Introduction

Streams and rivers provide crucial ecosystem services and are important habitats for aquatic life. However, past and present agricultural activities have exposed them to multiple stressors, drastically altering their natural chemical and hydro-morphological conditions (Allan 2004). Nutrient inputs from agricultural sources have altered natural stoichiometric ratios (Westphal et al. 2020; Wymore et al. 2016) and caused nutrient concentrations to rise above pristine levels (Bernhardt et al. 2017). Practices such as soil tillage, drainage channelization, and riparian clear-cutting have further deteriorated the morphological conditions of streams, increasing streambed light exposure (Tank et al. 2021) and deposition of fine sediment on its surface (U.S. EPA 2002). As a result, agricultural streams have impaired ecosystem structure (Schürings et al. 2022) and functions, with nitrate (NO3–) uptake being the most impacted ecosystem function (Brauns et al. 2022).

NO3– uptake in streams is largely mediated by microbial communities inhabiting the benthic and hyporheic compartments. The environmental conditions in the compartments play an important role in determining both rates and mechanisms of NO3– uptake. In the benthic compartment, factors like light availability and aerobic conditions favor autotrophic NO3– assimilation into biofilms. Furthermore, the presence of other nutrients, such as phosphorous, stimulates autotrophic biomass production (Romaní et al. 2004), increasing NO3– assimilation in biofilms (Hall et al. 2009). Conversely, in the hyporheic compartment, reduced light conditions create an environment that supports heterotrophic processes. Here, the presence of carbon derived from algal metabolites can enhance heterotrophic NO3– uptake, as this carbon is assimilated more quickly compared to the more recalcitrant substances, such as lignin or cellulose, originating from terrestrial ecosystems. Under anoxic conditions, NO3– is permanently removed through heterotrophic dissimilatory pathways i.e., denitrification or dissimilatory NO3– reduction to ammonium (DNRA) (Triska et al. 1989; Arango and Tank 2008).

Previous studies employing 15N–NO3– tracer techniques consistently showed that agricultural streams have higher whole-stream NO3– uptake than reference streams (i.e., LINX II as synthesized in Hall et al. 2009; Mulholland et al. 2008; Tank et al. 2021) and attributed it to an increase in photoautotrophic uptake (Bernot et al. 2006; Hall et al. 2009), and denitrification (Mulholland et al. 2008). While these studies provide valuable information on the overall effect of agricultural land use on whole-stream NO3– uptake, it remains unclear how specific agricultural stressors (e.g., nutrient, fine sediment) alter NO3– uptake rates in the benthic and hyporheic compartments and how this, in turn, affects whole-stream NO3– uptake.

Previous field studies could not disentangle the effect of specific agricultural stressors (i.e., increased light, nutrients, fine sediment) on whole-stream NO3– uptake as in the field stressors co-occur and often interact. Interacting stressors can intensify, reduce, or even produce contrasting effects that cancel individual responses out (Jackson et al. 2016; Morris et al. 2022). For instance, when considering as a response parameter NO3– uptake lengths (i.e., the average length the dissolved form of NO3– travels before being taken up), the positive effect of increased light in agricultural streams is diminished by the elevated NO3– concentration. Light tends to decrease uptake lengths by stimulating gross primary production (GPP) and consequently autotrophic assimilatory demand (Hall and Tank 2003). However, the high availability of NO3–, counteracts this effect, because only a smaller fraction of the entirely available NO3– is removed. Thus, these two stressors combined have an overall small effect on NO3– uptake lengths (Hall and Tank 2003). In addition, stressors can also indirectly affect NO3– uptake by changing the environmental conditions in the compartments. For example, fine sediment can diminish substrate stability in the benthic compartment, consequently reducing autotrophic biofilm development (Biggs et al. 1999; Myers et al. 2007). In the hyporheic compartment, instead, fine sediment reduces the water inflow, limiting the delivery of nutrients and oxygen. Here, the rapid depletion of the oxygen concentrations changes the redox conditions promoting dissimilatory processes such as denitrification when NO3– is not limiting (Böhlke et al. 2004; Holmes et al. 1996). Nutrient pollution can also lead to stoichiometric imbalances that affect uptake rates, for example, high levels of NO3– reduce the dissolved organic carbon (DOC) to NO3– ratio (DOC: NO3–). This reduced energy availability limits the demand for inorganic nitrogen, reducing the biofilm NO3– uptake efficiency (Mulholland et al. 2009; Wymore et al. 2016).

In this study, we quantified the individual and combined effects of two common agricultural stressors, i.e., fine sediment (FS), and the simultaneous increase in phosphorous and light levels (L&P) on benthic and hyporheic nitrate N uptake (N–NO3–) in a full factorial stream-side mesocosm experiment.

We hypothesized that:

  1. (1)

    In the absence of fine sediment, the simultaneous increase in light and phosphorous (L&P) increases benthic and hyporheic N–NO3– uptake due to stimulation of benthic autotrophic and hyporheic heterotrophic biofilms, respectively.

  2. (2)

    Under ambient light and nutrient conditions, fine sediment (FS) reduces benthic and hyporheic N–NO3– uptake due to the loss of stable substrates in the benthic compartment and clogging in the hyporheic compartment.

  3. (3)

    The stressors combined (L&P + FS) have additive effects on N–NO3– uptake in the two compartments i.e., the inhibiting effects of FS reduce the stimulating effects of L&P.

Methods

Experimental setup

The experiment was performed in a stream-side mobile mesocosm facility (MOBICOS) (Fink et al. 2020) located at the forested, upstream part of the Holtemme River (51° 49′ 00.7″ N, 10° 43′ 29.26″ E, Germany). This river section is characterized by a closed canopy and an undisturbed hydromorphology, moderate NO3– concentrations, and low soluble reactive phosphorus (SRP) concentrations compared to downstream reaches (Weitere et al. 2021). Surface water was continuously supplied to the flumes by a submersible pump (type XJ 40, ABS Sulzer, Germany). The inflowing water was filtered with a 50 μm self-cleaning filter (GEFA Process Technik GmbH, Model AP3 DA, Germany) to prevent the inflow of suspended particles and macroinvertebrates. The mesocosm setup comprised 24 flumes (72 cm length, 14 cm height, 8 cm width) (Online Resource 1, Fig. 1) filled with commercially available sediment previously washed and dried at 120 °C. The sediment height was 8 cm, allowing for a benthic (0–2 cm depth) and a hyporheic compartment (2–8 cm depth). The height of the water column was 4 cm. Four ceramic tiles (6 × 1 × 0.5 cm) were equidistantly placed in each flume perpendicular to the flow to facilitate water infiltration into the sediment. Discharge was set to a constant flow rate of 0.17 ± 0.006 L s−1 in each flume and continuously monitored with a magnetic inductive flow meter (IFM, model: SM8000, Germany). Water passed through a twin-wall polycarbonate honeycomb with an opening diameter of 5 mm to ensure laminar flow. The light was provided using one elongated LED lamp (Solar Stinger daylight 16 W, Germany) placed above each flume, and light: dark cycles were set to 10:14 h.

Treatments

We assigned the 24 flumes to four treatments: control, fine sediment (FS), light and phosphorous (L&P), and light and phosphorous combined with fine sediment (L&P + FS) (Table 1). In the control treatment, we mimicked the environmental conditions of the Holtemme River by using similar-sized gravel (2–4 mm) and comparable light intensities (i.e., 7.5 ± 1.5 µmol photons m−2 s−1). The FS treatment was created using fine sand (grain size 0.1–0.4 mm). The L&P treatment was created by increasing the light intensity from 7.5 ± 1.5 µmol photons m−2 s−1 to 51.2 ± 6.5 µmol photons m−2 s−1 and increasing SRP concentrations from below the quantification limit (< 0.003 mg l−1) to 0.04 mg l−1 by constantly supplying a KH2PO4 solution with a peristaltic pump (Watson–Marlow 205U, UK). To test the combined treatment effects, we simultaneously manipulated substrate, SRP, and light in the L&P + FS treatment (Table 1). Each treatment was replicated with six flumes. Flumes were operated under these conditions for two weeks.

Table 1 Overview of environmental conditions in each treatment
Full size table

We expected DOC limitation of heterotrophic microbial growth (Stutter et al. 2020; Graeber et al. 2021) to be of little relevance to our experiment due to the high bulk DOC concentrations in the stream. To ensure that DOC limitation was not the case, we tested in an auxiliary experiment with laboratory microcosms to which extent NO3– uptake was potentially limited by DOC availability. We used a protocol similar to Graeber et al. 2021. After two weeks of flume colonization, we sampled water from each flume outlet and we measured the uptake of DOC, dissolved N, and dissolved P fractions in a bioavailability experiment in the dark over an incubation time of eight days. We calculated the reactive DOC, reactive dissolved N, and reactive dissolved P concentrations and calculated the C:N:P ratio of reactive, hence microbially available macronutrients. To show the potential reactive DOC or P limitation of heterotrophic NO3– uptake, we used a ternary plot approach using the ggtern package in R (Graeber et al. 2021), in which we normalized the C:N:P to the median of a large analysis of bacterial C:N:P ratios (Godwin and Cotner 2018). We also assessed potential changes in dissolved organic matter (DOM) molecular composition via fluorescence spectroscopy and parallel factor analysis using the staRdom toolbox in R (Pucher et al. 2019). A detailed protocol and results can be found in Online Resource 2.

Tracer addition, sampling, and analyses

After two weeks of colonization, one flume per treatment was sampled to determine ambient isotope values (15N atom fraction, x(15N)) of benthic and hyporheic organic matter (OM). Subsequently, we added an enriched 15NO3– solution constantly for 24 h into the turbulent part of the flumes’ inlet section with syringe pumps (NE-1200 Twelve Channel Programmable Syringe Pump, New Era Pump Systems, USA). The addition was adjusted to achieve a final x(15N) of 30 in the stream water. After the addition, we collected sediment samples from the benthic and hyporheic compartments of each flume. To prevent sediment mixing between the benthic and hyporheic compartments, we extracted the upper 2 cm of sediment (benthic sample) and stored it in a sealed container. We then collected a sediment sample from the hyporheic compartment and stored it in another sealed container. To separate OM from the mineral substrate, we rinsed the samples multiple times with water and collected the OM fraction which passed through a 100 μm sieve. The hyporheic samples exhibited visibly lower OM content compared to the benthic samples. Therefore, a larger volume of sediment from the hyporheic samples had to be processed to obtain sufficient OM. The mineral fraction of the samples was transported to the laboratory and dried at 60 °C for at least 24 h before being weighed. To determine the volume of the sediment, the dry weight was divided by the specific density of the sediment, which was measured in the lab using a pycnometer (Blaubrand™).

In the laboratory, OM was freeze-dried (Delta 2–24 LSCplus, Germany) for at least 24 h, homogenized, and three replicates of each sample were weighed into tin caps for isotope analysis. Stable nitrogen isotopic composition was analyzed on an elemental analyzer (Flash 2000 Organic Elemental Analyzer, Thermo Fisher Scientific, Germany), directly connected via an open split system (ConFlo IV, Thermo Fisher Scientific, Germany) to an isotope ratio mass spectrometer (Delta V Advantage IRMS, Thermo Fisher Scientific, Germany).

During the analysis, only a minor enrichment was found (i.e., maximum x(15N) = 6.38). Thus, the normalization was done by analyzing the reference materials IAEA-311 (x(15N) = 2.05) (Parr and Clements 1991) and AS-3 (x(15N) = 0.36) (Kornexl et al. 1999) and applying a two-point calibration approach. The analytical precision was below 0.2. Particulate nitrogen (PN) concentrations of the OM were measured with an elemental analyzer (Elementar Analysen Systeme GmbH, Germany).

Calculation of N–NO3 – uptake

We used sediment samples collected before and after the isotope tracer addition to calculate nitrate N uptake (U g N–NO3– m−2 d−1) in the benthic and hyporheic compartments. In this experiment, we did not differentiate between assimilatory or dissimilatory processes, thus, we use the term “uptake” to refer to both processes. We followed the procedure proposed inMulholland 2004 and corrected all 15N values by the background values. Then, we calculated the excess atom fraction (AFE) to obtain uptake rates. For that, we first calculated the amount of tracer in the OM (15\({N}_{OM}\), g 15N) as:

$$^{{15}} N_{{OM}} = OM \times \left( {\frac{{\% N}}{{100}}} \right) \times AFE_{{OM}}$$

where OM (g dry weight) is the amount of organic matter within the samples, %N is the nitrogen elemental concentrations (%) of the OM, and \({AFE}_{OM}\) the excess 15N fraction of the OM. Then, we scaled the result by sediment volume (15\({N}_{Sed}\), g 15N cm−3):

$$N_{{Sed}} = \frac{{15N_{{OM}} }}{V}$$

where V (cm−3) is the volume of sediment from which the OM was collected. The volume of sediment was calculated by dividing the weight W(g) of the dry sediment by the sediment density d (g cm−3) as:

$$V=\frac{W}{d}$$

where dsand = 2.76 (g cm−3), dgravel = 2.54 (g cm−3). Subsequently, we calculated the uptake rate of the sediment (\({U}_{sed}\) g N–NO3– cm−3 d−1) as:

$${{U}_{sed}}_{}=\frac{{15 N}_{Sed}}{\left(\frac{{15 N}_{Flux}}{{N}_{flux}}\right)}$$

where

$$^{{15}} N_{{Flux}} = \left( {AFE_{{w,a}} {\text{*}}Q{\text{*}}C) - (AFE_{{w,b}} {\text{*}}Q{\text{*}}C} \right)$$
$${N}_{flux}=Q\text{*}C$$

15NFlux is the water tracer flux. For this, we subtracted from the total mass flux \(\left({AFE}_{w,a}\right)\) the background mass flux \(\left({AFE}_{w,b}\right)\) after multiplying both factors by C (N–NO3– concentration (mg l−1)) and Q (flume discharge (l s−1)).

We expressed uptake on an areal basis (\(\text{U}\), g N–NO3– m−2 d−1), i.e., N–NO3– uptake of a 1 m2 surface with a height of 1 cm by multiplying the \({U}_{sed}\) by 10,000. We finally summed the benthic uptake and the hyporheic uptake to obtain the total uptake (\({U}_{tot}\) g N–NO3– m−2 d−1).

Chlorophyll-a and microbial density

Chlorophyll-a concentrations in the sediment were measured from benthic samples collected after the two weeks of colonization. Sediment samples were immediately frozen at − 20 °C, and chlorophyll-a was analyzed via high-performance liquid chromatography with a Thermo Scientific UltiMate 3000 HPLC System (Dionex, Thermo Fisher Scientific Corporation, USA). Bacterial abundance was measured by collecting a separate sediment sample from the benthic and hyporheic compartments of each flume in sterile Falcon tubes. To detach cells from the sediment, we used established protocols as described by Chen et al. 2021. Cells were stained with DAPI and counted using an epifluorescence microscope at 100x magnification. A microscopic investigation was performed using a Zeiss AxioImager.Z2 epifluorescence microscope equipped with an HXP R 120 W/45 C UV Hg-vapor lamp, Colibri.2 LED illuminations and the following fluorescence filters: DAPI (365/10 nm excitation, 420 LP emission, FT 395 Beam Splitter), Alexa488 (472/30 excitation, 520/35 emission, 495 Beam Splitter), and Alexa594 (562/40 excitation, 624/40 emission, 593 Beam Splitter). The filter sets discriminated clearly between DAPI (365 nm) and natural autofluorescence of the diatom cells. Imaging was done with 100X oil objective numerical aperture N:A 1.4. The software for image acquisition allows for overlapping images acquired successively with different filter sets (Zen software from Carl Zeiss). Images are reported in Online Resource 3.

Physical-chemical parameters monitoring

Light intensity was measured with a frequency of five minutes using a light intensity data logger MX2202 (Onset, USA) placed on the sediment of three flumes per treatment. Planar oxygen sensors (SP-PSt6-YAU, PreSens Precision Sensing GmbH, Germany) were glued to the inner wall of three flumes for each treatment to monitor oxygen concentration. Oxygen was measured every 25 s at two depths: 2 cm (benthic) and 6 cm (hyporheic). Some cables connecting the planar sensors to the oxygen readers detached from the wall during the measurements, resulting in incomplete temporal data for certain treatments (specifically, 3 out of 9 incomplete datasets for the benthic and 2 out of 9 incomplete datasets for the hyporheic compartment). To ensure comparability of data among treatments, we selected the oxygen data from two flumes per treatment for each compartment where the oxygen data was complete and uninterrupted for a minimum of seven consecutive days (Online Resource 1, Figs. 2 and 3). Water samples for N–NO3–, nitrite (N–NO2−), ammonium (N–NH4+) were collected once at the outlet of each flume after the 2-week colonization period, while P was measured every 3rd day (details are reported in Online Resource 1).

Statistical analysis

To assess the individual and combined effect of L&P and FS on N–NO3– uptake, OM, bacterial abundance, and chlorophyll-a, we used General Linear Models (function glm, package: stats) with L&P and FS as fixed categorical factors. The resulting model was: intercept + L&P + FS + L&P* FS. If necessary, data was ln(x) transformed to improve its alignment with a Gaussian distribution. We also conducted a Tukey post-hoc test separately for benthic and hyporheic N–NO3– uptake to assess significant differences among the treatments (package: emmeans and multcomp). To classify the interaction type, we considered both the statistical significance of the interaction term in the GLM model (the p-value of L&P* FS) and the coefficient estimate of the interaction term. When the p-value of the interaction term was above 0.05, we concluded that there was no interaction, and the effect was classified as additive. Otherwise, we classified the interaction either as synergistic or antagonistic. The interaction was classified as antagonistic when the effect of a treatment was diminished by the presence of the other treatment (coefficient estimates were lower in L&P* FS than in the treatment alone). Conversely, when the effect of treatment was increased by the presence of the other treatment (coefficient estimates were higher in L&P* FS than in the treatment alone), then the interaction was classified as synergistic.

All tests were performed in R (R Core Team 2021).

Results

Treatment effects on compartmental N–NO3 – uptake

The addition of light and phosphorous (L&P) increased significantly N–NO3– uptake in the benthic compartment (Table 2L&P p < 0.001). Specifically, L&P increased 12-times benthic uptake in the treatment with gravel substrate (comparison Control vs. L&P, Fig. 1a and Online Resource 1 Table 1), while it increased 42-times benthic uptake in the treatment with fine sediment (comparison FS vs. L&P + FS, Fig. 1a and Online Resource 1 Table 1).

Fig. 1
figure 1

N–NO3– uptake in the benthic A and hyporheic B compartments in the different treatments (Control; FS: fine sediment; L&P: light and phosphorous; L&P + FS: light and phosphorous combined with fine sediment). Different letters indicate significant differences among treatments at the 0.05 level (Tukey post-hoc test). The Y-axis has been ln(x) transformed

Full size image

Benthic uptake in the FS treatment was, on average, 3-times lower than in the control (Fig. 1 and Online Resource 1 Table 1). However, differences were not significant (post-hoc test, comparison Control vs. FS, Fig. 1a), and the overall effect of FS on benthic uptake was not significant (Table 2 FS p > 0.05).

In the benthic compartment, the stressors did not interact (Table 2, L&PxFS p > 0.05) and benthic uptake in the combined treatments (L&P + FS) was comparable to the uptake observed in the L&P treatment (post-hoc test, comparison L&P vs. L&P + FS, Fig. 1a), signifying that the addition of FS did not affect the positive effect induced by L&P (i.e., additive effect).

In the hyporheic compartment, L&P and FS significantly increased N–NO3– uptake (Table 2, FS p < 0.001, L&P p < 0.001). Specifically, L&P increased 7-times hyporheic uptake in the treatment with gravel substrate (comparison Control vs. L&P, Fig. 1b, and Online Resource 1 Table 1), while FS increased 13-times hyporheic uptake at ambient nutrient and light conditions (comparison Control vs. FS, Fig. 1b and Online Resource 1 Table 1). Within the hyporheic compartment, the stressors interacted (Table 2, L&PxFS p < 0.001), with L&P weakening the effect of FS and reducing hyporheic uptake by ~ 2-fold (comparison FS vs. L&P + FS, Fig. 1b and Online Resource 1 Table 1) (i.e., antagonistic interaction).

Table 2 Summary of coefficient estimates from the GLM comparing responses between treatments in the two compartments
Full size table

Benthic uptake was 14, 25, and 30-times higher than hyporheic uptake in the Control, L&P, and L&P + FS treatments, respectively (Fig. 1 and Online Resource 1 Table 1). Conversely, in the FS treatment, benthic uptake was three times lower than hyporheic uptake (Fig. 1 and Online Resource 1 Table 1).

Treatment effects on chlorophyll-a, bacterial density, organic matter, and oxygen

Light and phosphorous (L&P) significantly increased chlorophyll-a concentration in the benthic compartment (Table 2, L&P p < 0.001) and chlorophyll-a was 5-fold higher in the L&P and L&P + FS treatment than in the control (and Table 3). Fine sediment (FS) increased significantly bacterial abundance in the hyporheic compartment (Table 2, FS p < 0.05). However, when L&P was added to FS (L&P + FS treatment) the stressors interacted antagonistically in the hyporheic compartment (Table 2, L&PxFS p < 0.05) and bacterial abundance decreased significantly (Table 3) (i.e., antagonism). In the benthic compartment, L&P and FS decreased the OM content, however, only FS effects were significant (Table 2, FS p < 0.05). L&P and FS interacted on benthic OM content (Table 2, L&PxFS p < 0.05), and OM content was higher in the combined treatment (L&P + FS) than in the two treatments alone (Table 3). The benthic compartment of all the treatments was oxygenated through the entire duration of the measurements (Table 3 and Online Resource 1 Fig. 2). In the hyporheic compartment of the treatments with fine sediment (i.e., FS and L&P + FS), we measured hypoxic and anoxic conditions (Table 3 and Online Resource 1 Fig. 3). In contrast, the hyporheic compartment of the flumes with a gravel bed (i.e., control and L&P) remained oxygenated throughout the experiment (Table 3 and Online Resource 1 Fig. 3).

Table 3 Parameters measured in the different treatments in the two compartments
Full size table

Auxiliary experiment on the potential DOC limitation

We measured DOC concentrations of on average 9.4 mg \({l}^{-1}\) (Fig. 3a, Online Resource 2). However, after 8 days of incubation in the dark in microcosms (Fig. 3b, Online Resource 2), only an average of 6.9% of the DOC was taken up by microbial heterotrophs indicating an average concentration of microbially reactive DOC of 0.6 mg \({l}^{-1}\). We also found no removal of dissolved OM fluorophores during the incubation but the production of a humic-like fluorophore (Fig. 4, Online Resource 2). When combining the reactive DOC concentration with the reactive N and P (concentrations of N–NO3– + ammonium N–NH4+ and SRP, and reactive dissolved organic nitrogen and P), we calculated reactive C:N:P ratios of 187:501:1 and 937:1326:1 for the control and FS treatments, respectively, and C:N:P ratios of 40:57:1 and 29:53:1 for the L&P and L&P + FS treatment, respectively (Table 3, Online Resource 2). These C:N:P ratios of the microbially reactive C, N and P constituents were lower in C relative to N compared to the literature-based C:N:P average for bacteria of 64:14:1 for all treatments and lower in C relative to P for the L&P treatment (Fig. 5, Online Resource 2).

Discussion

Fine sediment inputs and augmented light and phosphorus levels are common stressors in streams draining agricultural catchments. However, the individual and combined effects of these stressors on NO3– uptake are poorly understood. Here, we manipulated both stressors in a mesocosm experiment and quantified their individual and combined effects on nitrate N uptake (i.e., N–NO3–, hereafter “uptake”) in the benthic and hyporheic compartments. We found that stressors altered compartmental uptake rates but through different processes and intensities in the two compartments. Moreover, stressor interaction was compartmental dependent.

Stimulatory effects of light and nutrients are stronger on benthic than hyporheic uptake

We hypothesized that in the absence of fine sediment, the simultaneous increase in light and phosphorous (L&P) would increase benthic uptake due to stimulation of benthic autotrophic biofilm. Our results confirmed the hypothesis and we observed a 12-time increase in benthic uptake following the addition of L&P in the treatment with gravel (i.e., L&P treatment). Moreover, L&P stimulated autotrophs, resulting in a 5-fold higher benthic chlorophyll-a concentration in the L&P treatment than in the control. Assuming that algae were the dominant mediator of uptake in the benthic compartment, this finding indicates that uptake is more responsive to L&P enrichment than what can be inferred from chlorophyll-a concentration, which serves as an indicator of autotrophic biomass.

Different mechanisms might explain this substantial difference. First, light and nutrient-rich conditions increase benthic biofilm biomass (Romaní et al. 2004). However, they can also induce changes in the biofilm species composition. For example, under high L&P conditions, single-cell taxa, i.e., diatoms, can be replaced by filamentous, chain-forming green algae or cyanobacteria (Bourassa and Cattaneo 2000; Passy 2007; McCall et al. 2017). These taxa exhibit different physical characteristics, including variations in surface-to-volume ratio, nutrient uptake, and photosynthetic rates (Steinman et al. 1992). Larger cells have a higher NO3− uptake capacity and internal storage volume(Hillebrand et al. 2022) but a lower chlorophyll density per biovolume (Finkel et al. 2004). Thus, the higher increase in uptake compared to chlorophyll-a content could be attributed to the development of an algal community consisting of larger cells with higher uptake and NO3– storage capacity, albeit with reduced chlorophyll-a levels. In addition, the higher P availability and the development of a biofilm matrix could have greatly stimulated heterotrophic uptake, for example, by internal nutrient cycling (Besemer et al. 2009; Romaní et al. 2004) and/or by providing an increased surface area for bacterial growth in the benthic biofilm (Stock and Ward 1989).

We hypothesized that in the absence of fine sediment, L&P would increase hyporheic uptake due to stimulation of heterotrophic biofilms. Our results support this hypothesis since we observed a 7-time increase in hyporheic uptake following the addition of L&P in the treatment with gravel (i.e., L&P treatment). We assume that uptake was mostly attributable to assimilatory processes, as we always measured aerobic conditions in the hyporheic compartment. We suggest that heterotrophic uptake in the hyporheic compartment was stimulated not only by the addition of P in the L&P treatment but also by the presence of a more readily available source of carbon originating from the benthic algae, whose growth was enhanced in the L&P treatment. Our auxiliary experiment showed that when considering only microbially reactive DOC, N, and P (see Graeber et al. 2021; Stutter et al. 2018) there was a severe DOC limitation of N–NO3– uptake. Specifically, for all treatments, the molar C:N:P ratios of the microbially reactive C, N, and P constituents revealed lower C:N (< 1) compared to the literature-based C:N median of 4.6 for bacterial biomass (equaling 64(C):14(N):1(P) (Godwin and Cotner 2018). The P addition in the L&P treatment decreased the N:P ratio from > 500 to 53–57 and the C:P ratio from > 187 to 29–40, which are values closer to the typical bacterial N:P ratio = 14 and C:P ratios = 64 (Godwin and Cotner 2018, Fig. 5, Supplementary material 2). This suggests that heterotrophic uptake may have been DOC and P co-limited under ambient conditions (i.e., control and FS treatments), and our P additions in the L&P treatment may have released this limitation. Thus, the P addition, together with the release of DOM from the benthic algal community, could have increased the amount of bioavailable C and stimulated heterotrophic uptake. However, future studies should also assess the DOC composition change due to algal exudates and whether autochthonous DOC production alleviates allochthonous DOC limitation.

Finally, although L&P increased uptake rates in both compartments, it altered the relative contribution of the two compartments to whole-flume uptake. In the L&P treatment, benthic uptake was 25 times higher than hyporheic uptake, while in the control treatment, the ratio was 14:1. This suggests that in the absence of other stressors, L&P intensifies the significance of the benthic compartment in the uptake process, nearly doubling its relative contribution to the total uptake.

Contrasting effects of fine sediment on uptake in the benthic and hyporheic compartment

We hypothesized that under ambient light and nutrient conditions, fine sediment (FS) would reduce benthic uptake by decreasing biofilm development due to the less stable substrate.

Under ambient light and nutrient conditions, both uptake and chlorophyll-a content tend to be lower in the FS treatment than in the treatment with gravel (i.e., Control). However, the differences were not significant. Previous studies suggested that the reduced stability of FS (compared to gravel), can result in reduced periphyton growth (Biggs et al. 1999) and, consequently, lower uptake (Romaní et al. 2004). However, in our experimental setup, we did not witness any sediment movement, and the benthic biofilm in the Control and FS treatments showed similar development. Hence, we suggest that the lower (but not significant) uptake observed in the FS treatment might not be primarily due to reduced periphyton growth, but, instead to a diminished mass transfer from the water column to the benthic compartment. FS exhibits lower roughness compared to the gravel substrate and, consequently, lower turbulence near the streambed. Previous research has demonstrated that elevated near-bed turbulence enhances the downward transport of dispersal cells through the water column to the benthic compartment (Besemer et al. 2009) and increases nutrient fluxes (Risse-Buhl et al. 2017; Anlanger et al. 2021). Thus, we suggest that the lower turbulence above the benthic biofilm in the FS treatment might be the underlying cause for the observed reduced uptake compared to the control. Previous studies showed that heterogeneous substrates have higher algal biomass and productivity (Cardinale et al. 2002; Hoellein et al. 2009), and phosphate uptake (Marti and Sabater 1996) than homogeneous substrates. Based on our findings, this could potentially be extended to NO3– uptake, but further research is necessary to investigate this relationship more comprehensively.

We hypothesized that under ambient light and nutrient conditions, FS would reduce hyporheic uptake due to clogging. Our hypothesis was rejected as FS increased uptake and we measured a 13-fold higher hyporheic uptake in the FS treatment than in the control treatment. Since we measured anoxic conditions in the hyporheic compartment, we assume that the FS had stimulated dissimilatory process, i.e., denitrification or dissimilatory NO3– reduction to ammonium (DNRA). Previous studies under field conditions concluded that denitrification has a small contribution to total NO3– removal compared to assimilatory pathways (e.g., Mulholland et al. 2008). Nonetheless, denitrification rates tend to increase with elevated NO3– concentrations because NO3– is not limiting (Mulholland et al. 2008). These conditions are usually found in agricultural streams, however, in the study area, also forested streams, exhibit elevated NO3– concentrations (i.e., 1-1.5 N–NO3– mg l−1) due to atmospheric deposition. Hence, denitrification could occur whenever favorable conditions are present because NO3– is non-limiting. This might explain why a sole variation in the substrate type (i.e., from gravel to fine sand), under ambient light and P conditions, resulted in a 13-fold increase in uptake rates in the hyporheic compartment.

Notably, in the FS treatment, the role of the benthic and hyporheic compartments in terms of contribution to total uptake was reversed. Specifically, hyporheic uptake was larger than benthic uptake and the hyporheic compartment contributed to 70% of the entire flume uptake. This implies that the hyporheic compartment might play a major role in NO3– removal in forested streams when ambient NO3– concentrations are non-limiting and FS occurs, i.e., in pools. Moreover, under these conditions, rates of dissimilatory uptake in anaerobic hyporheic sediments (i.e., pools) might be comparable to rates of assimilatory uptake measured in more heterogeneous aerobic sediments (i.e., riffles).

Stressor interaction is compartmental-specific

We hypothesized that the combined effects of stressors on uptake would be additive in the benthic compartment. Specifically, we expected that the increase in uptake induced by L&P would be reduced by the presence of FS because FS reduces the formation of benthic biofilm. Our hypothesis was rejected because benthic uptake rates in the L&P + FS treatment were comparable to the ones measured in the L&P treatment, indicating that FS did not reduce uptake rates.

We previously argued that the reduced roughness of FS could potentially limit nutrient fluxes from the water column to the benthic compartment, leading to lower uptake rates in the benthic compartment compared to gravel. However, we did not observe this effect in the L&P + FS treatment, thus, we deduce that under increased L&P, the effect of FS becomes negligible. We speculate that L&P stimulated the formation of a large and more structurally complex biofilm (Proia et al. 2012) which was able to stabilize the underneath fine sediment (Valentine and Mariotti 2020) and form structures that increased the uptake of nutrients from the water column (Besemer et al. 2009).

In the hyporheic compartment, we hypothesized that the combined effects of stressors on uptake would be additive. Specifically, we expected that the presence of FS would reduce the increase in uptake induced by L&P, as FS prevents the inflow of water in the hyporheic compartment. Our proposed mechanism, and consequently hypothesis, proved to be incorrect. We observed an increase in the hyporheic uptake, which was attributable to FS and not to L&P, as hypothesized. FS created anoxic conditions in the hyporheic compartment of the L&P + FS treatment, which increased uptake through dissimilatory pathways. Although the same mechanism occurred in the hyporheic compartment of the FS treatment, hyporheic uptake rates in the L&P + FS treatment were 2-fold lower than in the FS treatment, indicating an antagonistic interaction between L&P and FS, with L&P reducing the effects of FS. We attribute this negative effect of L&P to bioclogging, i.e., the biological clogging derived from biofilm growth among interstitial spaces which can reduce water exchange between the stream channel and the subsurface (Aubeneau et al. 2016). Specifically, we suggest that in the L&P + FS treatment, L&P stimulated benthic biofilm, which grew among the sediment grains, reducing nutrient and carbon flow to the hyporheic compartment and subsequently decreasing uptake rates. This explanation supports previous studies which show that the benthic-hyporheic connectivity plays a crucial role in the hyporheic nitrogen uptake (Mendoza-Lera et al. 2019) by affecting solute transport and redox conditions (Caruso et al. 2017). Finally, while the stressors combined increased uptake rates in both compartments, their simultaneous occurrence further accentuated the disparity between benthic and hyporheic uptake i.e., the benthic uptake was 30 times higher than hyporheic uptake. This implies that in agricultural streams where fine sediment, light, and phosphorous occur simultaneously, the relative contribution of the hyporheic compartment to whole-stream uptake diminishes compared to less disturbed streams.

Conclusion

Previous field studies on NO3– dynamics concluded that whole-stream NO3– uptake increases in agricultural streams due to an increase in photoautotrophic uptake (Bernot et al. 2006; Hall & Tank, 2009), and denitrification (Mulholland et al. 2008). Here, we provide the mechanistic basis for the previously observed increase of whole-stream NO3– uptake in agricultural streams.

First, we demonstrate that nutrients and light are the primary drivers of the increase in NO3– uptake, while the effects of fine sediment were surprisingly modest. Second, we show that the hyporheic compartment does not remove a substantial amount of NO3–, although uptake rates tend to increase in response to agricultural stressors. Consequently, the observed rise in NO3– uptake rates in agricultural streams is mainly attributable to the stimulatory effects of light and nutrients on the benthic compartment.

Finally, we show that the type of stressor interactions is not only stressor-specific but also depends on the ecosystem compartment under study. Consequently, findings from one compartment can not be transferred to another, especially under multiple stressor scenarios. Indeed, disentangling multiple stressor effects in compartmentalized ecosystems such as streams requires considering all compartments and their contribution to whole-system functioning. We are beginning to understand how multiple interacting stressors affect stream functioning, but more mechanistic evidence is needed to disentangle whether additive or non-additive effects prevail in human-altered ecosystems.