Spatial Variation of Streamwater Chemistry in Two Swedish Boreal

Jan 22, 2005 - Streamflow generation in a large, alpine watershed in the southern Rocky Mountains of Colorado: Is streamflow generation simply the ...
0 downloads 3 Views 296KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 1463-1469

Spatial Variation of Streamwater Chemistry in Two Swedish Boreal Catchments: Implications for Environmental Assessment J O H A N T E M N E R U D * ,† A N D KEVIN BISHOP‡ Man-Technology-Environment Research Centre, Department of Natural Sciences, O ¨ rebro University, O ¨ rebro, Sweden, and Department of Environmental Assessment, Swedish University of Agricultural Sciences, Uppsala, Sweden

To evaluate the scale-dependent spatial variability of water chemistry within two Swedish boreal catchments (subcatchment areas 0.01-78 km2), samples were taken at every junction in the stream network during June 2000 and August 2002. The values of most chemical constituents spanned more than an order of magnitude, and the range was similar to that found in all of Northern Sweden by the national stream survey in 2000. According to the official assessment tools used in Sweden, the entire range of environmental status (for pH, absorbance, alkalinity, dissolved organic carbon (DOC)) and human acidification influence existed within these two study catchments. The water chemistry parameters were relatively stable at catchment areas greater than 15 km2. Sampling at that scale may be adequate if generalized values for the landscape are desired. However the chemistry of headwaters, where much of the stream length and aquatic ecosystem is found would not be characterized. Map parameters correlated to the variability in a key chemical parameter, DOC, but the best predictive map parameters differed markedly between catchments. This study highlights the importance of accounting for headwater spatial variability in environmental assessments of running waters, even in relatively pristine areas. The nature of drainage networks with many headwaters and progressively fewer downstream watercourses makes this a considerable challenge.

Introduction Water chemistry in running waters is generally evaluated on the basis of sampling at a limited number of points in space and time. A great deal of effort has been made to evaluate how discrete points in time are related to temporal variability of water chemistry (1-3). Another difficult challenge is how these discrete observations in space represent running waters in a landscape perspective where much of the stream length (and thus aquatic environment) is upstream from the observation points (4-6). Previous investigators have found considerable variability at points sampled across the landscape (7-10). This raises questions about how to conduct * Corresponding author phone: +46-19-301065; fax: +46-19303169; e-mail: [email protected]. Present address: ManTechnology-Environment Research Centre, O ¨ rebro University, SE70182 O ¨ rebro, Sweden. † O ¨ rebro University. ‡ Swedish University of Agricultural Sciences. 10.1021/es040045q CCC: $30.25 Published on Web 01/22/2005

 2005 American Chemical Society

environmental assessments which often want to characterize an entire landscape with regards to water chemistry and the aquatic ecosystem. This became apparent when an effort was made to use the Swedish Environmental Quality Criteria for Lakes and Watercourses (11) on data from a national survey of lakes and stream preformed, in year 2000, to characterize the acidification status of Swedish watercourses. The smallest catchment area in this survey was 15 km2, and subsequent analysis revealed that approximately half of the entire stream length was in catchments smaller than this (12). Not characterizing the status of these headwaters is a serious concern that contributed to the conclusion that an acidification assessment for streams could not be made at this point due to a lack of information (11). The European Union’s Water Framework Directive (WFD) will soon mandate such assessments for all surface waters in the EU (Dir 2000/60/EG). Many researchers have tried to identify a “representative elementary area” (REA) at which variation between nearby watercourses is reduced, so that a limited number of samples can represent the catchment area (13-15). The water chemistry signal from catchment areas smaller than the size given by the REA is implicitly regarded as “noise” to be ignored. In a landscape perspective, though, it is not at all clear that the headwater variability concealed by moving downstream to sample a REA is “noise” that should be disregarded. That “noise” could be the different habitats that are a fundamental part of the region’s biodiversity and ecological integrity (16). Since the capillary network of headwater streams comprises a large component of the aquatic environment, a “representative” downstream site will ignore this in assessing the aquatic ecosystems of a landscape. Even though small streams may not be suitable habitat for some species of special interest in watershed management (e.g. sport fisheries) they can still be a vital component of the downstream ecosystem that can support such species. This is a key message of the River Continuum Concept which emphasizes how the physicochemical structures of headwaters create habitats for specific communities which influence aquatic communities further downstream (17, 18). The large spatial and temporal variation in the water chemistry of headwaters compared to downstream was recognized long ago in Sweden (19). There are a variety of processes contributing to this change in water chemistry with size of stream. Downstream changes in water chemistry can be due to in-stream processes (chemical and biological), change of soil types, land-use and runoff generation processes along the stream (6, 20-23). The overriding factor in the reduction of variability between nearby streams as catchment size increases is the mixing of different sources as one moves downstream, in the absence of any systematic, landscape scale patterns. Despite the recognition of headwater variability, few studies have addressed the issue of how to characterize this variability in landscape-scale assessment strategies, and the need for such information from the boreal forest region has been identified (24). Predicting the water chemistry variability and designing efficient monitoring programs will require much more knowledge of small-scale variability and scalerelated structures in water chemistry than is currently available. A starting point for such work is knowledge of how samples at one or a few points in the catchment, complemented by map information, relates to the spatial distribution of water chemistry within the network. VOL. 39, NO. 6, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1463

TABLE 1. Percentage Soil and Land-Use at O, with Branches O1 and O2, and Sa Parameter

O

O1

O2

S

Peat Clay Silt Sand Glacial fluvium Till Rock/ thin soil Water Mixed forest Clear-cuts Other open land Arable Wetland Water

19.7 0.5 7.5 3.1 0.0 60.3 5.0 3.9 67.0 7.8 1.3 1.6 18.3 4.1

18.6 0.0 4.5 8.3 0.0 58.4 6.1 4.2 70.3 6.7 0.3 0.0 18.3 4.3

26.4 0.0 0.8 0.0 0.0 68.4 2.5 2.0 62.9 9.5 0.1 0.1 25.4 1.9

17.7 0.0 0.2 0.0 0.9 76.3 4.6 0.3 82.3 n.d. n.d. 0.1 17.3 0.3

a Soils (first 8 rows) are based on 1:50,000 scale soil maps (year 2000). Land-use (the 6 last rows) is based on 1:20,000 scale topographic maps (year 2001). Mixed forest is mostly coniferous trees. n.d. ) no data.

FIGURE 1. Map of the River O2 re basin location in Sweden. The two investigated catchments Ottervattsba1 cken (O), (with the western branch Hammonsba1 cken (O1) and the eastern branch Marraba1 cken (O2)), and So1 rba1 cken (S). Open circles ) samplings sites. Broad solid line ) surface water. In this study we sampled all the headwater streams and stream confluences in two relatively pristine Swedish boreal catchments under base flow conditions. Low flow conditions are representative of the situation experienced by the biota during most of the year (even though episodes of high flow are of great ecological significance). These catchments were chosen based on earlier surveys of water chemistry variability in these and at 10 other nearby “downstream” watercourses. Even though many chemical parameters were analyzed, this paper focuses on parameters routinely used in Sweden’s surface water monitoring program that can be assessed using the Swedish Environmental Quality Criteria (EQC). These parameters include pH, alkalinity, absorbance and dissolved organic carbon (DOC). In this region aquatic humic substances, measured as DOC, are dominant features of the water chemistry that strongly influence the pH, buffering capacity, nutrient concentrations, and bioavailability/toxicity of metals and organic pollutants (2, 25-27). DOC has a large impact on the biota as well (28, 29). Since DOC is one of the most important determinants of water chemistry and ecological status in boreal surface waters, the ability to predict the variability in this parameter using GIS data was explored. Study Area.The two catchments, Ottervattsba¨cken (78 km2) and So¨rba¨cken (63 km2) are located in the O ¨ re River basin, northern Sweden (Figure 1), with their outlets 50 km apart. Till is the dominant soil material (>60%) in these catchments, followed by peat 17-26% (Table 1). The most 1464

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 6, 2005

common soil type is podzol (with an average humus layer thickness of 7 cm), followed by wetlands. The riparian zone is often comprised of peat of varying thickness (30). The mean annual temperature is 1.0 °C, the precipitation is 650 mm/year (30-45% as snow) with an annual average discharge of 350 mm year and a calculated evapotranspiration of 300 mm year (31). Forests (approximately 82%) and mires (approximately 18%) dominate the landscape of these catchments, and there is almost no agriculture (Table 1). The dominant tree species are mixed stands of Norway spruce (Picea abies) and Scots pine (Pinus silvetris) with a minor contribution of hardwoods, mainly birch (Betula pubescens). Approximately 50% of Sweden’s area is covered by forest, and most of this forested area (70%) is coniferous (32). The study area represents a common landscape in Sweden. There is little overt human influence beyond low-intensity forestry. The pH is low (6.8 >0.2

0.02-0.05 10-25 4-8 6.5-6.8 0.10-0.20

0.05-0.12 25-60 8-12 6.2-6.5 0.05-0.10

0.12-0.2 60-100 12-16 5.6-6.2 0.02-0.05

>0.2 >100 >16 e5.6 e0.02

meq/L

Color was not analyzed in this survey but is included for comparison.

types (30). The whole catchment is above the highest postglacial coastline. Compared to O, the S catchment has a lower percentage of lakes (0.3%) than O (Table 1). About the same percentage of peat 18%, less silt (0.2%) and more till (76%) than O.

Methods A sample was taken approximately 10 m upstream from almost every stream junction (90% as seen on the national 1:50,000 topographic maps). In this article headwaters are the same as stream order 1 (35). Higher order streams are denoted “downstream”. At the beginning and at the end of the sampling day, runoff from the entire catchment was measured at a location where a stage-discharge relationship had been established (Hans Ivarsson, personal communication). The stream network of each catchment was sampled during 14th-20th June 2000, and during 19th-22nd August 2002 the O catchment was resampled. These periods of low discharge were chosen to maximize the likelihood of stable flow. This facilitates catchment mass balances and represents the hydrological situation during most of the year. A total of 61 stream junctions were sampled in Ottervattsba¨cken (O) year 2000, 33 from O1 (17 of which were headwaters) and 25 were from O2 (of which 11 were headwaters). In 2002 48 sites were resampled (the other sites were dry) plus 19 new sites (mostly downstream from the year 2000 headwater sampling sites), making for a total of 67 sites. For So¨rba¨cken (S) a total of 42 stream junctions were sampled year 2000, of which 21 were headwaters. So¨rba¨cken was not sampled year 2002. 1:20000 maps of topography and 1:50000 soil maps were used when 32 map parameters for each subcatchment were calculated using GIS. Spearman rank correlation was used to define the degree to which those map parameters could be used to predict the variability of DOC-concentrations. Chemical Analysis. Two bottles were collected at each location. One was a 500 mL dark glass bottle used for analysis of pH, electrical conductivity and absorbance. The other was a 1 L polyethylene flask from which aliquots were taken for measurement of other chemical parameters that were analyzed up to one month later. The stream temperature was measured in situ on the day of collection. The pH at 20 °C was measured with an electrode designed for low ionic strength (Orion model 9272). Absorbance at 254 and 420 nm wavelength was measured in a 5 cm quartz cuvette and on unfiltered samples (Hitachi U-1100). Electrical conductivity was measured at 20 °C (SDM 2010). Cl-, NO3-, SO42- and Fwere analyzed using capillary electrophoresis (CE) according to Romano et al. (36). Metals were analyzed using ICP-MS on HNO3 acidified samples. Alkalinity for year 2000 was titrated to two end-points (alktwp) of pH 4.5 and 4.2, according to the method of Ko¨hler et al. (25). For year 2002 the alkalinity (alk5.6) was measured according to the Swedish standard, pH ) 5.6. Alktwp was converted to alk5.6 using TOC (total organic carbon) and charge balance (25). In 2000 the samples for TOC and IC (inorganic carbon) analysis were made on frozen (-20 °C) samples and analyzed with a Shimadzu TOC-5000

at Lund University. For year 2002 samples were stored at 4 °C and analyzed on a Shimadzu TOC-V within a week. It has earlier been shown that DOC and TOC in the region of the O ¨ re River differ by less that 5% (25, 37), so the TOC values are referred to as DOC in the text. One commonly used characterization parameter of DOC is SUVA, Specific Ultra-Violet Absorbance, which is the absorbance at 254 nm divided by the amount of DOC (mg/ L). Gel filtration was performed with an HP Series 1100 (HPLC) equipped with TosoHaas column (TSKgel G 2000 SW, cutoff 1000 Dalton), at a flow rate of 1.0 mL/min and UV-detection at 254 nm and fluorescence detection (excitation 250 nm and emission 410 nm). Prior to the analysis the samples were filtered (0.22 µm, Osmonics) and the filtrate was diluted with running buffer (50 mM phosphate, pH 6.8). Polystyrene sulfonate standards (American Polymer Standard Corporation) were used for calibration of molecular weight (1200, 4400, 5900, 14800, 28200, 37500 and 60000 Dalton). Status Classification According to SEPA. The Swedish Environmental Quality Criteria (EQC) for Lakes and Watercourses (11) classifies surface waters on an integer scale from 1 to 5. The EQC make two separate assessments, one of Current Status, and the other of Human Influence. The class boundaries for Current Status are based on a mixture of statistical and subjective considerations to reflect the distribution of constituents in Swedish surface waters (11). Class 1 is for low values for TOC, color and absorbance, but high values of pH and alkalinity (Table 2). The assessment of Human Influence represents how far such influence has moved the observed situation from the natural situation. In this study, we evaluate Human Influence with regards to acidification. This assessment is based on a model that uses contemporary water chemistry and acid deposition to estimate changes in acidity status resulting from acid deposition, changes in alkalinity, and taking into account sulfate deposition (11). Human Influence class 1 ) Insignificant differences between the present and the reference conditions of pH and alkalinity, while 5 ) Extremely great differences. In their current form, EQC are intended as guidelines, but they are being used as the basis for defining the legally binding requirements of the European Union’s Water Framework Directive (WFD) in Sweden. The ambitious ecological goals of the new EU WFD also mandate assessment of all European waters, without specific reference to any minimum catchment size (Dir 2000/60/EG).

Results The headwaters displayed a large variation in water chemistry (Figure 2). Downstream, when the total catchment size surpassed approximately 15 km2, the concentrations were much less variable (Figure 2). For DOC the 5 and 95 percentiles (denoted just percentiles hereafter) in catchments < 15 km2 were 7.8 and 28 mg/L (median 16 mg/L, n ) 78), while the percentiles for larger catchments were 10 and 24 mg/L (median 12 mg/L, n ) 25) (Table 3). If only stream order 1 samples were used the percentiles are 5.5 and 31 mg/L (median 17 mg/L, n ) 49) VOL. 39, NO. 6, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1465

(Figure 2). According to a model of DOC acid/base properties (38) there was no difference in these properties with catchment size (data not shown). This is consistent with earlier findings that this model worked well across a range of lakes, streams, different flow conditions and seasons (25). The absolute range of variability was higher in the August 2002 sampling than in the same sample sites from 2000 when flows were not as low. For Ottervattsba¨cken (O) the measured average discharge was 0.47 mm/day ((0.20 mm, n ) 8) during the sampling 17th-20th of June 2000, while during 19th-22nd August year 2002 the mean discharge was 0.18 mm/day ((0.09 mm, n ) 5). The measured discharge averaged for So¨rba¨cken (S) was 0.54 mm/day ((0.22 mm, n ) 8) during sampling June 14th-18th year 2000. Both sampling periods were done during low water flow in the catchments, and year 2002 was driest. The relative ranking of concentrations for DOC, pH, alkalinity and absorbance between the different years, however was generally consistent (tested using Spearman rank of correlations, data not shown). This is exemplified by DOC (Figure 4). Spearman rank correlations stronger than (0.5, for at least one catchment, were found for 7 of 32 studied map parameters (Table 4, other data not shown). Percentage of peat correlation with DOC varied between 0.38 and 0.61. The better predictive parameters varied markedly between the catchments.

Discussion

FIGURE 2. Water chemistry in June 2000 as a function of catchment size. Ottervattsba1 cken O (0), with branches O1 (O) and O2 (4), and So1 rba1 cken, S (+). Black dots are the stream order. The 15 km2 minimum size of watercourse catchments in Sweden’s national inventory of stream chemistry year 2000 is denoted by the vertical gray line. Horizontal gray lines are the SEPA EQC (11) status class boundaries, see Table 2. (Table 3). The same pattern was found for pH, alkalinity and absorbance (420 nm), i.e., a higher variability in headwaters than downstream (Table 3). This pattern of variation was reflected in the Swedish EQC classification of status (gray horizontal lines in Figure 2) and degree of human influence (Figure 3). The full range of Current Status (pH, alkalinity, absorbance and DOC) and Human Influence (using alkalinity) was found in headwaters, while considerably fewer classes were found downstream. Catchment sizes smaller than 15 km2 were stream order 1 and 2, with only some stream order 3 (Figure 2). While the variability was always less downstream, the average values were not necessarily different. Electrical conductivity showed no difference between headwaters and downstream (which was generally low with an average of 31 µSm/cm). The quality aspects of DOC investigated in this study, SUVA and apparent molecular weight, did not show systematic changes with increasing subcatchment size (data not shown). Other parameters, however, did show differences in the mean values between headwaters and downstream. Temperature, for instance, was higher downstream 11.4 °C, compared to 9.2 °C in headwaters (Table 3). The mean headwater concentrations of DOC were higher than those at the outlet of each study catchment (Figure 2 and Table 3). However, the catchments O1, O2 and S exhibit three different patterns of DOC concentration change with increasing subcatchment size: from the greatest DOC decrease at O1, to almost no change at O2, while S is between 1466

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 6, 2005

Streams are the capillary network that embeds freshwater ecosystems in the landscape. Assessment of the stream network presents special challenges, as is illustrated by this study. The earlier data from watercourses in the range of 21 to 2939 km2 from the O ¨ rtra¨sk basin gave some indication that smaller streams were more variable and had higher concentrations of DOC (Figure 5). The sampling of all stream confluences in two catchments from that earlier study (where just the outlets had been sampled) revealed that the headwaters were much more variable than the data from Figure 5 would indicate if there was a linear increase in variability as catchment size decreased. In fact, the range of DOC, absorbance, pH and alkalinity found in the headwaters was almost as great as seen in the entire forested area of northern Sweden by the national survey of watercourses conducted in 2000 (Figure 6). Furthermore, the entire range of Current Status classes for DOC, pH and alkalinity and Human Influence classes of acidification were found in the two catchments (O+S, Figure 3) N.B. There is no classification system established for evaluating the Human Influence on DOC. Awareness of spatial variability is not new. Already in the first decades of the 20th century, the special character of headwaters had been noted in Sweden (19). Nonetheless, how to deal with this variability in monitoring and assessment has yet to be addressed. When the Swedish Environmental Protection Agency recently attempted to determine the acidification status of running waters using national monitoring data, the conclusion was that considerably more information about streams was needed before this assessment could be made (39). The aim of Sweden’s national inventory is to investigate the spatial pattern in the environment, as a complement to more temporally intensive sampling on selected sites. In Sweden there are roughly 395,000 km of streams as measured on the national 1:250,000 maps. The national survey measures some 800 points on this network, but only a few with an upstream catchment area smaller than 15 km2 (40). A large part of all streams length is thus overlooked in the national survey. In this study the headwaters, stream order 1, made up approximately 40% of the total catchment area. The relationship between stream order, using 1:50,000 maps, and catchment size (see Figure

TABLE 3. Median with 5 and 95 Percentiles (Italics) for Stream Temperature, Dissolved Organic Carbon (DOC), Alkalinity, pH and Absorbance at 420 nm (5 cm), Grouped by Catchment Size and Stream Order (Str ord)a Catchment size class

n

Str. temp. °C

DOC mg/L

Alkalinity meq/L

pH

420 nm 5 cm

15 km2 Str ord 1 Str ord 2-4

78 25 49 54

9.6 (6.6/14) 12 (10/15) 9.2 (6.5/13) 11.4 (8.4/15)

16 (7.8/28) 12 (10/24) 17 (5.5/31) 13 (9.0/26)

0.051 (-0.082/0.22) 0.15 (0.027/0.22) 0.053 (-0.10/0.22) 0.060 (0.0050/0.22)

5.9 (4.5/6.8) 6.7 (5.6/7.2) 5.8 (4.2/6.9) 6.2 (5.1/7.1)

0.29 (0.12/0.69) 0.23 (0.18/0.53) 0.30 (0.09/0.79) 0.23 (0.16/0.56)

km2

a

Stream order 1 ) headwaters. n ) number.

FIGURE 3. Classification of human influence with regards to acidification, based on alkalinity, according to SEPA EQC (11). Data from Ottervattsba1 cken O (0), with branches O1 (O) and O2 (4), and So1 rba1 cken, S (+) year 2000, n ) 98. Human Influence class 1 ) Insignificant, while 5 ) Extremely great difference between present value and the reference value. The 15 km2 minimum size of watercourse catchments in Sweden’s national inventory of stream chemistry year 2000 is denoted by the vertical gray line.

TABLE 4. Spearman Rank Correlations for the 7 of 32 Map Parameters that Had Correlations to Observed DOC Concentrations Which Were Stronger than ( 0.5 for at Least One Catchmenta Catchment

FIGURE 4. Ranking of DOC concentrations for year 2000 (x-axis) versus year 2002 (y-axis) at the same sampling site for O (0), with branches O1 (O) and O2 (4). Rank ) 1 is the highest DOC concentration, while 48 is the lowest concentration that year. Solid line is the one-to-one line with 2000 as reference year. 2) indicates that the Swedish National Survey in 2000 mostly sampled streams which had a stream order 3 or greater. The current Swedish monitoring strategy with 15 km2 as the lower limit in catchment size gives a relatively “representative” picture of the landscape, in that much of the variability upstream is hidden. This is consistent with a sampling strategy based on the idea of an REA where smallscale variability is often considered as “noise” to be suppressed (15). Even if much of the aquatic features and species valued by society are found downstream from the 15 km2 boundary for sampling, the headwater environment influences the downstream ecosystem (17), and provides much of the water and dissolved constituents. The water chemistry experienced by the biota in the headwaters is much more variable than downstream. In this study we have only focused on the water chemistry, but in other boreal streams, the number and type of species is likely to differ between

Parameter

O

O1

O2

S

Catchment size (km2) Lakes % Peat % Thin soil % Rock % Stream Density Average Elevation (m asl)

-0.37 -0.48 0.61 -0.68 -0.70 0.38 -0.42

-0.61 -0.81 0.52 -0.80 -0.81 0.73 -0.70

0.13 0.35 0.38 -0.11 -0.11 -0.40 0.23

-0.08 0.08 0.51 -0.34 -0.14 -0.24 0.02

a The 25 parameters where the correlations were less than 0.5 were Forest%, Clear-cuts%, Open Land%, Arable%, Wetland-Forest%, Wetland-Open%, Wetland-inaccessible%, Wetland Total%, Water+Wetland Total%, Clay%, Silt%, Sand%, Glacial fluvium%, Till%, 1-2-5 and 10 m broad zones of Total wetland % along Stream Length, above Highest Post-Glacial Coast Line %, Stream reach (m), Lake length (m), Total length (m), Stream Length/ direct distance, Slope, Slope direct distance. Stream density ) Stream Length (m) divided with Catchment size (km2).

headwaters and downstream (16). The headwaters could have species that do not occur downstream. Within the next decade, the European Union’s Water Framework Directive will mandate assessments of all waters, with no lower limit in catchment area currently specified. Knowledge of how the headwaters relate to what is observed downstream will be essential in devising strategies for assessing the multitude of headwaters. Strategies for assessing water chemistry in landscapes would benefit from knowledge of consistent patterns in the relationship between headwaters and downstream. While some properties did not change downstream such as electrical conductivity and SUVA, others did. Specifically, median DOC-concentrations and absorbance were higher in headwaters than at the outlet. Headwaters in the investigated region were expected, on average, to have higher DOC than downstream due to several factors including: the frequency of mires along headwaters, lakes VOL. 39, NO. 6, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1467

FIGURE 5. Open diamonds are average DOC from 12 catchments in the River O2 re Basin, each sampled between 5 and 30 times during the month of June 1991 and June 1992. Maximum and minimum values are the short black horizontal line, from (Hans Ivarsson and Mats Jansson, personal communications). Crosses are one sampling day in June 1996 (Catharina Pettersson, personal communication). The 15 km2 minimum size of watercourse catchments in Sweden’s national inventory of stream chemistry year 2000 is denoted by the vertical gray line. Horizontal gray lines are the status class boundaries for DOC (11), see Table 2.

FIGURE 6. A comparison of water quality in this study (black crosses, year 2000, number ) 103) and Sweden’s national inventory of stream chemistry year 2000 (gray circles, number ) 163). The vertical gray line at 15 km2 represents the minimum catchment size in the national inventory. The points plotted from the national inventory were from Sweden’s northern counties using all samples collected below tree line, with arable and urban land under 1%, and catchment size < 350 km2. Horizontal gray lines are the SEPA EQC (11) status class boundaries, see Table 2. (clarifying effects), in-stream/hyporheic biological geochemical processes that decrease the DOC concentrations, and the effect of inflowing groundwater with low concentrations of DOC downstream. The degree of difference from headwater to outlet, however, varied between the O1, O2 and S as exemplified by DOC (Table 3, Figure 2). Alkalinity and pH, in contrast to DOC, were generally lower in the headwaters (Table 3, Figure 2). Alkalinity and pH increases with catchment size are due to both lower DOC levels (and thus less 1468

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 6, 2005

organic acids) as well as more older groundwater entering the watercourses with more weathering products. One strategy for addressing the challenge of spatial variability would be to find map parameters that correlated to water chemistry. This would facilitate predictions of the variability across the landscape in water chemistry from the variation in map parameters. As has been found for other studies (21, 41, 42) correlations were indeed found between one of the key water chemistry parameters for this site (DOC) and map parameters. Peat percentage, cited as a good predictor of DOC in these other studies (21, 41, 42) had Spearman correlations between 0.38 and 0.61 on all 3 catchments (Table 4). Better correlations, however, were found for other parameters. The parameters that correlated best on the different catchments were also different, despite the proximity of the catchments and outward similarities. For instance percentage of lakes had a Spearman correlation of -0.81 (negative) on catchment O1, but 0.35 (positive) on O2 and 0.08 (none) on S (Table 4). A further complication in the interpretation of spatial surveys is that water chemistry varies with time, generally correlating to variation in discharge. Three days of sampling for each catchment were conducted during stable weather conditions, indicated by stable discharge. Any time (discharge) effect on the spatial variation is therefore considered to have had a negligible effect in this study, even though that must be considered in operational assessment when the data is collected across a variety of flow conditions. The two sampling occasions, during low flow conditions, for the Ottervattsba¨cken catchment also indicated that there was a consistency over time in the ranking of water chemistry parameters between the different headwaters (Figure 4). The concentration values, and thus the pattern in the landscape, would be different at high flow conditions. The extremes at high flow may be critical thresholds for the aquatic life, but the low flow conditions sampled in this study represent the situation during most of the year. In the forest landscape of Sweden, sampling at a scale of stream order 3 (or excluding catchments smaller than 15 km2) may be adequate if generalized values are desired for the landscape. The situation in the headwaters, however, would not be characterized. This is a serious omission because a large part of the stream length is found in these small streams. This study highlights the importance of accounting for spatial variability in assessments of running waters, even in relatively pristine areas. The nature of drainage networks with many headwaters and progressively fewer downstream watercourses complicates the assessment of a suitable catchment size, and there are indications that simple correlations to map information will not work across even a relatively uniform landscape. This, however, is a challenge that must be met for both national assessment, and international obligations under the EU Water Framework Directive. In this study, some patterns were observed, but much more such data is required to devise regional sampling strategies, and the ways to use GIS data to support those strategies.

Acknowledgments Financial support for this research was provided by the Knowledge Foundation. We would like to thank our fellow collaborators for excellent field- and laboratory work: Rose Cory, Tobias Eriksson, Evastina Grahn, Matthias Heinz, Ulf Juto, Jan Seibert, Anna Stenberg and Rasmus So¨rensen. Thanks to Jakob Nisell, Stephan Ko¨hler and Martin Erlandsson for good data processing. Many thanks also to Mats Jansson with co-workers and Anders Lo ¨ fgren with family for making it possible to use the “O ¨ rtra¨sk laboratory facilities”. The manuscript improved after we received valuable com-

ments from Bert Allard, Ishi Buffam, Anders Du ¨ ker, Stefan Karlsson, Stefan Lo¨fgren, and Johan To¨rnblom.

(23) Dawson, J. J. C.; Bakewell, C.; Billett, M. F. Is in-stream processing an important control on spatial changes in carbon fluxes in headwater catchments? Sci. Total Environ. 2001, 265 (1-3), 153167.

Literature Cited

(24) Findlay, S.; Sinsabaugh, R. L. Unravelling the sources and bioavailability of dissolved organic matter in lotic aquatic ecosystems. Mar. Freshwater Res. 1999, 50 (8), 781-90.

(1) Boyer, E. W.; Hornberger, G. M.; Bencala, K. E.; McKnight, D. M. Response characteristics of DOC flushing in an alpine catchment. Hydrol. Process. 1997, 11 (12), 1635-1647. (2) Sande´n, P.; Karlsson, S.; Du ¨ ker, A.; Ledin, A.; Lundman, L. Variations in hydrochemistry, trace metal concentration and transport during a rain storm event in a small catchment. J. Geochem. Explor. 1997, 58 (2-3), 145-155. (3) Mulholland, P. J.; Hill, W. R. Seasonal patterns in streamwater nutrient and dissolved organic carbon concentrations: Separating catchment flow path and in- stream effects. Water Resour. Res. 1997, 33 (6), 1297-1306. (4) Grieve, I. C.; Marsden, R. L. Effects of forest cover and topographic factors on TOC and associated metals at various scales in western Scotland. Sci. Total Environ. 2001, 265 (1-3), 143-151. (5) Eckhardt, B. W.; Moore, T. R. Controls on dissolved organic carbon concentrations in streams, southern Quebec. Can. J. Fish. Aquat. Sci. 1990, 47 (8), 1537-1544. (6) Aitkenhead, J. A.; Hope, D.; Billett, M. F. The relationship between dissolved organic carbon in streamwater and soil organic carbon pools at different spatial scales. Hydrol. Process. 1999, 13 (8), 1289-1302. (7) Sharp, W. E. A Topologically optimum water-sampling plan for rivers and streams. Water Resour. Res. 1971, 7 (6), 1641-1646. (8) Millard, S. P.; Yearsley, J. R.; Lettenmaier, D. P. Space-time correlation and its effects on methods for detecting aquatic ecological change. Can. J. Fish. Aquat. Sci. 1985, 42, 13911400. (9) Loftis, J. C.; McBride, G. B.; Ellis, J. C. Considerations of scale in water quality monitoring and data analysis. Water Res. Bul. 1987, 27 (2), 255-264. (10) Warry, N. D.; Hanua, M. The use of terrestrial ecoregions as a regional-scale screen for selecting representative reference sites for water quality monitoring. Environ. Man. 1993, 17 (2), 267276. (11) SEPA. Environmental quality criteria for lakes and watercourses; Report 4913, Swedish Environmental Protection Agency, Stockholm, Sweden, 2000. (12) Erlandsson, M. Assessment of acidification status in Swedish watercourses; Internal Report ISSN 1401-5765, Department of Environmental Assessment, Swedish University of Agricultural Sciences, Uppsala, Sweden (in Swedish with English summary), 2003. (13) Blo¨schl, G.; Grayson, R. B.; Sivapalan, M. On the representative elementary area (REA) concept and utility for distributed rainfallrunoff modelling. Hydrol. Process. 1995, 9 (3-4), 313-330. (14) Wolock, D. M.; Fan, J.; Lawrence, G. B. Effects on basin size on low-flow stream chemistry and subsurface contact time in the Neversink river watershed, New York. Hydrol. Process. 1997, 11 (9), 1273-1286. (15) Reggiani, P.; Schellekens, J. Modelling of hydrological responses: the representative elementary watershed approach as an alternative blueprint for watershed modelling. Hydrol. Process. 2003, 17 (18), 3785-3789. (16) Malmqvist, B. Aquatic invertebrates in riverine landscapes. Freshwater Biol. 2002, 47 (4), 679-694. (17) Vannote, R. L.; Minshall, G. W.; Cummins, K. W.; Sedell, J. R.; Cushing, C. E. The river continuum concept. Can. J. Fish. Aquat. Sci. 1980, 37 (1), 130-137. (18) Newbold, J. D.; Mulholland, P. J.; Elwood, J. W.; O’Neil, R. V. Organic carbon spiralling in stream ecosystems. Oikos 1982, 38 (3), 266-272. (19) Eriksson, J. V. Den kemiska denudationen i Sverige (The chemical denudation of Sweden); Report Band 5, nr 3, Swedish Meteorological and Hydrological Institute, Stockholm, Sweden (in Swedish with French summary), 1929; p 69. (20) Bilby, R. E. Role of organic debris dams in regulating the export of dissolved and particulate organic matter from a forested watershed. Ecology 1981, 62 (5), 1234-1243. (21) Hope, D.; Billett, M. F.; Cresser, M. S. A review of the export of carbon in river water - fluxes and processes. Environ. Pollut. 1994, 84 (3), 301-324. (22) Schindler, J. E.; Krabbenhoft, D. P. The hyporheic zone as a source of dissolved organic carbon and carbon gases to a temperate forested stream. Biogeochemistry 1998, 43 (2), 157174.

(25) Ko¨hler, S.; Hruska, J.; Bishop, K. Influence of organic acid site density on pH modelling of Swedish lakes. Can. J. Fish. Aquat. Sci. 1999, 56 (8), 1461-1470. (26) Jansson, M.; Bergstro¨m, A. K.; Drakare, S.; Blomqvist, P. Nutrient limitation of bacterioplankton and phytoplankton in humic lakes in northern Sweden. Freshwater Biol. 2001, 46 (5), 653-666. (27) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; John Wiley & Sons: Hoboken, New Jersey, USA, 2003; pp 314-325. (28) Giller, P. S.; Malmqvist, B. The Biology of Streams and Rivers, Biology of Habitats; Oxford University Press: Oxford, Great Britain, 1998. (29) Jansson, M.; Bergstro¨m, A. K.; Blomqvist, P.; Isaksson, A.; Jonsson, A. Impact of allochthonous organic carbon on microbial food web carbon dynamics and structure in Lake Ortrasket. Arch. Hydrobiol. 1999, 144 (4), 409-428. (30) Ivarsson, H.; Karlsson, L.-I. Geological and geochemical conditions in the River O ¨ re drainage basin, northern Sweden; Gerum report nr 15; Geography Department, Umeå University: Umeå, Sweden, 1992. (31) Alexandersson, H.; Karlstro¨m, C.; Larsson-McCann, S. Temperature and precipitation in Sweden 1961-90, Reference normals; Report Meteorologi nr 81; Swedish Meteorological and Hydrological Institute: Norrko¨ping, Sweden (in Swedish with English summary), 1991. (32) Marklund, K. (Ed). National Encyclopedia; Bra Bo¨cker Fo¨rlag: Ho¨gana¨s, Sweden (in Swedish), 1995. (33) Laudon, H.; Westling, O.; Bishop, K. Cause of pH decline in streamwater during spring melt runoff in northern Sweden. Can. J. Fish. Aquat. Sci. 2000, 57 (9), 1888-1900. (34) Bertills, U.; Lo¨vblad, G. (Eds). Kritisk belastning fo¨r svavel och kva¨ve. (Critical loads of sulphur and nitrogen); Report 5147; Swedish Environmental Protection Agency: Stockholm, Sweden (in Swedish), 2002. (35) Strahler, A. N. Quantitative analysis of watershed geomorphology. Trans. Am. Geophys. Union 1957, 38 (6), 913-920. (36) Romano, J.; Jandik, P.; Jones, W. R.; Jackson, P. E. Optimization of inorganic capillary electrophoresis for the analysis of anionic solutes in real samples. J. Chromatogr. 1991, 546 (1-2), 411421. (37) Ivarsson, H.; Jansson, M. Regional variation of dissolved organic matter in running waters in central northern Sweden. Hydrobiologia 1994, 286 (1), 37-51. (38) Ko¨hler, S.; Hruska, J.; Jo¨nsson, J.; Lo¨vgren, L.; Lofts, S. Evaluation of three organic equilibrium models using an in-situ titration of an organic rich stream in northern Sweden. Water Res. 2002, 36 (18), 4487-4496. (39) SEPA. Miljo¨kvalitetsmål 7, Bara naturlig fo¨rsurning (Environmental quality goal No. 7, Only natural acidification); Report 5000; Swedish Environmental Protection Agency: Stockholm, Sweden (in Swedish), 2003. (40) Wilander, A.; Johnson, K. R.; Goedkoop, W. Riksinventering 2000. En synoptisk studie av vattenkemi och bottenfauna i svenska sjo¨ar och vattendrag (National survey 2000. A synoptic study of water chemistry and benthic fauna in Swedish lakes and watercourses); Report 2003:1; Department of Environmental Assessment, Swedish University of Agricultural Sciences: Uppsala, Sweden (in Swedish), 2003. (41) Laudon, H.; Ko¨hler, S.; Buffam, I. Seasonal TOC exports from seven boreal catchments in northern Sweden. Aquat Sci. 2004, 66 (2), 223-230. (42) Allan, J. D.; Erickson, D. L.; Fay, J. The influence of catchment land use on stream integrity across multiple spatial scales. Freshwater Biol. 1997, 37 (1), 149-161.

Received for review April 8, 2004. Revised manuscript received November 29, 2004. Accepted December 1, 2004. ES040045Q VOL. 39, NO. 6, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1469