Contrasting E ﬀ ects of Bioturbation Studied in Intact and Reconstructed Estuarine Sediments

: Macrofauna can produce contrasting biogeochemical e ﬀ ects in intact and reconstructed sediments. We measured benthic ﬂuxes of oxygen, inorganic carbon, and nitrogen and denitriﬁcation rates in intact sediments dominated by a ﬁlter and a deposit feeder and in reconstructed sediments added with increasing densities of the same organisms. Measurements in reconstructed sediments were carried out 5 days after macrofauna addition. The degree of stimulation of the measured ﬂuxes in the intact and reconstructed sediments was then compared. Results conﬁrmed that high densities of bioturbating macrofauna produce profound e ﬀ ects on sediment biogeochemistry, enhancing benthic respiration and ammonium recycling by up to a factor of ~3 and ~9, respectively, as compared to control sediments. The deposit feeder also increased total denitriﬁcation by a factor of ~2, whereas the ﬁlter feeder activity did not stimulate nitrogen removal. Moreover, the e ﬀ ects of deposit feeders on benthic ﬂuxes were signiﬁcantly higher (e.g., on respiration and ammonium recycling) or di ﬀ erent (e.g., on denitriﬁcation) when measured in intact and reconstructed sediments. In intact sediments, deposit feeders enhanced the denitriﬁcation coupled to nitriﬁcation and had no e ﬀ ects on the denitriﬁcation of water column nitrate, whereas in reconstructed sediments, the opposite was true. This may reﬂect active burrowing in reconstructed sediments and the long time needed for slow growing nitriﬁers to develop within burrows. Results suggest that, in bioturbation studies, oversimpliﬁed experimental approaches and insu ﬃ cient preincubation time might lead to wrong interpretation of the role of macrofauna in sediment biogeochemistry, far from that occurring in nature.


Introduction
Bioturbation by macrofauna is a fascinating topic that has fueled a diversified body of scientific research [1][2][3][4][5][6]. Macrofauna couple the pelagic and benthic environments in various ways. For instance, feces and pseudofeces deposition by filter feeders, containing viable and active phytoplankton, displaces pelagic primary production at the sediment level [7,8]. Filter feeders activity, such as that of clams, may process the entire water column of shallow water bodies many times per day, assuring water transparency, increased light penetration, and benthic primary production [9][10][11]. Filter feeders also excrete large amounts of the mineralized phytoplankton biomass back into the water column [12][13][14]. Biodeposits can increase the organic content of sediments and stimulate both aerobic and anaerobic Water 2020, 12 metabolic pathways [15,16]. Filtration and excretion thus have the potential of maintaining pelagic and benthic primary production elevated. Deposit feeders, such as polychaete worms, explore surficial or deep sediment horizons to extract food and avoid predation. As oxygen (O 2 ) penetration in sediments is extremely limited, burrowing macrofauna have evolved different ways to cope with anoxia and potentially toxic concentrations of compounds such as ammonia or sulfide. Burrow architectures may vary among species, but most deposit feeders ventilate their burrows by injecting and circulating oxic bottom water or by extracting pore water from sediments [17][18][19]. Such a need to get rid of metabolic end-products and to import O 2 for their own respiration produces different biogeochemical side-effects on sediments [20]. For example, the volume of oxidized sediments in bioturbated areas is much higher than that in their not bioturbated counterparts, which means that the oxidation status of different elements is different depending on the presence of macrofauna. Ecosystem-level implications are multiple as oxidized sediments may retain elements avoiding their release to the water column or favor their permanent loss via coupled oxic-anoxic processes [21,22]. Macrofauna, through bioturbation but also feeding strategy and ecological interactions, affect both the physical and biological environments in sediments and in the water column. During their life span, macrofauna contribute to the creation and become part of a complex network of micro-and macro-organisms and of chemical and biological processes [23][24][25]. With this respect, little is known about the complex and intimate mutual interactions within holobionts, which are associations between microbes and macrofauna, and that affect benthic functioning [6,26].
Research activities in the field of bioturbation, with the aim of understanding density-dependent effects of macrofauna on selected processes, have largely adopted experimental approaches based on the addition of a single species to reconstructed sediments (e.g., sieved or freeze-thaw, in order to remove other macrofauna species, then homogenized and packed back into microcosms) [27][28][29][30]. Results from this kind of studies have allowed us to understand how important macrofauna is for a number of biogeochemical processes, including aerobic and anaerobic microbial respiration rates, nutrient recycling, reworking, and mixing of new and old organic matter and oxidation of chemically reduced metal pools [31][32][33]. The level of enhancement or inhibition of specific processes operated by macrofauna and measured with such approaches has been likely used to calibrate models, to upscale processes under varying macrofauna density or to propagate the effects to the water column [34]. However, such an approach may lead to a partial understanding of the true effect of macrofauna due to multiple factors. One of them is the absence of other macrofauna groups that may compete with, facilitate, or inhibit the targeted species. Another one is the time taken for an organism living in sediment, which is necessary to engineer the environment and, therefore, to change the chemical conditions and the biological communities (e.g., meiofauna and microbes). Due to these factors, we hypothesized that processes measured under in situ conditions may deeply differ from those measured shortly after sediment manipulation and macrofauna addition. We also hypothesized that such differences may vary along different macrofauna functional groups.
To test these hypotheses, we measured aerobic, anaerobic (denitrification), total respiration, and nutrient fluxes in intact and reconstructed sediments collected from two estuarine areas dominated by a filter feeder and a deposit feeder. Specific objectives were i) to analyze the effects of bioturbation on different benthic processes in intact and reconstructed sediments and ii) to evaluate whether differences between the two experimental conditions (i.e., intact versus reconstructed) depend upon macrofauna functional groups.

Study Sites
Samples were collected from the Sacca di Goro Lagoon (Figure 1), a shallow microtidal lagoon within the Po River Delta, Northern Italy (44.78-44.83 • N, 2.25-12.33 • E) [35]. The lagoon is approximately triangular in shape, has a surface area of 27 km 2 , an average depth of 1.5 m, and it Water 2020, 12, 3125 3 of 14 is connected to the sea by a large mouth. The lagoon presents a western portion influenced by the nutrient-rich freshwater inlet of the Po di Volano, a central portion with marine influence and an eastern, confined area cultivated with clams (Ruditapes philippinarum). The average annual temperature and precipitation in this area are 17 ± 1 • C and 631 ± 117 mm, respectively. For the aims of this study, two stations were selected as representative of a pristine and of a heavily impacted lagoon area and located in the western and in the eastern portion of the lagoon, respectively. Station 1 has turbid water due to the Po di Volano freshwater inputs and has muddy, highly bioturbated sediments. During summer, the macrofauna community is dominated by the polychaete Alitta succinea. Station 2 has transparent water and sandy sediments and is located within a licensed area for clam farming; the dominant macrofauna is represented by the clam R. philippinarum.
Water 2020, 12, x FOR PEER REVIEW  3 of 15   nutrient-rich freshwater inlet of the Po di Volano, a central portion with marine influence and an  eastern, confined area cultivated with clams (Ruditapes philippinarum). The average annual temperature and precipitation in this area are 17 ± 1 °C and 631 ± 117 mm, respectively. For the aims of this study, two stations were selected as representative of a pristine and of a heavily impacted lagoon area and located in the western and in the eastern portion of the lagoon, respectively. Station 1 has turbid water due to the Po di Volano freshwater inputs and has muddy, highly bioturbated sediments. During summer, the macrofauna community is dominated by the polychaete Alitta succinea. Station 2 has transparent water and sandy sediments and is located within a licensed area for clam farming; the dominant macrofauna is represented by the clam R. philippinarum. Figure 1. Panel A shows the location of the two sampling stations in Sacca di Goro Lagoon (Po River Delta, Northern Italy). Panel B shows reconstructed sediments from the two stations, added with no (C, control) or increasing densities of Alitta succinea (L, low, 3 ind core −1 , and H, high, 6 ind core −1 ) or Ruditapes philippinarum (L, low, 2 ind core −1 , and H, high, 4 ind core −1 ). Panel C shows randomly collected intact sediments from the two stations, with natural macrofauna community, dominated by the species indicated. Panel D shows a reconstructed sediment core with burrowing individuals of Ruditapes philippinarum.

Reconstructed Sediment Cores
Sediments and water were collected in July 2016. On 2 July 2016, nearly 50 L of sediment and 100 L of water were collected from the two stations. As both stations were shallow (<1 m depth), the water was collected by hand in 25 L tanks, whereas the sediment was collected with a shovel and transferred into 20 L buckets. Water and sediments were brought to the laboratories at Ferrara University within 2 h. Here, sediments from the two stations were (i) sieved (0.5 mm) in two tanks containing in situ water to remove large macrofauna, (ii) homogenized, and (iii) left to settle for 3 days. The tanks with sieved sediments and in situ water were maintained in the dark at in situ temperature, and the overlying water was kept oxic by air bubbling. Individuals of A. succinea and of R. philippinarum retrieved during the sieving were maintained alive in separate small aquaria with a few cm of sediments and 10 cm of well aerated water phase. On 5 July 2016, 9 cores (inner diameter = 8 cm, height = 30 cm) were subsampled from each tank. Water and sediment phases were subsampled by pushing vertically transparent Plexiglas cores in the tanks. Once retrieved, all cores were leveled in order to have a sediment height of 10 cm and a water column of 17 cm. Cores with muddy sediments from Station 1 were treated as follows: 3 cores were used as control, 3 cores were (B) shows reconstructed sediments from the two stations, added with no (C, control) or increasing densities of Alitta succinea (L, low, 3 ind core −1 , and H, high, 6 ind core −1 ) or Ruditapes philippinarum (L, low, 2 ind core −1 , and H, high, 4 ind core −1 ). (C) shows randomly collected intact sediments from the two stations, with natural macrofauna community, dominated by the species indicated. (D) shows a reconstructed sediment core with burrowing individuals of Ruditapes philippinarum.

Reconstructed Sediment Cores
Sediments and water were collected in July 2016. On 2 July 2016, nearly 50 L of sediment and 100 L of water were collected from the two stations. As both stations were shallow (<1 m depth), the water was collected by hand in 25 L tanks, whereas the sediment was collected with a shovel and transferred into 20 L buckets. Water and sediments were brought to the laboratories at Ferrara University within 2 h. Here, sediments from the two stations were (i) sieved (0.5 mm) in two tanks containing in situ water to remove large macrofauna, (ii) homogenized, and (iii) left to settle for 3 days. The tanks with sieved sediments and in situ water were maintained in the dark at in situ temperature, and the overlying water was kept oxic by air bubbling. Individuals of A. succinea and of R. philippinarum retrieved during the sieving were maintained alive in separate small aquaria with a few cm of sediments and 10 cm of well aerated water phase. On 5 July 2016, 9 cores (inner diameter = 8 cm, height = 30 cm) were subsampled from each tank. Water and sediment phases were subsampled by pushing vertically transparent Plexiglas cores in the tanks. Once retrieved, all cores were leveled in order to have a sediment height of 10 cm and a water column of 17 cm. Cores with muddy sediments from Station 1 were treated as follows: 3 cores were used as control, 3 cores were added with 3 individuals of A. succinea, and 3 cores were added with 6 individuals of A. succinea, to reproduce densities (600 and 1200 ind m −2 , respectively) comparable to those measured in situ. In a few minutes, all added polychaetes burrowed into the sediments. Cores with sandy sediments from Station 2 were treated as follows: 3 cores were used as control, 3 cores were added with 2 clams (R. philippinarum), and 3 cores were added with 4 clams, to simulate densities at farmed areas (400 and 800 ind m −2 , respectively). Within 30 min from clams addition all organisms burrowed within the sand, and only siphons were visible. All 9 + 9 cores were provided with stirring units (a rotating magnet suspended 5 cm above the sediment-water interface driven by an external motor at 40 rpm) and submersed in separate incubation tanks containing aerated and well mixed water from the two stations inside a temperature-controlled room maintained in the dark at the in situ temperature (25 • C).

Intact Sediments
On 9 July 2016, a second sediment and water sampling was carried out during which at each station 6 Plexiglas cores (inner diameter = 8 cm, height = 30 cm) were randomly collected by hand for flux and denitrification measurements and 5 Plexiglas cores (inner diameter = 4 cm, height = 20 cm) were collected for sediment characterization (for the methods see [14,36,37]). Cores with disturbed vertical profiles or with visually broken clam shells along walls were immediately discarded. Besides intact sediments, nearly 100 L of water from each site was collected for preincubation and incubation periods. Water column temperature, pH, and salinity were measured at the two sites by means of a multiple probe (mod 556, YSI Instrument, Yellow Spring, OH, USA). All the material was brought to the laboratories at Ferrara University within 2 h. Here, intact sediment cores from the two sites were submersed into two 50 L tanks containing in situ aerated and well mixed water inside a temperature-controlled room maintained in the dark at the in situ temperature. Intact cores for flux measurements were provided with a rotating magnet, suspended 5 cm above the sediment-water interface driven by an external motor at 40 rpm.

Dark Incubations for Flux and Denitrification Measurements
On 10 July 2016, 5 days after macrofauna addition in reconstructed cores and after an overnight preincubation of the intact sediment cores, two sequential dark incubations were performed for the measurements of net solute fluxes across the sediment-water interface and of denitrification rates. During the first incubation, measurements of benthic fluxes of dissolved O 2 , dissolved inorganic carbon (DIC), molecular nitrogen (N 2 ), and dissolved inorganic nitrogen (DIN = NH 4 were performed with a start-end approach [38]. The incubation lasted between 2 and 3 h in order to keep O 2 drops from the initial values between 25 and 40%. Incubation time was set according to regular check (0.5 h) of dissolved O 2 in the water phase via a microelectrode (OX-50, Unisense A/S, Aarhus, Denmark) through the cores lid. At time, zero water samples (n = 4) were collected from each incubation tank. At time, final water samples were collected from each core water phase through a valve in the cores lid. Incubation started when gas-tight lids were positioned on the top of each core. Water samples were collected via 60 mL glass syringes; 20 mL was disposed in 12 mL exetainers (Exetainer ® , Labco Limited, High Wycombe, UK) and added with Winkler reagents for O 2 measurements, 20 mL was disposed in 12 mL exetainers for DIC determination, which was measured with 0.1 N HCl titration with an automatic titration unit (ABU91, Radiometer, Copenhagen, Denmark) and a combined pH sensor [39]. An aliquot of 20 mL was filtered (Whatman GF/C, Cytiva, Marlborough, MA, USA) and transferred to scintillation vials for nutrient analysis. Dissolved inorganic nitrogen was determined spectrophotometrically. Ammonium (NH 4 + ) was determined using salicylate and hypochlorite in the presence of sodium nitroprussiate [40]. Nitrate (NO 3 − ) was determined after reduction to nitrite (NO 2 − ) in the presence of cadmium, and NO 2 − was determined using sulphanilamide and N-(1-naphthyl)ethylenediamine [41,42]. Fluxes were defined as the difference between final and initial concentrations of solutes in the water column multiplied by the volume of the core water phase and divided by the area of the sediment and the incubation time (as reported in [38]). The respiratory quotient RQ, which is the ratio of dissolved inorganic carbon to oxygen fluxes (absolute values), was calculated. After the incubation for flux measurements, lids were removed, the water in each tank was renewed, and the cores were submersed. Cores were maintained open and stirred in in situ oxygenated and well mixed water in the dark and at the in situ temperature. Approximately 5 h later, denitrification measurements started. The water level in each tank was lowered just below the cores opening, in order to separate the cores from the tank water. Labelled NO 3 − from a stock solution (20 mM, sodium nitrate-15 N, ≥98 atom%, Sigma-Aldrich, Merk KGaA, Darmstadt, Germany) was added to the water column of each core in order to have a final 15 NO 3 − concentration of 25 µM (Isotope Pairing Technique -IPT- [43]). Before and after 5 min from the addition of the labelled NO 3 − a subsample (5 mL) of the water phase was collected, filtered, and dispensed into 10 mL plastic vials for NO 3 − analysis. Labelled NO 3 − was allowed to diffuse into the anoxic horizon and then a gas-tight lid was positioned on the top of each core and the incubation started. The incubation length of the denitrification experiment overlapped that of the flux measurements. At the end of the incubation, the cores were opened, slurred, and samples were collected for labelled N 2 analyses via membrane inlet mass spectrometry (MIMS, Bay instrument, Easton, MD, USA) [44]. Water samples for labelled N 2 analyses were poisoned with ZnCl 2 to stop microbial activity. Rates of total denitrification, denitrification coupled to nitrification (Dn), and water column nitrate-driven denitrification (Dw) were calculated, as described in [43]. Denitrification efficiency (DE) was calculated as the ratio between denitrification and cumulative flux of N from sediment (N 2 + DIN) and expressed as a percentage [45]. The slurry in each core was then sieved (0.5 mm) in order to collect all macrofauna; macrofauna were recognized, and dry weight was quantified after drying the soft tissue at 70 • C to a constant weight.

Statistical Analysis
Differences among net solute fluxes or denitrification rates measured in the intact cores of the two stations and in the reconstructed cores of C, L, and H treatments were tested via one-way analysis of variance (ANOVA) followed by the post hoc Holm Sidak test. The dependency of solute fluxes or denitrification rates on macrofauna biomass were tested using linear regression analysis. Normality of residuals was confirmed numerically with a Shapiro-Wilks test.
The comparison of regressions between the biomass and measured process rates or calculated respiratory quotients and denitrification efficiency in intact and reconstructed cores were performed via analysis of covariance (ANCOVA). The level for statistical significance was set at 0.05. Statistical analyses were performed with R software v. 3.5.1 (The R Foundation for Statistical Computing, Vienna, Austria) [46].

General Features of the Study Sites
Station 1 had soft, organic sediments, with lower density and higher porosity, organic matter, C and N contents, and chlorophyll a as compared to Station 2 ( Table 1). The biomass of macrofauna was one order of magnitude higher at Station 2 mostly due to high size of the cultivated R. philippinarum (Table 1). At Station 1, the dominant organism retrieved from sediments was the deep burrower A. succinea, with a density of 870 ± 530 ind m −2 . Besides A. succinea, the macrofauna community at Station 1 included other small polychaetes and occasional individuals of Monocorophium insidiosum, Chironomus salinarius, Hydrobia spp., and Capitella capitata, whose pooled biomass was <10% of the total. At Station 2, R. philippinarum dominated the macrofauna community, with a density of 630 ± 490 ind m −2 . Besides clams, sediments also hosted a few individuals of M. insidiosum and Echinogammarus spp., whose pooled biomass was <1% of the total. Station 1 was influenced by the freshwater inputs of the Po di Volano and had lower salinity and higher concentrations of DIN as compared to Station 2, with NO 3 − as dominant chemical species (Table 1).

Benthic Respiration, DIN Fluxes, and Rates of Denitrification in Intact Sediment Cores
Respiration rates tended to be higher at Station 2, dominated by R. philippinarum, however differences were not significant ( Figure 2). Similar net O 2 uptake and DIC release at the two stations, despite large difference in organic matter content, were probably the result of much higher contribution of macrofauna metabolic activity in the sandy Station 2, sustained by clams. Both stations recycled large and not statistically different amounts of NH 4 + . At the muddy, organic-rich Station 1, such amounts are likely sustained by high microbial ammonification, whereas at Station 2, such amounts are likely excreted by clams [47]. Net inorganic carbon to NH 4 + efflux ratios averaged 6.9 ± 0.7 and 13.9 ± 4.7 at Station 1 and 2, respectively, suggesting in general the labile nature of the available organic matter. The low ratio measured at Station 1, very close to that of phytoplankton, may be due to much higher rates of NH 4 + oxidation within sediments via nitrification [22]. This is confirmed by higher net NO 2 − and NO 3 − fluxes, averaging 119.35 ± 5.35 and 122.27 ± 15.11 µmol N m −2 h −1 , respectively, and by denitrification rates of NO 3 − produced within sediments, all significantly higher at Station 1. Elevated nitrification rates at Station 1 are probably supported by the activity of burrowers, increasing O 2 availability within NH 4 + -rich pore waters [27,48,49].
Rates of total denitrification were higher at Station 1 due to higher denitrification coupled to nitrification and to higher denitrification of water column NO 3 − (Table 1). Denitrification efficiency was higher by a factor of 3 at Station 1 (15%) than at Station 2, due to similar NH 4 + regeneration but much higher denitrification. Despite such difference, at both sites the major dissolved inorganic N flux was towards the water column as recycled NH 4 + , whereas the fraction lost to the atmosphere was minor.

Benthic Respiration, DIN Fluxes and Rates of Denitrification in Reconstructed Cores
Increasing densities of A. succinea and R. philippinarum in reconstructed sediments significantly stimulated benthic respiration, including the consumption of O 2 and the production of DIC (Figure 3). The degree of stimulation of O 2 and DIC fluxes was different along with the density gradient, resulting in an increase in the respiratory quotient, approaching the unit in the high density treatments. Both macrofauna functional groups largely enhanced NH 4 + regeneration, whereas their net effects on NO 2 − and NO 3 − fluxes were less marked likely due to simultaneous stimulation of multiple microbial processes (e.g., nitrification and denitrification, Figure 3) [50]. The isotope pairing technique revealed a significant effect of A. succinea on the removal of water column NO 3 − via denitrification (Dw), whereas the same was not apparent with R. philippinarum [28]. Both macrofauna groups did not stimulate denitrification coupled to nitrification (Dn) (Figure 3). In reconstructed sediments, denitrification efficiency tended to decrease along with increasing macrofauna densities (p = 0.088 and p = 0.152 for A. succinea and R. phlippinarum, respectively) confirming a major effect of both deposit-feeding and filter-feeding macrofauna biomass on N recycling than on N losses [5]. Denitrification efficiencies overlapped values determined in intact sediments and averaged 15% (range 11-18%) and 5% (range 2-8%) in the A. succinea and R. philippinarum treatments, respectively. In the upper panel, the y-axes on the right reports the respiratory quotient (RQ), calculated as the ratio between DIC and O2 fluxes (absolute value). Where present, letters above bars indicate statistically significant differences. Averages ± standard errors (n = 6) are reported.

Benthic Respiration, DIN Fluxe,s and Rates of Denitrification in Reconstructed Cores
Increasing densities of A. succinea and R. philippinarum in reconstructed sediments significantly stimulated benthic respiration, including the consumption of O2 and the production of DIC ( Figure  3). The degree of stimulation of O2 and DIC fluxes was different along with the density gradient, resulting in an increase in the respiratory quotient, approaching the unit in the high density treatments. Both macrofauna functional groups largely enhanced NH4 + regeneration, whereas their In the upper panel, the y-axes on the right reports the respiratory quotient (RQ), calculated as the ratio between DIC and O 2 fluxes (absolute value). Where present, letters above bars indicate statistically significant differences. Averages ± standard errors (n = 6) are reported.
(p = 0.088 and p = 0.152 for A. succinea and R. phlippinarum, respectively) confirming a major effect of both deposit-feeding and filter-feeding macrofauna biomass on N recycling than on N losses [5]. Denitrification efficiencies overlapped values determined in intact sediments and averaged 15% (range 11-18%) and 5% (range 2-8%) in the A. succinea and R. philippinarum treatments, respectively. C indicates the control treatment, L the low density treatment (600 and 400 ind m −2 for polychaetes and clams, respectively), and H the high density treatment (1200 and 800 ind m −2 for polychaetes and clams, respectively). In the upper panels, the y-axes on the right reports the respiratory quotient (RQ), calculated as the ratio between DIC and O2 fluxes (absolute value). Where present, letters above bars indicate statistically significant differences. Averages ± standard errors (n = 3) are reported.
These results allow to perform simple linear regressions between measured process rates and macrofauna biomass to calculate the degree of enhancement of a certain process ( Table 2). The biomass of added A. succinea (10.3 ± 3.7 and 21.4 ± 4.8 gdw m −2 for L and H, respectively) was an order of magnitude lower than that of added R. philippinarum (222.2 ± 37.5 and 410.7 ± 68.8 gdw m −2 for L and H, respectively), however the effects produced on benthic biogeochemistry was comparable, except for denitrification rates. Burrowing A. succinea contributed less to respiration and excretion than R. philippinarum, but produced a much larger effect on microbial activity, compensating the different biomass. This interpretation is supported by the higher degree of stimulation of Dw by A. Figure 3. Dark fluxes of dissolved O 2 and DIC, DIN, and rates of denitrification measured in reconstructed sediments added with Alitta succinea (Station 1) and Ruditapes philippinarum (Station 2). C indicates the control treatment, L the low density treatment (600 and 400 ind m −2 for polychaetes and clams, respectively), and H the high density treatment (1200 and 800 ind m −2 for polychaetes and clams, respectively). In the upper panels, the y-axes on the right reports the respiratory quotient (RQ), calculated as the ratio between DIC and O 2 fluxes (absolute value). Where present, letters above bars indicate statistically significant differences. Averages ± standard errors (n = 3) are reported.
These results allow to perform simple linear regressions between measured process rates and macrofauna biomass to calculate the degree of enhancement of a certain process ( Table 2). The biomass of added A. succinea (10.3 ± 3.7 and 21.4 ± 4.8 g dw m −2 for L and H, respectively) was an order of magnitude lower than that of added R. philippinarum (222.2 ± 37.5 and 410.7 ± 68.8 g dw m −2 for L and H, respectively), however the effects produced on benthic biogeochemistry was comparable, except for denitrification rates. Burrowing A. succinea contributed less to respiration and excretion than R. philippinarum, but produced a much larger effect on microbial activity, compensating the different biomass. This interpretation is supported by the higher degree of stimulation of Dw by A. succinea. Denitrification in fact is not performed by deposit-feeding macrofauna but by the microbes growing along burrows that take advantage from the macrofauna-mediated bioirrigation, transporting NO 3 − -rich water within sediments [8,19,31]. Burrowing and ventilation activities moreover mobilize pore water solutes as NH 4 + , which is released in large amounts in A. succinea L and H treatments.
We speculate that fluxes measured in sediments with R. philippinarum are instead mostly sustained by clams excretion [51,52]. Table 2. Slope, intercept, level of significance, and r 2 values of linear regressions between solutes fluxes (µmol m −2 h −1 ), respiratory quotient, denitrification rates (µmol m −2 h −1 ), and efficiency and macrofauna biomass of Alitta succinea and Ruditapes philippinarum (g dw m −2 ) retrieved in intact and reconstructed cores from the two stations. For slopes and intercepts, standard errors are reported in brackets.  Table 2 reports the slopes of linear regressions between measured fluxes or calculated rates and biomass of A. succinea and R. philippinarum in intact and reconstructed sediment cores, whereas Table 3 reports the results of the ANCOVA where such regressions were compared. As a general outcome, the enhancement of specific processes by the two functional groups of macrofauna (e.g., oxygen uptake, NH 4 + efflux and denitrification of water column NO 3 − -Dw-) suggested much larger effects of the deposit feeder as compared to the filter feeder (Table 2). For example, the increase in benthic oxygen consumption per gram of dry weight of the burrower varied between 81 and 156 µmol g dw

Macrofauna Stimulate Differentially Biogeochemical Processes in Intact and in Reconstructed Sediments
This is likely due to the construction and irrigation of subsurface burrow structures, to the increased sediment-water exchange surface and to the stimulation of microbial activity within sediments [28]. Interestingly, the stimulatory effects of A. succinea on most measured benthic fluxes, with NO 3 − as the only exception, were generally different in intact and reconstructed sediments, whereas this was not true for R. philippinarum (Tables 2 and 3). In reconstructed sediment cores A. succinea produced much higher enhancement of O 2 respiration and of water column NO 3 − consumption via denitrification (by a factor of 2) and of NH 4 + recycling (by a factor 5) as compared to intact sediments. In unmanipulated sediments, A. succinea stimulated denitrification of NO 3 − produced via nitrification (Dn), which is expected, whereas this was not apparent in reconstructed sediments. In the latter, A. succinea stimulated the denitrification of water column NO 3 −, likely due to recently constructed burrows with sharp redox gradients and limited numbers of nitrifying bacteria on their walls. Besides differences in the degree of stimulation of certain processes, measurements in intact and reconstructed bioturbated sediments led sometimes to contrasting effects of macrofauna. For example, A. succinea in intact cores enhanced DIC retention in sediments, whereas the same organism in reconstructed cores stimulated DIC release to the water (Tables 2 and 3). Opposite effects by burrowing fauna were apparent also for NO 2 − fluxes.
Interestingly, the effects of R. philippinarum on solute fluxes in intact and reconstructed cores were not statistically different (Tables 2 and 3). We speculate that clams drive fluxes mostly through their metabolic activities but produce a limited effect, at least in the short-term, on sedimentary variables and on microbial communities within sediments.

Conclusions
Results of this study confirm the key role of bioturbating macrofauna in benthic biogeochemistry, with deposit feeders simulating the removal of nitrogen via denitrification and increasing the denitrification efficiency of the benthic system. Deposit feeders also increase the rates of oxygen consumption, partly due to chemical or biological oxidation of anaerobic metabolism end products. As such, they may buffer the toxicity of sulphides and act upon redox-sensitive processes such as P retention, ultimately producing positive feedbacks on benthic biodiversity and negative feedbacks on eutrophication. In the Sacca di Goro lagoon, clams are cultivated, and their density is orders of magnitude higher than under natural conditions. Clams are demonstrated to favor benthic-pelagic coupling, but with such densities, they may determine a too rapid (re)cycling of excreted nutrients associated to ingested phytoplankton. As clams produce a scarce stimulation of denitrification, they maintain low the denitrification efficiency, and they may promote (or sustain) eutrophic conditions.
The present study suggests that results of experiments with sediments and macrofauna manipulations need to be carefully considered. We demonstrated here that the degree of stimulation of commonly measured biogeochemical processes as aerobic, anaerobic (e.g., denitrification), and total respiration or nutrient fluxes by macrofauna can be substantially different when calculated in intact, not manipulated natural sediments compared to reconstructed microcosms. Reasons are probably multiple and include, first of all, the consequences of sediment sieving and packing, which deeply alter vertical gradients, organic matter distribution, the sediment redox, the pore water chemistry and likely affect important fractions of reactive components of the organic matter pools. Sieving also removes all the macrofauna in order to highlight the biogeochemical effects of a single species. This helps to have very clear, often linear, and highly significant results; on the other hand, the simplification of the community removes an unaccountable number of ecological interactions that shape the true functioning of the benthic ecosystems. Results obtained from manipulated sediments are real, as they are measured, but are likely never found in the field, due to oversimplification of the experimental approach, including the removal of chemical gradients and multiple ecological interactions. Aquatic ecologists often use manipulative approaches, where they remove species to analyze the effect of a few targeted organisms. Results from this work, and in particular, fluxes measured in manipulated sediments, suggest that clear and easy-to-interpret results are obtained from these manipulations. The intensity of process stimulation calculated in manipulative studies can then be coupled to data on macrofauna abundances in order to upscale the effects of bioturbation at the whole system level. In doing so, one can get results that are quite far from the reality. Something similar likely occurs when incubating a deep burrower or a clam in a small vial to calculate their respiration and excretion rates; they will try to dig through the glass walls during the incubation and their metabolic activity will be much higher as compared to when they are laying in their burrows, surrounded and protected by sediments. A major fraction of the effects of macrofauna on sediments are indirect, and associated to the way they affect microbial communities and metabolic activities and, with them, chemical microgradients within sediment, which are flux drivers. Measurements in manipulative experiments should also consider these aspects and consider or quantify how long it takes for macrofauna to produce a steady state in terms of microbial communities, activities, and chemical gradients in manipulated sediments. The latter require a long conditioning time before measurements, as the appropriate preincubation period is probably closer to weeks than to days.