatoms’ ice gliding motility opens new routes for understanding their survival within a harsh ecological niche and their migratory responses to environmental changes. Our work highlights the robust adaptability of ice diatoms in one of Earth’s most extreme settings.

Ice diatoms are important components of algal communities in polar regions, serving as the major primary producers prior to the spring phytoplankton bloom (1–3). These microorganisms exhibit extraordinary adaptations to the extreme conditions within sea ice (4–11). Polar ice caps envelop up to 13% of the Earth’s ocean surface and are characterized by internal porous structures filled with brine channels, creating a unique habitat with subfreezing temperatures and highly fluctuating salinity (12–14). Despite the harsh conditions, ice diatoms have developed specialized adaptations to remain active, including increased efficiency of metabolism and fluidity of lipid compositions (6, 15–18). While some biochemical adaptations that enable ice diatoms to thrive under extreme cold are known, their locomotive strategies remain largely unexplored.

Ice diatoms’ behavioral adaptations critically influence their survival and interaction with the icy habitat (19, 20). They are believed to “select” specific depths within the ice core that offer optimal light, nutrients, and salinity, enhancing their survival prospects (21–26). Furthermore, ice diatoms secrete substances such as ice-binding proteins (IBPs) and extracellular polymeric substances (EPS) (27–29). These substances, while protecting the diatoms from cold and balancing osmotic stress (16, 30), also alter the process of ice crystallization (31–34) and impact polar ocean carbon cycling (30, 35, 36). The dual ability of ice diatoms to relocate within the ice for survival and to reshape their habitat highlights the importance of understanding their motility mechanism. By migrating within brine channel networks in sea ice, which becomes permeable at temperatures above −5^°C (12), diatoms can spread and affect polar ecosystems, including initiating new under-ice blooms (37). It has been hypothesized that the diatoms living in brine channels might seed under-ice blooms due to increased ice porosity during warming spring (38). As polar regions undergo rapid changes in climate, it is urgent to understand the adaptive physiology of these unique species living in ice and how they might evolve with rising temperatures (32, 39, 40).

Despite the critical role of migration ability, there remains a significant gap in our understanding of ice diatoms’ motility in their native icy environments. Indirect empirical evidence, such as changing ice thickness leading to the relocation of algal colonies at the ice bottom, has provided estimates of diatom movement, suggesting a minimum velocity of approximately 1.5 cm per day (41). Discrepancies between observed cell densities and photosynthetic measurements in ice, unexplained by local growth rates alone, further suggest active migration of cells within the ice (42). Additionally, field experimental manipulations of snow depth, which influence light penetration, have demonstrated that ice diatoms can strategically reposition themselves within the ice column in a few days (43). These behavioral adjustments, believed to be adaptive responses to shifting light conditions, suggest an active engagement with their environment rather than passive diffusion (43). Following laboratory observations of an ice diatom species, Cylindrotheca closterium, reveal that it can move on the surface of a culture flask at 0 ^°C (43). However, translating this finding to the context of natural ice environments introduces uncertainties, because the physical and chemical properties of laboratory surfaces differ significantly from those of sea ice. The lack of direct observations within the actual ice context leaves critical questions about ice diatom motility, including the specific mechanisms they employ and their responses to environmental changes like temperature fluctuations.

In this study, we present direct cellular observations of ice diatoms demonstrating unique ice gliding ability across their natural habitat in sea ice. This ability reflects their adaptations to two critical factors: icy substrates and cold temperatures. First, we show that only ice diatoms can move on ice; temperate diatoms immediately lose motility upon contact due to lack of adhesion to ice. Second, we find that ice diatoms maintain enhanced motility at freezing temperatures. On glass substrates where both types can move, ice diatoms exhibit gliding speeds an order of magnitude faster than their temperate counterparts. To understand the mechanisms behind this superior cold motility, we investigate temperature-dependent factors governing gliding speed. By introducing traction force microscopy (TFM) tailored for diatoms, we map the forces applied and experienced by gliding diatoms, revealing distinct interactions with the substrate. Based on experimental insights, we develop a thermo-hydrodynamic model incorporating primary temperature-dependent factors for gliding motility, revealing strategies ice diatoms have evolved to maintain efficient movement under extreme cold conditions. Furthermore, we investigate population-level motility of ice diatoms and reveal temperature-dependent patterns characterized by gamma and chi distributions. As temperature increases, ice diatoms diversify gliding speeds, indicating adaptive behaviors that enhance ecological success under changing conditions.

Results

Distinct Ability of Arctic Diatoms to Glide in Icy Environments.

To investigate the dynamic behavior of ice diatoms within sea ice, we conduct a 45-d Arctic expedition aboard the R/V Sikuliaq, collecting ice cores from the first-year sea ice in the Chukchi Sea (70.3^° to 71.6^° N, 161.0^° to 165.8^° W) from 12 ice stations throughout the summer season in 2023 (June 15 to July 30) (Fig. 1A and SI Appendix, Table S1). These cores feature biota layers embedded in the ice, which appear brown and with reduced transparency at specific depths, as shown in Fig. 1A (ice station 21 at 71.3^° N, 164.6^° W).

Fig. 1.

Gliding motility of ice diatoms. (A) Left panel: Expedition route and ice stations in the Arctic. Upper Right panel: Biotic layer observed in an Arctic ice core from the Chukchi Sea; the core is shown turned 90 degrees, with its original upper surface oriented toward the right side of the image. (Scale bars, 20 mm.) Lower Right panel: Cross-sectional view of the ice core showing the region with the highest diatom density. (Scale bars, 2 mm.) (B) Time-lapse trajectories of ice diatoms gliding directly on an ice surface, imaged using subzero temperature controlled microscopy. (Scale bar, 100 $\mathrm{\mu }$m.) (C) Comparison of the gliding motility of ice diatoms (Navicula sp. and Pleurosigma sp., orange) and temperate diatoms (Navicula arenaria var. rostellata, Navicula sp. and Pleurosigma sp., blue) on both glass and ice substrates. At 0 ^°C, ice diatoms maintain robust motility. Temperate diatoms at the same temperature glide an order of magnitude slower on glass and completely lose gliding ability on ice. The comparison demonstrates the specialized cold and substrate adaptation of the ice-dwelling species. Error bars indicate SE from experiment-level means ($n=2$ or 3). Cell- and species-level data are shown in SI Appendix, Fig. S5. (D) Schematic of adhesion strength measurements: Poiseuille flow applies wall shear to diatoms on glass or ice surfaces (Upper panel). The critical shear stress ($>$10% cells detach) differs by substrate and cell type. Both ice and temperate diatoms remain attached on glass, whereas only ice diatoms retain adhesion on ice (Lower panel). (E) Average gliding speed as a function of temperature for three species of ice diatoms collected in the field: () Navicula sp., () Pleurosigma sp., () Entomoneis sp. Error bars indicate SE from replicate experiments for Navicula sp. (2 to 4 replicates; 20 to 150 cells per experiment) and cell-to-cell variability for Pleurosigma sp. (2 to 7 cells per temperature due to limited cell availability). (F) Temperature dependence of cellular motility and transport across the tree of life, illustrating that ice diatoms approach the lowest temperature limits of motility (44–55).

At each ice station, we systematically image the diatoms under high magnification (10$\times$, 20$\times$, 40$\times$) to understand the biodiversity of ice diatoms trapped in ice cores. Based on morphological markers (SI Appendix), we predominantly find pennate diatoms such as Navicula sp. (confirmed by DNA barcodes) together with Pleurosigma sp., Nitzschia frigida, Entomoneis sp., Pseudo-nitzschia sp., C. closterium, and Fragilariopsis sp. (each provisionally assigned) (SI Appendix, Figs. S2 and S3). Our microscopic observations further reveal that the diatoms presenting in the ice core can also be found in the water column.

While the presence of diatoms in ice cores has been reported for decades (2, 56), no cellular-scale observations of their behavior in ice exist. Ice is a scattering media, making direct observation challenging. We circumnavigate this issue by extracting diatoms from ice cores and redispersing them on thin ice film surfaces or within microice channels (see Materials and Methods for details). These preparations are then examined under bright field at desired temperatures using our customized temperature-controlled microscope (see Materials and Methods for details). Remarkably, we find that the Arctic pennate diatom species can actively glide on ice surfaces and within ice channels, as shown in Fig. 1B and SI Appendix, Fig. S4 and Movies S1 and S2. Although salinity in natural brine channels varies with temperature (26), the salinity in our experiments remains effectively constant (approximately 36 ppt), because the timescale of diatoms’ motility is much shorter than that of the ice film growth during our observations. At deeper supercooling at this salinity, freezing remains kinetically delayed. Heterogeneous nucleation triggers ice formation after a lag, leaving a prefreeze window for imaging (SI Appendix). Mean gliding speeds are calculated across the full temperature range down to −15^°C, but speed-distribution analyses, which require longer uninterrupted records, are restricted to T ≥ −7^°C under our experimental conditions, since ice nucleation rate (SI Appendix, Eq. S1) rises steeply with additional supercooling.

To illustrate the uniqueness of ice gliding motility, we compare ice diatoms (Navicula sp. and Pleurosigma sp.) with various temperate species (Navicula sp., N. arenaria var. rostellata, and Pleurosigma sp.) gliding on both glass and ice substrates. On glass surfaces at 0 ^°C, although both ice and temperate diatoms exhibit gliding motility, the ice diatoms glide significantly faster–nearly an order of magnitude greater than their temperate counterparts on average (ice diatoms, $1.92\pm 0.24$ $\mathrm{\mu }$m/s; temperate diatoms, $0.17\pm 0.09$ $\mathrm{\mu }$m/s), as shown in Fig. 1C and Movie S3. Cell-level data for individual speeds are provided in SI Appendix, Fig. S5. This enhanced motility reveals remarkable cold-adapted motility. Interestingly, on ice surfaces, temperate diatoms completely lose their ability to glide; any movement observed is limited to passive drifting due to uneven ice surfaces or self-interactions in aggregations (Fig. 1C and Movie S3). In contrast, ice diatoms maintain consistent gliding speeds comparable to those observed on glass, as shown in Fig. 1C. This stark contrast indicates the unique ability of ice diatoms to navigate on and through ice, suggesting that they have evolved specialized physiological traits to interact effectively with their icy habitat. These comparisons indicate two exceptional aspects regarding the adaptations of ice diatoms’ motility: i) the unique ability to interact with an icy substrate and ii) the motile resilience to cold temperatures.

Gliding motility in diatoms is widely accepted to rely on mucilage-mediated adhesion with the substrate to establish the traction necessary for movement: cells extrude mucilage threads through the raphe, anchor them to the substrate, and generate propulsion by pulling against these anchored threads (57–60). We quantify the adhesion strength of this mucilage by measuring the shear stress required to detach diatoms from surfaces in a fluidic channel (detailed in Materials and Methods and SI Appendix). Our quantitative measurements demonstrate that ice diatoms exhibit substantial adhesion strengths on both ice and glass surfaces at 0 ^°C. On ice substrates, ice diatoms withstand a critical shear stress (defined as the shear stress at which more than 10% of cells detach; see SI Appendix, Fig. S6 for details) of approximately 90 mPa. In contrast, temperate diatoms exhibit negligible adhesion to ice, detaching entirely at the lowest shear stress applied (1 mPa). On glass substrates, adhesion strengths are considerably higher, with critical shear stresses of approximately 563.2 mPa for ice diatoms and 1,126.6 mPa for temperate diatoms, as shown in Fig. 1D. These behaviors align with their respective motility on different substrates (Fig. 1E). The alignment between the mapping of motility and adhesion strength across different substrates suggests that the mucilage secreted by ice diatoms may include specific ice binding capabilities that enable attachment to ice and contribute to the unique gliding ability on ice (Discussion).

Temperature is a fundamental state variable which influences nearly all cellular processes, including those governing motility (61–64). Despite its significance, cellular motility under extreme cold conditions has not been extensively studied (45, 52, 53, 65, 66). To assess the impact of temperature variations on ice diatoms’ motility, we quantify the average moving speed of three prevalent wild ice diatoms in the Arctic, i.e., Navicula sp., Pleurosigma sp., and Entomoneis sp. The majority of data are collected within 24 to 48 h from the moment of sampling using the customized temperature-controlled microscope aboard the R/V Sikuliaq. Remarkably, all three species remain motile at freezing or subfreezing temperatures, and Navicula sp. is able to move down to −15^°C. Moreover, these species show a general trend where their average gliding speed increases with rising temperatures, reaching a plateau at around 10 to 12 ^°C (defined as optimal temperature), and then declining sharply at 18 to 22 ^°C (defined as maximum temperature), as shown in Fig. 1E. In contrast, temperate diatoms in the same genera (Pleurosigma sp. and Navicula sp.) exhibit an optimal motility temperature around 30 ^°C and a maximum deactivating temperature near 36 to 42 ^°C (see SI Appendix and SI Appendix, Fig. S7 for details). When exposed to colder temperatures, these temperate diatoms almost cease their motion at approximately −1^°C. The systematic shift of temperature response curves toward lower temperatures corresponds with the enhanced motility at freezing temperatures for ice diatoms (Fig. 1C) and highlights notable adaptations to cold. We explore the underlying mechanisms that counteract the inhibitory effects of extreme cold on gliding processes in the following section.

Having established the unique ability of ice diatoms to glide on ice substrates and their enhanced motility at cold temperatures, we wonder how this compares to the broader context of cellular motility across the tree of life. By surveying published results on the impact of temperature on motility in single cells and the related activity of motor proteins (44–55, 67–70), we create a comprehensive dataset that includes the lowest temperature experiments performed to date, as shown in Fig. 1F (up to 50^°C) and SI Appendix, Fig. S8. Notably, we find that ice diatoms exhibit gliding motility at the lowest recorded temperature limits in a Eukaryotic cell. Their adaptation across a broad range of temperatures provides a possible explanation for why they form dominant communities in ice cores (SI Appendix, Fig. S2) (1, 9, 56).

Strategies for Cold-Adapted Motility in Navicula sp.

To understand the strategies enabling ice diatoms to adapt their motility to freezing temperatures, we concentrate our investigations on an Arctic species, Navicula sp., which we isolate and fortunately culture from an ice core at Ice Station 87 (71.7^° N, 164.0^° W) in the Chukchi Sea (see Materials and Methods for details). Our observations of similar gliding speeds of ice diatoms at freezing temperatures on both ice and glass substrates (Fig. 1D) suggest that the dominant mechanisms controlling their gliding speeds are independent of the type of substrates. Therefore, for the mechanistic studies presented hereafter, we choose flat glass substrates for our controlled laboratory experiments to explore their gliding mechanisms and dissect their temperature-dependent motility.

Gliding motility is a fascinating mode of transport utilized across the tree of life, both in prokaryotic and eukaryotic cells (71–74). Although many studies have investigated the mechanisms of motility in pennate diatoms, a quantitative understanding of the key factors governing gliding speeds remains limited (75–78). Gliding motility in temperate diatoms is commonly interpreted through an actomyosin framework, known as adhesion motility complex, in which myosin motors pull mucilage threads rearward along stationary actin cables, propelling the cell forward (59, 60, 79–82), as schematically shown in Fig. 2A. Although alternative mechanisms have been proposed (83), previous inhibitor-based experiments (84), recent molecular evidence identifying myosin involvement in mucilage transport (85), and consistent observations of retrograde mucilage streaming along the raphe (86) collectively support this interpretation. Employing similar experimental approaches as in temperate diatoms, we investigate the internal mechanisms driving the motility of ice diatoms.

Fig. 2.

Physical factors governing the gliding speed of ice diatoms. (A) Schematic illustrates diatom gliding motility driven by acto-myosin and mucilage thread-based latching machinery. (B) Mucilage traces left by ice diatoms, Navicula sp., labeled with Wheat Germ Agglutinin (WGA), indicate movement paths (orange arrows). (Scale bar, 20 $\mathrm{\mu }$m.) (C) Three-dimensional visualization of actin cables (green) within Navicula sp., illustrates cytoskeletal structures crucial for gliding motility. (Scale bar, 10 $\mathrm{\mu }$m.) (D) Scanning electron micrograph of Navicula sp. reveals thin slits (raphes) on the diatom frustules (white arrows). (Scale bar, 3 $\mathrm{\mu }$m.) (E) Side view of a gliding ice diatom on a glass substrate shows a tilt resulting from torque competition between active and passive drag forces. (Scale bar, 10 $\mathrm{\mu }$m.) (F) A directional flux beneath Navicula sp. via raphe grooves is revealed using fluorescent polystyrene beads (0.5 $\mathrm{\mu }$m), indicating that tracer particles latched to the ventral raphe move opposite to diatom motion. The thick curve shows the diatom’s trajectory; thin curves show tracer particles’ trajectories. Inset: Schematic illustrates tracer particles latched to mucilage threads that are moving along the ventral raphe. The tracer particle movement reflects internal acto-myosin kinetics. (Scale bar, 20 $\mathrm{\mu }$m.) (G) Distribution of gliding velocities for diatoms and speeds of latched tracer particles, $v$, in the ventral region. (H) Average speed of tracer particle clusters attached to mucilage threads is independent of their size (projection diameter, $d$). The color map indicates intensity of kernel density estimation, $I$. (I) Comparison of temperature-dependent motility of ice diatoms absent of tracer particles, $V_{\mathit{av}}$ () and mucilage latched tracer particle velocity, $V_{\mathit{particles}}$ () reveals significant reduction from velocity generated by internal machinery to final gliding velocity. Error bars indicate SE from replicate experiments (2 to 6 replicates for raphe-particle streaming; 2 to 4 replicates for Navicula sp. gliding). Inset: Normalized velocity loss, $V_{\mathit{loss}}=(V_{\mathit{particles}}-V_{\mathit{av}})/V_{\mathit{particles}}$, demonstrates the temperature-dependent impact of external boundary conditions.

The Arctic Navicula sp. is kept at 3 ^°C with a 12-h day/night cycle. We examine their mucilage secretion by allowing these diatoms to glide on a glass surface for 12 h and staining the substrate with Wheat Germ Agglutinin (see Materials and Methods for details). Fine mucilage residues left on the substrate are directly visualized, as shown in Fig. 2B. Additionally, we establish the actin cytoskeleton architecture in the ice diatom by labeling their actin filaments with phalloidin (see Materials and Methods for details). Two actin cables running parallel to the long axis are colocated within the raphe region (Fig. 2C). The raphe structure in ice diatoms is revealed in scanning electron microscopy (SEM), as shown in Fig. 2D. These observations suggest that ice diatoms utilize a gliding mechanism similar to that of temperate species (60).

While powered by internal myosin motors, moving diatoms strongly interact with the substrate and the surrounding hydrodynamic environment. Side-view imaging of live gliding ice Navicula sp. indicates that these diatoms consistently adopt tilted orientations during movement (average tilt angle: $1.95\pm 0.{58}^{°}$; Fig. 2E, SI Appendix, Figs. S9 and S10 and Movie S4). The tilted angles observed in gliding ice diatoms correlate well to previous observations on temperate diatom, Craticula cuspidata (81). The observed tilt angles in both ice and temperate diatoms reflect the interplay between the torque in the vertical plane generated by propulsive forces from the adhesion motility complex and the opposing torque generated by external resistive forces acting off the cell’s center line on the cell body (Discussion).

Gliding motility of diatoms is governed by complex processes involving internal and environmental factors. To explore how temperature impacts these processes, we categorize the factors governing gliding speed into two temperature-dependent regimes: i) internal interactions, which encompass all temperature-modulated cellular dynamics such as the driving forces generated by the adhesion motility complex and dissipation within the thin raphe; ii) external interactions, which include all interactions outside the cell such as the temperature-dependent hydrodynamic drags, where temperature can alter the surrounding viscosities.

To quantify the effects of these regimes, we isolate internal contributions from external effects by introducing freely suspended $0.5$ $\mathrm{\mu }\mathrm{m}$ polystyrene tracer particles into the surrounding water, which are then captured beneath the diatom (ventral region). Interestingly, these tracer particles move opposite to the diatom’s travel direction, indicating attachment to mucilage threads connected to the internal myosin machinery (Fig. 2F and Movie S5). This opposing movement, which contrasts with the uncorrelated movement of dorsal particles (Movie S6), aligns with findings on temperate diatoms (86, 87) and suggests a generalized gliding mechanism that extends to ice-dwelling diatoms.

The movement of these ventral tracer particles provides a direct map of the internal machinery’s kinetics. The measurement relies on particles attaching to mucilage threads that are not anchored to the substrate (see the Inset of Fig. 2F); in this state, the system is effectively “decoupled” from the surface. Quantifying particle speeds reveal that when particle movement is present, diatom movement slows down significantly (Fig. 2G). The reduction arises because the myosin-generated force is partially diverted into driving the attached particles, rather than directly interacting with the substrate to generate traction and propel the cell forward. Because this measurement is isolated from the complexities of surface traction, the particle’s movement is a reliable proxy for the internal motor’s dynamics. Additionally, we notice that some tracer particles aggregate into clusters of varying sizes. Their movement speeds show no correlation with cluster size (Fig. 2H). This invariant speed across different loading forces suggests that the diatom maintains a uniform myosin motility rate within a certain loading regime. Thus, we consider the movement speed of the tracer particles to be a reasonable approximation of the internal speed of the myosin-mucilage threads involved in diatom motion.

Comparing the average speeds of tracer particles with the average speeds of diatoms without particle attachment shows a significant velocity reduction (Fig. 2I). This velocity loss persists across the entire temperature range explored and intensifies as temperature decreases, reaching approximately 80% of velocity loss at temperatures below $-5$ ^°C (Inset of Fig. 2I). This velocity reduction is likely due to external drags affecting the diatom more than the smaller particles. The temperature-dependent loss highlights the distinct impact of the external drags on diatom motility. Subsequently, we examine how a gliding diatom interacts with the environments.

Gliding motility necessitates close proximity to a substrate. Between the diatom surface and the substrate there is a gap filled with a thin film of mucilage solution [0.5 to 1 $\mathrm{\mu }$m (57)], which places the system within the lubrication regime and contributes substantially to viscoelastic drag. While previous studies have primarily focused on internal friction and classical Stokes drag from ambient fluid flow (85), the specific role of viscoelastic drag from mucilage beneath the diatoms has not been explicitly addressed.

The kymograph of ventral tracer particles, tracked relative to the moving cell (Fig. 3A), reveals a predominant, though not exclusive, behavior of gliding ice Navicula sp., providing a window into the drag effect. Particles are driven by the internal machinery from the cell tip to the center (diagonal trace) before stalling near the cell’s center, at which point they are passively dragged along with the cell body (vertical trace). The two-phase motion of active driving followed by passive dragging suggests the mucilage disconnects from the myosin near the cell’s center. The disconnection is likely due to a central discontinuity in the raphe (Fig. 3A). Although actin cables are continuous, allowing continuous movement of myosin (Fig. 2C), the raphe discontinuity interrupts movement of mucilage threads. The mucilage accumulates near the middle and eventually diffuses in the surrounding water. The discontinuous geometry suggests a division of the ventral region into a leading edge, where active “driving” occurs via adhesion motility complex, and a trailing edge, where “dragging” effects are predominantly observed.

Fig. 3.

Thermo-hydrodynamic model incorporating internal and external factors illustrates ice diatoms’ strategies for enhanced gliding speed in cold environments. (A) Upper panel: SEM image of an ice diatom showing a raphe discontinuity (white boxes) dividing the ventral region into leading and trailing edges. (Scale bar, 1 $\mathrm{\mu }\mathrm{m}$.) Lower Right panel: Kymograph from motility videos showing tracer particles latched onto the ventral raphe; particles on leading and trailing edges are highlighted above (white). Lower Left panel: Diatom’s velocity over time. In the diatom’s frame of reference, leading-edge particles moving backward along the raphe indicate active force generation. Particles near or in the trailing edge are passively dragged, demonstrating that mucilage released into the surrounding water contributes to the lubrication film. [Scale bars, $10$ $\mathrm{\mu }\mathrm{m}$ (horizontal) and 1 s (vertical).] (B) Two-dimensional traction force microscopy maps taken over time reveal driving forces consistently applied in the leading edge and competing drag forces experienced in the trailing edge. Arrows indicate direction of traction forces. (Scale bar, 10 $\mathrm{\mu }$m.) (C) Particle image velocimetry (plotted in diatom frame of reference) depicts ambient hydrodynamic drag experienced by a gliding ice diatom. Arrows indicate direction of flow and colors represent velocity magnitude. White lines depict streamlines. (Scale bar, 20 $\mathrm{\mu }$m.) (D) Schematic highlighting three regimes governing gliding motility: (I) Internal driving region from adhesion motility complex, and external drag regions from (II) ventral viscoelastic drag, and (III) ambient Stokes drag. (E) Comparative analysis of average gliding speeds for ice and temperate marine diatoms across varying temperatures, with lines representing the best fits to Eq. 1: () ice Navicula sp., () ice Pleurosigma sp., () temperate Navicula sp., () temperate Pleurosigma sp., () temperate Pinnularia sp., showing that the model captures the primary factors governing gliding speeds. (F) Comparison of activation enthalpy ($\mathrm{\Delta }H$), changes in heat capacity ($\mathrm{\Delta }C$) and sensitivity of mucilage viscosity to temperature ($B$) for both ice and temperate marine diatoms. The differences highlight the adaptive strategies allowing ice diatoms to glide in extreme cold.

Traction force microscopy (TFM), which measures forces at the external interface between the secreted mucilage and the substrate, provides direct evidence of where driving and dragging traction forces originate. Analysis of sustained directional movement of cells, as shown in (Fig. 3B and SI Appendix, Fig. S11 and Movie S7) shows that traction forces opposing the diatom’s forward movement, denoted as backward-directed traction forces, are predominantly localized at the leading edge. Direction-reversal experiments further confirm this traction-force distribution: upon reversing cell direction, the traction pattern flips accordingly, consistently repositioning backward-directed traction forces to the new leading edge (SI Appendix, Fig. S11). These backward-directed traction forces indicate that the diatom actively pulls the substrate backward relative to its direction of motion, thereby propelling itself forward. This interpretation aligns closely with observations from particle streaming in the ventral region, where beads adhered beneath the cell are consistently transported backward, opposite the diatom’s forward movement (Fig. 2F). Collectively, these complementary observations strongly support the adhesion motility complex model and establish that the backward-directed traction forces at the leading edge serve as the primary driving forces for gliding motility in the Arctic ice Navicula sp.

Conversely, traction forces aligned with the movement direction are primarily at the trailing edge. We differentiate two distinct physical mechanisms underlying these trailing-edge forces: i) active opposing forces due to bidirectional myosin motor activity (“tug-of-war”), consistent with the active internal dynamics and qualitative biofilm-substrate displacements described in prior studies (85, 88); ii) passive viscoelastic drag arising from mucilage threads continuously detached from myosin motors, subsequently suspended and sheared within the surrounding fluid (SI Appendix). Opposing forces observed during stalling periods illustrate the tug-of-war scenario (see kymograph in SI Appendix, Fig. S12, 1 to 2 s). In contrast, during sustained forward motion, trailing-edge traction predominantly arises from passive viscoelastic drag (see kymograph in SI Appendix, Fig. S12, 5 to 11 s). This passive drag scenario is also supported by the synchronous motion of tracer particles at the trailing edge (Fig. 3A).

The magnitude of ventral viscoelastic drag depends on the mucilage rheological properties. Since the mucilage secreted by gliding diatoms persists beneath the cell and becomes hydrated, we model it as a viscoelastic medium described by a standard linear solid (SLS). As continuous shear and dilution are expected to disrupt any long-lived network, the parallel spring in SLS can be neglected and the SLS reduces to its Maxwell limit (89) (SI Appendix). This representation inherently accounts for energy dissipation related to both viscous flow and the rupturing and peeling of mucilage strands from the substrate as the diatom advances. Under steady-state conditions, the shear stress is then proportional to an effective viscosity. The estimated viscoelastic drag (200 to 2,000 pN at $5\mathrm{\mu }\text{m}/\text{s}$) is on the same order of magnitude as the propulsive force generated by ice diatoms (1,200 pN; estimated from TFM; see SI Appendix). Therefore, that ventral viscoelastic drag places a substantial load on the motility machinery and thereby limits the steady-state gliding speed that a diatom can achieve.

To fully characterize external hydrodynamic resistances and verify the flow regime, we perform particle image velocimetry (PIV) measurements above a gliding diatom (Fig. 3C). The observed velocity fields confirm that the fluid dynamics align with a Stokes-flow regime, thus excluding alternative mechanisms such as self-generated osmotic or active flows.

Since these mechanisms are generalizable to both ice and temperate diatoms, we develop a model incorporating internal and external temperature-dependent terms to compare species and reveal cold-adapted strategies (Fig. 3D). Assuming steady-state motion, we balance the internal force (propulsion from myosin motors and friction from mucilage threads moving through the raphe) with external forces dominated by ventral viscoelastic drag, along with a smaller Stokes drag contribution: $F_{\mathrm{inner}}=F_{\mathrm{ventral}}+F_{\mathrm{stokes}}$, which yields an expression for the average velocity of a gliding diatom (details in SI Appendix):

$$\begin{array}{c}\hfill V_{\mathit{av}}=\frac{N\left(T\right)f_{\mathrm{single}}\left(T\right)}{{\eta }_{\mathrm{mucilage}}(T)\frac{\mathit{WL}}{h_0}\mathrm{\Phi }\left(k\right)+{\eta }_{\mathrm{water}}(T){\varsigma }_{\mathit{up}}L},\end{array}$$ [1]

where $T$ is the temperature, $N(T)$ is the number of active myosin motors whose associated mucilage strands are effectively adhered to the substrate, $f_{\mathrm{single}}(T)$ is the net force from a single mucilage-thread-myosin unit, combining myosin-generated propulsion and internal frictional resistance from mucilage threads sliding within the raphe, ${\eta }_{\mathrm{mucilage}}(T)$ and ${\eta }_{\mathrm{water}}(T)$ are the temperature-dependent viscosities of mucilage and water, respectively, $L$ is the length of diatom, $W$ is the width of the diatom, $\mathrm{\Phi }\left(k\right)$ is a geometric factor dependent on the ratio $k=h_1/h_0$, and $h_1$ and $h_0$ denote the front and back gap size between the bottom surface of the cell and the substrate respectively. The gap is defined as the distance between the ventral side of the diatom and the substrate.

We first account for temperature-dependent internal factors: $N(T)$ is governed by enzyme-catalyzed kinetics intrinsic to the myosin–actin interaction cycle, and $f_{\text{single}}(T)$ is controlled by conformational changes in myosin molecules (90, 91). These biological processes depend on temperature. A fundamental theory that describes the temperature dependence of such biological processes is the Eyring–Evans–Polanyi (EEP) transition state theory (92, 93). Based on EEP and under the assumption that the temperature dependence arises mainly from entropy changes, we model the internal force as (94): $F_{\mathrm{internal}}=N\left(T\right)f_{\mathrm{single}}(T)=f_0{N}_0{\tau }_0\frac{k_B}{h}{e}^{\frac{\mathrm{\Delta }{S}_0}{R}}{T}_0^{-\frac{\mathrm{\Delta }C}{R}}{\left(\frac{1}{T}\right)}^{-\left(\frac{\mathrm{\Delta }C}{R}+1\right)}{e}^{-\frac{\mathrm{\Delta }H}{\mathit{RT}}}$, where $N_0$, $f_0$, ${\tau }_0$ and $\mathrm{\Delta }{S}_0$ are the activated number of myosin motors, the force per motor, the characteristic time of the mechanochemical cycle, and the entropy change, respectively, all evaluated at the reference temperature $T_0$ (typically $293\mathrm{K}$). $k_B$, $h$, and $R$ are Boltzmann’s constant, Planck’s constant, and the gas constant. $\mathrm{\Delta }H$, the activation enthalpy, sets the enthalpic height of motors’ activation barrier, whereas $\mathrm{\Delta }C$, the heat-capacity change, governs how that barrier and the associated entropy shift with temperature.

We account for the impact of temperatures on environmental factors by incorporating the temperature-dependent viscosity of seawater, ${\eta }_{\text{water}}(T)$ (95, 96), and modeling the mucilage viscosity using the Vogel–Fulcher–Tammann (VFT) equation (97): ${\eta }_{\mathrm{mucilage}}(T)={\eta }_0{e}^{\left(\frac{B}{T-T_{\mathit{FV}}}\right)}$, where ${\eta }_0$ is the reference zero-shear-rate viscosity, $T_{\mathit{FV}}$ is the critical glass transition temperature, and $B$ characterizes the sensitivity of mucilage viscosity to temperature changes. Robustness analysis (SI Appendix) reveals that among these external factors, the temperature-dependent mucilage viscosity, ${\eta }_{\mathrm{mucilage}}(T)$, is the dominant parameter determining temperature-dependent gliding motility.

Based on temperature-dependent gliding motility of both ice diatoms and temperate diatoms, we fit Eq. 1 to our data (SI Appendix; parameters listed in SI Appendix, Table S3). The current model effectively describes the experimental data for both ice and temperate diatoms, as shown in Fig. 3E. The model captures the primary physical phenomena governing temperature-dependent motility.

The parameters from the EEP model ($\mathrm{\Delta }C$ and $\mathrm{\Delta }H$) and the VFT model ($B$) provide physical intuition for predicting the strategies that ice diatoms might have evolved to enhance motility under freezing conditions. At the molecular level, $\mathrm{\Delta }C$ and $\mathrm{\Delta }H$ collectively capture temperature sensitivities of processes such as myosin catalytic rates, actin–myosin binding, and intracellular ATP turnover, whereas $B$ likely reflects temperature-driven changes in the cross-link density of the extracellular mucilage network. Because all three quantities are obtained by fitting data within the thermo-hydrodynamic framework, the absolute values in Fig. 3 E and F are model-informed estimates, not direct experimental observables. Compared to temperate species, ice diatoms exhibit a smaller change in heat capacity ($\mathrm{\Delta }C$), suggesting they are better adapted to maintaining internal stability against temperature fluctuations (Fig. 3F). This stability is advantageous in consistently cold environments where large internal energy fluctuations can be detrimental. Additionally, a lower activation enthalpy ($\mathrm{\Delta }H$) reduces the enthalpic barrier, enhancing energy efficiency at freezing temperatures to sustain internal driving dynamics. In response to the external hydrodynamic effect in extreme conditions, ice diatoms show reduced sensitivity of mucilage viscosity to temperature variations, indicated by a smaller $B$ (Fig. 3F). Assuming similar viscosities at room temperature, the mucilage of ice diatoms increases in viscosity more slowly as temperatures decrease compared to that of temperate diatoms. This adaptation allows ice diatoms to maintain higher mucilage fluidity and effectively reduce resistance forces, thereby facilitating enhanced motility in freezing conditions.

Temperature-Dependent Motility in Gliding Population.

The dynamics of ice diatoms’ gliding motility extends beyond simple considerations of average velocity, as shown in Fig. 4A. Our analysis of the statistical distribution of diatom motility reveals two distinct regions in the probability density function: a sharp decline at low speeds followed by a secondary peak and a subsequent decline at higher speeds (Fig. 4A). This distribution shows a remarkable dependence on temperature. At lower temperatures, the secondary peak is less pronounced, largely obscured by the prevalence of nonmotile diatoms, resulting in a distribution that mostly decays. As temperatures increase, this secondary peak becomes more distinct, and the speeds at which this peak occurs also increase. Concurrently, as temperature rises, we observe a broadening range of speeds. This indicates that individual diatoms display more variability in their motility at increased temperatures.

Fig. 4.

Temperature-dependent motility patterns in diatom populations. (A) Trajectories and gliding speed distributions on a glass surface at $-$3, 6, and 18 ^°C, displayed from left to right. Upper panel: diatom trajectories. (Scale bar, 200 $\mathrm{\mu }$m.) Lower panel: the corresponding probability density functions, $p$, of gliding speeds $V$ in red, fitted with a combination of a gamma distribution and a chi-distribution tail in black. (B) The shape parameter $k_s$ from the chi distribution rises with temperature, reflecting a progressively less-skewed speed distribution and suggesting that additional motility modes become engaged. (C) Normalization of $k_s$ by the velocity ratio $V_{\mathit{av}}/V_{\mathit{particle}}$, showing how motility diversity correlates with efficiency across different temperatures.

These patterns deviate from a Gaussian distribution, which is typically indicative of random and uncorrelated movements. The absence of a Gaussian profile suggests that diatom motility is influenced by more complex internal and external factors rather than being purely stochastic. By analyzing the mean square displacement, we establish that their motion is superdiffusive, closely approximating ballistic movement (SI Appendix and SI Appendix, Fig. S13). This implies that, once set in motion, ice diatoms maintain their trajectory over longer timescales than would be expected under normal diffusive conditions.

The statistical distribution of ice diatoms’ motility aligns well with a model combining a gamma distribution and a chi-distribution tail, as shown in Fig. 4A:

$$\begin{array}{c}\hfill f\left(V\right)=c\frac{{\beta }^{\alpha }{V^{\alpha }}^{-1}{e}^{-\beta V}}{\mathrm{\Gamma }\left(\alpha \right)}+\left(1-c\right)\frac{2V^{k_s-1}{e}^{-V^2/2{\sigma }^2}}{2^{k_s/2}\mathrm{\Gamma }\left(k_s/2\right){\sigma }^{k_s}},\end{array}$$ [2]

where $\alpha$ and $\beta$ are the gamma-shape and scale parameters, $c$ is the weight coefficient, $\sigma$ sets the active-speed scale, and $k_s$ is the effective degree of freedom. The first term in Eq. 2 describes pause intervals, whereas the second term reduces to the two-dimensional (2D) Maxwell–Boltzmann speed distribution when $k_s=2$ (SI Appendix, Fig. S14). This limit provides a baseline for uncorrelated motion in a 2D plane, so any fitted value of $k_s\ne 2$ quantifies the additional motility modes that become engaged as temperature rises.

In the speed distribution, the second peak is primarily described by the chi distribution (SI Appendix and SI Appendix, Fig. S15). The spread of the motility distribution is characterized by the fitted $k_s$, which increases with temperature, reaching a plateau at around 12^°C, as shown in Fig. 4B.

Physically, $k_s$ measures how many independent motility modes are simultaneously engaged. The increase in $k_s$ indicates not just a speed variation, but a strategic diversification in movement patterns. Such diversification suggests that highly motile cells can rapidly colonize newly formed nutrient patches or brine channels, while slower cells can conserve energy, buffering populations during unfavorable periods. Interestingly, $k_s$ can be rescaled by the efficiency ratio of internal to external speeds, as shown in Fig. 4C. The rescaled result is a constant, which indicates that ice diatoms optimize their energy expenditure relative to the external conditions they encounter. As diatoms encounter less resistance, or as their internal mechanisms become more efficient, they can afford a greater diversity in their movement strategies with respect to energy expenditure. The interrelationship between temperature, motility diversity, and efficiency ratios reveals an adaptive strategy. It suggests that ice diatoms are not merely responding passively to temperature changes but are actively modulating their motility strategies in a way that maximizes their ecological success under varying thermal conditions.

Discussion and Outlook

The gliding motility of ice diatoms in sea ice suggests finely tuned adaptations to their cold, dynamic habitat, highlighted in two aspects: distinct interaction with icy environments and resilience to cold temperatures.

The unique ability of ice diatoms to glide on ice, enabling them to thrive in conditions that immobilize other marine diatoms, correlates with their particular ability to adhere to ice through mucilage. The ice adhesion ability of ice diatoms, revealed and quantified by our shear flow assays, could be potentially enabled by well-known ice-binding proteins (IBPs) in mucilage threads. IBPs have been widely identified in ice diatoms (98) to protect them from the cold by preventing recrystallization of ice (31–34). However, the role of IBPs in aiding ice diatom mobility has not been documented. Certain IBPs exhibited in psychrophilic bacteria, like MpIBP_RIV, have demonstrated the ability to facilitate adhesion to ice (99). Considering the hypothesis suggesting that IBPs in ice diatoms originate from psychrophilic bacteria (27, 100, 101), it is plausible that ice diatoms might utilize similar mechanisms to adhere to ice. Further investigations of chemical components in the mucilage that contribute to the unique “skating” ability of ice diatoms, potentially including IBPs, will deepen our understanding of the mechanism of adhesion in gliding motility on ice.

Additionally, the ice gliding motility suggests that the motor proteins of ice diatoms have adapted to low temperature conditions. A recent study has focused on identifying unique myosin motors in temperate diatoms (85). More broadly, myosin motor-based single molecule studies have enabled us to understand the entire catalytic cycle of molecular motors walking (69). These measurements have been mostly limited to room temperature studies and have not systematically explored colder regimes. Our results set a low-temperature record for the quantitative characterization of diatom gliding, a process understood to be dependent on the actin–myosin system (60, 85). Future work will involve the integration of molecular-level studies with thermodynamic modeling to dissect the molecular machinery for these remarkable organisms that retain motility at subzero temperatures. Specifically, molecular investigations will clarify which molecules contribute to cell propulsion and how their conformational changes and interactions depend on temperature.

Ice diatoms have also evolved excellent adaptations to secure enhanced motility at cold temperatures by showing a systematic shift of their temperature responsive motility toward cold. This study integrates internal molecular machinery and external hydrodynamic factors to dissect mechanisms controlling the gliding speed of diatoms. The staining of mucilage trails and the visualization of the actin within the raphe region in ice diatoms confirm the internal basis of force generation in ice diatom motility (60, 85). The observations of particle movements in the ventral region of gliding diatoms allow us to quantify the impact of viscoelastic drags, distinct from internal driving mechanisms on diatom motility. While our modeling analysis assumes a steady-state scenario, future work will incorporate time-dependent processes to account for elastic contributions.

The hydrodynamic effects not only lead to a loss of output speed from the myosin motor but also tune the instantaneous dynamic behavior of diatoms in response to the environment. The correlation between cell tilt angle and instantaneous speed (SI Appendix, Fig. S10) suggests that cell shape and orientation play important roles in influencing traction distribution and motility efficiency. Systematic studies exploring how variations in cell shape across different diatom species influence motility and traction distribution represent a valuable direction for future research.

Employing traction force microscopy (TFM), we demonstrate the interactions between gliding ice diatoms and their substrates in real time. The localized traction patches observed in ice Navicula sp. (with leading-edge traction as the primary active force generation site and trailing-edge traction primarily representing drag forces; see Fig. 3B, SI Appendix, Figs. S11 and S12 and Movie S7) distinguish diatom motility from crawling animal cells, whose numerous, rapidly remodeling focal adhesions generate a fluctuating, spatially extended field of traction forces (74, 102). Furthermore, in contrast to recent findings in apicomplexan motility (74), where self-organized continuous actin flows drive gliding, diatoms rely on stationary actin cables along which myosin motors intermittently transport discrete mucilage threads, leading to localized propulsion.

Field observations of sharply delimited biotic layers within sea-ice cores point to an active depth-selection strategy by motile ice diatoms (Fig. 1A and SI Appendix, Table S1). Illumination and nutrient availability are two prominent, though not exclusive, axes of this optimization (103, 104). Light intensity falls steeply with depth, creating a vertical gradient (105, 106). Conversely, nutrient concentrations generally increase toward brine channels that connect to underlying seawater (2, 104). Our finding shows that ice diatoms can glide directly on the ice matrix and present superdiffusive, nearly ballistic trajectories, giving them the capacity to traverse brine-channel networks rapidly. If, as documented for several temperate pennate diatoms (81, 107, 108), ice diatoms also possess phototactic or chemotactic responses, this motility would allow them to reposition into microniches where illumination and nutrient supply are simultaneously optimal. Resource tracking, however, is only part of the challenge for ice-dwelling diatoms. Brine channels form a tortuous three-dimensional lattice with pore diameters ranging from micrometers to millimeters (103); this architecture imposes physical constraints likely to shape diatom trajectories and attachment sites (Movie S2). Future investigations about how ice diatoms adapt and navigate inside such confined conduits will reveal the mechanical limits of their depth-selection behavior. Superimposed on these physical constraints are pronounced chemical fluctuations, most notably in salinity (103). Brine salinity can vary by orders of magnitude as temperature and melt regimes change (109, 110). Although the effects of these salinity shifts on ice diatom motility and adhesion have not yet been experimentally assessed, the dramatic salinity variation suggests plausible impacts, warranting systematic investigation. Future field measurements of depth-dependent light, nutrients, salinity, and brine channel geometry, coupled with controlled laboratory studies of phototaxis and salinity tolerance, are therefore required to explore this integrated framework of resource, chemical, and physical drivers of ice-diatom ecology and, by extension, polar ecosystem stability.

In conclusion, our study reveals a distinct aspect of how ice diatoms navigate sea ice, bridging ecology and cell physiology. Ice diatoms exhibit ice gliding motility, which is absent in temperate counterparts. The unique ice gliding ability allows ice diatoms to search for optimal light, salinity, and nutrient conditions within a dynamically changing landscape. The cold-adapted strategies predicted by our model provides insights for future research on how these psychrophilic microorganisms optimize motility under cold stress, including adaptations in motor protein conformations and rheological properties of mucilage. The observed temperature-dependent distributions in population motility reflect not just physical responses but strategic population adaptations. By linking movement statistics to biological energy expenditure and environmental interactions using a scaling analysis, we suggest that ice diatoms actively modulate energy utilization and movement strategies to optimize motility at the population level.

The finding of ice gliding motility and the insights into energy-efficient movement can be integrated into ecological models to better predict phytoplankton dynamics within sea ice and inform adaptation strategies in other extremophilic microorganisms. These findings also add a dynamic perspective to climate prediction models for polar ecosystems under environmental change (111).

Materials and Methods

Temperature-Controlled Fluorescent Microscopy Setup.

Our studies utilize a facility-grade widefield fluorescent microscope (Squid, Cephla), custom-built with temperature-controlled moduli, as shown in Fig. 5A. This portable system is ideal for field studies due to its adaptability and mobility. Detailed configurations of the Squid microscope can be found in ref. 112.

Fig. 5.

Experimental setups. (A) Subzero temperature-controlled fluorescent microscope equipped with a cooling stage and incubator. (B) Temperature feedback control of the cooling stage. (C) Setup for observing behavior of ice diatoms on ice surface. (D) Setup for quantifying the adhesion strength of ice diatoms on ice.

Two complementary systems permit both steady-state and rapid-shift experiments (Fig. 5A): i) a Peltier-conditioned chamber encloses the entire microscope for long-term transmitted-light imaging; ii) a Peltier cooling stage, mounted directly on the objective turret, drops the sample from 20 to $-20$ ^°C within a minute, enabling fast, fluorescence-based assays of temperature-responsive behavior. Diatoms are monitored via their autofluorescence excited at approximately 470 and 640 nm. Both temperature-controlled systems exploit the Peltier effect to absorb heat from the cold side and release it to the hot side (Fig. 5B). The chamber uses a fan-cooled module (Laird SAA-170-24-22). The stage employs a higher-capacity element (Same Sky CP39255074H-2) coupled to a liquid heat exchanger for superior hot-side dissipation, ensuring fast, sustained supercooling capacity (Fig. 5B). To achieve precise thermal regulation, a thermistor (Amphenol SC30F103A) attached to the cold face feeds a PID controller (8TCM-X107), which continuously adjusts drive current to hold the set temperature with minimal deviation.

Arctic Ice Cores Postprocessing in the Field.

At each Arctic ice station, 10 cm-diameter ice cores are cut into 13 cm sections. Segments bearing the biota layer are transferred to a large volume of 1 $\mathrm{\mu }$m-filtered seawater (1:5, $\mathrm{v}o{l}_{\mathit{ice}}$/$\mathrm{v}o{l}_{\mathit{seawater}}$) and held at 3 ^°C until the ice melted. After the diatoms settled, the diatom-rich sediment was harvested for further field experiments and for establishing laboratory cultures. Detailed culturing procedures are provided in SI Appendix.

Systems Designed for Investigating Ice Diatom Gliding Motility within Ice Environments.

For on-ice assays, a 3 mm layer of fresh water is frozen on the bottom of a Petri dish to create the ice substrate. Ice-cooled seawater containing suspended diatoms is gently overlaid, and the dish is maintained at 0 ^°C. After 30 to 60 min the cells settle at the ice–water interface and the system reaches thermal equilibrium; time-lapse imaging then begins (Fig. 5C). Diatom motility within solid ice is examined in custom ice microfluidic channels; fabrication and imaging details are provided in SI Appendix. Additionally, the temperature-controlled motility measurements are described in detail in SI Appendix.

Adhesion strength is quantified by the wall shear stress $\tau$ required to detach diatoms. Once cells have settled, seawater is pumped through the channel with a peristaltic pump (Kamoer DI-Pump). The corresponding volumetric flow rate is converted to $\tau$ using the Poiseuille solution for a rectangular duct. Glass assays use commercial µ-Slides (5 mm wide $\times$ 0.8 mm deep; ibidi). For ice assays, we fabricate 60 mm-long channels (4 mm wide $\times$ 8 mm deep) from polydimethylsiloxane (PDMS) and acrylic; the lower 4 mm are filled with water and frozen, creating a 4 mm-thick ice bottom wall and a 4 mm-deep flow passage above (Fig. 5D). Observations are made at the channel midpoint to minimize entrance and exit effects. Detailed adhesion measurements are provided in SI Appendix and SI Appendix, Fig. S6.

Full protocols for fluorescent and SEM imaging, raphe-particle streaming, particle-image velocimetry, and traction-force microscopy are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Discoveries of ice gliding diatoms in the Arctic Ocean.

Movie S2.

Ice diatoms gliding in ice microchannels.

Movie S3.

Ice and temperate diatoms gliding on glass vs. ice substrates.

Movie S4.

A gliding ice diatom interacts with substrate at tilted angles.

Movie S5.

Tracer particle streaming along the ventral raphe of ice diatoms.

Movie S6.

Tracer particle streaming along the dorsal raphe of ice diatoms.

Movie S7.

Traction forces generated by gliding ice diatoms.

Data, Materials, and Software Availability

The data that support findings of the manuscript are available on Dryad (https://doi.org/10.5061/dryad.3xsj3txt2) (113). All other data are included in the manuscript and/or supporting information.