Marine sediment cores were constructed and then incubated for 7 d. The total oxygen uptake (TOU) was measured as the indicator of total community respiration, which primarily represents mineralisation of refractory organic C. In parallel, we quantified fresh organic matter respiration (FOMR) in the same cores by addition of ¹³C labelled substrates and determining the subsequent release of ¹³C labelled dissolved inorganic C (DIC).

To obtain insights into the TOU and FOMR of different components of the benthic community, as well as to assess the interaction between microorganisms and macrofauna, we applied the following four treatments when constructing experimental cores: (1) Control: natural, intact sediment cores. Respiration is due to prokaryotes and macrofauna, and their interaction. (2) Defaunated: sediment cores that were defaunated by inducing anoxia, and exposed again to overlying oxygenated water. Respiration is dominated by prokaryotes (with some meiofauna present), but macrofauna are excluded. (3) Restocked: sediment cores were first de-faunated (by inducing anoxia), and then exposed again to overlying oxygenated water and re-stocked with a controlled macrofaunal community. Respiration is due to prokaryotes and a controlled biomass of macrofauna, and their interaction. (4) Fauna: sediment cores were constructed that contain only clean construction sand, to which macrofauna were introduced. Respiration is due to macrofauna. A control with only clean construction sand was run, but TOU data was not acquired due to instrument problems. However, we expect microbial respiration to be small in these construction sand cores compared to that of the macrofauna added.

The experiment with the four treatments was conducted twice using different ¹³C labelled substrates. In a first experiment, ¹³C labelled algal detritus from an axenic culture (¹³C-AA) was added, which allowed tracing of C into the microbial biomass. In the second experiment, we added natural microphytobenthos cultured in the presence of ¹³C labelled bicarbonate(¹³C-MPB), thus providing a fully natural fresh C source.

If there were no interactions between components of the benthic community, respiration rates measured in the “fauna” treatment can simply be added to those from the “defaunated” treatment and would equal the rates measured in the re-stocked treatment (macrofauna and microbes + meiofauna together). Deviations from this expectation are indicative of positive (i.e. inverted microbial loop) or negative (i.e. competitive) interactions between components of the benthic community.

Experiments were conducted in June 2010 and June 2011 at the Netherlands Institute for Sea Research (Yerseke, the Netherlands). Sediment cores and filtered seawater were collected from nearby intertidal sites in the Oosterschelde estuary. Key experimental details are listed in Table 1. Surface sediment cores (19.4 and 14.3 cm inner diameter for ¹³C-AA and ¹³C-MPB respectively) were collected in acrylic tubes, and after a short transit to the laboratory (< 2 h), they were kept in darkness in a climate-controlled room at ambient temperature with overlying filtered seawater (0.2 µm pore size) at in-situ salinity (Table 1). Overlying water was oxygenated using air stones, except for when periods oxygen consumption rates were determined (see description below).

Three replicate cores were subjected to each of the 4 treatments in the two experiments. De-faunation of sediment cores for the defaunated and restocked treatments was conducted by asphyxiation as in described Rao et al. (2014), which leaves the sediment stratification intact (as opposed to defaunation by sieving). To this end, anoxic conditions were induced by purging the overlying seawater in the core with N₂ gas for several hours and then sealing the cores with gas-tight lids for 4–6 d. After this anoxic period, the cores were opened and the overlying water was exchanged and re-aerated with air stones. Dead organisms that had migrated to the sediment surface were first removed with tweezers. The cores were subsequently left undisturbed for one day to allow the re-oxidation of reduced compounds that had accumulated in the surface layer of sediment. After one day of reaeration, a mix of fauna (Table 2 and further below) was added at the surface of restocked treatment cores, and were allowed migrate into the sediment. Cores were then acclimated again for 1–2 d before being amended with ¹³C labelled organic detritus. After that the cores were incubated for 7 d.

De-faunation by inducing anoxia was selected in preference to de-faunation by sieving. It was felt that sieving would cause extensive changes to sediment structure and composition which would have more potential to introduce artefacts than the possibility of live fauna remaining in the sediment following induction of anoxia. Presence of live fauna in the de-faunated treatment was minimal, and is reported below.

Based on background knowledge about the sampling site (Daggers et al., 2020), we knew a priori that the fauna at the sampling location predominantly consists of the polychaetes Hediste diversicolor, Arenicola marina and Heteromastus filiformis, the gastropod Hydrobia ulvae and the bivalve Cerastoderma edule. These species were therefore selected for the re-stocked and fauna treatment and introduced into cores at densities that simulated the natural faunal community (Table 2 and results).

We acknowledge that greater replication will always strengthen an experiment, but benefits have to be balanced against practical constraints (space, volume of sediment, operator time). Although some experiments use 4 or 5 replicates per treatment, the triplicate replication used here is in line with similar experiments in the literature (e.g. Moodley et al., 2000; Sweetman and Witte, 2008; van Nugteren et al., 2009a; Rossi et al., 2009), and is sufficient for showing difference between treatments despite natural variability (see results).

For the “axenic algae” (¹³C-AA) experiment, the marine diatom Skeletonema costatum was axenically cultivated in ¹³C-labelled medium. The resulting algal cells were 28.25 atom % and 14.49 atom % ¹³C for two separate batches. A slurry of freeze-dried, ¹³C-labelled biomass (395 ± 11 mg C m⁻²) was carefully mixed into to the water column and allowed to settle onto the sediment-water interface (so that the whole surface area was more or less homogeneously covered with labelled substrate).

For the “microphytobenthos” (¹³C-MPB) experiment, microphytobenthos was collected at the study site at the same time as the sediment cores. The top millimetres of sediment were scraped off at locations where distinctly brown patches (indicative of high MPB biomass) were present. This sediment/microphytobenthos mixture was enriched with ¹³C through incubation in a white plastic culture box (0.6 m × 0.4 m) that was placed outside (ambient temperature) and covered with a transparent lid (natural light). The thin layer of sediment in the culture box was topped with a thin layer of ambient seawater (∼ 5 mm) to prevent dehydration. The next day, 0.136 g of ¹³C-labelled sodium bicarbonate (NaH¹³CO₃, 99 %; Cambridge Isotope Laboratories) was dissolved in 50 mL of filtered seawater and introduced into the culture. This label addition was repeated daily for 7 d, after which the labelled microphytobenthos was harvested by scraping off the top several millimetres. This mixture was homogenised, frozen in liquid nitrogen (to kill the MPB cells and prevent respiration activity by MPB during the sediment core incubations) and stored until further usage at -18 °C. The chlorophyll-a concentration of this slurry was determined on 3 subsamples using standard fluorometry methods (Aminot and Rey, 2001). The resulting concentration (37 ± 5 µg g⁻¹) was converted to C using a conversion factor of 40 (Stephens et al., 1997) resulting in an estimated 1.5 ± 0.2 mg C g⁻¹. Cores were amended with 12.5 cm³ of slurry (density 2.0 g cm⁻³), which was added using a pipette, and allowed to settle onto the sediment surface over several hours. Each core hence received 3.08 mmol of C from MPB (corresponding to 2.30 g C m⁻²). The ¹³C labelling level of the MPB was unknown, but this does not prevent calculation of respiration rates from the measured ¹³C-DIC production.

In both experiments, cores were incubated for 7 d after addition of labelled algae, with repeated measurements of O₂ consumption and ¹³C-DIC release during this period (see below). At the termination of the experiment, sediment cores were sub-sampled using plastic syringes and samples were frozen at -18 °C. The remaining sediment in each core was sieved through a 1 mm mesh. Fauna retained on the mesh were picked, and their wet biomass was recorded, after which specimens were frozen for further analysis.

Benthic respiration was measured in all cores through total O₂ uptake (TOU, i.e. proxy for total community respiration, primarily of refractory organic C) and release of ¹³C-DIC (i.e. proxy for respiration of fresh, labile algae) at several time points: before and straight after addition of isotopically labelled algae, and every 1.5 d for 7 d thereafter.

At the beginning of each respiration measurement, the overlying water of each core was sampled for dissolved oxygen (DO), dissolved inorganic carbon (DIC), and ¹³C of DIC. Cores were then sealed with custom-built gas-tight lids, excluding all air bubbles, and incubated for 2–5 h until O₂ saturation in the overlying water had fallen to ∼ 70 %. During the closed incubation, core top water was stirred continuously. At the end of each respiration measurement, core top water was again sampled for the parameters listed above. After respiration measurements, the overlying water in each core was exchanged to avoid build-up of (toxic) metabolic products, and kept aerated by gentle bubbling with air.

Samples for O₂ analysis were collected in glass Winkler bottles with ground glass stoppers and known volumes. Bottles were allowed to overflow copiously before MnSO₄ and KI in KOH solutions were added and stoppers inserted. Samples were shaken for 30 s, and stored in at 4 °C before analysis within 2 d. Samples were titrated against standardised thiosulphate solution using a micro-titration set-up.

Dissolved inorganic carbon (DIC) samples (20 mL) were stored in crimp-cap vials, and preserved with HgCl₂ (20 µL of saturated solution). Vials were stored at 4 °C, inverted with the caps standing in water to prevent the exchange of CO₂ with the atmosphere. Samples were analysed for DIC concentration and δ13C as detailed in Moodley et al. (2000), using a Carlo Erba MEGA 540 gas chromatograph coupled to a Finnigan Delta S isotope ratio mass spectrometer, following creation of a He headspace in each sample vial. Standards used were acetanilide, and the IAEA standard CH-6. Repeat analysis of standard materials yielded precision of ± 4.4 % for DIC concentrations, and ± 0.09 ‰ for δ13C.

Sediment samples from the axenic algae experiment were analysed for ¹³C incorporation into bacterial phospholipid fatty acids (PLFAs) using a modified Bligh-Dyer extraction after Middelburg et al. (2000). Lipids were extracted at room temperature in a mixture of chloroform, methanol and water, before being loaded onto silicic acid columns. Phospholipid fatty acids were eluted in methanol and derivatised to fatty acid methyl esters (FAMEs) using methanolic NaOH. The C12:0 and C19:0 FAMEs were used as internal standards. Samples were separated by gas chromatography using a BPX70 column, combusted in a Thermo GC combustion II interface, and isotopic ratios were measured using a Thermo Delta + isotope ratio mass spectrometer.

The Total Oxygen Uptake (TOU) of the sediment was calculated from the difference in the total amount of dissolved O₂ present (i.e. O₂ concentration × chamber volume) between the start and end of each closed incubation, divided by the time elapsed in each measurement (Δt), and normalised to the surface area of the cores (SA), i.e., TOU = (O₂ end - O₂ start) /Δt/ SA.

Release of ¹³C-DIC was determined from the difference in total amount of ¹³C in each chamber (i.e. DIC concentration × chamber volume × At % ¹³C DIC) between the start and end of the incubation, divided by the duration of the incubation (Δt), and normalised to the surface area (SA) of the cores, i.e. ¹³C-DIC Release = (¹³C end - ¹³C start) /Δt/ SA.

Cumulative TOU and ¹³C-DIC release were calculated by multiplying each of the measured rates described above by the time periods between closed TOU / ¹³C incubations. These were then summed to produce estimates of cumulative TOU and ¹³C-DIC release over the whole experiment for each treatment.

Uptake of ¹³C into bacterial biomass in the ¹³C-AA experiment was calculated by first subtracting naturally present ¹³C based on analysis of unlabelled sediment. Presence of ¹³C in the bacterial indicators i-C14:0, i-C15:0, ai-C15:0 and i-C16:0 was then summed, and scaled up based on these compounds representing 14 % of total bacterial PLFAs, and PLFAs representing 5.6 % of total bacterial biomass (Boschker and Middelburg, 2002).

Statistical analysis of data was performed using Minitab 18. Differences between treatments were investigated using either one-way ANOVA or KruskalWallis, depending on whether data were normally distributed, determined using the Anderson-Darling normality test. For tests of respiration rate data values of n ranged from 10 to 24, and for faunal ¹³C labelling from 5 to 17. In some cases tests for difference were conducted between treatments for data from each day separately, in which case n=3. We recognise that checking the distribution of such a small group of data is not necessarily possible, and also that use of non-parametric tests in this situation does carry a risk of not identifying patterns which are in fact present. We note that the questions of which statistical tests are most appropriate, and whether statistical testing should be included at all, are ones on which different statisticians and readers are unlikely to agree. We feel that the approach we have taken is justifiable, but acknowledge that any approach to statistical testing which we could take would be open to differences of opinion.

The living macrofaunal biomass recovered from the control treatment at the end of the ¹³C-AA experiment (4.8 ± 3.2 g wet weight per core) was far greater than that recovered from the defaunated treatment (0.9 ± 1.0 g wet weight per core). This illustrates that asphyxiation removed > 80 % of the fauna, but still a restricted anoxia-tolerant community (Hediste, Arenicola) survived. The natural biomass present (in the control treatment) was lower than anticipated, and so biomass added to the restocked and fauna treatments (18–21 g wet weight per core, Table 3) was four times higher than the control treatment. Very few dead organisms were seen in treatments where fauna were added, with the majority recovered alive at the end of the experiment. Note that macrofauna biomass data were not recorded for the ¹³C-MPB experiment.

Total Oxygen Uptake rates showed substantial variation and ranged from 9–91 mmol O₂ m⁻² d⁻¹ in the ¹³C-AA experiment, and 7–241 mmol O₂ m⁻² d⁻¹ in the ¹³C-MPB experiment (Fig. 1). TOU values were generally higher in the ¹³C-MPB experiment compared to the ¹³C-AA experiment. Due to problems with the oxygen measurement technique, data is lacking for the control and fauna treatments in the ¹³C-AA experiment during the first 2 d after feeding.

In the ¹³C-AA experiment, TOU showed a slight decrease over time in the restocked treatment, but no clear temporal pattern in the other treatments (Fig. 1a). In the ¹³C-MPB experiment all treatments displayed a similar temporal pattern, with maximal TOU values immediately after algal addition, and TOU values returning to pre-feeding levels after ∼ 6 d (Fig. 1b). Differences in TOU between treatments were apparent in both experiments (Kruskal-Wallis p < 0.001 for both experiments). In the ¹³C-AA experiment, TOU values were always higher in the re-stocked cores compared to other treatments (Mann-Whitney pairwise comparisons p < 0.001). There was also a significant difference between the control and defaunated treatments, while other pairs of treatments were not significantly different (Mann-Whitney pairwise comparisons p = 0.004, 0.62, and 0.012 for control vs. defaunated, control vs. fauna, and defaunated vs. fauna, respectively). In the ¹³C-MPB experiment rates were higher in the control and re-stocked treatments than in the defaunated and fauna only treatments (Kruskal-Wallis, p≤0.001). TOU values in the control and restocked treatments (Mann-Whitney, p=0.130) and in the defaunated and fauna only treatments (Mann-Whitney, p=0.516) were not significantly different from each other.

The cumulative TOU (i.e. the total O₂ consumed during each 7 d experiment) was higher in the ¹³C-MPB experiment compared to the ¹³C-AA experiment. Cumulative TOU showed a similar pattern between treatments in both experiments (Fig. 2a) and was maximal in the restocked treatment, then followed by the control, and finally the defaunated and fauna only treatments (Fig. 2a). Due to the high variability, significant differences between treatments could be identified for the ¹³C-AA experiment (ANOVA, p < 0.001, groupings shown in Fig. 2a), but not for the ¹³C-MPB experiment (ANOVA, p=0.052).

Fresh organic matter respiration (FOMR) rates were measured as the release of ¹³C -DIC and ranged between 0.04–1.85 mmol C m⁻² d⁻¹ in the ¹³C-AA experiment, and 0.01–4.38 mmol C m⁻² d⁻¹ in the ¹³C-MPB experiment (Fig. 3). Fresh organic matter respiration rates were substantially higher in the ¹³C-MPB experiment, but generally showed a similar time evolution in both experiments. Rates were always highest immediately after feeding, and declined rapidly thereafter, reaching constant levels after ∼ 5 d (Fig. 3). Differences between treatments were most apparent during the first 2 d after feeding. For the ¹³C-AA experiment, the re-stocked and fauna treatments showed slightly higher initial FOMR rates (Fig. 3a). For the ¹³C-MPB experiment, the defaunated treatment showed higher initial rates (Fig. 3b). Due to the marked change in rates over time, significant differences in rates between treatments were only apparent on individual days. Significant differences between treatments were present 5 d after feeding in the ¹³C-AA experiment (Kruskal-Wallis p=0.04), and 1, 2 and 7 d after feeding in the ¹³C-MPB experiment (Kruskall-Wallis, p=0.029, 0.038 and 0.034, respectively). However, pairwise Mann-Whitney U tests were not sufficiently powerful to show which pairs of treatments were significantly different on those days.

The cumulative FOMR was higher in the ¹³C-MPB experiment by a factor ∼ 2–8 compared to the ¹³C-AA experiment for different treatments, and showed different patterns in the two experiments (Fig. 2b). In the ¹³C-MPB experiment cumulative FOMR was maximal in the defaunated and fauna only treatments (ANOVA, p=0.011, groupings shown in Fig. 2b). In the ¹³C-AA experiment there was no significant difference in cumulative FOMR between treatments (ANOVA, p=0.061).

For each time point the TOU/FOMR ratio was calculated. The ratio ranged from ∼ 50–1100 for the ¹³C-AA experiment and from ∼ 19–958 for the ¹³C-MPB experiment (Fig. 4). There was a significant difference in TOU : FOMR ratios between the treatments in the two experiments (¹³C-AA experiment ANOVA, p=0.014; ¹³C-MPB experiment Kruskall-Wallis, p < 0.001). Post-hoc testing showed that for the ¹³C-AA experiment the restocked treatment had significantly higher ratios than the defaunated treatment, and that the other two treatments were not significantly different from any other. Further, the fauna treatment, although not being statistically significantly different, appeared most similar to the defaunated treatment (Fig. 4). Similarly, in the ¹³C-MPB experiment, all treatments were significantly different from each other (Mann-Whitney, p < 0.001–0.002), except for the defaunated and fauna treatments, which were not significantly different (Mann-Whitney, p=0.948).

Uptake of ¹³C into bacterial biomass was quantified by PLFA analysis in the ¹³C-AA experiment (Fig. 5) and predominantly occurred in the surface sediment (0–1 cm), with uptake values 10-fold higher than the subsurface sediment (9–10 cm). Differences were notable between treatments: ¹³C uptake into bacterial biomass was not detectable in the fauna-only treatment, and ranged up to a maximum of 0.052 mg C g⁻¹ of wet sediment in the defaunated treatment (Fig. 5). Bacterial ¹³C uptake appeared to be maximal in the defaunated treatment (Fig. 5), but due to high variability in the control treatment, the observed differences between the control, defaunated and restocked treatments were not statistically significant.

Uptake of ¹³C into macrofaunal biomass was quantified in both the ¹³C-AA and ¹³C-MPB experiments and varied between the two experiments (Fig. 6). All taxa showed uptake of labelled fresh organic matter in both experiments, providing δ13C values up to 786 ‰ in the ¹³C-AA experiment and up to 286 ‰ in the ¹³C-MPB experiment (Fig. 6).

It should be noted that the C dose used in the two experiments varied (395 ± 11 and 2300 mg C m⁻² for the ¹³C-AA and ¹³C-MPB experiments, respectively), and therefore direct comparison of faunal C uptake or labelling intensity between experiments is not possible. However, comparisons can be made regarding relative labelling levels of different taxa within each experiment, and these showed significant differences in labelling between taxa (Kruskal-Wallis, p< 0.001 for the ¹³C-AA experiment and ANOVA, p< 0.001 for the ¹³C-MPB experiment). Hydrobia ulvae and Hediste diversicolor showed the highest labelling in both experiments, consistent with their high motility and surface deposit feeding habits. In contrast, the sessile and deep-living taxa Arenicola marina, Cerastoderma edule and Macoma balthica showed a lower labelling intensity (Fig. 6). Data for Heteromastus filiformis illustrated how variable the feeding can be within a single macrofaunal taxon, with low labelling in the ¹³C-AA experiment, and high labelling in the ¹³C-MPB experiment. This may be due to a feeding preference by Heteromastus filiformis, or could be a result of differences between the experiments in terms of C dose or other site-specific factors.

For the ¹³C-AA experiment, the wet weight of the macrofauna were measured, allowing quantification of total added C uptake by the macrofauna. Hydrobia ulvae was excluded from this calculation due to uncertainties in wet weight data. Macrofaunal C uptake ranged from 15.9 mg C m⁻² in the defaunated treatment up to 61.5 mg C m⁻² in the restocked treatment (Fig. 6c). Macrofaunal uptake was generally higher in the restocked treatment than in other treatments, however variability in faunal biomass meant the differences were not statistically significant (Kruskal-Wallis, p = 0.192). Further, when total C uptake data from the ¹³C-AA experiment were pooled by taxon, there was a significant difference in uptake accounted for by different taxa (Kruskal-Wallis, p=0.044), with Arenicola marina and Hediste diversicolor each showing significantly more C uptake than Heteromastus filiformis, and Macoma balthica (Mann-Whitney, p = 0.027–0.030, Fig. 6d).

In the ¹³C-AA experiment sufficient ¹³C pools were quantified to allow a carbon budget to be calculated (with Hydrobia excluded from macrofaual uptake, as mentioned above). In the control and defaunated treatments a mean of 17.5 ± 5.5 % of the added ¹³C was recovered from biologically processed pools (fauna, bacterial biomass and respiration). The restocked treatment showed the highest percentage of biologically processed ¹³C (24.2 %), with particularly high uptake into macrofauna (Fig. 8). We presume that the ¹³C that could not be accounted for remained predominantly in the sediment, although a portion will have been converted to DOC. Data are not available to confirm this.