Sea cucumbers reduce chromophoric dissolved organic matter in aquaculture tanks

Time-series in the big-volume tanks

We monitored during more than one year two aquaculture tanks located at the iMareNatural S.L. facilities (http://www.imarenatural.com) in Motril, Spain (36°44′38″N, 3°35′59″W). Each tank of 50,000 liters of capacity was connected directly with the coastal water by one inlet pipe (inlet water) and the water from each tank was released by one outlet pipe located in the bottom of the tank (effluent). The seawater was pumped into the tanks at a continuous flow of 1,200 l h⁻¹. Therefore, water residence time in the tanks was ca. 42 h and the total annual effluents accounted for 10,512 m³. In one of the tanks, 811 ± 125 individuals of the primary species, the sea anemone A. sulcata, and 93 ± 3 adults of sea cucumbers Holothuria tubulosa (≈80%) and Holothuria forskali (≈20%) were included (hereafter designated as + holothurian tank). In the other tank only 690 ± 87 individuals of the primary specie were included (hereafter designated as—holothurian tank). Sea anemones were placed on floating plastic boxes in the surface of the tanks and holothurians were free in the bottom and walls of the tanks. Sea anemones were fed with about 900–1,800 g of fresh chopped fish, mainly Scomber scombrus (Chintiroglou & Koukouras, 1992) twice per week.

Water samples for different chemical and biological analysis were taken biweekly from July 17th 2013 to August 20th 2014. Each sampling day, we took water samples from the inlet pipe, the center of the two tanks using a telescopic stick with a plastic beaker located in its extreme and from their corresponding effluents. To avoid that light can affect absorption measurements, we immediately took the CDOM samples in pre-combusted (4 h at 500 °C), acid-cleaned, amber glass bottles. They were kept in ice during transportation to the laboratory (about one hour from the tanks). Water samples were filtered through pre-combusted Whatman GF/F glass fiber filters of 0.7 µm nominal pore size and the <0.7 µm fraction was used for the optical characterization of chromophoric dissolved organic matter.

Chromophoric Dissolved Organic Matter (CDOM)

Absorption spectra of dissolved organic matter provide information on CDOM concentration and other qualitative properties. Absorption coefficients at specific wavelengths (e.g., 325 nm and 443 nm) are used as proxies of CDOM concentration, and the spectral slopes and spectral ratios, which are largely independent of the concentration, are surrogates of CDOM origin, molecular weight and chemical structure (Weishaar et al., 2003; Twardowski et al., 2004; Helms et al., 2008; Nelson & Siegel, 2013; Martínez-Pérez et al., 2017).

CDOM absorbance spectra were recorded at wavelengths from 200 nm to 750 nm at 1-nm interval using an UV/VIS Perkin Elmer spectrometer with a 10 cm-quartz cuvette. The spectrophotometer was connected to a computer with Lambda 25 software. The detection limit of the spectrophotometer (0.001Absorbance) corresponds to a CDOM absorption coefficient detection limit of 0.02 m⁻¹. Spectrum corrections due to residual scattering by fine size particle fractions, micro-air bubbles, or colloidal material present in the sample were performed by subtracting the average of the absorption between 600 and 700 nm (Green & Blough, 1994).

CDOM absorption coefficients, a_λ, were calculated using the next equation: (1)${\displaystyle a_{\lambda }=2.303\frac{\mathrm{Absorbance}\left(\lambda \right)-\mathrm{Absorbance}\left(600-700\right)}{l}}$ where a_λ is the absorption coefficients in m⁻¹ at each λ wavelength, Absorbance (λ) is the absorbance at wavelength λ, Absorbance (600–700) is the average absorbance from 600 to 700 nm, 2.303 is the factor that converts decadic to natural logarithms, l is the cuvette path length in m⁻¹.

Spectral slopes describe the shape decay of absorption coefficients vs. wavelengths. Slopes were calculated from the linear regression of log-transformed absorption coefficients in the wavelength bands 275–295 nm (S_275−295) and 350–400 nm (S_350−400) (Helms et al., 2008). The spectral slopes for both wavelength ranges were calculated as in Eq. (2). (2)${\displaystyle a_{\lambda }=a_{\lambda \mathrm{ref}}{e}^{-S\left(\lambda -{\lambda }_{\mathrm{ref}}\right)}}$ Where λ is the selected wavelength in nm, a_λ is the absorption coefficient at λ wavelength in m⁻¹, a_λref is the absorption coefficient at a reference wavelength λ_ref, and S is the spectral slope. The spectral slope ratio (S_R) was calculated as the ratio of the spectral slope from 275 nm to 295 nm (S_275−295) to the spectral slope from 350 nm to 400 nm (Helms et al., 2008).

CDOM absorption usually decreases exponentially as wavelengths increase. Therefore, the shorter the wavelength, the more sensitive to changes is (Fig. 1). Despite a low sensitivity (Fig. 1), the visible wavelength at 443 nm is used in remote sensing studies due to its correspondence with the satellite sensor (Siegel et al., 2005; Ortega-Retuerta et al., 2010). Therefore, the spectral slope from 275 nm to 295 nm is the most sensitive optical parameter of CDOM changes (Helms et al., 2008), but this parameter is not quantitative. CDOM quantity was measured as absorption coefficient at 325 nm (a₃₂₅), since this wavelength is the most common in the literature (Nelson & Siegel, 2013; Catalá et al., 2015) and has a higher sensitivity than 443 nm (Fig. 1).

Ancillary data

Basic parameters as temperature (°C), pH, salinity (psu), total dissolved solids (TDS), and conductivity (mScm-1) were measured in the tanks using a multi-parameter HANNA probe (HI9828 model; Woonsocket, RI, USA). Total organic carbon (TOC) concentration was measured as non-purgeable organic carbon after a high-temperature catalytic oxidation using a Shimadzu TOC-V CSN. Samples, by triplicate, were acidified with hydrochloric acid and purged for 20 min to eliminate the remaining dissolved inorganic carbon. Three to five injections were analyzed for each sample. Standardization of the instrument was done with potassium hydrogen phthalate. Particulate organic matter (POM) was obtained filtering between 1.5 and 2.0 l of water through pre-weighed and pre-combusted (500 °C for 4 h) Whatman GF/F glass fiber filters. The filters containing all the solids were dried at 60 °C for >24 h and reweighed to determine the total mass (mineral + organic matter). Then, the organic fraction was burned by combusting the filters at 500 °C for 6 h; finally, the filters were reweighed again to determine the mineral residue. POM was obtained after the subtraction of the mineral residue to the total mass. The concentration of chlorophyll-a was determined spectrophotometrically after pigment extraction with methanol (APHA, 1992). Bacteria abundance was determined in triplicate using flow cytometry (Gasol & Del Giorgio, 2000) with a FACScalibur Becton Dickinson cytometer equipped with a laser emitting at 488 nm. Data were processed using Cell quest software.

Short-term experiments

To test the statistical significance of the results obtained in the time-series, we performed three short-term (3 days) experiments. Each experiment was carried out in seven tanks of 300 liters that contained a floating plastic box with 80 individuals of A. sulcata per tank and consisted of two treatments: + holothurians (+H) and –holothurians (−H). At the initial time, in three of the tanks we included 10 individuals of H. tubulosa in each tank. These three tanks are the replicates of the +holothurians treatment. The other four tanks only contained the 80 individuals of A. sulcata and represent the replicates of the –holothurians treatment. The experiment 1 was carried out from 6th to 9th October 2017, the experiment 2 from 27th to 30th October 2017, and the experiment 3 from 3rd to 6th November 2017. During the duration of each experiment the anemones were not fed to control the net effect of holothurian activity. At the initial and final time we took samples for the optical characterization of dissolved organic matter. To analyze the samples we followed the same procedures used in the time-series.

Statistical analysis

To compare the time-series of the CDOM optical parameters in the inlet water with the effluents from the tank with holothurians and the tank without holothurians we performed paired t-test (for normally distributed variables) and Wilcoxon matched pairs test (for not-normally distributed variables) using the Statistica software (V8). These statistical analyses ameliorate the problem of temporal pseudoreplication in this type of studies (Millar & Anderson, 2004). Correlations between CDOM optical parameters and potential controlling factors were performed using Statistica software (V8). In the short-term experiments to test the statistical significance of the presence of holothurians on the CDOM optical parameters we performed analysis of variance (ANOVA) comparing the tanks with holothurians (+H) with the tanks without holothurians (−H) using Statistica software (V8).

Time-series of the optical parameters in the big-volume tanks

In the inlet waters, the a₃₂₅ values ranged from 0.06 to 0.83 m⁻¹ (Table S1) and in the effluents of +holothurian and –holothurian tanks from 0.06 to 0.79 m⁻¹ and from 0.37 to 1.27 m⁻¹, respectively (Tables S2 and S3). The absorption coefficients of the inlet waters (i.e., coastal waters of Western Mediterranean Sea) were similar to those ones found in other coastal waters (Catalá et al., 2013; Nima et al., 2016) or in the open Mediterranean Sea (Bracchini et al., 2010; Organelli et al., 2014). Systematically, throughout the time-series, the effluent of the –holothurian tank showed higher a₃₂₅values (grey triangles in Fig. 2A) than the effluent of the + holothurian tank (red squares in Fig. 2A). These last values were similar to the a₃₂₅ values in the inlet waters (white circles).

Spectral slopes and spectral slope ratios are qualitative parameters, which are independent of the CDOM concentration. The higher the spectral slope, the smaller the DOM molecular weight is (Helms et al., 2008). The slope in the band from 275 nm to 295 nm (S_275−295) is an optical parameter particularly sensitive to environmental changes as solar radiation or salinity (Helms et al., 2008; Helms et al., 2013; Catalá et al., 2013). In the inlet water, the values of S_275−295 ranged from 10 to 38 µm⁻¹ (Table S1) and in the effluents of +holothurian and –holothurian tanks from 6 to 28 µm⁻¹ and from 13 to 40 µm⁻¹, respectively (Tables S2 and S3). In the inlet water, the values were similar to those reported for coastal and estuary waters, usually characterized with lower slopes (∼15–25 µ m⁻¹) than the values for the open ocean (∼25–50 µ m⁻¹) (Helms et al., 2008; Helms et al., 2013; Catalá et al., 2015). Like the a₃₂₅ values, the S_275−295 values in the effluent waters of the –holothurian tank (grey triangles in Fig. 2B) showed consistently higher values than the inlet water (white circles in Fig. 2B) and in the effluents of the +holothurian tank (red squares in Fig. 2B). In the inlet waters, the spectral slope ratios (S_R) ranged from 0.6 to 2.6 (Table S1) and in the effluents of +holothurian and –holothurian tanks from 0.5 to 3.1 and from 0.4 to 3.9, respectively (Tables S2 and S3). The S_R values in –holothurian effluents (grey triangles in Fig. 2C) showed consistently higher values than the inlet water (white circles in Fig. 2C) and +holothurian effluent (red squares in Fig. 2C).

To assess if the presence of holothurians can modify significantly CDOM we performed paired t-test or Wilcoxon matched pair tests pooling all the time-series data (Table 1). In the Fig. 3 we show all the time-series data pooled in median values, 25–75% percentiles and non-outliers values. The a₃₂₅ values in the –holothurian tank and effluent (Fig. 3A, grey boxes) were significantly higher that the values in the +holothurian tank and effluent and the inlet water (Table 1). A similar effect was found for the spectral slope (S₂₇₅₋₂₉) (Fig. 3B, grey boxes), and the spectral ratios (S_R) (Fig. 3C, grey boxes). Indeed, we observed higher CDOM concentration with higher spectral slopes (surrogate of smaller molecular size; Helms et al., 2008) in the tank without holothurians than in the tank with holothurians. Therefore, holothurians appear to reduce significantly CDOM concentration, particularly of compounds with comparatively lower molecular weight making the spectral slopes smaller. On the other hand, the differences between the inlet water and the +holothurian tank and effluent water, although significant, were less relevant (Fig. 3, Table 1).

These results suggest that effluent of the monoculture of A. sulcata increases CDOM in comparison with the inlet water, which could affect the recipient coastal waters. These higher CDOM values in the –holothurian tank in comparison with the values in +holothurian tank could be related to: (1) a higher abundance of bacteria and their metabolic by-products that produce CDOM or (2) a higher concentration of particulate organic matter (derived from uneaten food, detritus and microbial cells) which disaggregation in dissolved compounds also produce CDOM. In both cases, an increment in CDOM concentration is expected in absence of holothurians. Several studies have shown that bacteria and phytoplankton can produce CDOM as metabolic by-products (Nelson, Siegel & Michaels, 1998; Nelson, Carlson & Steinberg, 2004; Ortega-Retuerta et al., 2009; Romera-Castillo et al., 2010; Catalá et al., 2015; Catalá et al., 2016). However, we did not find significant relationships between the a₃₂₅ values and the concentration of chlorophyll-a (+holothurian r² = 0.009, p = 0.513; -holothurian r² = 0.002, p = 0.804) or the abundance of bacteria (+holothurian r² = 0.014, p = 0.403; -holothurian r² = 0.068, p = 0.095) (Fig. 4A). Therefore, phytoplankton and bacterial carbon processing appear to have a minor importance in these tanks. In contrast, we found significant and positive relationships between the concentration of particulate organic matter (POM) and the a₃₂₅ values in the +holothurian tank and effluent water (red squares; r² = 0.41, p = 0.002; regression line a₃₂₅ = 0.20 + 0.079 POM) and the a₃₂₅ values in the –holothurian tank and effluent water (grey triangles; r² = 0.20, p = 0.006; regression line a₃₂₅ = 0.42 + 0.102 POM) (Fig. 4B). Therefore, POM concentration in the tanks appears to be the main driver of CDOM changes. POM disaggregation into dissolved components is a common process in coastal waters (He et al., 2016), particularly under sunny conditions (Shank et al., 2011; Pisani, Yamashita & Jaffé, 2011). Holothurians consume several components of POM as phytoplankton cells, bacteria, uneaten food, animal feces, and transparent exopolymer particles (Hudson et al., 2005; Slater & Carton, 2009; Navarro et al., 2013; Yokoyama, 2013; Wotton, 2011). Since holothurians reduce POM concentration by consumption in these tanks (SM Sadeghi-Nassaj, GL Batanero, IP Mazuecos, C Alonso, I Reche, in preparation), we think that POM disaggregation into DOM in the tank with holothurians was significantly lower than in the tank without holothurians where a relevant fraction of POM might have been converted into CDOM. In addition, recently Brothers, Lee & Nestler (2015) have demonstrated a direct uptake of free amino acids in several tissues as the respiratory trees, epidermis, and oral tentacles of a sea cucumber species (Parastichopus californicus) during the visceral regeneration. It is well known that amino acids such as the tyrosine and tryptophan are able to absorb light in the ultraviolet band (e.g., Catalá et al., 2013). Therefore, a direct and selective assimilation of free amino acids by holothurians could also explain a reduction in the a₃₂₅ values and in the spectral slopes.

Short-term experiments

To corroborate the results obtained in the time-series, we performed three short-term (3 days) experiments in smaller tanks manipulating the presence of holothurians. At the initial time, we did not observe significant differences in the optical parameters between both treatments (+H and –H) indicating the experiments started with identical conditions with the exception of the presence of holothurians (Table 2). However, at the final time after three days, the presence of holothurians (+H treatment) significantly reduced the values of the CDOM absorption coefficients at 325 nm (a₃₂₅) in two out of three experiments (Fig. 5A red bars, Table 2) and in one out of three experiments at 443 nm (a₄₄₃) (Fig. 5B red bars, Table 2). In the three experiments, at the final time, the presence of holothurians (+H treatment) reduced significantly the spectral slopes (S_275−295) (Fig. 5C red bars, Table 2). This variability in the statistical significance for the different optical parameters is related to the inherent higher sensitivity of CDOM as wavelengths are shorter (see Fig. 1).