Physico-chemical properties of the sediments
Physico-chemical properties in the sediment–water interface have an essential impact on the P binding of the sediments. Based on the O2 concentrations in the near-bottom water, the surface layers of sediments S2 and S7 were considered oxic and those of S3, S4, S5 and S6 hypoxic/reduced (Table 2). At the time of sample sectioning, sediment S1 was visually evaluated as oxic due to its light brown 0–2 cm surface layer. In oxic sediments the redox-sensitive Fe oxyhydroxides are assumed to be capable of binding P. Organic rich muddy sediments have often fluffy surface layers and high water contents. Accordingly, the water content in the sediments ranging from 66 to 92% (Table 2) correlated positively with TC (2094 − 7256 µmol g−1) and TN (647 − 952 µmol g−1) (t = 4.916 and 4.847, respectively, df = 6, p = 0.003). Sediments S1, S3 and S4 were considered organic rich since they were significantly higher in TC than sediments S2, S5, S6 and S7 (t = 5.045, df = 6, p = 0.004). The latter four sediments, in turn, had higher SSA values (17.4 − 21.0 m2 g−1) indicating that they had potentially larger surface area for the sorption reactions than the previous ones (12.8 − 16.1 m2 g−1).
As common in the investigated sea area, all sediments were relatively high in TFe (> 700 µmol g−1) and TAl (1635 − 2614 µmol g−1) but clearly low in Ca  (Table 2). TP in the sediments ranged typically from 36 to 52 µmol g−1. However, sediment S2 differed from others by its clearly higher TP, TFe and TMn contents (see Table 2). In the portioning of the fresh sample, sediment S2 was found to contain abundantly small Fe–Mn concretions (diameter range of 0.5 − 3 mm). The concretions were avoided during the portioning. However, although the sieving of the freeze-dried sample removed the biggest concretion fragments (Ø > 2 mm), it was impossible to exclude them completely as the dry concretions were very fragile and broke during sample handling. The presence of the concretions explains the high TP, TFe, TMn and SSA in this sediment. The concretions also explain the freeze-drying-induced elevation in ascorbate-extractable Fe in sediment S2 from 29.5 to 94.9 µmol g−1 and Mn from 5.7 up to 89.2 µmol g−1. Correspondingly, in sediment S3, freeze-drying increased the ascorbate-extractable Fe from 8.3 to 22.6 µmol g−1, while Mn increased only from 0.3 to 0.4 µmol g−1. In this originally hypoxic sediment, the low easily reducible Mn indicates its reduction-induced dissolution. The freeze-dried sediment S1 was the highest in ascorbate-extractable Fe, 92.2 µmol g−1.
Effects of freeze-drying on phosphorus exchange
In the three investigated sediments, freeze-drying increased the desorbable P and the EPC0 and decreased the P sorption, as reflected in lower Q values. These findings are supported by earlier studies with soils and sediments reporting similar responses both to freeze-drying [28, 43] and air-drying [10, 44]. The freeze-drying-induced changes resulted in more gently bending isotherm graphs (Fig. 2) pointing to a lowered P buffering capacity (Table 3, kEPC0, k5, k20 and k100) as compared with the corresponding fresh samples. In the investigated sediments, the freeze-drying-induced changes were parallel but differed in their magnitude. The sediments varied in their physico-chemical properties allowing us to detect possible reasons for the observed changes. We assessed the differences in P sorption between the treatments at various stages of the isotherm graphs, which is not usually done in the isotherm studies.
The originally oxic sediment S1 was least affected by the freeze-drying. Freeze-drying increased desorbable P only slightly and resulted in similar increase in EPC0 (Fig. 3). Sediment S1 had the steepest isotherms and thus the greatest ability to sorb P (Fig. 2). The high P sorption can be explained by the abundance of easily reducible Fe (FeAsc 92.2 µmol g−1) reflecting presence of poorly crystalline Fe oxyhydroxides in the sediment. Freeze-drying decreased the P sorption significantly only at the highest P addition (QP100) (see Online Resource Table 1). However, the buffering capacities in lower P additions were ca four to five times higher in the fresh than in the freeze-dried samples (kEPC0, k5 and k20; Table 3) and the difference was even greater at the highest P addition (k100). However, the k values of the replicates deviated most in fresh sediment S1 since its isotherm was almost linear while the Freundlich fitting usually describes well particularly bending isotherms .
In originally hypoxic sediment S3, the isotherm graphs were more gently bending than in originally oxic sediment S1 (Fig. 2), presumably due to the lower amount of Fe oxyhydroxides (FeAsc 22.6 µmol g−1). The freeze-drying-induced changes were also more marked than in sediment S1 and seen in significantly lower P sorption (e.g. the Q values) nearly at all P additions (see Online Resource Table 1). The differences in the P buffering capacities between the treatments were clear at the beginning of the isotherm (kEPC0 and k5) being ca seven times higher in the fresh than in the freeze-dried samples (t = 18.096, df = 2, p = 0.003 and t = 6.748, df = 2, p = 0.021, respectively), but they decreased with increasing P additions (Table 3). Also, in sediment S3, desorbable P and EPC0 increased only slightly upon freeze-drying (Fig. 3).
Sediments S1 and S3 were similar in their OM but differed in their original O2 status and Fe oxyhydroxide content. Since in both sediments freeze-drying lowered the P sorption it is likely that the drying resulted in changes in their sorption surfaces. Raiswell et al.  showed with synthetic ferrihydrite that freeze-drying and the following storage of the dry sediments (at room temperature) may lead to aggregation of poorly crystalline Fe oxyhydroxides. Poorly crystalline Fe oxyhydroxides have a high affinity for P , but their structural changes, for instance along with ageing, weakens the P sorption . The originally oxic sediment S1 was four times higher in Fe oxyhydroxides than the originally hypoxic sediment S3. Thus, the decrease in the P sorption was less pronounced in sediment S1 having more Fe oxyhydroxides available for the P sorption than sediment S3, regardless of the possible Fe aggregation during freeze-drying. Therefore, in sediment S1, the weakening sorption of P was not clearly seen until at the highest P additions. In fact, the notably high P buffering capacity k100 in fresh sediment S1 indicates the abundance of unoccupied adsorbents and a high P sorption potential left even after the highest P addition. This additional information highlights the benefit of investigating the buffering capacities at different stages of the isotherm graph. It is also noteworthy that the freeze-dried samples had been stored for different periods prior to the isotherm experiment: S1 for 1.5 and S3 for 4 months. Thus, we cannot exclude the potential effect of the storage time on the results of the dried samples, although the changes are presumably slow .
Previous studies have reported that air-drying causes rupture of OM with consequent release of P [10, 43, 48]. In sediments S1 and S3, the slight increases in desorbable P (and the concomitant increase in EPC0) can originate from ruptured OM (Fig. 3). It is also noteworthy that the isotherm experiments with these sediments were performed using dissimilar solid-to-solution ratios (1:200 in S1 and 1:300 in S3) as explained in the Materials and methods section. It is possible that the smaller solid-to-solution ratio (i.e. the lower particle concentration) solely explains the greater increase in desorbable P and EPC0 in freeze-dried sediment S3 than in sediment S1. Furthermore, Hjort  found freeze-drying to promote dissolution of Fe from fractions representing OM in anoxic OM-rich sediments. Thus, the freeze-drying-induced increase in the ascorbate-extractable Fe in sediment S3 may originate from changes in the sediment OM. However, since the freeze-drying decreased the P sorption, it suggests that the released Fe did not participate in the P sorption.
The originally oxic concretion-rich sediment S2 showed the lowest P sorption and, thus, had the most gently bending isotherm graphs (Fig. 2). The Q values were consistently lower in the freeze-dried than in the fresh samples, but the buffering capacities were similar in both treatments (Table 3). The difference in Q values between the treatments was greater (ca 0.2 µmol g−1) than in sediment S1 (0.01 − 0.08 µmol g−1) but smaller than in sediment S3. However, the freeze-drying increased markedly desorbable P (t = 19.399, df = 2, p = 0.003), resulting in a corresponding increase in EPC0 (Fig. 3). Presumably, this released P was occluded in the Fe–Mn concretions prior to their disruption during drying and sieving of the sample. This conclusion is supported by the drying-induced three- and 15-fold increases in ascorbate extractable Fe and Mn, respectively, and by the exceptionally high content of TP, TFe and TMn in the sediment. According to Winterhalter , some surface sediments in the Gulf of Finland contain concretions high in Fe, Mn and P. Furthermore, part of the P in the concretions is found to be dithionite extractable (Lukkari pers. comm.) assumed to be bound onto reducible Fe compounds, mainly Fe oxyhydroxides .
Air-drying of originally submerged sediments has been reported to elevate EPC0. For instance, in the study of Qiu and McComb  the increase was 0.4 µmol l−1, whereas Kerr et al.  recorded an increase as high as 5–5.8 µmol l−1. Simpson et al.  compared the effects of air-drying and freeze-drying with the corresponding fresh fine (< 2 mm) stream sediments having originally low EPC0 (< 0.65 µmol l−1). In their study, the freeze-drying either increased or decreased EPC0 by 0.3 µmol l−1, the effects being rather modest compared with those recorded in the air-drying. In our previous study with 22 freeze-dried surface sediments collected from the coastal northern Gulf of Finland, the EPC0 values ranged from 0.2 to 38 µmol l−1 . Thus, the freeze-drying-induced increase of ca 0.5 µmol l−1 recorded in the present study (i.e. in the sediments that did not contain concretions) is relatively small. This highlights the importance of evaluating the impact of freeze-drying on the P exchange in the context of the observed scale of the investigated parameters in the given research area.
Phosphorus exchange in artificial seawater and ambient water solutions
Another goal of the present study was to find out whether ASW would produce similar P exchange isotherm as ambient water. The ambient water of sediments S1 and S4 − S7 contained 1.3 − 3.4 µmol l−1 PO4-P, 0.1 − 7.6 µmol l−1 NO3-N, 1.3 − 5.8 µmol l−1 NH4-N and 14.3 − 39.5 µmol l−1 SiO4-Si (Online Resource Table 2). The highest PO4-P and NH4-N concentrations were recorded in the ambient water of sediment S4. The pH is known to affect the P sorption. Therefore, we determined pH of the equilibration solutions at the lowest and the highest P additions (pH-P0 and pH-P100) and found out that they were generally higher in the ambient water than in the ASW (Online Resource Table 2). The biggest differences (0.4 − 0.5) in pH values between the ASW and the ambient water were recorded in sediments S4, S5 and S7, but their impact was not clearly visible in the isotherms (Fig. 4).
In sediments S1, S5, S6 and S7, the P exchange isotherms produced with ambient water and ASW were similar, while in sediment S4 they differed to some extent (Fig. 5). Generally, the P sorption (Q values) was slightly higher in the isotherms produced when using the ambient waters (supplementary data Table 3), probably due to the higher P concentration (1.3 − 3.5 µmol l−1) in the ambient water than in ASW. Also, the ionic composition being more diverse in the ambient water than in ASW solutions, presumably resulted in a higher ionic strength in the former, even though the salinities of the ASW solutions were adjusted to correspond to those of the ambient water. Increasing ionic strength enhances the P sorption . Nevertheless, the variation in the Q values (on average 0.2 − 1.0 µmol g−1) between the two different solutions was close to that recorded between the replicate samples and in most of the sediments the isotherm graphs almost equalled.
In sediment S4, ASW produced a steeper P exchange isotherm than the ambient water. This was reflected in P buffering capacities being somewhat higher in the ASW (≤ 2.7 l g−1) (Table 4). Sediment S4 was higher in OM (TC 8.7%) than the other investigated sediments (TC 3.6 − 7.5%). This was presumably reflected in the properties of its ambient near-bottom water and, thus, in the lower P buffering capacities in the sediment. Dissolved organic matter (DOM) can contain organic anions that can disturb the P sorption by competing for the same sorption sites  and, thus, lower the P buffering capacity of the adsorbent. The highest PO4-P and NH4-N concentrations in the ambient water of sediment S4 can refer to the presence of mineralization products, which supports this conclusion. In the other sediments, however, the k values were rather similar. Also, the EPC0 values as well as the isotherm graphs produced with the ambient water and the ASW were almost equal (Fig. 6). Our results suggest that ASW can be used to replace ambient water in the P exchange studies without significant effects on P sorption. In fact, in the P exchange studies emphasizing the comparison of the sediment properties, ASW may even be a better option than ambient water, as ASW minimizes the interference of the unknown solution matrix in the experiment.
Noteworthy matters regarding phosphorus exchange experiments
The results of this study highlight that when examining the effects of dissimilar treatments or experimental conditions on the P exchange isotherms, it is important to assess the magnitude of the changes in a proper context. For instance, differences that are close to the analytical detection limit of P can be considered minor. Thus, when considering the detection limit of the P analysis (0.05 µmol l−1; ) and the range in EPC0 values in brackish water sediments (up to 38 µmol l−1; ), it is justified to state that the freeze-drying-induced changes on the easily exchangeable P and on EPC0 (0.1 − 0.5 µmol l−1) in sediments S1 and S3 were practically quite small. Furthermore, the analytical uncertainty assessed by means of repeated measurements (n = 20) of the commercial control solutions containing 0.42, 4.2 and 10.5 µmol P l−1 was on average 13, 6 and 5%, respectively. This should be considered, particularly when examining the differences in EPC0 values and the equilibrium P concentrations at low P additions. We also assessed the impact of the solid-to-solution ratio on P exchange by executing an additional experiment using freeze-dried sediment S2 with the same experimental setup with a wider solid-to-solution ratio (1:200) than used when comparing the fresh and the freeze-dried subsamples of sediment S2 (1:50) (Fig. 6). Decreasing the solid-to-solution ratio from 1:50 to 1:200 elevated EPC0 by 1.5 µmol l−1 and the P desorbed from the sediment (i.e. Q0) by 1.5 µmol g−1. This result highlights that also the solid-to-solution ratio used should be considered when interpreting or comparing the results of the isotherm experiments. Furthermore, also the possible influence of the isotherm fitting on the calculated P exchange parameters should be taken into account.
As shown in the present as well as in some previous studies, freeze-drying and storage of fresh samples affect sediment properties [20, 28]. In particular, originally reduced sediments are subjected to fast changes when in contact with atmospheric O2 [20, 55]. Thus, they are expected to exhibit chemical changes if stored unprotected from oxidation. When storing of sediments prior to P exchange experiments cannot be avoided, the practices to be used should be carefully planned and preferentially tested according to the goals of the research. Furthermore, the storage period of the samples to be compared should be kept equal. The use of fresh sediment samples is preferable if accurate weighing, according to known dry matter content, is possible and the experiments can be carried out in N2 atm soon after the sampling and if the sampling and the storage can be performed shielded from atmospheric O2. As can be seen in the isotherms of fresh sediments S2 and S3, the volumetric portioning resulted in more marked deviation among the replicates than weighing (see Fig. 2b).