Advection Not Dispersion and Transient Storage Controls Streambed Nutrient and Greenhouse Gas Concentrations

Streams and rivers are globally important in the carbon and nitrogen cycles due to high carbon and nitrogen turnover rates and contribute disproportionately to global greenhouse gas (GHG) emissions relative to their areal coverage. The hyporheic zone may be a hotspot of biogeochemical reactivity within fluvial ecosystems resulting in high rates of nutrient attenuation and associated GHG production. Controls on streambed nutrient cycling and particularly GHG production remain insufficiently understood. In this study, porewater concentrations of nutrients (NH4+, NO3-, NO2-) and GHGes (CO2, CH4, N2O) were measured alongside surface water breakthrough curves (BTC) of conservative (uranine) and reactive tracers [resazurin (raz)-resorufin (rru)] to provide insights into often assumed correlations between in-stream advective transport and transient storage metrics, and streambed biogeochemistry. Streambed biogeochemical concentrations were significantly correlated with advective transport time but not with dispersion and transient storage. The effect of advective transport time varied between chemical species, with NH4+, CO2, and CH4 exhibiting positive correlations and NO3-, NO2-, and N2O displaying negative correlations with advective transport time and vice versa for long-term storage. These findings increase knowledge of the relationship between hydrological drivers and streambed chemistry, potentially highlighting areas of the streambed expected to have elevated nutrients and GHGs. This improved understanding may allow chemical species to be effectively targeted by morphological restoration, which will aid in effective pollution and climate remediation.


INTRODUCTION
Streams and rivers play an important role in global biogeochemical cycling, processing large proportions of the carbon and nitrogen entering them annually (Seitzinger et al., 2006;Cole et al., 2007;Trimmer et al., 2012). Streams can perform important ecosystem functions, reducing nitrogen pollution through nutrient attenuation, and improving ecosystem health, water quality, and stream habitats (Triska et al., 1993;McMahon and Böhlke, 1996;Brunke and Gonser, 1997).
Within streams and rivers the majority of metabolic activity and biogeochemical cycling occurs within the sediment, resulting in "hotspots" of nutrient attenuation and GHG production in these areas (McClain et al., 2003;Lautz and Fanelli, 2008;Trimmer et al., 2012). Mixing of groundwater and surface water can create conditions for increased reactivity due to substrate supply and increased residence times, thus increasing the time available for reactions to occur (Findlay, 1995;Boulton et al., 1998;McClain et al., 2003;Lautz and Fanelli, 2008;Pinay et al., 2009;Boano et al., 2010;Krause et al., 2011;Zarnetske et al., 2011). Solutes in surface water typically enter and leave the streambed through hyporheic exchange flowpaths resulting in transient storage within the sediments (Bencala and Walters, 1983;Bencala et al., 2011).
Similarly, "hotspots" and "hot moments" of reactivity may occur in surface flow due to changes in advective transport times and transient storage increasing residence times of water and its chemical constituents. Advective transport times, dispersion and transient storage can, therefore, be important controls on biogeochemical cycling, within both surface and sub-surface flows (Mulholland et al., 1997;Boulton et al., 1998;Gooseff et al., 2005;Pinay et al., 2009;Zarnetske et al., 2011) and are controlled by local hydrology and geomorphology (Harvey and Bencala, 1993;Brunke and Gonser, 1997;Jin et al., 2009;Schmadel et al., 2016).
Conservative tracers (e.g., chloride, uranine) are commonly used to determine advective transport times, dispersion and transient storage indices in streams and rivers, which enable these parameters to be compared between streams, sub-reaches, and differing geomorphologies (Harvey and Wagner, 2000;Gooseff et al., 2005;Mason et al., 2012;Ward et al., 2013a,b;Schmadel et al., 2016). Reactive tracers [e.g., resazurin (raz), nitrate] may be used to investigate specific processes such as aerobic respiration and nutrient uptake (Haggerty et al., 2009;Zarnetske et al., 2011;González-pinzón et al., 2012;Harvey et al., 2013;, providing additional information on reactive solute transport. Despite the known importance of streambed biogeochemical cycling, understanding of controls on biogeochemistry in these environments remains insufficiently understood. Here, we investigate the influence of advective transport times, dispersion and transient storage, and aerobic microbial respiration on streambed chemistry within an agricultural stream. We aim to determine the relationship between hydrological metrics and aerobic respiration, and streambed chemistry and if the relationships (positive or negative) vary with the concentration of chemical constituents. We hypothesize that advective transport, dispersion and transient storage, and ecosystem respiration drive variations in chemical concentrations within the streambed and that the direction of the relationship will vary depending on chemical species.

Study Site
Experiments were conducted over four seasons (July 2016, October 2016, February 2017, March 2017 in the Wood Brook, a small, lowland stream at the Birmingham Institute of Forest Research, Staffordshire, UK (https://www.birmingham.ac. uk/research/activity/bifor/index.aspx). Each sampling campaign included a solute tracer experiment and streambed porewater sampling. The catchment upstream of the monitoring location is dominated by arable land (predominantly potato and winter wheat crops; Blaen et al., 2017a) with patches of deciduous woodland (Figure 1). The experimental reach consisted of a 1 km section starting in arable land, before flowing along the border between arable land and a patch of deciduous woodland, and into the woodland itself. In channel sediment varied between the upstream section, which contains sand-dominated sediment and the downstream section, which contains gravel-dominated sediment (indicated by the black bar in Figure 1). The regional groundwater aquifer is Permo-Triassic sandstone, overlain with 10 m thick glacial till deposits with 15-60 cm of sandy clay sediment on top of those (Blaen et al., 2017a). The average channel width was 2 m and average water depth was 13.9 cm. A continuous monitoring station was located ∼2 m downstream of site D (Figure 1) and measured a suite of parameters including river level (converted to discharge using a stage-discharge rating curve), specific conductance, temperature, dissolved oxygen, pH, dissolved organic carbon, and NO − 3 (Supplementary Figure 1). The stream discharge measured at the continuous monitoring station ranged from 5 L s −1 in October to 40 L s −1 in February. Further details on how parameters were measured, including the stage-discharge rating curve, are provided in Blaen et al. (2017a).

In-stream Tracer Injections
A slug injection of both conservative (uranine) and reactive (resazurin) tracers was conducted seasonally 250 m upstream of the first fluorometer (180 m upstream in July) (Blaen et al., 2018). Tracers were dissolved in stream water immediately before injection so that a tracer-in-solution method was used (Moore, 2005). 1.2 g of uranine and 8 g of resazurin were injected in October, February, and March, and 2 g of uranine and 8 g of resazurin were injected in July. Quantities were chosen to ensure tracer detection by in-stream fluorometers (GGUN-FL30, Albillia Sàrl, Switzerland) (Lemke et al., 2013). In July, uranine concentrations at site A were higher than the fluorometer detector can analyse, therefore, the mass of uranine used was lower in the other seasons compared to July. Tracers were released at a distance upstream of the study reach long enough to ensure full mixing in the water column. The resazurin-resorufin "smart" tracer system has been widely used in streams over the past decade as a proxy for aerobic respiration (Haggerty et al., 2008(Haggerty et al., , 2009. Resazurin is irreversibly reduced to resorufin when used as a terminal electron acceptor by aerobic microorganisms, as both resazurin and resorufin are fluorescent dyes, their concentrations are easily determined through fluorometry (Haggerty et al., 2008). Subsequently, the rate of transformation of resazurin to resorufin measured through stream tracer experiments provides an estimate of whole-stream aerobic respiration (Gonzálezpinzón et al., 2014).
Tracer injections were performed in late afternoon or early evening to minimize the possibility of tracer photo-degradation (Cai and Stark, 1997;Haggerty et al., 2008), the location of the stream within woodland also minimized photo-degradation through shading. Breakthrough curves (BTC) of uranine, resazurin, and resorufin were measured simultaneously at four sites A, B, C, and D (Supplementary Figures 2-4), creating three sub-reaches AB sand (sand-dominated sediment), BC sand (sanddominated sediment), and CD gravel (gravel-dominated sediment) within the study reach (Figure 1 and Table 1). Breakthrough curves were measured in-situ using on-line field fluorometers (GGUN-FL30) capable of measuring uranine, resazurin, and resorufin simultaneously at 10 s resolution (Lemke et al., 2013), resulting in higher temporal resolution BTCs and increased information on BTC tailing. Different wavelengths were used to excite uranine, resorufin, and resazurin (470, 525, and 570 nm) so they were detected simultaneously with minimal interference from each other (Lemke et al., 2013). Fluorometer calibration (including cross-calibration) was performed in-situ immediately before injection to allow calibration with the stream water of the study site. The fluorescence intensity of uranine, resazurin, and resorufin is affected by pH below 7.5 (Lemke et al., 2013), the pH ranged between 7.4 and 8.6 during our experiments with only 4% of measurements below 7.5 (Supplementary Figure 1).
The resulting BTCs were corrected for variations in temperature and turbidity (Blaen et al., 2017b), data was selected from the relevant time period of the BTC, and was corrected for any drift in the tracer signal between the beginning and end of the BTC. The drift correction was applied by averaging the baseline concentration prior to the arrival of the BTC and correcting the BTC so that the baseline after the end of the tail had passed was equal to this initial baseline concentration. Where the BTC was clearly being detected (the arrival of the BTC could be seen) but the background concentration was below zero, the background was corrected to zero, likewise background concentrations above The quantity column indicates the counts of each feature. Also presented are the average slope, width, depth, and length of the experimental reaches, the dominant sediment type present in each reach, and the diameter that 90% of particles are smaller than (µm) from grain size analysis of sediment sieved at 2 mm.
zero were corrected down to zero as uranine, resorufin, and resazurin do not naturally occur in streams. Some background concentrations were negative because they were either below the limit of detection or because the calibration curve was not perfect close to zero.

Calculation of Solute Transport and Storage Parameters
We analyzed conservative BTCs through the determination of transport, and short-and long-term storage metrics following (Ward et al., 2013a;Schmadel et al., 2016) due to the high uncertainty and susceptibility to equifinality of parametric approaches to stream solute BTC analysis (e.g., Wagner and Harvey, 1997;Lees et al., 2000;Mason et al., 2012;Ward et al., 2013a). Details of the metrics and how they were calculated are given in Table 2, t w and TSI were normalized by dividing by t ad and the first temporal moment was normalized by the recovered transit time distribution (Schmadel et al., 2016). The window of detection represents the maximum tracer recovery time and provides the boundary between short-term and longterm storage (Harvey and Bencala, 1993;Harvey et al., 1996;Wagner and Harvey, 1997;Harvey and Wagner, 2000;Ward et al., 2013b), allowing determination of the maximum length of short-term storage processes captured by each tracer experiment. Due to some localized increases in storage, especially at site C, CV, and γ values could often not be computed as this led to negative values of central moments; therefore, CV and γ were used here instead. The final hydrological metrics used for the analysis in this study are highlighted in bold in Table 2 and are M 1 , t ad , t w,norm , TSI norm , CV, γ, and M loss . The rate of transformation of resazurin-to-resorufin was used as a proxy for integrated whole-stream aerobic respiration. Following Knapp and Cirpka (2018) the zeroth temporal moments of resazurin and resorufin were normalized by those of uranine and the apparent transformation rate of resazurin to resorufin was calculated from the normalized zeroth temporal moments. The zeroth temporal moment is calculated from the integrals of the BTC (Knapp and Cirpka, 2018).

Porewater Sampling
Porewater samples were collected seasonally from three sites (Figure 1), shortly prior to or following solute tracer experiments to minimize interference to both porewater samples and BTCs while ensuring similar conditions were present. Porewater samples were collected at 10 and 20 cm depth from multilevel mini-piezometers (Rivett et al., 2008;Krause et al., 2013), with sites 1 and 2 containing four multilevel mini-piezometers and site 3 containing seven multilevel mini-piezometers ( Figure 1B). Sampling locations were chosen across bedforms so that a range of upwelling and downwelling sites were included in the dataset, resulting in vertical hydraulic gradients that were positive and negative (−0.24 to 0.28, Supplementary Table 1). Vertical hydraulic gradients varied with discharge between seasons resulting in some measurement locations switching from upwelling to downwelling sites across seasons. Water samples were filtered through 0.45 then 0.22 µm Nylon syringe filters (Thames Resteck, UK) into sterile centrifuge tubes and subsequently frozen until analysis.
Greenhouse gas samples were produced in the field from porewater samples using the headspace equilibrium method (McAuliffe, 1971). A headspace of 14 ml ultrapure helium was introduced to a syringe containing 7 ml of porewater sample and shaken for 2 min to equilibrate gases between the headspace and the water. The gas sample in the headspace was injected into a pre-evacuated exetainer and stored at room temperature, in the dark, until analysis. Further details can be found in Comer-Warner et al. (2019, 2020.

Laboratory Analysis
Concentrations of NH + 4 , NO − 3 , and NO − 2 in the porewater samples were determined using a continuous flow analyser (Skalar San++) and concentrations of CO 2 , CH 4 , and N 2 O were measured using two gas chromatographs (GCs) (Agilent 7890A). Henry's constant was then used to convert the concentration in the gas sample (hence the headspace) to the concentration in the porewater sample (Wilhelm et al., 1977;Hudson, 2004). Details on analysis, accuracy, precision, and limit of detection for the above analyses can be found in Comer-Warner et al. (2019, 2020. Concentrations across all piezometers and depths for each site and season were averaged to give a site-and season-dependent concentration for each chemical species.

Statistical Analysis
All statistical analyses were conducted using the R V 3.3 (R Development Core Team, 2011). Spearman's rank correlation coefficients were calculated to assess relationships between variables in the dataset (Harrell Miscellaneous package; Harrell and Dupont, 2020), due to the large number of comparisons made p-values were controlled for Type I errors using the False Discovery Rate method (Benjamini and Hochberg, 1995). The spearman test was preferred to a pearson test as this is non-parametric and allows non-linear relationships to also be explored. The exploration of non-linear relationships is crucial as the relationship between hydrology and streambed chemistry is not necessarily expected to be linear. A principal component analysis was performed using the Rattle GUI R software and package to further explore how variables related to each other. Prior to the ordination data were log-transformed, centered around their means, and values standardized to the standard deviation of the dataset (i.e., converted to Z-scores). To assess differences in streambed chemistry between study reaches and sample dates a MANOVA (Adonis function-Vegan package; Oksanen et al., 2020) was conducted. The MANOVA method partitions distance (in this case Euclidean distance) among explanatory variables and uses a permutation procedure (999 permutations) based on pseudo-F ratios.

Comparisons of Hydrological Metrics and Streambed Chemistry
The focus of this study was the relationships between hydrology and streambed chemistry, therefore, spatial and temporal variability of hydrological metrics, resazurin-resorufin turnover, and streambed chemistry are shown in Figure 2 and Supplementary Figure 5, but are not considered in detail here. CO 2 , CH 4 , and NH + 4 concentrations were generally highest in the sand-dominated sites, whereas NO − 3 , NO − 2 , and N 2 O concentrations were generally highest in the gravel-dominated site ( Table 3). Seasonal variations were low with no significant differences observed with respect to CO 2 and NO − 2 , whereas, NH + 4 , N 2 O, and CH 4 concentrations were highest in Autumn and NO − 3 concentrations were highest in Winter (Table 3) Streambed CO 2 , CH 4 , and NH + 4 showed a positive correlation with M 1 and t ad , which was significant except between CO 2 and t ad (r > 0.64, adjusted p-values between <0.01 and 0.04, Figure 3 and Supplementary Figures 7-9). Streambed CO 2 , CH 4 , and NH + 4 were negatively related to M loss (r between −0.42 and −0.68, Figure 3 and Supplementary Figures 7-9), which was only significant between CO 2 and M loss (adjusted p-value = 0.04, Supplementary Figures 7-9). Streambed CO 2 , CH 4 , and NH + 4 were not significantly correlated with t w,norm , TSI norm , CV, or γ (r between −0.07 and 0.41, adjusted p-values >0.05, Figure 3 and Supplementary Figures 7-9).
Principal components 1 and 2 described ∼80% of the variation in the measured parameters. The relationships between the variables were consistent with the results of the correlation analysis. The distribution of reach observations in PCA space (components 1 and 2) indicated differences between reaches; in particular, CD gravel was separated from AB sand and BC sand ( Figure 4A). However, observations overlapped between seasons, therefore, the structure of the data is spatially and not seasonally 3 | The mean and one standard deviation of streambed nutrient and greenhouse gas concentrations across sites and seasons.

DISCUSSION
Streambed chemical concentrations in the agricultural, lowland stream studied here were related, in some cases, to advective transport ( M 1 and t ad ) and long-term storage ( M loss ) but not dispersion and transient storage ( t w,norm , TSI norm , CV, or γ). The direction and significance of the relationships varied between chemical species with CO 2 , CH 4 , and NH + 4 positively related to advective transport time and NO − 3 , NO − 2 , and N 2 O negatively related. Hence, the relationship between reach-scale hydrology, and streambed chemistry may not be as strong as previously thought, with transient storage having less influence than advective transport time through a stream reach.
The lack of a relationship between CO 2 concentration and resazurin-resorufin turnover was unexpected given the wide use of resazurin as a proxy for aerobic respiration (Haggerty et al., 2009;González-pinzón et al., 2012;. Here, CO 2 concentration was measured rather than production or concentration change, which may explain the lack of relationship observed. Additionally, sediment CO 2 concentration and instream respiration may not be closely related and resazurinresorufin may only serve as a proxy for in-stream respiration, not streambed respiration, if hyporheic exchange is low. Further investigation is required to explore the causes of the lack of relationship as there could be wider implications for the use of resazurin to quantify metabolically active transient storage and reach-scale aerobic respiration, especially because hyporheic respiration can account for a high proportion of total stream respiration (Battin et al., 2003a,b). The resazurin-resorufin system, therefore, may not accurately quantify hyporheic zone respiration, especially with short windows of detection resulting from the use of instantaneous tracer injections. Significant limitations in the ability of resazurin-to-resorufin transformation to reproduce respiration rates calculated from diel-oxygen profiles have also been observed (Knapp and Cirpka, 2018), which provides further evidence for caution when using the resazurin-resorufin tracer system.

Comparisons of Hydrological Metrics and Streambed Chemistry
The positive correlation between CO 2 , CH 4 , and NH + 4 indicates that concentrations of these chemical species were high within the same areas of the streambed, likely due to the breakdown of organic matter and anaerobic conditions within the streambed (Baker et al., 2000;Duff and Triska, 2000). These conditions decrease the rate of nitrification, therefore, less NH + 4 is converted to NO − 3 and conditions are suitable for denitrification to reduce NO − 3 , explaining why the negative correlation between these two nutrients was not significant. The positive correlation between NO − 3 , NO − 2 , and N 2 O likely reflects their involvement in similar reaction processes e.g., nitrification and denitrification (Kendall, 1998;Duff and Triska, 2000;Trimmer et al., 2006). A significant correlation was expected but was only observed between NO − 3 and N 2 O.
Findings here support previous relationships between GHG fluxes and nutrients that were significant and positive between CO 2 and NH + 4 , CH 4 and NH + 4 , and N 2 O and NO − 2 , and significant and negative between CO 2 and NO − 3 , CH 4 and NO − 3 , and N 2 O and NH + 4 (Vidon and Serchan, 2016b), indicating nutrient controls on streambed GHGs. These correlations can also be explained by flowpath length or reaction times with higher residence times found at longer flowpath lengths and times typically resulting in increased concentrations of CO 2 , CH 4 , and NH + 4 and decreased NO − 3 concentrations. Higher concentrations of CO 2 , CH 4 , and NH + 4 , at longer advective transport times is likely due to more time for biogeochemical reactions to occur with these species accumulating in the streambed (Pinay et al., 2009;Zarnetske et al., 2011;Boodoo et al., 2017). Conversely, lower concentrations of NO − 3 , NO − 2 , and N 2 O, at longer advective transport times illustrates the conversion of these products to other intermediate or final products due to increased time for biogeochemical reactions to occur (Pinay et al., 2009;Zarnetske et al., 2011;Quick et al., 2016;Hampton et al., 2020). Previous work has shown streambed GHG concentrations are controlled by river hydrological stage and hyporheic exchange (Hampton et al., 2020;Villa et al., 2020). Contrasting influence of flowpaths and downwelling/upwelling zones indicate that CH 4 increases along flowpaths and is higher in anoxic, upwelling zones whereas N 2 O has been found to vary with flowpath or remain unchanged along flowpaths (Hlaváčová et al., 2005;Reeder et al., 2018).
The negative relationship observed between NO − 3 and transport time may reflect the dominance of nitrification over denitrification at shorter residence times that has previously been documented (Zarnetske et al., 2011;, although this was not significant. These results may support findings that hyporheic zone solute transport is driven by advection and increased uptake of N occurs at higher travel times (Bukaveckas, 2007;Vidon and Serchan, 2016a), although a direct comparison is not possible as concentrations not reaction rates were measured here and streambed residence times were not measured. Conversely, no significant relationship between advective transport and ecosystem respiration is also possible (Blaen et al., 2018). A non-linear relationship between NO − 3 , NO − 2 , and N 2 O and advective transport time was expected as these chemical species are produced and then consumed with increasing residence times (Burgin et al., 2013;Marzadri et al., 2014;Quick et al., 2016;Hampton et al., 2020).
A significant relationship between the chemical species and dispersion and transient storage was not observed. As streambed chemical species were measured these were expected to be directly linked to dispersion and transient storage as exchange of surface water into the streambed increases biogeochemical reactivity and residence time affects C and N concentrations (Hall et al., 2002;McClain et al., 2003;Thomas et al., 2003;Faulkner and Campana, 2007;Lautz and Fanelli, 2008). The relationship between dispersion and transient storage, and nitrogen and carbon cycling is complex with varying relationships previously observed. Dispersion and transient storage have been found to both control and have no effect on N and C cycling (Jones et al., 1995;Hall et al., 2002;Bernot et al., 2010;Baker et al., 2012). The varying relationship may be due to dispersion exerting only secondary control on solute transport in the hyporheic zone . In some cases M loss was negative and may result from fluorometer locations, i.e., stream reaches, intersecting longer flowpaths where tracer mass enters the streambed in one reach and is returned to the surface flow in the next reach.
Dispersion and transient storage are affected by both surface and sub-surface processes. The metrics calculated here may have been dominated by surface processes or reflect storage processes occurring at short timescales resulting in a lack of relationship (Harvey and Bencala, 1993;Ward et al., 2013b). Further evidence is provided by the significant correlation between CO 2 and longterm storage, which may indicate that flow paths with relatively high residence times control streambed chemistry rather than flow paths with short residence times. Further influence on these relationships are indicated by the spatial and temporal groupings shown in Figure 4. The sand-dominated reaches overlap and are separated from the gravel-dominated reach; therefore, the sediment type appears to have a large control on the streambed chemistry, independent of the hydrology. It is crucial, therefore, to consider the effects of sediment type on hydrologic metrics when investigating relationships between streambed chemistry and hydrology. Differences in carbon and nitrogen cycling between the sand-dominated and gravel-dominated reaches have been observed at this site (Comer-Warner et al., 2019, 2020. These studies found the sand-dominated sediments were characterized by higher rates of microbial activity and denitrification with lower concentrations of nitrate, nitrite and lower rates of recalcitrant C uptake than the gravel-dominated sediments. Denitrification was also more complete in the sanddominated sediments, these characteristics resulted in sanddominated sediments having higher concentrations of CO 2 , CH 4 , and CH 4 :CO 2 ratios, and lower concentrations of N 2 O than the gravel-dominated sediments (Comer-Warner et al., 2019, 2020. Seasonally there was no significant difference in streambed chemistry, while there were large variations in discharge and hydrological metrics, providing further evidence for additional controlling factors on streambed chemistry. This may explain some of the unexpected observations discussed above.

Methodological Considerations
In this study point-scale streambed chemistry measurements were compared to reach-scale hydrological metrics and aerobic respiration. We acknowledge that there is a discrepancy in scale in the dataset but believe that relating point and reach-scale observations is the next challenge to understanding controls on biogeochemical cycling. To reduce some of the issues arising from differences in scale we sampled from four to seven piezometers at each streambed sampling site. The piezometers were located through bedforms with upwelling and downwelling locations sampled, to prevent a point measurement representing one extreme of a flow path (Supplementary Table 1). Future studies could improve this design by sampling multiple bedforms within each experimental reach and measuring streambed hydrology concurrently with surface hydrology. This would also aid in controlling for local factors influencing chemical concentrations independent of hydrology, such as sorption of NH + 4 onto clay sediments.

CONCLUSIONS
Advective transport time was related to streambed biogeochemical concentrations, with higher concentrations of CO 2 , CH 4 , and NH + 4 and lower concentrations of NO − 3 , NO − 2 , and N 2 O at longer advective transport times. Long-term storage also exhibited a significant, positive relationship with CO 2 . The effect of advective transport time and long-term storage was dependent on chemical species. NH + 4 , CO 2 , and CH 4 exhibited positive relationships and NO − 3 , NO − 2 , and N 2 O exhibiting negative relationships with advective transport time. The opposite pattern was observed with long-term storage. Dispersion and transient storage did not indicate significant controls on any of the measured chemical species.
These findings contribute to community knowledge of the hydrological drivers of streambed GHGs. They indicate that further to controls of hyporheic residence times previously determined, processes affecting advective transport and not dispersion or transient storage may have the greatest effect on streambed nutrient and GHG concentrations. These hydrological drivers interrelate with other environmental parameters, which are also influenced by local hydrology. Sediment type, substrate availability, and season have also been found to influence streambed chemistry at this site (Comer-Warner et al., 2019, 2020. Further understanding of the effects of distal drivers (e.g., discharge) on hydrological variables and therefore, streambed chemistry should enable predictions of hyporheic nutrient cycling and GHG emissions aiding in effective pollution and climate remediation. Improved understanding may also allow morphological restoration to effectively target nutrient pollution and GHG emissions, which will aid in effective pollution and climate remediation. This improved understanding may allow chemical species to be effectively targeted by morphological restoration, which will aid in effective pollution and climate remediation.
Further investigation into the effects of hydrological metrics on C and N cycling is crucial to improve understanding of global C and N budgets and effects of global change on streams and rivers. Here, streambed concentrations of GHGs were measured and future work should focus on the relationships between hydrological metrics, and reaction rates and fluxes to provide further insight into links between hydrology and streambed chemistry. This, combined with the work presented here, will enable the effects of short-term and long-term hydrological variation, e.g., from storm events or climate change, on GHG dynamics to be better understood. Previous work shows that streambed N 2 O is highest at intermediate residence times (Zarnetske et al., 2011;Quick et al., 2016), however, previously streambed dynamics have not been linked directly with advection, dispersion or transient storage. Installing benthic chambers to measure sediment-water fluxes of GHG and measuring concentration over time will help to further constrain these processes. Improved understanding of the relationships between hydrogeomorphology and advection, dispersion and transient storage will enable conditions advantageous for the reduction of pollution and GHG emissions to be determined. Including streambed GHG emissions and separated transient storage (surface and subsurface) in further investigations would be beneficial (Briggs et al., 2009;Krause et al., 2011). Additionally, considering legacy effects of hydrologic conditions prior to tracer experiments, which may affect streambed chemistry for longer periods and therefore, during tracer experiments where hydrology may have changed, is crucial to further our understanding.

DATA AVAILABILITY STATEMENT
The datasets generated for this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NERC's Environmental Information Data Centre, doi: 10.5285/00601260-285e-4ffa-b381-340b51a7ec50 and doi: 10.5285/5b34d963-d0f0-465e-b395-e955b89e1cd7.

AUTHOR CONTRIBUTIONS
SC-W, DG, and SK conceived of and planned the experiment. SC-W, PB, NB, and AB were involved in data acquisition. SC, CL, and KK were involved in data analysis. SC-W wrote the manuscript and all other authors contributed to manuscript editing. DG and SK provided funding. All authors contributed to the article and approved the submitted version.