Particulate organic matter as a food source for a coral reef sponge

Sponges are ubiquitous aquatic, sessile, filter feeders. In some marine ecosystems, such as Antarctica or certain tropical reefs, they constitute a significant proportion of the benthic biomass(Barthel and Gutt, 1992; Barnes and Bell, 2002). Sponges' ability to satisfy their metabolic needs in diverse niches has a major influence on their world distribution.

Interestingly, coral reef sponges survive in frequently oligotrophic seas. Sponges are heterotrophic filter feeders, thriving on small (≤50 μm)organic particles (Reiswig,1971a). Organic matter in coral reefs can be roughly divided into three main types (Ribes et al.,2003). The first is dissolved organic matter (DOM). Some sponges that are densely packed with symbiotic bacteria (`bacteriosponges') are known to extract considerable amounts of DOM from the water (e.g. Reiswig, 1981; Yahel et al., 2003). According to Ribes et al. (Ribes et al.,1999), insignificant amounts of DOM are also expected to be absorbed by sponges not containing large amounts of bacterial symbionts(Hadas et al., 2006), although other cases have been recently demonstrated(De Goeij et al., 2008).

The second type is live particulate organic matter (LPOM), composed of various planktonic cells. Sponge removal efficiency (RE) of some LPOM is well documented and similar among most species studied: high RE of particles of about 1 μm in size (mainly prokaryotic cells), and decreased RE of larger particle sizes (e.g. Reiswig,1971a; Yahel et al.,2003).

The third type is non-living particulate organic matter (hereafter`detritus'), which is a structurally and chemically diverse fraction deriving from a range of different sources (Wilson et al., 2003). Detritus significance as a food source for marine sponges is poorly documented.

Four decades ago the nutritional balance of three Caribbean sponges was comprehensively studied (Reiswig,1971a; Reiswig,1971b; Reiswig,1974), comparing the amount of particulate organic carbon (POC)ingested by the sponges to their respiration rate and estimated growth rate. Reiswig concluded that one symbiont-free sponge [(Crypto) Tethya crypta] was able to meet its energetic needs by ingesting POC whereas another species (Mycale sp.) was just short of balancing its energetic needs (Reiswig,1974). A similar difference was found between Dysidea avara, suggested to ingest only particulate organic matter (POM)(Ribes et al., 1999), and Halichondria panicea, which requires additional sources(Thomassen and Riisgard,1995). Whether this inconsistency stems from a difficulty in encompassing all feed sources available for a sponge or reflects interspecies variance is unclear. In recent years there has been an accumulation of data that show that some sponge species can absorb significant amounts of dissolved organic carbon (Yahel et al.,2003; De Goeij et al.,2008). Moreover, organic nitrogen, a key nutritional element, has been overlooked and no information exists on its availability to sponges.

Two main methods have been employed to study sponge feeding. In the first method, water is sampled directly and simultaneously from the oscula (outflow)stream and close to the sponge surface just before entering its body. This method was applied in situ(Reiswig, 1971b; Yahel et al., 2005) and in the laboratory (Stuart and Klumpp,1984). The second method measures removal rate of food particles resulting from filtration activity in a closed system. This method was applied in a simple aquarium system while feeding the sponges with cultured non-photosynthetic bacteria (NPB) and algae(Thomassen and Riisgard,1995), underwater in chambers using natural plankton(Ribes et al., 1999), and in funnels (Witte et al., 1997; Ribes et al., 2003).

The Red Sea sponge Negombata magnifica (Keller 1889) was used to study the nutrition of a filter feeder in an oligotrophic (coral reef)environment. This branching sponge lives (at 3–60 m depth) either exposed or inside caves (Ilan,1995). It seldom exhibits photosynthetic activity (E.H.,unpublished data), and is virtually microsymbiont-free (bacteria were seldom observed in histological sections) (Gillor et al., 2000).

The northern Red Sea has relatively low chlorophyll-a (chl-a) levels(0.1–0.5 μg l^–1), mostly contributed by ultraplankton (cell size <8 μm) organisms(Yahel et al., 1998). This ultraplankton fraction shows a large temporal variation in abundance and composition, due to seasonal water column mixing and stratification(Lindell and Post, 1995). The present study assesses the contribution of natural POM to N. magnifica's diet, by determining the RE of the various food particles at different concentrations, measuring the amounts of POC and particulate organic nitrogen (PON) available to the sponge, determining the filtration rate, and estimating the flux of organic carbon and nitrogen to the sponge. The information obtained on sponge nutrition contributes to a greater understanding of the sponge's role in coral reef food webs and to developing artificial systems for sponge culture.

Thirty Negombata magnifica fragments were cut from 10 individual sponges growing in the northern-most area of the Gulf of Aqaba, Red Sea. These fragments had been attached to PVC plates one year prior to the laboratory experiment, as described by Hadas et al.(Hadas et al., 2005). This preparation enabled experimenting with intact sponges on inert substratum. Prior to experiments, all sponges were taken out of the water for a few seconds, placed on an electronic scale for wet mass (WM) determination (mass of the plates was recorded preceding attachment), and ranged between 50 g and 100 g. All were kept in the sea at a depth of 10 m, seaward of the Interuniversity Institute (IUI) in Eilat, Israel. Sponges were transferred(attached to plates) to the laboratory for acclimation at least 24 h prior to each experiment. The sponges were kept in the experimental aquarium with a steady flow of 1.5 l min^–1. All experiments were conducted at the IUI facilities during winter and spring. Seawater (22.5±0.5°C),pumped from 3 m deep, was continuously supplied to the laboratory-maintained sponges.

The sponge RE of the different POM fractions was measured by comparing particle concentrations in water samples collected simultaneously from the sponge inflowing and outflowing water(Reiswig, 1971b; Yahel et al., 2003; Yahel et al., 2005). Unlike the large uni-oscular sponges used in the cited studies, N. magnificahas many small (diameter 1–6 mm) oscula (E.H., unpublished data); thus,the conventional methods required adaptation.

To evaluate the RE of the different types of particulate matter by N. magnifica, all planktonic cells smaller than 50 μm were counted. Total particulate organic carbon and nitrogen content of the water flowing in and out of the sponge were also measured. To test sponge RE of different particle concentrations and types, measurements were conducted during winter and spring, when large fluctuations in plankton concentration prevail at the study site (Lindell and Post, 1995). Samples were collected when the water column was partially mixed (December),when spring phytoplankton bloom occurred (March and April), and during water stratification (June).

No single method measures all POM in the sea. The terminology used here for the different planktonic groups is derived from the measurement methods applied and not from a standard terminology. All planktonic organisms smaller than 8 μm were termed ultraplankton, after Lindell and Post(Lindell and Post, 1995). This group consists of three cell types: the cyanobacterium Synechococcussp., the eukaryotic ultraphytoplankton (EUP), and the remaining cells, here termed NPB. All cells >8 μm, whether photosynthetic or not, were termed large plankton. Non-living particles were termed detritus.

Ultraplankton cells (<8 μm) were identified and counted using a flow cytometer (FACScan, Becton Dickinson, Franklin Lakes, NJ, USA) equipped with a 488 nm Argon ion laser, applying procedures described by Marie et al.(Marie et al., 2000). Briefly,1.5 ml seawater samples were fixed with glutaraldehyde (G5882, Sigma Chemical Co., St Louis, MO, USA) at a final concentration of 0.1%, frozen in liquid nitrogen and stored at –80°C until analysis. For the photosynthetic cell counts, 450 μl of sampled water were analyzed. Synechococcussp. was identified based on its phycoerythrin orange fluorescence, and EUP was identified by the red fluorescence of its chlorophyll. To count the NPB, all ultraplanktonic cells were marked with the DNA stain SYBR™ Green-I(Molecular Probes, Inc., Eugene, OR, USA, ref: S-7563) and 30 μl of them were then analyzed in the flow cytometer. The NPB number was calculated by subtracting the number of photosynthetic cells from the total count obtained by SYBR Green measurement. As a standard for all measurements 0.95 μm yellow–green fluorescent beads (Polysciences, Warrington, PA, USA, ref:71825) were used. FACS list mode files were processed with the custom-designed software CYTOWIN (V. 4.1, D. Vaulot, CNRS, Roscoff, France).

Mean cell size of inflowing and outflowing EUP was compared using the flow cytometer output data. This comparison relied on the flow-cytometry signature of chlorophyll red fluorescence intensity, assuming that higher intensity is correlated to higher cell volume (Simon et al., 1994; Green et al.,2003). Although cell volume and chlorophyll content are not easily correlated because of differences caused by light adaptations, the present study compared paired samples that included the entire cell population entering the sponge and the expelled portion of this population. Thus, under these experimental conditions red fluorescence was considered a more direct indicator of cell size than the more commonly used light side scattering(Cunningham and Buonnacorsi,1992).

Water samples of ≤1 liter were filtered through a 5 μm polycarbonate filter (Poretics, Gorinchem, The Netherlands), using only gravitation to minimize cell destruction. Samples were stained with SYBR Green to detect non-autofluorescent cells. Following sample preparation, cells were counted using an epifluorescence microscope at a magnification of X1000 (Labophot-2,Nikon, Tokyo, Japan, equipped with a high pressure mercury lamp, model HB-10101AF). Excitation ranged between 450 nm and 490 nm, with a barrier to emission of 520 nm.

The total amount of particulate organic carbon and nitrogen in the sponge inflowing and outflowing water were directly measured with a C:N analyzer.

Water samples were collected, as described previously, from eight different sponge individuals during two weeks. Samples were filtered on pre-combusted(four hours at 550°C) and pre-weighed GF/F (Whatman, Maidstone, Kent, UK)filters until the filter clogged, usually after about 4 liters. The inflow water was pre-filtered through a 50 μm mesh size filter to eliminate the large particles that cannot enter the sponge through its ostia (incurrent openings). After filtration, the filters were dried for 24 h at 50°C and then stored at –20°C until analysis.

Prior to analysis, the filters were acidified for 8 h with hydrochloric acid fumes to eliminate any traces of the inorganic carbon prevailing in the vicinities of coral reefs. Following acidification, the samples were dried for 24 h, then weighed to determine the total organic matter on each filter. The filters were cut into halves and, after weighing one half, the filter was inserted into a tin capsule and loaded into an autoanalyzer sampler. Glycine was used as a standard and measured after every fourth sample. The amount of carbon and nitrogen was determined using a glycine standard curve.

The GF/F filter pore size is 0.8 μm. Thus, some NPB are not retained on the filter (Altabet, 1990). To estimate the amount of live POC not retained, the number of NPB cells present in the filtered water was counted using flow cytometry.

Because RE of Synechococcus sp. by N. magnifica was nearly 100% (present study), based on Thomassen and Riisgard(Thomassen and Riisgard,1995), it was determined that filtration rate equals the clearance rate of this microorganism. The experimental system design was based on the principles for a steady-state system, reviewed by Riisgard(Riisgard, 2001).

This measurement was assumed to be valid only if the RE of Synechococcus sp. was not affected by reduced cell concentrations in the steady-state system when cell concentration was lower than the ambient concentrations by about 20% (present study). To test this assumption, the inflowing cell concentration was artificially reduced and the RE of Synechococcus was measured. The amount of cells in the water was controlled by mixing different ratios of fresh seawater with 0.2 μm filtered seawater (Suporlife 200, PALL, Biopharmaceuticals, East Hills, NY,USA).

Sponge nitrogen excretion was estimated by measuring the amount of total ammonia nitrogen (TAN) it excreted in a steady-state system(Riisgard, 2001). Because the water TAN concentration was expected to be low, the highly sensitive fluorometric method described by Holmes et al.(Holmes et al., 1999) and adapted by David (David, 2002)was used.

Nine N. magnifica individuals attached to PVC plates were each mounted in a 1 liter metabolic chamber, constantly supplied with fresh seawater (23±1.0°C). Water flow rate was determined by measuring the chamber outflowing water volume over time. After a short acclimation(about 60 min), when most of the sponge oscula were open, the water entering and exiting the metabolic chamber was sampled simultaneously.

Water samples were each divided into four test tubes each containing 4 ml. A measured amount of 0.1, 0.2 or 0.3 ml of ammonia standard (50 μmol l^–1, Sigma Chemical Co.) was added to three of the test tubes; thus, used as an internal standard. Aliquots of each sample were excited at 365 nm and the emission at 492 nm was measured. The TAN concentration was determined by extrapolating the resulting regression line to the y intercept [detailed description in David(David, 2002)]. Because of this test's sensitivity to TAN contaminations, if the regression's R² was less than 0.99, the entire measurement was discarded.

The carbon and nitrogen contents of bacterioplankton were calculated separately for GF/F retained and non-retained NPB (with estimated cell size of 0.056 μm³ and 0.04 μm³, respectively)(Lee and Fuhrman, 1987). The ratio between carbon and nitrogen amount (mass) to cell volume(μm³) is 0.38 and 0.11, respectively, and the C:N ratio for NPB is about 3.7. Hence, the carbon and nitrogen amount for the retained NPB is 20 fg C cell^–1 and 5.4 fg N cell^–1. For the non-retained NPB, estimations are 15 fg C cell^–1 and 4 fg N cell^–1. The carbon and nitrogen amounts of Synechococcus sp. are 200 fg C cell^–1 and 50 fg N cell^–1 (Caron et al.,1995). From the estimates of these elements' content in phytoplankton cells we used Montagnes et al.(Montagnes et al., 1994)conversion factors, as best covering our measurement techniques. Amounts of organic carbon and nitrogen for large planktonic cells were calculated for each cell size: C (pg cell^–1)=0.109×(μm³)^0.991 and N(pg cell^–1)=0.0172×(μm³)^1.023(Montagnes et al., 1994).

Statistical analyzes were performed using JMP IN (release: 5.0.1a, SAS Institute Inc., Cary, NC, USA). Statistical analysis of the RE (expressed as percentages) used an arcsin transformation of the square root of removal ratios. A probability smaller than 0.05 was considered significant unless stated otherwise. To test if the RE of the three ultraplankton cell types were mutually independent, pair-wise Pearson product-moment correlation was used. To determine differences in RE between the three ultraplankton cell types and the four sampling periods, two-way analysis of variance (ANOVA) model was used. Tukey Kramer test was used for multi-comparison analyzes.

The sponges removed all naturally occurring ultraplankton types from the water but at different RE (Fig. 1). Synechococcus sp. was removed at mean efficiency of 99.2±0.2% (Fig. 1A) and mean RE of the NPB cells (Fig. 1B) and EUP (Fig. 1C) were 85.7±1.3% and 84.9±2.0%, respectively.

The RE of the three cell types varied independently from each other(pair-wise correlations P>0.05). Analysis of differences in RE between the three planktonic cell types revealed that Synechococcussp. had significantly (P<0.001) higher RE than the other two cell types, and this remained constant through all cell concentrations(Table 1).

The mean NPB cell RE was not significantly different from that of the EUP cells. However, whereas no linear regression was found between NPB RE and ambient cell concentration (N=27, R²=0.1 P=0.09), such a relationship was found for the EUP cells(N=28, R²=0.84, P<0.001, y=1.3–1.4×10^–5x).

Mean red fluorescence intensity of EUP increased significantly(7.9±1.6 fluorescence arbitrary units) in the outflowing EUP cells relative to the inflowing cells (paired t-test; N=25, t=4.8, P<0.001).

Almost all of the large eukaryotic cells were photosynthetic (mean volume 341 μm³ cell^–1). Different planktonic types dominated different seasons: small dinoflagellates in December; diatoms in March (particularly the genus Pseudonitzchia); and small flagellates in April.

Negombata magnifica removed large eukaryotic cells at a mean efficiency of 37.0±0.87% (Fig. 2), with great variability: from about 70% (December) to zero removal (in April for some of the examined specimens). Mean plankton cell concentration increased from December to March and decreased in April(99.1±8.2 cell ml^–1; 572.9±127.5 cell ml^–1; and 257.2±35.6 cell ml^–1,respectively). A comparison of RE between these sampling periods revealed a significant difference (one-way ANOVA, F_2,17=10.3, P<0.01). A multi-comparison test showed a significant difference in cell removal rates in April (19.5±2.5%) relative to those in December and March (54.6±3.9% and 36.9±2.5%, respectively), with no significant difference between the latter two sampling periods(Table 1). Although an inverse relationship exists between RE and cell concentration, calculation of algae quantities indicates that the sponges cleared from the water an average of 108 cells ml^–1 in March, 92 cells ml^–1 in April,and only 54 cells ml^–1 in December.

Cell size distribution analysis showed that 60% of the algae were smaller than 20 μm³. The selectivity index of all seven size categories was not significantly different from zero (t-test, comparison to zero as a fixed value).

In testing the GF/F RE of ultraplankton no EUP and Synechococcussp. cells were recovered in the filtered water. Conversely, 37.1±1.2%of the ambient NPB was found in the filtered water; thus, this ratio of NPB was reduced from the bacteria number that was found in the FACS analysis in order to avoid an over estimation of the contribution of this nutrient source when comparing it with the total nutrient amounts that were found in the C:N analyzer. The nutrient amounts contributed by the unretained bacteria were added to the total nutrient budget.

Total C:N analysis demonstrated that both POC and PON were significantly removed (paired t-test, N=8, t=4.9, P<0.05) by all eight sponges(Fig. 3). Mean C and N concentrations were 48.46±5.69 μg C l^–1 and 6.45±0.7 μg N l^–1 in inflow water and 18.84±1.74 μg C l^–1 and 2.04±0.19 μg N l^–1 in outflow water. A significant difference was found in the POM C:N ratio of the ambient water and the sponge outflow water(8.6±0.2 and 10.8±0.3, respectively; paired t-test, N=8, t=5.4, P<0.001).

The amount of live PON (LPON) in the inflow water was 5.4±0.24 μg N l^–1, comprising 84% of the total PON, while prokaryotic cells contributed 77% to the LPON pool(Fig. 4A). The amount of live POC (LPOC) in the inflow water was calculated at 22.6±1.1 μg C l^–1, making it 46.6% of the total amount of POC. Prokaryotic organisms contributed 69% of the LPOC (Fig. 4B).

The mean FR was 10.8±1 ml min^–1g^–1 sponge (WM; N=7). Cell concentration did not differ between the control tank and ambient water, N. magnificacompletely removed Synechococcus in all tested cell concentrations. A significant linear regression model was found between cell concentration and the amount of Synechococcus sp. removed(y=0.997x+1.1, R²=0.99, P<0.05, N=12).

TAN released by N. magnifica to the water was 2.7±0.8 nmol TAN min^–1 g^–1 (WM), equivalent to 54.4±16.1 μg N day^–1 g^–1 (WM, Fig. 5).

The major feeding strength of sponges stems from their nearly non-selective filtration of a wide variety of nutritional sources. Measuring, identifying and quantifying these different types of organic particles are complex,requiring many different measurement methods. Additionally, these food sources are unevenly distributed in time and space, and their nutritional value varies.

Nutritional value analysis showed that organic carbon and nitrogen were unevenly distributed among the different food particles. Within living particles, carbon and nitrogen were directly calculated (from the conversion factors), while the potential contribution of detritus to sponge feeding was calculated by subtracting nutrient amounts in the live particles from total POM. We estimated that 54% of the total available POC was from detritus,almost equal to the LPOM fraction. Among LPOM, eukaryotes and prokaryotes had similar organic carbon contents. By contrast, 84% of the calculated organic nitrogen came from LPOM, with prokaryotes being the major source of PON for the sponge diet, contributing 65% of the total organic nitrogen.

As the POM pool is not totally digested by sponges, using only the plankton-to-nutrients conversion factors would lead to overestimating the actual quantity of nutrients available for sponge nutrition. Food digestibility has not been defined for sponges. We therefore used the digestibility definition as applied for fish [nitrogen digestibility = gross nitrogen intake – gross nitrogen excretion(Cho and Kaushik, 1990)]. The problem here is that some ingested particles are not digested; for example,some apparently intact algae were observed leaving the sponge. To overcome this conceptual problem, outflowing particles were considered as undigested by the sponge. Because each sample was the integration of ca. 12 h of sponge activity, it is suggested that the reduction in particle concentration between inflowing and outflowing water represents food particle digestibility and not any temporary status.

Because most PON (84%) was contributed by LPON, by assuming that the sponge digested the remaining 16% of PON from an unresolved source (not LPON), we could calculate the minimal nitrogen digestibility as 79% {[1–(total PON_outflow–LPON_outflow)×(total PON_inflow–LPON_outflow)^–1]×100}. A comparable value (78%) was found for the bivalve Mactra veneriformis (Hiwatari et al.,2002). Although food digestibility calculated here is within the accepted range, it is interesting that this sponge maintains a relatively high digestibility of such diverse food types. Based on nitrogen digestibility and changes in C:N ratios, we estimate carbon digestibility as at least 73.5%[1–(1–nitrogen assimilation efficiency)×10.8×8.6^–1].

Negombata magnifica removed all unicellular planktonic organisms from the water. RE of the different cell types were not correlated to each other, unlike Witte et al.'s (Witte et al., 1997) contention for polar sponges. Prokaryotic cells in all Red Sea naturally occurring concentrations were removed at an efficiency ranging from approximately 100% for Synechococcus sp. to about 85%for the NPB, similar to another Red Sea sponge, Theonella swinhoei(Yahel et al., 2003). The difference in RE between Synechococcus sp. and the NPB can be attributed to different population structures; whereas the former is relatively homogeneous, the latter includes small bacteria down to a diameter of 0.4 μm (Lee and Fuhrman,1987), possibly retained at a lower efficiency than the larger ones, similar to the RE of 0.5 μm beads by Crambe crambe(Turon et al., 1997). Because Prochlorococcus sp. cells are almost absent from the water during winter (Lindell and Post,1995), this organism was not expected to appear in most measurements. Had the latter cell type been present in the samples, it would have been included in the NPB fraction.

The RE of EUP was concentration dependent, ranging from 65% to 90%. Analysis of the mean fluorescence intensity of the sponge inflowing and outflowing EUP, indicated that the smaller cells were retained more efficiently than the larger ones. This can explain the fluctuations found in the RE of EUP cells. The size of these cells might have changed(Arin et al., 2002) due to typical temporal variations in nutrient levels in the Gulf of Aqaba(Lindell and Post, 1995).

The ability to efficiently capture prokaryotes, the reduced ability to capture bigger EUP and larger plankton cells, and the non-selectivity toward any of the large planktonic groups, suggest that N. magnifica's filtration mechanism comprises two parallel working systems.

The first system, probably operated by the flagellated choanocyte cells,specializes in capturing the smaller cells and is responsible for removing Synechococcus sp., NPB cells and the smaller EUP. The maximal particle size this system captures (not determined in the present study), is probably set by the prosopyle (about 5 μm) size(Weissenfels, 1992). This system's RE approaches 100% for about 1 μm particles, at all natural cell concentrations, with decreased efficiency for the smallest particles.

In the second filtration system, the aquiferous system lining pinacocytes probably capture the larger food particles (large EUP and above), as its filtration efficiency is lower than that of the former. RE varies and is partially saturated at higher natural planktonic concentrations. The pinacocytes' ability to capture small particles as NPB is probably limited. This suggested that N. magnifica filtration mechanism is in accordance with data obtained for C. crambe(Turon et al., 1997).

Detritus can be an important nutrient source (at least of organic carbon)for sponge feeding, and we hypothesize that, like live cells, its removal is size dependent. However, detrital particles cannot be separated from the sponge-generated particles (faecal pellets) retained on a GF/F filter. Therefore, RE of the ingested detritus cannot be directly measured as described for other sponges (Wolfrath and Barthel, 1989). Using an indirect calculation [(total POM)_in–(total POM)_out–(live POM)_in+(live POM)_out], we estimated that detritus ingested by the sponge contributed at least 10.6±4.8 μg C l^–1 (approximately one-third of the ingested organic carbon). Detritus-derived organic nitrogen seems nearly negligible, because the intake was below detection level (t-test, t=0.05, d.f.=7, P=0.9).

Negombala magnifica filtration rate was within the higher range of other sponges, whether tropical (Reiswig,1974; Yahel et al.,2003) or temperate (Riisgard et al., 1993; Turon et al.,1997). The relatively high filtration rate might be adaptive in increasing nutrient flux into such sponges residing in oligotrophic environments.

The mean daily amounts of POC and PON absorbed by N. magnificawere 480 μg C day^–1 g^–1 (WM) sponge and 76.6 μg N day^–1 g^–1 (WM) sponge,including the additional GF/F filter non-retained NPB fraction. Assuming that 0.46 mg C corresponds to 1 ml of O₂, and a mean respiration rate of 37 nmol O₂ min^–1 g^–1 (WM) sponge(Hadas et al., 2008), the carbon amount corresponding to this respiration rate is 0.54 mg C day^–1 g^–1 (WM) sponge. In the present study the sponge absorbed 85% of the carbon amount required for respiration,although not all of it is digestible by the sponge, and more organic carbon is needed for the production of biomass. The amount of nitrogen excreted as TAN was found to be 54.4±16.1 μg N day^–1g^–1 (WM) and more nitrogen might be excreted as nitrite and nitrate. Moreover, the nitrogen allowance should be higher because more nitrogen is deposited for sponge biomass during growth. It was previously found that N. magnifica can grow by up to 0.55% of its body mass per day, and protein constitutes 7.5% of its biomass(Hadas et al., 2005). These data indicate that the daily amount of nitrogen deposited in sponge biomass is 60 μg N day^–1 g^–1 (WM). Thus, the daily nitrogen requirement seems to be twice the absorbed amount [76.6 μg N day^–1 g^–1 (WM) sponge].

The discrepancy in both carbon and nitrogen budgets could be explained by sponge ingestion of nutrients undetermined in the present study, such as the colloids in the size range of between 0.8 μm and 0.2 μm that were not retained on the GF/F filters and could not be quantified (like the bacteria). Moreover, it is hypothesized that N. magnifica removes smaller organic matter from the water at a decreasing efficiency(Fig. 6). This overlooked nutrient source, mostly DOM (e.g. Hadas et al., 2006), can easily supply all the sponge's nutritional requirements, in potentially containing significant nutrient quantities(Yahel et al., 2003).

The nutrient budget suggested in the present study might have some potential sources of error, mainly as a result of the inability to maintain N. magnifica in a fully controlled environment for an extended period of time. Consequently, the data were integrated from several distinct experiments that differed in time and in sponge individuals. The dependence on a natural food source (instead of feeding the sponge with known feed particles) is probably the most significant source of error, because the quality and amount might change over time. In the future, when controlled systems for sponge culture will become available, the nutrient budget will be refined.

The present study highlights the importance of detritus as a carbon source in sponge feeding, and the significance of prokaryotic organisms as the dominant particulate organic nitrogen source in the coral reef. The apparent deficiency in the nutrient budget suggests that N. magnifica feeds on a broader than previously estimated spectrum of particles, and must absorb significant amounts of DOM in order to meet its nutritional requirements,despite the fact that this sponge contains very low amounts of microsymbionts. This hypothesis might also be valid for other coral reef heterotrophic suspension-feeders such as bivalves, tunicates and soft corals.

ABBREVIATIONS

 
• A

  mean of all field counts

 
• B

  number of fields on a filter

 
• CL

  clearance rate

 
• C_out,C_in

  TAN concentration in the outflow and inflow water,respectively

 
• C₁,C₂

  cell concentrations in inflow and outflow water, respectively

 
• chl-a

  chlorophyll-a

 
• DOM

  dissolved organic matter

 
• E

  index for selectivity

 
• EUP

  eukaryotic ultraphytoplankton

 
• FL

  flow rate

 
• FR

  filtration rate

 
• IUI

  Interuniversity Institute (in Eilat, Israel)

 
• LPOC

  live particulate organic carbon

 
• LPOM

  live particulate organic matter

 
• LOPN

  live particulate organic nitrogen

 
• NPB

  non-photosynthetic bacteria

 
• POC

  particulate organic carbon

 
• POM

  particulate organic matter

 
• PON

  particulate organic nitrogen

 
• RE

  removal efficiency

 
• r_i,n_i

  cell abundances in the i bin in the inflow and outflow,respectively

 
• TAN

  total ammonia nitrogen

 
• V

  volume of water filtered

 
• W

  sponge wet mass

 
• WM

  wet mass