Separate phases of metal partitioning behaviour in freshwater lakes that receive varying degrees of atmospheric contamination and have low concentrations of suspended solids were investigated to determine the applicability of the distribution coefficient, KD. Concentrations of Pb, Ni, Co, Cu, Cd, Cr, Hg and Mn were determined using a combination of filtration methods, bulk sample collection and digestion and Inductively Coupled Plasma-Mass Spectrometry (ICP-MS). Phytoplankton biomass, suspended solids concentrations and the organic content of the sediment were also analysed. By distinguishing between the phytoplankton and (inorganic) lake sediment, transient variations in KD were observed. Suspended solids concentrations over the 6-month sampling campaign showed no correlation with the KD (n = 15 for each metal, p > 0.05) for Mn (r2 = 0.0063), Cu (r2 = 0.0002, Cr (r2 = 0.021), Ni (r2 = 0.0023), Cd (r2 = 0.00001), Co (r2 = 0.096), Hg (r2 = 0.116) or Pb (r2 = 0.164). The results implied that colloidal matter had less opportunity to increase the dissolved (filter passing) fraction, which inhibited the spurious lowering of KD. The findings conform to the increasingly documented theory that the use of KD in modelling may mask true information on metal partitioning behaviour. The root mean square error of prediction between the directly measured total metal concentrations and those modelled based on the separate phase fractions were ± 3.40, 0.06, 0.02, 0.03, 0.44, 484.31, 80.97 and 0.1 μg/L for Pb, Cd, Mn, Cu, Hg, Ni, Cr and Co respectively. The magnitude of error suggests that the separate phase models for Mn and Cu can be used in distribution or partitioning models for these metals in lake water.
Citation: Gormley-Gallagher AM, Douglas RW, Rippey B (2015) The Applicability of the Distribution Coefficient, KD, Based on Non-Aggregated Particulate Samples from Lakes with Low Suspended Solids Concentrations. PLoS ONE 10(7): e0133069. https://doi.org/10.1371/journal.pone.0133069
Editor: Liping Zhu, Institute of Tibetan Plateau Research, CHINA
Received: September 3, 2014; Accepted: April 3, 2015; Published: July 22, 2015
Copyright: © 2015 Gormley-Gallagher et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: Data have been deposited to Figshare: http://dx.doi.org/10.6084/m9.figshare.1448893.
Funding: This work was funded by the Department for Employment and Learning (DEL) Northern Ireland. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Colloidal matter in lakes is composed of clays, labile organic matter, hydrous metal oxides, and phytoplankton artefacts and the particles typically range in size from 0.1 μm to 1.0 μm in diameter . Its size overlaps with fundamental biogeochemical phases of metal cycling in lakes; the dissolved and particulate phases. The dissolved metal fraction is generally defined as the metals passing through a 0.45 μm filter [2, 3] and the particulate fraction is an aggregate of two or more properties, predominately the phytoplankton and lake sediment, which frequently range in diameter size from 0 μm to 0.1 mm . The influence of colloids on metal partitioning behaviour in lakes is therefore difficult to operationally define, which is of particular significance when calculating the distribution (or partition) coefficient, KD.
The distribution coefficient is the most common and simplest method of estimating the extent of contaminant retardation from particles to water . It is calculated as the ratio between metals in the particulate and dissolved phases [6, 7] and is used widely in prioritizing site remediation and waste management decisions . However, consensus is building that the use of KD as a descriptor for metal partitioning between solids and water is probably unsuitable .
There are limited studies on KD in natural aquatic environments , and those available often record great variations in KD as a result of biogeochemical factors and limitations of measurement [11, 9, 12, 7]. There are two related problems with KD. Firstly, to collect a concentrated sample of particles in the water column, sediment traps are frequently used , which not only exposes the sample to potential metal contamination for longer, but also produces time-averaged values that are input to the KD formula with other parameters not reflective of the same timeframe. The concentrations of metals in the top 5 cm of sediment from the lakebed are often used to overcome this . However, as mentioned, the particulate phase typically represents aggregated phytoplankton and lake sediment, both of which can have contrasting metal affinities that are not static in time or space. This means that, by not distinguishing between the particulate fractions, KD may have little power in estimating partitioning behaviour because there may be dissimilar mechanisms controlling the distribution between the phases.
Secondly, the presence of colloids in water with a high suspended solids (SS) mass can disguise the true distribution of metals, because as SS increases, so too does the concentration of colloidal matter, which typically increases dissolved (filtrate) metal concentrations . Therefore, an increase in colloids with SS may appear to dilute the particulate phase, lowering the KD. This typically results in an inverse relationship between KD and particle concentration, i.e. the particle concentration effect (PCE ). In recognition of such problems, Benoit et al.  suggests it would be valuable if more research was carried out on the behaviour of KD in freshwater systems that have a markedly low SS (≈ 0.1 mg/L) where the influence of colloids is minimised. Yet the PCE has been observed under conditions whereby the concentrations of colloidal matter are eliminated .
What remains unknown is whether the predicted inverse relationship between KD and particle concentration is still evident when the phytoplankton and lake sediment are measured individually (i.e. non-aggregated particulate samples) and then incorporated into the KD formula. This should be investigated in freshwater systems with a low SS so to discount the influence of colloids. To address this, we set out to: (1) investigate lake systems with a considerably low SS mass; (2) calculate separately the phytoplankton and lake sediment fractions of the particulate phase in order to yield metal KDs that reflect transient conditions; and (3) examine the relationship of these KD with SS.
Investigations were undertaken in three lakes that have been shown to receive varying degrees of metal contamination in the UK . That is, one lake was selected in each of the regions that receive high, medium and low metal contamination from the atmosphere (Fig 1). Due to the need for appropriate lacustrine data on the relationship between metals in the phytoplankton and the dissolved phase [19, 20, 21, 22], it was considered important to obtain such data from a range of metal contaminated regions in order to address any variations. Also, the three lakes receive metal contamination solely from atmospheric deposition—and thus metal contamination from runoff or direct discharges would not influence our results [23, 18]. Additionally, the size and bathymetry of the lakes meant that regular sediment resuspension events (and by association high SS loads) would be unlikely to influence our investigation outcome [24, 25, 26].
The study defined the regions by determining the metal burden from the background Pb concentrations (in lake sediment cores at depths dated pre-1900) and the Pb flux to the sediment at each core slice post-1900. The yellow circles in the low, medium and high regions of Pb contamination depicts the locations of Loch Coire nan Arr, Loch Doilet and Loch Urr, respectively.
In the following detailed site descriptions, lake surface area, perimeter, altitude, grid reference, catchment area and distance from the sea and to the nearest village were calculated and/or obtained using the OS Landranger Memory-Map V5 edition (2006) for northern and southern Scotland (Licence number PU 100034184). The maximum lake depths were based on collected field data, while catchment geology, vegetation and soil type were derived from Patrick et al. .
Loch Coire nan Arr has a surface area of 13.21 ha, a maximum lake depth of 11 m and a catchment area of 8.45 km2 (Table 1). It is the most northerly of the three sites and lies in the region of low metal contamination from the atmosphere . The catchment is dominated by steep corrie cliffs, and the lake itself fills a large deep sandstone corrie that was carved by deglaciation at the end of the Pleistocene. Loch Coire nan Arr is one of the six UK sites represented in the UNECE International Co-operative programme on Assessment and Monitoring of Acidification of Rivers and Lakes . Permission for sampling the site was obtained from The Applecross Trust, a conservation charity responsible for the management of the lake (contact: email@example.com).
Loch Doilet has a surface are of 51.55 ha, a maximum lake depth of 16 m and a catchment area of 33.51 km2 (Table 1). The lake, lying northwest of the Ben Nevis Mountain range, is the largest of the three lakes and has received moderate metal contamination from the atmosphere . The catchment rises from the lake to a peak of approximately 720 m. The dominant soil types are peats, which are eroded on the uppermost reaches of the catchment . Permission for sampling the site was obtained from the Forestry Commission Scotland, a UK non-ministerial government department responsible for the management of the lake (contact: firstname.lastname@example.org).
Loch Urr has a surface area of 47 ha with a maximum lake depth of 13 m (Table 1). It lies in the Dumfries and Galloway region of south-west Scotland, an area that has received high metal contamination from the atmosphere . The lake drains the smallest of the three catchments with an area of 7.73 km2. The underlying geology is mainly composed of granite / gneiss and the land-use is confined to low-intensity sheep grazing . Permission for sampling the site was obtained from the Urr District Salmon Fisheries Board, a board of the Galloway Fisheries Trust charity set up to protect the lake and its catchment (contact: email@example.com).
Sampling campaigns were conducted on five occasions: March, May, June, July and September 2007 at each of the three lakes. The deepest parts of the lakes were located for sampling using a LPS-1 Digital Handheld Depth Sounder (Vexilar USA) and hydrographic charts from a survey completed with a MIDAS Surveyor GPS Echosounder (Fig 2). Depth profiles of conductivity, temperature and dissolved oxygen were recorded at each sampling station with a YSI 556 Multi Probe System. Lake water was sampled from 6 m below the lake surface using a Perspex Ruttner sampler . The reason for sampling at 6 m was to ensure sampling was taken from a consistent area where surface and bottom waters are completely mixed. Calculating the specific surface mixed layers was outside the scope of this study (see ). However, it has been previously shown that the free water surface over which wind acts upon in medium-sized lakes, such as those investigated in this study (according to, for example, Leppäranta’s  classification), is 6 m [32, 25, 33, 34].
The hydrographic survey was completed with a MIDAS Surveyor GPS Echosounder supplied with a 210kHz transducer (Valeport, Devon, UK) at the beginning of this study. Approximate depth is shown in metres and a change in depth is indicated by the grey/colour scale. The X denotes where the sampling took place.
Using the Perspex Ruttner sampler, on each sampling campaign at the three lakes, three 1 L water samples were collected for SS analysis and an additional twelve 50 ml water samples were extracted for metal, phytoplankton and chlorophyll a analysis. Below details the methods used to process the samples and determine metal content in the phases investigated. All equipment coming into contact with samples for metal analysis were prepared to reduce metal contamination and prevent adsorption losses . For all phases, the XSeriesI ICP-MS (ThermoFisher Scientific) was used for the analysis of Pb, Cu, Co, Hg, Co and Mn in the samples. The precision of the laboratory analysis was assessed from triplicate analysis of reference material and field samples and accuracy by using blanks . Additional methods for assessing error in calculations are detailed below.
At each lake, three 50 ml water samples were stored in vials pre-prepared with glutaraldehyde (50%, Electron Microscopy grade, EMS, U.S.A.) to produce a final concentration of 2% (v/v). A Nikon-5400 inverted light microscope at 40x magnification was used to identify, count and measure the phytoplankton, following the procedures described by Olrik et al. . Cell counts were converted to counts per volume of lake water under the assumption that 1 mm3/L-1 was equivalent to 1 mg/L-1 wet weight . Cell volumes and surface areas were calculated using the geometric equations of Hillebrand et al. . The cell surface area and volume calculations were collated with cell counts per volume of lake water to equate the surface area and biomass per volume of lake water.
Net-haul phytoplankton samples were collected and acidified on-site using the approach set out by Ho et al. . The samples were digested by treatment with hydrofluoric, nitric and perchloric acid (following Bock ). An empty beaker (a reagent blank) and two samples of certified reference material (Chinese stream sediment, GBW 07301, LGC Promochem, UK) were included with every batch. Metal concentrations in the standing crop of phytoplankton (MPhy, mg/L) were calculated through (Eq 1): (1) where PhyBulk is the concentration of metals per unit mass of phytoplankton (mg/kg), and Biomass is the phytoplankton biomass in the water column at the time of sampling (kg/L).
In addition, the concentration of the pigment chlorophyll a was determined in the water samples because it is an indicator of phytoplankton biomass . The pigment was extracted from the filtered samples into 90% V/V methanol, and the detection was performed with a Shimadzu UV-Mini 1240 spectrophotometer set at an emission wavelength of 665 nm.
Lake sediment analysis
A piston corer was used to extrude the upper 5-cm long of lakebed sediment . The material was processed and analysed using the approach outlined for the phytoplankton material. The percentage weight loss on ignition (LOI) was determined on the sediment core sample following the methods set out by Heiri et al. .
For all sampling occasions at each lake, the 3 x 1 L water samples collected were used to gravimetrically determine SS by filtering through aliquots of known volume (1 L at a time) of lake water through oven-dried pre-weighted GF-C filter paper . The concentration of metals in the SS (MSed, mg/L) was determined through: (2) where SS is the suspended solids concentration (kg/L), Biomass is as Eq 1 (kg/L), SedBulk is the concentration of metals per unit mass of bulk sediment collected from the lakes (mg/kg), and LOI is the percentage weight lost on ignition of the sediment samples, which is proportional to the organic content of the samples (%).
The dissolved phase
At each lake, water samples (plus blank samples of Millipore Milli-Q water) were acidified on-site in three 50 ml pre-prepared vials to 2% (w/v) HNO3− (ARISTAR grade) following filtration through acid-washed 0.45 μm filters. Metal concentrations were determined in the filtrate to represent the dissolved constituents .
The partition coefficient
The partition coefficient KD (l kg-1 dw-1) was calculated through Eq 3 . (3) where CPart is the particulate metal concentration (i.e. MPhy + MSed) in kg L-1, CDiss is the dissolved (filter passing) concentration (kg L-1) and SS is the suspended solids concentration in mass dry weight (dw) per volume (kg dw L-1).
Accuracy of using the separate phase fractions
The accuracy of using the separate phase fractions to model the particulate and dissolved metal concentrations was assessed by firstly measuring the metal content in (non-filtered) lake water samples. At each lake, three 50 ml water samples (plus blank samples of Millipore Milli-Q water) were acidified on-site to 2% (w/v) HNO3− (ARISTAR grade) for the direct determination of total metal concentrations from spot samples.
Next, the modelled total metal concentrations were determined—representing lake sediment, phytoplankton and dissolved metal concentrations at a steady-state (Eq 4, mg/L).(4)
The errors between the modelled steady-state concentrations (Eq 4) and the measured concentrations were approximated by calculating the root mean square error of modelled concentrations with the difference in the measured and modelled concentrations, . This method is used to assess the accuracy of the modelled values , although a separate data set would be necessary for an independent test of the model [47, 48].
Additionally, as a number of factors are included in the calculations of the total metal concentrations, the error in each of those factors were propagated through to the overall error in the modelled values. This was completed by calculation of the maximum (CMax) and minimum (CMax) total metal concentrations (Eq 5). (5) where SSMax is the maximum suspended solids concentration (kg/L), BiomassMax is the maximum biomass of phytoplankton in the water column at the time of sampling (kg/L), SedBulk is as Eq 2, MPhyMax is the maximum concentration of metals in the standing crop of phytoplankton in the water column (mg/L) and CDissMax is the maximum concentration of metals in the filtered water fraction (mg/L). Likewise, the minimum total metal concentrations were calculated according to the following Eq 6.(6)
Results and Discussion
The suspended solids concentration at the time of sampling showed no correlation with the KDs (n = 15 for each metal) of Mn (r2 = 0.0063), Cu (r2 = 0.0002, Cr (r2 = 0.021), Ni (r2 = 0.0023), Cd (r2 = 0.00001), Co (r2 = 0.096), Hg (r2 = 0.116) or Pb (r2 = 0.164) (Fig 3). The significance of the regression analysis indicates that there was no relationship between the two variables (p > 0.05).
The results in Fig 3 conflict with reported patterns of KD decrease with increasing SS due to the dilution of suspended matter and the resuspension of less contaminated sediment [6, 7, 5], i.e. the particle concentration effect (PCE ). As mentioned, the PCE can be attributed to colloidal matter in waters with a high SS range, which increases dissolved (filtrate) metal concentrations—lowering the KD [49, 4]. In Benoit et al.’s study , they reported that as SS concentration declined, the rate of decrease in KD with SS slowed down to the point that KD became near constant when the SS concentration was low at around 1 mg/L. It is also notable that the SS range in this study, i.e. 0.2–10.8 mg/L (Fig 4), was an order of magnitude lower than that investigated by Benoit et al. (i.e. 1–100 mg/L).
Data are presented as the mean ± standard deviation (SD).
These results can be understood further by examining the calculations behind the relationship of KD and SS. The concentration of SS is inevitably included in the determination of particulate metals in the water column. The association between KD and SS may then be viewed as a spurious correlation because, as Eq 7 demonstrates, if the logarithms are taken of both sides on Eq 3 and are divided by log(SS), we get : (7) It can be ascertained from Eq 7 that the slope of the correlation between log(KD) and log(SS) is -1. Considering this spurious slope, KD values would decline with the higher (supposedly) dissolved fraction. This inevitable correlation may be particularly noticeable in samples containing high concentrations of colloids because this can increase filtrate metal concentrations and thus increase the denominator in Eq 3. This is in agreement with Hart and Lowery  who warned that because a high colloidal fraction decreases the highest values of KD, particle concentration has more influence on KD. Lindstrom  also recommends that the use of KD in modelling is discontinued because the ratio that includes SS concentrations is greatly affected by spurious correlations and that the particulate fraction (Cpart: total metal) be used as a simple alternative.
The potential influence of colloids on the behaviour of KD is also manifested in the differences between the mean KD of the metals (Fig 3). The mean KD (l kg-1 dw-1) of the metals ranges in the order of (n = 15 for each metal) Cd (1.2 x 106) > Pb (2.7 x 105) > Co (1.9 x 105) >Mn (3.1 x 104) > Cu (3.2 x 103) > Cr (2.2 x 103) > Ni (1.2 x 103) > Hg (82.0). In theory this suggests the high to low range of KDs corresponds to a decreasing reactivity with particulate matter . However, as previously described, the KD values may be lowered due to a colloidal matter increase. It may be more feasible that the variations across the mean KDs of the metals relate to the tendency of each metal to form organic complexes [15, 1]. For example, it has been observed that organic Co complexes in natural waters are weak and relatively less abundant than the organic complexes of Ni [53, 54]. Therefore, Co has a higher mean KD than Ni, as observed here, because of the relatively low influence of colloidal Co interactions.
The results in Fig 4 also highlight reproducibility errors in the SS measurements. This could be because the low SS concentrations meant that some of the measurements are near/below the limit of detection of the published method, i.e. 0.5 mg/L [55, 56]. As SS concentrations are included in KD calculations, this highlights an important related problem—that KD is difficult to measure not only in higher SS lakes due to the enhanced problem of colloids as described, but also in low SS lakes due to the limitations of measurement using standard methods. Another interpretation relates to biogeochemical factors. On the occasions with high reproducibility error—firstly Loch Doilet in June, this particular sampling date was an extremely windy occasion and the lake showed its highest dissolved oxygen (12.0 mg/L) and conductivity (24.2 ppm) readings. This suggests the variation might be related to Hilton’s processes in a lake [24, 25, 26]. Regarding all three lakes in September, Loch Urr and Loch Doilet had its highest chlorophyll a (23 μg/L and 2.7 μg/L respectively), while Loch Coire nan Arr had its second highest chlorophyll a reading (3.3 μg/L)–suggesting that productivity could influence the variation. This means that SS may be higher/lower at times because of other factors (apart from methods)–which inevitably affects the KDs.
It is also important to point out that if the SS data with poor reproducibility were omitted on this basis, there would be little or no effect on the findings depicted in Fig 3. That is, on excluding the three data with the highest SD (Loch Doilet in June, and Loch Coire nan Arr and Loch Urr in September), SS concentrations over the sampling campaign still show no correlation with the KD (n = 12 for each metal in this case, p > 0.05) for Mn (r2 = 0.022), Cu (r2 = 0.198, Cr (r2 = 0.078), Ni (r2 = 0.074), Cd (r2 = 0.026), Co (r2 = 0.004), Hg (r2 = 0.046) or Pb (r2 = 0.299). Furthermore, when we performed this test also omitting September’s result for Loch Urr, there remains no correlation between the SS and KD (n = 11 for each metal in this case, p > 0.05) for Mn (r2 = 0.022), Cu (r2 = 0.20, Cr (r2 = 0.085), Ni (r2 = 0.073), Cd (r2 = 0.029), Co (r2 = 0.005), Hg (r2 = 0.049) or Pb (r2 = 0.3).
The totals (in μg/L) of the separate phase fractions (i.e. those modelled based on Eq 4) are compared against the concentrations determined in the spot samples taken at the time of sampling in Fig 5. The model errors are ± 3.40, 0.06, 0.02, 0.03, 0.44, 484.31, 80.97 and 0.1 μg/L for Pb, Cd, Mn, Cu, Hg, Ni, Cr and Co respectively. When the ratio of model error to the maximum concentration measured is used to determine the relative performance of the model , it performs from best to worst in the following order; Cu (≈ 0.0003), Mn (≈ 0.0002), Pb (≈ 0.299), Co (≈ 0.327), Hg (≈ 0.379), Ni (≈ 0.42), Cd (≈ 0.483) and Cr (≈ 0.542).
The 1 to 1 line, maximum and minimum modelled concentration due to propagation of the maximum and minimum concentrations in each phase, and the root mean square error of model prediction (RMSE, μg L-1) are shown.
In theory, Fig 5 suggests that Cu and Mn behaved the most conservatively in the lakes. To suggest a metal behaves conservatively in the context of the model used here (Eq 4) means that changes in the distribution between the phases are accounted for in the mixed phase total [57, 58]. The apparent conservative behaviour of Mn and Cu has been exhibited elsewhere. For example, it has been noted that, Mn in its carbonate form MnCO3 (rhodochrosite) is not involved in oxidation-reduction reactions and can therefore behave more conservatively than, for example, Co [59, 60]. In the case of Cu, the high level of Cu complexation to ligands present in natural waters can enhance the resistance of Cu to changes in pH [61, 62].
There are, however, four main elements of uncertainty to this interpretation. Firstly, it is possible that the modelled (Eq 4) and directly measured total metal concentrations are similar because the filtered metal fraction used to obtain the modelled values was determined from the filtrate of the same lake water used to obtain the measured values. Secondly, the dissolved (filtrate) fraction is likely to represent a greater a degree of contamination than the other fractions because phytoplankton and lake sediment measurements were based on concentrated samples and then calculated down to the specific fraction they represent. Thirdly, it is important to note that the data is based on samples taken from the deepest area of the lakes—and these are assumed to represent the whole lake. This means that the degree of sediment focusing is not accounted for , that spot samples of lake water can represent the SS, phytoplankton and total metal content in the lakes, and that the top 5 cm of lakebed sediment represented (minus the organic content) the metal concentrations in the SS. Fourthly, due to the size range of colloidal matter potentially overlapping with the other phases in biogeochemical cycles [1, 4], it is operationally difficult to distinguish between the phases and, therefore, any result inevitably reflects some concentrations in two or more phases. Nevertheless, the correlation of the modelled and measured Mn and Cu values in Fig 5 cannot be ruled out as a reflection of conservative behaviour as the analysis was handled in an identical manner for the three lakes and thus useful comparisons can be made.
The absent correlation between the KD of each metal and SS does not support the PCE theory. Imperative to our interpretation was the considerably low concentration range of SS in the three lakes during this investigation. This meant colloidal matter had less opportunity to increase the dissolved (filter passing) fraction, which inhibited the spurious lowering of the KDs and subsequent PCE. These findings conform to the increasingly documented theory that the use of KD in modelling may mask true information about metal partitioning behaviour.
Based on a simple mixing model used to estimate the particulate and dissolved Cu, Mn, Pb, Co, Hg, Ni, Cd and Cr concentrations, Cu and Mn appeared to behave the most conservatively in the lakes. This is in agreement with reports on the stability of Cu and Mn in the literature. Interpretation of these findings is limited by a potential degree of contamination in the filtrate samples compared to the measurements derived from bulk samples, inherent difficulties in operationally defining each phase and underlying assumptions that the samples taken represent the whole lake. The magnitude of the errors in relation to the ranges of concentration suggests that the separate phase models for Mn and Cu can be used in distribution or partitioning models for these metals in lake water.
Conceived and designed the experiments: AGG RD BR. Performed the experiments: AGG. Analyzed the data: AGG RD BR. Contributed reagents/materials/analysis tools: AGG RD BR. Wrote the paper: AGG RD BR.
- 1. Zhang XB, Wen AB. Current changes of sediment yields in the upper Yangtze River and its two biggest tributaries, China. Global Planet Change. 2004; 41(3–4): 221–227.
- 2. Hurley PJ, Shafer MM, Cowell SE, Overdier JT, Hughes PE, Armstrong DE. Trace metal assessment of Lake Michigan tributaries using low-level techniques. Environmental Science and Technology. 1996; 30(6): 2093–2098.
- 3. Atkinson CA, Jolley DF, Simpson SL. Effect of overlying water pH, dissolved oxygen, salinity and sediment disturbances on metal release and sequestration from metal contaminated marine sediments. Chemosphere. 2007; 69(9): 1428–1437. pmid:17568653
- 4. Jiang L, Schofield OME, Falkowski PG. Adaptive evolution of phytoplankton cell size. American Naturalist. 2005; 166(4): 496–505. pmid:16224705
- 5. Opuene K, Agbozu IE. Relationships between heavy metals in shrimp (Macro brachium felicinum) and metal levels in the water column and sediments of Taylor Creek. International Journal of Environmental Research. 2008; 2(4): 343–348.
- 6. Balls P. The partition of trace metals between dissolved and particulate phases in European coastal waters: A compilation of field data and comparison with laboratory studies. Netherlands Journal of Sea Research. 1989; 23(1): 7–14.
- 7. Nguyen TH, Goss K, Ball WP. Polyparameter linear free Energy relationships for estimating the equilibrium partition of organic compounds between water and the natural organic matter in soils and sediments. Environmental Science and Technology. 2005; 39(4): 913–924. pmid:15773462
- 8. US.EPA. Understanding variation in partition coefficient, Kd, values. United States Environmental Protection Agency. 2012. Available: http://www.epa.gov/radiation/cleanup/402-r-99-004.html.
- 9. Veselý J, Majer V, Kučera J, Havránek V. Solid-water partitioning of elements in Czech freshwaters.Applied Geochemistry. 2001; 16: 437–450.
- 10. Palleiro L, Rodriguez-Blanco ML, Taboada-Castro MM, Taboada-Castro MT. The influence of discharge, pH, dissolved organic carbon and suspended solids on the variability of concentration and partitioning of metals in a rural catchment. Water, Air and Soil Pollution. 2013; 224(8): 1651.
- 11. Shafer MM, Overdier JT, Hurley JP, Armstrong D, Webb D. The influence of dissolved organic carbon, suspended particulates, and hydrology on the concentration, partitioning and variability of trace metals in two contrasting Wisconsin watersheds (U.S.A.). Chemical Geology. 1997; 136: 71–97.
- 12. Gaiero DM, Probst JL, Depetris PJ, Bidart SM, Leleyters L. Iron and other transition metals in Patagonian riverborne and windborne materials: geochemical control and transport to the southern South Atlantic Ocean.Geochimica et Cosmochimica Acta. 2003; 67(19): 3603–3623.
- 13. Henderson GM, Heinze C, Anderson RF, Winguth AME. Global distribution of the 230Th flux to ocean sediments constrained by GCM modelling. Deep-Sea Research I. 1999; 46: 1861–1893.
- 14. Last WM, Smol JP. Tracking environmental change using lake sediments. Volume 1: Basin analysis, coring, and chronological techniques. Dordrecht: Kluwer Academic Publishers; 2001.
- 15. Pham MK, Garnier JM. Distribution of trace elements associated with dissolved samples (<0.45 μm– 1 nm) in freshwater using coupled (frontal cascade) ultrafiltration and chromatographic separations. Environmental Science and Technology. 1998; 32: 440–449.
- 16. Benoit G, Oktay-Marshall S, Cantu A, Hood E, Coleman C, Corapcioglu M, et al. Partitioning of Cu, Pb, Ag, Zn, Fe, A1, and Mn between filter-retained particles, colloids, and solution in six Texas estuaries. Marine Chemistry. 1994; 45: 307–336.
- 17. Turner S. A binary aqueous component model for the sediment-water partitioning of trace metals in natural waters. Environmental Science and Technology. 2007; 41: 3977–3983. pmid:17612178
- 18. Rippey B, Douglas RW. Reconstructing regional-scale lead contamination of the atmosphere (1850–1980) in the United Kingdom and Ireland using lake sediments. Global Biogeochemical Cycles. 2004; 18:
- 19. Reynolds GL, Hamilton-Taylor J. The role of planktonic algae in the cycling of Zn and Cu in a productive soft-water lake. Limnology and Oceanography. 1992; 37: 1759–1769.
- 20. Chen CY, Folt CL. High plankton densities reduce mercury biomagnification. Environmental Science and Technology. 2005; 39: 115–121. pmid:15667084
- 21. Wang WX, Dei RCH. Metal stoichiometry in predicting Cd and Cu toxicity to a freshwater green alga Chlamydomonas reinhardtii. Environmental Pollution. 2006; 142: 303–312. pmid:16310914
- 22. Croteau M-N, Luoma SN, Pellet B. Determining metal assimilation efficiency in aquatic invertebrates using enriched stable metal isotope tracers. Aquatic Toxicology. 2007; 83: 116–125. pmid:17467071
- 23. Murray JW. Mechanisms controlling the distribution of trace elements in oceans and lakes. In: Hites RA, Eisenreich S, editors. Sources and Fates of Aquatic Pollutants. Washington DC: American Chemical Society; 1987. pp. 153–184.
- 24. Hilton J. A conceptual framework for predicting the occurrence of sediment focusing and sediment redistribution in small lakes. Limnology and Oceanography. 1985; 30: 1131–1143.
- 25. Douglas RW, Rippey B. The random redistribution of sediment by wind in a lake. Limnology and Oceanography. 2000; 45(3): 686–694.
- 26. Douglas RW, Rippey B, Gibson CE. Estimation of the in-situ settling velocity of particles in lakes using a time series sediment trap. Freshwater Biology. 2003; 48: 512–518.
- 27. Patrick SDT, Monteith T, Jenkins A. UK Acid Waters Monitoring Network: The First Five Years—Analysis and Interpretation of Results, April 1988—March 1993. London: ENSIS Publishing; 1995.
- 28. Juggins S, Flower RJ, Battarbee RW. Palaeolimnological evidence for recent chemical and biological changes in U.K. Acid Waters Monitoring Network sites. Freshwater Biology. 1996; 36(1): 203–219.
- 29. Sykes JM, Lane AMJ, George DG. The United Kingdom Environmental Change Network: Protocols for Standard Measurements at Freshwater Sites. Centre for Ecology and Hydrology, UK, National Environmental Research Council; 1999.
- 30. Monismith SG, MacIntyre S. The surface mixed layer in lakes and reservoirs. Encyclopedia of Inland Waters, ScienceDirect. 2009; 636–650.
- 31. Leppäranta M. Freezing of lakes and the evolution of their ice cover. Berlin: Springer-Verlag; 2015.
- 32. Caddick MX, Baumberg S, Hodgson DA, Phillips-Jones MK. Microbial responses to light and time. Society for General Microbiology, Symposium 56: Cambridge, UK: Cambridge University Press; 1998.
- 33. Baldwin DS,Gigney H, Wilson JS, Watson G, Boulding AN. Drivers of water quality in a large water storage reservoir during a period of extreme drawdown. Water Research. 2008; 42(19): 4711–4724. pmid:18804256
- 34. MacIntyre S, Clark JF, Jellison R, Framb JP. Turbulent mixing induced by nonlinear internal waves in Mono Lake, California. Limnology and Oceanography. 2009; 54(6): 2255–2272.
- 35. Yu M, Sun D, Huang R, Tian W, Shen W, Zhang H, et al. Determination of ultra-trace gold in natural water by graphite furnace atomic absorption spectrophotometry after in situ enrichment with thiol cotton fiber. Analytica Chimica Acta. 2003; 479: 225–231.
- 36. Long SE, Martin TD, Martin ER. Method 200.8- Determination of trace elements in waters and wastes by Inductively Coupled Plasma-Mass Spectrometry. Office of Research and Development, United States Environmental Protection Agency; 1990.
- 37. Olrik K, Blomqvist P, Brettum P, Cronberg G, Eloranta P. Methods for quantitative assessment of phytoplankton in freshwaters, Part I. Stockholm: Swedish Environmental Protection Agency; 1998: Rapport 4860.
- 38. Smith VH. Predictive models for the biomass of blue-green algae in lakes. Water Resources Bulletin. 1985; 21(3): 433–439.
- 39. Hillebrand H, Durselen CD, Kirschtel D, Pollingher U, Zohary T. Biovolume calculation for pelagic and benthic microalgae. Journal of Phycology. 1999; 39: 1145–1159.
- 40. Ho T-Y, Wen L-S, You C-F, Lee D-C. The trace-metal composition of size-fractioned plankton in the South China Sea: biotic versus abiotic sources. Limnology and Oceanography. 2007; 52(5): 1776–1788.
- 41. Bock R. A handbook of decomposition methods in analytical chemistry. International Textbook Company; 1979.
- 42. Boyer JN, Kelble CR, Ortner PB, Rudnick DT. Phytoplankton bloom status: Chlorophyll a biomass as an indicator of water quality condition in the southern estuaries of Florida, USA. Ecological Indicators. 2009; 9(6): S56–S57.
- 43. Heiri O, Lotter AF, Lemcke G. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. Journal of Paleolimnology. 2001; 25: 101–110.
- 44. Arain MB, Kazi TG, Jamali MK, Jalbani N, Afridi HI, Shah A. Total dissolved and bioavailable elements in water and sediment samples and their accumulation in Oreochromis mossambicus of polluted Manchar Lake. Chemosphere. 2008; 70(10): 1845–1856. pmid:17889926
- 45. Johansson K, Bergback B, Tyler G. Impact of atmospheric long range transport of lead, mercury and cadmium on the Swedish forest environment. Water, Air, and Soil Pollution. 2001; Focus 1: 279–297.
- 46. Birks HJB, Line JM, Juggins S, Stevenson AC, Ter Braak CJF. Diatoms and pH reconstruction. Philosophical Transactions of the Royal Society B. 1990; 327: 263–278.
- 47. Bennion H, Juggins S, Anderson NJ. Predicting epilimnetic phosphorus concentrations using an improved diatom-based transfer function and its application to lake eutrophication management. Environmental Science and Technology. 1996; 30: 2004–2007.
- 48. Rippey B, Rose CL, Douglas RW. A model for lead, zinc, and copper in lakes. Limnology and Oceanography. 2004; 46(6): 2256–2264.
- 49. Balistrieri LS, Murray JW, Paul B. The geochemical cycling of trace elements in a biogenic meromictic lake. Geochimica Cosmochimica Acta. 1994; 58: 3993–4008.
- 50. Hart GL, Lowery B. Axial-radial influence of porous cup soil solution samplers in a sandy soil. Soil Science Society of America Journal. 1997; 61(6): 1765–1773.
- 51. Lindström M. Predictive modelling of heavy metals in urban lakes. Ph.D. Thesis, Acta Universitatis Upsaliensis. 2000. Available: http://www.diva-portal.org/smash/get/diva2:166022/FULLTEXT01.pdf.
- 52. Gagnon C, Saulnier I. Distribution and fate of metals in the dispersion plume of a major municipal effluent. Environmental Pollution. 2003; 124: 47–55. pmid:12683982
- 53. Kozelka PB, Bruland KW. Chemical speciation of dissolved Cu, Zn, Cd, Pb in Narragansett Bay, Rhode Island. Marine Chemistry. 1998; 60: 267–282.
- 54. Millero FJ. Speciation of metals in natural water. Geochemical Transactions. 2001; 2(8): 56–64.
- 55. APHA. Standard methods for the examination of water and wastewater, 19th Edition. Washington DC: American Public Health Association; 1995.
- 56. McKergow L, Pritchard M, Elliott AH, Duncan MJ, Senior AK. Storm fine sediment flux from catchment to estuary, Waitetuna—Raglan Harbour, New Zealand. New Zealand Journal of Marine and Freshwater Research. 2010; 44(1): 53–76.
- 57. Danielsson L, Magnusson B, Westerlund S, Zhang K. Trace metals in the Gota River estuary. Estuarine Coastal and Shelf Science. 1983; 17: 73–85.
- 58. Vignati DAL, Dworak T, Ferari B, Koukal B, Loizeau JL, Minouflet M, et al. Assessment of the geochemical role of colloids and their impact on contaminant toxicity in freshwaters: an example from the Lambro-Po system (Italy). Environmental Science and Technology. 2005; 39(2): 489–497. pmid:15707048
- 59. Lindsay WL. Chemical equilibria in soils. New York: John Wiley and Sons; 1979.
- 60. Tack FMG, Callewaert OWJ, Verloo MG. Metal solubility as a function of pH in contaminated, dredged sediment affected by oxidation. Environmental Pollution. 1996; 91(2): 199–208. pmid:15091441
- 61. Kogut MB, Voelker BM. Kinetically inert Cu in coastal waters. Environmental Science and Technology. 2003; 37(3): 509–518. pmid:12630466
- 62. Meylan S, Odzak N, Behra R, Sigg L. Speciation of copper and zinc in natural freshwater: comparison of voltammetric measurements, diffusive gradients in thin films (DGT) and chemical equilibrium models. Analytica Chimica Acta. 2004; 510: 91–100.