Effects of Macrofaunal Recolonization on Biogeochemical Processes and Microbiota—A Mesocosm Study

Abstract

Macroinvertebrates are widespread in lake sediments and alter sedimentary properties through their activity (bioturbation). Understanding the interactions between bioturbation and sediment properties is important given that lakes are important sinks and sources of carbon and nutrients. We studied the biogeochemical impact of macrofauna on surface sediments in 3-month-long mesocosm experiments conducted using sediment cores from a hypoxic, macrofauna-free lake basin. Experimental units consisted of hypoxic controls, oxic treatments, and oxic treatments that were experimentally colonized with chironomid larvae or tubificid worms. Overall, the presence of O₂ in bottom water had the strongest geochemical effect and led to oxidation of sediments down to 2 cm depth. Relative to macrofauna-free oxic treatments, chironomid larvae increased sediment pore water concentrations of nitrate and sulfate and lowered porewater concentrations of reduced metals (Fe²⁺, Mn²⁺), presumably by burrow ventilation, whereas tubificid worms increased the redox potential, possibly through sediment reworking. Microbial communities were very similar across oxic treatments; however, the fractions of α-, β-, and γ-Proteobacteria and Sphingobacteriia increased, whereas those of Actinobacteria, Planctomycetes, and Omnitrophica decreased compared to hypoxic controls. Sediment microbial communities were, moreover, distinct from those of macrofaunal tubes or feces. We suggest that, under the conditions studied, bottom water oxygenation has a stronger biogeochemical impact on lacustrine surface sediments than macrofaunal bioturbation.

1. Introduction

Bioturbation is defined as the translocation of particles (reworking) and movement of water (ventilation) through sediments by living organisms [1]. Ventilation causes the advective and diffusive exchange of solutes between sediments and overlying water (bioirrigation). This leads to input of oxygen (O₂) into otherwise anoxic sediment layers and can cause fluctuations between anoxic/oxic conditions within sediments [1,2]. Reworking due to macrofaunal burrowing, feeding, and defecation activities moves sediment particles in an undirected or directed manner and can thereby mix, oxidize, and introduce organic matter to surface sediments [1,2]. The dominant mode of bioturbation, and thus the biogeochemical and sedimentological impact of bioturbation on sediments, varies between different macrofaunal species and habitats [1].

In addition bioturbation affects biogeochemical processes by influencing microbial activity [3]. Mechanical breakdown of large into small, more accessible organic particles during macrofaunal feeding promotes microbial colonization and organic matter mineralization [4]. Pumping of high-energy electron acceptors, e.g., O₂, nitrate (NO₃⁻), by ventilation into deeper layers, where these electron acceptors are otherwise absent, interferes with the vertical separation of microbial respiration reactions based on energy yields [5] and causes coexistence of otherwise competing reactions [6,7]. Ventilation can also stimulate microbial activity by removing toxic metabolites, such as H₂S [8], or by providing food to microorganisms (“gardening”) [9]. In addition, macrofaunal reworking can decrease microbial populations, e.g., through grazing and deposit-feeding [1]. While most studies have focused on the impact of bioturbation in marine sediments, the effects of bioturbation on freshwater sediments are less studied and understood. The main bioturbators of freshwater sediments include fish, clams, insect larvae and oligochaetes [10,11].

Particularly chironomid larvae and oligochaetes are important and abundant bioturbators across many freshwater sediments [12]. Many chironomid larvae build sedimentary tubes from salivary silk and ventilate these with overlying water, thereby causing bioirrigation of the surrounding sediment [13,14]. Ventilation by chironomid larvae also creates redox interfaces and affects nutrient cycling [1,15] by increasing fluxes of phosphate (PO₄³⁻) and ammonium (NH₄⁺) to bottom water [16,17], or by promoting phosphorus removal by adsorption [18]. Input of O₂ by ventilation may also stimulate nitrification and subsequent denitrification in surrounding sediment [19,20].

The majority of freshwater oligochaetes fall into the family Tubificidae. Many members of this family build dense burrow networks into deeper parts of the sediment [21] and feed head-down several centimeters below the surface and defecate at the sediment surface [12]. This activity transports reduced sediment to the surface, where the sediment rapidly oxidizes. The subsequent downward transport of oxidized sediment through the addition of new fecal layers increases the redox potential in surface sediments [22]. Though tubificids typically affect nitrogen (N) and P cycling less than chironomids [20,23], they can also increase the efflux of phosphate and ammonium to overlying water [23,24] or influence rates of nitrification and denitrification [20].

Increased anthropogenic nutrient inputs, especially P, stimulate water column primary production and thereby increase organic carbon (OC) loading and OC burial in lakes through a process called eutrophication [25,26,27]. Increased OC loading increases O₂ consumption and can lead to water column hypoxia or even anoxia, and thereby severely impact ecosystem functioning [27]. Periods of hypoxia or anoxia for weeks to months kill macroinvertebrates or force these to retreat to oxic parts of lakes [28,29]. Decreased bottom water O₂ concentrations furthermore alter macrofaunal communities due to differences in physiological tolerance to hypoxic or anoxic conditions among macrofaunal species [30]. Thus, macrofaunal assemblages are studied as indicators of eutrophication [10,31]. To prevent anoxia, the decrease of P input and artificial aeration of water columns have been implemented and led to the macrofaunal recolonization of sediment in eutrophic lakes [27].

So far, little is known about the impact of sediment macrofauna on microbial community structure in lake sediments, even though the observed effects of macrofauna on biogeochemical processes are largely mediated by microorganisms. Among the limited number of studies, several have found that microbial population size does not change or increases slightly due to lacustrine sediment bioturbation [32,33], while others have suggested that predation by oligochaetes may reduce microbial population size [21]. Further studies have shown that bioturbation locally changes sediment bacterial community structure [34,35], e.g., by leading to elevated abundances of methane- and iron (Fe)-oxidizing bacteria in tubes of chironomid larvae [9].

In this study, we investigate experimentally how the switch from severe bottom water hypoxia to oxic conditions, as occurring during the reversal of eutrophication (oligotrophication) or due to artificial aeration of eutrophic lakes, and subsequent colonization by chironomid larvae and tubificid worms, affects sediment biogeochemistry and microbial community structure. To this end, we artificially oxygenated and macrofaunally colonized sediment cores from the hypoxic, macrofauna-free deep basin of Lake Zurich (Switzerland) in the laboratory. Over a period of 3 months, we compare the response of redox-sensitive biogeochemical processes and microbial community composition across hypoxic controls (C), oxic treatments (O), oxic treatments supplemented with chironomid larvae (L), and oxic treatments supplemented with tubificid oligochaete worms (W). We study the impact of oxygenation and bioturbation based on (1) biogeochemical properties of solute (anions, cations) and solid phases (acid extractable Fe, Chl a, luminophore beads), (2) microprofiling of O₂, redox potential, pH, and H₂S, and (3) microbial community analyses based on 16S rRNA gene copy numbers and sequence compositions. Our results provide novel insights into how bottom water oxygenation and recolonization by macrofauna affect sedimentary C, N, P and Fe cycling and microbial community composition in lake sediments.

2. Materials and Methods

2.1. Core and Macrofaunal Sampling

We obtained 24 sediment cores from the deep basin of Lake Zurich at 137 m water depth (47°16.995 N, 8°35.624 E). The bottom water at the station is hypoxic and sediments are devoid of macrofauna [36]. Cores were on average ~50 cm long and were obtained by gravity coring using 60 cm long and 15 cm wide core liners (UWITEC, Mondsee, Austria). Immediately after retrieval, cores were capped and cooled in ice water during transport to the laboratory at Eawag (Kastanienbaum, Switzerland), where cores were stored at 10 °C until the start of the experiments. An additional 10 sediment cores were obtained from both the shallow sublittoral zone of Lake Lucerne at 24 m water depth (47°00.051 N, 8°20.218 E) and from the deepest part of Lake Baldegg at 66 m water depth (47°11.929 N, 8°15.613 E). These two stations were chosen because of previously described macrofaunal abundance and community structure. Macrofauna at the station in Lake Lucerne almost exclusively consist of chironomid larvae (dominant groups: Procladius sp., Micropsectra sp., Macropelopia fehlmanni, Tanytarsus sp., Sergentia coracina), whereas the station in Lake Baldegg is dominated by oligochaete worms (dominant groups: unclassified Tubificidae without bristles, Limnodrilus hoffmeisteri, L. profundicula) [37]. Cores with macrofauna were also cooled with ice water during transport and stored at 10 °C at the laboratory, however, overlying water was bubbled with air to prevent hypoxia. Macrofauna were sampled by sieving through 200 µm mesh.

2.2. Experimental Design

Cores were incubated inside a custom-made flow-through aquarium through which lake water from 42 m water depth in Lake Lucerne (~9 °C) was constantly pumped at 80 l/h (Figure 1 and Figure S1). The inorganic tracer bromide (Br⁻) was injected to the inflow at a constant rate using a peristaltic pump (final concentration: 500 µM) and used to track ventilation. Before starting the experiments, luminophore beads were added to the sediment surface to trace sediment reworking [38].

There were six replicate cores for each experimental unit. Hypoxic controls consisted of sediment cores with core caps. For all oxic treatments caps were removed to allow for exchange with O₂-rich lake water. Chironomid larvae treatments (L) treatments were inoculated with 30 chironomid larvae from Lake Lucerne (~1700 individuals per m²), whereas worm (W) treatments were inoculated with 150 oligochaetes from Lake Baldegg (~8500 individuals per m²). These macrofaunal abundances corresponded to ~150% of the natural abundances at the sampled stations [37]. Sampling was performed after 6 (T1), 14 (T2), 27 (T3), 43 (T4), 63 (T5), and 82 d (T6). At each timepoint one core per treatment was destructively sampled. To minimize spatial biases, cores were arranged randomly inside the aquarium.

Due to decreases in chironomid larvae abundances at 14 and 27 d, where only 6–12 larvae were recovered in L treatments, we added additional larvae (~20) at day 31. In all subsequent timepoints we recovered >25 larvae per core during sampling. The initial declines might have been caused by nightly migrations into the water column [12] followed by washout, by mortality, including predation by Procladius sp. and M. fehlmannii, or by metamorphosis.

2.3. Monitoring and Sampling

The dynamics of pH, redox potential, and O₂ in the top 0–4 cm of sediment were monitored weekly and then bimonthly using microsensors. Additional measurements were performed 1–2 days prior to destructive sampling.

For pore water and sediment sampling, cores were taken out of the aquarium. Bottom water was sampled using a syringe, porewater from 0.25, 0.5, 0.75, 1, 1.5, 2 cm was sampled using microrhizons (0.2 µm pore size, Rhizosphere), and at 3, 4, 6, 8, 12, 16, 20 and 28 cm using normal rhizons (10 cm porous part, 0.2 µm pore size, Rhizosphere). All rhizons were inserted horizontally via predrilled holes (sealed by transparent tape) on the side of the core liner. To prevent air (O₂) contamination, microrhizons were flushed with nitrogen gas immediately before insertion, whereas for normal rhizons with 3-way stop cocks the initial 1 mL of pore water was discarded. Furthermore, starting at 43 d, modified normal rhizons, that had been sealed at the base with a 2-cm piece of shrink tubing, were used. These modified rhizons effectively prevented contamination of porewater samples with bottom water that in some cases had been drawn down between the liner and sediment core. All pore water samples were immediately fixed using 1 M HCl for cations (Fe²⁺, Mn²⁺, NH₄⁺; final pH: 4–5) and 1 M NaOH for anions (SO₄²⁻, NO₃⁻, PO₄³⁻, Br⁻) and stored at −20 °C.

After pore water sampling, sediment ‘subcores’ were taken from the top 0–6 cm of sediment by vertically inserting 5 and 20 mL cut-off syringes. The 5 mL syringes were closed with parafilm and immediately frozen at −80 °C for later analyses of DNA/RNA, solid-phase Fe pools, and sediment solid phase redox state (i.e., electron donor capacity (EDC)/electron acceptor capacity (EAC)). The 20 mL syringes for analyses of luminophores and chlorophyll a (chl a) were frozen at −20 °C. Prior to laboratory analyses, frozen subcores were extruded and sectioned into their appropriate depth intervals. Deeper sediment layers (6, 8, 12, 16, 20, 28 cm) were stored in the same way, but were sampled by extruding sediment and horizontally inserting cut-off syringes at each target depth.

2.4. Analyses

2.4.1. Microsensor Analyses

We determined the vertical distributions of O₂, redox potential, pH, and H₂S (at 82 d only) with 100 µm tip size microelectrodes using a field multimeter system with a micromanipulator (Unisense, Aarhus, DK). We always measured two sets of cores (time points) at each monitoring or before each sampling. Per each core and time point, multiple (3–5) replicate profiles were measured to check for spatial heterogeneity. The pH ranged from 7.7 to 6.6 and had distinct, diagenetically controlled profiles with no local peaks indicative of active cable bacteria (profiles in Figure S2). H₂S concentrations were below detection to very low (<3.5 µM) (detailed profiles in Figure S3).

2.4.2. Pore Water Analyses

[Fe²⁺] and [Mn²⁺]. Pore water concentration profiles of Fe²⁺ and Mn²⁺ were measured to determine the distributions of microbial Fe and Mn cycling. In total, 2 mL of pore water, that had been fixed with 20 µL of 30% HCl (Suprapur, trace metal certified, Sigma-Aldrich, Darmstadt, Germany), were analyzed by Inductively Coupled Plasma-Optical Emission Spectroscopy (5100, Agilent Technologies, Santa Clara, USA) after dilution with Milli-Q water and compared to multi element standards (solution IV, MERCK, Certipur, Darmstadt, Germany).

[NH₄⁺]. Concentrations of NH₄⁺ were determined spectrophotometrically on 96 well plates according to Kempers and Kok [39]. The absorbance of blue indophenol complexes (modified Berthelot reaction) was measured in duplicates on a spectrophotometer (Synergy HT, BioTek photometer, Bad Friedrichshall, Germany) at 640 nm against a blank (MQ, pH 4). Standards and blanks were made using ammonium chloride (Sigma-Aldrich, analytical grade) dissolved in Milli-Q water.

[SO₄²⁻, NO₃⁻, NO₂⁻, Br⁻]. Inorganic anion concentrations were measured on an ion chromatograph (IC, DIONEX DX-ICS-1000, Thermo Fisher Scientific, Waltham, USA) equipped with an AG14A guard column, an AS14A main column and a 4 mm suppressor (ADRS). The eluent was a solution of 8 mM Na₂CO₃ and 1 mM NaHCO₃ prepared on the day of the measurement. Standards were prepared for each ion separately by dissolving their sodium salts (all Sigma-Aldrich, analytical grade) in Milli-Q water. [NO₂⁻] was negligible throughout the experiments and is not presented in the results.

2.4.3. Solid-Phase Analyses

Bioavailable sedimentary iron. For quantifications of biologically available particulate Fe(II) and Fe(III), frozen subcores were cut into 2 mm slices, immediately placed into 5 mL HCl (0.5 M), and weighed. After extraction (30 min, shaking) the extract was split into two subsamples. To determine Fe(II), 40 µL of extract were mixed with 2 mL of 0.02% ferrozine in 50 mM HEPES at pH 7 [40] and then quantified photometrically on a Plate Reader (Synergy HT, BioTek). For total Fe, 1 mL of extract was mixed with 0.2 mL hydroxylamine (1.5 M) to reduce Fe(III) to Fe(II) and then measured as above. Fe(III) concentrations were the difference of total Fe and Fe(II). Standards consisted of dilution series of 100 mM FeSO₄ in 0.5 M HCl. To get insights into more crystalline Fe phases, sediment was sequentially also extracted using 3M HCl for 24 h at 60 °C for cores sampled after 27, 63, and 82 d. Afterwards, extracts were pH corrected using NaOH and analyzed as above, except that the reducing agent was 6.25 M hydroxylamine.

To determine changes in the redox state of the sediment with depth, we quantified the electron donating capacity (EDC) and electron accepting capacity (EAC) using mediated electrochemical analysis [41,42]. Further information about the experimental setup and analysis are given in Text S1.

Please see Text S1 for further information.

Luminophore beads. To quantify macrofaunal sediment reworking, luminophore beads consisting of fluorescent paint-covered silt particles with 30–40 µm diameter (D50: 30–40 µM, Environmental Tracing, Helensburgh, UK) were added at the start of the experiment. Depth distributions of beads (0.25, 0.5, 0.75, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, and 6 cm) were analyzed after 6, 43, 63 and 82 d. Frozen sediments were extruded from 20 mL cut-off syringes and sliced into 2 mm depth intervals. Samples were weighed, dried at 50 °C, and homogenized. Afterward each sample was placed in a plastic petri dish and distributed evenly by gently shaking until a monolayer of luminophore beads was observed. Photographs of the whole petri dish were taken using a Sony 6000a camera under UV light and all luminophores on the pictures were counted by eye in Image J. For each petri dish three pictures were taken, with petri dishes being shaken between pictures.

Chlorophyll a. Sediment chl a was quantified after 82 d as a proxy for the content of fresh algal organic carbon (OC). Chl a was extracted from ~0.3 g of wet sediment in the dark for 20 h with 90% acetone (2.5 mL) at −20 °C [43], and quantified by spectrophotometry (UV-VIS, VARIAN, Cary, 50 Scan, Agilent Technologies, Santa Clara, SC, USA) before and after acidification according to Lorenzen [44] to quantify contents of chl a and pheopigments. Ratios of chl a to pheopigments were used as a proxy for OM freshness.

Nucleic acid extraction from sediment. DNA and RNA were extracted simultaneously following lysis protocol I in Lever, et al. [45] using 100 µL of 10 mM sodium hexametaphosphate to prevent adsorptive losses. The detailed extraction protocol can be found in the, Texts S2 and S3.

DNA extraction from macrofauna. Prior to extraction, macrofaunal samples (whole specimens of larvae and worms) were cleaned with molecular grade water and cut into pieces using an ethanol wiped and flame sterilized scalpel. From a subset of specimens the gut was separated from the rest of the body using hypodermic sterile needles and extracted separately (see also [37]). Macrofaunal DNA was extracted following lysis protocol I in Lever, et al. [45]. For further details see Text S2.

Quantitative polymerase chain reaction (qPCR). For methodological information on the reverse transcription and quantification of rRNA gene copy numbers, we refer the reader to section Text S4.

Next Generation Sequencing (NGS). NGS was performed on a MiSeq (Illumina Inc., USA) using 10% PhiX. For 16S, the universal prokaryotic 16S rRNA gene primer pair Univ519F (CAG CMG CCG CGG TAA) and Univ802R (TAC NVG GGT ATC TAA TCC) [46,47] was used. For details of library preparation please see Text S5.

2.4.4. Bioinformatic Analyses

A first quality check with FastQC was performed followed by trimming read ends and merging the pairs into amplicons using the software seqtk and flash (N:103, usearch v11.0.667_i86linux64, Trim R1: 15, Trim R2: 20, FLASH v1.2.11, Min Overlap: 15, Max Overlap: 300, Max Mismatch Density: 0.25), data were dereplicated to obtain unique amplicons using usearch:fastx_uniques. Primer sites were trimmed using usearch (usearch v11.0.667_i86linux64, Amplicon range: 100–600, Number of mismatches: 1, Coverage: full-length, Wildcards enabled: IUPAC codes). Afterwards a quality filtering step using prinseq (PRINSEQ-lite 0.20.4, Size Range: 100–600, GC Range: 30–70, Min Q Mean: 20, Number of Ns: 0, Low Complexity: dust/30) was performed. Then ZOTUs were generated using usearch (UNOISE3), 97% clustering was performed (UPARSE), and a denoising (error correction) was performed using usearch:unoise3. Amplicons were mapped to ZOTUs to generate count tables using usearch:otutab. Bacterial phylogenetic assignments were done with SINTAX using the SILVA_128 database (v11.0.667_i86linux64). Archaeal 16S rRNA gene assignments were done in ARB (www.arb-home.de) by neighbor-joining phylogenetic trees. The ARB database was based on a SILVA 16S database with manually optimized alignments that had been expanded with 16S gene sequences from whole-genome studies and updated to state-of-the-art phylogenetic nomenclature. All further analyses (NMDS, CAP) were performed using RStudio under R version 3.5.2 and packages phyloseq, vegan, and ggplot2 [48].

2.4.5. Statistical Analyses

Pairwise comparisons between porewater-dissolved and solid-phase chemicals were performed based on Wilcoxon signed-rank tests to assess statistical differences between treatments.

3. Results

3.1. Visual Observations

The sediment initially retrieved from Lake Zurich was dark brown and clearly laminated in the upper 10 cm. With time, the surface turned to a lighter brownish color except in hypoxic control (C) treatments. This color change reached deeper layers and was more blurry in W (≥2 cm) compared to O or L treatments (0.75–1 cm) (Figure 2d,e). In W treatments a dense network of tunnels established to 20 cm within <7 days after the start of the experiment. Worms were often observed at the sediment surface, with their heads buried and tail ends actively undulating above the sediment water interface (Figure 2c). Fine, filamentous worm feces accumulated on the sediment surface over time (Figure 2a). After addition to cores, larvae quickly buried themselves into the sediment and subsequently remained hidden. Diverse burrow and tube types were formed, confirming the presence of diverse chironomid communities. Some burrows were only visible as holes from the sediment surface. Others had tubes (chimneys) that extended 1–2 cm above the sediment surface (Figure 2h). Ventilated, U-shaped burrows with a lighter color indicating sediment oxidation were also observed (Figure 2f). Larval feces consisted of roundish pellets that were deposited on the sediment surface (Figure 2b).

3.2. Microsensor Measurements of O₂ Concentrations and Redox Potential

Average O₂ microprofiles for 27, 63, and 82 d are shown in Figure 3 (for individual profiles and all time points, see Figure S4). O₂ microprofiles confirm the hypoxic, perhaps even anoxic, conditions in hypoxic controls. The O₂ at low concentrations in bottom water was likely introduced during the opening of cores for measurements. Across all oxic treatments, O₂ concentrations in the bottom water are between 240 and 260 μM. The O₂ penetration depth increases over time from ~3.4 to ~8.2 mm and was similar across oxic treatments, except that O₂ penetration depth was significantly (p < 0.05) lower in W compared to O and L treatments after 82 d. O₂ concentrations above the sediment surface, moreover, vary more between replicates in W compared to O and L treatments (Figure S5).

Redox potential (E_h) values ranged from +600 mV at the surface to −200 mV at 4 cm in all oxic treatments (Figure 4). E_h values in the C treatment showed reducing conditions with all values <40 mV. E_h values in O, L and W treatments ranged from +400 to +600 mV at the sediment surface. In O treatments, E_h decreased with depth to ~0 mV. The E_h in L and W treatments also decreased with depth, but more gradually. After 82 d, W treatments showed significantly (p < 0.05) higher E_h values below 2 cm compared to L and O treatments.

3.3. Pore Water Concentrations of Anions and Cations

Despite showing considerable fluctuations with depth, which could in part result from heterogeneous distributions of macrofauna and their activities, porewater geochemical profiles are influenced in consistent ways by the presence of O₂ and different macrofaunal groups.

3.3.1. Bromide (Br⁻)

Br⁻ was absent from controls as these treatments were kept closed throughout the experiment. O, L, and W treatments have similar bottom water Br⁻ concentrations (~400 µM) throughout the experiment, an indication that [Br⁻] was constant in tank water (Figure 5). Over time, Br⁻ penetration depths in O, L and W treatments increased from 8 (6 d) to >20 cm (82 d). Matching known bioturbation activities, Br⁻ penetration into deeper layers is fastest in L and slowest in O treatments, as evidenced by significantly (p < 0.05) higher Br⁻ concentrations in L compared to O treatments.

3.3.2. NO₃⁻ and SO₄²⁻

NO₃⁻ was nearly absent from hypoxic controls, whereas all oxic treatments had significant NO₃⁻ concentrations in bottom water (49–73 µM). Compared to O treatments, L treatments had significantly higher NO₃⁻ concentrations between 2 and 8 cm. W also had elevated NO₃⁻ concentrations after 82 d, but only in the top 2 cm. In hypoxic controls, bottom water [SO₄²⁻] decreased from 97 μM at the beginning to 7 μM after 3 months, whereas [SO₄²⁻] remained at ~160 µM in all oxic treatments. Similar to [NO₃⁻] treatments, [SO₄²⁻] in larval treatments was significantly (p < 0.05) elevated between 2 and 8 cm compared to W and especially O treatments (Figure 5).

3.3.3. Mn²⁺ and Fe²⁺

Dissolved Mn²⁺ and Fe²⁺ were present in bottom water of controls, but close to 0 µM in bottom water of all oxic treatments. In all treatments, [Mn²⁺] increased to ~15 cm (~250 µM) and stayed constant or decreased slightly below, whereas [Fe²⁺] increased consistently below 2 cm (Figure 5). Compared to O treatments, Mn²⁺ and Fe²⁺ concentrations were significantly (p < 0.05) lower from 2–8 cm in L treatments. This difference was most pronounced during the first two months.

3.3.4. NH₄⁺

NH₄⁺ was present in bottom water of controls but not oxic treatments, which also had lower concentrations in the top 2 cm (Figure S8). While [NH₄⁺] below 2 cm overlapped and fluctuated significantly with depth in all treatments, L treatments had significantly (p < 0.05) lower [NH₄⁺] throughout the experiments than all other treatments. W treatments only showed slightly (not significant, p = 0.22) lower [NH₄⁺] compared to O at later timepoints (>63 d).

3.4. Solid Phase Analyses

3.4.1. Iron and EAC/EDC

Depth profiles of the % contribution of Fe(III) to total bioavailable iron Fe(II), all extracted with 0.5 M HCl, are shown in Figure 6 (for contents of bioavailable Fe(II) and Fe(III) for all time points see Figure S9). Fe(II) dominated everywhere except in surface sediments of oxic treatments at later time points. Despite the higher redox potential of W treatments there were no clear differences in Fe(III) trends between O, L, and W treatments. Additionally, EDC analyses confirm the increasingly reducing conditions with sediment depth, and the lower ratio of EDC:EAC confirms the more oxidizing conditions in surface sediment—again with no differences between O, L, and W treatments (Figure S10).

3.4.2. Luminophores and Chlorophyll a

Luminophore counts and Chl a data suggest minimal sediment reworking by fauna and minimal differences in degradation rates of labile organic matter between treatments (Figures S11 and S12).

3.5. Microbial Community Composition

3.5.1. qPCR Values of Bacterial and Archaeal 16S rRNA Genes

Bacterial and archaeal gene copies showed little systematic variation with depth and over time (Figure S13). Bacterial copy numbers varied around 2 × 10⁹ copies per g⁻¹ wet sediment. The only noteworthy change was a decrease in gene copies in C and W treatments after 82 d. Archaeal gene copies showed the same pattern but had a lower average abundance (~1 × 10⁸ copies per g⁻¹ sediment). In addition, Bacteria-to-Archaea Ratios (BARs) increased over time in surface sediments of oxic treatments. This trend was also present in BARs that were calculated based on reverse-transcribed RNA copies (Figure S13).

In addition to these treatment-related sedimentary trends, we examined BARs in whole specimens, fecal pellets, and tubes/burrow walls of larvae and worms (Figure 7). Sediment BARs showed the lowest ratios (24.2 ± 24.3, n = 108). Worm specimens (185.3 ± 93.2, n = 10), tubes (187.0 ± 132, n = 13), and feces (176.8 ± 121.6, n = 9) had highly similar BARs that were on average 8–10 fold higher than bulk sedimentary BARs. Larval specimens (238.0 ± 161.3, n = 13) and feces (211.6; 256.5, n = 2) had similar BAR values to worm specimens and feces, however, larval tubes (53.2 ± 28.5, n = 6) had significantly lower BAR values that were intermediate between those of sediments and larval specimens and feces.

3.5.2. Amplicon Sequencing of 16S rRNA

An overview of the bacterial and archaeal read numbers (NGS) and zOTU numbers can be found in Table S1.

3.5.3. Sediment Microbial Community Zonation Patterns across Experimental Treatments

Bacterial and archaeal communities in hypoxic controls were highly similar to those recently reported for the same location [49] throughout the experiment but changed significantly in surface sediments of experimental treatments. Since bacterial and archaeal RNA sequence compositions were highly similar to archaeal 16S rRNA gene sequence compositions (Figures S14 and S15), we only discuss 16S rRNA gene sequence data in the following text.

3.5.4. Bacteria

NMDS analyses showed that at the end of the experiment bacterial communities in surface sediment (upper 1.5 cm) of all oxic treatments cluster away from the hypoxic control (Figure 8A). This difference was absent from deeper samples, where communities in all four treatments follow the same depth-related trend, indicating no impact on O₂ or bioturbation. Surface sediments of oxic treatments mostly aligned with concentrations of O₂ and other electron acceptors (SO₄^2–, NO₃^–), as well as redox potential (Figure S16). This suggests that changes in O, L, and W treatments are mainly driven by oxygenation and/or oxidation of surface sediment. In deeper sediment layers, bacterial communities correlated mostly with Fe²⁺ and other reduced compounds (NH₄⁺ and Mn²⁺). Phylogenetic comparisons of bacterial communities across the four treatments (Figure 9; for details see Text S7) suggest that among the dominant phyla (Proteobacteria, Bacteroidetes, Actinobacteria, Planctomycetes, Verrucomicrobia, Chloroflexi, Omnitrophica, Cyanobacteria, Latescibacteria) there was a gradual increase in α-(mostly Caulobacterales), β-(Nitrosomonadales, Burkholderiales and Rhodocyclales) and γ-(Methylococcales, Cellvibrionales) Proteobacteria in O, L, and W treatments through time. These increases were accompanied by decreases in δ-Proteobacteria (mostly Synthrophobacterales), Actinobacteria (Actinobacteria, Acidimicrobiia, Thermoleophilia), Planctomycetes (Planctomycetacia), Chloroflexi (Dehalococcoidi, Anaerolineae), Omnitrophica, Latescibacteria, and Cyanobacteria (Cyanobacteria, chloroplasts). By comparison, bacterial communities in hypoxic controls changed little with depth or over time.

3.5.5. Archaea

While archaeal communities also clustered according to sediment depth, no clear differences between the four treatments can be observed in NDMS plots at the end of experiments (Figure 8A). This is in line with CAP analyses, which showed that archaeal community composition does not strongly correlate with measured chemical data (Figure S16). Archaeal community depth profiles were stable through time (Figure 9; for details see Text S7) and indicate dominance of Pacearchaeota, Woesearchaeota, Diapherotrites and Euryarchaeota. While Pacearchaeota showed almost no depth-related changes, relative abundances of Woesearchaeota decreased slightly below 4 cm, whereas those of Altiarchaeales and Euryarchaeota increased with depth. Two uncharacterized phylum-level groups, which we call Unknown Asgardarchaeota Phylum III and Woesearchaeota-related phylum, had elevated abundances in surface sediments.

3.5.6. Comparison of Microbial Communities in Sediment, Faunal Specimens, Burrow Structures, and Feces

NMDS analyses indicate strong community differences between sediments and macrofaunal biomass, feces, and burrow structures (Figure 8B), with macrofaunal biomass representing a variable but distinct end member for larvae and worms in Bacteria and for larvae only in Archaea. Bacterial communities in burrow structures and feces clustered with or between macrofaunal biomass and oxic surface sediments in Bacteria. This pattern was also evident in archaeal communities from larval burrows and feces. By contrast, archaeal communities from worm biomass, burrows, and feces were strongly scattered and showed no clear clustering patterns.

Bacterial communities in tubes, feces, and faunal specimens at higher phylogenetic levels were also different from those in sediment and showed general trends that were similar to those in NMDS plots (Figure 9). Relative abundances of β-, ε- and γ-Proteobacteria were clearly higher, whereas contributions of Omnitrophica and Spirochaetes were lower in tubes of larvae and worms compared to surface sediments. Worm and larval feces were similar to worm and larval tubes, respectively, but had higher fractions of ε-Proteobacteria. Whole worms and larval samples had very different, less diverse and more variable bacterial DNA sequences than the other sample categories. Whole worms contained mainly α-, β-Proteobacteria and/or Fusobacteria, while whole larvae were dominated by α-, β- and/or γ-Proteobacteria or Firmicutes.

Archaeal communities of fauna, burrow structures, and feces were dominated by few archaeal taxa (Figure 9). Larval and worm tubes and feces were, similar to sediments, dominated by Diapherotrites, Pace- and Woesearchaeota. Archaeal communities of whole organisms were highly variable, with dominant members including, Pace- and/or Woesearchaeota in worms, and Diapherotrites, Eury-, Pace-, Thaum-, or Woesearchaeota in larvae.

4. Discussion

We investigated the effect of bottom water oxygenation and chironomid larval and oligochaete worm colonization in previously macrofauna-free, hypoxic sediment to gain a better understanding of how lake oligotrophication and bottom water reoxygenation affect biogeochemical processes and microbial communities in lacustrine surface sediment. We found that the presence of O₂ alters pore water and solid phase geochemical conditions in surface sediments and leads to a gradual but clear change in microbial community composition over time. We, moreover, showed that the presence of macrofauna adds to the effect of O₂ conditions, but was generally weak and more strongly influences biogeochemical parameters than microbial community composition. Larvae and worms had distinct effects on sediment biogeochemistry with larvae mainly altering pore water geochemical gradients, whereas worms raised the redox potential of the sediment. While differences in the microbial community composition were very small between O, L and W treatments, communities of microorganisms were highly distinct from sediments in macrofaunal feces, tubes, and biomass. In the following, we discuss the influences of O₂ and the added effects due to ventilation of larvae and reworking by worms on sediment biogeochemistry and the microbial community structure, as well as possible reasons for the observed weak treatment effects of macrofauna addition.

4.1. The Effect of Bottom Water O₂ and Macrofaunal Activity on Sediment Biogeochemistry

The presence of O₂ in sediment-overlying water was the main driver of changes in pore water and solid-phase biogeochemistry in our experiment. In hypoxic controls, the oxic lake water-derived electron acceptors NO₃²⁻ and SO₄²⁻ were strongly depleted relative to oxic treatments. Similarly, concentrations of redox-sensitive reduced solutes, Fe²⁺ and Mn²⁺ (Figure 3 and Figure 5), which were produced in the sediment cores through microbial Fe(III) and Mn(IV) reduction and were nearly absent in bottom water of oxic treatments, accumulated in water overlying hypoxic controls. The strong impact of O₂ was also evident from lower redox potentials (Figure 4), Fe(III) contents (Figure 6), and EACs (Figure S10) in surface sediments of controls compared to oxic treatments.

By comparison, the biogeochemical impact of macrofauna was minor. In contrast to other studies we observed no decreased O₂ penetration (Figure 3), which previous studies on chironomid larvae have interpreted to indicate increased aerobic microbial activity [23,50,51]. Presence of larvae elevated dissolved electron acceptor concentrations (NO₃⁻ and SO₄²⁻), especially from 2 to 6 cm. These increases were likely due to ventilation activities which led to bioirrigation and increased nitrification and S oxidation through the input of O₂ (Figure 5). Bioirrigation was likely also responsible for the decreased concentrations of reduced compounds (Fe²⁺, Mn²⁺, NH₄⁺) in the same layers (Figure 5). Bioirrigation likely diluted these compounds with compound-depleted bottom water and stimulated in situ chemical and biological oxidation with O₂ and possibly NO₃⁻.

The impact of chironomid larvae on porewater chemistry in the bioirrigated layer is supported by ratios of pore water [NO₃⁻], [SO₄²⁻], [Mn²⁺], [Fe²⁺], [PO₄³⁻] and [NH₄⁺] to the conservative tracer [Br⁻]. These indicate increased ratios of [NO₃⁻] and [SO₄²⁻] to [Br⁻], and decreased ratios of [Mn²⁺] and [Fe²⁺] to [Br⁻] in L treatments compared to O and W treatments from 2 to 8 cm (Appendix A, Figure A1). The absence of a clear larval impact in the top 2 cm is likely related to dominance of diffusive mixing and advection driven porewater exchanges with bottom water in this surface layer.

Compared to larvae, worms had an overall lower biogeochemical impact on surface sediments that mainly manifested itself in an increase in redox potential. This increase in redox potential was likely related to reworking through the observed head-down, bottom-up conveyor feeding, which visibly oxidized the top ≥2 cm of sediment. Similar increases redox potential have been previously noted for bioturbated treatments with the tubificids Tubifex spp. and C. riparius [35]. Yet, it remains unclear which chemical transformations are responsible for these increases in redox potential in W treatments. Clear shifts in measured iron pools from Fe(II) toward Fe(III) were absent (Figure 6), Figure S9), suggesting that—if significant oxidation of solid-phase Fe(II) occurred—it took place in (labile) iron pools that made up minor portions of bulk iron extracts. Even though it is known that sedimentary OM can also act as a redox-active phase [52,53], changes in EAC and EDC measurements do not indicate a significant impact of worms on the redox potential of organic compounds (Figure S10).

Notably, despite the increased redox potential and visible oxidation of the top ≥2 cm of sediment, luminophore counts suggest no significant reworking in W treatments. It is possible that reworking was mainly confined to the surface centimeter, though this does not explain the observed redox potential and color changes. Alternatively, the constantly observed, undulating movements of worm tails above the sediment surface may have significantly increased input of oxic overlying water into the top 1–2 cm of sediment and thus driven the observed changes in redox potential. Such locally enhanced exchanges between surface sediments and overlying water were, however, not visible in porewater concentrations of electron acceptors or reduced species. Thus, as for larvae, clear impacts of worm bioirrigation on porewater geochemistry in the top layer of sediment may have been masked by significant diffusive and advection-driven mixing with bottom water.

4.2. Effect of Bottom Water O₂ on Sediment Microbial Communities

Reflecting the fact that the biogeochemical impact of O₂ was strongest in surface sediments, microbial community changes in response to O₂ mainly occurred within the top 2 cm of sediment. Hereby statistical tests suggest changes in O₂, NO₃⁻, SO₄²⁻, and Fe²⁺ concentrations and redox potential as likely drivers of microbial community changes (Figure S16). O₂ presence can explain the increased relative abundances of chemoorganotrophic aerobes, such as Caulobacterales (α-Proteobacteria), Cellvibrionales (γ-Proteobacteria) or Sphingobacteriia (Bacteroidetes) [54,55,56,57] and aerobic methanotrophs (Methylococcales, mainly Methylococcaceae (γ-Proteobacteria) [58,59]. The latter increased ~10-fold in surface sediments of oxic treatments. Similarly, aerobic ammonium and nitrite oxidizers, such as Nitrosomonadales (β-Proteobacteria) [60] and Nitrospirae [61], increased by factors of ~7 and 2, respectively, in oxidized layers. In addition, sequence contributions of metabolically versatile Burkholderiales (β-Proteobacteria, mainly Comamonadaceae) [62,63] doubled, including sequences of the denitrifying genus Pelomonas [64], which suggests that besides nitrification also denitrification was enhanced in surface sediments of oxic cores. The increased importance of denitrification is matched by the increase in facultatively denitrifying H₂S oxidizing Thiotrichaceae (Thioploca, Beggioatoa) and Helicobacteraceae (Sulfuricurvum, Sulfurimonas) [65] in surface sediments of oxic treatments.

Matching the geochemical impact of O₂, the increase in aerobic and denitrifying taxa in surface sediments of oxic treatments was accompanied by decreases in likely anaerobes (Figure 8). These include groups typically growing under anoxic or low O₂ conditions such as Bacteroidia [66] (Bacteroidetes), Omnitrophica [67], Anaerolineae [68] and Dehalococcoidia [69] (both Chloroflexi), Clostridia [70] (Firmicutes), and Latescibacteria [71,72]. In addition, SO₄²⁻ reducing and syntrophic secondary fermenters such as Syntrophaceae (Syntrophus, Desulfomonile, Smithella and Desulfobacca) and SO₄²⁻ reducing Desulfarculaceae (both δ-Proteobacteria) decreased [73,74]. The decrease in SO₄²⁻ reducing taxa despite the increase in SO₄²⁻ concentrations in oxic treatments can be explained with the anaerobic growth requirements of most sulfate reducers.

Despite the importance of microbial iron cycling in sediments of Lake Zurich [49], microbes involved in Fe cycling were generally low in relative abundances. Slight increases in ferrous Fe oxidizers (Gallionellaceae, β-Proteobacteria, Nitrosomonadales) [75] suggest an increase in aerobic iron oxidation in oxic treatments. Yet, other iron-oxidizing Acidimicrobiales (Actinobacteria) [76] decreased, and taxa that are known to include iron reducers, i.e., Rhodobacteraceae (α-Proteobacteria), Holophagaceaea (Acidobacteria), Bacillaceae (Firmicutes), Deferribacteres and Geobacteraceae (δ-Proteobacteria,), Pseudomonaceae and Acidiferrobacteraceae (both γ-Proteobacteria) [77] were rare (<0.5% mra) and showed no trend with treatments or sediment depth.

Although archaeal populations decreased in response to O₂, in line with a previous study reporting clear dominance of Bacteria over Archaea in O₂-exposed surface sediments [78], we observe no clear changes in archaeal community structure over time (Figure 8 and Figure 9). A potential explanation is that changes in community composition to environmental changes, e.g., O₂ exposure, take longer in Archaea than Bacteria due to the long generation times of many archaeal taxa.

4.3. Effect of Chironomid Larval and Worm Bioturbation on Sediment Microbial Communities

Despite being comparatively minor, additional changes in microbial community composition occur in bioturbated treatments relative to macrofauna-free oxic treatments. Matching the more gradual redox profiles in L and W cores (Figure 2 and Figure 4) and previous studies reporting correlations between redox potential and microbial community structure [35,79,80], bacterial communities in bioturbated cores change more gradually from oxidized surface sediment to reduced deeper layers compared to non-bioturbated treatments (Figure 9). By contrast, clear changes in nitrifying and denitrifying Bacteria (Rhodocyclales, Nitrosomonadales, Burkholderiales or Rhizobiales) or nitrifying Thaumarchaeota, whose activities have been previously reported to increase in bioturbated sediments [20,81,82,83,84], are absent. The only notable exception is perhaps the increased percentage of methane-oxidizing Methylococcales in surface sediment (0–0.5 cm) of W treatments (16%) compared to oxic non-bioturbated treatment (7%), but this was only observed at the final time point. Thus—under the experimental conditions studied—O₂ presence in bottom water had a much greater impact on microbial community structure of bulk sediment than chironomid larval or tubificid worm bioturbation activity.

Instead, differences in microbial communities in bioturbated treatments were highly localized, being most evident in macrofauna themselves followed by macrofaunal feces and burrow structures (Figure 8 and Figure 9). Consistent with a recent study on the same locations [37], faunal microbiomes were dominated by putatively parasitic as well as mutualistic α-, β-Proteobacteria and/or Fusobacteria (worms), and α-, β- and/or γ-Proteobacteria or Firmicutes (larvae) (for a detailed discussion of potential metabolic functions and host–microorganism interactions see [37]). Interestingly, microbial communities in larval tubes and burrows were very similar to those in oxidized surface sediments (Figure 8 and Figure 9). This contrasts with previous studies documenting major differences in microbial communities between larval tubes and surrounding sediment [9,82] due to increased growth of methane-cycling microorganisms in tubes [9,15,85]. Rather than larval tubes or burrows, our study suggests that tubificid burrow communities, which have to our knowledge not been studied before, are clearly different from surrounding sediment. Compared to oxidized surface sediment or larval structures, worm burrows were more clearly dominated by β- and γ-Proteobacteria and also had higher percentages of ε-Proteobacteria. These differences in β-Proteobacteria were mainly due to an increase of Burkholderiales (unclassified Comamonadaceae and Oxalobacteraceae) and Methylophilales (unclassified Methylophilaceae) by a factor of 2–3. However most pronounced was the increase in ε-Proteobacteria, driven by a 16-fold increase in Sulfuricurvum (Helicobacteraceae) in worm burrows compared to oxic surface sediment. This suggests elevated proportions of aerobic and/or denitrifying, methylotrophic, and in particular S-oxidizing microbial taxa in worm burrows compared to surface sediment.

Microbial communities in worm and larval feces were very similar to those in corresponding burrow structures (Figure 8 and Figure 9). The only conspicuous difference was an on average higher percentage of ε-Proteobacteria in worm feces. This increase was mainly driven by a ~4-fold increase of S-oxidizing Campylobacterales (Helicobacteraceae, Sulfuricurvum) compared to worm burrows. Given the proposed preferential metabolism of proteinaceous organic matter by tubificid worms, the relative increase in S-oxidizing microbial taxa in feces could be related to the elevated release of reduced inorganic S species by microbial fermentation of S-containing amino acids.

4.4. Possible Reasons for Weak Bioturbation Effects

The observed macrofaunal impact on microbial community structure was low but matched the overall minor biogeochemical impact of W and L compared to O treatments. Other studies have reported more pronounced effects of bioturbation on nutrient fluxes and microbial reaction rates in freshwater sediment [3,78,82].

A possible reason for the minor impact of chironomid bioturbation is that we used diverse, natural assemblies of chironomid species at close to natural abundances [37]. Chironomid larval taxa at the sampling site [37] are dominated by generalistic surface detritus-feeding and filter feeding taxa (e.g., Micropsectra sp., Tanytarsus sp.) [86], as well as predators (e.g., Procladius sp.). By comparison, most previous studies added mono-species communities of, e.g., Chironomus plumosus or C. riparius, at much higher population densities (e.g., 14,000 individuals per m² in [87] compared to ~1500 individuals m⁻² in our experiments), and with larger body size and body mass. The average body length of our chironomid larvae was around 1–1.5 cm, which is considerably smaller compared to 2–2.5 cm body lengths reported for 4th instar larvae of C. plumosus in many previous studies [50,88,89]. This size difference could be important, as bioirrigation rates are known to increase with size in chironomids [90].

As for L treatments, we also used natural assemblages of oligochaetes in W treatments. The dominant oligochaetes at the sampling site were L. hoffmeisteri and unidentified Tubificidae [37]. L. hoffmeisteri feeds mainly at 2–6 cm sediment depth and has been reported to have high reworking rates [91]. While we observed the typical head-down deposit feeding of tubificids in our experiments, we also observed the formation of galleries of worm burrows to >20 cm sediment depth. Worms were frequently seen moving through these established burrow networks. Previous research has shown that these burrow networks are widespread in eutrophic lakes of central Switzerland and that sediment reworking by this mode of living is minimal [37]. Thus, the worm specimens that were used in the experiments could have engaged in unknown, deep feeding activities that only involved minimal sediment mixing (also see [37]). An additional factor may have been worm size. Previously it was shown that reworking activity and mixing depths are dependent on worm body length, and increase by an order of magnitude in worms with body lengths around 10 cm compared to worms with lengths around 5 cm [91,92]. Oligochaetes used here were typically ≤3 cm long, consistent with the observed low reworking rates. Furthermore, several, though not all, previous studies that have reported strong reworking by tubificids applied worm densities that were higher (e.g., 10,000 individuals m⁻² [93]) than in our experiments (~8500 individuals m⁻²).

Finally, the impact of macrofaunal bioturbation on sediment geochemical conditions (especially solid phases) and microbial community structure might take longer than our study period of 3 months. Indeed, a recent study found long-term effects of chironomid bioturbation on P immobilization in sediment in a 3.5-yr experiment [18]. In our case, changes in bioavailable Fe pools were evident in W treatments during the second half of the experiments, indicating that mineral transformations due to bioturbation were not completed at the end of our experiments. Microbial generation times also need to be considered in the context of experimental duration, since these can vary greatly from hours and days [94] to weeks [95] to months or even longer in Archaea [96]. Slow population response times due to low growth rates could explain the rather slow change in microbial community structure, especially in archaeal communities. However, we did observe significant changes in microbial communities due to oxygenation and in macrofaunal burrows. Thus, slow microbial growth rates are unlikely to be the only reason for the overall minor impact of bioturbation on microbial communities in the sediments studied.

5. Conclusions

Our study provides novel insights into the impacts of reoxygenation and macrofaunal recolonization on sediment biogeochemistry and microbial community composition. Based on mesocosm experiments involving intact sediment cores from a previously hypoxic and macrofauna-free lake basin, we show that oxygenation of bottom water alone has a stronger impact on sediment biogeochemistry and microbial community structure than the additional presence of bioturbating macrofauna. The comparatively smaller impact of bioturbation in the sediments studied than observed in numerous past studies could be in part due to the use of natural chironomid larval and tubificid worm communities at close-to-natural densities and size distributions in our work. By contrast, most previous studies on this topic have involved sediment amendments with mono-species assemblages of larvae or worms, at population densities and size distributions that are considerably higher than those in natural sediments of the lakes studied. Though more research, e.g., involving longer time scales, more locations, and additional experimental setups is necessary, our study demonstrates the feasibility and importance of using natural communities and population densities to study their impact on elemental cycles and their microbial agents in lake sediments.