Impact of sampling techniques on the concentration of ammonia and sulfide in pore water of marine sediments

Method I – rhizon CSS samplers inserted directly into sediment cores through 4-mm holes in capped Plexiglas liners, at 5-cm intervals, and connected to 3-ml syringes. The pore water extraction lasted for about 20 min. We collected about 2.5–2.8 ml of pore water. The following sediment depths were sampled: 5, 10, 15, 20, 25, 30 cm bsf (below sea floor; core No. 1, 2, 3) and additionally 35 cm bsf (core No. 1 and 2). The sample of 5 cm bsf from core No. 1 for ammonia determination was rejected due to contamination. The resolution applied in this method was rather low compared to typical studies of short cores from the coastal marine environment (usually 0.5–2 cm). However, our objective was to obtain results in order to compare the extraction methods, while the exact layer-to-layer variations in parameters were a secondary issue. Rhizon samplers have been widely used for pore fluid sampling from soils for over 20 years (Knight et al. 1998; Sigfusson et al. 2006). They basically consist of a hydrophilic porous polymer tube (pore diameter of 0.15 μm) and PVC tubing. They can be connected either to a syringe or to a vacuum test tube; pore water is obtained from a sediment core by vacuum-driven suction (detailed characteristics of rhizon samplers can be found in e.g. Seeberg-Elverfeldt et al. 2005). This method eliminates contact with atmospheric air, ensures very little core disturbance during sampling and gives filtrated samples that are almost free of bacteria. The range of sediment dewatering for a pore water volume of ~3 ml is max 0.5 cm; cylindrical area of r = 0.5 cm around a rhizon, close to the sediment edge (Seeberg-Elverfeldt et al. 2005). Therefore, sampling at a very high resolution, e.g. 0.5 cm, is impossible using rhizons. The sorption should be reduced as the samplers are made of inert polymer, which is expected to have no ion-exchange properties.

Method II – centrifugation of approx. 9 ml of a sediment subsample (20 min, 3500 rpm rot/min, in 10 ml tubes) directly from capped Plexiglas core liners, collected within 30 min after the core collection, using syringes with the luer tip removed (diameter of approx. 10 mm) through predrilled 11-mm holes, at 5-cm intervals. The range of the sediment layer from which each subsample was collected was 1.1 cm. Pore waters for sulfides and ammonia were sampled at 5 cm resolution: for sulfides – from sediment depths of 5, 10, 15, 20, 25 cm bsf (core No. 4); 7, 12, 17, 22, 27 cm bsf (core No. 5) and 4, 9, 14, 19 cm bsf (core No. 6); and for ammonia – at 5, 10, 15 cm bsf (cores No. 4); 7, 12, 17 cm bsf (core No. 5); and 4, 9, 14 cm bsf (core No. 6). Contact with atmospheric air was limited to a minimum when transferring a sample to centrifuge

tubes. Different depths of sediment sampling resulted from not equal length of the cores in liners, which had been pre-drilled every 5 cm. In the case of ammonia, sediment sampling from only three depths of each core in this method was due to accidental disturbance of deeper parts of the cores during sampling for sulfides and the subsequent decision to reject samples from these layers for further analysis. The concentrations of sulfides and ammonia in cores No. 5 and 6 were linearly interpolated to obtain values for 5, 10, 15 etc. cm bsf. This is not a typical sampling procedure. However, it can be used to extract pore waters for the analysis of oxygen-reactive compounds, especially from long cores that are usually not sectioned in a traditional way. In this study, it was used for comparison as it eliminates the effect of sample filtration (we used turbidity correction instead; the absorbance value of a sample without reagents was subtracted from the measured absorbance value for that sample with reagents; Rainwater & Thatcher 1960), core disturbance during sampling (sectioning), and additionally it ensures limited contact with atmospheric air.

Method III – centrifugation (20 min, 3500 rpm, in 40 ml tubes) of 5-cm sediment sections pushed out of core liners. Sediment sectioning was carried out within 1 h after core retrieval. Vertical sediment edges of each section were discarded. Concentrations of parameters measured in each 5-cm section were assigned to the middle depth of the section, i.e. the 0–5 cm sediment layer – 2.5 cm bsf; the 5–10 cm sediment layer – 7.5 cm bsf; the 10–15 cm sediment layer – 12.5 cm bsf, etc. Samples were not filtered after centrifugation, however, the correction for turbidity of pore water was applied. Some studies do not recommend filtering of samples for parameters related to gas content (e.g. Rainwater & Thatcher 1960) or toxicity (e.g. Ankley et al. 1994). The procedure of Method III is usually carried out under anaerobic atmosphere to avoid reactions that change redox potential of anoxic sediments. In our case, samples were exposed to atmospheric air during sectioning and transferring into centrifuge tubes as one of our objectives was to determine the effect of oxidation. Sediment cores in this method are usually disturbed during the extrusion from core liners, especially when they contain gaseous compounds (e.g. methane that migrates up the cores after decompression). The results obtained with this method are mean values for pore waters of each sediment section and are not the values for a specific sediment depth, as in the case of the first and second method. The thickness of the sections was selected as 5 cm to obtain averaged profiles of the examined parameters with the same number of subsamples as in Method I and II.

For the purpose of this study, pore water samples for sulfides and ammonia were analyzed spectrophotometrically, immediately after the extraction (within

max 2 h after core retrieval), in a ship-based laboratory. The samples were stored in the dark at 5°C for a short period between the extraction of pore water from the cores and the analysis. Subsamples of 2 ml were analyzed for sulfides (sum of H₂S, HS- and S^2- ), using the methylene blue method described by Fonselius et al. (1999). The limit of detection for sulfides was 1 μM and RSD < 3% (based on calibration curves). In the case of ammonia (sum of NH₄ ⁺ and NH₃), subsamples of 0.5 ml were diluted with milli-Q water and analyzed according to the standard indophenol blue method used in marine chemistry (Grasshoff et al. 1999). The limit of detection for ammonia was 2 μM and RSD < 3% (based on calibration curves). The statistical analysis (Statistica v.10 software) was carried out to compare the obtained results.

Ammonia concentrations in pore waters from different sediment depths, obtained using Method I (rhizon samplers) were generally similar for all three cores (Fig. 1a). The concentrations ranged from 1.88 to 3.29 mM, from 1.11 to 3.05 mM and from 1.02 to 2.92 mM for cores 1, 2 and 3, respectively. The mean RSD value for this method was 8.05%.

Figure 1

In Method II (centrifugation of subsamples obtained with luer tip removed syringes), the obtained concentrations were within the following ranges: 1.07–1.75, 1.23–1.84, 0.93–1.44, for cores 4, 5, 6, respectively. The least similar ammonia values were measured in pore waters collected from the depth of 15 cm bsf – the difference between cores 5 and 6 was almost twofold, while the most similar values were those from the levels of 5 and 10 cm bsf (Fig. 1b). The mean RSD for this method (depths 5, 10, 15 cm bsf) was 17.02%.

The highest values of ammonia were obtained for pore water profiles of sediments sampled using Method III (core sectioning + centrifugation; Fig. 1a,b,c). They did not differ significantly between the cores (7, 8 and 9; Fig. 1c). The concentration ranges for each core were as follows: 1.11–3.62 mM (core 7), 1.47–3.69 mM (core 8) and 1.21–4.13 mM (core 9; Fig. 1c). Again, differences between the average concentration profile for this method and specific values in each core were generally < 10% (mean RSD for this method = 8.23%)

Sulfide concentrations in pore water samples extracted using the first method (rhizon samplers) ranged from 315 to 495 μM, from 299 to 618 μM and from 315 to 535 μM for cores 1, 2 and 3, respectively (Fig. 2a). The difference between the values for specific sediment depths of the three cores, expressed as the mean RSD, was 12.23%. The largest values were obtained for sediments at depths of 10 cm (RSD

Figure 2

= 18.24%) and 25 cm bsf (RSD = 26.04%), and the smallest ones for 20 cm bsf (RSD = 4.80%) and 35 cm bsf (RSD = 3.27%).

The concentrations of sulfide in pore waters of each core sampled using Method II (centrifugation of sediment subsamples collected using cut syringes) ranged from 185 to 365 μM, from 194 to 346 μM and from 301 to 386 μM for cores 4 and 5, 6, respectively (Fig. 2b). The profiles of sulfide concentration obtained with this procedure were the most similar among all three tested methods. The difference between the average profile and the values in individual cores, expressed as the mean RSD, was 9.07%. At a depth of 25 cm bsf, however, the average sulfide concentration for this method was almost two times lower than for the first method, i.e. the mean value of 189 μM for Method II and 379 μM for Method I.

The third method (core sectioning + centrifugation of sediment sections) produced the least repeatable sulfide concentration values. The concentration ranges for the cores were as follows: 29–298 μM (core 7), 53–389 μM (core 8) and 31–470 μM (core 9; Fig. 2c). The shapes of the sulfide profiles in the sediment column were completely different for each core sampled using this method (mean RSD value of 51.59%). For most layers, a particular sulfide value was sometimes over two times higher than the value for the same level but a different core, e.g. sulfide concentration in the layer of 5–10 cm bsf of cores 8 (373 μM) and 9 (117 μM). Moreover, the lowest sulfide values obtained with this method were measured in pore waters of the bottom layer of each core (Fig. 2c), i.e. 29 μM (30–35 cm bsf, core 7), 53 μM (30–35 cm bsf, core 8), 31 μM (25–30 cm bsf, core 9).

Based on the statistical analysis, it was found that a method used for pore water sampling had a statistically significant impact (Kruskal–Wallis test, p = 0.001) on the concentration of ammonia in the sediment pore water. The shape of the average ammonia profile was quite similar for each sampling method (Fig. 3). This was particularly clear for the cores in Method I and III. The concentration values for ammonia measured in pore water obtained by sediment sectioning and centrifugation (Method III) were on average 35% higher than the concentrations in samples from cores collected by Method I (rhizon samplers) and Method II (centrifugation of subsamples obtained with cut syringes). Based on the ammonia profiles (Fig. 1) and considering the factors related to each method, we observed that core sectioning, disturbance and contact with atmospheric air (all corresponding to Method III) led to the highest concentrations compared to factors related to the two other methods.

Figure 3

At pH values of pore waters from the Gulf of Gdansk (6.44–8.04; Brodecka-Goluch & Łukawska-Matuszewska 2018; Brodecka 2013; Łukawska-Matuszewska & Kiełczewska 2016), most of the ammonia content (> 97%) in pore waters is present in the form of the ammonium ion (Emerson et al. 1975) and therefore, the potential escape of gaseous

NH₃ while sectioning a core (Method III) is negligible for the results of the analysis. However, in the case of core disturbance occurring during sectioning, the loss of about one third of hydrogen sulfide was reported (see Results and Discussion subsections related to sulfide). Assuming that with the typical pH range of pore water samples (on average ~7), about 50% of sulfide was present in the form of hydrogen sulfide (e.g. Holmer & Hasler-Sheetal 2014), the loss of 30% of H₂S led to a pH increase of about 0.5–2, depending on the exact in situ pH values of pore waters (from a range of 6.44–8.04 – typical for pore water in the sediments from the Gulf of Gdansk). We assume that as a result of H₂S loss and/or oxidation, pH of pore waters in Method III was changed (Chapman et al. 2002) in relation to samples in Method I and II. According to the literature, pH of samples may affect the results of ammonia determination by the indophenol blue method (Tzollas et al. 2010; Crompton 2006 and references therein). The mechanism of reactions that occur during ammonia analysis using the indophenol blue method is complicated (Aminot et al. 1997) and it largely depends on pH. The formation of monochloramine (reaction intermediate) requires a pH of 8–11.5 and the obtained indophenol is fully oxidized at pH of 10.8 (Grasshoff et al. 1999). In our study, as we previously assumed, the initial pH and buffer capacity of samples subjected to ammonia determination was different in Method I and II vs Method III. This factor may have been responsible for the differences in the ammonia profiles obtained by Method III.

Some authors (e.g. Zadorojny et al. 1973) observed the interference of amino acids in the indophenol method. Although amino acids concentrations in seawater are relatively low (0.5 μM; Grasshoff et al. 1999), the amounts found in the sediments are considerably greater. It has been found that the concentrations of amino acids in pore water of surface sediments are one or two orders of magnitude higher than those in the overlying water (Mintrop & Duinker 1994 and references therein). Moreover, these organic compounds can be adsorbed on clay minerals present in the sediments (Friebele et al. 1981). Despite the fact that amino acids were not analyzed in pore water samples in the present study, we speculate that to some extent they may have interfered with the ammonia measurements. They may have caused differences between the results obtained for immediately filtered and extracted pore water (Method I, rhizon samplers) and pore water collected after core sectioning and centrifugation of the whole sediment sections (Method III), where there was a greater possibility of amino acids adsorption on sediment particles (clay minerals). It has been observed that amino acids can serve as positive (e.g. L-lysine-HCl, L-threonine) or negative (e.g. L-cysteine, L-glutamic acid or L-phenylalanine) artifacts in the ammonia determination by the indophenol blue method. Accordingly, they can increase the final result of ammonia concentration by ~3% or decrease by 5-13% (Zadorojny et al. 1973). However, previous studies tested the effect of amino acids for ammonia concentrations of about one order of magnitude lower than in pore waters.

The next factor that cannot be ignored in the discussion of pore water extraction methods is filtration as it may change the concentration of an analyte in pore water (Ankley et al. 1994). Despite the fact that Song et al. (2003) and Schrum et al. (2012) successfully applied rhizon samplers (as a method based on filtration) for ammonia measurements in pore waters from lake and marine sediments and obtained accurate results, similar to those obtained by other methods applied for comparison, Ibánhez & Rocha (2014) identified and described the NH₄⁺ adsorption on PES membranes used in rhizons. They found that the highest potential for this adsorption occurs at low temperature and low salinity, and with a small volume of filtered water (< 2 ml). For example, at 5°C in a ~350 μM NH₄⁺ solution the adsorption is about 28 nM cm^-2. According to Ibánhez & Rocha (2014), adsorption capacity of PES membranes is almost two times higher at < 5°C than at 20°C. On the other hand, the impact of salinity is associated with ion competition for adsorption sites. At NH₄⁺ concentrations < 400 μM, the effect of adsorption was observed for salinity values up to 10 PSU (Ibánhez & Rocha 2014). The authors also suggested pH as a factor affecting the NH₄⁺ sorption capacity of membranes (as it controls the equilibrium between NH₄⁺ and NH₃), however, this was not further researched by Ibánhez & Rocha (2014). In the present study, one of the objectives was to check whether rhizon samplers provide lower ammonia concentrations due to filtration (pore size of 0.15 μm) and subsequent adsorption. The results obtained in our study did not provide a clear answer as the concentrations in Method I (rhizon samplers) were lower than in Method III (core sectioning + centrifugation, no filtration), but similar to the concentrations in Method II (centrifugation of sediment subsamples without core sectioning, no filtration). These results would rather indicate a factor related to Method III, which contributed to the higher concentration, e.g. interferences in the indophenol blue method caused by changes in pH due to hydrogen sulfide escape/oxidation. In general, the case of ammonia adsorption on PES membranes used in rhizons needs further research in order to obtain clear results, especially with regard to physical and chemical conditions that potentially promote this process in brackish environments.

Based on the obtained results, we believe that the differences in pH, and partly also the effect of amino acids, were the main causes of higher ammonia concentrations obtained by Method III compared to Methods I and II. However, the possibility that sample filtration (rhizons in Method I) led to the adsorption of ammonia cannot be completely excluded.

The statistical analysis conducted for sulfide results confirmed that the method of pore water sampling had a statistically significant impact on sulfide concentrations (Kruskal–Wallis test, p = 0.0001). The difference in the concentrations was evident not only in the case of average profiles for each method (Fig. 4), but also in the case of profiles for cores sampled using the same method (Fig. 2a,b,c). This was probably caused by the gaseous nature of the examined component. The highest sulfide concentrations were measured in cores sampled by

Figure 4

rhizons (Method I) and the lowest ones in sediments that were pushed out, cut and centrifuged (Method III; Fig. 4). The lower concentrations of sulfide obtained using Method II and III can be explained by its upward migration and escape (in Method III during core sectioning), followed by oxidation (during exposure of sediment sections to open air in Method III; very limited exposure in Method II). The highest rates of escape and oxidation were determined in the core that was sectioned as the last one (core No. 9) compared to core No. 7 that was cut first, about 1 h before core No. 9 (average concentration of sulfide = 230 μM for core No. 7 vs 203 μM for core No. 9). In addition, the concentration of sulfide in the topmost sediment layer was much higher (470 μM) in the last core (No. 9) compared to core No. 7 (298 μM), which also confirms the existence of the upward migration.

Taking into account the range of pH values of pore waters in the Gulf of Gdansk mentioned in the previous subsection and based on the relationships between sulfide speciation and pH (e.g. Holmer & Hasler-Sheetal 2014), it can be assumed that about 30–80% of sulfide in the measured samples was present in the form of dissolved, non-volatile HS", and the rest in the form of volatile H₂S. Thus, during our experiment we tried to minimize the time between the core retrieval and sectioning to avoid any additional artifacts. Nevertheless, the unavoidable process of the core decompression, the subsequent sectioning (sediment disturbance) and at least partial exposure of individual sediment sections to atmospheric air in Method III resulted in the greatest hydrogen sulfide loss and oxidation compared to the two other methods (Schulz 2000; Cline & Richards 1969). Part of the volatile H₂S escaped from the cores and another part was oxidized to sulfates (Chapman et al. 2002). In addition, centrifugation of sediments in Method II and III led to further H₂S loss from pore water to the headspace of centrifugation tubes (Henry's law). Based on the differences between (1) Method I and II, and (2) Method II and III, we calculated the effect of centrifugation and core disturbance with subsequent oxidation, respectively. The loss of H₂S as a result of core disturbance/sectioning and sulfide oxidation (loss of ~20% of sulfide) was greater than that resulting from centrifugation (loss of ~10% of sulfide; Fig. 4). In contrast, the loss of hydrogen sulfide in Method I was minimal as rhizon samplers ensured no contact with air during extraction. Finally, the difference between Method I and Method III with regard to sulfides was on average about 30%.

Based on the applied methods, we were not able to verify whether the filtration process (Method I) led to any adsorption of sulfides on the PES membrane. However, even if any adsorption occurred, it was probably negligible compared to the negative effect of core sectioning (core disturbance and oxidation) or centrifugation on sulfide concentrations.