Shallow‐lake sediments release nutrients by complete destratification events

Abstract

The development of phytoplankton communities in hypereutrophic shallow lakes, often used for aquaculture, is not fully understood and can sometimes be unpredictable. Focusing on the abiotic factors that regulate their succession, we recorded short‐term mixing events in a shallow lake and examined their relationship with nutrient release from sediments in the lab. In situ measurements reveal a dynamic cycle of mixing and stratification during summer, when the lake mostly stratifies during the day and mixes at night, depending on wind conditions. The studied lake was stratified 45% of the time and, on average, mixed every 1.5 days. In terms of hydrodynamics, the velocities of surface and bottom waters are similar in magnitude, regardless of whether conditions are calm or windy. Stirred‐core experiments recreated both lake hydrodynamic regimes and the observed patterns of destratification at the study site. Temporal destratification experiments show that the sediment releases more solutes during complete mixing than during partial destratification, due to an increase in sediment/water concentration gradients. This results in more phosphorus and ammonia being released, and more nitrate being consumed by sediments, when the water column is fully mixed compared to when a bottom layer remains unmixed. The effect of dissolved oxygen did not directly influence nutrient release by Fe‐P compounds dissolution, as oxygen above the sediment did not fall below 50% saturation, but mixing enhanced the transport of electron acceptors to the sediment. The cycle of stratification and mixing appears to be a key factor in internal loading under oxic conditions.

Core Ideas

• During summer, shallow lakes exhibit a dynamic mixing‐stratification regime.

• Temporal stratification followed by complete destratification releases nutrients from sediments.

• The periodicity of these episodes may shape nutrient dynamics in the water column.

Plain Language Summary

In shallow, nutrient‐rich lakes often used for aquaculture, the growth of algae is hard to predict. This study explored how short‐term mixing of lake water affects the release of nutrients from sediments. During summer, the lake alternated between mixing at night and stratifying during the day, with mixing happening roughly every 1.5 days. Lab experiments mimicked these conditions and showed that complete mixing of the water column led to a higher release of nutrients (like phosphorus and ammonia) from sediments compared to incomplete mixing. However, this effect depended on the nutrients accumulating during earlier stratified periods. In summary, the study found that occasional mixing stirs up nutrients from sediments, fueling phytoplankton growth.

Abbreviation

SWI

sediment‐water interface

1. INTRODUCTION

The frequency of harmful cyanobacterial blooms is increasing worldwide, along with water pollution and climate change (Gobler, 2020). Shallow lakes are the most common type of waterbody and are more susceptible to eutrophication and cyanobacteria growth than deep lakes (Qin et al., 2020), because of the more significant influence of the sediment bed, which allows rapid nutrient cycling even under hypoxic conditions (Phillips, 2008).

These highly productive, shallow lakes are often used for controlled fish production. Whereas fish farming in shallow lakes dates back to 2300 years before the present (Beveridge & Little, 2002), the development of phytoplankton assemblages in shallow lakes remains partially understood and sometimes unpredictable (Scheffer, 2004). Understanding the internal dynamics of limiting nutrients, such as phosphorus (P) and nitrogen (N), could help explain sudden changes in phytoplankton assemblages. Nutrient cycling responds to the interaction between biogeochemistry and hydrodynamics (Higashino et al., 2008; Huettel et al., 2003; Umlauf et al., 2023).

Shallow lakes show extremely dynamic stratification regimes (Søndergaard, Nielsen, Johansson, et al., 2023), resulting in significant variations of dissolved oxygen (DO) (Goncharov et al., 2024) and carbon cycling (Andersen et al., 2017). Thus, the cycling of nutrients must be highly dynamic in such polymictic waterbodies. Abiotic factors such as light exposure and nutrient availability (despite high trophic levels) primarily regulate the growth of phytoplankton, with P and N being the key limiting nutrients (Kajgrová et al., 2024). Consequently, phytoplankton can exhibit sudden changes in population as a response to nutrient dynamics, with cyanobacteria typically being the most sensitive photosynthetic organism (Yang et al., 2023). As for biotic factors, the zooplankton community in aquaculture ponds typically lacks large filtering organisms, resulting in a significant food web gap between bred fish (artificially fed with cereals) and the microbial community, which comprises by microzooplankton, microalgae, bacteria, cyanobacteria, fungi, protists, and oomycetes (Pechar, 1995). Therefore, most nutrient cycling occurs at the sediment‐water interface (SWI) rather than within the water column.

In the summer, when sufficient light is available and the external load is low, internal loading becomes relevant for primary production (Orihel et al., 2017). Whereas P internal loading mostly—but not always—depends on oxygen availability in bottom waters (Hupfer & Lewandowski, 2008), oxygen concentration itself depends on hydrodynamics (Higashino, 2018). During the summer, when phytoplankton blooms, mainly convective cooling and wind action drive the hydrodynamics of shallow lakes, except for those with low retention times due to significant inflows and outflows (Bouffard & Wüest, 2019). The interaction between diurnal heating (primarily due to solar radiation) and night cooling results in temporal stratification, also called periodic mixing (Gantzer & Stefan, 2003). As a result, shallow lakes can stratify during the day and mix during the night, yet wind can alter this regime (Holgerson et al., 2022). The question is whether the cooling and destratification that occurs overnight are sufficient for complete mixing and transport of the released nutrients from the benthic water layers located immediately above the sediment.

During stratification, oxygen depletion in the deeper waters above the Fe‐rich sediment can lead to the reduction of nitrates and iron (Fe), resulting in the dissolution of iron (hydr)oxides and the release of P, as well as the production of ammonia through dissimilative nitrate reduction (Yuan et al., 2022). However, a high proportion of P and N in the form of ammonia also originates from the decomposition of organic matter by bacteria (Wetzel, 2001). The proportion of P forms varies depending on several factors, including the chemical compositions of the catchment soils, human activities, and the internal cycle of the pond itself. Anderson, Johengen, Godwin, et al. (2021) captured the timing of P release from sediment in situ in Lake Erie. Mixing episodes can restart or change the speed of chemical and microbial reactions in the system by reoxygenating bottom waters (Orihel et al., 2015). Therefore, periodic mixing and oxygen depletion near sediments observed in shallow lakes may serve as a switch mechanism for benthic fluxes (Wilkes, 2019). Likewise, stratification periods may work as a capacitor by accumulating nutrients before their release. The speed of destratification, whether smooth (convection only) or abrupt (convection plus wind), affects the extent of mixing throughout the water column, resulting in either complete or incomplete destratification. Such processes might influence the dynamics of benthic fluxes (Anderson, Johengen, Miller, et al., 2021). However, the net effect of this alternation between stratification and mixing regimes on nutrient release is unknown for oxic shallow waters. This study aims to investigate the impact of short‐term stratification cycles on benthic fluxes. To achieve this, we followed the summer stratification regime of an aquaculture shallow lake (a fishpond) in close detail and tested the impact of these mixing processes on the release of nutrients from sediments in the laboratory. This study offers a deeper understanding of the role of hydrodynamics in internal nutrient cycling within shallow lakes.

2. MATERIALS AND METHODS

2.1. Study site and in situ measurements

Municky fishpond (Figure S1) is an artificial shallow lake (area: 105 ha, maximum depth: 2.2 m) used for carp aquaculture located in the region of South Bohemia, Czech Republic in Central Europe, with a long tradition of fish farming since the 12th century. This pond can serve as a representative model of the thousands of similar lakes scattered throughout the region.

Water temperature was recorded as in Goncharov et al. (2024), using six platinum resistance temperature detectors (Pt100) located at different water depths, 0.1, 0.5, 0.9, 1.1, 1.3, and 1.4 m, with a Minilog data logger (Fiedler AMS, Ltd.,). These were attached to a steel pole of a temporary frame that was fixed to the lake bottom (Frame 1; Figure S1). Data were collected every 10 min from June 1 to September 31, 2020. When analyzing this time series, we considered that the lake was in a mixed state (not stratified) when the temperature difference between the lowest and the uppermost layers was less than 0.5°C. We chose this threshold parameter as a practical rule of thumb. An informative discussion about different values used in literature can be found in Goncharov et al. (2024).

Both air temperature and wind speed and direction were recorded hourly using a weather station (WH1081, FrequencyCast) installed in temporary Frame 1 (Figure S1), 80 cm above the water surface, for the same period as the water temperature measurements.

Core Ideas

• During summer, shallow lakes exhibit a dynamic mixing‐stratification regime.

• Temporal stratification followed by complete destratification releases nutrients from sediments.

• The periodicity of these episodes may shape nutrient dynamics in the water column.

Water velocities were recorded during late summer capturing two scenarios of interest: under calm (August 11–14, Frame 1 in Figure S1, water depth 1.35 m) and windy (August 25–29, Frame 2 in Figure S1, water depth 1.23 m) conditions. An acoustic Doppler profiler (Aquadoop, Nortek) was deployed at the bottom of the lake, near the temporary frames, in an upward‐looking position, sampling at 1 Hz, and covering 30 cells, each 5 cm thick. Surface (20% of water depth) and bottom (80% of water depth) water velocities were computed by orthogonal sum of the 3D velocity components. Mean velocities were obtained by running a 2‐min moving average.

2.2. Sediment characterization and treatment

Sediment material used for incubations was collected using an Ekman grab sampler. It was taken from the surface layer of sediment, which was approximately 5 cm thick but could reach up to 10 cm at its maximum depth. The aim was to avoid hitting the deeper, compressed sediment with different structural and chemical properties, and to collect only the more homogenized surface layer. The collected sediment was left to settle, and the water was decanted in the laboratory. The sediment was sieved through a 2‐mm mesh to remove organic debris, which was present at a low density but could affect the installation of minipeepers and pore water analyses. After homogenization, made to ensure reproducibility and equal starting conditions, the sediment was transferred equally to 20 cm diameter, 40‐cm height plexiglass incubation tubes.

Immediately afterward, sediment cores were gently filled with tap water without chlorine and remained in this state for 2 weeks. This period was dedicated to establishing new stable biogeochemical balances. During this time, the water column was mixed at the lowest possible speed (1 V, as described in this section) to prevent stratification, as previously confirmed using a dye.

The collected sediment was analyzed for dry weight content (105°C for 4 h) and organic matter as loss on ignition (LoI; 550°C for 2 h). The total concentrations of the main chemical elements were determined with 8800 Triple Quadrupole ICP‐QQQ (Agilent Technologies) after freeze‐drying and mineralization of the sample using nitric‐perchloric acid digestion (Kopáček et al., 2001). Particle size distribution was determined using a Mastersizer 2000 particle size analyzer with a Hydro MU 2000 wet dispersion unit).

The sediment used in the experiment was characterized by a dry matter content of 15.5%, LoI values of 18.6%, and a distribution of fine‐grained particles. These characteristics, together with the average total concentrations of the chemical elements are presented in Table 1.

TABLE 1.

Main characteristics of surface sediments from Municky fishpond.

Dry content	Organic matter	Particle size (µm)			Average total concentration of chemical elements (mg g−1)
(%)	(%)	d10	d50	d90	Mg	Al	P	S	Ca	Mn	Fe
15.5	18.6	7	29	186	3.8	70.5	1.4	3.5	8.6	0.7	32.0

2.3. Stirred cores experiment—Effect of water velocity

We incubated three sediment cores (20 cm diameter, 20‐cm sediment bed, 16‐cm water column; Figure S2) to evaluate the effect of water velocity on benthic fluxes. Two volumes of inner stagnant and outer moving water were created in each core by inserting a cylinder at its center (32 cm long and 8.9 cm diameter). The outer water was stirred by plastic grid‐palettes powered by a 2–16 V engine. The cores were incubated with minimal stirring (1 V) while monitoring DO levels for 1 week, until stabilization (no changes in oxygen concentrations) was achieved. Flow speed was calibrated against voltage in a separate core replicate by tracking suspended particles (Section S.3 of Supporting Information). For the experiment, we stirred the cores at 2.3, 5.0, and 13 V to achieve flow velocities of 0.5, 1.5, and 4.5 cm s⁻¹, respectively, corresponding to friction velocities of 0.06, 0.13, and 0.27 cm s⁻¹ (validated by Acoustic Doppler Velocimetry). Flow regimes correspond to laminar (Re = 555), transition (Re = 1665), and turbulent (Re = 4995). Unfortunately, the core used for laminar flow had porewater leakage from the bottom (about 20 mL), which slightly biased these results.

We measured vertical profiles of porewater solute concentrations with gel minipeepers (diffusive equilibrium in thin layers [DET]) with a 5‐mm resolution, according to Krom et al. (1994). The DET method establishes a concentration equilibrium between the dissolved ions in the pore water and the water contained in the gel. A polyacrylamide gel (91% water by volume), approximately 2 mm thick, 9 cm long and 2.5 cm wide, was placed in a Plexiglas frame and covered with a polyethersulfone membrane with a 0.2 µm porosity. The resulting probe was preincubated in oxygen‐free H₂O (bubbled, covered with pure argon for 24 h) to remove all DO from the gel. The probe was then inserted into a sediment core and left for 24 h. Afterward, it was removed from the sediment, dismantled, and the gel sectioned into 5‐mm thick layers using a scalpel. The individual strips were extracted with H₂O (V = 2.5 mL) for 24 h in glass vials, after which they were transferred into another set of vials filled with 1% HNO₃ and extracted for an additional period of 24 h. The determination of the solutes in H₂O and HNO₃ followed, and the residual gel was discarded. The main anions and cations (Cl⁻, N⁻, NH4⁺, N‐NO₂ ⁻, N‐NO₃ ⁻, and SO₄ ²⁻) back‐equilibrated with H₂O were analyzed using ion chromatography (ThermoFisher Scientific, ICS 5000, Dionex) while other chemical elements (Al, Ca, Fe, K, Na, Mg, Mn, and P) were analyzed in both H₂O and 1% HNO₃ extractions with 8800 Triple Quadrupole ICP‐QQQ (Agilent Technologies) and sum together. The acid back‐extraction was used to dissolve artificial precipitates freshly formed inside of the gel after removing the probe from the anoxic sediment.

Each core had two minipeepers: one in stagnant water and one in moving water, which we exposed for 12 h. During this time, cores were covered with aluminum foil to prevent light exposure. DO microprofiles were recorded at the end of this experiment, 10 min after stopping the stirring, using a micro‐optode (PSt7, PreSens) operated with a micromanipulator (accuracy 10 µm) with a sampling rate of 1 Hz, and steps of 100 µm, and 90 s for equilibration. In the case of solutes measured with minipeepers, a concentration interface height was defined as the distance from the SWI at which porewater reaches half of its normalized solute concentration, that is, $\frac{C-C_{\min }}{C_{\max }-C_{\min }}=0.5.$

2.4. Stirred cores experiment—Recreating stratification cycles

To recreate the effect of destratification on benthic fluxes, we incubated two additional sediment cores with the same characteristics as those in the previous experiment. The water column of each core was heated on top by infrared heating panels. The dimensions of the electric heating mat for each core were 14 cm × 15 cm, with an output of 5 W. These were positioned on the upper surface of the cores to heat the water surface. Heating was applied for 22 h to develop stratification. After stopping the heat, water spontaneously started to mix by convection due to the temperature difference with the air. At this moment, one of the two cores was stirred to recreate the wind effect on the lake, while the other was left without stirring to simulate a calm night. This mixing period lasted for 2 h for both cores, allowing for both complete and incomplete mixing to be achieved. Then, the water was heated again in both cores for an additional period of 22 h (without stirring). Water temperature was monitored at six different depths, respectively (with the same thermistors as for field records).

To track solutes over time, water samples (2.5 mL) were collected from the water column by pipetting at surface water (approximately 5 cm below the water surface) and bottom water (approximately 5 mm above the sediment surface) at discrete intervals. In the latter case, water was slowly suctioned to prevent porewater withdrawal or sediment resuspension. To estimate benthic fluxes, we also collected integrated water samples (10 mL) at the beginning and end of the experiment. This was achieved by slowly suctioning water uniformly from the bottom to the surface throughout the entire water column. All water samples were filtered (glass fiber, 0.4 µm) and analyzed via ion chromatography and inductively coupled plasma mass spectrometry for the same elements as for minipeepers.

Fluxes for all solutes were estimated based on the concentration changes in the water column as follows:

$$\mathrm{Flux}=\frac{C_{\mathrm{f}}-C_{\mathrm{i}}}{A\mathrm{\Delta }t}V$$

where $C_{\mathrm{f}}$ and $C_{\mathrm{i}}$ are the final and initial concentrations (µg L⁻¹), $A$ is the contributing sediment area (0.025 m²), $\mathrm{\Delta }t$ is the time interval (48 h), and $V$ is the water volume (4 L).

3. RESULTS

3.1. Lake temporal stratification

The Municky reservoir exhibits a dynamic stratification‐mixing regime in summer (Figure 1). On average, it was completely mixed 55% of the time, with a frequency of 1.5 days. However, mixed or stratified periods can last up to 6 days depending on air temperature and wind conditions. In general, the lake stratifies during the day and mixes at night. This is due to the temperature difference between the air and the water surface, which leads to convective mixing. Nocturnal cooling sometimes does not lead to the mixing of the entire water column, as seen on July 23, but complete destratification is more common than incomplete mixing. Mostly, the interplay between wind action and air–water temperature differences defines destratification patterns.

FIGURE 1.

Summer stratification dynamics of Municky fishpond characterized by water temperature (left axis) recorded at different water depths, as shown in the legend. Each panel represents a month's time series, from top to bottom: June, July, August, September. The lake mixed 55% of the time (in green), mostly during the night (in gray). Wind episodes higher than 3 m s⁻¹ (in bars, right axis) explained most periods of several days without stratification.

The hour of the day when the wind blows influences the net result of lake stratification. An afternoon wind promotes stratification, while an early morning or night wind would cause destratification. This is because the wind increases the heat flow in both directions: either from air to water (warming) or from water to air (cooling), depending on the temperature difference. Figure 2 illustrates this process by comparing water and air temperatures (Figure 2a,b), wind speed and direction (Figure 2c,d), and water velocities (Figure 2e,f) from two periods of 5 days each: calm (Figure 2a,c,e) and windy (Figure 2b,d,f) conditions. Typical consecutive hot summer days lead to periodic diurnal stratification. Mixing occurs every night unless warm afternoon breezes blow beforehand (e.g., on August 10 at 8:00 p.m., August 11 at 5:00 p.m., and August12 at 1:00 p.m.). In this case, the lake remains stratified during the night with mixing delayed until dawn. On the other hand, windy conditions prevent stratification by lowering air temperatures and increasing water mixing. Strong winds blew in August from the 26th day at 7:00 a.m. to the 27th at 9:00 p.m. In situ wind speeds peaked at 11.6 m s⁻¹. The morning winds, which began blowing at 7:00 a.m., together with a significant shift in wind direction, from northeast (77°) to southwest (270°), prevented diurnal stratification on August 26. As the windy conditions persisted, the lake remained mixed for 2 days. The surface water velocity reached 7.9 cm s⁻¹ (1.7 cm s⁻¹ average) during this windy episode compared to 5.1 cm s⁻¹ (1.6 cm s⁻¹ average) under calm conditions. In both cases, bottom velocities remain below 5.7 cm s⁻¹, within the same order of magnitude as surface velocities (1.3 and 1.5 cm s⁻¹ average for windy and calm conditions, respectively). Therefore, these calm versus windy conditions show only subtle differences in terms of hydrodynamics, despite the observed differences in stratification.

FIGURE 2.

Contrast between calm (a, c, and e) and windy (b, d, and f) conditions in a shallow lake during late summer. Water and air temperature (a and b), wind speed and direction (c and d), and water velocities (e and f). Under calm conditions, mixing occurs every night unless warm afternoon breezes postpone it until dawn (e.g., August 10 at 8:00 p.m., August 11 at 5:00 p.m., and August 12 at 1:00 p.m.), while both surface and bottom water velocities remain below 5.1 cm s⁻¹. Under windy conditions, strong winds prevented stratification for 2 days, resulting in surface water velocities that peaked at 7.9 cm s⁻¹ but remained of the same order of magnitude as the bottom velocities.

3.2. Effect of mixing intensity

Experiments with stirred cores demonstrate that the mixing/no‐mixing regime in the bottom water alters the availability of nutrients across the SWI (Figure 3). Oxygen (Figure 3a), remained present throughout the water column and the uppermost 2 mm of the sediment. At the same time, mixing affects the concentrations of P and ammonia in the first 2 and 4 cm of sediment, respectively (Figure 2b,c), resulting in substantial decrease in concentration in the water above the sediment.

FIGURE 3.

Nutrient concentration gradients developed under different flow regimes across the sediment‐water interface (SWI) in an experiment with stirred cores. The oxygen microprofile (a) was measured with an optode sensor, while P (b) and ammonia (c) profiles were obtained with gel minipeepers. Note that oxygen was measured in another core replicate. Gradients for other compounds are shown in Figures S4 and S5.

In general, SWI concentration gradients of P, ammonia, and other solutes show no substantial difference among flow regimes (Figures 2c, 3b, and 4; Figures S4 and S5). However, compared to stagnant waters, establishing a flow regime considerably decreases the height of the concentration interface (about 1 cm for turbulent flow; Figure 4). Therefore, it can be hypothesized that hydrodynamics at these levels of turbulence act more like an on/off switch mechanism for benthic exchange rather than a continuous regulator by flow intensity.

FIGURE 4.

Gradients of normalized solute concentrations across the sediment‐water interface (SWI) measured with gel minipeepers during the stirred cores experiment. Concentration interface height was defined as the height at which porewater reaches half of its normalized concentration, that is, $(C-C_{\min })/(C_{\max }-C_{\min })=0.5$, which is depicted for the (inner) stagnant and (outer) mixed volumes of each sediment core.

We attribute the nutrient gradients observed at the SWI to biogeochemical processes, as inert compounds such as Cl and Na did not show an interfacial gradient (Figure S4).

3.3. Complete versus incomplete mixing

The experiment with heated cores resembles a cycle of diurnal stratification, nighttime mixing, and diurnal stratification again (Figure 5). During 2 h without heating, the two tested cores mix through convective cooling. The stirred core completely destratifies its water column by mimicking wind action, whereas the core without stirring does so incompletely through convective mixing only. In the latter case, a bottom layer remains unmixed. The water column of both cores remains under oxic conditions, with approximately 83% DO saturation (Figure S6), regardless of the development of stratification.

FIGURE 5.

A cycle of stratification, mixing, and re‐stratification is applied in the experiment with heated cores. During the mixing period (2 h without heating highlighted in gray), one core (a) was stirred to simulate wind action, while the second one (b) was left unstirred with convective mixing only. Temperature recorded at different depths as shown in legend.

Both experimental cores accumulate solutes in the benthic water due to stratification and lack of oxygen at the SWI. Whereas benthic concentrations keep increasing in the unstirred core, these fall in the stirred one (Figure 6c,d) due to water column homogenization.

FIGURE 6.

Time series of P (a and c) and ammonia (b and d) in surface water (a and b) and bottom water (c and d) of heated‐cores experiment. After the mixing period (in gray), the stirred core showed a substantial increase in both concentrations in surface water as compared to the non‐stirred core. On the contrary, the non‐stirred core accumulated nutrients in the bottom layer.

The higher concentrations of phosphorus and ammonia in the surface water layer of the stirred core demonstrate the importance of complete destratification loading nutrients into the water column (see Figure 6; Figure S7). Meanwhile, an increase in the concentration of P and ammonia in the bottom water of a water column that is not completely mixed might be observed (Figure 6; Figure S8). Therefore, incomplete destratification leads to the accumulation of nutrients in layers above the sediments, thus with low effect of sediment on surface waters. Over time, these benthic concentrations are expected to converge to an equilibrium with SWI concentrations. Thus, fluxes would decrease as the SWI gradient decreases.

Regarding other solutes present in porewater, each compound has a different gradient located at a different height. Those sharp gradients situated close to the SWI must result in high fluxes. The quantification of benthic fluxes (Table 2; Table S1) shows that, in general, complete destratification releases more solutes than incomplete destratification. In terms of limiting nutrients, it releases more P and ammonia, and consumes more nitrate. In both cases, a benthic layer concentrated with solutes remained at the end of the second stratification period (Figure 6; t = 48 h). We estimated the thickness of this layer based on the consistency between final concentrations obtained with point and integrated water samples (Section S.5 in the Supporting Information; Figure S9). The volumes of surface (mixed) and bottom (unmixed) layers were calculated by fitting their respective concentrations with the concentrations in integral sample, which covers the whole water column. Based on this model, we suggest concentration benthic layer thicknesses of 22.4 and 4.8 mm for the stirred and unstirred cores, respectively (Figure S10). The model is only used to estimate the thickness of the layer based on two concentrations, of which the one above the sediment is probably only an average value of the steep gradient at the sediment/water interface. We therefore do not assume a clear and stable separation of the two layers, as observed in temperature data (Figure 5).

TABLE 2.

Benthic fluxes (mg m² d⁻¹) from experimental cores based on solute concentration changes observed in integrated water samples after complete and incomplete destratification (see Table S1).

Destratification	Na	Mg	P	K	Ca	Mn	Fe	Cl	N‐NO2	N‐NO3	SO4	N‐NH4
Complete	98.8	75.1	1.21	60.9	619	2.51	−0.04	62.7	0.04	−17.8	68.5	45.1
Incomplete	90.6	57.4	0.56	45.6	466	2.76	−0.01	39.2	0.13	−15.3	−155	37.9
Difference (%)	9	31	115	34	33	−9	209	60	−69	16	−144	19

To summarize, complete destratification—for example, promoted by wind or strong convection—supplies the water column with nutrients that previously accumulated in a benthic layer during diurnal stratification. This process restores SWI gradients for the next stratification period, which enables high desorption and ammonification rates combined with weak solute diffusivities. After a subsequent cycle, a new concentrated bottom layer, ranging from millimeters to centimeters in scale, remains just above the sediment (likely unavailable for algae uptake) unless disrupted again by mixing. The periodicity of this gradient restoration affects the overall nutrient concentrations in the lake.

4. DISCUSSION

4.1. Recurrent mixing promotes release

The studied shallow lake exhibits a dynamic stratification and mixing regime during summer (Figure 1), indicating a significant influence of hydrodynamics on the exchange of nutrients with sediments. Experiments with lake sediment incubated in the lab show that hydrodynamics generally affect SWI concentration gradients as an on/off mechanism. This was evidenced by vertical porewater concentration profiles of nutrients, which show subtle differences among laminar, transition, and turbulent conditions but significant as compared with stagnant conditions (Figure 3). Lab experiments resembling a stratification/mixing cycle showed an accumulation of nutrients in benthic waters that were released after complete destratification (Figure 6). Surprisingly, slow mixing performed under oxic conditions in the water column enhanced release of nutrients from sediments. Stratification affects oxygen concentrations in porewater (Figure 3a) but maintains it at levels that are high enough to impact P dissolution and ammonia production (Osafo, 2016); meanwhile solute diffusivity rates in benthic waters can decrease by three orders of magnitude (Lorke et al., 2002). As a result, nutrients accumulate in porewater and the benthic layer, causing an upward migration of the SWI concentration gradient. However, as the porewater equilibrates with sediment particle‐surface concentrations, accumulation rates slow down and attenuate this vertical migration of the nutrient concentration gradients. On the contrary, mixing episodes increase turbulent diffusivity. As a result, available‐accumulated nutrients diffuse into the water column, causing a downwards migration of the SWI gradient, which restores the conditions for a new release cycle (Figure 7a).

FIGURE 7.

Schematic of solute concentrations in a temporarily stratified shallow lake. (a) Vertical profile of solute concentrations across the sediment‐water interface (SWI). Theoretically, the SWI concentration gradient migrates upward during stratification and downward during mixing. The alternation between these two scenarios should promote interfacial exchange. (b) Hypothetical concentration time series of an inert and a reactive solute in the water column. A reactive solute, such as P, undergoes significant changes after mixing episodes with complete destratification (in gray).

The alternation of these two scenarios should promote nutrient release. Indeed, the number of (complete) mixing episodes roughly predicted P concentration changes observed in the lake from the beginning of June to the end of July, before primary production started to increase substantially (not shown). Considering that the experiment with complete mixing released 1.6 mg of P m⁻² cycle⁻¹ (P concentration increased from 3 to 13 µg L⁻¹ in a volume of 4 L from a sediment surface of 0.025 m²), each cycle should increase P concentration in the lake by 2.0 µg L⁻¹ (assuming 0.8 m water average depth). Accounting for 37 mixing episodes observed during this period, it yields an expected concentration increase of 74 µg L⁻¹ (112 µg L⁻¹ if the lake had mixed daily) compared to the 26 and 55 µg L⁻¹ registered in the lake surface and bottom waters, respectively (unpublished raw data from lake monitoring). Since P is a highly reactive nutrient, we expect that mixing episodes affect its concentration in the water column in a distorted staircase fashion (Figure 7b). As many natural and dynamic processes occur in lakes, laboratory incubation experiments cannot simulate and mimic all of them. Thus, the transfer of results to a natural situation is never straightforward. One such issue is the homogeneity of sediment and its diagenetic profiles, which were not maintained in our experiment. However, due to the high abundance of common carp in the pond and their benthic feeding habits, the upper 10 cm of sediment is likely to be constantly homogenized. According to Orihel et al. (2017), our estimation of P fluxes falls within the range of fluxes expected for oxic conditions. Still, it is relatively low, considering the lake's trophic level and pH conditions (around 8.1, based on lake monitoring measurements). We attribute these differences to the difficulty of reproducing in situ conditions in the experimental cores.

The periodicity of this alternation between stratification and mixing regimes is significant, as solute accumulation in benthic water and porewater occurs on a timescale of hours to days during stratification periods. Meanwhile, the restitution of the SWI solute concentration gradients should occur on a timescale of minutes to hours (almost spontaneously) during mixing, after complete destratification.

4.2. Implications

As a net result, periodic mixing should cause more nutrient release than either persistent stratification or continuous mixing. Whereas we only partially tested this hypothesis through experiments with stirred cores, other studies converge to a similar conclusion (Søndergaard, Nielsen, Skov, et al., 2023; Welch & Cooke, 2005). In this respect, the frequency of stratification disruption appears to be more relevant than the intensity of mixing. As shown experimentally, mixing intensity mostly affects solute availability in porewater and slightly increases penetration depth. As the porewater concentration gradient migrates upward during stratification, porewater concentrations equilibrate with those of nutrients in sediment particles, thereby recessing the particle‐porewater equilibration processes.

The timing and intensity of nutrient release should have an impact on algal growth, depending on light conditions and nutrients concentrations. We expect that algae might uptake and accumulate nutrients during nocturnal mixing and use them later during diurnal photosynthesis, according to their quota flexibility (Grobbelaar, 2003). On the other hand, phytoplankton can decrease the concentration of available nutrients during the stratification period, resulting in lower primary productivity, and the accumulation of nutrients, along with reduced compounds in the benthic water. Both oxygen production resulting from the increased nutrient concentration in the euphotic zone, eventually limited by turbidity, and oxygen consumption for oxidation of reduced compounds accumulated at the benthic water appear simultaneously. The final oxygen concentration will depend on the fitness of phytoplankton and the amount of reduced compounds. When a longer stratification period occurs, a lower final oxygen concentration will be in the water column.

Other practical implications refer to lake monitoring. Our results highlight the importance of employing careful sampling strategies, particularly in terms of the chosen times and depths of measurements, as previously noted by Goncharov et al. (2024). Sediment incubation experiments suggest the existence of a benthic layer that is enriched with nutrients (approximately 2–50 times more concentrated than surface water). Its thickness typically ranges from millimeters to centimeters and may change over time due to stratification or mixing conditions. Ignoring the presence of this layer might affect the quantification of nutrients available within the water column of a given water body under research. Moreover, the stratification dynamics observed in the lake not only challenge in situ sampling strategies in shallow waters, but also sediment flux estimations based on incubation of sediment cores. Most sediment incubations are performed under stagnant conditions either with aerobic or anaerobic treatment, while studies considering mixing are less common (Viollier et al., 2003). Our experiments show higher benthic nutrient fluxes for complete than for incomplete destratification, both under aerobic conditions. Thus, nutrient flux estimates based on sediment incubations should consider this effect by taking into account realistic shallow water hydrodynamics.

After studying the mixing patterns of a shallow, polymictic lake in summer and assessing its potential effect on solute exchange with sediments, key open questions arise. The solute release rates from sediments to porewater are not yet fully understood. We do not know yet if interfacial porewater and benthic water concentrations would recover within 36 h after a mixing episode, which is the average mixing frequency of the studied lake. Mesocosm experiments have definitively demonstrated this over periods of 7 days under anaerobic conditions (Wilkes, 2019), which spans a much larger timescale than the one studied here. In our experimental cores, ammonia stabilized its concentrations in benthic water within approximately 30 h under incomplete stratification; meanwhile P concentration kept increasing after 48 h (Figure 6). This time was required to equilibrate with the water volume of the experimental core. In the lake, it should take a different time depending on the differences in water depth and vertical diffusivity between the core and the lake.

The findings presented here are of relevance for aquaculture, as they contribute to a better understanding of the internal nutrient cycling in shallow lakes. In turn, the temporality and intensity of this cycling, among other factors, should explain sudden changes in primary production that are sometimes unpredictable in these waterbodies.

5. CONCLUSIONS

The stratification regime in shallow lakes can be very dynamic with recurrent mixing episodes. For example, our study lake was stratified 45% of the time and mixed every 1.5 days on average. However, both scenarios could last up to 6 days. Lake surface and bottom velocities exhibit similar order of magnitude, regardless of whether conditions are calm or windy. Flow intensity affects benthic fluxes more as an on/off switch rather than as a gradual regulator. We deduce, from laboratory experiments with heated and stirred sediment cores, that during nighttime cooling, abrupt destratification episodes can release more nutrients from sediments to surface waters than those occurring slowly, resulting in incomplete destratification before sunrise. Therefore, recurrent mixing episodes must be key drivers of benthic fluxes. The nutrient release was not directly linked to oxic/anoxic cycling at the sediment surface, followed by Fe‐P compounds dissolution, as is often described, but the complete mixing also enhanced the flux of electron acceptors, including oxygen. Depending on the frequency of mixing, the benthic layer can concentrate, accumulate, and release solutes periodically into the water column, thereby pumping up nutrients to meet the demand of algae. Further studies are needed to evaluate this mechanism at different periodicities, in conjunction with algae growth.

AUTHOR CONTRIBUTIONS

Jiří Jan: Conceptualization; data curation; investigation; methodology; writing—original draft. Felipe Breton: Data curation; formal analysis; methodology; visualization; writing—original draft. Jakub Borovec: Conceptualization; funding acquisition; investigation; methodology; project administration; resources; supervision; validation; writing—review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Supplemental material

Jan, J. , Breton, F. , & Borovec, J. (2025). Shallow‐lake sediments release nutrients by complete destratification events. Journal of Environmental Quality, 54, 2061–2073. 10.1002/jeq2.70092