Filtering Activity and Nutrient Release by the Keratose Sponge Sarcotragus spinosulus Schmidt, 1862 (Porifera, Demospongiae) at the Laboratory Scale

: Sponges are an important constituent of ﬁlter-feeder benthic communities, characterized by high ecological plasticity and abundance. Free bacteria constitute an important quota of their diet, making them excellent candidates in aquaculture microbial bioremediation, where bacteria can be a serious problem. Although there are studies on this topic, certain promising species are still under investigation. Here we report applied microbiological research on the ﬁltering activity of Sarcotragus spinosulus on two different concentrations of the pathogenic bacterium Vibrio parahaemolyticus in a laboratory experiment. To evaluate the effects of the ﬁltration on the surrounding nutrient load, the release of ammonium, nitrate, and phosphate was also measured. The results obtained showed the efﬁcient ﬁltration capability of S. spinosulus as able to reduce the Vibrio load with a maximum retention efﬁciency of 99.72% and 99.35% at higher and lower Vibrio concentrations, respec-tively, and remarkable values of clearance rates (average maximum value 45.0 ± 4.1 mL h − 1 g DW − 1 ) at the highest Vibrio concentration tested. The nutrient release measured showed low values for each considered nutrient category at less than 1 mg L − 1 for ammonium and phosphate and less than 5 mg L − 1 for nitrate. The ﬁltering activity and nutrient release by S. spinosulus suggest that this species represents a promising candidate in microbial bioremediation, showing an efﬁcient capability in removing V. parahaemolyticus from seawater with a contribution to the nutrient load.


Introduction
Marine sponges (Phylum Porifera) are ancient metazoans that dominate many of the hard-bottom benthic habitats around the world along a wide geographical distribution and depth range [1,2]. These sessile organisms are benthic filter-feeders with a high capability to filter huge amounts of water (0.002-0.84 mL s −1 cm 3 of sponge tissue) through their aquiferous system [3][4][5] and to retain a wide range of 0.1-50 µm organic particles, including phytoplankton, heterotrophic eukaryotes, bacteria, and viruses with a retention efficiency of up to 99% for nano and picoplankton [6][7][8][9][10][11][12]. In addition, sponges play a relevant role in benthic-pelagic coupling [13,14], serve as mediators of the biogeochemical flow by respiring organic matter and facilitating the consumption and release of nutrients, such as ammonium, nitrate, and phosphate [15].
In highly anthropized marine environments, such as intensive or confined mariculture systems, the excessive release of excreta from farmed species and organic matter from uneaten feed create favorable conditions for pathogenic bacteria growth, especially Vibrio, responsible for diseases and high mortality in target species, with consequent economic losses [31][32][33][34][35][36][37]. To overcome this problem, the use of antibiotics has spread despite the increase in production costs and the negative consequences on farmed species and the surrounding environment. Indeed, antibiotic residues can remain in products for human consumption and antibiotics released into the environment can induce the development and spread of antibiotic-resistant bacteria in the food chain [38].
The co-cultivation of sponges in association with mariculture plants may be considered an eco-friendly alternative to prevent and control the growth and spread of bacteria, pathogenic and non-pathogenic, in aquaculture waste [19,21,24,26,40]. sponge cultivation may be suitable for the eco-sustainable supply chain of biomass for certain target species [27,41,46,47]. The sponge biomass obtained in polyculture systems has considerable potential from a commercial point of view, having good appeal for hobbyists as well as cosmetic and natural bioactive compound companies (e.g., [25]).
Zoo-remediation is a poorly considered approach to reduce aquatic pollution, primarily due to ethical reasons. In the case of invertebrate animal species, while overcoming ethical issues, further criticism, such as the availability of the appropriate amount of biomass to obtain a valuable result, management of the zoo-remediator biomasses, survival skills in critical conditions, and excessive collection efforts on wild populations, needed to be addressed [17,48]. To find sustainable solutions to these issues, recently an in situ innovative integrated multitrophic aquaculture (IMTA) system in a Mediterranean fish farm, in which explants of the keratose sponge Sarcotragus spinosulus Schmidt 1862 (Porifera, Demospongiae) were co-cultured, showed promising survival and growth performances with a doubling of the sponge biomass after one year of rearing [49]. To date, no studies are available on the filtering performance and nutrient release [15,25], despite representing a deeply studied species in the research of basic and applied biology (e.g., microbiology, mariculture and the extraction of bioactive compounds) [50][51][52]. Conversely, the natural products that can be extracted from this species are well-known (e.g., polyprenylhydroquinones) [53] and have drawn particular attention due to the wide spectrum of their antibacterial, antiviral, anti-inflammatory, and cytotoxic activities [52].
In this paper, the filter-feeding activity of S. spinosulus on the bacterial load was investigated in laboratory conditions by estimating the clearance rate and retention efficiency versus the Gram-negative halophilic bacterium Vibrio parahaemolyticus (family Vibrionaceae). Data were also related to the release of nutrients (NH 4 + , NO 3 − , and PO 4 3− ). Thus, the present study represents a contribution to the knowledge of the filtering activity and nutrient release of S. spinosulus, which can permit a better focus on its suitability as a microbial bioremediator within sustainable mariculture facilities.

Studied Species
Sarcotragus spinosulus Schmidt 1862 (Porifera, Demospongiae, Keratosa, Dictyoceratida, Irciniidae) is a massive horny sponge, common in Mediterranean coastal environments, occurring in shallow waters and also just below the tide line [54][55][56]. Among the Mediterranean demosponges, this species can be considered one of the most light-tolerant, being screened by a thick layer of superficial pigmented tissue (pinacoderm) made by a large number of melanocytes and a dense bacterial simbiocortex [55]. The species is considered of high ecological plasticity, being able to live both in high-energy vertical cliffs and in low-energy semi-enclosed bays with a high sedimentation regime [47,55,57].

Sponge Sampling
Sponge specimens of S. spinosulus were randomly collected from Mar Grande of Taranto (40 • 21 N-17 • 18 E) by scuba diving at a depth of 5-10 m (T = 20 • C) in January 2019. The samples were carefully detached from the substratum, immediately transported to the laboratory within cooled bags, then cleaned of sediment and macrofouling organisms with seawater, and kept in an aquarium containing 100 L of artificial filtered seawater (AFSW) (0.22 µm pore size filters, Millipore). Samples underwent acclimatization for 2 days before testing in a temperature-controlled room (20 • C), and the water was substituted twice with new AFSW to avoid water contamination by bacteria and particulate matter (starved sponge specimens).

Experimental Procedures
The laboratory experiment aimed to investigate the filtering activity of the demosponge Sarcotragus spinosulus at two different Vibrio parahaemolyticus concentrations and to evaluate its nutrient release (NH 4 + , NO 3 − , and PO 4 3− ). The CIRPS 4253 V. parahaemolyticus strain, part of our laboratory collection [58], was used to prepare AFSW with 10 4 CFU mL −1 (concentration C1) and 10 6 CFU mL −1 (concentration C2). CIRPS 4253 was grown in 3% NaCl Luria-Bertani (LB, OXOID, Milan, Italy) broth and incubated overnight (O/N) at 37 • C. The concentration of viable bacteria (CFU mL −1 ) in the O/N culture was calculated using a standard viable count assay. Briefly, 0.1 mL of the serially diluted bacterial culture was plated on 3% NaCl LB, and the plates were incubated O/N at 37 • C. The bacterial colonies formed on each plate were counted, and the CFU mL −1 was calculated with respect to the dilution factor and the volume plated. The test to evaluate the CFU mL −1 of the O/N culture was performed in triplicate.
The filtering experiment was performed in triplicate and consisted of 30 tanks (10 tanks per experiment) placed in a temperature-controlled room at 20 • C (the average seasonal seawater temperature during the sponge harvest period) with continuous airing and artificial lighting (16:8 light/dark, light intensity 250 µmol m −2 s −1 ) and each filled with 3 L of AFSW. The tanks were kept under the same experimental conditions during the three experimental procedures.
A total of 10 tanks per experiment were set up. Six tanks contained one starved sponge specimen each, two of which were aseptically inoculated with an aliquot of the CIRPS 4253 O/N culture to reach the concentration C1 (treatment tanks, T_C1); similarly, two tanks were inoculated to reach the concentration C2 (treatment tanks, T_C2); two tanks were not inoculated and used as a control to test the possible release of bacteria belonging to the family Vibrionaceae due to the sponge (sponge control tanks, SC). Four tanks did not contain sponges: two of which were considered as negative control tanks containing only AFSW (NC), and the last two tanks were considered as positive controls inoculated with the CIRPS 4253 O/N culture at the final concentrations C1 and C2 (PC_C1 and PC_C2, respectively).
The bacterial viable count and content of the nutrients were evaluated in each tank at five sampling times (0, 2, 4, 24, and 48 h). The viable bacterial count was performed at each sampling time by spreading 0.1 mL of serially diluted seawater samples onto thiosulfatecitrate-bile salt-sucrose agar (TCBS, OXOID, Milan, Italy) with 2% NaCl, a selective and differential medium for halophilic Vibrio. The plates were incubated at room temperature (22 ± 2 • C) for 24-48 h. Colonies with a yellow color (considered to be V. parahaemolyticus) formed in each plate were counted, and the CFU mL −1 was calculated with respect to the dilution factor and the volume plated. The data were reported as the mean value ± the standard error (SE) of each experimental tank set.
The well-being of the sponge specimens was monitored throughout the experiment by observing the sponge surface and the osculum openings. At the end of the experiment, the volume of each specimen was measured by means of a graduated beaker (125 ± 28 mL, mean value), then the sponges were dried in preweighed aluminum foil at 100 • C for 24 h and weighed to determine the dry weight (DW, mean value 44.8 ± 10 g).

Filtering Activity Assessment
At each sampling time, the retention efficiency (R) was calculated as a percentage for the difference in bacterial concentrations by the following equation: where C t0 is the initial bacterial concentration and C tx is the bacterial concentration at each successive sampling time [42]. The clearance rate (CR) was estimated following the equation given by Coughlan [59], which measures the bacterial removal from the seawater as a function of time T, volume V of water used in the filtering experiment, and sponge size W: The data were reported as weight-specific clearance rates and expressed in milliliters per hour per gram of dry sponge tissue (mL h −1 g DW −1 ).

Nutrient Analysis
To evaluate the contribution of S. spinosulus in terms of the dissolved inorganic nutrients in the surrounding environment, the release of ammonium (NH 4 + ), nitrate (NO 3 − ), and phosphate (PO 4 −3 ) in the seawater at each sampling time during the experiment was measured.
The ammonium content was monitored with a pH meter (HI 5222, Hanna Instruments, Woonsocket, RI, USA) equipped with an ammonium ion-selective electrode (HI 4101, Hanna Instruments, Woonsocket, RI, USA), calibrated in the range 0-10 mg L −1 according to the manufacturer's instructions and expressed as mg NH 3 L −1 [60,61].
The nitrate content was determined spectrophotometrically with a Beckman DU 6400 spectrophotometer according to [62] with minor modifications. Seawater samples (1.25 mL) were treated with 0.025 mL HCl and the specific absorbance, measured at 220 nm, was adjusted by subtracting the nonspecific absorbance at 275 nm due to the interference of organic compounds. The nitrate concentration was determined by referring to a standard curve with a range of 0-10 mg L −1 and expressed as mg NO 3 − L −1 . The phosphate content was determined according to Strickland and Parsons [63] by the spectrophotometric determination of a blue phosphomolybdic complex that specifically absorbs at 882 nm. The phosphate concentration was determined by referring to a standard curve with a range of 0-5 mg PO 4 3− L −1 . Finally, the sponge excretion rate (E) was calculated by multiplying each nutrient concentration value for the water volume in a tank, dividing for the dry weight (DW) biomass per unit time (h). Consequently, the nutrient excretion rates were expressed as micromoles N or P per gram of dry weight per hour (µmol g DW −1 h −1 ) according to [64]: where Nc t0 and Nc tx were the nutrient (NH 4 + , NO 3 − , and PO 4 3− ) contents in seawater at the initial time t 0 and at each sampling time (t 1 -t 4 ), respectively.

Statistical Analysis
All experimental data were computed as dependent variables using PERMANOVA as an approach similar to parametric ANOVA. Univariate PERMANOVA tests were run on Bray-Curtis similarity matrices with 9999 permutations [65]. The bacterial concentration (C, 3 levels) and time (t, 5 levels) factors were used to detect differences in the CFU, retention efficiency (R), clearance rate (CR), nutrient concentrations (NH 4 + , NO 3 − and PO 4 3− ) and excretion rate (E) in t × C interactions. Each interaction was individually analyzed using Univariate PERMANOVA tests with the same experimental design. If necessary, transformed data in a Bray-Curtis similarity matrix with 9999 permutations was used to perform the analyses [65]. If it was impossible to obtain enough permutations for PERMANOVA analysis, the reference p was obtained using a permutation simulation test (Monte Carlo test). The pairwise test was applied to discover statistically significant differences in each pair of factor levels based on the significance value of PERMANOVA/Monte Carlo tests. All analyses were conducted using Primer v6+ PERMANOVA software [66].

Filtering Capability
The sponge health status was assessed visually during both the starvation period (2 days) and the 48 h experiment showing no signs of stress neither to the external surface nor to changes on oscula openings.

Statistical Analysis
All experimental data were computed as dependent variables using PERMANOVA as an approach similar to parametric ANOVA. Univariate PERMANOVA tests were run on Bray-Curtis similarity matrices with 9999 permutations [65]. The bacterial concentration (C, 3 levels) and time (t, 5 levels) factors were used to detect differences in the CFU, retention efficiency (R), clearance rate (CR), nutrient concentrations (NH4 + , NO3 -and PO4 3-) and excretion rate (E) in t × C interactions. Each interaction was individually analyzed using Univariate PERMANOVA tests with the same experimental design. If necessary, transformed data in a Bray-Curtis similarity matrix with 9999 permutations was used to perform the analyses [65]. If it was impossible to obtain enough permutations for PER-MANOVA analysis, the reference p was obtained using a permutation simulation test (Monte Carlo test). The pairwise test was applied to discover statistically significant differences in each pair of factor levels based on the significance value of PER-MANOVA/Monte Carlo tests. All analyses were conducted using Primer v6+ PER-MANOVA software [66].

Filtering Capability
The sponge health status was assessed visually during both the starvation period (2 days) and the 48 h experiment showing no signs of stress neither to the external surface nor to changes on oscula openings.
The Vibrio concentration increased in the control tanks (PC_C1 and PC_C2), reaching, at the end of the experiment, the same mean value (4.9 ± 0.1 × 10 6 ). In the negative control (NC), no bacterial colonies were registered during the experiment. As for the SC control tanks, the appearance of bacteria was observed starting from t1. The bacterial concentration in these tanks remained at lower levels (from 63 ± 23 at t1 to 25 ± 2.89 at t4) than each other tanks (five orders of magnitude lower than the PCs), without subsequent increases.
The retention efficiency of S. spinosulus in removing bacteria, calculated as a percentage change in the bacterial concentration between two successive times (retention efficiency = R), was affected by the initial concentration of bacteria (C), the time (t), and their interaction (t × C) (univariate PERMANOVA, df = 3, pseudo-F = 7.3315, p = 0.01). Significant differences were highlighted both between the two Vibrio concentrations and over time. At t1 (two hours from the beginning), R showed lower values in the T_C1 tanks compared to the T_C2 ones (t1:33.3% and 68.5%, respectively; pairwise test, T_C2 >> T_C1) (Figure 2a) This relationship was maintained throughout the experiment and the differences among treatments and controls continually increased. The pairwise comparison as a function of the Vibrio concentration highlighted a significant decrease over time in both treatments (Pairwise test, TC_C1: t0 > t1 > t2 > t3 >> t4 and TC_C2: t0 >> t1 = t2 >> t3 >> t4). At the end of the experiment, after 48 h (t4), the Vibrio concentrations showed the highest decrease in the treatment tanks, reaching values that were three (T_C1 = 67 ± 7) and four (T_C2 = 383 ± 41) orders of magnitude lower than the initial concentrations.
The Vibrio concentration increased in the control tanks (PC_C1 and PC_C2), reaching, at the end of the experiment, the same mean value (4.9 ± 0.1 × 10 6 ). In the negative control (NC), no bacterial colonies were registered during the experiment. As for the SC control tanks, the appearance of bacteria was observed starting from t1. The bacterial concentration in these tanks remained at lower levels (from 63 ± 23 at t1 to 25 ± 2.89 at t4) than each other tanks (five orders of magnitude lower than the PCs), without subsequent increases.
The retention efficiency of S. spinosulus in removing bacteria, calculated as a percentage change in the bacterial concentration between two successive times (retention efficiency = R), was affected by the initial concentration of bacteria (C), the time (t), and their interaction (t x C) (univariate PERMANOVA, df = 3, pseudo-F = 7.3315, p = 0.01). Significant differences were highlighted both between the two Vibrio concentrations and over time. At t1 (two hours from the beginning), R showed lower values in the T_C1 tanks compared to the T_C2 ones (t1:33.3% and 68.5%, respectively; pairwise test, T_C2 >> T_C1) (Figure 2a), and reached values close to 100% at t4 (48 h) (99.35% and 99.72% at C1 and C2, respectively). This latter value was reached for the T_C2 samples as early as t3 (24 h) (pairwise test t3: T_C1 << T_C2; t4: T_C1 = T_C2). The clearance rate was calculated considering the volume of water processed by the sponge for a certain time (CR: mL h −1 g DW −1 ± SE), as shown in Figure 2b. The maximum values of the clearance rates were recorded at t2 for T_C1 (16.6 ± 0.9) and at t1 for T_C2 (45.0 ± 4.1) (Pairwise test, t x C, on factor t: T_C1: t1 = t2 >> t3 < t4; T_C2: t1 >> t2 = t3 < t4). The mean value over a 48 h trial (t4) was 7.4 ± 0.2 at the lower V. parahaemolyticus concentration tested (T_C1) and 8.7 ± 0.9 at the higher concentration tested (T_C2). The relationship between the treatments (T_C1 and T_C2) is superimposable to that of retention efficiency. Statistical analysis revealed a significant relationship as a function of time (t) and of the interaction between time and the initial bacterial concentration (C) (univariate PER-MANOVA, df = 3, pseudo-F = 3.5463, p = 0.009).

Nutrient Release
The nutrient release (mg L −1 ± SE) from S. spinosulus during the laboratory experiments showed low values for each considered nutrient category (NH4 + , NO3 -, and PO4 3-), which were less than 1 mg L −1 for ammonium and phosphate and less than 5 mg L −1 for nitrate ( Figure 3). The highest nutrient concentrations were found in each tank containing sponges (T_C1, T_C2, and SC), while, in each control tank (PC_C1, PC_C2, and NC), the nutrient values remained very low.
Regarding the ammonium, a considerable increase was recorded in the tanks containing the sponges (T_C1, T_C2, and SC), with final values (t4) higher than 0.5 mg L −1 (Figure 3a). Statistical analysis showed a significant relationship of NH4 + concentration for the factor time (t), concentration (C), and their interaction (t × C). The a posteriori test at t4 highlighted significant differences between all tanks containing sponges (Pairwise test: T_C1> T_C2 >> SC) and no difference between the positive controls (Pairwise test: PC_C1 = PC_C2). The highest ammonium concentrations were recorded in the T_C1 and T_C2 treatments (0.80 ± 0.01 and 0.71 ± 0.38, respectively) with T_C1 > T_C2. In the negative control (NC) samples, the NH4 + remained zero during the whole experiment. The clearance rate was calculated considering the volume of water processed by the sponge for a certain time (CR: mL h −1 g DW −1 ± SE), as shown in Figure 2b. The maximum values of the clearance rates were recorded at t2 for T_C1 (16.6 ± 0.9) and at t1 for T_C2 (45.0 ± 4.1) (Pairwise test, t × C, on factor t: T_C1: t1 = t2 >> t3 < t4; T_C2: t1 >> t2 = t3 < t4). The mean value over a 48 h trial (t4) was 7.4 ± 0.2 at the lower V. parahaemolyticus concentration tested (T_C1) and 8.7 ± 0.9 at the higher concentration tested (T_C2). The relationship between the treatments (T_C1 and T_C2) is superimposable to that of retention efficiency. Statistical analysis revealed a significant relationship as a function of time (t) and of the interaction between time and the initial bacterial concentration (C) (univariate PERMANOVA, df = 3, pseudo-F = 3.5463, p = 0.009).

Nutrient Release
The nutrient release (mg L −1 ± SE) from S. spinosulus during the laboratory experiments showed low values for each considered nutrient category (NH 4 + , NO 3 − , and PO 4 3− ), which were less than 1 mg L −1 for ammonium and phosphate and less than 5 mg L −1 for nitrate ( Figure 4). The highest nutrient concentrations were found in each tank containing sponges (T_C1, T_C2, and SC), while, in each control tank (PC_C1, PC_C2, and NC), the nutrient values remained very low.
Regarding the ammonium, a considerable increase was recorded in the tanks containing the sponges (T_C1, T_C2, and SC), with final values (t4) higher than 0.5 mg L −1 (Figure 4a). Statistical analysis showed a significant relationship of NH 4 + concentration for the factor time (t), concentration (C), and their interaction (t × C). The a posteriori test at t4 highlighted significant differences between all tanks containing sponges (Pairwise test: T_C1> T_C2 >> SC) and no difference between the positive controls (Pairwise test: PC_C1 = PC_C2). The highest ammonium concentrations were recorded in the T_C1 and T_C2 treatments (0.80 ± 0.01 and 0.71 ± 0.38, respectively) with T_C1 > T_C2. In the negative control (NC) samples, the NH 4 + remained zero during the whole experiment. The nitrates also showed a continuous increase over time in the tanks with sponges (T_C1, T_C2, and SC) while each other tank showed almost constant values, or a slight decrease (Figure 4b). The NO 3 − concentration was affected by the initial concentration (C), time (t), and the interaction between the two factors (C × t) (univariate PERMANOVA, df = 8, pseudo-F = 10.094, p = 0.001). At the end of the experiment, the highest values were recorded in T_C1 (4.24 ± 0.24), greater than SC and T_C2 and the other tanks (pairwise test, t4: T_C1> SC> T_C2> NC = PC_C2 = PC_C1).
The phosphates showed a general increase over time (Figure 4c), albeit with some slight variations between times t2 and t3, reaching a maximum value near 0.20 mg L −1 . PERMANOVA highlighted significant differences as a function of time and in the interaction between the time and concentration (C × t) (univariate PERMANOVA, df = 8, pseudo-F = 11.367, p = 0.0001). The highest values at the end of the experiment (t4) were recorded in the SC samples (0.21 ± 0.01), followed by the two treatments T_C1 (0.18 ± 0.01) and T_C2 (0.15 ± 0.01), the negative control NC (0.09 ± 0.001), and the two positive controls PC_C2 and PC_C1 (pairwise test, t4: SC> T_C1> T_C2 >> NC >> PC_C2 = PC_C1). The nitrates also showed a continuous increase over time in the tanks with sponges (T_C1, T_C2, and SC) while each other tank showed almost constant values, or a slight decrease (Figure 3b). The NO3 -concentration was affected by the initial concentration (C), time (t), and the interaction between the two factors (C × t) (univariate PERMANOVA, df = 8, pseudo-F = 10.094, p = 0.001). At the end of the experiment, the highest values were recorded in T_C1 (4.24 ± 0.24), greater than SC and T_C2 and the other tanks (pairwise test, t4: T_C1> SC> T_C2> NC = PC_C2 = PC_C1).
The phosphates showed a general increase over time (Figure 3c), albeit with some slight variations between times t2 and t3, reaching a maximum value near 0.20 mg L −1 . PERMANOVA highlighted significant differences as a function of time and in the interaction between the time and concentration (C × t) (univariate PERMANOVA, df = 8, pseudo-F = 11.367, p = 0.0001). The highest values at the end of the experiment (t4) were recorded in the SC samples (0.21 ± 0.01), followed by the two treatments T_C1 (0.18 ± 0.01) and T_C2 (0.15 ± 0.01), the negative control NC (0.09 ± 0.001), and the two positive controls PC_C2 and PC_C1 (pairwise test, t4: SC> T_C1> T_C2 >> NC >> PC_C2 = PC_C1).
The excretion rate (E, µmol g DW −1 h −1 ) of nutrients by S. spinosulus-calculated by relating the concentration of nutrients, the sponge biomass, and the processed water-is given in Figure 4. The highest values were recorded for ammonium (0.73 at t4), followed by nitrates (0.23 at t1) and phosphates (0.02 at t2). A decrease in E was observed for the ammonium and nitrates during the experiment, except for ammonium in T_C2, which contrarily showed an increase (Figure 4). The excretion rate (E, µmol g DW −1 h −1 ) of nutrients by S. spinosulus-calculated by relating the concentration of nutrients, the sponge biomass, and the processed water-is given in Figure 4. The highest values were recorded for ammonium (0.73 at t4), followed by nitrates (0.23 at t1) and phosphates (0.02 at t2). A decrease in E was observed for the ammonium and nitrates during the experiment, except for ammonium in T_C2, which contrarily showed an increase (Figure 4).

Discussion and Conclusions
The polyculture of fish with organisms at different food web levels has considerable environmental and economic potential, particularly if edible and/or non-edible species with a potentially high commercial value are co-cultured. Among the non-edible organisms, Porifera (sponges) represents a valuable candidate due to its key role in organic matter recycling and sustainable and commercially appealing biomass production [45].
The filtering activity and nutrient release by S. spinosulus obtained in the present study demonstrate that this species represents a valuable candidate in microbial bioremediation, showing the efficient capability in removing V. parahaemolyticus from seawater in the laboratory experiment. The results obtained showed that S. spinosulus effectively controlled the growth of V. parahaemolyticus in the laboratory experiment. This finding is rather appealing due to the high pathogenicity of Vibrio spp. for both humans and aquaculture animals. V. parahaemolyticus, indeed, is a human bacterial pathogen widely occurring in marine environments, frequently isolated from a variety of seafood, including bivalves, crustaceans, and fish [67]. Vibriosis is currently responsible for most disease outbreaks in

Discussion and Conclusions
The polyculture of fish with organisms at different food web levels has considerable environmental and economic potential, particularly if edible and/or non-edible species with a potentially high commercial value are co-cultured. Among the non-edible organisms, Porifera (sponges) represents a valuable candidate due to its key role in organic matter recycling and sustainable and commercially appealing biomass production [45].
The filtering activity and nutrient release by S. spinosulus obtained in the present study demonstrate that this species represents a valuable candidate in microbial bioremediation, showing the efficient capability in removing V. parahaemolyticus from seawater in the laboratory experiment. The results obtained showed that S. spinosulus effectively controlled the growth of V. parahaemolyticus in the laboratory experiment. This finding is rather appealing due to the high pathogenicity of Vibrio spp. for both humans and aquaculture animals. V. parahaemolyticus, indeed, is a human bacterial pathogen widely occurring in marine environments, frequently isolated from a variety of seafood, including bivalves, crustaceans, and fish [67]. Vibriosis is currently responsible for most disease outbreaks in aquaculture [31][32][33][34]68].
Although the release of Vibrionaceae found during our experiment in sponge control tanks could question their potential application for bioremediation by appearing to be a problem rather than a solution, as questioned by some authors [69], the recorded concentrations in the sponge control tanks are negligible compared to the treatments (four/six orders of magnitude lower), and are likely related to the response of the specimens to the experimental conditions. We emphasized, however, that these bacteria appeared after two hours from the beginning of the experiment and remained at low concentrations up to the end. The final balance between the bacteria removed and those that appeared is strongly in favor of removal.
The retention efficiency (R) values found in the present experiment represent a further encouragement to the use of this sponge in aquaculture. Our findings showed R values up to 99.72%, in line with further studies that reported retention efficiencies ranging from 70% to 99% for small suspended particles such as Vibrio spp. [4,11,42,[70][71][72]. Our experiment showed that the retention efficiency increased gradually, reaching the maximum value after 48 h at C1 and after 24 h at C2. This latter evidence indicates that the retention of Vibrio cells by S. spinosulus is positively related to their greater availability in the experimental tanks.
The clearance rate (CR), indicating a measure of the food depletion as a function of time, the cleared water volume, and the sponge size [10,59], estimated at both Vibrio concentrations tested, demonstrated the worthy filtering performances of S. spinosulus. The highest CR value was registered at the highest Vibrio concentration (C2), with values ranging between 8.7 and 45.0 mL g DW −1 h −1 . The highest value was quickly observable two hours after the start of the experiment highlighting that, as for R, the increased availability of Vibrio positively affected the rate of bacterial concentration change for this sponge. S. spinosulus, at its maximum filtering activity, was able to clean up a water volume of 17 times its volume in 1 h.
Although the comparison of the CR between different sponge species is challenging due to the intraspecies variability, the effect of sponge size, the morpho-physiological features, and the different units in which CR are expressed [4,25,73], our results are comparable with those reported for other Mediterranean species (Supplementary Table S1). A careful analysis of the results requires further considerations related to the characteristics of the tested sponge. S. spinosulus is attributed to high microbial abundance sponge species (HMA), hosting an abundant and diversified microbial community (two/four magnitude orders of bacteria per gram of sponge tissue higher than the surrounding seawater) and lower pumping rate than low microbial abundant (LMA) species [5,74].
The CR found here is of the same order of magnitude reported for Mediterranean HMA sponge species at a comparable investigation time and sponge size (Supplementary Table S1) [5,25]. Although the specimens of S. spinosulus used in our experiment were larger than the closely related Mediterranean species compared, such as Ircinia variabilis and Spongia officinalis, the CR here found is in line with or greater than that measured at a comparable time evaluation (Supplementary Table S1).
In the present study, the nutrient release (ammonium-NH 4 + , nitrate-NO 3 − , and phosphate-PO 4 −3 ) from S. spinosulus was measured for its contribution to the nutrient overload in the surrounding seawater. Our results are in line with those reported in the literature, confirming, as for other Mediterranean demosponges, the behavior of the studied HMA sponge S. spinosulus as a nutrient source (Supplementary Table S2) [15,75].
The major contribution is due to the release of ammonium with a positive relationship with the availability of bacteria in the tanks. For nitrogen, the final ammonium release (t4) was about eight times higher than the initial one, while the nitrates showed only a doubling of values. At the low bacterial experimental concentration (C1), a higher ammonium and nitrate release was observed with respect to the treatment C2 and SC (sponge control), likely linked to the decrease of bacteria in the water due to the sponge filtering activity. The contribution of the sponge in the release of phosphates in any experimental condition was negligible.
Metabolic processes in sponges occur at the cellular level and cannot be neglected in the role of the associated microbial community in the sponge metabolic balance. Large metabolic differences (filtration rates, nutrient flow, etc.) between LMA and HMA sponges were documented [15]. Sponges feed on both particulate organic matter (POM) and dissolved organic matter (DOM). Some sponge species have high value for use in IMTA due to their ability to convert DOM into POM, making it available for other suspension feeders and detritivores [45,76]. Among POM can be counted several types of planktonic cells: primarily picoplankton (mainly bacterioplankton and phytoplankton) and partly nanoplankton (e.g., diatoms), but also non-living particles (i.e., debris).
As for DOM, the ability of sponges to remove or release dissolved organic or inorganic compounds depends on photoautotrophic and chemotrophic processes mediated by the associated microbial community (such as archaea, bacteria, cyanobacteria, yeasts, and also diatoms) [15]. The POM utilization produces dissolved organic compounds, ammonium, and phosphate, which can be released directly into the water but can be transformed in the processes of nitrification, photoautotrophy, denitrification, and/or anammox (oxidation of ammonia in the absence of O2) by the associated microbiomes particularly in HMA species [15,[77][78][79][80].
In addition, the sponge microbiota can influence whether the holobiont acts as a net source or sink of bioavailable nitrogen [81,82] and can be capable of releasing nitrogen at ecologically relevant values in oligotrophic marine environments [83,84]. Regarding the phosphorus flow, sponges are considered as sources of PO 3 + independently of whether they are HMA or LMA species; moreover, research demonstrated the ability to store intracellular polyphosphate granules in three reef species mediated through symbiotic microorganisms [75,85]. Therefore, sponges with their microbiota can affect both the quantity and speciation of inorganic nutrients, making them available to nearby primary producers [77].
Our findings showed that S. spinosulus (attributed to HMA) acts as a source of inorganic nitrogen since, due to its microbiota, the ammonium produced in the metabolic processes is nitrified to NOx [51]. This nitrogen availability could facilitate the growth of primary producers, such as phytoplankton, which can be exploited by further filter feeders, such as bivalves or could be utilized by seaweed, thus, underlining its utility in IMTA systems.
The sponges grown in IMTA carry out their filtration activity by removing both the organic matter from the water and harmful particles (bacteria, viruses, and also fecal pellets) and converting these into food for other invertebrates, operating an important bypass from DOM to POM. Some of these bacteria could be pathogenic for both fish and humans; thus, their removal could represent a useful tool for reducing the use of antibiotics in aquaculture, acting in parallel for good environmental quality, on the organoleptic qualities of the farmed products and on the problem of antibiotic resistance. Considering the rapid expansion of the aquaculture sector, combining complementary filter-feeder macroinvertebrates, such as sponges with traditional mariculture, could allow aquaculture to reach environmental sustainability of mariculture, minimizing the microbial impacts.
In conclusion, we provided evidence that S. spinosulus is able to remove the inoculated V. parahaemolythicus from seawater in test tanks at different concentrations, showing better performance at the higher concentration, with a contribution to the nutrient load. The promising survival and growth performance already obtained by this species in a Mediterranean IMTA system [49] highlights the ability of this sponge species to withstand the environmental conditions of an aquaculture facility. In addition, the biomass obtained [49] appears to be sufficient to implement the rearing system over time, thus avoid-ing ethical problems due to the depletion of wild stocks. S. spinosulus represents an effective mediator and bioremediator in integrated multitrophic aquaculture systems.