Hydrodynamically‐Driven Deposition of Mud in River Systems

The riverine transport and deposition of mud is the primary agent of landscape construction and evolution in many fluvial and coastal environments. Previous efforts exploring this process have raised uncertainty regarding the effects of hydrodynamic and chemical controls on the transport and deposition of mud, and thus the constructions of muddy coastal and upstream environments. As such, direct measurements are necessary to constrain the deposition of mud by river systems. Here, we combine laboratory evidence and a field investigation in the Mississippi River delta to explore the controls on the riverine transport and deposition of mud. We show that the flocculation of mud, with floc diameters greater than 10 μm, in freshwater is a ubiquitous phenomenon, causing the sedimentation of mud to be driven by changes in local hydrodynamics, and thus providing an explanation for how river systems construct landscapes through the deposition of mud in both coastal and upstream environments.


Introduction
The transport and deposition of riverine mud is the primary agent of landscape construction and evolution across the majority of Earth's surface (Dyer, 1989;FAO, 1988;Kemper & Rosenau, 1984;Lamb et al., 2020;Young & Southard, 1978).Fluvial systems are distribution networks that transport cohesive sediment to construct floodplains, deltas, and estuaries found across the Earth's surface and in the sedimentary record on Mars.While muddy landscapes may not be entirely composed of mud, even small amounts of cohesive sediment may increase the bulk strength of otherwise non-cohesive sand, thus exerting a controlling influence on landscape morphology and morphodynamics (Dunne & Jerolmack, 2018, 2020;Julian & Torres, 2006;Kothyari & Jain, 2008).As such, constraining the transport and deposition of mud is vital to improve our understanding of landscape evolution.
Canonically, flocculation of mud occurs in deltaic and estuarine environments where the presence of cations in marine water neutralizes the repellent surface charge on clay minerals, allowing for particles to aggregate (Hunter & Liss, 1982;Mietta et al., 2009;Sutherland et al., 2015;Whitehouse, 2000).Additionally, various field and laboratory investigations have demonstrated that flocculation will occur in the presence of a high concentration of organics, particularly in response to the introduction of long-chain organic polymers such as Xanthum gum (Abolfazli & Strom, 2023;Deng et al., 2019Deng et al., , 2023;;Gregory, 1978;Mietta et al., 2009;Zeichner et al., 2021).However, a recent compilation of suspended sediment vertical concentration profiles has suggested that flocs form across a range of geochemical conditions in freshwater environments (Lamb et al., 2020;Nghiem et al., 2022).Given such a range of environmental conditions, and in combination with the presence of deposited mud upstream, flocs likely form across the full range of environmental conditions.
To date, few studies have directly measured the in-situ size and characteristics of flocs transported in rivers, and, as such, our understanding of the controls on floc size, how flocs are transported, and how fluvial systems deposit mud and construct muddy landscapes remains relatively unconstrained (Lamb et al., 2020;Nghiem et al., 2022;Osborn et al., 2021).Through a combination of laboratory experiments and a field investigation, we provide direct evidence that the baseline state for cohesive sediment transported in water is flocculated, which in turn facilitates mud settling out from suspension under standard turbulent environmental conditions, thus creating the muddy channel beds and floodplains observed on Earth and Mars (Lapôtre et al., 2019;Matsubara et al., 2015).

Particle Settling and Transport
Sediment transport is typically described by three categories: bedload, suspended load, and washload (Church, 2006;Rouse, 1937).Bedload is the trundling of grains along the bed, more typical in gravel-bedded rivers or a few specific sandy environments (Church, 2006;Devauchelle et al., 2011;Dunne & Jerolmack, 2018).Grains transported as suspended load make lengthy hops and are supported by the flow for long periods of time.In contrast, washload is the consistent suspension of particles by the flow that never settle out from the flow, and thus do not engage with bed material.These three sediment transport phase spaces are delineated by the Rouse Number (P), which contrasts the grain settling velocity (w s ) with the fluid shear velocity (u * ) of the flow as: where κ is von Kármán's constant (Von Kármán, 1930), usually taken to be κ = 0.41 (García, 2008;Nezu & Nakagawa, 1993;Nezu & Rodi, 1986).Transport states categorize grains for which P < 0.8 as washload, 0.8 ≤ P ≤ 2.5 as suspended load, and P > 2.5 as bedload.The key difference between suspended load and washload is that sediment transported as suspended load will temporarily deposit and exchange with the channel bed, as such P = 0.8 can be defined as the threshold of bed engagement, P B (Figure 1b).For transported sediment of Rouse number P ≥ P B , the bed over which sediment is transported will have a composition that reflects the composition of the transported sediment.
To test this supposition, we present an analysis of the distribution of shear velocities exerted on floodplains by overbank flows (Figure 1c).Using instantaneous flow data from USGS gaging stations (n = 55), we examine the cumulative distribution function of instantaneous flow data for rivers for which the presence of mud in the floodplains/river banks can be inferred from the combination of their bed grain size and bankfull shear stress (Dunne & Jerolmack, 2018, 2020).The bankfull shear velocity, calculated from the channel geometry using the depth-slope product, was subtracted from the distribution of fluid shear velocities and filtered for shear velocity values in excess of bankfull.This provides a coarse estimate of the u * values exerted on the floodplain by the overbank flow during floods.
Assuming an average clay particle size of 2 μm and an average floc size of 100 μm, and thus average floc settling velocity of 1 mm s 1 (Fall et al., 2021;Lamb et al., 2020;Strom & Keyvani, 2011), we find that under no flow conditions does P for unflocculated particles exceed P B , and thus unflocculated particles occupy solely the washload phase space during overbank flow.As such, overbank flow cannot deposit unflocculated clay particles.
In contrast, the CDFs of silt and flocculated particles passes through the full range of Rouse phase spaces, and as such it is more probable for flocculated cohesive sediment to be deposited and able to construct the observed/ inferred muddy floodplains.While this analysis is indeed a rough estimate of overbank shear velocities (e.g., changes in hydraulic roughness between floodplains and channels) and uses an assumed floc settling velocity that can be either increased or decreased by the presence of organic matter, silt, and sand (Fall et al., 2021;A. J. Manning et al., 2010;Tran & Strom, 2017), it provides an indication that for landscapes composed of fine-grained sediment across Earth and Mars, settling of cohesive sediment must be wide-spread for a range of environmental conditions.As such, settling cannot be solely driven by the chemical changes in the river water across the transition from fresh to saline environments.

Laboratory Evidence of Mud Flocculation in the Absence of Added Salts
A suite of experiments was conducted in laboratory mixing tanks to explore the response of suspended mud floc size distributions to changes in salinity, type and quantity of organics present, and the composition of inorganic clays and silts.Abolfazli and Strom (2023) have highlighted that the response of suspended mud floc size distributions in laboratory mixing tanks to an increase in salinity is dependent on the type and quantity of organics present and the composition of the inorganic clays and silts.Yet, all laboratory mixing-tank experiments on clay minerals and natural mud consistently point to the prevalence of flocs in the absence of marine salinity.For example, pure kaolinite has been found to flocculate better in pure deionized water than it does in deionized water with the addition of NaCl (Figures 2a and 2b), similar to the observation of Abolfazli and Strom (2023) and Partheniades (2009).While flocs composed of pure inorganic sediment are typically smaller than those observed with natural mud, the manyfold increase in size from constituent grains (D 50 = 5-10 μm) to floc size (D 50 = 45 μm) under the simple combination of just clay minerals and deionized water is notable.
This finding is in agreement with other studies on the sedimentation of cohesive sediment that showed negligible change in the zeta potential (and thus colloidal stability) of clay particles in saline water as salinity was increased after the addition of salt to the solution (Chassagne et al., 2009;Seiphoori et al., 2021).Many studies have shown that salt can, at times, interact with sediment and organic matter to change the size distribution of suspended mud flocs (Abolfazli et al., 2024;Abolfazli & Strom, 2022, 2023;Deng et al., 2023;Mietta, 2010;Verney et al., 2009).Yet, the point we are making with these examples is that mud can and typically does exist in a flocculated state without the presence of marine water.It is, therefore, likely that mud normally exists in a flocculated state in riverine environments as suggested by Droppo and Ongley (1994) and shown in Osborn et al. (2023) for portions of the Mississippi River.
Furthermore, as part of the study presented in this paper, samples of natural bed sediment and stream water were put into the same laboratory mixing tanks from both the lower reaches of the Mississippi River (near Venice, LA) and its headwaters (i.e., Stroubles Creek near Blacksburg, VA) all produced flocs of substantial size without the addition of salt (Figures 2c and 2d); see Abolfazli (2023) for experimental methods.The ambient salinity in both freshwater test cases was less than 0.1 PSU, contained a mixture of clay mineral types and organic material, and had flocs of substantial size (e.g., the D 50 of images C and D are 146 and 173 μm, respectively).

Field Investigation
Given the discrepancy between the canonical need for saline water to initiate the flocculation of mud, and both laboratory evidence and recent data compilations (Chassagne et al., 2009;Lamb et al., 2020;Seiphoori et al., 2021;Zeichner et al., 2021) suggesting that mud flocs form naturally in freshwater river systems, direct measurements of floc sizes across a range of natural environments is required.The spatially dynamic physical and chemical nature of the Fluvial to Marine Transition Zone (FMTZ), the region over which a river transitions from a freshwater to a saline environment, presents an ideal opportunity to explore the controls on the size and settling rate of mud flocs carried in suspension by the river.
We conducted a field investigation in the FMTZ of the Mississippi River, starting in the main channel (MC) of the Mississippi River near the Bonnet Carré Spillway (station ID: BC, 200 km upstream from Head of Passes [HOP]), and proceeded down the MC to near Venice, LA (station ID: MC, 20 km upstream from HOP), and to the ocean through the South Pass (SP) channel (station IDs: SP1-10, 2.5-20 km downstream from HOP) (Figure 3a).This survey was conducted in late June 2020 under a discharge range of 16,310-25,683 m 3 s 1 .At each site, we deployed the FlocARAZI (Osborn et al., 2021) (Figure 3a inset), a conductivity, temperature, and depth sensor, a P6 suspended sediment sampler (Nittrouer et al., 2011), and a Shippek grab sampler.These devices allowed us to measure floc size, water salinity, suspended sediment concentration, and bed grain-size composition from the channel thalweg.Using the FlocARAZI, we see the depth-averaged distribution of floc sizes at each station and note an abrupt shift in the floc-size distribution around station SP5B (Figure 3b).Suspended sediment samples were collected using a P6 sampler and were later dosed with sodium hexametaphosphate, sonicated, and analyzed using a LISST-Portable XR (Pomázi & Baranya, 2020) to determine the grain size of the constituent particles of the observed flocs (Figure 3b).The decrease in floc size coincides with both an abrupt increase in salinity of the river water and in the mud content, determined through proportion of silt and clay by mass of the bed sediment (Figures 3c and 3d).
Measurements of channel depth were made using a LOWRANCE fish finder, and water surface slopes were calculated using river discharge following the formulation developed by Nittrouer et al. (2011).We determine u * via the depth-slope product.This method is particularly useful during periods of high discharge (Dong et al., 2019).Based on the measured floc size, constituent grain size, and assumed density of 2,650 kg m 3 for the constituent grains, and floc fractal dimension η f = 2.5, we calculate the floc settling velocity following the formulation of Strom and Keyvani (2011) for the flocs observed at each station.From the floc w s and u * values, calculated P for the floc-size distribution at each station shows that under the high discharge conditions during data collection, the observed floc-size distribution overwhelmingly occupies the washload Rouse phase space (P < 0.8) (Figure 4a).However, we note a gradual change in the size of the flocs observed in suspension between the MC and SP.
Given SP's downstream position relative to MC and the general similarity between the floc-size distribution at MC and SP1-5, we infer that the MC supplies its floc-size distribution to SP.Given the decrease in floc size in the lower half of SP, the fate of the larger flocs from the MC is in question.Using the floc-size distribution from station MC and the shear velocity values calculated at each station, we calculate the Rouse number for the MC floc-size distribution as it traverses SP (Figure 4b).In this case, the Rouse number of the coarser portion of the floc-size distribution, illustrated by the D 75 and D 84 , exceed P B .This shift across the P B boundary at SP3 corresponds to a general shift in bed-sediment composition from nearly pure sand to 50% or greater silt-clay content by mass.While there is indeed variability in bed composition downstream from SP3, likely due to seasonal variation in fluid shear stress and rate of deposited mud erosion (Galler & Allison, 2008), the correlation between the Rouse number exceedance of P B for the coarse fraction of the MC flocs with the shift in bed composition from sand to mud is indicative of hydrodynamically-driven deposition of flocs on the river bed.
An identical survey was attempted in January 2021, however, the SP locations were unreachable due to channel navigability issues.Measurements of floc-size distributions in the MC at station MC were made under comparable discharge conditions to the summer (18,293 m 3 s 1 ).Water temperature was 6.3°C in the winter versus 27.9°C in the summer.Results show seasonality could have a profound effect on the depth-averaged distribution of the median floc size: D 50 was 37% larger, and variance of floc D 50 being 128% greater, during the than winter.

Discussion
Our study shows through in-situ measurements that flocculation in both field and laboratory environments is a ubiquitous phenomenon that greatly increases the ability of cohesive sediment to settle out from suspension.The tendency for clay minerals in water to flocculate, regardless of the presence of dissolved salts or organic matter, offers a potential explanation for how rivers transport and deposit cohesive sediment.This leads to the development of cohesive floodplains, even upstream of marine water intrusions or in the absence of organic material, such as is understood to-date on Mars.Our field results show a consistent state of flocculation for mud transported by the Mississippi River.A trend of decreasing floc size downstream toward the marine outlet, from a median floc size of approximately 100-50 μm, corresponds to a decrease in shear velocity, and thus ability of the flow to maintain floc suspension.Observed increases in salinity moving down the pass, from 0 to 7 PSU, did not result in a corresponding increase in suspended floc size.While this finding may appear to be in contrast to previous assertions that flocculation initiates or intensifies as rivers enter saline, coastal environments, previously observed changes in floc size within this region could also be due to a range of other physical and biochemical influences within the coastal zone.
The onset of bed interaction with the exceedance of P B leads to the accumulation of mud on the bed.The diminishing capacity of the flow to maintain a washload transport state (P < 0.8) for the floc-size distribution due to decreasing u * allows for the coarse fraction of the floc-size distribution to exceed the P B and interact with the bed through either ephemeral or perennial deposition under flow conditions that would otherwise prohibit the deposition of unflocculated clay particles.Comparisons of floc-size distributions between summer and winter months demonstrates a strong effect on the depth-averaged median floc-size distribution, despite comparable discharge and shear conditions.We attribute this 37% increase in the median floc size and 128% increase in the variance of the floc size distribution to increased biological activity due to warmer water temperatures in summer months leading to enhanced production and availability of organic matter, which has been previously demonstrated to result in the formation of larger flocs (Eisma, 1986;Verney et al., 2009;Zeichner et al., 2021).Seasonal variations in floc size, driven by changing levels of biological activity, and fluid shear stress, driven by changes in discharge and/or form drag, likely result in spatially variable deposition of mud on the channel bed.
In the regions of the study area without deposition of mud on the bed, our observed median floc sizes were approximately 100 μm.This is consistent with the range of back-calculated and observed floc sizes in previous studies of flocs in a variety of other freshwater environments and laboratory conditions (Abolfazli & Strom, 2023;Lamb et al., 2020;Lawrence et al., 2022;Nghiem et al., 2022;Osborn et al., 2023;Soulsby et al., 2013;Zeichner et al., 2021).We find that, despite the range of floc sizes observed in the field between stations, the D 50 of the floc constituent particles remained constant at approximately 14 μm.This indicates that the observed suspended sediment behaviors are driven by changes in the floc-size distribution and not by changes in the constituent grain size.Previous studies of Mississippi River suspended sediment composition have shown that the fine portion of the suspended sediment load (≤62.5 μm) is comprised of a mixture of various silt sizes and clay mineralogies (Johns & Grim, 1958;Johnson & Kelley, 1984).Additionally, previous laboratory investigations have demonstrated that, while flocs do contain the silt fraction of the suspended sediment load, floc size is unaffected by silt content and driven by clay particle interactions (Tran & Strom, 2017).
For all analyzed suspended sediment samples, we find particles that occupy the clay-size range (≤2 μm) comprise, on average, less than 5% of the cumulative volume distribution of the constituent particles (see Dunne et al., 2024), despite the sonication and dispersant treatment that the flocs were subjected to.Given that approximately 95% of the cumulative volume concentration of the treated sediment is outside the size range of clay, and that field evidence demonstrates that the sampled suspended sediment was indeed flocculated at the time of sampling, we propose that the average constituent grain size of 14 μm is indicative of the presence of hyperstable clay aggregates within the flocs that even vigorous sonication and the use of chemical dispersants were not able to disaggregate.This is consistent with previous work that has suggested that mud flocs should not be thought of as structures comprised of individual cohesive grains with uniform density and stability, but rather sizedependent path of a hierarchy of aggregate stability where larger, more fragile flocs are comprised of smaller, more stable aggregates (Fall et al., 2021;Krone, 1963;Michaels & Bolger, 1962;Soulsby et al., 2013;Van Leussen, 1988).
Data suggest that order 1 μm diameter clay particles, and possibly organic materials (Fall et al., 2021), appear to be able to form hyper-stable order 10 μm diameter stable aggregates, which in turn appear to form semi-stable order 100 μm diameter aggregate.This hierarchy of stability offers a possible explanation for the reactive nature of floc size to environmental forcings, including turbulent shear or biological activity, despite the ubiquitous tendency of clay in water to flocculate.Bonds between the particles that comprise small, hyper-stable aggregates are strong.However, bonds between the hyper-stable aggregates are relatively weaker and thus more likely to form or break under various environmental conditions, resulting in the macroscopic floc size being driven by external environmental forcings.Further work will be necessary to explore the controls on the yield strength of bonds between aggregates of varying sizes and bonds of various origin.
Our findings demonstrate through field and laboratory studies that the flocculation of clay in water is a ubiquitous phenomenon that allows for the deposition of cohesive sediment by river systems both on the bed of river channels and during overbank flows.While floc size does appear to be enhanced by biological activity and/or the Geophysical Research Letters 10.1029/2023GL107174 presence of organic matter, the first order control on the deposition of mud by river systems, and thus the construction of muddy channels and floodplains found across a range of environments on Earth and Mars, is changes in local hydrodynamic stresses and the capacity of a flow to transport flocculated mud as either washload or suspended load.

Figure 1 .
Figure 1.(a) Schematic of the flocculation process.(b) Schematic comparing washload and suspended load transport.(c) Cumulative Distribution Function (CDF) of Rouse Number of 100 μm flocs during overbank flows on floodplains.Line color is indicative of particle size.Background color is indicative of transport type phase space.

Figure 3 .
Figure 3. (a) Map of station locations and Head of Passes (HOP) within the Mississippi River Delta (image source: Planet).Inset image: FlocARAZI used to image flocs.(b) Particle size distribution at each station.Line color is indicative of particle size percentile within the distribution.(c) Salinity at each station.(d) Bed sediment % silt and clay by mass at each station.

Figure 4 .
Figure 4. (a) Rouse number (P) of floc-size distribution observed at each station.Line color is indicative of particle-size percentile within the distribution.Background color indicates transport type phase space.(b) Rouse number (P) of main channel (MC) floc-size distribution at each station.Line color is indicative of particle size percentile within the distribution.Point color is indicative of % silt and clay by mass in the bed sediment at each station.Background color is indicative of transport-type phase space.(c) Comparison on depth-averaged median floc-size distribution at station MC between summer and winter months.