Sediment gravity flows are the primary process by which sediment and organic carbon are transported from the continental margin to the deep ocean. Up to 40% of the total marine organic carbon pool is represented by cohesive extracellular polymeric substances (EPS) produced by microorganisms. The effect of these polymers on sediment gravity flows has not been investigated, despite the economic and societal importance of these flows. We present the first EPS concentrations measured in deep-sea sediment, combined with novel laboratory data that offer insights into the modulation of the dynamics of clay-laden, physically cohesive sediment gravity flows by biological cohesion. We show that EPS can profoundly affect the character, evolution, and runout of sediment gravity flows and are as prevalent in deep oceans as in shallow seas. Transitional and laminar plug flows are more susceptible to EPS-induced changes in flow properties than turbulent flows. At relatively low concentrations, EPS markedly decrease the head velocity and runout distance of transitional flows. This biological cohesion is greater, per unit weight, than the physical cohesion of cohesive clay and may exert a stronger control on flow behavior. These results significantly improve our understanding of the effects of an unrealized biological component of sediment gravity flows. The implications are wide ranging and may influence predictive models of sediment gravity flows and advance our understanding about the ways in which these flows transport and bury organic carbon globally.


Clay, inherently associated with organic matter, is the most abundant sediment type on Earth (Hillier, 1995). Recent advances in our understanding of the properties of clay have redefined our models of submarine sediment gravity flows (SGFs) in both modern and ancient environments (Wright and Friedrichs, 2006; Barker et al., 2008; Sumner et al., 2009). SGFs are volumetrically the most significant sediment transport process in the ocean, they can pose destructive hazards to offshore infrastructure, and their deposits form major hydrocarbon reservoirs (Talling, 2014). Understanding and prediction of SGF behavior therefore have scientific, economic, and social importance.

Clay-rich SGFs are governed by the ability of clay minerals to aggregate, or flocculate (McCave and Jones, 1988). Laboratory experiments of clay-rich SGFs show that at sufficiently high concentrations of clay, flocs bind to form a network that behaves as a gel, increasing viscosity and suppressing shear-generated turbulence (Baas and Best, 2002; Baas et al., 2009). These flows, and their deposits, are radically different to fully turbulent flows and highlight the importance of understanding how cohesive material affects SGFs.

Extracellular polymeric substances (EPS) are secreted by microorganisms in many environments, from rivers and estuaries to hypersaline systems and deep-sea hydrothermal vents (Decho and Gutierrez, 2017). The adhesion of this exopolymer to sediment grains forms a matrix of EPS, sediment, and single-cell organisms called a biofilm, which is considered to be the primary mechanism by which benthic microorganisms stabilize the sediment they inhabit (Tolhurst et al., 2002). Small concentrations of EPS (CEPS < 0.063% by weight, wt%), representing baseline levels in estuarine sediment, can increase the development time of bed forms exponentially (Malarkey et al., 2015). EPS research to date has focused on coastal environments. The presence of EPS in deep-sea sediment and their impact on SGFs is not known.

EPS represent up to 40% of the marine organic carbon pool (Falkowski et al., 1998). The burial of organic carbon (Corg) in marine sediment represents the second largest sink of atmospheric CO2 (Galy et al., 2007). However, global carbon budget studies typically estimate Corg flux by measuring the particulate organic matter settling through the water column and accumulating in marine sediment (Martin et al., 1987; Muller-Karger et al., 2005; Decho and Gutierrez, 2017). SGFs are not currently considered to be a significant Corg transport and burial process in Corg flux models on continental margins. This is surprising, since fine-grained sediment flows have been found to transport the majority of Corg (de Haas et al., 2002). Here, we present the first measurements of EPS from deep-sea sediment cores, and, through novel experiments, we test if EPS-derived biological cohesion is capable of inhibiting turbulence and intensifying cohesive SGF behavior. This has fundamental implications for (1) understanding SGF behavior in the natural environment, (2) models of Corg flux on continental margins and estimating Corg burial, as well as (3) reconstruction of past environments from ancient deposits and predicting hydrocarbon reservoir properties.


To explore the impact of biological cohesion on the dynamics of cohesive SGFs, series of experimental SGFs were generated with and without EPS. These flows were generated in a 5-m-long, 0.2-m-wide, and 0.5-m-deep, smooth-bottomed lock-exchange tank (Fig. DR1 in the GSA Data Repository1). The reservoir was filled with a mixture of kaolinite clay (volumetric concentration Cclay = 5–23 vol%), EPS, and seawater. Xanthan gum was used as a proxy for natural EPS (cf. Tan et al., 2014). Xanthan gum shares chemical similarities with a wide variety of EPS and is widely used as a substitute for EPS in marine ecology, soil science, and sediment stability research (see the Data Repository). The range of CEPS used in the experiments (0–0.268 wt%), was informed by seabed sediment cores obtained during RV Tangaroa cruise TAN1604, from 127 to 1872 m depths in the Hauraki Gulf, North Island of New Zealand (Figs. DR2 and DR3; Table DR1). These EPS data are based on bulk carbohydrate content, collected using the standard assay method of DuBois et al. (1956). The maximum CEPS recorded in the TAN1604 cores was 0.260 wt%, with an average value of 0.139 wt%. For comparison, background EPS content ranges from 0.01 to 0.1 wt% measured in estuarine sediment and from 0.1 to 0.67 wt% measured in freshwater sediment (Gerbersdorf et al., 2009; Malarkey et al., 2015).

In the laboratory, we compared the head velocity, Uh, and runout distance of clay-only control flows with equivalent flows containing EPS to test if biological cohesion intensifies cohesive flow behavior. Uh versus horizontal distance was measured using a high-definition video camera. Flow runout distances (maximum deposit extent from the lock gate) were recorded, except for flows that traveled the length of the tank (Table 1). To examine the impact of EPS on floc dynamics, samples were extracted for floc size and density analyses, using the LabSFLOC-2 (Laboratory Spectral Flocculation Characteristics (Manning et al., 2007) method (see the Data Repository). Each flow was classified visually following Hermidas et al. (2018).


The clay-only control flows generated turbidity currents at Cclay = 5–15 vol% and top transitional plug flows (TTPFs; Hermidas et al., 2018) at Cclay = 22–23 vol%, allowing us to examine the effects of EPS in fully turbulent flows and transitional flows experiencing turbulence suppression by physical cohesion (Table 1). Turbidity currents traveled the length of the tank and generated shear waves along their upper interface with the ambient fluid (Table 2). The TTPFs featured a dense, laminar lower “plug” layer with coherent fluid entrainment structures (Baker et al., 2017; Figs. DR4 and DR5) that transitioned upward into a dilute turbulent layer (Table 2).


Adding EPS to the turbidity currents with Cclay < 15 vol% produced no visual changes in flow behavior and no measurable differences in the Uh profiles (Figs. DR6 and DR7). At Cclay = 15 vol%, the EPS-laden flows (Table 1) were also visually indistinguishable from turbulent clay-only flow F07. However, the Uh of F08 and F09 decreased more rapidly than for F07 between 4 m and 4.2 m along the tank, resulting in lower Uh values at 4.6 m (Fig. 1). At the highest CEPS, F10 began to decelerate rapidly at 2.5 m and halted at 3.9 m.

TTPFs at Cclay = 22 vol% and Cclay = 23 vol% exhibited distinct decreases in Uh and runout distance as EPS were added (Fig. 2; Fig. DR8). Normalized to the maximum head velocity, Uh,max, and runout distance of the clay-only flows, the combined Cclay = 22 vol% and Cclay = 23 vol% data are strongly correlated with CEPS (Figs. DR9 and DR10). In Cclay = 22 vol% flows, CEPS ≤ 0.089 wt% still produced TTPFs with lower Uh,max and shorter runout distance than the control (Fig. DR8), whereas CEPS ≥ 0.133 wt% dramatically reduced upper boundary mixing, producing a distinct interface with the ambient fluid characteristic of plug flows (Fig. DR11; Hermidas et al., 2018). Flows F14 and F15 also lacked coherent fluid entrainment structures. These flows achieved Uh,max less than half of the Cclay = 22 vol% clay-only and clay-EPS TTPFs, and en masse freezing significantly reduced runout distance (Fig. DR8). At Cvol = 23 vol%, this change from TTPF to plug flow occurred at CEPS ≥ 0.087 wt%. At the highest CEPS of 0.259 wt% in F20, the slurry slid out of the reservoir for 0.6 m (Hampton et al., 1996).


Our experiments demonstrate that the strong biological cohesion imparted by EPS in the seabed extends to SGFs at the CEPS values found in deep-sea and coastal environments. Uh and runout distance were reduced by adding EPS to the TTPFs and the densest turbidity currents. At higher clay and EPS concentrations, the biological cohesion caused flow transformation, suggesting that EPS limit shear turbulence by increasing the strength of the bonds between clay flocs. These flows are likely to be non-Newtonian with a yield stress, as in aqueous xanthan gum solutions and xanthan-kaolinite mixtures (Song et al., 2006; Hamed and Belhadri, 2009). This biological cohesion was greater per unit weight than the physical cohesion. At Cclay = 22 vol%, the addition of 0.133 wt% EPS induced a flow transformation from TTPF to plug flow, substantially reducing the runout distance and Uh,max. A similar reduction in runout distance would require an increase in clay concentration from 22% to 25% in an EPS-free flow (Baker et al., 2017).

We hypothesize that EPS strengthen cohesion between flocs and assist in building a network of interconnected flocs (Leppard and Droppo, 2004). EPS in TTPFs are brought into contact with more clay particles than in turbidity currents, allowing the EPS to strengthen the particle network, further suppressing turbulence and encouraging laminar flow (Table 2).

To test this hypothesis, samples of fluid were taken from clay-only TTPF F16 and clay-EPS TTPF F17 for floc size and density comparison using LabSFLOC-2 (Fig. DR12). As each sample was released into the settling column, the gel underwent gravitational settling and broke into flocs. Only 10% of the flocs in F16 were larger than 200 μm, compared to 55% in F17. The clay-EPS flocs had a lower density than the clay-only flocs, and this difference increased up to ten times as floc size increased. The dominance of large, water-rich flocs in clay-EPS F17 implies that the clay-EPS gel was more resistant to separation into smaller flocs under shear during static settling. We further infer that within flows, these water-rich clay-EPS flocs are more resistant to shear turbulence than the clay-only flocs, and that this difference increases with increasing clay and EPS concentrations, resulting in more laminar flow behavior.


The influence of EPS was observed in experimental turbulent, transitional, and laminar SGFs, each with complex and variable flow properties. Scaling of these flows to natural prototypes is a nontrivial task, and standard methods (e.g., Froude, Reynolds, and Shields numbers, and distorted geometric scaling) are unlikely to be valid without modification (Iverson, 1997; Marr et al., 2001; Mulder and Alexander, 2001). However, the experimental SGFs may be used as analogues for natural SGFs based on fundamental physical principles. The experimental flows traveled at <0.55 m s−1; suitable analogues are hyperpycnal river discharge, and weak submarine SGFs triggered by earthquakes, which have velocities of 0.3–1.6 m s−1 (Talling et al., 2013), and natural SGFs that have decelerated to a similar velocity. The observed turbulence suppression is likely to apply to other flow types, such as wave- and current-driven sediment flows and fluid mud flows on shelves, which rely on turbulence to suspend particles (Traykovski et al., 2000). Turbulence modulation in higher-velocity flows, and in flows with higher turbulence intensity, is likely to occur at higher concentrations of EPS and clay than in these experiments.


The settling of particulate Corg to the seafloor is regarded as a key process in the global carbon cycle. However, measurements show that only 1%–4% of this material is buried within the seabed (Decho and Gutierrez, 2017). Large amounts of particulate Corg are delivered to continental shelves via rivers (0.8 PgC yr−1; Liu et al., 2010) and by in situ primary production (6.2 PgC yr−1; Chen, 2010). The majority is remineralized in the water column, and only a small amount is preserved on the shelf (0.2 PgC yr−1; de Haas et al., 2002; Chen, 2010; Liu et al., 2010). In contrast, continental slopes are important carbon sinks. Approximately 40%–50% of Corg transported from the shelf to the continental slope (0.5 PgC yr−1; Chen, 2010) is buried at water depths <500 m. The remaining 50%–60% is stored at >800 m (Muller-Karger et al., 2005).

The transport of Corg down continental slopes by SGFs has not received much attention in these calculations. However, large-scale SGFs on continental margins have the potential to move large quantities of terrestrial and marine Corg to the deep sea (e.g., Mountjoy et al., 2018) while also providing the rapid burial that promotes the preservation of Corg. The ability of EPS to alter SGF behavior, as shown here, implies that EPS change the spatial distribution of organic carbon sinks. Moreover, the increased cohesion of mixed clay-EPS flows promotes en masse deposition, which helps the burial and preservation of Corg by reducing oxygen exposure times (Burdige, 2007; Hedges and Keil, 1995). In the context of the global carbon cycle, a better understanding of the ability of EPS to influence the dynamics of its transport medium may improve numerical models of Corg fluxes on continental margins, particularly where they are dominated by fine-grained sediment input and high primary, and therefore EPS, production. Our deep-sea cores showed EPS levels similar to coastal environments, where EPS are widely recognized for their contribution to the Corg pool (Morelle et al., 2017). These results highlight that current global carbon budgets may underestimate Corg flux by not considering Corg transport via SGFs and the contribution of in situ EPS to the Corg sink in deep-sea sediment.


Increasing numbers of studies highlight the importance of considering physical cohesion when interpreting and modeling SGF processes, but the insights into biologically cohesive SGFs presented here indicate the need to recognize the potent effects of EPS in future studies. At the concentrations found in modern estuarine, coastal, and deep-marine environments, EPS have proven to be capable of changing flow behavior and reducing the deposit runout distance and Uh,max of cohesive flows. This research will improve our predictive models of these volumetrically significant and hazardous events. Further studies are needed to better constrain and quantify the effects of EPS in process models for cohesive and noncohesive SGFs, and their wider environmental relevance for geological history and the global carbon cycle.


The Australian Government Research Training Program Scholarship funded Craig’s Ph.D. candidature. An International Association of Sedimentologists Postgraduate Award Grant funded Craig’s visit to Bangor University (Bangor, UK). The UK Natural Environment Research Council grant NE/1027223/1 (COHBED project) enabled this research to be undertaken using the flume facility built by Rob Evans (Bangor University). We thank Brian Scannell, Edward Lockhart, and Connor McCarron for help in the laboratory. We thank the officers, crew, and scientific complement for core collection and processing on RV Tangaroa cruise TAN1604, with funding via the New Zealand Strategic Science Investment Fund to the National Institute of Water & Atmospheric Research’s National Centres of Coasts and Oceans and Climate, Atmosphere, and Hazards. Irvine Davidson (University of St. Andrews, UK) is thanked for his help in measuring extracellular polymeric substance concentrations in the Hauraki Gulf cores. We are also grateful for the constructive feedback from reviewers Peter Talling, Lawrence Amy, and Esther Sumner, which greatly improved this manuscript.

1GSA Data Repository item 2020028, description of laboratory methods, flume tank set-up, vertical profiles of EPS concentrations and locations of TAN1604 cores, images of experimental flows, head velocity profiles of 5, 10, and 22 vol% Cclay flows, scatter plots of normalized run-out distances and normalized Uh,max for 22 and 23 vol% Cclay flows, and scattor plot of floc size and settling velocity for 23 vol% clay-only and clay-EPS flows measured using the LabSFLOC-2 methodology, is available online at http://www.geosociety.org/datarepository/2020/, or on request from editing@geosociety.org.
Gold Open Access: This paper is published under the terms of the CC-BY license.