- Email: [email protected]
Is in-stream processing an important control on spatial changes in carbon fluxes in headwater catchments?
The Science of the Total Environment 265 Ž2001. 153᎐167
Is in-stream processing an important control on spatial changes in carbon fluxes in headwater catchments? J.J.C. DawsonU , C. Bakewell, M.F. Billett Department of Plant & Soil Science, Cruickshank Bldg., Uni¨ ersity of Aberdeen, Aberdeen AB24 3UU, UK Received 11 February 2000; accepted 20 March 2000
Abstract Data on small-scale spatial variations in instantaneous fluxes and concentrations of dissolved organic carbon ŽDOC., dissolved inorganic carbon ŽDIC. and free carbon dioxide ŽCO2 . are presented for a small acidic headwater stream in NE Scotland. Chloride is used as a conservative element to estimate additional, diffuse inputs of water into the main stem of the stream, other than those from tributaries. Downstream changes in instantaneous carbon fluxes were calculated and then used to estimate losses and gains of carbon within the stream system. Dissolved organic carbon concentrations in the stream ranged from 1.19᎐6.06 mg ly1 at its source to a maximum of 10.0᎐25.3 mg ly1 as the stream passed through deep peats; DOC concentrations then declined in the lower part of the catchment. DIC concentrations were initially low, increased to 1.5᎐3.0 mg ly1 and then decreased to 0.1᎐1.65 mg ly1 at the lowest site. Free CO2 concentrations increased from 0.35 mg ly1 at the stream source to 3.30 mg ly1 as the stream passed through the peat dominated area. Continually high inputs of CO2-rich water Ž) 6.0 mg ly1 . from tributaries maintained these high concentrations in the main stem, until approximately 1.74 km downstream, when there was a rapid decline in concentration. Significant changes in DOC, DIC and CO2 fluxes occur over a distance of 2.7 km downstream from the stream source to the catchment outlet. Between 5.64᎐41.5 mg C sy1 as DOC and 2.52᎐16.2 mg C sy1 as DIC are removed from the water column. Between 6.81 and 19.0 mg C sy1 as CO2 is lost along the stream length as progressive equilibration with the atmosphere occurs. We estimate that 11.6᎐17.6% of the total DOC flux . is removed from streamwater by in-stream processes. Dissolved inorganic carbon ŽHCOy 3 and free CO2 losses are in excess of nine times its measured flux at the outlet of the catchment. These results suggest that in-stream processing of DOC and DIC and outgassing of CO2 are important controls on the spatial variability of carbon fluxes within headwater streams in upland catchments dominated by organic-rich soils. 䊚 2001 Elsevier Science B.V. All rights reserved. Keywords: Carbon; Catchment; Spatial changes; Dissolved organic carbon; Free CO 2 ; In-stream processing
U
Corresponding author. Tel.: q44-1224-272693. E-mail address: [email protected] ŽJ.J.C. Dawson.. 0048-9697r01r$ - see front matter 䊚 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 0 . 0 0 6 5 6 - 2
154
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
1. Introduction Concentrations of DOC, DIC and free CO2 are affected by both biotic and abiotic processes within streams ŽHope et al., 1994.. Although DOC is thought to be an important link in the energy budget of streams ŽWetzel, 1992., in headwaters it is assumed to be of mainly allochthonous origin derived from terrestrial organic matter ŽVannote et al., 1980; Fiebig et al., 1990.. This assumption has been supported by work showing a strong correlation between soil carbon pools and streamwater DOC concentrations, particularly in small-scale Ž- 5 km2 . catchments ŽAitkenhead et al., 1999.. Relationships between different variables and riverine DOC fluxes in catchments similar to the study stream, showed that soil carbon content and percent peat cover were important in controlling riverine carbon fluxes ŽHope et al., 1997.. However, as stream order increases, terrestrial organic carbon inputs become less significant and autochthonous primary production and organic transport from upstream are likely to become more important ŽVannote et al., 1980.. Dissolved organic carbon in streams consists of approximately 20% low molecular weight compounds; carbohydrates, amino acids, peptides, nucleic acids and carboxylic acids, which potentially provides stream biota with a huge energy source ŽThomas, 1997.. The remainder tends to be refractory and is composed of phenolics, fulvic, humic and hydrophilic acids ŽThurman, 1985.. Dissolved organic carbon can be immobilised on the stream bed, where it becomes available for biofilms ŽFiebig and Lock, 1991. that act as major transformers of DOC in headwater streams ŽMcDowell, 1985; Fiebig and Lock, 1991.. Stream morphology and abiotic adsorption of DOC may lead to the retention and removal of DOC within the stream ŽMcDowell, 1985; Hope et al., 1994.. Adsorption of organic material, as well as sedimentation and coagulation, is usually controlled by the physical chemistry at the solidrliquid interface ŽDavis, 1982.. Spatial changes in free CO2 have been previously studied in a number of upland streams ŽRebsdorf et al., 1991; Pinol ˜ and Avila, 1992; Neal and Hill, 1994; Dawson et al., 1995.. In upland
catchments, CO2 concentrations in the soil atmosphere are usually 10᎐100 times higher than concentrations in the above ground atmosphere, the CO2 either being lost by diffusion to the soil surface or dissolved in soil water before reaching the stream system ŽSkiba and Cresser, 1991.. In the upper reaches of catchments, concentrations of free CO2 in streamwater are usually high and decrease rapidly downstream until CO2 concentrations are close to equilibrium with the atmosphere. The export of carbon as CO2 has been estimated to constitute approximately 10᎐30% of the total carbon flux from a number of catchments ŽKempe, 1982; Neal and Hill, 1994; Dawson et al., 1995.. This flux is dependent upon biotic in-stream processes such as photosynthesis, respiration and decomposition ŽHoffer-French and Herman, 1989; Rebsdorf et al., 1991.. Outgassing of CO2 to the atmosphere potentially constitutes a significant loss of terrestrially derived carbon and previous studies on streams in Alaska suggest that it makes a significant contribution to the overall land᎐atmosphere carbon flux ŽKling et al., 1991, 1992.. Streamwater CO2 concentration is also controlled by soil᎐water and groundwater inputs. Inputs of CO2 to the stream are regulated by factors such as temperature, soil moisture, O2 and nutrient availability and soil organic matter content ŽHoward and Howard, 1993; Jones and Mulholland, 1998.. Regulation of soil respiration by temperature causes seasonal patterns in soil CO2 , with concentrations higher in the summer and lower in the winter ŽCastelle and Galloway, 1990.. Stream turbulence, velocity, depth and gradient will also influence CO2 losses to the atmosphere ŽRebsdorf et al., 1991.. Downstream spatial changes in DOC, DIC and CO2 can therefore be controlled by two different types of processes; Ž1. spatial differences in quantity and composition of soil water inputs from different parts of the catchment as the stream flows through different soil types; and Ž2. gains or losses of carbon due to in-stream processes. This paper describes work in a small organicrich headwater catchment which aims to measure gains and losses of DOC, DIC and CO2 due to in-stream processing along the main stem. Area
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
weighted discharge was calculated for every sampling site on the main stem and tributaries. Additional Ždiffuse. inputs of water between two contiguous sites were quantified by the difference in chloride fluxes between them, assuming chloride behaves conservatively ŽFerrier et al., 1990.. The amount of within stream carbon loss was compared to the total carbon flux from the catchment outlet, to determine whether in-stream processes are an important factor in controlling the overall carbon flux.
2. Materials and methods 2.1. Study area Spatial variability in carbon fluxes was studied in Brocky Burn, a small headwater stream in Glen Dye, NE Scotland ŽFig. 1.. Brocky Burn is approximately 2.7 km in length, has a catchment area of 1.3 km2 and drains in a southerly direction from the Cairn of Edendocher Ž549 m.. It is a typical acidic upland moorland stream with a stony bed and limited algae growth. Mean streamwater chemistry for Brocky Burn from 26 samples, collected at 14-day intervals between June 1996 and June 1997, is presented in Table 1 ŽSmart, unpublished data.. Table 1 Basic streamwater solute chemistry collected fortnightly Ž n s 26. between June 1996 and June 1997 ŽSmart, unpublished data. a Determinand
Mean
Min.
Max.
pH Ca2q Mg 2q Naq Kq Cly SO42y NOy 3 Gran alkalinity SiO 2
5.17 2.03 0.70 5.22 0.31 7.43 3.64 0.05 55.8 7.82
4.24 1.04 0.44 2.71 0.18 4.99 2.30 0.00 y67.7 2.99
6.87 3.42 1.00 6.32 0.52 8.72 5.10 0.21 164 11.9
a All determinands are in mg ly1 except for Gran-alkalinity Žmol c ly1 . and pH.
155
The Brocky Burn catchment consists largely of heather Ž Calluna ¨ ulgaris, Erica tetralix and Erica cinerea. moorland with rushes and grasses present in the riparian zone and flush sites. The catchment is part of a working estate and is managed for grouse by periodic burning of heather. There was limited burning covering a small part of the study area in the year prior to sampling. The upper part of the catchment slopes steeply from the Cairn of Edendocher towards the central area of the catchment, which is dominated by deep peats. The Brocky Burn catchment is underlain by granite and granite-derived colluvial and fluvioglacial material. The soils consist of 59% hill peat, 22% peaty podzols, 19% rankers and - 1% fluvisols. The stream first rises on poorly developed rankers and then passes through deep peats, which vary in thickness from 0.5 to 5.0 m ŽDeacon, personal communication.. Further down the catchment, the peat adjacent to the stream is mixed with boulders of glacial origin before the gradient increases again in the lower section of the catchment, where peaty podzols predominate ŽRees et al., 1989.. Mean annual precipitation values ŽSeptember 1996᎐September 1998. in the upper and lower parts of the catchment were 1196 mm and 1131 mm, respectively. Average summer discharge, continually measured at a flume located at the outflow of Brocky Burn ŽNO615832., was 0.0259 m3 sy1. Annual loss by evapotranspiration ŽSeptember 1996᎐September 1998., calculated from the difference between precipitation and discharge, was 25.6%. 2.2. Sampling protocol To determine the extent of in-stream processing of carbon, intensive water sampling was undertaken to measure small-scale spatial changes in the concentrations and instantaneous fluxes of carbon and chloride. The stream was sampled on three occasions during the summers of 1997 and 1998 Ž14r07r1997, 11r06r1998 and 7r07r1998.. Samples were collected from 26 sites along the length of Brocky Burn and also on each of its tributaries ŽFig. 1.. Samples on the main stem
156
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
Fig. 1. The sampling sites Ž䢇. along the main stem of Brocky Burn and its associated tributaries. The flume is situated at the lowermost site. Discharge was measured on the three sampling dates at 0.0082 m3 sy1 Ž14r07r1997., 0.0184 m3 sy1 Ž11r06r1998. and 0.0040 m3 sy1 Ž07r07r1998..
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
Fig. 2. Diagram to illustrate the sampling strategy associated with the tributaries and main stem.
were also collected close to confluences with tributaries in order to quantify the influence of the tributary on the ionic composition of the streamwater, after full mixing of the tributary with the main stem had occurred ŽFig. 2.. The distance between successive sample sites on the main stem varied from 25 to 200 m. A 40-ml glass vial was used to collect each sample for the determination of DOC and DIC and a 125-ml acid-washed plastic bottle for Cly analysis. Samples for free CO2 , pH and temperature were taken less intensely at sites on the main stem and tributaries. Carbon dioxide samples were collected in 20-ml nylon gas syringes following headspace equilibration ŽKling et al., 1991.. The sample for the determination of pH was collected in a 60-ml plastic luer-lock syringe with a stopper fitted under water to avoid outgassing of CO2 . Samples were transported back to the laboratory and stored at 4⬚C prior to analysis. At the lowermost site, discharge from the catchment was measured at a continuous monitoring station. The stream was close to base flow on each of the three sampling dates. 2.3. Analysis of water samples Dissolved organic carbon and DIC were determined by ultra-violet oxidation ŽLABTOC, Pollution and Process Monitoring.. Samples were not filtered as discharge was close to base flow and particulate organic carbon ŽPOC. concentrations were negligible Ž- 1 mg ly1 .. Chloride was determined by ion-chromatography ŽDionex Series 4500I.. Samples were analysed for free CO2 using headspace analysis ŽKling et al., 1991, 1992; Hope
157
et al., 1995.. A 20-ml headspace was created in a 60-ml luer-lock syringe and equilibrated with the streamwater sample for 1 min by vigorously shaking within the stream. A 10᎐15-ml sample of the headspace gas was then transferred into a nylon gas syringe and analysed within 24 h for its CO2 concentration using a GC ŽChrompack 9001. with attached methanizer and FID. A GK2401C combined glassrreference electrode ŽRadiometer, Copenhagen. was used to measure pH within an enclosed system to prevent outgassing of CO2 and a shift in the bicarbonate-CO 2 equilibria. 2.4. Calculation of in-stream processing Discharge was measured directly at the lowermost site and indirectly at all other sites by area weighting as follows: Qsite s
Catchment area Ž km2 . at the site Catchment area Ž km2 . at the flume = QFlume
Ž1.
This assumes that precipitation distribution and runoff per unit area are uniform over the entire catchment. This is a reasonable assumption for a small Ž1.3 km2 . catchment, particularly as samples were collected at or near base flow conditions. Initially, the instantaneous fluxes of the three carbon determinands ŽDOC, DIC and free CO2 . and chloride were calculated at each sample site to determine the importance of in-stream processing of carbon. For tributaries joining the main stem, the sum of the instantaneous chloride fluxes for sites W and X should equal the flux at site Y provided: Ž1. chloride behaves as a conservative element; and Ž2. there are no additional inputs or losses of water between the three individual sampling points ŽFig. 2.. The chloride flux was therefore used to estimate unmeasured inputs of water not accounted for by sampling and then used to adjust the data for each carbon determinand, to take account of any additional input. Chloride was used, as it is considered to behave as a conservative element within catchments and is therefore unaffected by in-stream processes ŽFerrier et al., 1990.. The main assumption involved is
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
158
that the amount of additional chloride from unmeasured diffuse sources Žbank seepage and groundwater inputs. between two points on the stream channel, can be used to estimate additional inputs of dissolved carbon from similar sources. This underlying assumption is evaluated in detail in the discussion. The instantaneous flux for each site was calculated as follows: Flux s C = Q
Ž2.
where C s concentration of each determinand Žmg ly1 . and Qs discharge Žl sy1 .. Once the instantaneous flux data had been calculated for each site, further calculations were required to identify the presence of any in-stream processing of carbon. Firstly, the additional inputs of water between sites were calculated using the instantaneous chloride flux data. The instantaneous flux at Site Y on the main stem should equal the flux at the next downstream site ŽZ., assuming no additional inputs or losses of water ŽFig. 2.. The downstream flux of chloride is, however, larger than that at the upstream site because it has a slightly greater catchment area. The ‘additional’ Žunmeasured. input of water was calculated by using the ratio of chloride fluxes between sites Y and Z:
GainrLoss of C s C Flux at Y y wŽ C Flux at W qC Flux at X. = Ž EW ª Y .x
Ž6.
All the gains or losses between each sampling site down the full length of the stream were then summed to give a record of the cumulative gain or loss of carbon from the stream source to the catchment outlet Finally, the sum of the discharges and fluxes from all the tributaries, including Site 3 Žeffectively the first tributary ., were calculated. This was to ascertain the contribution tributaries make to the discharge and fluxes of individual determinands at the catchment outlet.
Discharge Ratio Ž EYª Z . s Cly Flux at Zr Cl y Flux at Y
Ž3.
Using this value, gains or losses of carbon can be calculated within the stream between the two sites: GainrLoss of C s C Flux at Z y Ž C Flux at Y= EYª Z . Ž4. Where a tributary ŽX. joins the main stem between two sites ŽW and Y in Fig. 2., the following equations were used: Discharge Ratio Ž E W ª Y . s Cly Flux at Yr Ž Cly Flux at W qCly Flux at X.
Ž5.
Fig. 3. Downstream spatial variability in Ža. DOC, Žb. DIC and Žc. free CO2 concentrations along the main stem of Brocky Burn.
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
159
3. Results 3.1. Discharge Although stream discharge on the three sampling dates varied, all sampling was at or near base flow and values were lower than the average summer discharge measured at Brocky Burn Ž0.0259 m3 sy1 .. The highest discharge Ž0.0184 m3 sy1 . was measured at the flume on 11r06r1998. On the other two sampling dates, discharge values of 0.0082 m3 sy1 Ž14r07r1997. and 0.0040 m3 sy1 Ž07r07r1998. were recorded. Discharge remained steady during all three sample collection periods. 3.2. Spatial ¨ ariability in DOC, DIC, free CO2 concentrations and pH Downstream variations in the concentrations of DOC, DIC and free CO2 on all three sampling dates are shown in Fig. 3. DOC varied with discharge, with the highest concentrations occurring on the 11r06r1998. The DOC concentrations at the stream source ranged from 1.19 to 6.06 mg ly1 and increased to a maximum of 10.0᎐25.3 mg ly1 as the stream passed through the deep peats in the central area of the catchment. DOC concentrations then declined to between 7.90 and 19.0 mg ly1 at the outlet. Concentrations of DIC were more variable than DOC and were also relatively low. DIC concentrations were lowest Ž- 1.5 mg ly1 . when discharge was at its highest on 11r06r1998. On the other days, DIC concentrations were low near the stream source and increased to between 2.0 and 3.0 mg ly1 in the vicinity of the peat dominated area 0.85 km downstream. By 1.6 km downstream, concentrations of DIC on all three sampling dates started to fall and continued to decrease to between 0.1 and 1.65 mg ly1 at the lowermost site. All three sets of downstream samples showed consistent and reproducible trends in CO2 concentrations. Free CO2 concentrations in the main stem ranged from 0.28 to 0.40 mg ly1 in streamwater derived from poorly developed rankers to 2.15᎐3.28 mg ly1 when the stream
Fig. 4. Downstream spatial variability in pH along the main stem of Brocky Burn.
passed through the central area of deep peat. Two clear peaks are observed in this part of the catchment, before CO2 concentration decreased to 0.20᎐0.28 mg ly1 at the outlet. The lowest concentrations of CO2 occurred on the day associated with the highest discharge. Streamwater pH was closely related to discharge, the highest discharge producing the lowest pH value. pH decreased from between 5.87᎐6.38 at the top of the catchment and 4.48᎐5.55 as the water entered the peat-rich area, 0.85 km downstream from the source ŽFig. 4.. From this point onwards, pH steadily increased to 5.01᎐6.71 at the lowermost site. Stream temperature increased from a low of 6.1᎐7.9⬚C at the source to 10.1᎐ 16.3⬚C at the lowermost site. 3.3. Tributaries The DOC concentrations of the tributaries show marked differences from those of the main stem, reflecting different source areas within the catchment. Table 2 shows the changes in all carbon determinand concentrations of the main stem sites above and below the confluence with each tributary. DIC concentrations in the main stem were influenced by a number of major tributaries entering the stream, particularly T2 Žaverage concentration 3.00 mg ly1 ., T4 Ž3.40 mg ly1 . and T6 Ž4.10 mg ly1 .; in all cases DIC concentrations
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
160
Table 2 DOC, DIC and free CO 2 concentrations of tributaries compared to the two sites on the main stem above and below the tributary a Sampling date 14r07r1997
11r06r1998
07r07r1998
Site
DOC
DIC
CO 2
DOC
DIC
CO 2
DOC
DIC
CO 2
3 T1 4 6 T2 7 8 T3 9 10 T4 11 13 T5 14 15 T6 16 19 T7 20 24 T8 25
2.62 ns ns 5.77 5.53 6.04 6.25 23.10 8.34 9.72 7.64 9.25 9.58 6.69 ns 10.01 2.14 7.98 8.82 nf 8.80 7.59 nf 7.31
2.95 ns ns 2.25 4.13 3.01 2.93 2.93 2.65 2.23 4.25 2.30 2.66 1.85 ns 2.79 5.26 3.30 2.72 nf 2.58 1.73 nf 2.33
ns ns ns ᎐ 5.34 2.87 ns 4.85 2.54 ns ns 1.67 1.80 3.43 ns 1.9 6.07 ns 0.32 nf ns 0.23 nf ns
17.6 28.9 20.7 24.3 20.6 23.4 23.0 27.8 24.9 24.7 20.4 24.6 24.7 24.2 25.3 24.2 12.8 22.9 23.8 30.2 24.1 21.8 4.74 19.5
1.79 0.55 0.95 1.45 2.39 1.00 1.68 1.46 1.09 1.23 2.24 1.17 0.94 0.46 0.75 1.09 2.94 1.11 0.14 0.01 0.00 0.00 0.29 0.12
ns ns ns ᎐ 3.29 2.15 ns 1.53 1.85 ns ns 1.03 0.94 1.61 ns 1.72 4.35 ns 0.38 0.47 ns 0.33 1.26 ns
4.09 10.3 4.97 6.82 6.37 6.81 7.14 18.4 8.90 8.81 6.59 8.87 9.92 11.8 10.7 10.5 2.88 8.76 8.96 nf 9.26 8.60 nf 8.40
1.19 1.67 1.19 1.08 2.34 2.06 2.12 2.96 2.52 2.43 3.70 2.27 2.38 0.85 2.20 1.99 4.04 2.50 1.84 nf 2.23 1.99 nf 1.58
ns ns ns ᎐ 4.64 3.28 ns 3.51 2.58 ns ns 1.52 1.96 1.42 ns 2.19 6.04 ns 0.36 nf ns 0.30 nf ns
All determinands are expressed in mg ly1 Žnf s no flow., Žns s not sampled..
a
exceeded those on the main stem. Free CO2 concentrations on the main stem were maintained by high CO2 inputs from the tributaries, particularly T2 Ž4.40 mg ly1 ., T3 Ž3.30 mg ly1 . and T6 Ž5.50 mg ly1 .. The tributaries only accounted for 50.7% of the total discharge measured at the flume and 52.0% to the chloride flux ŽTable 3.. However, the mean
percent contribution made to the DOC flux at the final site by the tributaries was 63.5%. This value was statistically different Ž P- 0.05. from the mean percent contribution made by the tributaries to both the chloride flux and discharge at the lowermost site. The contributions from the tributaries to the DIC and CO2 fluxes Ž333% and 636%, respectively. reflect the significant losses of
Table 3 The contribution Ž%. of the tributaries and Site 3 Ž1st tributary . to total discharge and fluxes at the lowermost site in Brocky Burn Determinand
14r07r1997
11r06r1998
07r07r1998
Mean
S.D.
Discharge Cly DOC DIC CO 2
47.2 48.3 61.2 93.4 875
56.5 57.7 65.7 826 353
48.6 49.9 63.5 79.8 679
50.7 52.0 63.5 333 636
5.0 5.0 2.2 427 263
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
both determinands from streamwater in the main stem above the lowermost site. 3.4. Gains and losses of carbon The estimated cumulative gains and losses, calculated using the Cly data, of DOC, DIC and free CO2 downstream for each sample run are shown in Fig. 5. Although on all three sampling days a similar pattern emerged, the magnitudes of
161
the fluxes varied, depending upon discharge. The greater the discharge, the greater the gains or losses in the flux of each determinand. Over the whole stream section, all three carbon determinands showed a net loss of carbon due to instream processes. Initially, at the uppermost sites there was a net gain of DOC. At 1.05 km downstream from the source, on the 2 low flow days, there was a sharp change from net gain to net loss; this continued
Fig. 5. Cumulative gains and losses of DOC, DIC and free CO2 downstream along Brocky Burn on Ža. 14r07r1997, Žb. 11r06r1998 and Žc. 07r07r1998.
162
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
with a number of fluctuations until 2.00 km downstream. At this point the net loss of DOC increased markedly. On the day with the highest flow Ž11r06r1998., the initial net gain in the upper part of the catchment was maintained until 2.00 km downstream, where there was also a sharp change to net loss; this persisted until the catchment outlet. Along the length of the stream, between 5.64 and 41.5 mg C sy1 was removed as DOC from the streamwater column. There were only small cumulative gains and losses of DIC from the stream source to 1.92 km downstream, when the magnitude of the losses and gains increased on all 3 days. These changes culminated in a final net loss of between 2.52 and 16.2 mg C sy1 as DIC. Although the free CO2 fluxes on all 3 days initially showed a very small net gain Ž1 mg C sy1 ., by 1.00 km downstream this changed to a net loss. Although there was a slight gain of CO2 between 1.50 and 1.75 km, this was not enough to prevent a net loss of CO2 at the outlet of between 6.81 and 19.0 mg C sy1 . A comparison of the net losses of carbon determinands as a percent of the overall fluxes at the outlet at the catchment ŽTable 4., showed that between 11.6% and 17.6% of the total DOC flux at the lowermost site was lost along the stream length. On the 2 days of lower discharge, 67.1% and 43.4% of the DIC flux at the lowermost site was lost by in-stream processes. However, on the higher discharge day, the DIC flux at the lowermost site was smaller due to the low inorganic carbon concentrations. Thus, the net loss of DIC far exceeded the final flux at the flume. Net losses of free CO2 are greater than the CO2 flux measured at the flume by 370% on the highest discharge day and 841᎐997% on the two lower discharge days.
4. Discussion 4.1. Spatial ¨ ariability in DOC, DIC and free CO2 concentrations Although DOC concentrations are known to be strongly influenced by discharge ŽGrieve, 1984.,
Table 4 Net loss of carbon calculated as a percentage of the flux at the outlet of the catchment Sampling date
DOC
DIC
Free CO 2
14r07r1997 11r06r1998 07r07r1998
11.6 11.9 17.6
67.1 1102 43.4
996 369 840
spatial patterns in DOC along streams are closely related to the soil changes within the catchment ŽRees et al., 1989.. The increased concentrations of DOC as the stream passes from rankers to peat was more marked during highest flow than on the 2 lower flow days, reflecting changes in hydrological pathway. After the DOC concentration reaches a maximum in the area of deepest peats, it began to decrease when the stream passes into an area of peaty podzols as the gradient of the stream increased. Although inputs of DOC from peaty podzols may be substantially less than DOC inputs to the stream from peats, the decrease in DOC concentration is relatively small in the lower part of the catchment. This may be due to the high upstream inputs of DOC not being significantly diluted further downstream. Dissolved inorganic carbon concentrations were variable along the catchment but always - 3.5 mg ly1 , reflecting the acidic nature of the stream and its low groundwater inputs ŽReid et al., 1981.. When flow was at its highest Ž11r06r1998., the pH was - 5 in the main section of the stream, shifting the carbonate equilibria in favour of free CO2 and maintaining low HCO3y concentrations. Although there are inputs of DIC from tributaries ŽTable 3., which maintain the HCOy 3 concentration in the upper and middle parts of the catchment, DIC is increasingly diluted downstream. The streamwater chemistry of individual tributaries may therefore show a greater groundwater influence, because of lower water volumes associated with fracture patterns in the underlying granite. Free CO2 concentrations in Brocky Burn are within the range reported elsewhere for headwater streams ŽRebsdorf et al., 1991; Pinol ˜ and Avila, 1992; Neal and Hill, 1994; Dawson et al., 1995.. CO2 concentrations are initially low, possi-
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
bly due to low soil respiration rates in the poorly developed mineral soils at the top of the catchment and the ease of escape to the atmosphere through thin coarse textured soils. A marked increase in CO2 concentration in the stream associated with the peat may be due to a marked decrease in the rate of CO2 escape from peaty soils to the atmosphere due to their low porosity ŽSkiba and Cresser, 1991.. This is further enhanced by continual inputs of CO2-rich water from tributaries, which have been reported before in a similar upland stream ŽDawson et al., 1995.. Tributaries tend to have a greater concentration of CO2 as they are shorter in length and thus have less distance to degas before reaching the main stem. There is a rapid decrease in CO2 concentrations at 1.74 km downstream from the source as inputs of CO2 from tributaries cease and the soils within the catchment change from peats to peaty podzols. Outgassing of CO2 from the main stream therefore becomes the dominant process in controlling CO2 concentration. These reduced inputs may be due to differences in the water holding capacity of the soils; deep wet peats will store dissolved CO2 in the soil solution more readily than the well aerated mineral soils and thus slow down exchange with the atmosphere. The decrease in concentration also corresponds to an increase in stream gradient and turbulence, which enhances the outgassing of CO2 to the atmosphere ŽRebsdorf et al., 1991.. CO2 concentrations eventually drop to a value just above that corresponding to equilibration with the atmosphere, as the high inputs from further up the catchment have been lost and inputs from the mineral soils lower down the catchment are less significant. 4.2. Tributaries Tributaries contribute to only 50.7% of the total discharge measured at the catchment outlet. This has implications for the concept of using mixtures of water from tributaries and sub-catchments to model and predict concentrations and solutes in river systems. The main sources of the large amount of unaccounted water are bank seepage and groundwater inputs through the
163
stream bed. Losses of water from the catchment through the soils and underlying granite will be negligible because the area is strongly indurated with a low permeability soil horizon forming an effective seal ŽReid et al., 1981.. Another possible source of input to the streams, are point sources of subsurface flow, such as the flush areas in the lower part of the catchment ŽFig. 1.. Although flushes ŽT7 and T8. were not flowing on 2 of the 3 sampling days, inputs were probably still occurring. The contribution of the tributaries to the overall discharge from the catchment on the higher flow day was 9% higher than that on the 2 other days. The fact that the contribution of tributaries to the overall catchment chloride flux was very similar to that for discharge, suggests that chloride does behave as a conservative ion in the system. DOC, DIC and CO2 contributions from tributaries show significant differences compared to discharge. This reflects the losses of these determinands from streamwater upstream of the lowermost site. 4.3. Assumptions in calculating carbon losses It is necessary to estimate the additional input of DOC, DIC and CO2 from diffuse sources Žbank seepage, groundwater inputs through the streambed., because only 51% of the flow into the stream can be sampled directly in the catchment. The assumption made is that the percent increase in the instantaneous chloride flux between two successive sample points Ž25᎐200 m apart. can be used to estimate the increase in DOC, DIC and CO2 fluxes caused by unmeasured inputs of water. Chloride concentrations from diffuse inputs are therefore assumed to be constant for the small distances between contiguous sites, which may not always be the case. Differences in chloride concentrations might be caused by weathering of chloride containing minerals Žnot present in this catchment . or changes in evapotranspiration rates. However, on individual sample days, the chloride concentration range of the stream did not vary by ) 0.75 mg ly1 along its length. This suggests that additional inputs of chloride-rich or chloride-poor water does not occur between adjacent sites.
164
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
The difference between Ž1. the sum of the measured upstream instantaneous flux and the estimated DOC, DIC and CO2 flux from diffuse sources; and Ž2. the instantaneous flux at the next downstream site, enables losses or gains caused by in-stream processing to be estimated. The approach used to estimate amounts of in-stream processing may lead to an underestimate of the amount of DOC lost by processing within the stream. This is because mean concentrations of DOC measured in soil solutions derived from two depths in the riparian zone and two depths in peat bogs in Brocky Burn are higher than streamwater concentrations and vary from 36 to 45 mg ly1 ŽBillett, unpublished data.. Also, soil solution DOC concentrations, derived from the Bs horizon of a podzol in an adjacent catchment in Glen Dye, had a mean DOC concentration of 28.3 mg ly1 ŽKennedy, unpublished data.. These concentrations are significantly higher than the estimated concentrations of DOC from diffuse sources, which are - 10 mg ly1 . This can be calculated by dividing the differences in the instantaneous DOC fluxes by the amount of additional water introduced between two sites. In terms of DIC and free CO2 , which already have large within-stream losses compared with the final flux at the outlet from the catchment, the underlying assumption based on Cly would also tend to underestimate the amounts of these determinands lost by in-stream processing. 4.4. Gain and loss of DOC Within the stream channel there was a net gain in DOC in the upper part of the catchment on all three occasions. Production of DOC within the water column might occur from the breakdown of coarse or fine POC by microbial action ŽVannote et al., 1980.. Net losses of DOC occur by 1.05 km downstream on the low discharge days and by 2.25 km downstream on the high flow day. Losses of DOC are due to biotic processes, such as biofilm respiration and adsorption onto algae ŽThomas, 1997. and to abiotic uptake, such as adsorption onto mineral surfaces Žincluding Fe and Al oxides and hydroxides. and precipitation
of POC ŽDavis, 1982.. Total Al concentration in this stream varied between 67 and 99 g ly1 for a similar time of year ŽStutter, unpublished data.; this could provide a potential substrate for significant adsorption and removal of DOC. Overall, there appear to be a series of interactive processes controlling gains and losses of DOC within the water column, which are influenced by flow conditions and the degree of biotic activity within the stream. Continual increases in the amount of DOC lost in the lower reaches of the catchment, suggest that losses of DOC by biological uptake and abiotic adsorption are important. The two low flow days followed a period of dry weather, whereas the higher flow sampling was carried out at the tail-end of the descending limb of the hydrograph. This period of high flow may have removed much of the biofilms and algae from the stream environment. It is possible that this shifts the equilibria in the catchment, particularly in the top and central areas, towards processes that favour gains of DOC Že.g. formation of DOC from fine POC. and away from loss processes Žbiological uptake by metabolism and adsorption.. 4.5. Gain and loss of DIC Very little change in instantaneous DIC fluxes occurred in the upper reaches of the catchment until 1.74 km downstream from the source, when a net loss of HCO3y occurred, possibly due to increased biological activity within the stream. However, DIC uptake is likely to be minimal because other carbon sources also exist for biota. These losses may also be connected with the carbonate equilibria within the stream. As CO2 is removed from the stream by outgassing, HCO3y reacts with the high concentrations of Hq present in the acidic stream forcing the equilibria back towards CO2 . There are also a number of groundwater inputs of HCO3y from tributaries. However, no tributaries occur in the lower part of the catchment, so losses of HCO3y are not replenished from groundwater or any other source and the magnitude of the net loss of HCO3y continues to increase.
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
4.6. Gain and loss of free CO2 The majority of CO2 is lost from the stream system by outgassing, as streamwater supersaturated in CO2 equilibrates with the atmosphere. CO2 may also be lost from the stream during daylight hours by photosynthesis ŽRebsdorf et al., 1991; Pinol ˜ and Avila, 1992.; we were not, however, able to differentiate between the two processes. Although greater amounts of CO2 are probably lost from the stream by outgassing, the net loss is lower than expected due to the replenishment of CO2 in the main stem from tributaries. That occurs until losses by outgassing become far greater than inputs at sites in the lower part of the catchment. Also in the lower section of the stream, rates of outgassing have decreased and there is a levelling out of net CO2 loss. In-stream photosynthesis will further increase CO2 loss especially during sunnier days. The similarities exhibited in the net losses and gains of DIC and CO2 are not unexpected due to the chemical link between the two via the carbonate equilibria. This link may also be enhanced by the method of analysis because samples measured for DIC will contain minor amounts of CO2 , which remains after equilibration with the atmosphere ŽHope et al., 1995.. 4.7. O¨ erall effect on total carbon fluxes The percent losses of DOC, DIC and free CO2 compared with the fluxes at the outlet are shown in Table 4. In-stream DOC loss was between 11.6 and 17.6% of its final flux from the catchment. This is similar to losses reported in other work by Fisher and Likens Ž1973., where a 34% energy loss was reported in a stream within the Hubbard Brook Experimental Forest, New Hampshire. However, this energy loss does not just represent DOC, which was estimated as 47% of the total energy input to the stream. Hubbard Brook is also a more productive higher pH system with the stream receiving significant litter inputs from a mixed conifer-hardwood forest. High rates of removal of added bovine manure extract Ž31%. and jewel-weed leachate Ž19%. were observed by Kuserk et al. Ž1984.. The contrast in the removal
165
rates between the two types of added carbon suggests that natural differences in within stream processing are dependent on substrate type. In comparison to Brocky Burn, both these studies were undertaken in higher order streams where processing of DOC would be enhanced by higher pH values and temperatures. The losses of DIC and free CO2 measured at Brocky Burn reflect the greater losses of these determinands along the stream length compared to the relatively small flux at the outlet of the catchment. In particular, the flux of CO2 to the atmosphere constitutes a significant loss of CO2 from the streamwater, which is not measured at the catchment outlet.
5. Conclusions Two main processes control spatial concentrations of streamwater DOC, DIC and free CO2 in the Brocky Burn catchment. The main control on streamwater DOC concentrations are the spatial differences between soil water inputs from different parts of the catchment as the stream flows through different soil types. However, in-stream processes modify these DOC inputs, thus providing a secondary control on the spatial variability of DOC concentrations and fluxes. Although instream processes reduce the final DOC flux at the catchment outlet by 11.6᎐17.6%, hydrological pathways control initial amounts of carbon reaching the stream. Similar processes also control the concentrations and fluxes of DIC and CO2 in the headwater. However, in-stream processes Žincluding outgassing of CO2 . have significantly more influence on spatial changes occurring in these two determinands than on DOC. On all three sampling occasions, spatial changes in the individual determinands showed similar downstream trends and reflect different provenances of DOC, DIC and free CO2 in the catchment. This is exemplified by the increase in DOC and CO2 concentrations as the stream enters the carbon-rich peat areas in the central part of the catchment, after flowing through the poorly developed mineral soils at the head of the catchment. Thus, stream sources are not always supersaturated with respect to CO2 ŽDawson et al.,
166
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
1995., but depend on soil CO2 concentrations supplying the stream at that point. The data presented here are from a small headwater stream draining an upland catchment dominated by organic soils. In larger, higher order river systems in-stream processing would probably have a greater effect on DOC flux because they tend to be less acidic and have higher water temperatures. Chloride appears to behave conservatively in this headwater stream and can be used to determine diffuse inputs of water from bank seepage and groundwater into the main stem. This assumption possibly under-estimates the amount of in-stream processing because the carbon concentration of the water derived from diffuse sources is likely to be higher than the value which has been assumed. Moreover, in-stream processing along the length of the tributaries, prior to reaching the main stem, was not accounted for in our estimate of in-stream losses. Sampling, however, was carried out in summer daylight hours when conditions for in-stream processing would be expected to be close to optimal. There would almost certainly be a reduction in the importance of in-stream processing during the colder lower light periods of the year.
Acknowledgements The authors would like to thank Fasque Estates for allowing access to the site and their support for the research. J.J.C. Dawson wishes to thank the Natural Environment Research Council for providing financial support and Douglas Low for assistance with mapping. C. Bakewell acknowledges receipt of an E.C. scholarship. References Aitkenhead JA, Hope D, Billett MF. The relationship between dissolved organic carbon in streamwater and soil organic carbon pools at different spatial scales. Hydrolog Process 1999;13:1289᎐1302. Castelle AJ, Galloway JN. Carbon dioxide dynamics in acid forest soils in Shenandoah National Park, Virginia. Soil Sci Soc Am J 1990;54:252᎐257. Davis JA. Adsorption of natural dissolved organic matter at
the oxiderwater interface. Geochim Cosmochim Acta 1982;46:2381᎐2393. Dawson JJC, Hope D, Cresser MS, Billett MF. Downstream changes in free carbon dioxide in an upland catchment from northeastern Scotland. J Environ Qual 1995;24: 699᎐706. Ferrier RC, Jenkins A, Miller JD, Walker TAB, Anderson HA. Assessment of wet deposition mechanisms in an upland Scottish catchment. J Hydrol 1990;113:285᎐296. Fiebig DM, Lock MA. Immobilization of dissolved organic matter from groundwater discharging through the streambed. Freshwater Biol 1991;26:45᎐55. Fiebig DM, Lock MA, Neal C. Soil water in the riparian zone as a source of carbon for a headwater stream. J Hydrol 1990;116:217᎐237. Fisher SG, Likens GE. Energy flow in Bear brook, New Hampshire: an integrative approach to stream ecosystem metabolism. Ecol Monogr 1973;43:421᎐439. Grieve IC. Concentrations and annual loading of dissolved organic matter in a small moorland stream. Freshwater Biol 1984;14:533᎐537. Hoffer-French KJ, Herman JS. Evaluation of hydrological and biological influences on CO2 fluxes from a karst stream. J Hydol 1989;108:189᎐212. Hope D, Billett MF, Cresser MS. A review of the export of carbon in river water: fluxes and processes. Environ Pollut 1994;84:301᎐324. Hope D, Dawson JJC, Cresser MS, Billett MF. A method for measuring free CO2 in upland streamwater using headspace analysis. J Hydol 1995;166:1᎐14. Hope D, Billett MF, Milne R, Brown TAW. Exports of organic carbon in British Rivers. Hydrolog Process 1997;11: 325᎐344. Howard DM, Howard PJA. Relationships between CO2 evolution, moisture content and temperature for a range of soil types. Soil Biol Biochem 1993;25:1537᎐1546. Jones JB, Mulholland PJ. Influence of drainage basin topography and elevation on carbon dioxide and methane supersaturation of stream water. Biogeochemistry 1998;40:57᎐72. Kempe S. Long-term records of CO2 pressure fluctuations in fresh waters. In: Degens ET, editor. Scientific Committee on Problems of the Environment ŽSCOPE.rUnited Nations Environment Programme ŽUNEP. ᎏ Transport of Carbon and Minerals in Major World Rivers, 52. Hamburg, Germany: University of Hamburg, 1982:91᎐332. Kling GW, Kipphut GW, Miller MC. Arctic lakes and streams as gas conduits to the atmosphere: implications for tundra carbon budgets. Science 1991;251:298᎐301. Kling GW, Kipphut GW, Miller MC. The flux of CO2 and CH4 from lakes and rivers in Arctic Alaska. Hydrobiology 1992;240:23᎐36. Kuserk FT, Kaplan LA, Bott TL. In situ measures of dissolved organic carbon flux in a rural stream. Can J Fish Aquat Sci 1984;41:964᎐973. McDowell WH. Kinetics and mechanisms of dissolved organic carbon retention in a headwater stream. Biogeochemistry 1985;1:329᎐352.
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167 Neal C, Hill S. Dissolved inorganic and organic carbon in moorland and forest streams: Plynlimon, mid-Wales. J Hydrol 1994;153:231᎐243. Pinol ˜ J, Avila A. Streamwater pH, alkalinity, pCO2 and discharge relationships in some forested Mediterranean catchments. J Hydrol 1992;131:205᎐225. Rebsdorf A, Thyssen N, Erlandsen M. Regional and temporal variation in pH, alkalinity and carbon dioxide in Danish streams, related to soil type and land use. Freshwater Biol 1991;25:419᎐435. Rees RM, Parker-Jarvis F, Cresser MS. Soil effects on water chemistry in three adjacent upland streams at Glen Dye in northeast Scotland. Water Res 1989;23:511᎐517. Reid JM, MacLeod DA, Cresser MS. Factors affecting the chemistry of precipitation and river water in an upland catchment. J Hydrol 1981;50:129᎐145.
167
Skiba U, Cresser MS. Seasonal changes in soil atmospheric CO2 concentrations in two upland catchments and associated changes in river water chemistry. Chem Ecol 1991; 7:217᎐225. Thomas JD. The role of dissolved organic matter, particularly free amino acids and humic substances, in freshwater ecosystems. Freshwater Biol 1997;38:1᎐36. Thurman EM. Organic geochemistry of natural waters. Dordrecht, Holland: NijhoffrJunk Publishers, 1985:497pp. Vannote RL, Minshall GW, Cummins KW, Sedell JR, Cushing CE. The river continuum concept. Can J Fish Aquat Sci 1980;37:130᎐137. Wetzel RG. Gradient dominated ecosystems: sources and regulatory functions of dissolved organic matter in freshwater ecosystems. Hydrobiology 1992;229:181᎐198.
Is in-stream processing an important control on spatial changes in carbon fluxes in headwater catchments? J.J.C. DawsonU , C. Bakewell, M.F. Billett Department of Plant & Soil Science, Cruickshank Bldg., Uni¨ ersity of Aberdeen, Aberdeen AB24 3UU, UK Received 11 February 2000; accepted 20 March 2000
Abstract Data on small-scale spatial variations in instantaneous fluxes and concentrations of dissolved organic carbon ŽDOC., dissolved inorganic carbon ŽDIC. and free carbon dioxide ŽCO2 . are presented for a small acidic headwater stream in NE Scotland. Chloride is used as a conservative element to estimate additional, diffuse inputs of water into the main stem of the stream, other than those from tributaries. Downstream changes in instantaneous carbon fluxes were calculated and then used to estimate losses and gains of carbon within the stream system. Dissolved organic carbon concentrations in the stream ranged from 1.19᎐6.06 mg ly1 at its source to a maximum of 10.0᎐25.3 mg ly1 as the stream passed through deep peats; DOC concentrations then declined in the lower part of the catchment. DIC concentrations were initially low, increased to 1.5᎐3.0 mg ly1 and then decreased to 0.1᎐1.65 mg ly1 at the lowest site. Free CO2 concentrations increased from 0.35 mg ly1 at the stream source to 3.30 mg ly1 as the stream passed through the peat dominated area. Continually high inputs of CO2-rich water Ž) 6.0 mg ly1 . from tributaries maintained these high concentrations in the main stem, until approximately 1.74 km downstream, when there was a rapid decline in concentration. Significant changes in DOC, DIC and CO2 fluxes occur over a distance of 2.7 km downstream from the stream source to the catchment outlet. Between 5.64᎐41.5 mg C sy1 as DOC and 2.52᎐16.2 mg C sy1 as DIC are removed from the water column. Between 6.81 and 19.0 mg C sy1 as CO2 is lost along the stream length as progressive equilibration with the atmosphere occurs. We estimate that 11.6᎐17.6% of the total DOC flux . is removed from streamwater by in-stream processes. Dissolved inorganic carbon ŽHCOy 3 and free CO2 losses are in excess of nine times its measured flux at the outlet of the catchment. These results suggest that in-stream processing of DOC and DIC and outgassing of CO2 are important controls on the spatial variability of carbon fluxes within headwater streams in upland catchments dominated by organic-rich soils. 䊚 2001 Elsevier Science B.V. All rights reserved. Keywords: Carbon; Catchment; Spatial changes; Dissolved organic carbon; Free CO 2 ; In-stream processing
U
Corresponding author. Tel.: q44-1224-272693. E-mail address: [email protected] ŽJ.J.C. Dawson.. 0048-9697r01r$ - see front matter 䊚 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 0 . 0 0 6 5 6 - 2
154
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
1. Introduction Concentrations of DOC, DIC and free CO2 are affected by both biotic and abiotic processes within streams ŽHope et al., 1994.. Although DOC is thought to be an important link in the energy budget of streams ŽWetzel, 1992., in headwaters it is assumed to be of mainly allochthonous origin derived from terrestrial organic matter ŽVannote et al., 1980; Fiebig et al., 1990.. This assumption has been supported by work showing a strong correlation between soil carbon pools and streamwater DOC concentrations, particularly in small-scale Ž- 5 km2 . catchments ŽAitkenhead et al., 1999.. Relationships between different variables and riverine DOC fluxes in catchments similar to the study stream, showed that soil carbon content and percent peat cover were important in controlling riverine carbon fluxes ŽHope et al., 1997.. However, as stream order increases, terrestrial organic carbon inputs become less significant and autochthonous primary production and organic transport from upstream are likely to become more important ŽVannote et al., 1980.. Dissolved organic carbon in streams consists of approximately 20% low molecular weight compounds; carbohydrates, amino acids, peptides, nucleic acids and carboxylic acids, which potentially provides stream biota with a huge energy source ŽThomas, 1997.. The remainder tends to be refractory and is composed of phenolics, fulvic, humic and hydrophilic acids ŽThurman, 1985.. Dissolved organic carbon can be immobilised on the stream bed, where it becomes available for biofilms ŽFiebig and Lock, 1991. that act as major transformers of DOC in headwater streams ŽMcDowell, 1985; Fiebig and Lock, 1991.. Stream morphology and abiotic adsorption of DOC may lead to the retention and removal of DOC within the stream ŽMcDowell, 1985; Hope et al., 1994.. Adsorption of organic material, as well as sedimentation and coagulation, is usually controlled by the physical chemistry at the solidrliquid interface ŽDavis, 1982.. Spatial changes in free CO2 have been previously studied in a number of upland streams ŽRebsdorf et al., 1991; Pinol ˜ and Avila, 1992; Neal and Hill, 1994; Dawson et al., 1995.. In upland
catchments, CO2 concentrations in the soil atmosphere are usually 10᎐100 times higher than concentrations in the above ground atmosphere, the CO2 either being lost by diffusion to the soil surface or dissolved in soil water before reaching the stream system ŽSkiba and Cresser, 1991.. In the upper reaches of catchments, concentrations of free CO2 in streamwater are usually high and decrease rapidly downstream until CO2 concentrations are close to equilibrium with the atmosphere. The export of carbon as CO2 has been estimated to constitute approximately 10᎐30% of the total carbon flux from a number of catchments ŽKempe, 1982; Neal and Hill, 1994; Dawson et al., 1995.. This flux is dependent upon biotic in-stream processes such as photosynthesis, respiration and decomposition ŽHoffer-French and Herman, 1989; Rebsdorf et al., 1991.. Outgassing of CO2 to the atmosphere potentially constitutes a significant loss of terrestrially derived carbon and previous studies on streams in Alaska suggest that it makes a significant contribution to the overall land᎐atmosphere carbon flux ŽKling et al., 1991, 1992.. Streamwater CO2 concentration is also controlled by soil᎐water and groundwater inputs. Inputs of CO2 to the stream are regulated by factors such as temperature, soil moisture, O2 and nutrient availability and soil organic matter content ŽHoward and Howard, 1993; Jones and Mulholland, 1998.. Regulation of soil respiration by temperature causes seasonal patterns in soil CO2 , with concentrations higher in the summer and lower in the winter ŽCastelle and Galloway, 1990.. Stream turbulence, velocity, depth and gradient will also influence CO2 losses to the atmosphere ŽRebsdorf et al., 1991.. Downstream spatial changes in DOC, DIC and CO2 can therefore be controlled by two different types of processes; Ž1. spatial differences in quantity and composition of soil water inputs from different parts of the catchment as the stream flows through different soil types; and Ž2. gains or losses of carbon due to in-stream processes. This paper describes work in a small organicrich headwater catchment which aims to measure gains and losses of DOC, DIC and CO2 due to in-stream processing along the main stem. Area
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
weighted discharge was calculated for every sampling site on the main stem and tributaries. Additional Ždiffuse. inputs of water between two contiguous sites were quantified by the difference in chloride fluxes between them, assuming chloride behaves conservatively ŽFerrier et al., 1990.. The amount of within stream carbon loss was compared to the total carbon flux from the catchment outlet, to determine whether in-stream processes are an important factor in controlling the overall carbon flux.
2. Materials and methods 2.1. Study area Spatial variability in carbon fluxes was studied in Brocky Burn, a small headwater stream in Glen Dye, NE Scotland ŽFig. 1.. Brocky Burn is approximately 2.7 km in length, has a catchment area of 1.3 km2 and drains in a southerly direction from the Cairn of Edendocher Ž549 m.. It is a typical acidic upland moorland stream with a stony bed and limited algae growth. Mean streamwater chemistry for Brocky Burn from 26 samples, collected at 14-day intervals between June 1996 and June 1997, is presented in Table 1 ŽSmart, unpublished data.. Table 1 Basic streamwater solute chemistry collected fortnightly Ž n s 26. between June 1996 and June 1997 ŽSmart, unpublished data. a Determinand
Mean
Min.
Max.
pH Ca2q Mg 2q Naq Kq Cly SO42y NOy 3 Gran alkalinity SiO 2
5.17 2.03 0.70 5.22 0.31 7.43 3.64 0.05 55.8 7.82
4.24 1.04 0.44 2.71 0.18 4.99 2.30 0.00 y67.7 2.99
6.87 3.42 1.00 6.32 0.52 8.72 5.10 0.21 164 11.9
a All determinands are in mg ly1 except for Gran-alkalinity Žmol c ly1 . and pH.
155
The Brocky Burn catchment consists largely of heather Ž Calluna ¨ ulgaris, Erica tetralix and Erica cinerea. moorland with rushes and grasses present in the riparian zone and flush sites. The catchment is part of a working estate and is managed for grouse by periodic burning of heather. There was limited burning covering a small part of the study area in the year prior to sampling. The upper part of the catchment slopes steeply from the Cairn of Edendocher towards the central area of the catchment, which is dominated by deep peats. The Brocky Burn catchment is underlain by granite and granite-derived colluvial and fluvioglacial material. The soils consist of 59% hill peat, 22% peaty podzols, 19% rankers and - 1% fluvisols. The stream first rises on poorly developed rankers and then passes through deep peats, which vary in thickness from 0.5 to 5.0 m ŽDeacon, personal communication.. Further down the catchment, the peat adjacent to the stream is mixed with boulders of glacial origin before the gradient increases again in the lower section of the catchment, where peaty podzols predominate ŽRees et al., 1989.. Mean annual precipitation values ŽSeptember 1996᎐September 1998. in the upper and lower parts of the catchment were 1196 mm and 1131 mm, respectively. Average summer discharge, continually measured at a flume located at the outflow of Brocky Burn ŽNO615832., was 0.0259 m3 sy1. Annual loss by evapotranspiration ŽSeptember 1996᎐September 1998., calculated from the difference between precipitation and discharge, was 25.6%. 2.2. Sampling protocol To determine the extent of in-stream processing of carbon, intensive water sampling was undertaken to measure small-scale spatial changes in the concentrations and instantaneous fluxes of carbon and chloride. The stream was sampled on three occasions during the summers of 1997 and 1998 Ž14r07r1997, 11r06r1998 and 7r07r1998.. Samples were collected from 26 sites along the length of Brocky Burn and also on each of its tributaries ŽFig. 1.. Samples on the main stem
156
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
Fig. 1. The sampling sites Ž䢇. along the main stem of Brocky Burn and its associated tributaries. The flume is situated at the lowermost site. Discharge was measured on the three sampling dates at 0.0082 m3 sy1 Ž14r07r1997., 0.0184 m3 sy1 Ž11r06r1998. and 0.0040 m3 sy1 Ž07r07r1998..
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
Fig. 2. Diagram to illustrate the sampling strategy associated with the tributaries and main stem.
were also collected close to confluences with tributaries in order to quantify the influence of the tributary on the ionic composition of the streamwater, after full mixing of the tributary with the main stem had occurred ŽFig. 2.. The distance between successive sample sites on the main stem varied from 25 to 200 m. A 40-ml glass vial was used to collect each sample for the determination of DOC and DIC and a 125-ml acid-washed plastic bottle for Cly analysis. Samples for free CO2 , pH and temperature were taken less intensely at sites on the main stem and tributaries. Carbon dioxide samples were collected in 20-ml nylon gas syringes following headspace equilibration ŽKling et al., 1991.. The sample for the determination of pH was collected in a 60-ml plastic luer-lock syringe with a stopper fitted under water to avoid outgassing of CO2 . Samples were transported back to the laboratory and stored at 4⬚C prior to analysis. At the lowermost site, discharge from the catchment was measured at a continuous monitoring station. The stream was close to base flow on each of the three sampling dates. 2.3. Analysis of water samples Dissolved organic carbon and DIC were determined by ultra-violet oxidation ŽLABTOC, Pollution and Process Monitoring.. Samples were not filtered as discharge was close to base flow and particulate organic carbon ŽPOC. concentrations were negligible Ž- 1 mg ly1 .. Chloride was determined by ion-chromatography ŽDionex Series 4500I.. Samples were analysed for free CO2 using headspace analysis ŽKling et al., 1991, 1992; Hope
157
et al., 1995.. A 20-ml headspace was created in a 60-ml luer-lock syringe and equilibrated with the streamwater sample for 1 min by vigorously shaking within the stream. A 10᎐15-ml sample of the headspace gas was then transferred into a nylon gas syringe and analysed within 24 h for its CO2 concentration using a GC ŽChrompack 9001. with attached methanizer and FID. A GK2401C combined glassrreference electrode ŽRadiometer, Copenhagen. was used to measure pH within an enclosed system to prevent outgassing of CO2 and a shift in the bicarbonate-CO 2 equilibria. 2.4. Calculation of in-stream processing Discharge was measured directly at the lowermost site and indirectly at all other sites by area weighting as follows: Qsite s
Catchment area Ž km2 . at the site Catchment area Ž km2 . at the flume = QFlume
Ž1.
This assumes that precipitation distribution and runoff per unit area are uniform over the entire catchment. This is a reasonable assumption for a small Ž1.3 km2 . catchment, particularly as samples were collected at or near base flow conditions. Initially, the instantaneous fluxes of the three carbon determinands ŽDOC, DIC and free CO2 . and chloride were calculated at each sample site to determine the importance of in-stream processing of carbon. For tributaries joining the main stem, the sum of the instantaneous chloride fluxes for sites W and X should equal the flux at site Y provided: Ž1. chloride behaves as a conservative element; and Ž2. there are no additional inputs or losses of water between the three individual sampling points ŽFig. 2.. The chloride flux was therefore used to estimate unmeasured inputs of water not accounted for by sampling and then used to adjust the data for each carbon determinand, to take account of any additional input. Chloride was used, as it is considered to behave as a conservative element within catchments and is therefore unaffected by in-stream processes ŽFerrier et al., 1990.. The main assumption involved is
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
158
that the amount of additional chloride from unmeasured diffuse sources Žbank seepage and groundwater inputs. between two points on the stream channel, can be used to estimate additional inputs of dissolved carbon from similar sources. This underlying assumption is evaluated in detail in the discussion. The instantaneous flux for each site was calculated as follows: Flux s C = Q
Ž2.
where C s concentration of each determinand Žmg ly1 . and Qs discharge Žl sy1 .. Once the instantaneous flux data had been calculated for each site, further calculations were required to identify the presence of any in-stream processing of carbon. Firstly, the additional inputs of water between sites were calculated using the instantaneous chloride flux data. The instantaneous flux at Site Y on the main stem should equal the flux at the next downstream site ŽZ., assuming no additional inputs or losses of water ŽFig. 2.. The downstream flux of chloride is, however, larger than that at the upstream site because it has a slightly greater catchment area. The ‘additional’ Žunmeasured. input of water was calculated by using the ratio of chloride fluxes between sites Y and Z:
GainrLoss of C s C Flux at Y y wŽ C Flux at W qC Flux at X. = Ž EW ª Y .x
Ž6.
All the gains or losses between each sampling site down the full length of the stream were then summed to give a record of the cumulative gain or loss of carbon from the stream source to the catchment outlet Finally, the sum of the discharges and fluxes from all the tributaries, including Site 3 Žeffectively the first tributary ., were calculated. This was to ascertain the contribution tributaries make to the discharge and fluxes of individual determinands at the catchment outlet.
Discharge Ratio Ž EYª Z . s Cly Flux at Zr Cl y Flux at Y
Ž3.
Using this value, gains or losses of carbon can be calculated within the stream between the two sites: GainrLoss of C s C Flux at Z y Ž C Flux at Y= EYª Z . Ž4. Where a tributary ŽX. joins the main stem between two sites ŽW and Y in Fig. 2., the following equations were used: Discharge Ratio Ž E W ª Y . s Cly Flux at Yr Ž Cly Flux at W qCly Flux at X.
Ž5.
Fig. 3. Downstream spatial variability in Ža. DOC, Žb. DIC and Žc. free CO2 concentrations along the main stem of Brocky Burn.
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
159
3. Results 3.1. Discharge Although stream discharge on the three sampling dates varied, all sampling was at or near base flow and values were lower than the average summer discharge measured at Brocky Burn Ž0.0259 m3 sy1 .. The highest discharge Ž0.0184 m3 sy1 . was measured at the flume on 11r06r1998. On the other two sampling dates, discharge values of 0.0082 m3 sy1 Ž14r07r1997. and 0.0040 m3 sy1 Ž07r07r1998. were recorded. Discharge remained steady during all three sample collection periods. 3.2. Spatial ¨ ariability in DOC, DIC, free CO2 concentrations and pH Downstream variations in the concentrations of DOC, DIC and free CO2 on all three sampling dates are shown in Fig. 3. DOC varied with discharge, with the highest concentrations occurring on the 11r06r1998. The DOC concentrations at the stream source ranged from 1.19 to 6.06 mg ly1 and increased to a maximum of 10.0᎐25.3 mg ly1 as the stream passed through the deep peats in the central area of the catchment. DOC concentrations then declined to between 7.90 and 19.0 mg ly1 at the outlet. Concentrations of DIC were more variable than DOC and were also relatively low. DIC concentrations were lowest Ž- 1.5 mg ly1 . when discharge was at its highest on 11r06r1998. On the other days, DIC concentrations were low near the stream source and increased to between 2.0 and 3.0 mg ly1 in the vicinity of the peat dominated area 0.85 km downstream. By 1.6 km downstream, concentrations of DIC on all three sampling dates started to fall and continued to decrease to between 0.1 and 1.65 mg ly1 at the lowermost site. All three sets of downstream samples showed consistent and reproducible trends in CO2 concentrations. Free CO2 concentrations in the main stem ranged from 0.28 to 0.40 mg ly1 in streamwater derived from poorly developed rankers to 2.15᎐3.28 mg ly1 when the stream
Fig. 4. Downstream spatial variability in pH along the main stem of Brocky Burn.
passed through the central area of deep peat. Two clear peaks are observed in this part of the catchment, before CO2 concentration decreased to 0.20᎐0.28 mg ly1 at the outlet. The lowest concentrations of CO2 occurred on the day associated with the highest discharge. Streamwater pH was closely related to discharge, the highest discharge producing the lowest pH value. pH decreased from between 5.87᎐6.38 at the top of the catchment and 4.48᎐5.55 as the water entered the peat-rich area, 0.85 km downstream from the source ŽFig. 4.. From this point onwards, pH steadily increased to 5.01᎐6.71 at the lowermost site. Stream temperature increased from a low of 6.1᎐7.9⬚C at the source to 10.1᎐ 16.3⬚C at the lowermost site. 3.3. Tributaries The DOC concentrations of the tributaries show marked differences from those of the main stem, reflecting different source areas within the catchment. Table 2 shows the changes in all carbon determinand concentrations of the main stem sites above and below the confluence with each tributary. DIC concentrations in the main stem were influenced by a number of major tributaries entering the stream, particularly T2 Žaverage concentration 3.00 mg ly1 ., T4 Ž3.40 mg ly1 . and T6 Ž4.10 mg ly1 .; in all cases DIC concentrations
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
160
Table 2 DOC, DIC and free CO 2 concentrations of tributaries compared to the two sites on the main stem above and below the tributary a Sampling date 14r07r1997
11r06r1998
07r07r1998
Site
DOC
DIC
CO 2
DOC
DIC
CO 2
DOC
DIC
CO 2
3 T1 4 6 T2 7 8 T3 9 10 T4 11 13 T5 14 15 T6 16 19 T7 20 24 T8 25
2.62 ns ns 5.77 5.53 6.04 6.25 23.10 8.34 9.72 7.64 9.25 9.58 6.69 ns 10.01 2.14 7.98 8.82 nf 8.80 7.59 nf 7.31
2.95 ns ns 2.25 4.13 3.01 2.93 2.93 2.65 2.23 4.25 2.30 2.66 1.85 ns 2.79 5.26 3.30 2.72 nf 2.58 1.73 nf 2.33
ns ns ns ᎐ 5.34 2.87 ns 4.85 2.54 ns ns 1.67 1.80 3.43 ns 1.9 6.07 ns 0.32 nf ns 0.23 nf ns
17.6 28.9 20.7 24.3 20.6 23.4 23.0 27.8 24.9 24.7 20.4 24.6 24.7 24.2 25.3 24.2 12.8 22.9 23.8 30.2 24.1 21.8 4.74 19.5
1.79 0.55 0.95 1.45 2.39 1.00 1.68 1.46 1.09 1.23 2.24 1.17 0.94 0.46 0.75 1.09 2.94 1.11 0.14 0.01 0.00 0.00 0.29 0.12
ns ns ns ᎐ 3.29 2.15 ns 1.53 1.85 ns ns 1.03 0.94 1.61 ns 1.72 4.35 ns 0.38 0.47 ns 0.33 1.26 ns
4.09 10.3 4.97 6.82 6.37 6.81 7.14 18.4 8.90 8.81 6.59 8.87 9.92 11.8 10.7 10.5 2.88 8.76 8.96 nf 9.26 8.60 nf 8.40
1.19 1.67 1.19 1.08 2.34 2.06 2.12 2.96 2.52 2.43 3.70 2.27 2.38 0.85 2.20 1.99 4.04 2.50 1.84 nf 2.23 1.99 nf 1.58
ns ns ns ᎐ 4.64 3.28 ns 3.51 2.58 ns ns 1.52 1.96 1.42 ns 2.19 6.04 ns 0.36 nf ns 0.30 nf ns
All determinands are expressed in mg ly1 Žnf s no flow., Žns s not sampled..
a
exceeded those on the main stem. Free CO2 concentrations on the main stem were maintained by high CO2 inputs from the tributaries, particularly T2 Ž4.40 mg ly1 ., T3 Ž3.30 mg ly1 . and T6 Ž5.50 mg ly1 .. The tributaries only accounted for 50.7% of the total discharge measured at the flume and 52.0% to the chloride flux ŽTable 3.. However, the mean
percent contribution made to the DOC flux at the final site by the tributaries was 63.5%. This value was statistically different Ž P- 0.05. from the mean percent contribution made by the tributaries to both the chloride flux and discharge at the lowermost site. The contributions from the tributaries to the DIC and CO2 fluxes Ž333% and 636%, respectively. reflect the significant losses of
Table 3 The contribution Ž%. of the tributaries and Site 3 Ž1st tributary . to total discharge and fluxes at the lowermost site in Brocky Burn Determinand
14r07r1997
11r06r1998
07r07r1998
Mean
S.D.
Discharge Cly DOC DIC CO 2
47.2 48.3 61.2 93.4 875
56.5 57.7 65.7 826 353
48.6 49.9 63.5 79.8 679
50.7 52.0 63.5 333 636
5.0 5.0 2.2 427 263
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
both determinands from streamwater in the main stem above the lowermost site. 3.4. Gains and losses of carbon The estimated cumulative gains and losses, calculated using the Cly data, of DOC, DIC and free CO2 downstream for each sample run are shown in Fig. 5. Although on all three sampling days a similar pattern emerged, the magnitudes of
161
the fluxes varied, depending upon discharge. The greater the discharge, the greater the gains or losses in the flux of each determinand. Over the whole stream section, all three carbon determinands showed a net loss of carbon due to instream processes. Initially, at the uppermost sites there was a net gain of DOC. At 1.05 km downstream from the source, on the 2 low flow days, there was a sharp change from net gain to net loss; this continued
Fig. 5. Cumulative gains and losses of DOC, DIC and free CO2 downstream along Brocky Burn on Ža. 14r07r1997, Žb. 11r06r1998 and Žc. 07r07r1998.
162
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
with a number of fluctuations until 2.00 km downstream. At this point the net loss of DOC increased markedly. On the day with the highest flow Ž11r06r1998., the initial net gain in the upper part of the catchment was maintained until 2.00 km downstream, where there was also a sharp change to net loss; this persisted until the catchment outlet. Along the length of the stream, between 5.64 and 41.5 mg C sy1 was removed as DOC from the streamwater column. There were only small cumulative gains and losses of DIC from the stream source to 1.92 km downstream, when the magnitude of the losses and gains increased on all 3 days. These changes culminated in a final net loss of between 2.52 and 16.2 mg C sy1 as DIC. Although the free CO2 fluxes on all 3 days initially showed a very small net gain Ž1 mg C sy1 ., by 1.00 km downstream this changed to a net loss. Although there was a slight gain of CO2 between 1.50 and 1.75 km, this was not enough to prevent a net loss of CO2 at the outlet of between 6.81 and 19.0 mg C sy1 . A comparison of the net losses of carbon determinands as a percent of the overall fluxes at the outlet at the catchment ŽTable 4., showed that between 11.6% and 17.6% of the total DOC flux at the lowermost site was lost along the stream length. On the 2 days of lower discharge, 67.1% and 43.4% of the DIC flux at the lowermost site was lost by in-stream processes. However, on the higher discharge day, the DIC flux at the lowermost site was smaller due to the low inorganic carbon concentrations. Thus, the net loss of DIC far exceeded the final flux at the flume. Net losses of free CO2 are greater than the CO2 flux measured at the flume by 370% on the highest discharge day and 841᎐997% on the two lower discharge days.
4. Discussion 4.1. Spatial ¨ ariability in DOC, DIC and free CO2 concentrations Although DOC concentrations are known to be strongly influenced by discharge ŽGrieve, 1984.,
Table 4 Net loss of carbon calculated as a percentage of the flux at the outlet of the catchment Sampling date
DOC
DIC
Free CO 2
14r07r1997 11r06r1998 07r07r1998
11.6 11.9 17.6
67.1 1102 43.4
996 369 840
spatial patterns in DOC along streams are closely related to the soil changes within the catchment ŽRees et al., 1989.. The increased concentrations of DOC as the stream passes from rankers to peat was more marked during highest flow than on the 2 lower flow days, reflecting changes in hydrological pathway. After the DOC concentration reaches a maximum in the area of deepest peats, it began to decrease when the stream passes into an area of peaty podzols as the gradient of the stream increased. Although inputs of DOC from peaty podzols may be substantially less than DOC inputs to the stream from peats, the decrease in DOC concentration is relatively small in the lower part of the catchment. This may be due to the high upstream inputs of DOC not being significantly diluted further downstream. Dissolved inorganic carbon concentrations were variable along the catchment but always - 3.5 mg ly1 , reflecting the acidic nature of the stream and its low groundwater inputs ŽReid et al., 1981.. When flow was at its highest Ž11r06r1998., the pH was - 5 in the main section of the stream, shifting the carbonate equilibria in favour of free CO2 and maintaining low HCO3y concentrations. Although there are inputs of DIC from tributaries ŽTable 3., which maintain the HCOy 3 concentration in the upper and middle parts of the catchment, DIC is increasingly diluted downstream. The streamwater chemistry of individual tributaries may therefore show a greater groundwater influence, because of lower water volumes associated with fracture patterns in the underlying granite. Free CO2 concentrations in Brocky Burn are within the range reported elsewhere for headwater streams ŽRebsdorf et al., 1991; Pinol ˜ and Avila, 1992; Neal and Hill, 1994; Dawson et al., 1995.. CO2 concentrations are initially low, possi-
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
bly due to low soil respiration rates in the poorly developed mineral soils at the top of the catchment and the ease of escape to the atmosphere through thin coarse textured soils. A marked increase in CO2 concentration in the stream associated with the peat may be due to a marked decrease in the rate of CO2 escape from peaty soils to the atmosphere due to their low porosity ŽSkiba and Cresser, 1991.. This is further enhanced by continual inputs of CO2-rich water from tributaries, which have been reported before in a similar upland stream ŽDawson et al., 1995.. Tributaries tend to have a greater concentration of CO2 as they are shorter in length and thus have less distance to degas before reaching the main stem. There is a rapid decrease in CO2 concentrations at 1.74 km downstream from the source as inputs of CO2 from tributaries cease and the soils within the catchment change from peats to peaty podzols. Outgassing of CO2 from the main stream therefore becomes the dominant process in controlling CO2 concentration. These reduced inputs may be due to differences in the water holding capacity of the soils; deep wet peats will store dissolved CO2 in the soil solution more readily than the well aerated mineral soils and thus slow down exchange with the atmosphere. The decrease in concentration also corresponds to an increase in stream gradient and turbulence, which enhances the outgassing of CO2 to the atmosphere ŽRebsdorf et al., 1991.. CO2 concentrations eventually drop to a value just above that corresponding to equilibration with the atmosphere, as the high inputs from further up the catchment have been lost and inputs from the mineral soils lower down the catchment are less significant. 4.2. Tributaries Tributaries contribute to only 50.7% of the total discharge measured at the catchment outlet. This has implications for the concept of using mixtures of water from tributaries and sub-catchments to model and predict concentrations and solutes in river systems. The main sources of the large amount of unaccounted water are bank seepage and groundwater inputs through the
163
stream bed. Losses of water from the catchment through the soils and underlying granite will be negligible because the area is strongly indurated with a low permeability soil horizon forming an effective seal ŽReid et al., 1981.. Another possible source of input to the streams, are point sources of subsurface flow, such as the flush areas in the lower part of the catchment ŽFig. 1.. Although flushes ŽT7 and T8. were not flowing on 2 of the 3 sampling days, inputs were probably still occurring. The contribution of the tributaries to the overall discharge from the catchment on the higher flow day was 9% higher than that on the 2 other days. The fact that the contribution of tributaries to the overall catchment chloride flux was very similar to that for discharge, suggests that chloride does behave as a conservative ion in the system. DOC, DIC and CO2 contributions from tributaries show significant differences compared to discharge. This reflects the losses of these determinands from streamwater upstream of the lowermost site. 4.3. Assumptions in calculating carbon losses It is necessary to estimate the additional input of DOC, DIC and CO2 from diffuse sources Žbank seepage, groundwater inputs through the streambed., because only 51% of the flow into the stream can be sampled directly in the catchment. The assumption made is that the percent increase in the instantaneous chloride flux between two successive sample points Ž25᎐200 m apart. can be used to estimate the increase in DOC, DIC and CO2 fluxes caused by unmeasured inputs of water. Chloride concentrations from diffuse inputs are therefore assumed to be constant for the small distances between contiguous sites, which may not always be the case. Differences in chloride concentrations might be caused by weathering of chloride containing minerals Žnot present in this catchment . or changes in evapotranspiration rates. However, on individual sample days, the chloride concentration range of the stream did not vary by ) 0.75 mg ly1 along its length. This suggests that additional inputs of chloride-rich or chloride-poor water does not occur between adjacent sites.
164
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
The difference between Ž1. the sum of the measured upstream instantaneous flux and the estimated DOC, DIC and CO2 flux from diffuse sources; and Ž2. the instantaneous flux at the next downstream site, enables losses or gains caused by in-stream processing to be estimated. The approach used to estimate amounts of in-stream processing may lead to an underestimate of the amount of DOC lost by processing within the stream. This is because mean concentrations of DOC measured in soil solutions derived from two depths in the riparian zone and two depths in peat bogs in Brocky Burn are higher than streamwater concentrations and vary from 36 to 45 mg ly1 ŽBillett, unpublished data.. Also, soil solution DOC concentrations, derived from the Bs horizon of a podzol in an adjacent catchment in Glen Dye, had a mean DOC concentration of 28.3 mg ly1 ŽKennedy, unpublished data.. These concentrations are significantly higher than the estimated concentrations of DOC from diffuse sources, which are - 10 mg ly1 . This can be calculated by dividing the differences in the instantaneous DOC fluxes by the amount of additional water introduced between two sites. In terms of DIC and free CO2 , which already have large within-stream losses compared with the final flux at the outlet from the catchment, the underlying assumption based on Cly would also tend to underestimate the amounts of these determinands lost by in-stream processing. 4.4. Gain and loss of DOC Within the stream channel there was a net gain in DOC in the upper part of the catchment on all three occasions. Production of DOC within the water column might occur from the breakdown of coarse or fine POC by microbial action ŽVannote et al., 1980.. Net losses of DOC occur by 1.05 km downstream on the low discharge days and by 2.25 km downstream on the high flow day. Losses of DOC are due to biotic processes, such as biofilm respiration and adsorption onto algae ŽThomas, 1997. and to abiotic uptake, such as adsorption onto mineral surfaces Žincluding Fe and Al oxides and hydroxides. and precipitation
of POC ŽDavis, 1982.. Total Al concentration in this stream varied between 67 and 99 g ly1 for a similar time of year ŽStutter, unpublished data.; this could provide a potential substrate for significant adsorption and removal of DOC. Overall, there appear to be a series of interactive processes controlling gains and losses of DOC within the water column, which are influenced by flow conditions and the degree of biotic activity within the stream. Continual increases in the amount of DOC lost in the lower reaches of the catchment, suggest that losses of DOC by biological uptake and abiotic adsorption are important. The two low flow days followed a period of dry weather, whereas the higher flow sampling was carried out at the tail-end of the descending limb of the hydrograph. This period of high flow may have removed much of the biofilms and algae from the stream environment. It is possible that this shifts the equilibria in the catchment, particularly in the top and central areas, towards processes that favour gains of DOC Že.g. formation of DOC from fine POC. and away from loss processes Žbiological uptake by metabolism and adsorption.. 4.5. Gain and loss of DIC Very little change in instantaneous DIC fluxes occurred in the upper reaches of the catchment until 1.74 km downstream from the source, when a net loss of HCO3y occurred, possibly due to increased biological activity within the stream. However, DIC uptake is likely to be minimal because other carbon sources also exist for biota. These losses may also be connected with the carbonate equilibria within the stream. As CO2 is removed from the stream by outgassing, HCO3y reacts with the high concentrations of Hq present in the acidic stream forcing the equilibria back towards CO2 . There are also a number of groundwater inputs of HCO3y from tributaries. However, no tributaries occur in the lower part of the catchment, so losses of HCO3y are not replenished from groundwater or any other source and the magnitude of the net loss of HCO3y continues to increase.
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
4.6. Gain and loss of free CO2 The majority of CO2 is lost from the stream system by outgassing, as streamwater supersaturated in CO2 equilibrates with the atmosphere. CO2 may also be lost from the stream during daylight hours by photosynthesis ŽRebsdorf et al., 1991; Pinol ˜ and Avila, 1992.; we were not, however, able to differentiate between the two processes. Although greater amounts of CO2 are probably lost from the stream by outgassing, the net loss is lower than expected due to the replenishment of CO2 in the main stem from tributaries. That occurs until losses by outgassing become far greater than inputs at sites in the lower part of the catchment. Also in the lower section of the stream, rates of outgassing have decreased and there is a levelling out of net CO2 loss. In-stream photosynthesis will further increase CO2 loss especially during sunnier days. The similarities exhibited in the net losses and gains of DIC and CO2 are not unexpected due to the chemical link between the two via the carbonate equilibria. This link may also be enhanced by the method of analysis because samples measured for DIC will contain minor amounts of CO2 , which remains after equilibration with the atmosphere ŽHope et al., 1995.. 4.7. O¨ erall effect on total carbon fluxes The percent losses of DOC, DIC and free CO2 compared with the fluxes at the outlet are shown in Table 4. In-stream DOC loss was between 11.6 and 17.6% of its final flux from the catchment. This is similar to losses reported in other work by Fisher and Likens Ž1973., where a 34% energy loss was reported in a stream within the Hubbard Brook Experimental Forest, New Hampshire. However, this energy loss does not just represent DOC, which was estimated as 47% of the total energy input to the stream. Hubbard Brook is also a more productive higher pH system with the stream receiving significant litter inputs from a mixed conifer-hardwood forest. High rates of removal of added bovine manure extract Ž31%. and jewel-weed leachate Ž19%. were observed by Kuserk et al. Ž1984.. The contrast in the removal
165
rates between the two types of added carbon suggests that natural differences in within stream processing are dependent on substrate type. In comparison to Brocky Burn, both these studies were undertaken in higher order streams where processing of DOC would be enhanced by higher pH values and temperatures. The losses of DIC and free CO2 measured at Brocky Burn reflect the greater losses of these determinands along the stream length compared to the relatively small flux at the outlet of the catchment. In particular, the flux of CO2 to the atmosphere constitutes a significant loss of CO2 from the streamwater, which is not measured at the catchment outlet.
5. Conclusions Two main processes control spatial concentrations of streamwater DOC, DIC and free CO2 in the Brocky Burn catchment. The main control on streamwater DOC concentrations are the spatial differences between soil water inputs from different parts of the catchment as the stream flows through different soil types. However, in-stream processes modify these DOC inputs, thus providing a secondary control on the spatial variability of DOC concentrations and fluxes. Although instream processes reduce the final DOC flux at the catchment outlet by 11.6᎐17.6%, hydrological pathways control initial amounts of carbon reaching the stream. Similar processes also control the concentrations and fluxes of DIC and CO2 in the headwater. However, in-stream processes Žincluding outgassing of CO2 . have significantly more influence on spatial changes occurring in these two determinands than on DOC. On all three sampling occasions, spatial changes in the individual determinands showed similar downstream trends and reflect different provenances of DOC, DIC and free CO2 in the catchment. This is exemplified by the increase in DOC and CO2 concentrations as the stream enters the carbon-rich peat areas in the central part of the catchment, after flowing through the poorly developed mineral soils at the head of the catchment. Thus, stream sources are not always supersaturated with respect to CO2 ŽDawson et al.,
166
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167
1995., but depend on soil CO2 concentrations supplying the stream at that point. The data presented here are from a small headwater stream draining an upland catchment dominated by organic soils. In larger, higher order river systems in-stream processing would probably have a greater effect on DOC flux because they tend to be less acidic and have higher water temperatures. Chloride appears to behave conservatively in this headwater stream and can be used to determine diffuse inputs of water from bank seepage and groundwater into the main stem. This assumption possibly under-estimates the amount of in-stream processing because the carbon concentration of the water derived from diffuse sources is likely to be higher than the value which has been assumed. Moreover, in-stream processing along the length of the tributaries, prior to reaching the main stem, was not accounted for in our estimate of in-stream losses. Sampling, however, was carried out in summer daylight hours when conditions for in-stream processing would be expected to be close to optimal. There would almost certainly be a reduction in the importance of in-stream processing during the colder lower light periods of the year.
Acknowledgements The authors would like to thank Fasque Estates for allowing access to the site and their support for the research. J.J.C. Dawson wishes to thank the Natural Environment Research Council for providing financial support and Douglas Low for assistance with mapping. C. Bakewell acknowledges receipt of an E.C. scholarship. References Aitkenhead JA, Hope D, Billett MF. The relationship between dissolved organic carbon in streamwater and soil organic carbon pools at different spatial scales. Hydrolog Process 1999;13:1289᎐1302. Castelle AJ, Galloway JN. Carbon dioxide dynamics in acid forest soils in Shenandoah National Park, Virginia. Soil Sci Soc Am J 1990;54:252᎐257. Davis JA. Adsorption of natural dissolved organic matter at
the oxiderwater interface. Geochim Cosmochim Acta 1982;46:2381᎐2393. Dawson JJC, Hope D, Cresser MS, Billett MF. Downstream changes in free carbon dioxide in an upland catchment from northeastern Scotland. J Environ Qual 1995;24: 699᎐706. Ferrier RC, Jenkins A, Miller JD, Walker TAB, Anderson HA. Assessment of wet deposition mechanisms in an upland Scottish catchment. J Hydrol 1990;113:285᎐296. Fiebig DM, Lock MA. Immobilization of dissolved organic matter from groundwater discharging through the streambed. Freshwater Biol 1991;26:45᎐55. Fiebig DM, Lock MA, Neal C. Soil water in the riparian zone as a source of carbon for a headwater stream. J Hydrol 1990;116:217᎐237. Fisher SG, Likens GE. Energy flow in Bear brook, New Hampshire: an integrative approach to stream ecosystem metabolism. Ecol Monogr 1973;43:421᎐439. Grieve IC. Concentrations and annual loading of dissolved organic matter in a small moorland stream. Freshwater Biol 1984;14:533᎐537. Hoffer-French KJ, Herman JS. Evaluation of hydrological and biological influences on CO2 fluxes from a karst stream. J Hydol 1989;108:189᎐212. Hope D, Billett MF, Cresser MS. A review of the export of carbon in river water: fluxes and processes. Environ Pollut 1994;84:301᎐324. Hope D, Dawson JJC, Cresser MS, Billett MF. A method for measuring free CO2 in upland streamwater using headspace analysis. J Hydol 1995;166:1᎐14. Hope D, Billett MF, Milne R, Brown TAW. Exports of organic carbon in British Rivers. Hydrolog Process 1997;11: 325᎐344. Howard DM, Howard PJA. Relationships between CO2 evolution, moisture content and temperature for a range of soil types. Soil Biol Biochem 1993;25:1537᎐1546. Jones JB, Mulholland PJ. Influence of drainage basin topography and elevation on carbon dioxide and methane supersaturation of stream water. Biogeochemistry 1998;40:57᎐72. Kempe S. Long-term records of CO2 pressure fluctuations in fresh waters. In: Degens ET, editor. Scientific Committee on Problems of the Environment ŽSCOPE.rUnited Nations Environment Programme ŽUNEP. ᎏ Transport of Carbon and Minerals in Major World Rivers, 52. Hamburg, Germany: University of Hamburg, 1982:91᎐332. Kling GW, Kipphut GW, Miller MC. Arctic lakes and streams as gas conduits to the atmosphere: implications for tundra carbon budgets. Science 1991;251:298᎐301. Kling GW, Kipphut GW, Miller MC. The flux of CO2 and CH4 from lakes and rivers in Arctic Alaska. Hydrobiology 1992;240:23᎐36. Kuserk FT, Kaplan LA, Bott TL. In situ measures of dissolved organic carbon flux in a rural stream. Can J Fish Aquat Sci 1984;41:964᎐973. McDowell WH. Kinetics and mechanisms of dissolved organic carbon retention in a headwater stream. Biogeochemistry 1985;1:329᎐352.
J.J.C. Dawson et al. r The Science of the Total En¨ ironment 265 (2001) 153᎐167 Neal C, Hill S. Dissolved inorganic and organic carbon in moorland and forest streams: Plynlimon, mid-Wales. J Hydrol 1994;153:231᎐243. Pinol ˜ J, Avila A. Streamwater pH, alkalinity, pCO2 and discharge relationships in some forested Mediterranean catchments. J Hydrol 1992;131:205᎐225. Rebsdorf A, Thyssen N, Erlandsen M. Regional and temporal variation in pH, alkalinity and carbon dioxide in Danish streams, related to soil type and land use. Freshwater Biol 1991;25:419᎐435. Rees RM, Parker-Jarvis F, Cresser MS. Soil effects on water chemistry in three adjacent upland streams at Glen Dye in northeast Scotland. Water Res 1989;23:511᎐517. Reid JM, MacLeod DA, Cresser MS. Factors affecting the chemistry of precipitation and river water in an upland catchment. J Hydrol 1981;50:129᎐145.
167
Skiba U, Cresser MS. Seasonal changes in soil atmospheric CO2 concentrations in two upland catchments and associated changes in river water chemistry. Chem Ecol 1991; 7:217᎐225. Thomas JD. The role of dissolved organic matter, particularly free amino acids and humic substances, in freshwater ecosystems. Freshwater Biol 1997;38:1᎐36. Thurman EM. Organic geochemistry of natural waters. Dordrecht, Holland: NijhoffrJunk Publishers, 1985:497pp. Vannote RL, Minshall GW, Cummins KW, Sedell JR, Cushing CE. The river continuum concept. Can J Fish Aquat Sci 1980;37:130᎐137. Wetzel RG. Gradient dominated ecosystems: sources and regulatory functions of dissolved organic matter in freshwater ecosystems. Hydrobiology 1992;229:181᎐198.