Bedform characteristics and biofilm community development interact to modify hyporheic exchange

.

. This dynamic ecotone (i.e. meeting point) act as a biogeochemical 'hot spot' creating an interface for the exchange between chemical solutes (Boulton et al., 1998;Gandy et al., 2007) and aquatic biota (Jones and Mulholland, 2000).
It is well established that hyporheic exchange influences mixing characteristics. This is governed by a range of environmental parameters including: flow rate, sediment composition and channel morphology, particularly bedform (Marion et al., 2002;Cardenas et al., 2008;Bottacin-Busolin et al., 2009;Boano et al., 2014;Clark et al., 2018). These factors have the capacity to impact the extent to which dissolved pollutants carried in water come into contact with degrader communities and thus, control their environmental fate (Scharper et al., 2018).
Increased anthropogenic activity including mining, urban and industrial development, dam construction and agriculture can disrupt the bio-hydrological connections within the hyporheic zone (Hancock et al., 2002). As such, it is vital that a wider understanding of the controls governing hyporheic exchange are investigated. This is necessary if strategies are to Aubeneau et al., 2016). However, there is a need to determine how bedform morphology (i.e. amplitude and wavelength) influences biofilm formation and the feedback on hyporheic exchange, which remains largely unknown.
Changes in the water-sediment exchange can interact with and modify water flow patterns (Nogaro et al., 2013;Battin et al., 2016). For example, previous research has demonstrated a reduction in interstitial water flow around benthic biofilms (Bottacin-Busolin et al., 2009;Orr et al., 2009;Larned et al., 2011), resulting in alterations in solute residence time distributions (Aubeneau et al., 2016). In addition, the growth of these benthic biofilms on sediments clogs the interstitial pore spaces within the sediment layers (Battin et al., 2008;Aubeneau et al., 2016). This can alter the hydraulic conductivity of the sediment, impeding water penetration and solute transfer within the hyporheic zone (Nogaro et al., 2010;Aubeneau et al., 2016).
In-situ studies have observed variability in biofilm growth in accordance with sediment composition, specifically particle size (Nogaro et al., 2010). Aubeneau et al. (2014) noted a 25% greater exchange rate in channels with coarse gravel particles (5 cm) versus those with finer sediment (0.5 cm). This is partly attributed to the smaller infiltration capacity of finer sediments. As such, nutrient penetration is reduced, restricting the layer and depth of surface biofilm growth (O'Connor & Harvey, 2008).
There have been a number of studies that have examined how the physical attributes (i.e. al., 2017). However, few studies have studied the mutual impact and interactions of bed morphology and biofilm on the hyporheic zone exchange (e.g. Bottacin-Busolin et al. 2014).
As such, a mechanistic understanding of how hyporheic exchange is regulated and controlled is underdeveloped due to the complexity of natural systems.
Here, we present an innovative approach to examine how bedform features such as dune wavelength and sediment size can interact with biofilm communities at both i) the sedimentwater interface and ii) bed profile depth, using an experimental flume system. Using fluorescent dye tracer experiments we then examine how this feeds back on hyporheic exchange and discuss the role of pumping in driving both biofilm growth and feedbacks on fluvial mixing. To our knowledge, the work presented here is the first attempt to explicitly consider the entire bed profile (i.e. sediment size, bedform, dune wavelength) in assessing both the physical and biological impacts on hyporheic exchange. This is of functional importance for environmental remediation and restoration strategies, with functioning streambed interfaces also aiding in wider catchment nutrient cycling and maintenance of healthy streambed communities .

Water extraction site
ml -1 ) was collected from Wellesbourne (52°11'51.7"N 1°36'09.3"W) and was first passed through a 200 μm mesh and then a 38 μm mesh to remove particulate matter. The water was then transferred to the flumes within 2 hours of initial collection.

Flume design
A custom-built recirculating flume experimental system was constructed enabling tracer studies to be conducted in a replicate system with identical physical and hydraulic characteristics. The experimental system consisted of 9 recirculating rectangular flume channels, comprised of 3 separate units each containing 3 flumes. Each flume was 1.98 m long, 0.1 m wide and 0.2 m deep ( Figure 1). The flumes were constructed from 0.01 m thick glass allowing for a clear visual observation of the experiment. To maintain the structural rigidity of the system an aluminium frame was constructed to hold the flumes in place. The frame had adjustable legs which aided in generating sufficient flow depth along the channel length for different bedform morphologies. The slopes placed on the flumes created a gravity chute down which water could flow. The water flowed over a weir, positioned downstream, which could be adjusted accordingly to achieve uniform flow. The water then proceeded down a pipe and through a flow meter where a Grundfos UPS15 recirculating pump then pumped the water through additional piping back into the flume system ( Figure 1). Average flow depth and bed height was fixed at 0.03 m and 0.1 m respectively, with a discharge of 12 L min -1 to ensure sufficient water mixing.

Range of experimental flow conditions
J o u r n a l P r e -p r o o f

Journal Pre-proof
In the experiments we systematically varied bed-profiles. Experiments were run over a maximum of 20 days with data collected every 2 days. Three profiles were chosen: i) flat, ii) undulating with two dunes ( = 1 m), iii) undulating with multiple dunes ( = 0.2 m) ( Figure   2a). Two sizes of smooth synthetic glass beads (VWR International) were applied for each of the bedding profiles. The glass beads may have affected the penetrance of light into the sediment and thus, permitted greater biofilm depth penetration. However, the choice to use glass beads minimised sorption as a factor, as the synthetic glass beads were smooth and impermeable, adding uniformity to the bed sediment. The mean particle sizes chosen for investigation were 5 mm and 0.5 mm to represent gravel and coarse sand environmental conditions, respectively. In summary a total of six bedding profiles were investigated.
Each set of bed profiles was allocated its own bank of three flumes enabling each experiment to be run in triplicate. Experiments were first conducted using the 0.5 mm particle size, after which the flumes were cleaned with ethanol before bed profiles were created with the 5 mm particle size.
The artificial substrates were poured into the bed section of the recirculating flume to an average bed height of 0.1 m, with a maximum dune height of 0.11 m from peak to base ( Figure 2a). Templates were constructed for each of the bed profiles and sediment weight used in each system was logged.

Analytical method to determine hyporheic exchange
J o u r n a l P r e -p r o o f Hyporheic exchange in natural systems is controlled by a complex array of both biological, chemical and physical processes. It can be modelled using an effective dispersion coefficient (D) which encompasses these complex interactions and can be formulaically displayed as, where: β is the sediment diffusion correction term, D b is the biodiffusivity (L 2 t -1 ; i.e. biologically mediated diffusive transport of solutes and solids in sediments), D d is the dispersion coefficient (encompassing turbulent diffusion and pumping, L 2 t -1 ) and D m is the molecular diffusion coefficient (L 2 t -1 ), all of which can be obtained empirically or through modelling (Berg et al., 1998).
O'Connor & Harvey (2008) developed a succession of scaling relationships that related effective diffusion coefficients to a variety of fluid flow and sediment characteristics, a concept originally proposed by Richardson & Parr (1988). A measured value of effective diffusion over a sediment can be determined indirectly using a conservative tracer (e.g.
Rhodamine) in a closed system (i.e. laboratory flume). In this case the concentration (C) is initially 0 in the bed sediment. The concentration in the overlying water column (C wc ) is equal to an initial concentration (C 0 ). As the water recirculates, hyporheic exchange will drive the tracer into the sediment interstitial pore space resulting in a decrease in concentration in C wc over time. As such, the slope of a temporal concentration profile (i.e. C wc versus t 1/2 ) can be used to calculate D, briefly D = (

Biofilm growth
J o u r n a l P r e -p r o o f

Journal Pre-proof
Biofilm growth was measured i) temporally on the sediment surface every 48 hours for 20 days and ii) within the bed profile (i.e. every 1 cm in 10 cm depth) after 20 days. The flume units were placed in a controlled environment room, light was used to stimulate the photosynthetic growth of algae and cyanobacteria, and therefore accelerate biofilm growth within the interstitial pore spaces. Fluorescent daylight bulbs were covered with Lee DS 226 UV filter screens to prevent transmission of light wavelengths < 380 nm. The lights were set on a timer to provide 16 hours of light and 8 hours of dark. The collected river water acted as an inoculum to permit microbial growth in the system.
Modifications to the experimental design were made to permit the quantification of in-situ microbial growth within each unit ( Figure 2b). This included the addition of 2.5 cm 2 length glass sides on the surface to act an extractable synthetic surface (Kowalczyk et al., 2013) from which to extract and quantify surface biofilm ( Figure 2b). The slides were placed in triplicate along the length of the flume starting at t = 0, spaced at 20 cm intervals along the sediment bed. During the experiment three slides were extracted every 48 hours from the same locations on each experimental unit to asses biofilm growth rate on the surface of the sediment bed. Biofilm was resuspended and carbohydrate was quantified using a phenol-H 2 SO 4 assay (Kowalczyk et al. 2016, andfurther described in Dubois et al. 1959) as a proxy for biomass.
The sediment bed was also cored in order to quantify biomass growth within the bed profile at depth. Due to the destructive sampling involved, coring was only undertaken at the end of the tracer experiment (i.e. after 20 days, to a depth of 10 cm). Sediment cores were taken every 20 cm (Figure 2b), in triplicate, along the sediment bed using perspex corers (length 10 cm; diameter 2 cm). Once removed the cores were frozen and dissected into sections of 1 cm J o u r n a l P r e -p r o o f intervals using a scalpel. Each 1 cm segment was placed into a 25 ml Falcon tube with 5 ml of water and vortexed to suspend the biofilm. Following biofilm extraction, the biofilm biomass was analysed for carbohydrate content in the same manner as described above.

Statistical analysis 2.6.1 Biofilm growth
We applied a multiple linear regression (MLR) approach to test whether sediment size (two levels factor: 5 mm and 0.5 mm, explanatory variable) and bedform type (three level factor: flat, two-dune, multi-dune, explanatory variable), control the biofilm biomass (response variable, number of samples (n) = 540) within the bed sediment at the final timepoint (20 days). Subsequently, we also tested under the same approach the effect of sediment size (explanatory variable) and bedform type (explanatory variable) on the surface biofilm biomass over time (response variable, n = 198). In this instance, we also included time (continuous covariant) in the model as a controlling variable for the non-independence of the repeated observations over time.
The generated MLR model coefficients were used to calculate the modelled relationship between factors and compared to the observed values. All output results from the regression models are provided in the Supplementary Material (Tables S1-S2; Figure S3), alongside the model metric outcomes (standardized beta values, t statistic = t, degrees of freedom = df, adjusted r 2 and p value) and the number of samples (N).

Hyporheic exchange
J o u r n a l P r e -p r o o f

Journal Pre-proof
The initial slope of the concentration profile (i.e. C wc versus t 1/2 ) referred to hereafter as the 'exchange gradient' was used as a metric of hyporheic exchange. We selected biofilm, sediment size and bedform as the likely controlling factors governing the hyporheic exchange. A MLR model was applied to the dataset to produce a unique model integrating biofilm biomass (explanatory variable), sediment size (explanatory variable) and bedform (explanatory variable) as predictors for hyporheic exchange (response variable; n = 198; of freedom = df, adjusted r 2 and p value) and the number of samples (N).
The performance of all the models were assessed using the following metrics: a) The coefficient of determination (R 2 ) for the regression between measured and modelled biofilm amounts/ hyporheic exchange.
b) The Nash-Sutcliffe efficiency (Nash and Sutcliffe, 1970), a measure of goodness of fit between the modelled and measured data i.e.: where 'measured' is the measured data, 'modelled' is the modelled amount estimated using the various predictors and 'mean' is the mean measured data. The closer the NSE is to 1 the J o u r n a l P r e -p r o o f stronger the model fit. A value of 0 or lower indicates that the model performs no better than the mean of the data (Nash and Sutcliffe, 1970).
Statistical analysis (including: descriptive statistics, alongside multi-linear and linear regression) was performed using Graphpad Prism v8 (GraphPad Software, Inc., www.graphpad.com) and SPSS v25 (SPSS Inc., Chicago, Ill., USA). The threshold for statistical significance was set at a confidence level of 95% but greater significance was noted. Normality and homogeneity of variance was tested on the residuals using the Shapiro-Wilk test and Levene's tests, repectively.

Biofilm growth
Microbial growth was measured on the surface and at depth across all experiment flume systems (Figure 3 to 6). Biofilm growth gradually increased over the course of all experiments ( Figure 3) and was concentrated on the sediment surfaces, irrespective of sediment size or bedform (Figure 4 to 6).
The maximum biofilm biomass was 105.1  6.1 and 115.1  6.0 μg carbohydrate equivalent mm -3 , respectively for the 5 mm and 0.5 mm sized particles, within the flat bed system, stabilising around day 10. The two-dune ( = 1 m) system reached peak biofilm growth, across both particle sizes, by approximately day 10 ( Figure 3) at which peak biofilm growth plateaued. Biofilm growth was slower within the multiple-dune system ( = 0.2 m) across both sediment sizes but reached the highest levels from day 12 (Figure 3). Biomass reached a J o u r n a l P r e -p r o o f peak of 129.1  6.4 and 106.1  5.8 μg carbohydrate equivalent mm -3 , respectively for the 5 mm and 0.5 mm sized particles for the two-dune system and 131.2  7.2 and 149.4  8.7 μg carbohydrate equivalent mm -3 within the multi dune system.
After 20 days there was a trend for higher levels of surface biofilm growth on the 0.5 mm compared to the 5 mm sediment particle size across all systems, and biomass reduced with depth across all experimental systems (Figure 4 to 6). In the flat bed system there was a gradual reduction within the 5 mm sediment systems, with only a 40% reduction in biomass at a depth of 6 cm ( Figure 4). Conversely, the 0.5 mm sediment systems displayed a rapid reduction in biomass with depth, with a 42% reduction in biomass observed after a depth of only 2 cm (Figure 4). By contrast, two-dune system generally did not exhibit sudden drops in the maximum biofilm growth until approximately 3 cm in depth ( Figure 5), with biomass levels in the multi-dune beds typically only dropping rapidly after a depth of around 4 cm ( Figure 6). Biofilm depth infiltration extent varied depending on the dune sample location from where the core was extracted (i.e. up-slope, slope peak etc; Figures 5 and 6). Using the generated model coefficients (Tables S1-2, Supplementary Material) biofilm biomass was plotted as a function of sediment size and bedform across time ( Figure 7a) and J o u r n a l P r e -p r o o f with depth ( Figure 7b). As expected, both the modelled and measured data displayed a significant correlation between biofilm biomass growth over time ( Figure 7a) and with depth ( Figure 7b), across all sediment sizes. Days explained 71% to 79% of the biofilm data across the 0.5 mm and 5 mm sediment systems (Figure 7 a), respectively, with depth having a greater control in biomass growth within the 5 mm sediment systems, explaining over 70% of the biofilm data, in contrast to 64% in the 0.5 mm sediment systems (Figure 7b). Both modelled temporal and depth biomass data correlated strongly to the measured data producing r 2 values of 0.76 and 0.68, respectively (Table 1; Figure S3; Supplementary Material). This corresponded to NSE metrics of 0.74 and 0.90 for the depth and temporal data sets, respectively (Table 1).

Hyporheic exchange
To elucidate whether the growth of microbial communities on the flume bedforms impacted hyporheic exchange, the trace data was normalised to permit the identification of temporal changes in exchange gradients. A MLR analysis and model was constructed using sediment size, bedform and biomass as predictors for hyporheic exchange over time. The MLR analysis releveled that 70% of the variance in hyporheic exchange was controlled by a combination of these three factors. Using the model coefficients hyporheic exchange was modelled across all experimental systems. There was a weak but significant correlation between the modelled and measured hyporheic exchange data (Table 1; Figure S4, Supplementary Material). However, the low overall NSE metric, 0.38, suggested a poor overall fit between the modelled and measured hyporheic exchange data, using sediment size, bedform and biomass as predictors.

Journal Pre-proof
A linear regression between biomass and hyporheic exchange (Figure 8a) revealed there was a strong correlation between these two parameters within the two dune (r 2 = 0.60, p<0.0001, df = 31) and flat dune (r 2 = 0.45, p <0.0001, df =31) 0.5 mm sediment systems, with no significant relationship between biofilm and hyporheic exchange reported in any of the 5 mm sediment systems (Figure 8b). Here, increased bedform complexity resulted in a larger hyporheic exchange, regardless of the biofilm present. This trend is emulated within the multi dune 0.5 mm sediment system, which also displayed consistently higher hyporheic exchange gradients than the flat or two dune systems when biomass was below 75 g mm -3 (Figure 8a).
A predictive model for hyporheic exchange, within the flat and two-dune 0.5 mm sediment systems, was built using biofilm alone (Table 1; Figure S4, Supplementary Material). This revealed a moderate goodness of fit (r 2 = 0.54; p <0.0001; df = 64, NSE 0.60) between the measured and modelled hyporheic exchange data sets.

Surface biofilm growth
Utilising river water as inoculum in the experimental systems provided an insight into the process of microbial community growth in sediment beds. However, Harvey et al., (2019) demonstrate that to achieve this there must be a balanced ratio between hyporheic residence to reaction time. Otherwise an imbalance will cause the reactants to be used up promoting the storage of biochemically inactive water (Harvey et al., 2019).

Biofilm growth with depth
In general, higher amounts of biofilm developed on the smaller experimental substrate (0.  , 2016). However, in general, the 5 mm particle size permitted greater autotrophic biofilm depth penetration across all bedforms. This trend is supported by observations from several in-situ based river studies (Nogaro et al. 2010;Parker et al. 2018). Depth, particle size and bedform were strong predictors for biofilm growth in the sediment bed. This is likely attributable to porosity which is known to influence the exchange of flows and energy in fluvial systems (Sawyer & Cardenas, 2009;Nogaro et al., 2010). The larger pore spaces within the 5 mm system will permit more water and light to penetrate the entirety of the sediment bed, as well as maintain aerobic conditions (Elliott & Brooks, 1997), which could also contribute to a more intensive exchange of matter. As such, the porous structure may allow the biofilm matrix to spread deeper into the sediment bed (Packman et al., 2004) as well as encourage interstitial habitat colonisation.
By the same logic the restricted interstitial pore space in the 0.5 mm sediment systems will inhibit effective diffusion into the sediment bed, confining nutrients and biofilm growth to the sediment surface layers (Nogaro et al., 2010). This is reflected in the rapid reduction in surface biomass, for all 0.5 mm particle size test studies, in response to sediment core depth.
The feedback from microbial growth will promote rapid bioclogging of the interstitial pore spaces creating impermeable bedforms. This feature has been shown to alter flow dynamics with Haggerty et al. (2014) reporting an increase in transient storage times by a factor of 4 in response to biofilm clogging.
dropping rapidly after a depth of around 4 cm. The presence of dunes increases hydraulic roughness by protruding into the flow (Paarlberg et al. 2009). High pressure regions develop upstream of the obstruction with low pressure forming downstream, creating a pressure gradient, which pumps through the obstruction and increases flow connectivity (Tonina & Buffington, 2007). As such, the greater number of bedforms within the dune systems may be creating a strong pumping mechanism, permitting nutrients to be propelled further into the sediment bed, supporting biofilm growth. This could also explain why biofilm growth on the sediment surface was slower within the multi-dune systems.
Pumping may be enabling the biofilm matrix to overcome the physical constraints of the 0.5 mm sediment. As such, biofilm is likely controlled by the light and nutrient deficit experienced with increased sediment depth. This could explain why biomass levels tended to be greater on the up and peak slope locations within the multi-dune and two-dune systems, where the force of pumping is likely greater. Equally, the lower pressure experienced after the dune structure would explain why the down-slope and trough sections tended to exhibit lower biomass concentrations with increasing sediment depth. From an ecological perspective, these biological drivers emphasise the importance of considering not only the magnitude of the hyporheic exchange flux but also the flow pathways within the hyporheic zone, which play an equally crucial role in modulating biogeochemical river processes .

Hyporheic exchange mediating factors
Impact of microbial growth on the hyporheic exchange was quantified by repeat discrete dye injections into the flume systems. Hyporheic exchange was controlled by a combination of sediment size, bedform and biofilm, with sediment size the dominant controlling factor. This is visually supported by the faster rate of exchange in systems with a larger sediment particle size (i.e. 5 mm) and in systems with bedforms present. These findings are in-keeping with the J o u r n a l P r e -p r o o f theoretical background regarding sediment size and bedform structures (Elliott and Brookes, 1997;O'Connor and Harvey, 2008;Magliozzi et al. 2018). Within the 0.5 mm sediment systems biofilm was the dominant controlling factor, and predictor, of hyporheic exchange within flat and two dune environments. In these sediment systems, exchange gradients became progressively slower with increased biofilm. This implies that the microbial growth and associated bioclogging, over time, impeded the rate of hyporheic exchange. This is reinforced by Aubeneau et al. (2016) noting decreases in conservative solute transport in response to biofilm growth.
Conversely, the multi-dune system exhibited the opposite trend to the 0.5 mm sediment flat and two-dune systems with no significant change in exchange rates identified in relation to biofilm growth. This implies that microbial growth had limited impact on the movement of water through the sediment, despite increased biofilm overtime. Similarly, exchange rates were consistent across all 5 mm sediment experimental series, despite changes in biofilm growth. This suggests that the 5 mm sediments used in the test system are too large to permit sufficient quantities of microbial growth on the surface to impede effective diffusion. This is supported by greater biofilm growth at depth within the 5 mm particle systems. This could also explain why the model to predict hyporheic exchange as a function of biofilm, sediment size and bedform performed poorly within the 0.5 mm multi-dune system and all 5 mm experimental systems. The integration of more intermediate sediment sizes within the experimental framework would help in modelling the predictive forces of bedform and sediment size in response to hyporheic exchange.
Here, bedform was identified as the major controlling factor determining exchange rates, with a greater number of dunes contributing to higher rates of hyporheic exchange. This, coupled with the consistent levels of exchange reported in the multi-dune 0.5 mm sediment J o u r n a l P r e -p r o o f system adds support to the concept of pumping. Pumping will become 'activated' from the increased presence of bedforms, reflected in more consistent and higher exchange rates in the multi-dune system, irrespective of sediment size. This mechanism was sufficient to force the Rhodamine tracer through fine particles and thus, overcome the i) small interstitial pore spaces and ii) excess clogging of the small interstitial pore spaces overtime. Greater advective mass transfer between the water column and stable sediment bed will create a positive cascade effect within the system. Faster water exchange will shorten residence times and lower microbial growth. This is supported by the slower biofilm growth rates within the multi-dune system. This will, in turn, reduce interstitial pore space clogging and further increase the rate of exchange between the water and sediment column. In a river environment these mechanisms can work to increase material transport ( It is commonly recognised that hyporheic flow depends on high and low permeability regions within the stream bed (Sawyer et al., 2009;Packman et al., 2002). Laboratory recirculating flume experiments have identified differences in the flow pathways over permeable versus impermeable beds of the same topography (Cooper et al., 2017). In general, turbulence over permeable beds is less intense resulting in greater energy flow and momentum transport. The force generated from pumping enhances the permeability potential of the sediment. As such, J o u r n a l P r e -p r o o f energy levels within the dune environments will likely be higher further contributing to higher rates of material transport across the sediment-water interface.
However, it is also important to note that biofilm growth is cyclically associated with changes to morphology and physiological states (Hall-Stoodley & Stoodley, 2002)  This study uses biomass as a surrogate for total biofilm growth, ignoring the structural and functional taxonomic diversity that makes up the biofilm matrix. This paper identifies a potential reciprocal interaction of hydrodynamics within the physical structure and biofilm growth in the pore space. This interaction could have significant repercussion on the biotransformation of pollutants and nutrient recycling. Research by Scheidweiler et al. (2003) explores this further, providing evidence that biofilms are able to differentiate and actively remodel their matrix, in porous environments, in order to exploit their surrounding space.
This can increase their carrying capacity and function as hyporheic bioreactors. This is further supported by Battin et al. (2003) noting the functional ability of biofilms to increase hydrodynamic transient storage of stream water and retention of suspended particles, feeding back into biochemical process changes. Subsequently, further analysis into the succession of biofilms both over time and depth would make a natural extension to the current investigation. Furthermore, given the impact of bedform and its link with advective pumping, J o u r n a l P r e -p r o o f Journal Pre-proof further investigations should seek to explore how bedform height also impacts hyporheic exchange.

Conclusion
This research highlights the important and interconnected feedback mechanisms involved in both driving and sustaining water-sediment interactions. Biological changes in the flume systems exerted a strong control on the conservative transport of our solute, supported by the fact that no modifications were made to the flume systems during each experiment series and microbial communities were freely allowed to proliferate. The extent of biological control was strongly influenced by the environment's physical parameters including bedform structure and particle size. These physical controls, in turn, feedback into the biological system and can work to dampen or intensify the biological influence on solute transport.
Bedform structures have the potential to overcome the flow limited effects of biofilm clogging which is enhanced in systems with larger sediment sizes. This finding is of functional importance for both river management and stream restoration design but also water quality. More bedform features and greater hyporheic exchange rates could aid in driving water purification, nutrient recycling and altering streambed community organisation.
As such, understanding more about the coupling between microbial and hyporheic flow represents an important frontier of research for ecosystem scientists. Table 1. Summary of goodness of fit metrics for the models. R 2 values and slope of regression lines between modelled and measured biofilm and hyporheic exchange data for the model approaches, along with Nash-Sutcliffe efficiency (NSE) values. n = number of samples.