Urban Land-Use Influences on Transport of Heavy ... - Springer Link

URBAN LAND-USE INFLUENCES ON TRANSPORT OF HEAVY METALS TO LAKES AND CONCENTRATIONS IN FISH M. LINDSTRÖM and L. HÅKANSON∗ Department of Earth Sciences, Uppsala University, Villav. 16, S-752 36 Uppsala, Sweden (∗ author for correspondence, e-mail: [email protected])

(Received 4 February 1999; accepted 25 October 2000)

Abstract. To investigate the urban land-use influences on transport of heavy metals to lakes and metal concentrations in fish liver (perch and crucian carp), ten lakes in the Stockholm have been investigated. The catchment area characteristics have been studied in detail and put on a GIS-platform. The morphometry, water quality and sedimentological characteristics of the lakes have been mapped and the metal concentrations in water, sediments and fish liver analysed. Evidently, metal concentrations in fish depend on many internal characteristics related to the lake food-web and predation pressure, water chemistry (which influences metal partitioning coefficients and hence pelagic and benthic metal transport pathways), and lake morphometry and hydrodynamics (which influence transport patterns and internal loading). However, many of these causal agents regulating metal concentrations in water, sediments and fish, depend on catchment area characteristics. One aim of this paper has been to quantitatively rank how the studied factors influence metal concentrations in lake water and fish using statistical methods to find out how much of the variability among the lakes that can be statistically explained by these factors. This should provide an interesting base for further studies on the role of other factors, not accounted for in this study, and about causal mechanisms. The results may also, hopefully, be used to address questions related to remedial measures – what can actually be done to reduce metal transport from urban areas to lakes, and metal concentrations in fish, and what can be expected from such remedies? Normed catchment heavy metal loads to the lakes have been calculated from models. Two-step regression models consisting of catchment ‘size’ and catchment ‘urban status’ parameters could explain large parts of the variations in the metal fluxes among the lakes. These two main clusters of catchment variables are defined and motivated in this work. Cu was found to be the metal most dependent on ‘urban status’ followed by Cr. Pb had about equal influences from the ‘size’ as from the urban catchment ‘status’ variables. Cd, Hg, Ni and Zn were more influenced by the ‘size’ than by the ‘urban status’. For most metals, concentrations in fish liver were found to be most correlated to general urbanity parameters and Cu and Pb to communications variables. Keywords: fish, lakes, land-use, metals, sediments, statistical models, urban areas

1. Introduction Heavy metals transported from catchments to water emanate from several sources, including agriculture and other types of land-use, natural weathering from rocks and soils (Salomons and Förstner, 1984), emissions from anthropogenic activities and atmospheric depositions from local and/or regional sources (e.g., Tarvainen et al., 1997). Heavy metal concentrations of different aquatic environments (such as Water, Air, and Soil Pollution: Focus 1: 119–132, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands.

120

M. LINDSTRÖM AND L. HÅKANSON

Figure 1. Location map of the Stockholm study lakes.

rivers, ground water and lakes) have been shown to depend on urban parameters. High metal concentrations in samples of runoff water from vehicle service and parking areas were found by, e.g., Malmqvist (1983) and Pitt et al. (1995). None of these studies, however, quantifies the anthropogenic load in relation to the natural. Large pools of heavy metals are building up in the technosphere of urban environments (SNV, 1996; Lohm et al., 1997). One aim of this work is to identify and quantify how urban land-use influence the heavy metal transport from catchment to lakes. Influences of urban land-use on the transport of heavy metals to lakes is related to the present anthropogenic load as well as to the natural (a defined/reference) situation. Identification of the present anthropogenic load of heavy metal to lakes gives a basis for assessing possible remedies. Increased metal concentrations in fish is used as an indicator of pollution in different contexts (Håkanson, 1984). Increased mercury concentrations in pike and perch is a known effect of acidification (Håkanson et al., 1990; Andersson et al., 1995) as well as contamination from industrial point sources. Some urban factors have been shown to cause increasing metal concentrations in fish, e.g., Campbell (1994) showed increased metal concentrations in fish living in storm water treatment ponds and Lenat and Crawford (1994) a disturbed fish community in an urban site compared to an agricultural. To the best of our knowledge, no study has, however, revealed how different urban activities influence metal concentrations in

URBAN LAND-USE INFLUENCES ON TRANSPORT OF HEAVY METALS

121

Figure 2. Description of the catchment area classification system. Six main classes, each divided into two to eight subclasses. Most of the classes belong to the catchment ‘urban status’ cluster, other are related to the lake/catchment ‘size’ cluster.

fish. This work has used an urban land-use gradient to correlate concentrations in fish liver to a number of urban land-use parameters. Our studies concern metals in fish liver, since most metals are known to be concentrated in liver and kidney and the concentrations of metals in the liver give a more relevant measure of metal contamination of lakes than data from, e.g., muscle tissue (see, e.g., Håkanson and Uhrberg, 1981; Håkanson, 1984). From the methodological presuppositions, our main aim is to give a quantitative ranking based on empirical data of the factors in the catchment area which influence metal concentrations in fish.

2. Methods and Data Ten lakes and catchments in the Stockholm region covering as wide a range of lake and catchment characteristics as possible where chosen for this study. The lakes are rather small, with areas from 0.04 to 0.79 km2 and maximum depths from 2.3 to 13.8 m. The urban influence on the catchment areas ranges from 0 to almost 100% and mostly consists of suburban residential areas with both single- and multifamily houses and city centre features. There are few large industries or known large point sources for metal emissions in these catchment areas. Most lakes are eutrophic and have pH in the range from 6.6 to 8.0. Figure 1 gives a location map of the study area and Table I a compilation of data for the lakes and their catchments. Table I also lists all the factors tested in the following statistical analyses. Lake water was sampled on 5 occasions during 1996. Heavy metal concentrations were analysed using ICP-MS for all metals except Hg, which was analysed

122


TABLE I Compilation of lake data, morphometry, water chemistry and catchment characteristics Parameter Lake morphometry Maximum depth Mean depth Theoretical water retention time Volume development (= 3×Dm/Dmax) Area of drainage area

Min

Mean

Max

Unit

Abbreviation

m m yr –

Dmax Dm T Vd

km2

ADA

– mS m−1 mg L−1 m µg L−1 µg L−1

pH Cond Alk Secchi TN TP

2.3 1.31 0.1 1.52

5.1 2.90 1.3 1.74

13.8 7.16 4.9 1.99

0.57

1.95

4.74

Water chemistry (mean values for 1996) pH 6.58 Conductivity 7.23 Alkalinity 12.42 Secchi depth 1.3 Total N concentration 548.2 Total P concentration 13.4

7.62 44.26 120.8 9 2.5 913.5 57.8

8.04 98.82 241.4 4.1 1474.2 116.6

Metal concentrations in perch liver Cd 0.018 Cr 0.012 Cu 5.2 Hg 0.019 Ni 0.028 Pb 0.017 Zn 65 No. of perch 2 Weight of perch 13.2

0.409 0.030 25.3 0.058 0.054 0.052 106.96 7 107.3

1.398 0.07 144.07 0.093 0.104 0.098 135.83 10 303.4

µg/g ds µg/g ds µg/g ds µg/g ws µg/g ds µg/g ds µg/g ds – g

Metal concentrations in crucian carp liver Cd 0.005 Cr 0.013 Cu 4.2 Hg 0.009 Ni 0.038 Pb 0.078 Zn 60 No. of crucian carps 2 Weight of crucian carps 139.5

0.033 0.025 8.8 0.015 0.069 0.143 86.86 5 385.9

0.108 0.046 15.6 0.024 0.134 0.263 154 10 946.3

µg/g ds µg/g ds µg/g ds µg/g ws µg/g ds µg/g ds µg/g ds – g


123

TABLE I (continued) Parameter

Min

Mean

Max

Unit

Abbreviation

61723 36172 0 0 0 0 0 92 0 0 0 0 0 0 0 0 211756 0 0 0 0 0 0

294561 239605 52389 173232 119419 18503 3009 464792 31088 6513 9989 38051 48977 304997 292983 127060 964607 238190 726416 8642 670 28 56

942405 759344 177446 407381 347611 45677 17251 1797315 166788 21101 48805 183992 223470 1725874 667392 442698 2164703 594434 1720409 26261 3287 76 97

m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 % %

Water area Wetland Comm Road P-lot P-house Tbuild Ind Oth.env.aff. CC Inst Mul-fam One-fam OL Grass nat mnat lnat Other Curoof BIx AIx

Catchment area characteristics Total water area Lake area Wetlands area Total communications area Roads Parking-lot area Parking-house area Total building area Legislated industry area Other environmental affecting area City centre area Institutions Multi family house One family house Total open land area Grass area Total nature area Minor nature area Larger nature area Total other area Copper roof area Index of total building influences Index of total anthropogenic influences

with Fluorescens, by a certified laboratory (SGAB, Luleå, Sweden). Standard water chemical variables (pH, alk, cond, total phosphorous and total nitrogen, etc.) were analysed by Stockholm Water according to the ‘Swedish standard’. Surface sediment samples from 5 accumulation area sites per lake were collected in 1996 (using methods presented by Håkanson and Jansson, 1983) and analysed in a standard way for heavy metals (Cd, Cr, Cu, Ni, Pb, Zn) by flame AAS following digestion with HCl:HNO3 (1:3), 100 ◦ C for 1 hr and room temperature for 20 hr. Hg in the sediment samples was analysed by MeAna, Uppsala, Sweden, according to Uhrberg (1981). During 1996 fish of two species, one omnivore, perch (Perca fluviatilis) from 9 lakes and one benthivore crucian carp (Carassius carassius), from 7 lakes were

124


caught by net. Pooled samples of the fish livers from each lake were analysed for heavy metal concentrations by MeAna (see Table I). From empirical data of water and sediment concentrations, Lindström and Håkanson (2000) designed and tested a dynamic mass-balance model assessing the main fluxes to, from and within each lake discussed in this study. Lindström (2000) used multiple regression models to correlate calculated loads of these metals from the catchments to numerous catchment characteristics assumed to influence metal transport from land to water. The urban status of the catchments of the lakes were described according to a system presented in Figure 2, where all land and water areas were classified depending on natural and urban land-use and factors influencing metal transport to the lakes. The best models to calculate metal fluxes to lakes were based on a size parameter and an urban catchment status parameter as model variables. The model variables are here transferred to the same scales by a standard scoring (a normalisation procedure) to facilitate a comparison of the parameter coefficients, by subtraction of the mean (MV) and division by the standard deviation (SD). This allows a dimensionless comparison of how the different model variables (x) influences the catchment load of metals.

3. Results and Discussion Among the catchment parameters there are some clusters of functionally and statistically related parameters. Table II gives a simple cluster analysis which shows that, e.g., the total communication areas are strongly correlated with the parkinglot areas (r = 0.78). This is logical since they both represent the communications cluster. Some variables are, however, more or less correlated by chance, e.g., the minor nature areas (mnat) are also well correlated with the parking-lot areas (r = 0.95). This illustrates that all regression parameters must be seen in the light of the cluster they represent. As can be seen in Table II, the areas of copper roofs represent general urban influences. In the same way, parks are related to the communications cluster. The ‘size’ cluster is, as shown in Figure 2, represented by several related lake and catchment area size variables. Some of the catchment description classes correlate with the ‘size’ parameters more or less by chance. As an opposite reference cluster to the ‘size’ cluster, all other variables are lumped together and called the ‘urban catchment status’ cluster. 3.1. L AND - USE INFLUENCES ON TRANSPORT OF HEAVY METALS Table III presents regression models as normalised regression coefficients and statistics. The model variables, the ‘size’ and the ‘urban catchment status’ clusters, can describe the variation in metal transport from land to lakes with high statistical degrees of explanation, r2 . Since all model variables are transformed to yield as


125

Figure 3. The two parameter normalised regression plane for Pb, together with points for nine of the lakes. Catchment load depends on a size parameter (T = theoretical lake water retention time) and a catchment urban status parameter (road = area of roads in the drainage area). The two slopes in each direction represent the relative strengths of the two given model variables.

normal frequency distributions as possible (see Håkanson and Lindström, 1997) and the parameters are relatively independent, the equations may be visualised by a linear plane (Figure 3). The relative strength of the two model variables are represented by the slope coefficients of the plane (regression equation) in each direction. As can be seen from Table III, Cu has a higher slope for the urban catchment status variable than for the size variable and for Cr and Pb the size and the urban status coefficients are about the same. For Cd, Hg, Ni and Zn the slope of the size variable is much higher than for the urban status variable. From a lake management point of view, it is of great interest to be able to differentiate between different kinds of pollution sources, since this is important for determining which remedial actions are likely to be successful and to get realistic expectations of the result of the remedial measures and, hence also, to get important information for cost-benefit analyses. This work has differentiated the local (= lake specific) load from the catchment (proportional to the catchment urban status

126


TABLE II Cluster analysis, giving the correlations among the x-variables with the highest r-values. This table identifies cluster groups for all the given variables. For abbreviations see Table I Variable

Best correlating variables (r)

Cluster

P-lot

mnat road TP comm water area V AIx oth.env.aff Inst one-fam Curoof road mnat P-lot comm Curoof BIx gard P-lot, Tbulid AIx Tbuild TP comm, road comm mnat ADA road TP Tbuild OL lnat OL ADA gard nat mul.fam ind V

Communications cluster

P-house

Curoof oth.env.aff

Ind

AIx

BIx

Park

Grass Nat

Lnat

(0.95) (0.92) (0.87) (0.78) (0.79) (0.77) (0.76) (0.82) (0.88) (–0.91) (–0.92) (0.88) (–0.88) (0.90) (0.86) (0.76) (0.82) (0.78) (–0.77) (0.67) (0.78) (0.75) (0.73) (0.72) (0.94) (0.92) (0.86) (0.85) (0.81) (0.80) (0.76) (0.95) (0.80) (0.80) (0.75) (0.95) (0.86) (0.68) (0.66)

Size

AIx /Copper roofs? General urban influences, AIx /not communications


AIx

AIx /Total building areas

Communications cluster /size /General urban influences

open land Non urban areas /Size?

? /Size


TABLE II (continued) Variable

Best correlating variables (r)

Cluster

Mnat

comm park road TP track ind P-lot mul-fam lnat bath mnat area, V oth.env.aff ADA Tbuild DR road track Secchi Curoof Dm, Dmax Secchi T wetland comm road oth.env.aff Tbuild ind wetland P-lot Ind imperm surf TP nat grass park ADA other

General urban influences /g AIx

CC

Mul-fam

One-fam

Wetland

Ditch

Inst

Vd

Other areas

OL

(0.95) (0.92) (0.86) (0.76) (0.89) (0.72) (0.71) (0.73) (0.86) (0.84) (0.74) (0.67) (–0.91) (0.83) (0.82) (0.81) (0.77) (–0.99) (0.64) (–0.63) (0.88, 0.81) (0.78) (0.76) (0.74) (0.69) (0.67) (–0.91) (–0.62) (–0.58) (0.51) (0.81) (0.76) (0.92) (0.70) (0.80) (0.76)/ (0.73) (0.72) (0.72)

City centre?

Size

Size /Tbulid

Wetlands /not AIx Size /Wetlands


Lake form Non-anthropogenic. /? Communications?

Open land

127

128


TABLE III Best models for the given metals (compiled from Lindström, 2000, the predicted variable is the metal transport (kg yr−1 ) from land to lake). F-ratio, r2 , number of lakes (n), size slope coefficient, model variable, status slope coefficient and model variable. The slope coefficients illustrate the relative strengths of the model variables. All regressions, except for Hg, are 95% significant. For abbreviations see Table I Me

F

r2

n

Size coeff.

Variable

Status coeff.

Variable

Cd Cr Cu

7 9 26 6 6a 7 28 16 12

0.94 0.84 0.90 0.81 0.88 0.97 0.99 0.90 0.97

7 8 7 9 8 8 7 9 7

0.94 0.53 0.82 0.46 0.78 0.79 1.03 0.68 0.97

log(area) log(T) log(water) log(V) log(T) log(water) log(water) log(T) log(area)

0.38 0.65 0.98 0.68 0.44 0.30 0.29 0.55 0.33

log(Curoof) log(park) log(Curoof) log(BIx) log(park) log(park) log(Curoof) log(road) log(Curoof)

Hg Ni Pb Zn

a 90% Significance.

parameter) from the regional background load, which influences all investigated lakes in more or less the same manner, and is proportional to the lake/catchment area size variables (see Lindström, 2000). 3.2. L AND - USE EMISSIONS AND EFFECTS ON BIOTA Heavy metal concentrations in fish liver have been correlated against transformed relative values of the catchment parameters. The results are presented in Table IV. The parameters are grouped into clusters of functionally and statistically related parameters. According to Table II the following 6 clusters likely exist in this material: 1) communications, 2) size, 3) general urban influences (AIx), 4) open land, 5) natural, non-urban, areas and 6) wetlands. The Cd concentrations in perch liver correlate negatively with the communications cluster. Both perch and crucian carp show a negative correlation with Vd, the lake form factor, which does not belong to any clear cluster and represents internal lake conditions, like resuspension (see Håkanson and Peters, 1995). A low Vd characterises lakes with large shallow areas, which indicates high internal loading due to high wind/wave induced resuspension and higher metal concentrations in fish. For crucian carp, Cr correlates with lake depth and both positively and negatively with parameters from the communications cluster. For both species of fish there are correlations between Cu and the communications cluster as well as the size cluster. Hg for crucian carp has a correlation with the communications cluster and perch


129

TABLE IV Regressions showing the relations between metal concentrations in pooled samples of fish liver and different land-use and lake morphometry parameters (p < 0.05, i.e., 95% significance). Abbreviations, see Table II Eq. no.

F>

r2

n

Regression equation

1 2 3

15 37 8

0.88 0.86 0.84

8 8 7

log(Cdperch) = 0.65 – 1.67 × Vd – 1.34 × log(road) log(Cdperch) = –1.72 – 0.63 × log(inst) log(Cdperch) = 8.96 – 4.52 × Vd + 0.97 × log(wetland)

4 5 6

21 30 36

0.97 0.96 0.84

8 6 9

log(Cuperch) = –0.01 + 33.6 × (P-lot) – 0.28 × log(other) log(Cuperch) = 2.15 + 159.4 × (P-house) + 0.32 × log(Curoof) log(Cuperch) = 0.80 + 27.0 × (P-lot)

7

14

0.74

7

log(Hgperch) = –0.58 + 0.43 × log(wetland)

8 9

9 7

0.88 0.61

7 7

log(Niperch) = –1.07 + 0.58 × log(park) + 0.34 × log(BIx) log(Niperch) = –0.55 + 0.60 × log(park)

10 11

12 9

0.75 0.71

6 6

log(Pbperch) = –0.59 + 0.20 × log(park) log(Pbperch) = –0.65 + 0.30 × log(oth.env.aff.)

12 13 14 15

11 8 18 6

0.91 0.79 0.72 0.49

9 9 9 9

log(Znperch) = 2.98 + 0.08 × log(Tbuild) – 0.14 × log(ADA) log(Znperch) = 1.78 + 0.0026 × AIx + 0.29 × log(Secchi) log(Znperch) = 2.10 + 0.07 × log(Tbuild) log(Znperch) = 1.90 + 0.0022 × AIx

16 17

43 9

0.97 0.88

6 7

log(Cdcrucian carp) = 2.88 – 1.89 × Vd + 1.21 × log(comm) log(Cdcrucian carp) = –3.01 + 0.0093 × AIx – 0.84 × log(water)

18 19 20

24 9 34

0.97 0.97 0.87

6 7 7

log(Crcrucian carp) = –1.44 + 0.95 × log(Secchi) + 0.53 × log(comm) log(Crcrucian carp) = –1.98 + 1.35 × log(Dm) – 0.002 × AIx log(Crcrucian carp) = –2.03 + 1.17 × log(Dm)

21 22

29 8

0.85 0.69

7 6

log(Cucrucian carp) = 1.31 + 0.47 × log(OL) log(Cucrucian carp) = 1.40 + 0.17 × log(other)

23

64

0.94

6

log(Hgcrucian carp) = –1.26 + 0.22 × log(other)

24 25 26 27

28 13 12 9

0.88 0.73 0.71 0.65

6 7 7 7

log(Nicrucian carp) = –1.39 – 0.25 × log(lnat) log(Nicrucian carp) = –1.77 + 1.02 × log(Dmax) log(Nicrucian carp) = –1.32 + 9.38 × (P-lot) log(Nicrucian carp) = –1.44 + 0.005 × AIx

130


TABLE IV (continued) Eq. no.

F>

r2

n

Regression equation

28 29

15 10

0.76 0.68

7 7

log(Pbcrucian carp) = –1.04 + 10.9 × (P-lot) log(Pbcrucian carp) = –1.60 – 0.85 × log(DR)

30 31 32

17 15 8

0.96 0.75 0.62

6 7 7

log(Zncrucian carp) = 1.54 + 81.1 × (P-house) – 0.19 × log(inst) log(Zncrucian carp) = 1.87 + 69.6 × (P-house) log(Zncrucian carp) = 1.82 + 7.03 × (P-lot)

has a correlation with wetland areas. Ni in perch liver correlates with park, which represents the communications cluster. For Pb, perch only has weak correlations with general urban parameters and Pb in crucian carp has a correlation with the communications cluster. For Zn, both perch and crucian carp liver concentrations correlate with the general urbanity and communication clusters. The positive correlations between Cdperch and Hgperch versus the wetland areas are likely due to an increased catchment transport with increased catchment outflow (≈ wetlands) areas. Metals in outflow areas (≈ wetlands) are generally retained for a shorter time as compared to inflow areas (≈ dry land areas), where the metals have to pass through several vertical soil layers to reach the ground water and finally the lake (Håkanson and Peters, 1995). Mercury is known to be transported to lakes by humic matter related to wetlands (Meili, 1991), so these statistical models agree with and support results obtained from causal analysis. This work demonstrates different urban influences from different types of catchment land-use and also indicates to the role of the lake form factor (Vd) and internal lake processes to determine fish Cd concentrations. For some fish and some metals (perch: Cd, Cu; crucian carp: Pb), the communications cluster is the main determining factor for the metal concentrations while for other it is the general urban parameters (perch: Ni; crucian carp: Cd) and for another yet (crucian carp Ni and Zn), it is the ‘size’ variables.

4. Conclusions This work has quantified the anthropogenic catchment load of metals to ten urban (Stockholm) lakes in relation to the background. Regression models with one ‘size’ and one ‘urban catchment status’ parameter proved to explain very significant parts of the variations in metal transport to the lakes from the catchments.


131

Cu was found to be the most ‘urban’ related metal followed by Cr. Pb had about equal influences from the ‘urban’ and the ‘size’ parameters while Cd, Ni and Zn were more influenced by the ‘size’ than the ‘urban’ parameters. The landuse influences heavy metal concentrations also in fish liver. Interesting correlations have been shown for all the seven metals, and r2 -values greater than 0.9 are found for Cd, Cr, Cu, Hg and Zn for fish liver versus different catchment area parameters. These regressions provide a ranking of the factors influencing metal transport to lakes from urban catchments (catchment ‘size’ and ‘urban status’) and factors influencing biouptake in fish (different types of urban land-use). Such knowledge is essential in remedial contexts, e.g., to identify sources of metal pollution and to get realistic expectations of a given remedy.

Acknowledgements This work was financially supported by the National Swedish Environmental Protection Agency and the Stockholm Environment and Health Protection Administration. We are also grateful for all the help we have received from different people from the local authorities and from the Stockholm and Huddinge fishing associations.

References Andersson, P., Borg, H. and Kärrhage, P.: 1995, Water, Air, and Soil Pollut. 80, 889–892. Campbell, K. R.: 1994, Arch. Environ. Contam. Toxicol. 27, 352–356. Håkanson, L. and Uhrberg, R.: 1981, Undersökningar i Kolbäcksåns vattensystem. XIII. Metaller i fisk och sediment, Natl. Swed. Environ. Prot. Board, SNV PM 1408, Uppsala, 215 p. (in Swedish). Håkanson, L. and Jansson, M.: 1983, Principles of Lake Sedimentology, Springer-Verlag, Berlin, 316 p. Håkanson, L.: 1984, Arch. Hydrobiol. 101, 373–400. Håkanson, L., Andersson, P., Andersson, T., Bengtsson, Å., Grahn, P., Johansson, J.-Å., Kvarnäs, H., Lindgren, G. and Nilsson, Å.: 1990, ‘Measures to reduce mercury in fish’, Final report from the Liming-mercury-cesium project, SNV PM 3818, 189 p. Håkanson, L. and Peters, R. H.: 1995, Predictive Limnology – Methods for Predictive Modelling, SPB Academic Publishing, Amsterdam, 464 p. Håkanson, L. and Lindström, M.: 1997, Ecol. Model. 99, 171–201. Håkanson, L.: 1999, Water Pollution – Methods and Criteria to Rank, Model and Remediate Chemical Threats to Aquatic Ecosystems, Backhuys Publishers, Leiden. Lenat, D. R. and Crawford, J. K.: 1994, Hydrobiologia 294, 185–199. Lindström, M.: 2001, Water, Air, and Soil Pollution: Focus 1(3–4), 103–118. Lindström, M. and Håkanson, L.: 2000, ‘A model to calculate heavy metal load to lakes dominated by urban runoff and diffuse inflow’, Manuscript, Inst. of Earth Sci., Uppsala University (submitted). Lohm, U., Bergbäck, B., Hedbrant, J., Jonsson, A., Svidén, J., Sörme, L. and Östlund, C.: 1997, Databasen Stockhome. Flöden och accumulation av metaller i Stockholms teknosfär, 138 p. (in Swedish).

132


Malmqvist, P.-A.: 1983, Urban Stormwater Pollutant Sources, 371 p. Meili, M.: 1991, ‘Mercury in Boreal Forest Lake Ecosystems’, Doctoral Dissertation, Uppsala University. Pitt, R, Field, R., Lalor, M. and Brown, M.: 1995, Wat. Environ. Res. 67, 260–275. Salomons, W. and Förstner, U.: 1984, Metals in the Hydrocycle, Springer, Heidelberg, 349 p. SNV: 1996, Metaller i Stad och Land – kretslopp och kritisk belastning, Lägesrapport 1996, 65 p. (in Swedish). Tarvainen, T., Lahermo, P. and Mannio, J.: 1997, Water, Air, and Soil Pollut. 94, 1–32. Uhrberg, R.: 1982, Anal. Chem. 54, 1906–1908.