Charge density of total organic carbon in Finnish lakes

Environmental Pollution 77 (1992) 107-113

Charge density of total organic carbon in Finnish lakes Pirkko Kortelainen Water and Environment Research Institute, National Board of Waters and the Environment, PO Box 250, SF-O0101 Helsinki, Finland

Ion balance calculations were used to estimate the charge density of total organic carbon (TOC) and for calibrating the model of Oliver et al. (1983) for Finnish lakes. The database consisted of samples collected in the autumn of 1987 from 955 randomly selected lakes covering the whole country. Organic anion concentrations were estimated as the differencebetween E cations and ~ inorganic anions. The charge density of TOC was estimated by dividing the organic anion concentration by TOC. The median charge density was 7-5 #eq (mg TOC) -l. In lakes with pH values lower than 4.5 the average charge density was 5.2 geq (rag TOC)-mand it increased to 10-3/zeq (rag TOC)-~ in lakes with pH higher than or equal to 7.0. The calibrated model explained 79% of the observed variability in organic anion concentration. The value for the total amount of acidic functional groups per tool organic carbon in the model of Oliver et al. (1983) was slightly higher compared to the Finnish calibration. At pH values higher than 4.4 the Finnish calibration gives lower organic anion concentrations compared to the model of Oliver et al. (1983). INTRODUCTION

quently, the chemical characteristics of the material will vary between different isolation schemes (e.g. Aiken, 1988; Hedges, 1990; Shuman, 1990). There is no general consensus on the best measure of organic acidity (Mulholland et al., 1990). In the pH values existing in natural waters, carboxylic acids are the most important acidity sources. XAD resins are widely used to isolate dissolved organic carbon (DOC). The amount of base used to titrate the isolated sample to pH 7-9 is generally defined as carboxyl acidity (see Shuman, 1990). Because all methods of measuring the organic acidity are laborious and consequently not applicable for routine water sample analysis, most of the studies that have documented the contribution of organic matter to surface water acidity have used indirect estimates of organic acidity. Organic acidity has been found to be important in eastern Canada (e.g. Gotham et al., 1986; Jeffries et al., 1986; Kerekes et al., 1986), in Minnesota, in the upper Great Lakes area, and in Florida (Brakke et al., 1988; Eilers et all., 1988a, 1988b; Landers et al., 1988) as well as in Finland (Kortelainen & Mannio, 1988, 1990; Kortelainen et al., 1989; K/imhri et al., 1991). Organic acids dissociate across a wide range of pH values, but are not completely dissociated in natural waters. The dissociation of organic acids is a function of the pKa value distribution of the acids. Perdue and

Peatlands cover about one third of the land area in Finland. The ditching and draining of peatlands for forestry has been very intensive: to date about half the area of peatlands has been ditched. Coniferous forests are predominant, and 66% of the land area is presently used for forestry (The Finnish Forest Research Institute, 1990). Mineral soils are largely podzolic, having developed from till or glaciofluvial deposits mainly on granitic bedrock. The topography is flat and lakes are predominantly small and shallow. High concentrations of organic matter are characteristic features in Finnish surface waters. Despite decades of research, organic matter in aquatic environments remains poorly characterized at the molecular level. Aquatic organic matter is a complex mixture of organic acids with smaller amounts of neutral and basic components (e.g. David & Vance, 1991). A problem of all organic acids, and humic substances in particular, is that their measured characteristics are often dependent upon the isolation procedure. Isolation and fractionation procedures are seldom perfectly efficient; one commonly loses a fraction of the material or it is transformed in some way by the procedure. ConseEnviron. Pollut. 0269-7491/92/$05.00 © 1992 Elsevier Science

Publishers Ltd, England. Printed in Great Britain 107

108

P. Kortelainen

co-workers suggested a continuous distribution model to describe the extent of proton binding of organic matter across the whole pH range (Perdue et aL, 1980, 1984; Perdue, 1985). The most frequent values were reported at pKa 3.7 and 12.5, corresponding to carboxylic and phenolic functional groups, respectively. The charge density of organic carbon is a function of the pK, value distribution of organic acids. Two approaches, based on different assumptions, have been used to estimate the contribution of organic anions to the charge balance of a solution. Chemical equilibrium calculations use an organic acid model that has been calibrated from potentiometric titrations of naturally occurring organic solutes (Oliver et aL, 1983; Perdue et aL, 1984), whereas ion balance calculations are based on the principle of electroneutrality. Using the electroneutrality approach, the charge density of organic carbon can be estimated by dividing the calculated organic anion concentration (E c a t i o n s - X inorganic anions) by organic carbon. A survey of 955 randomly selected lakes covering the whole country was conducted in autumn 1987 as part of the Finnish Research Project on Acidification (see Forsius et aL, 1990a, 1990b; Ktimtiri et aL, 1991). A large number of chemical variables as well as major watershed characteristics were determined. The objective of the present study was to estimate the charge density of total organic carbon (TOC) in Finnish lakes using ion balance calculations. In addition, organic anion concentrations estimated by ion balance calculations were used to calibrate the model of Oliver et al. (1983), similar to Driscoll et al. (1989).

MATERIAL AND METHODS The lakes for the survey were randomly selected by a two-stage cluster sampling from two separate subregions, together covering the whole country. The selection was weighted by lake density. Lakes of sizes 0.01-10 km 2 were included in the statistical sampling of southern and central Finland. In northern Finland only lakes of sizes 0-1-10 km 2 were included. Four of the selected lakes were not visited owing to difficult field conditions. In addition, five lakes were eliminated after field sampling, due to heavy waste water loading. The water quality of the lakes can be considered representative for small lakes in Finland. Water samples from the surface layer (1 m) of the lakes were taken during the autumn overturn in 1987. A detailed description of the selection of the lakes as well as the analysis methods is given in Forsius et al. (1990b). Ion balance calculations

Total organic carbon (TOC) was measured by uvpersulphate-oxidation followed by IR gas measure-

ments. Colour was measured by the visual comparison method and chemical oxygen demand (CODM,) titrimetrically with KMnO4. The major inorganic cations, Ca, Mg, Na and K, were measured by flame-AAS, and C1 and SO4 by ion chromatography. Fluoride was measured by ion-selective electrode. PO4, NO3, NH4 and Mn were determined colorimetrically. Total monomeric AI and non-labile monomeric AI were measured by the method of R~geberg & Henriksen (1985). Labile monomeric A1 was calculated as a difference between total monomeric AI and non-labile monomeric AI, and it was assumed to have a charge value of 3 in ion balance calculations, pH was measured on a Radiometer pH meter, and the Gran method was used for alkalinity determinations. In the present study, the principle of electroneutrality was used for charge balance analysis. This approach was justified on the basis of the strong relationship between organic anion and TOC (r = 0.870, P < 0.001), CODM, (r = 0-874, P < 0.001) and colour (r = 0.832, P < 0.001). The charge density of TOC was estimated by dividing the organic anion concentration by TOC: ([Ca2÷+ Mg2+ + Na ÷ + K ÷ + NH~ + Mn 2÷ + A13÷ + H ÷] - [SO~+ CI- + NO; + F- + HCO;])/TOC. Gran alkalinity (obtained using the pH 3.7-4-4 regression results) was used as an estimate for HCO;, although these measurements include, in addition to HCO;, some organic alkalinity (see Cantrell et al., 1990). This causes an underestimation of the organic anion concentration calculated as the difference between the main cations and inorganic anions. The Gran alkalinity error in the ion balance calculations is probably to some extent compensated by metal complexation. Both free and complexed metal concentrations are determined when measuring base cation concentrations by atomic absorption spectrometry. If the cations are partly complexed with organic matter, the organic anion content will be an overestimate for the free organic anion content. Based on model calculations, Tipping et al. (1991) concluded that only small amounts of Ca 2÷ and Mg 2+are organically bound at the pH values existing at natural waters. In a recent paper, Urban et al. (1990) concluded that only a minor part of manganese is organically bound. Driscoll et al. (1989) included manganese in ion balances when calibrating the model of Oliver et al. (1983). Consequently, manganese was included in the ion balances of the present study, although the contribution was minor. Iron was not included in the ion balance calculations, as it exists predominantly as an Fe(II)-organic complex at the pHs of most natural waters (Perdue et al., 1976; Stumm & Morgan, 1981). It is difficult to determine the uncertainty associated with organic anion concentration estimated by charge balance. Organic anion concentrations are determined through the analysis of many ions, so uncertainty in these estimates may be large, because of the uncertainty associated with the analysis of each individual ion.

Charge density of total organic carbon in Finnish lakes

However, assuming that the errors associated with sampling and analysis are randomly distributed, the uncertainty in the determination of cationic solutes may be offset to some extent by the uncertainty in the determination of anionic solutes in charge balance calculations. Lakes with charge densities lower than 0 or higher than 25 (n = 23) were excluded from the present database. The database in the present study consisted of 955 lakes.

Model calibration In the model of Oliver et al. (1983), dissociated organic anion concentrations are estimated using the equation:

[A-] =

K.[mDOC]/(K.+

[H÷])

(1)

where K, is the mass action quotient of fulvic and humic acids and m represents the concentration of the total amount of acidic functional groups per mol of dissolved organic carbon (DOC). The parameter m was measured by potentiometric titrations of isolated and purified organic acids from highly coloured surface waters. Oliver et al. (1983) found that the mass action quotient changes with pH over the course of the titration, and that the fulvic acids from two different aquatic sources exhibited almost the same dissociation behaviour as a function of pH, which was identical to earlier studies (Gamble, 1970; Perdue et al., 1980) as well as to the recent study of Cantrell et al. (1990). Oliver et al. (1983) modelled the pica value by the following: pKa = a + b p H - c(pH) 2

(2)

When substituted into eqn (1), this p K , value allows the prediction of organic anion concentration as a function of pH and DOC. In the present study total organic carbon (TOC) measurements were used instead of DOC. Organic anion concentrations calculated by ion balances were used to calibrate the constants a, b, c and m in eqns (1) and (2), using a derivative-free, non-linear least squares routine (SAS, 1989).

109

consequently lower leaching from catchment areas in the colder environment. In the full data set, 26% of the highly humic (TOC > 10mg litre -1) lakes with low sulphate concentrations (*SO4< 30/zeq litre -1) were strongly acidic (alkalinity < 0 p.cq litre-l). These lakes are located mainly in central Finland. The pH value of the lakes in the present material ranged from 4.1 to 8-0 (median 6-3) and the TOC concentration from 1-3 to 46.6 mg litre -I (median 12 mg litre-I). Calculated organic anions were the main anions in the data set (median 92 boeq litre-~). Organic anions dominated in acidic lakes, indicating that a substantial fraction of the overall acidity of such lakes was of natural origin (see Fig. 1). At current deposition levels of sulphate, the pH of humic lakes in Finland is determined to a greater extent by high TOC concentrations than by non-marine sulphate in many areas. However, in the southern parts of the country, where the acidic deposition is highest, the anthropogenic acidity commonly exceeds the catchment-derived organic acidity (see Kortelainen & Mannio, 1990; K~m~iri et aL, 1991). The highest base cation concentrations are found in the same regions. Therefore, pH of the humic lakes is not essentially lower at high deposition areas, but depends on concentrations of all ions. The median charge density was 7.5/zeq (mg TOC) -~. In 10% of the lakes, charge densities were greater than 12 tLeq (mg TOC) -1 and in 10% less than 5.3/~eq (mg TOC) -I (Table 1). The charge density increased with increasing pH (r = 0.479, P < 0.001), alkalinity (r = 0.213, P < 0.001), and base cations (r = 0.174, P < 0.001). Also charge density increased with decreasing TOC concentrations (r = -0.461, P < 0.001). In lakes with pH values lower than 4.5, the average charge

A-/TOC =5.2 5.6 6.4 7.3 8.1 9.3 10.3 (#eq mg -1 ) T0C =21 19 18 16 14 10 7 {rag litre -1) 100n=19 67 118 156 241 279 75 Alkalinity Fluoride

(upper) I

RESULTS AND DISCUSSION Organic acidity and charge density of TOC in Finnish lakes In previous papers (Kortelainen et aL, 1989; Kortelainen & Mannio, 1990; Kam~tri et al., 1991) based on the present material, it has been demonstrated that organic acidity strongly affects the acidity of lakes in Finland. TOC concentrations in lakes are high, especially in central parts of the country. One of the most important reasons for the lower TOC concentrations in northern Finland is probably the lower primary production and

I Organic anions

l m Nitrate (lower) Non-marine sulphate ,o

0

L.O

5.0

pH

6.0

7.0

F i g °1o Mean values for total organic carbon (TOC) and charge density (A-/TOC) for lakes in seven pH classes. Percentage composition of anions is also displayed.

110

P Kortelainen Table I. Chemical characteristics of the lakes

Variable pH Conductance TOC Colour Ca + Mg + Na + K Gran alkalinity Organic anion (A-) ° Charge density (A-/TOC)

10th Percentile mS m - ~ mg litre -~ mg litre -~ Pt p.eq litre -t /zeq litre -~ /,teq litre -~ /~eq (mg TOC) -j

5.0 1-7 4.8 25 120 - 1 40 5.3

Median 6.3 3-1 12 100 250 75 92 7.5

90th Percentile 6.9 6.3 24 220 500 220 170 12

° A- [Ca2+ + Mg2+ + Na ÷ + K ÷ + NH~ + Mn 2÷ + AI3+ + H ÷] - [SO~- + C1- + NO3 + F- + HCO3]. AP ÷ is labile monomeric AI (Rogeberg & Henriksen, 1985), assumed to have a charge value of 3 in ion balance calculations. Gran alkalinity was used as an estimate for HCOf. =

density was 5.2/zeq (mg TOC) -~ (average TOC = 21 mg litre -t) and it increased to 10.3/zeq (mg TOC) -~ in the lakes with p H higher than or equal to 7.0 (average TOC = 7 m g litre -~) (Fig. 1). Tipping et al. (1991) pointed out that, although charge density is very closely related to pH, the relationship is seldom linear because, besides proton-binding, the binding of metal ions, especially aluminium, may be significant. In the present data set, the close relationship between pH and TOC (r -- -0.508, P < 0.001) is most probably reflected also in a charge density-TOC relationship. When lakes are stratified according to pH and TOC into different groups the charge density increases with increasing pH, but at high pH values also with decreasing TOC (Table 2). The lowest charge density values were found near the Gulf of Bothnia and in the eastern parts of the country, where the number of acid humic lakes is high. The highest charge density values were found in northern Finland, where the proportion of clear water-high pH lakes is highest. Thompson (1986) demonstrated that the organic anion contribution to the alkalinity titrations is largest for samples with high DOC concentrations and high pH values. In the present data base, the alkalinity contribution to ion balances is minor at pH values lower than 5.0 (see Fig. 1). However, it is probable that, in humic lakes with higher pH values, charge densities are somewhat Table 2. Median charge density of total organic carbon (TOC) (peq (rag TOC) -~) in Finnish lakes in different pH and TOC classes TOC<5 5_20 (mg litre -t) pH < 5-0 5.0 < pH < 6.0 6.0 - 6-5

7.1 (n = 5) 6.8 (n=20) 10 (n=8) 13 (n=68)

5.8 (n = 12) 6.9 (n=35) 8.6 (n=70) 9-2 (n=162)

5.2 (n = 27) 6.8 (n=123) 7.5 (n= 123) 8-1 (n= 113)

5.6 (n = 42) 6.5 (n=96) 6-9 (n=40) 7.4 (n=ll)

underestimated due to the dissociation of organic acids in the Gran alkalinity measurements. Charge density in Finnish lakes compared to other countries In the Norwegian survey (Henriksen et al., 1988b), which was targeted at sensitive areas, lakes were predominantly clearwater acidic lakes located mainly at high deposition areas. Ninety per cent of the lakes had TOC < 6 mg litre-', and 30% of the lakes had pH ___5.0. In the Norwegian data set, the charge density was generally 4-5/~eq (mg TOC-I). In the most acidic lakes (pH < 4.8) charge densities decreased to 2-5-3/zeq (mg TOC) -~, and values of TOC < 2 mg litre -I contributed negligible amounts of dissociated organic anions (Henriksen et al., 1988a,b). In the Eastern Lake Survey in the USA (Landers et al., 1988), the pH values were higher than in the Norwegian or Finnish survey. Organic matter concentrations were higher than in the Norwegian survey, but lower than in the Finnish survey. The charge density increased with increasing pH. At pH 5 the mean charge density in different areas ranged from about 2 to 5/~eq (mg DOC)-t, whilst at pH 7 the range was from about 6 to 14/zeq (mg DOC) -~ (Driscoll et al., 1989). When comparing charge density of organic matter from different sources, pH differences of the samples have to be kept in mind. For example, many Norwegian surface waters have low pH values, consequently charge densities are low. In addition to the different distributions of lake water pH values, the nature of organic carbon, and consequently charge densities, can be expected to vary regionally, temporally, or as a function of water residence time or other limnological factors. Many Norwegian watersheds can be characterized by having thin organic soil horizons and considerable bedrock exposure, as well as steep topography. Finnish watersheds have thicker, finer-textured soils with higher proportions of poorly drained peaty soils. Precipitation amounts and run-off are higher in Norway than in Finland. As a consequence, there is a longer residence

Charge density o f total organic carbon in Finnish lakes

111

p K , = 1.1pH - 0.041(pH) 2 6"01_1171

/..,

Oliver et al.

5.61--

~

ELS

(Oliver et al., 1983) (4) p K , = 1.31pH - 0-07(pH) 2 (Driscoll et al., 1989)

°,o . °

--

/

/

....

S" .....

4.0 3.6

70

,'l

I

[A-] = Ko[9.2TOCJ/(Ka + [H÷]) i

I

(5)

The value for the total concentration of acidic functional groups per mol organic carbon (parameter m in eqn 1) in the Finnish calibration was slightly lower compared with the model of Oliver et al. (1983) and the calibrated value of Driscoll et al. (1989):

,° °°

// t,./, -

(3)

pKo = 0.96 + 0.90pH - 0.039(pH) 2

~.'"

5.2--

~g

(Finland)

I

I

i

(Finland)

(6)

[A-] = Ka[IODOC]/(K~ + [H+]) (Oliver et al., 1983) (7)

b

[A-] = K~[ll.7DOC]/(K a + [H÷]) (Driscoll et aL, 1989) (8)

60 // Gr

~°°°°*°°° ° ° * ° °° **° °°* ° °°° °°° `

50

d .9 g

~° ..p"/""

._¢-(2 40 °°o°

°°°°°

)

O

30 3.5

4.5

5.5 pH

6.5

7.5

Fig. 2. Comparison of Oliver et al. 's (1983) model results for (a) pKo and (b) organic anion concentration using model parameters of Oliver et ai. (1983), calibrated values of the US Eastern Lake Survey (ELS) by Driscoll et al. (1989) and Finnish calibrated values. Calculations of organic anion concentrations were made assuming a TOC of 6 mg litre-~ (=500/~mol C litre-t; compare to Driscoll et al., 1989).

The calibrated model was able to account for 79% of the observed variability in the organic anion concentration in the Finnish lakes (see Fig. 3). At pH values lower than 5.1, the Finnish calibration gives lower p K , values compared to Oliver et al. (1983), while the Driscoll et aL (1989) calibration gives higher pK~ values. At the pH values higher than 5-1, both calibrations give higher pK~ values compared to Oliver et aL (1983) (see Fig. 2a). Compared to the Finnish calibration the discrepancy in apparent pKa values is 0.3 units at pH 7.0, whilst compared to the US Eastern Lake Survey data set the discrepancy is about 0.4 units at pH 7.0. Driscoll & Bisogni (1984) reported that the pK~ values for waters low in alumino-organic solutes were approximately 0.5 pH units lower than for lakes with elevated concentrations of alumino-organic solutes. The model of

350 a

time of water in soils and greater contact with organic material occurring in a thicker soil horizon in Finland than in Norway, which may result in differences in the quality of organic carbon in the run-off water.

250 "7 A _

Model calibration

6

~

A •

•

A •

A

, A

~r (u ~L

,~ 150

Driscoll et aL (1989] calibrated the Oliver et aL (1983) model using a reduced version of the US Eastern Lake Survey data set. They found that it was difficult to calibrate the model for two subregions. However, model convergence could be achieved if the intercept (parameter a in ¢qn 2) was constrained at zero (similar to calibrations obtained for other subregions). Also for the Finnish data set, model convergence was achieved only when the intercept was set to zero. Results of the model calibration for parameters b and c were close to the values proposed by Oliver et aL (1983) (see Fig. 2a):

• .•'.~=4qmJllb~"

p

50 0 0

50

150 250 ZCot -ZAn (/~=q litre q )

350

Fig. 3. Relationship between organic anion content estimated by ion balance calculations (~ C a t - X An) and by the calibrated model of Oliver et al. (1983) (A-). The regression equation was: y = 1.0x + 5.4, r2 = 79%, n = 954.

112

P. Kortelainen

Oliver et aL (1983) was based on potentiometric titrations of preconcentrated and purified organic acids. The carboxyl acidity, measured from preconcentrated fractions which have been stripped of complexed metals (e.g. AI and Fe), is often higher than charge densities estimated by ion balance calculations. In many studies, 10/.teq (mg C) -t has been accepted as a median value for the experimentally measured carboxyl acidity of aquatic humus (see, for example, Oliver et aL, 1983; Perdue et aL, 1984; Cantrell et aL, 1990). The differences in the values of parameters in the models have implications for estimating organic anion concentration in surface waters. According to Driscoll et aL (1989), hypothetical calculations of organic anion concentration were made for water containing 500/xmol C litre -t over a range o f p H values (3-5-7.5), using model parameters provided by Oliver et al. (1983), values obtained from the calibration of Driscoll et al. (1989) and values from the calibration of the present data set (see Fig. 2b). The differences in calculated organic anion concentrations are a function of both pH values and organic carbon concentrations. At low pH values, the Finnish calibration gives higher organic anion concentrations compared to the two models. The Finnish calibration gives lower organic anion concentrations at pH values higher than 4-4 compared to Oliver et aL (1983) and at pH values higher than 4.9 compared to Driscoll et aL (1989). The discrepancy increases when TOC increases. The discrepancy in calculated organic anion concentrations is due to the different values for total acidity (parameter m in eqn 1) as well as different apparent pKa values in the three models. Some of the discrepancy between the models is most probably due to the underestimation of organic anions in the Finnish data set, because Gran alkalinity was used as an estimate for HCO~- in the ion balance calculations.

CONCLUSIONS In a heterogeneous lake material no simple relationship between organic carbon and the organic acid fraction can be expected, because TOC measures a variety of organic carbon forms which can vary greatly in their acid-base characteristics. The charge density of TOC increases with increasing pH, although it is probable that a considerable amount of organic acids is dissociated already at the lowest pH values in the Finnish data set. Compared to the model of Oliver et al. (1983) the Finnish calibration gives lower values for total concentration of acidic functional groups per mol organic carbon, as well as for organic anion concentrations at pH values higher than 4.4. The differences in parameters between the models may reflect differences in the composition of organic carbon, the treatment (isolation,

purification) of natural organic matter in the study of Oliver et al. (1983), as well as the underestimation of organic anion concentrations in the Finnish data set, because Gran alkalinity was used in ion balance calculations as an estimate for HCO3. Although ion balance calculations are a very rough measure, they are useful in estimating the contribution of organic acidity to the acidity of surface waters, especially in large survey studies, where direct organic acid measurements are too laborious. The results of the present survey also support the conclusion of Driscoll et aL (1989) that the calibration of the model of Oliver et aL (1983) is a useful tool in estimating the organic acid dissociation and the contribution of organic matter to the acidity of surface waters, in different regions.

ACKNOWLEDGEMENTS Martin Forsius, Jaakko Mannio and Maximilian Posch are acknowledged for helpful discussions during preparation of this paper, and David Brakke and Mark David for useful comments on the manuscript.

REFERENCES

Aiken, G. R. (1988). A critical evaluation of the use of macroporous resins for the isolation of aquatic humic substances. In Humic Substances and Their Role in the Environment, ed. F. H. Frimmel & R. F. Christman. Wiley, Chichester. Brakke, D. F., Landers, D. H. & Eilers, J. M. (1988). Chemical and physical characteristics of lakes in the Northeastern United States. Environ. Sci. TechnoL, 22, 155-63. Cantrell, K. J., Serkiz, S. M. & Perdue, E. M. (1990). Evaluation of acid neutralizing capacity data for solutions containing natural organic acids. Geochim. Cosmochim. Acta, .54, 1247-54. David, M. B. & Vance, G. F. (1991). Chemical character and origin of organic acids in streams and seepage lakes of central Maine. Biogeochemistry, 12, 17-41. Driscoll, C. T. and Bisogni, J. J. (1984). Weak acid/base systems in dilute acidification lakes and streams in the Adirondack region of New York State: In Modeling of Total Acid Precipitation Impacts, ed. J. L. Schnoor. Butterworth, Boston. Driscoll, C. T., Fuller, R. D. & Schecher, W. D. (1989). The role of organic acids in the acidification of surface waters in the Eastern U.S. Water Air Soil Pollut., 43, 21-40. Eilers, J. M., Brakke, D. F. & Landers, D. H. (1988a). Chemical and physical characteristics of lakes in the Upper Midwest, United States. Environ. Sci. Technol., 22, 164-72. Eilers, J. M., Landers, D. H. & Brakke, D. F. (1988b). Chemical and physical characteristics of lakes in the southeastern United States. Environ. Sci. Technol., 22, 1727. Forsius, M., Kamari, J., Kortelainen, P., Mannio, J., Verta, M. & Kinnunen, K. (1990a). Statistical lake survey in Finland: regional estimates of lake acidification. In Acidification in Finland, ed. P. Kauppi, P. Anttila and K. Kentt~imies. Springer-Verlag, Berlin and Heidelberg.

Charge density of total organic carbon in Finnish lakes Forsius, M., Malin, V., Makinen, I., Mannio, J., Kamari, J., Kortelainen, P. & Verta, M. (1990b). Finnish lake acidification survey: survey design and selection of lakes. Environmetrics, 1(1), 73-88. Gamble, D. S. (1970). Titration curves of fulvic acid: the analytical chemistry of a weak acid polyelectrolyte. Can. J. Chem., 48, 2662-9. Gorham, E., Underwood, J. K., Martin, F. B. and Ogden, J. G. (1986). Natural and anthropogenic causes of lake acidification in Nova Scotia. Nature (Lond), 324, 451-3. Hedges, J. I. (1990). Compositional indicators of organic acid sources and reactions in natural environments. In Organic Acids in Aquatic Ecosystems, ed. E. M. Perdue and E. T. Gjessing. Wiley, Chichester. Henriksen, A., Brakke, D F. & Norton, S. A. (1988a). Total organic carbon concentrations in acidic lakes in southern Norway. Environ. Sci. Tecbnol., 22, 1103-5. Henriksen, A., Lien, L., Traaen, T. S., Sevaldrnd, I. S. & Brakke, D. F. (1988b). Lake acidification in Norway-present and predicted chemical status. Ambio, 17, 259-66. Jeffries, D. S., Wales, D. L., Kelso, J. R. M. & Linthurst, R. A. (1986). Regional chemical characteristics of lakes in North America. Part I: Eastern Canada. Water Air Soil Pollut., 31, 551-69. K~imari, J., Forsius, M., Kortelainen, P., Mannio, J. & Verta, M. (1991). Finnish lake survey: present status of acidification. Ambio, 20(1), 23-7. Kerekes, J., Beauchamp, S., Tordon, R. & Pollock, T., (1986). Sources of sulphate and acidity in wetlands and lakes in Nova Scotia. Water Air Soil Pollut., 31,207-14. Kortelainen, P. & Mannio, J. (1988). Natural and anthropogenic acidity sources for Finnish lakes. Water Air Soil Pollut., 42, 341-52. Kortelainen, P., Mannio, J., Forsius, M., Kam~ri, J. & Verta, M. (1989). Finnish lake survey: the role of organic and anthropogenic acidity. Water Air Soil Pollut., 46, 235-49. Kortelainen, P. & Mannio, J. (1990). Organic acidity in Finnish lakes. In Acidification in Finland, ed. P. Kauppi, P. Anttila & Kenttamies. Springer-Verlag, Berlin and Heidelberg. Landers, D. H., Overton, W. S., Linthurst, R. A. & Brakke, D. F. (1988). Eastern lake survey: regional estimates of lake chemistry. Environ. Sci. Technol., 22, 128-35. Mulholland, P. J., Dahm, C. N., David, M. B., Di Toro, D. M., Fisher, T. R., Hemond, H. F., K6gel-Knabner, I., Meybeck, M. H., Meyer, J. L. & Scdell, J. R. (1990). What are the temporal and spatial variations of organic acids at

113

the ecosystem level? In Organic Acids in Aquatic Ecosystems, ed. E. M. Perdue & E. T. Gjessing. Wiley, Chichester. Oliver, B. G., Thurman, E. M. & Malcolm, R. L. (1983). The contribution of humic substances to the acidity of colored natural waters. Geochim. Cosmochim. Acta, 47, 2031-5. Perdue, E. M., Beck, K. C. & Reuter, J. H. (1976). Organic complexes of iron and aluminium in natural waters. Nature (Lond.), 260, 418-20. Perdue, E. M., Renter, J. H. & Ghosal, M. (1980). The operational nature of acidic functional group analyses and its impact on mathematical descriptions of acid-base equilibria in humic substances. Geochim. Cosmochim. Acta, 44, 1841-51. Perdue, E. M., Reuter, J. H. & Parrish, R. S. (1984). A statistical model of proton binding by humus. Geochim. Cosmochim. Acta, 48, 1257-63. Perdue, E. M. (1985). Acidic functional groups in humic substances. In Humic Substances in Soil, Sediment, and Water, ed. G. R. Aiken, D. M. Mcknight, R. L. Wershaw & P. MacCarthy. Wiley, New York. Rogeberg, E. J. S. & Henriksen, A. (1985). An automatic method for fractionation and determination of aluminium species in fresh waters. Vatten, 41, 48-53. SAS (1989). SAS Institute Inc., SAS/STAT User's Guide, Version 6, Fourth Edition, Volume 2, Cary, NC. Shuman, M. S. (1990). Carboxyl acidity of aquatic organic matter: possible systematic errors introduced by XAD extraction. In Organic Acids in Aquatic Ecosystems, ed. E. M. Perdue & E. T. Gjessing. Wiley, Chichester. Stumm, W. & Morgan, J. J. (1981). Aquatic Chemistry. An Introduction Emphasizing Chemical Equilibria in Natural Waters, 2nd ed. Wiley, New York. The Finnish Forest Research Institute (1990). Yearbook of Forest Statistics, 1989. Folia Forestalia 760, Helsinki, Finland, p. 246. Tipping, E., Woof, C. & Hurley, M. A. (1991). Humic substances in acid surface waters; modelling aluminium binding, contribution to ionic charge-balance, and control of pH. War. Res., 25, 425-35. Thompson, M. (1986). The Effects of Organic Anions on Alkalinity Measurements in Organic-rich Waters in Nova Scotia. NWRI Contrib. 86-9, Nat. Water Res. Inst, Burlington, Ontario, Canada. Urban, N. R., Gorham, E., Underwood, J. K., Martin, F. B. & Ogden, J. G. (1990). Geochemical processes controlling concentrations of Al, Fe, and Mn in Nova Scotia lakes. Limnol. Oceanogr., 35(7), 1516-34.