Trace metal concentrations in lake and overbank sediments in ...

Research article

Trace metal concentrations in lake and overbank sediments in southern Norway S. Rognerud 7 D. Hongve 7 E. Fjeld 7 R.T. Ottesen

Abstract As, Be, Cd, Co, Cr, Cu, Hg, Ni, Pb, V, Se and Zn concentrations were determined and compared in lake and overbank sediments from 33 catchments without local pollution sources in southern Norway. There were no significant differences in concentrations of Be, Co, Cr, Cu, Ni, and V in overbank and pre-industrial lake sediments. In areas with shallow overburden, and significant influence from long-range atmospheric pollution, concentrations of As, Cd, Hg, Pb, Se, and Zn in overbank sediments were probably modified by vertical percolating water. In such areas, we suggest using lake sediments as a better sampling medium for mapping pre-industrial concentrations. Pre-industrial lake sediments yield natural concentrations of Hg and Se, which consist of both geogenic and natural atmospheric deposition. Important covariables like organic carbon content, Fe oxides, and fine mineral fraction were generally higher in preindustrial lake sediments as compared to overbank sediments. By adjusting for such differences overbank sediments could be used as an alternative in mapping background concentrations of trace metals in regions with few lakes.

Introduction

Natural background concentrations of metals in the environment vary widely both regionally and locally. In some areas, trace metals may reach levels that are harmful to ecosystems (Painter and others 1994). The natural distribution of metals should therefore be included into assessments of the critical loads of metal pollutants to terrestrial and aquatic ecosystems. Geological maps offer little information on the abundance and spatial distribution of trace metals. (Ottesen and others 1989; Davenport 1990). During weathering, mineral constituents are released to natural waters. Some trace metals are adsorbed on surfaces of suspended material or may be incorporated in the lattice of minerals such as Fe and Mn oxides or in organic matter. Particles originating in catchments are deposited in lakes, river banks, and flood plains. Lake sediments have been used to display spatial variation patterns of trace elements in the Canadian shield area (Coker and others 1979; Garrett and others 1990; Kerr and Davenport 1990), and Scandinavia (Johansson 1985; Rognerud and Fjeld 1993), indicating that lake sediments are a useful medium in evaluating the background concentration of metals (Painter and others 1994). Overbank sediments, sediments deposited from suspension on floodplains in stages of overbank flow, have been used as a sampling medium in geochemical mapping in FennoscanKey words Trace metals 7 Lake sediments 7 dia and China (Ottesen and others 1989; Bølviken and Overbank sediments 7 Geochemical mapping others 1990; Shen and Yan 1995; Langedal 1997). Many factors influence natural background concentrations in drainage sediments. Concentrations in pre-industrial lake sediments may be modified by authigenic processes occurring in lake waters and by diagenetic processes within sediments. Pre-industrial overbank sediments may be inReceived: 19 February 1999 7 Accepted: 17 April 1999 fluenced by percolation of soil water and by terrestrial S. Rognerud (Y) organic matter buried during floods. Norwegian Institute for Water Research, Sandvikaveien 41, Recent monitoring of acid atmospheric deposition in N-2312 Ottestad, Norway Norway show obvious trends from low values in the D. Hongve north to high values in the southern and southwestern National Institute of Public Health, P.O. Box 4404 Torshov, regions (Tørseth and Semb 1995). Precipitation monitor0403 Oslo, Norway ing also provides evidence of the long-range transport from continental Europe and Great Britain of various inE. Fjeld Norwegian Institute for Water Research, P.O. Box 173, N-0411, dustrial pollutants, including trace metals (Berg and othOslo, Norway ers 1994). Steinnes and Njåstad (1995) and Berg and Steinnes (1997) have inferred enrichment of metals in R.T. Ottesen mosses and the organic surface layer of natural soils in Geological Survey of Norway, P.O. Box 3006, N-7004, Trondheim, Norway the southernmost Norway to be results of atmospheric

Environmental Geology 39 (7) May 2000 7 Q Springer-Verlag

723

Research article

deposition. The similarity between depositional maps and the distribution pattern for many trace metals in recent lake sediment (Rognerud and Fjeld 1993) and overbank sediments (Øyen and others 1990) is striking. An inevitable question is, therefore, to which degree these sediments mirror natural background concentrations, and to which extent the observed distribution patterns are due to recent atmospheric deposition of pollutants. The objectives in this study were: (i) to examine and compare the concentrations of As, Be, Cd, Co, Cr, Cu, Hg, Ni, Pb, Se, V and Zn in recent and pre-industrial sediments of lakes with overbank sediments, sampled in the same catchments in southern Norway; (ii) to indicate the environmental factors influencing concentrations in these two sampling media; and (iii) to evaluate the use of lake and overbank sediments in regional geochemical mapping.

sands may also occur in the bottom of the valleys. The vegetation consists of pristine alpine or forest areas with no, or scarce influences from agriculture. In Table 1 summary statistics of morphometric and hydrological characteristics of the sampling sites are given. Lake water Water samples were collected from 1 m depth in the center of each lake. Chemical water quality variables were measured (Table 1). In general, the lakes were oligotrophic, slightly acidic to neutral with soft water, and low to medium in dissolved organic carbon (DOC). The concentration of chloride is an indication of the influence of the atmospheric deposition of seasalts and trace metals originating in the marine environment. The chloride concentrations indicate a broad range of influence from seasalts.

Lake sediments We used a modified Kajak-Brinkhurst (KB) gravity corer (Mudroch and Azcue 1995). The closing of the valve by a Materials and methods messenger has been changed to a thin, flexible membrane which closes automatically when the water flow through We sampled 33 catchments in southern Norway selected the tube stops. In the open position it ensures an unresfrom a previous national survey of overbank sediments tricted water flow through the corer during lowering and (Øyen and others 1990). The sampling sites cover a wide penetration. The 75-cm-long polyacrylic tubes (44-mm range of conditions regarding bedrock geology, thickness inner diameter) were sharpened at the lower end. The of overburden, amount of precipitation, and atmospheric function of the corer has been carefully tested in sedidepositions of acids and metals in modern times (Fig. 1). ments with low to high concentration of organic matter. Each catchment has a lake downstream of the site of A portable echosounder was used to locate the coring site sampled overbank sediment. The river valleys in the east- of the deepest part of the lake or at an appropriate subern part of the sampled areas are covered by till, glaciobasin. Only cores with apparently undisturbed sedimentfluvial gravels and sands. The western parts have a thin water interfaces were accepted. The core sampler was overburden with many outcrops especially in areas with a usually lowered slowly to the bottom, but in a few lakes steep relief. However, glaciofluvial and fluvial gravels and with exceptionally firm sediments it was necessary to drop the sampler from a distance above the bottom to obtain cores representative for pre-industrial sediment samples. The cores were 30–50 cm long. Fig. 1 The cores were extruded and sectioned on site to minimLocation of the study sites (np33) showing (A) bedrock dominated by minerals resistant to chemical weathering, (B) pH ize disturbance of the flocculent surface sediments. Each sample was homogenized and freeze-dried before further in precipitation, and (C) concentrations (mg/g dry weight) of analyses. The results discussed in this paper are from lead in terrestrial mosses. Maps redrawn from Nordic Council of Ministers (1993) subsamples from the 0–1-cm slices (surface sediment)

724


Research article

Table 1 Summary statistics of lake specific data and some chemical variables in lake water. SE standard error of the mean (np33) Variable

Unit

Mean

SE

Median

Range

Lake surface Lake depth Catchment area Water retention Mean annual runoff pH DOC Chloride

km 2 m km 2 year l s –1 km –1

6.8 43 324 0.47 46.2 5.6 2.0 3.4

2.0 4.2 54.8 0.13 4.2 0.11 0.2 0.4

1.6 38 200 0.17 42.5 5.6 1.7 2.8

0.4–45.1 11–125 11–1308 0.01–3.0 13–110 4.8–7.1 0.4–4.3 0.7–10.2

mg C/L mg Cl/L

and homogenized subsamples of a 5-cm slice from the deepest, pre-industrial part of the core, here called preindustrial sediment, representing spans of 100–400 years of sedimentation. (See forthcoming section for the dating procedure.) They will therefore have been deposited over a long time period, which also most probably would have covered periods with floods that may have produced overbank sediments. A contamination factor (Cf) is defined as the ratio of the concentration in surface lake sediment to pre-industrial sediment concentration.

only 27 samples could be analyzed. Pb was included in the second series as a control of recovery, and there was no significant differences between these series. Thus, substitute values from the originally NGU analysis of Pb could be used for the four missing samples. The empirical ratio between loss on ignition (LOI) and organic carbon (OC) in the other samples were used to estimate OC values in the four missing samples. The age of the pre-industrial sediments were estimated in a subset of ten lakes using the 14C-method. The subset was carefully selected from lake size, depth, catchment area, annual runoff and loose deposits, to be sure to inOverbank sediments clude the lakes with the highest sedimentation rates. DatOverbank sediments are produced when major floods oc- ing was done by the Radiological Dating Laboratory, The cur in a river system and new sediment sources open. A Norwegian University of Science and Technology, Trondvertical section through overbank sediments reflects the heim. Only one dating was made for each sediment core history of sedimentation over time, and a composite and the assessed age is the mean value for the 5-cm slice sample of such a section will give an integrated picture of of assumed pre-industrial sediment. The correct historic the chemical and mineralogical conditions from a large average age for the youngest samples was, with 68% number of sediment sources opened during floods (Ottes- probability, between AD 1405 and 1500 while the oldest en and others 1989). It is difficult, however, along instawas from BC 810–535. Episodes leading to sudden ble river reaches where the river reworks and mixes the changes in sedimentation rate can, almost invariably, be alluvial sediment, to detect the true pre-industrial conidentified by visual inspection of the cores. Since all our centrations in the bottom parts of the overbank sediment cores had a homogeneous appearance we assume even profiles (Macklin and others 1994; DeVos and others sedimentation rates. The slice thickness is without impor1996; Swennen and others 1998). tance for the requirement that they should originate beThe overbank sediment samples in this study were colfore the Industrial Revolution (before AD 1750). lected at distances of 2–200 m from present-day streams. The pH values were measured using a Radiometer PHM A vertical section through the sediment was cut with a 63 equipped with a combined electrode GK2401 C. DOC spade. Excluding the upper 10 cm the sample was taken and chloride were determined with a Technicon AutoAfrom the rest of the profile. The sediments consisted of nalyzer II, using Industrial method 451–76 W (on-line dimostly coarse-grained material but, unfortunately, the gestion with persulfate and UV irradiation) and 99–70 W particle size distribution was not measured. The data (mercury thiocyanate method) respectively. used here are part of a large national survey of overbank Statistical methods sediments carried out by the Norwegian Geological SurThe multivariate relations between the trace elements and vey (NGU), Trondheim. the explanatory variables (Al, Fe, and OC) were analyzed Analytical methods by redundancy analysis (RDA). RDA is a constrained The overbank sediments were dried and sieved to obtain form of multivariate multiple regression. It is intermethe less than 0.063-mm fraction, which was analyzed at diate between principal component analysis (PCA) and NGU for Al, Be, Co, Cr, Cu, Fe, Mn, Ni, Pb, Se, V and Zn separate multiple regressions of each of the dependent using the analytical methods as for lake sediments (ICPvariables. RDA is a constrained form in that the underlyAES). Pb, Cd and As were later analyzed in 29 out of the ing dimensions in the Y-matrix (latent variables) are 33 samples (no material was left over in four samples) forced to be multiple linear regressions of the indepenusing the same methods as for lake sediments. For Hg, dent variables (the Z-matrix). For a detailed treatment on


725

Research article

In the lake sediments, the concentrations were significantly higher for As, Cd, Cu, Hg, Pb, Se, and Zn in the surface sediments than in the pre-industrial sediments (Fig. 2). No such differences were found for the covariables OC, Fe, and Al (Fig. 2, Table 2). The contamination factor was highest for Pb and decreased in the order As, Hg, Cd, Cu, Zn, and Se. However, the absolute difference in concentrations (mg g –1 d.w.) between surface and preindustrial sediments generally decreased in the order: Pb (107), Zn (68), Cu (13), As (5), Se (0.7), Cd (0.4), and Hg (0.15). The concentration of OC was an important explanatory variable for concentrations of As, Cd, Hg, Pb, and Se, except for As, Cd, and Pb in the pre-industrial sediments (Fig. 3). The slope of the regressions were, generally, similar in surface lake sediments and the overbank sediments, although the range in OC concentrations was twice as wide in surface lake sediments. There was no correlation between OC and the elements, As, Cd, and Pb, in the pre industrial sediments. Only Hg and Se were positively correlated to OC in both modern and pre-inResults dustrial sediments (Fig. 3). The other trace elements were positively associated to Fe and Al concentrations. Generally, there were systematically higher concentrations In the RDA, Fe and Al formed two correlated gradients in pre-industrial lake sediments than in the overbank se- (Fig. 4). OC formed a third gradient, correlated to the Al diments for Cd, Se, and Zn, but significantly lower for As and Fe gradients in surface lake sediments, but uncorreand Pb (Table 2). The two media differed in their conlated to Fe in pre-industrial lake sediments and Al in centration of important scavenging factors OC, Fe, and overbank sediments. The eigenvectors indicated that the Al. These covariates were significantly higher in the lake first RDA axis in the ordination diagram for overbank-, sediments than in overbank sediments (Table 2). Thus surface- and pre-industrial-sediments accounted for 68%, the strength of the association to these covariables can be 69% and 63%, of the variation in the data set, respectivean important factor explaining the difference in concenly. The second RDA axis accounted for 26%, 27%, and trations of trace metals. 22%. OC was the variable with greatest loading on the RDA, see Ter Braak and Prentice (1988). The RDA was done with an algorithm from Ter Braak and Prentice (1988), implemented in MATLAB 5.2 (MathWorks, Natic, MA). RDA leads to an ordination diagram that simultaneously displays the main relationships of the dependent variables (the Y-matrix) with the independent variables (the Z-matrix). In the ordination diagram dependent variables are indicated by points, whereas the independent variables are indicated by arrows. The variables represent different gradients in the data set. The variables pointing in roughly the same direction indicate that they are positively correlated; variables crossing at right angles indicate near-zero correlation, and variables pointing in opposite directions indicate high negative correlation. The length of the variables (the arrows or distances from origin to the points) are proportional to the rate of change in their direction, and indicate the importance of the variables in the analysis.

Table 2 Summary statistics of variables in the different sediment media. All concentrations are given as mg g –1 dry weight, except Fe, Al, Mn and OC (organic carbon) which are given in % dry weight.

726

Depth (cm)

Surface sediments 0–1

Variable

Mean

SE

Al As Be Cd Co Cr Cu Fe Hg Mn Ni OC Pb Se V Zn

2.12 7.81 1.87 0.70 10.1 20.5 32.3 5.37 0.24 0.24 13.5 14.1 131 2.70 65.9 155

0.81 1.08 0.27 0.11 1.35 1.7 2.6 0.50 0.025 0.16 1.3 1.1 15.2 0.25 9.4 18.1

Sections of lake sediments studied are surface sediment (0–1 cm) and pre-industrial sediment (30–50 cm). SE standard error of the mean; np33, except for As, Cd, OC (np29) and Hg (np27) in overbank sediments

Pre-industrial sediment 30–50

Overbank sediment 10–rest of profile

Range

Mean

SE

Range

Mean

SE

Range

0.81–4.27 0.48–32.7 0.9–7.27 0.06–3.12 1.4–27.7 8.1–55.5 7.0–69.2 1.5–12.9 0.03–0.60 0.01–5.54 4.9–39.1 1.6–27.2 8.9–412 0.37–6.25 17.7–348 37.6–481

2.31 2.56 1.92 0.26 11.0 17.5 19.5 4.80 0.087 0.11 11.1 15.6 24.5 1.90 49.5 87.0

0.11 0.47 0.23 0.03 1.4 2.0 2.1 0.57 0.008 0.05 1.1 1.1 2.9 0.17 3.4 8.2

1.45–3.95 0.41–12.0 0.9–6.2 0.11–0.66 1.3–31.0 4.77–53.1 7.60–55.1 1.5–14.8 0.03–0.25 0.01–1.43 3.2–27.2 4.0–29.1 8.0–76.0 0.61–4.65 19.6–107 38.6–206

1.6 3.09 1.31 0.15 9.4 18.2 17.2 2.6 0.065 0.032 10.2 4.5 34.8 0.66 41.5 49.4

0.11 0.53 0.12 0.02 0.9 1.9 1.8 0.2 0.011 0.006 1.2 0.7 5.3 0.12 3.0 6.2

0.5–3.0 0.47–12.5 0.4–3.1 0.04–0.53 3.6–24.3 3.5–56.1 4.4–47.7 0.6–4.9 0.014–0.25 0.01–0.19 2.1–28.9 0.6–14.8 8.9–115.5 0.1–2.8 9.0–79.6 11.5–186


Research article

pre-industrial sediments, although V and Be loaded weakly on the OC gradient.

Discussion

Fig. 2 Box-and-whisker plot of contamination factors (Cf) of trace elements, showing medians, upper and lower quartiles, and the 10th and 90th percentiles [Cfpthe ratio between concentrations in surface (0–1 cm)] and pre-industrial sediments (30–50 cm); np33, except for As and Cd (np29) and Hg (np27) in overbank sediments

first RDA axis in the ordination diagram of the overbank sediments, whereas Al was the variable with the greatest loading on the second RDA axis. In surface lake sediments Al and OC loaded equally on the first RDA axis, whereas Al was the variable with the greatest positive loading on the second RDA axis. Fe and Al were nearly equal, and the greatest loading on the first RDA axis in pre-industrial sediments. All independent (explanatory) variables loaded negatively on the second RDA axis, but OC was the most important variable. Among the dependent variables As, Cd, Hg, Pb, and Se formed a coherent group in the ordination diagrams of the overbank sediments and the surface lake sediments. This group was primarily associated with the OC gradient. The trace elements Be, Co, Cr, Cu, Ni, V, and Zn formed another coherent group primarily associated with the Fe and Al gradient. In the ordination diagram of the pre-industrial sediments Hg and Se in the first coherent group were still associated to the OC gradient, but As and Cd were associated to the Al and Fe gradient, whereas Pb was not associated with any of the gradients (lying close to the center). The other coherent group was generally still associated to the Al and Fe gradient also in the

The overbank sediments in Fennoscandia are derived mostly from till, a material which commonly has moved rather short distances (some hundred meters) away from the parent bedrock (Ottesen and others 1989). Thus the overbank sediments, as well as the minerals of the preindustrial lake sediments should be derived from the local bedrock. However, by comparing trace metal concentrations in lake and overbank sediments from the same catchments, it seems as metals originating from longrange atmospheric transport have influenced the overbank sediments in the southernmost areas. A probable mechanism for this is complexation with dissolved humic substances from the O-horizon and precipitation of the organic metal complexes deeper in the mineral soil. There were also indications that the acidity of the percolating water may have impoverished the minerogenic reservoir of Cd and Zn in the overbank sediments. Concentrations of Cu, Cr, Ni, V, Be, and Co were insignificantly different in overbank and pre-industrial lake sediments and strongly associated with the minerogenic part of the sediments. The influence of Al, Fe and OC The statistical analysis indicated that the concentrations of Al, Fe and OC were important covariables for trace metal concentrations in sediment media. This is consistent with other studies emphasizing the importance of these variables as scavenging agents in lakes (Fjeld and others 1994; Rognerud and others 1998). The positive association of As, Cd, Hg, Pb, and Se, to the OC gradient in the ordination diagrams from overbank and surface lake sediments reflectes a generally high affinity to organic ligands for these elements (Jackson and others 1980; Santschi 1988). The minerogenic part of profundal sediments in oligotrophic lakes are typically dominated by fine grained particles (less than 0.063 mm) consisting of Al-silicate minerals (Håkanson and Jansson 1983). Generally, Al concentrations in sediments increase with decreasing particle size (Salomons and Förstner 1984). Thus, Al can be used as an indicator variable for the abundance of fine-grained particles in the sediments (de Groot and others 1982; Mudroch and Azcue 1995). The positive association of Be, Co, Cr, Cu, Ni, and Zn to the Al gradient from the ordination diagrams is consistent with the observation that Ni, Zn, and Cu can substitute for Al in the alumino-silicates, or they all can be adsorbed on surfaces or between layers of the alumino-silicates (Loring 1976; Sparks 1995). The statistical analysis showed that concentration differences of covariables can have a significant influence on trace metal concentrations. Generally, the concentrations of covariables were higher in lake sediments than overbank sediments which emphasize that, under the prevail-


727

Research article

Fig. 3 Correlation between trace elements and organic carbon (OC) in overbank sediments, surface lake sediments (0–1 cm) and pre-industrial lake sediments (30–50 cm) for those elements correlated to organic carbon in lake surface sediments

728


ing hydrological conditions, lakes are more efficient traps than overbank sediments. The higher Al concentrations in lake sediments indicate a generally higher concentrations of fine particles (in the less than 0.063-mm fraction). The increased mobility of Al observed in acidic catchments can decrease the concentration in upper soil and increase the sedimentation of Al hydroxides in lakes (White and Driscoll 1985).

Research article

Fig. 4 Ordination diagram based on redundancy analysis of trace metal concentrations in surface lake sediments, pre-industrial lake sediments, and overbank sediments with respect to concentrations of organic carbon (OC), Al and Fe (arrows)

However, we observed no significant increase in Al concentrations up-core, indicating a moderate, if any, effect of recent Al hydroxide accumulation. Although the concentrations of organic matter in pre-industrial overbank sediments are commonly low (Ottesen and others 1989), we observed significant amounts of OC in many of our overbank sediments (Table 2). This can be derived from plant fragments buried in the sediment or subsequent precipitation of humic substances originating in surface humic layers. Three pieces of evidence are consistent with the argument that OC in the overbank sediments has been supplied from the O-horizon: (i) microscopic examination did not reveal any distinct remains of plant cells; (ii) the high correlation between OC and Fe in overbank sediments indicate that samples may have been collected from B horizons where precipitation of Fe with adsorbed humic compounds is common; (iii) the OC concentration was highest in overbank sediments from the southernmost part of Norway. Here, the overbank sediments were generally thin (20–50 cm) and contained small fractions of fine-grained particles (T. Volden, NGU, pers. comm.). Under such circumstances humic substances and contaminants can easily be transported with percolating water (McCarthy and Zachara 1989). Differences between OC and Al concentrations in surface and pre-industrial sediments were insignificant. This is consistent with the observation that degradation of organic matter in consolidated lake sediments is a very slow process (Louchouarn and others 1993), and grain size distribution is mainly a function of stable lake properties (Rowan and others 1992). Oxidation of redox sensitive elements like Fe and Mn may cause a slight enrichment in surface sediments. We assume the effect of bioturbation in profundal sediments of these oligotrophic forest and subalpine lakes to be small as this zone is inhabited by a scarce society of mostly small chironomids, pisidia and oligochaets (Brundin 1949). Fluxes of metals in pore water are generally towards the lower concentrations in lake water (Stumm and Schnoor 1995). It is therefore reasonable to assume that the pre-industrial sediments were not significantly influenced from modern contaminants by mechanical or diffusional transport. Concentrations of trace metals in lake sediments The Cf values indicate a significant modern atmospheric deposition of As, Cd, Hg, Pb, and Se, which has also been measured in terrestrial mosses from the same region (Berg and Steinnes 1997). There was a pronounced shift in the association between the trace metals and covariates from reference to surface lake sediments. As, Cd, and Pb in the pre-industrial sediments were associated to the minerogenic constituents (Al, Fe), whereas in surface sediments they were associated with OC. Naturally occur-


729

Research article

to soil components. For Pb, this is in concordance with other studies in Sweden (Bergqvist and others 1989), Germany (Heinrichs and Mayer 1977) and the United States (Siccama and others 1980). As and Pb is strongly bound to dissolved organic matter even under acidic conditions (Bergkvist and others 1989; Kabata-Pendias and Pendias 1984), and both elements are known to migrate vertically in well-drained soils (Kabata-Pendias and Pendias 1984; Turner and others 1985; Friedland and others 1992; Driscoll and others 1994). Our interpretation is consistent with the results of Eden and Bjørklund (1994) that overbank sediments in southern parts of Fennoscandia have been enriched by Pb through atmospheric deposition, and with showing that concentrations of Pb in surface soil have not changed over the last decade in southern Norway in spite of a significant atmospheric deposition (Berthelsen and others 1995) and that the B-horizon was slightly enhanced by Pb-humus complexes in the same region (Steinnes and Njåstad 1995). Lead was located near the center of the ordination diagram from the pre-industrial sediments indicating a weak association to other covariates. Lead occurs in solid solution replacing K and Ca in minerals and on sorption sites, and it is the least mobile among trace elements in natural soils (Kabata-Pendias and Pendias 1984). We are confident that the observed Pb concentrations in pre-industrial lake sediments reflect concentrations close to equilibrium with bedrock background concentrations. In contrast to Pb, As loaded strongly on the Fe gradient as observed in many geochemical surveys (Davenport and others 1993). Arsenate, the common form of As, sorbs strongly to Fe hydroxide phases (Hamilton-Taylor and Davison 1995). The Cd and Zn concentrations were significantly lower in the overbank sediment than in lake pre-industrial sediments. These elements are mobilized in acidified soils (Bergqvist and others 1989), and may drain to the streams. Elevated concentrations have been observed in lakes from the most acidified regions of our study (Skjelkvåle and others 1996) as well as an acidification of groundwater (Henriksen and Kirkhusmo 1986). The chloride concentrations of the lakes indicate that some catchments were significantly influenced by deposition of seasalts. Zn can be removed from soils because it readily exchange with sea-salt cations (Steinnes and Njåstad 1995). The ordination diagrams demonstrated the close associaComparison of trace metal concentrations in tion between Zn and the minerogenic fraction. Thus, we overbank sediments and pre-industrial lake assume that the percolation of acid water (and water ensediments riched with seasalt at some locations) may have impoverThe concentration of important scavenging agents (Al, Fe ished the minerogenic Zn reservoir in the overbank sediand OC) were generally higher in lake sediments than in ments. This could also be the case for Cd, but the close overbank sediments. Thus, it was unusual that the highassociation between Cd and OC indicates a slight conest concentrations of As and Pb occured in the overbank tamination of the organic fraction in overbank sediments sediments. However, our data is consistent with the hyby atmospheric deposition. A pH driven translocation of pothesis that acidic, humic waters must have percolated Cd and Zn has recently been observed in Norway (Berthrough the overbank sediments at some locations and thelsen and others 1995) and Germany (Kalbitz and facilitated migration of As and Pb from the contaminated Wennrich 1998). forest floor into the underlying mineral soil horizon, The concentrations of Be, Co, Cu, Cr, and Ni in these two where these elements and humic substances are adsorbed media were not significantly different. In the ordination ring Cd and Pb are commonly associated with fine grained Fe-Al-silicates (Loring 1976; Salomons and Förstner 1984), whereas As, Cd, and Pb deposited from the atmosphere are associated with OC in the catchments (Bergqvist and others 1989) and in runoff water (Johansson and Iverfeldt 1994; Borg and Johansson 1989). Thus, these references support our finding that Pb, As and Cd in our pre-industrial sediments were mainly dependent on the geochemistry in the catchments. However, Pb has a long pollution history (Renberg and others 1994), but many 210Pb dated sediment studies indicate that the most significant Pb deposition from the atmosphere is past 1750–1850 (summarized by Norton and Kahl 1992). In contrast to As, Cd and Pb, Hg and Se were closely associated with OC in pre-industrial sediments. Hg occurs naturally as gaseous Hg, which allow Hg to be widely circulated in the atmosphere (Slemr and Langer 1992). A significant part of the atmospheric Hg concentration is natural, and accumulation rates of geogenic Hg are generally low compared to atmospheric-derived Hg even in pre-industrial sediments (Nater and Grigal 1992; Swain and others 1992). Thus, the close association between Hg and OC in both surface and pre-industrial sediments strongly indicates that natural atmospheric deposition was a significant source for Hg concentrations in pre-industrial times. The marine environment is the most significant source of Se in Norwegian forest soils (Berg and Steinnes 1997), due to natural methylation and atmospheric transport (Mosher and Duce 1987). The deposition of natural organic Se-compounds is reflected in a strong correlation between Se and soil OC (Gustafsson and Johansson 1992). Thus, as was the case for Hg, we suggest that the close association between Se and OC in surface and pre-industrial sediments reflects the natural atmospheric deposition of Se. In the ordination diagrams, Co, Cr, and Ni were associated with the Al and Fe gradient in the pre-industrial sediments as well as in surface sediments. These elements had low Cf-values indicating low or insignificant anthropogenic supply. The marine environment is not a significant atmospheric source for these elements and they are not naturally methylated (Pacyna 1995). Thus, grain size (Al content) and metal concentration in parent material are the most important explanatory variables for sediment concentrations of these elements.

730


Research article

diagrams they loaded strongly on the Al and Fe gradients, indicating associations with the inorganic sediment fraction. Generally, these elements have similar concentrations in till and overbank sediments (Eden and Bjørklund 1994), indicating associations with particles characteristic for the catchments geochemistry. There were no significant differences in V concentrations between the media. In the ordination diagrams, V was located between the OC gradient and the Al-Fe gradient, but with low loadings on both gradients. This reflects weak associations to organic matter and fine particles. We assume that the association to OC reflects that V is an important plant nutrient (Kabata-Pendias and Pendias 1984). The weak association to Al and Fe probably reflects that V is enriched in fine-grained particles derived from Fe-rich mafic rocks (Kabata-Pendias and Pendias 1984).

Summary Concentrations of As, Cd, Hg, Pb, Se, and Zn in pristine forest and subalpine overbank sediments in southern Norway seem to be modified by percolation of surface water influenced by polluted atmospheric deposition. Although, important covariates like OC and Al were generally higher in pre-industrial lake sediments, overbank sediments had higher concentrations of As, Hg and Pb, which have significant sources in long-range atmospheric transport. Percolation of acid water is the probable reason why the minerogenic reservoirs of Cd and Zn in overbank sediments have been impoverished compared with pre-industrial lake sediments. Hg and Se were the only elements strongly associated to OC in pre-industrial lake sediments. Their principal source is natural atmospheric deposition. Thus, pre-industrial lake sediments reflect natural conditions, but overestimate geogenic concentrations. Concentrations of Be, Co, Cr, Cu, Ni, and V in overbank sediments and pre-industrial lake sediments were almost identical, although concentrations of important scavengers like Fe oxides and Al were generally higher in lake sediments. The indication that acidic, humic surface water has percolated through the overbank sediments, did not affect concentrations of these elements significantly, probably due to insignificant anthropogenic atmospheric depositions and a strong integration of each metal within alumino-silica minerals or strongly adsorbed to minerals. Generally, overbank sediments and pre-industrial lake sediments are both suitable media in mapping background concentrations of trace metals. However, in areas with shallow overburden and significant modern deposition of acids and trace elements, we suggest pre-industrial lake sediments are appropriate. Acknowledgements This study was financed by the Norwegian State Pollution Control Authority and Directorate for Nature Management. We thank B. Bølviken and T. Volden at the Geological Survey of Norway for access to the overbank samples, professor M. Abdullah at the University of Oslo for help with

organic carbon determinations, and to T. Andersen, Norwegian Institute for Water Research, for implementation of RDA in MATLAB. We also thank professor Steve Norton, Maine University, for helpful comments to the manuscript.

References Berg T, Steinnes E (1997) Use of mosses (Hylocomium splendense and Pleurozium schreberi) as biomonitors of heavy metal deposition: from relative to absolute deposition values. Environ Pollut 98 : 61–72 Berg T, Røyset O, Steinnes E (1994) Principal component analysis of data for trace elements and main components in precipitation falling on Norway. Environ Monit Assessm 31 : 259–273 Bergqvist B, Folkeson L, Berggren D (1989) Fluxes of Cu, Zn, Pb, Cd, Cr and Ni in temperate forest ecosystems. A literature review. Water Air Soil Pollut 47 : 217–289 Berthelsen BO, Steinnes E, Solberg W, Jingsen L (1995) Heavy metal concentrations in plants in relation to atmospheric heavy metal deposition. J Environ Qual 24 : 1018–1026 Bølviken B, Kullerud G, Loucks RR (1990) Geochemical and metallogenic provinces:a discussion initiated by results from geochemical mapping across northern Fennoscandia. J Geochem Explor 39 : 49–90 Borg H, Johansson K (1989) Metal fluxes to Swedish forest lakes. Water Air Soil Pollut 47 : 427–441 Brundin L (1949) Chironomiden und andere Bodentiere der südschwedischen Urbirgsseen. Inst of Freshwater Res Drottningholm Rep 30 Coker WB, Hornbrook EHW, Cameron EM (1979) Lake sediment geochemistry applied to mineral exploration. In: Hood PJ (ed) Geophysics and geochemistry in the search for metallic ores. Geol Surv Can Econ Geol Rep 31 : 435–478 Davenport PH (1990) A comparison of regional geochemical data from lake and streams in northern Labrador; implications for mixed-media geochemical mapping. J Geochem Explor 39 : 117–151 Davenport PH, Christopher TK, Vardy S, Nolan LW (1993) Geochemical mapping in Newfoundland and Labrador: its role in establishing geochemical baselines for the measurement of environmental change. J Geochem Explor 49 : 177–200 de Groot AJ, Zschuppe KH, Salomons W (1982) Standardization of methods of analysis for heavy metals in sediments. Hydrobiologia 92 : 689–695 DeVos W, Ebbing J, Hindel R, Schalich J, Swennen R, Van Keer I (1996) Geochemical mapping based on overbank sediments in the heavily industrialized border area of Belgium, Germany and The Netherlands. J Geochem Explor 56 : 91–104 Driscoll CT, Iverfeldt A, Otton JK (1994) Trace metal speciation and cycling. In: Cerny J, Moldan B (eds) Biogeochemistry of small catchments. J Wiley, New York, pp 299–322 Eden P, Bjørklund A (1994) Ultra-low density sampling of overbank sediment in Fennoscandia. J Geochem Explor 51 : 265–289 Fjeld E, Rognerud S, Steinnes E (1994) Influence of environmental factors on heavy metal concentration in lake sediments in southern Norway indicated by path analysis. Can J Fish Aquat Sci 51 : 1708–1720 Friedland AJ, Craig BW, Miller EK, Herrick, GT, Siccama TG, Johnson AH (1992) Decreasing lead levels in the forest floor of the northeastern USA. Ambio 21 : 400–403 Garrett RG, Banville RMP, Adcock SW (1990) Regional geochemical data compilation and map preparation, Labrador, Canada. In: Darnley AG, Garrett RG (eds) International geochemical mapping. J Geochem Explor 39 : 91–116


731

Research article

Gustafsson JP, Johansson L (1992) Selenium retention in the organic matter of Swedish forest soils. J Soil Sci 43 : 461–472 Håkanson L, Jansson M (1983) Principles of lake sedimentology. Springer, Berlin Heidelberg New York Hamilton-Taylor J, Davison W (1995) Redox-driven cycling of trace elements in lakes. In: Lerman A, Imboden DM, Gat JR (eds) Physics and chemistry of lakes. Springer, Berlin Heidelberg NewYork, pp 217–263 Heinrichs H, Mayer R (1977) Distribution and cycling of major and trace metals in two central European forest ecosystems. J Environ Qual 9 : 111–118 Henriksen A, Kirkhusmo LA (1986) Water chemistry of acidified aquifers in southern Norway. Water Qual Bull 11 : 1–6 Jackson TA, Kipphut G, Hesslein RH, Schindler DW (1980) Experimental study of trace metal chemistry in softwater lakes at different pH levels. Can J Fish Aquat Sci 37 : 387–402 Johansson K (1985) Mercury in sediments in Swedish forest lakes. Verh Int Ver Theor Angew Limnol 22 : 2359–2363 Johansson K, Iverfeldt Å (1994) The relation between mercury content in soil and the transport of mercury from small catchments in Sweden. In: Watras CJ, Huckabee JW (eds) Mercury pollution. Integration and Synthesis. Lewis Publ, Boca Raton, pp 323–328 Kabata-Pendias A, Pendias H (1984) Trace elements in soils and plants. CRS Press, Boca Raton Kalbitz R, Wennrich R (1998) Mobilization of heavy metals and arsenic in polluted wetland soils and its dependence on dissolved organic matter. Sci Total Environ 209 : 27–39 Kerr A, Davenport PH (1990) Application of geochemical mapping techniques to a complex Precambrian shield area in Labrador, Canada. J Geochem Explor 39 : 225–247 Langedal M (1997) Dispersion of tailings in the Knaben-Kvina drainage basin, Norway, 1 : evaluation of overbank sediments as sampling medium for regional geochemical mapping. J Geochem Explor 58 : 157–172 Loring D (1976) The distribution and partition of zinc, copper and lead in the sediments of Saguenay fjord. Can J Earth Sci 13 : 960–971 Louchouarn P, Lucotte M, Mucci A, Pichet P (1993) Geochemistry of mercury in two hydroelectric reservoirs in Quebec, Canada. Can J Fish Aquat Sci 50 : 269–281 Macklin MG, Ridgway J, Passmore DG, Rumsby BT (1994) The use of overbank sediments for geochemical mapping and contamination assessment: results from selected English and Welsh floodplaines. Appl Geochem 9 : 689–700 McCarthy JF, Zachara JM (1989) Subsurface transport of contaminants. Environ Sci Technol 23 : 496–502 Mosher BW, Duce RA (1987) A global atmospheric selenium budget. J Geophys Res 92 : 13289–13298 Mudroch A, Azcue JM (1995) Manual of aquatic sediment sampling. Lewis Publishers, London Nater EA, Grigal DF (1992) Regional trends in mercury distribution across the Great Lakes states, north central USA. Nature 358 : 139–141 Nordic Council of Ministers (1993) Nordic environmentcondition, development and threats (in norwegian). Report NORD 11. Copenhagen Norton SA, Kahl JS (1992) Paleolimnological evidence of metal pollution from atmospheric deposition. In: Verry ES, Vermette SJ (eds) 1991 Oct 8. Philadelphia, PA. Gen Tech Rep NC-150. US Dept of Agricul, Forest Service, North Central Forest Experiment Station, St. Paul, MN Ottesen RT, Bogen J, Bølviken B, Volden T (1989) Overbank sediment: a representative sample medium for regional geochemical mapping. J Geochem Explor 31 : 257–277

732


Øyen Ø, Bølviken B, Nilsen T (1990) Geochemical characteristics of norwegian communities using data from overbank sediments (in norwegian). Norw Geol Surv, Trondheim, Rep 90–015 Pacyna JM (1995) The origin of Artic air pollutants: lessons learned and future research. Sci Total Environ 160/161 : 39–53 Painter S, Cameron EM, Allan R, Rouse J (1994) Reconnaissance geochemistry and its environmental relevance. J Geochem Explor 51 : 213–246 Renberg I, Persson MW, Emteryd O (1994) Pre-industrial atmospheric lead contamination detected in Swedish lake sediments. Nature 368 : 323–326 Rognerud S, Fjeld E (1993) Regional survey of heavy metals in lake sediments in Norway. Ambio 22 : 206–212 Rognerud S, Skotvold T, Fjeld E, Norton SA, Hobæk A (1998) Concentrations of trace elements in recent and pre-industrial sediments from Norwegian and Russian Arctic lakes. Can J Fish Aquat Sci 55 : 1512–1523 Rowan BJ, Kalff J, Rasmussen JB (1992) Profundal sediment organic content and physical character do not reflect lake trophic status, but rather reflect inorganic sedimentation and exposure. Can J Fish Aquat Sci 49 : 1431–1438 Salomons W, Förstner U (1984) Metals in the Hydrocycle. Springer, Berlin Heidelberg New York Santschi PH (1988) Factors controlling the biogeochemical cycles of trace elements in fresh and coastal waters as revealed by artificial radioisotopes. Limnol Oceanogr 33 : 848–866 Shen X, Yan M (1995) Representativity of wide-spaced lowerlayer overbank sediment geochemical sampling. J Geochem Explor 55 : 231–248 Siccama TG, Smith WH, Mader DL (1980) Changes in lead, zinc, copper, dry weight, and organic matter content of the forest floor of white pine stands in central Massachusetts over 16 years. Environ Sci Technol 14 : 54–56 Skjelkvåle BL, Henriksen A, Vadset M, Røyset O (1996) Trace elements in norwegian lakes – Preliminary results from 473 lakes. Norwegian Institute for Water Research Rep 3457–96 Slemr F, Langer E (1992) Increase in global atmospheric concentrations of mercury inferred from measurements over the Atlantic Ocean. Nature 355 : 434–437 Sparks DL (1995) Environmental soil chemistry. Academic press, New York Steinnes E, Njåstad O (1995) Enrichment of metals in the organic surface layer of natural soil: identification of contributions from different sources. Analyst 120 : 1479–1483 Stumm W, Schnoor J (1995) Atmospheric depositions: Impact of acids on lakes. In: Lerman A, Imboden DM, Gat JR (eds) Physics and chemistry of lakes. Springer, Berlin Heidelberg New York, pp 185–215 Swain BE, Engstrom DR, Brigham ME, Henning TA, Brezonik PL (1992) Increasing rates of atmospheric deposition in the Midcontinental North America. Science 257 : 784–786 Swennen R, van der Sluys J, Hindel R, Brusselmann A (1998) Geochemistry of overbank and high order stream sediments in Belgium and Luxembourg: a way to assess environmental pollution. J Geochem Explore 62 : 97–70 Ter Braak CJF, Prentice IC (1988) A theory of gradient analysis. Adv Ecol Res 18 : 271–317 Tørseth T, Semb A (1995) Sulphur and nitrogen deposition in Norway, status and trends. Water Air soil Pollut 85 : 623–628 Turner RS, Johnson AH, Wang D (1985) Biogeochemistry of lead in McDonalds Branch Watershed, New Jersey Pine Barrens. J Environ Qual 14 : 305–317 White JR, Driscoll CT (1985) Lead cycling in an acidic Adirondack lake. Environ Sci Technol 19 : 1182–1187