Environmental Science Processes & Impacts View Article Online
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
PAPER
Cite this: Environ. Sci.: Processes Impacts, 2013, 15, 485
View Journal | View Issue
Linking fluorescence spectroscopy to diffuse soil source for dissolved humic substances in the Daning River, China Hao Chen,* Bing-hui Zheng* and Lei Zhang Dissolved organic matter collected in Daning River (China) in July 2009 was investigated with parallel factor analysis (PARAFAC) and fluorescence spectroscopy with the aim of identifying the origin of dissolved humic substance (HS) components. Two HS-like fluorescence components (peak M and C) with excitation/ emission (ex/em) maxima at 305/406 nm and 360/464 nm showed relatively uniform distribution in the vertical direction for each sampling site but a trend of accumulation down the river, independent of the highly heterogeneous water environment as implicated by water quality parameters (i.e., water temperature, algae density, chlorophyll a, dissolved oxygen, dissolved organic carbon, pH, conductivity and turbidity), while an amino acid/protein-like component (peak T; ex/em ¼ 280/334 nm) was quite variable in its spatial distribution, implying strong influence from point sources (e.g. sewage discharge) and local microbial activities. The fluorescence intensity (Fmax in Raman units) at these ex/em wavelength pairs fell in the range of 0.031–0.358, 0.051–0.224 and 0.026–0.115 for peak T, M and C, respectively. In addition, the Fmax values of peak C covaried with M (i.e. C ¼ 0.503 M, p < 0.01, R2 ¼ 0.973). Taken together, these results indicate that peak M and C originated primarily and directly from the same soil sources that were diffusive in the catchment, but peak T was more influenced by local
Received 16th May 2012 Accepted 6th November 2012
point sources (e.g. wastewater discharge) and in situ microbial activities. This study presents new insights into the currently controversial origin of some HS components (e.g. “peak M”, as commonly referred to in the literature). This study highlights that natural water samples should be collected at various depths
DOI: 10.1039/c2em30715d
in addition to along a river/stream flow path so as to better evaluate the origin of HS fluorescence
rsc.li/process-impacts
components.
Environmental impact Dissolved humic substances (HS) play an essential role in the environment. Although uorescence spectroscopy coupled with a well-established chemometrics (parallel factor analysis; PARAFAC) has proven to be a promising tool in monitoring dissolved HS components in natural water, the geochemical origin of some HS uorescence components (e.g. peak M ) still remains controversial. This study presents some results to show that a special sampling strategy applied to a special river catchment in conjunction with correlation analysis nally pointed to a different viewpoint compared with the one currently accepted in the literature, and indicates that such peak M in freshwater may also be attributed directly to soil origin. This study therefore gives insight into the origin of dissolved humic substances.
Introduction The concentration of humic substance (HS)-related dissolved organic carbon (DOC) has been observed to increase in natural water, and the increase originates in part from global warming, which has the potential to alter the decomposition rate of soil organic matter (SOM, an important source of dissolved HS) and rainfall patterns, thereby resulting in increased transfer of soil
State Environmental Protection Key Laboratory of Estuarine and Coastal Research, Chinese Research Academy of Environmental Sciences, Beijing 100012, China. E-mail: [email protected]; [email protected]; Fax: +86-10-84913914; Tel: +86-10-84913914
This journal is ª The Royal Society of Chemistry 2013
HS to dissolved phases (i.e. natural leaching of HS).1–3 The increase of HS concentration in freshwater ecosystems is causing increasing concerns,4–9 mostly due to the potential of dissolved HS to be involved in a range of physiochemical and biological processes that are of relevance to environmental and human health. For example, dissolved HS play a central role in inuencing the transport of pollutants as well as their speciation and bioavailability in aquatic environments.10 Photomineralization of dissolved HS may lead to the release of bioavailable inorganic nutrients that may be subsequently utilized by aquatic microbes, thus increasing the odds of algal bloom and threatening drinking water source security.11 In addition, the traditional chlorination processes applied to HS in
Environ. Sci.: Processes Impacts, 2013, 15, 485–493 | 485
View Article Online
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Environmental Science: Processes & Impacts drinking water works can produce carcinogenic by-products. Dissolved HS may also be subject to heterotrophic uptake, but the process is dependent on the sub-components of HS and on nutrient availability.12,13 For large freshwater ecosystems like globally large rivers, a slight change in the composition of dissolved HS as a result of photochemical and biological processes will impose considerable impact on carbon cycling on a large geographical scale.4,5,14 Numerous analytical efforts have been devoted to tracing the generation and removal of dissolved HS in freshwater ecosystems. To this end, uorescence spectroscopy has shown to be a promising tool in that it offers low instrumental maintenance and high detection sensitivity and is easy to measure; more importantly, modern uorescence instrumentation can be readily incorporated in on-line monitoring systems for water environment monitoring on a regular basis.15 In addition, uorescence excitation emission matrix (EEM) coupled with a well-established chemometrics tool (parallel factor analysis, PARAFAC) has provided a new dimension to the conventional uorescence analysis, allowing for retrieval of a great deal of information regarding the composition of the organic carbon pool and separation of HS-like uorescence signatures from non-HS signatures (e.g., protein) in natural water.7,13,16 Although uorescence spectroscopy cannot directly provide a value for the concentration of HS, it does offer a potentially sensitive approach to evaluate the relative abundance of HS in natural water. Numerous studies have reported multiple HS uorescence components over a range of natural water samples, but the origin of HS components still remains controversial. Most of the studies identied the HS origin by comparing the emission and excitation spectra of the HS components with previously reported spectra associated with “known” origins. For example, HS uorescence components referred to as “peak A”, “peak M” and “peak C” showing distinct uorescence excitation/emission centers were rst reported by Coble17 in a study on seawater. In contrast to other peaks, peak M was considered to be generated in a marine environment;17 as such a peak or similar has also been observed in other DOM studies on seawater and brackish water,18–20 with microbial origin proven or suggested to be involved for peak M or its analogues in those studies. However, some studies21 indicated that peak M or similar may also have a terrestrial origin in addition to the traditionally accepted marine microbial origin. Moreover, similar peaks have also been observed in freshwater ecosystems, and are attributed to aquatic microbial origins.12,13,22 It should be noted that spectral similarities between a peak M in a new study and a reported peak M in previous studies does not necessarily mean that the new peak M should have the same origin. This uncertainty is in part due to the inability of linking uorescence spectral features to specic molecular structures. Therefore, it still remains uncertain whether peak M or similar in some freshwater watersheds could be primarily attributed to soil origins rather than microbial origins. Given the diffuse nature of soil sources along a river, it might be expected that an accumulation trend for HS may be observed down the river ow path, and such accumulation has been 486 | Environ. Sci.: Processes Impacts, 2013, 15, 485–493
Paper observed in some river DOM studies,4 but the literature studies do not look into the distribution of HS uorescent components at different sampling depths. Fluorescence data at different depths is supposed to give more insight into HS origins. For example, if peak M preferentially exists in the river surface water rather than in the middle and bottom water, this may indicate that photochemical processes are likely involved in peak M generation. Moreover, auxiliary water quality parameters (e.g. water temperature, algal density, chlorophyll a concentration) measured in the eld may also add new dimensions to the uorescence data interpretation, as these parameters may directly tell whether the water environment in the eld is heterogeneous with respect to sampling depths and sites. Provided that a given water column is heterogeneous in terms of these parameters, a relatively uniform distribution of HS at different depths may provide further support to the hypothesis of a diffuse soil source rather than microbial origin, because the latter origin is susceptible to local heterogeneous water environment. The objectives of the present work were to: (1) identify the primary uorescent HS components in the Daning River (China) watershed by applying EEMs-PARAFAC to water samples collected at different depths and sites, (2) evaluate the spatial distribution of the HS uorescence signatures in relation to other non-HS uorescence signatures and selected water quality parameters, and (3) evaluate whether diffusive soil sources may play a role in accounting for HS uorescence signatures in the catchment.
Experimental details Sampling The Daning River was selected for this study due to several reasons related to the Yangtze River, which is the world's thirdlargest23 and China's longest river (ca. 6397 km in length), having the largest drainage basin (ca. 1.8 million km2 in area), the highest annually averaged discharge of 960 billion cubic meters, and providing the most important freshwater sources to approximately one third of the population in China. The operation of the Three Gorges Dam (TGD), the world's largest hydropower station (http://en.wikipedia.org/wiki/ Three_Gorges_Dam), in the Yangtze River will change the hydrology of the watershed. The increased water residence time upstream of the TGD may be as high as 77 days or even more in backwater areas,24 thereby presumably inuencing the natural leaching of HS into the water body. Increasing attention has been devoted to evaluating the effects of the TGD on DOM dynamics in various regions of the Yangtze River watershed.25,26 The Daning River (162 km in length, 136 m3 s1 in mean discharge) is a tributary of the Yangtze River, located in the upstream region of the TGD and owing through a series of valleys which are collectively called the Three Little Gorges and are a scenic spot of increasing popularity. The Daning River serves as a back-up drinking water source for a nearby city called Wushan City which normally uses mountain springs and streams to produce drinking water. Due to the operation of the TGD, the water level has increased in the Daning River resulting
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Paper
Environmental Science: Processes & Impacts
in low ow velocity (ca. 0.00–0.05 m s1) and high water retention time, while the water level may decrease by nearly 30 meters in seasons when the TGD is fully opened to facilitate fast dissipation of the incoming water from upstream to mitigate ooding. Ever since the operation of the TGD, its potential adverse impacts on the ecosystem have been brought to attention. In this regard, the monitoring of dissolved HS in the Daning River watershed is of interest and has been incorporated into a long-term environment monitoring programme that aims to evaluate how the whole ecosystem in the Yangtze River watershed is varying in response to TGD operation and global warming. A eld sampling campaign was conducted on 17 July 2009. During July, the TGD was opened to maintain a relatively low water level at approximately 150 m relative to sea level. As illustrated in Fig. 1, eight sites were surveyed for water sampling. Water from three layers (surface, middle and bottom) at each sampling site (2–8) was collected. Site 1 was located most upstream relative to other sites and was shallow (3 m; Table 1) during our sampling course; therefore, only the surface and bottom water in this site was collected. We therefore had 23 water samples from eight sites for study. Sites 1 and 2 were far from Wushan City and therefore were presumably under the weakest anthropogenic inuences. Sites 3–6 were located in valleys where forested landscape dominated. Sites 7 and 8 were near the conuence point of the Yangtze River and the Daning River, where Wushan City is located and wastewater effluent is discharged. As the sampling was conducted in a dry season without rainfall, the surface runoff of different types of DOM end members from nonpoint sources (e.g., agricultural land and forested land) during the sampling was minimized, if any occurred, and this sampling strategy was different from those
Fig. 1
employed in previous studies, e.g. Stedmon and Markager20 that intentionally collected samples under various weather conditions. Additionally, we did not use an event-based sampling strategy (e.g., during algal bloom) during our sampling period like other literature studies did, and thus the uorescence signature in this study would represent largely the background uorescence signatures of DOM in the watershed. The water samples were collected with Niskin water samplers (HYDRO-BIOS, Waterloos Inc., China). Water depth, pH, temperature, turbidity, conductivity, concentration of dissolved oxygen (DO) and chlorophyll a as well as blue-green algae (cyanobacteria) density were measured in situ by a multi-probe system YSI 6600 installed with several probes. A YSI 6131 probe was deployed to detect the blue-green algae density by observing the uorescence from a uorescent pigment (phycocyanin) that only exists in cyanobacteria,11 while a YSI 6025 probe measured the concentration of chlorophyll a by observing its uorescence emitted from all types of algae present that contained chlorophyll a. Therefore, blue-green algae density may not necessarily show correlation with chlorophyll a concentration.
Laboratory methods The water samples were immediately transported on ice back to the laboratory where the waters were passed through 0.45 mm nylon membrane lters (Millipore) that had been rinsed twice with high purity water (2 200 mL; resistivity > 18 MU, DOC # 0.06 mg L1) prior to ltration. The ltrate was stored at 4 C in the dark before optical and DOC measurements were performed within one day. EEMs were measured on a Hitachi 4500 uorometer. Excitation and emission wavelengths ranged from 250 to 580 nm and 280 to 600 nm with 5 nm and 2 nm intervals, respectively.
Sampling stations in the Daning River watershed. The arrows designate the water flow direction.
This journal is ª The Royal Society of Chemistry 2013
Environ. Sci.: Processes Impacts, 2013, 15, 485–493 | 487
View Article Online
Environmental Science: Processes & Impacts Table 1
Water quality parameters measured in situ at different water layers (S for surface; M for middle; B for bottom) for 8 sampling sites
Site
Layer
Water temperature ( C)
Dissolved oxygen (mg L1)
Sampling depth (m)
Conductivity (ms cm1)
pH
Algae density (cells per L)
Chlorophyll a (mg L1)
Turbidity (NTU)
1
S B S M B S M B S M B S M B S M B S M B S M B
19.83 19.77 26.24 19.65 18.76 30.67 23.87 19.25 30.9 24.2 19.34 29.95 21.24 18.64 26.31 24.65 24.48 25.07 25.09 25.1 25.09 25.04 24.89
7.74 6.99 7.57 6.95 0.4 9.06 5.69 2.47 9.08 5.45 7.04 10.4 5.61 7.82 6.88 5.26 4.8 5.93 5.15 5.08 6.02 5.06 3.25
0 3 0 10 20 0 10 20 0 13 25 0 15 30 0 13 25 0 20 35 0 40 80
238.7 239.6 255.8 216.8 203.9 257.8 257.6 270.9 251.3 276.3 226.3 262.7 271.5 212.6 264.8 265.2 268.4 206.8 206.8 261.6 260.4 260.7 470.2
8.0 8.1 8.4 8.0 7.9 8.7 7.7 7.8 8.9 7.8 7.9 8.7 7.8 7.9 7.9 7.8 7.8 7.7 7.7 7.8 7.8 7.8 7.7
27989 30132 26267 32482 20428 33514 30037 56696 42247 26621 19301 42567 28152 25016 28758 38063 45284 54228 61116 73591 64987 79980 19907
0.42 0.50 3.10 0.64 0.21 8.98 0.29 16.23 12.90 1.01 0.18 17.97 0.96 0.59 2.60 1.24 1.00 1.32 1.26 1.30 1.23 1.40 2.53
52.2 90.5 7.2 63 2.8 8.6 49.2 20.2 10.3 41.2 5.2 19.8 48.8 29.1 58.1 84.7 88.5 216.8 271.3 328.4 300.4 411 3.2
2
3
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Paper
4
5
6
7
8
The bandwidths of the excitation and emission light were set to 5 nm. The day-to-day lamp intensity uctuation was corrected by comparing with the intensity of a pure water Raman scattering peak at 397 nm generated by a lamp light at 350 nm. Excitation light intensity incident on the uorescence cuvette as a function of lamp output wavelength, and uorescence detector sensitivity as a function of uorescence wavelength, were both corrected as recommended by the uorometer user's manual. The scattering light signals in the EEMs were eliminated from the analysis according to an approach employed previously,7,27,28 which consisted in replacing the data region containing scattering light signals with “NaN” (not a number) using a home-written program implemented in a Matlab 6.5 soware package (Mathworks, Natick, US). The UV-Vis absorbance at wavelengths of 250 through 600 nm with 1 nm intervals was measured in 1 cm quartz cuvettes on a Shimadzu 1700 spectrophotometer in double-beam mode. Freshly prepared pure water (DOC < 0.06 mg L1) was used as absorption reference. The uorescence intensities were corrected for inner lter effects (IFEs) according to a method proposed by MacDonald et al.29 and later improved by Gu and Kenny.30 This method has proven efficient in IFE correction for uorescence data in previous PARAFAC studies.25,27,31,32 The concentrations of DOC were determined by high temperature combustion on a Shimadzu TOC-VCPH instrument with a detection limit of 0.06 mg L1. A non-purgeable DOC measurement mode was employed, in which the sample was acidied to pH 2 by aliquots of concentrated HCl and then purged with high purity (>99.995%) oxygen gas to remove inorganic carbon before determination of DOC concentration. The non-purgeable mode is suitable for natural water that
488 | Environ. Sci.: Processes Impacts, 2013, 15, 485–493
contains high concentrations of dissolved inorganic carbon relative to DOC.33,34 It is well-known that the natural water in the Yangtze River and its tributaries ows through carbonate minerals and contains high concentrations of dissolved inorganic carbon.35,36 Average DOC concentrations in triplicate measurements are reported in this study.
Modeling and statistics The corrected EEMs (23 sample 161 emission 67 excitation) were modeled by PARAFAC following the procedures employed elsewhere.7,25 A core consistency (CC) diagnosis25,37 was employed to judge whether an appropriate number of components were correctly chosen to model the EEMs. If the CC value is relatively high (>80) for an n-component model, while much lower (99%) explained variance percentage should be observed for an appropriate PARAFAC modeling in most cases.31 Fluorescence intensities at peak excitation/emission wavelength pairs of a given component, referred to as Fmax, were computed according to the component's spectral features (loading) and relative abundance (score) reported by PARAFAC models.27 Following Stedmon et al.,38 Fmax was divided by the integrated under-peak intensity of the Raman bands of pure water generated by an excitation of 350 nm on the same uorometer as for EEMs measurements, and was reported in Raman units (nm1) in this study. Linear regressions between any two components' Fmax values were undertaken with use of an R soware version 2.10.1. A linear regression can be considered statistically signicant
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Paper
Environmental Science: Processes & Impacts
when the p-value of the slope is 0.05, the intercept may be considered to not differ signicantly from zero, and as such the linear equations can be forced to have a zero intercept.
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Fluorescence-based diagnostics Fluorescence index (FI) was calculated as the uorescence intensity ratio at 450 nm and 500 nm at an excitation of 370 nm.39 The humication index (HIX) was dened as the ratio
of the integrated uorescence intensity during emission at a wavelength of 435–480 nm to the sum of the integrated uorescence intensity between wavelengths of 435–480 nm and 300–345 nm.40,41 Both FI and HIX were computed based upon the corrected uorescence data.
Results The water quality parameters measured in situ are presented in Table 1. Note that relatively low DO concentrations (
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
PAPER
Cite this: Environ. Sci.: Processes Impacts, 2013, 15, 485
View Journal | View Issue
Linking fluorescence spectroscopy to diffuse soil source for dissolved humic substances in the Daning River, China Hao Chen,* Bing-hui Zheng* and Lei Zhang Dissolved organic matter collected in Daning River (China) in July 2009 was investigated with parallel factor analysis (PARAFAC) and fluorescence spectroscopy with the aim of identifying the origin of dissolved humic substance (HS) components. Two HS-like fluorescence components (peak M and C) with excitation/ emission (ex/em) maxima at 305/406 nm and 360/464 nm showed relatively uniform distribution in the vertical direction for each sampling site but a trend of accumulation down the river, independent of the highly heterogeneous water environment as implicated by water quality parameters (i.e., water temperature, algae density, chlorophyll a, dissolved oxygen, dissolved organic carbon, pH, conductivity and turbidity), while an amino acid/protein-like component (peak T; ex/em ¼ 280/334 nm) was quite variable in its spatial distribution, implying strong influence from point sources (e.g. sewage discharge) and local microbial activities. The fluorescence intensity (Fmax in Raman units) at these ex/em wavelength pairs fell in the range of 0.031–0.358, 0.051–0.224 and 0.026–0.115 for peak T, M and C, respectively. In addition, the Fmax values of peak C covaried with M (i.e. C ¼ 0.503 M, p < 0.01, R2 ¼ 0.973). Taken together, these results indicate that peak M and C originated primarily and directly from the same soil sources that were diffusive in the catchment, but peak T was more influenced by local
Received 16th May 2012 Accepted 6th November 2012
point sources (e.g. wastewater discharge) and in situ microbial activities. This study presents new insights into the currently controversial origin of some HS components (e.g. “peak M”, as commonly referred to in the literature). This study highlights that natural water samples should be collected at various depths
DOI: 10.1039/c2em30715d
in addition to along a river/stream flow path so as to better evaluate the origin of HS fluorescence
rsc.li/process-impacts
components.
Environmental impact Dissolved humic substances (HS) play an essential role in the environment. Although uorescence spectroscopy coupled with a well-established chemometrics (parallel factor analysis; PARAFAC) has proven to be a promising tool in monitoring dissolved HS components in natural water, the geochemical origin of some HS uorescence components (e.g. peak M ) still remains controversial. This study presents some results to show that a special sampling strategy applied to a special river catchment in conjunction with correlation analysis nally pointed to a different viewpoint compared with the one currently accepted in the literature, and indicates that such peak M in freshwater may also be attributed directly to soil origin. This study therefore gives insight into the origin of dissolved humic substances.
Introduction The concentration of humic substance (HS)-related dissolved organic carbon (DOC) has been observed to increase in natural water, and the increase originates in part from global warming, which has the potential to alter the decomposition rate of soil organic matter (SOM, an important source of dissolved HS) and rainfall patterns, thereby resulting in increased transfer of soil
State Environmental Protection Key Laboratory of Estuarine and Coastal Research, Chinese Research Academy of Environmental Sciences, Beijing 100012, China. E-mail: [email protected]; [email protected]; Fax: +86-10-84913914; Tel: +86-10-84913914
This journal is ª The Royal Society of Chemistry 2013
HS to dissolved phases (i.e. natural leaching of HS).1–3 The increase of HS concentration in freshwater ecosystems is causing increasing concerns,4–9 mostly due to the potential of dissolved HS to be involved in a range of physiochemical and biological processes that are of relevance to environmental and human health. For example, dissolved HS play a central role in inuencing the transport of pollutants as well as their speciation and bioavailability in aquatic environments.10 Photomineralization of dissolved HS may lead to the release of bioavailable inorganic nutrients that may be subsequently utilized by aquatic microbes, thus increasing the odds of algal bloom and threatening drinking water source security.11 In addition, the traditional chlorination processes applied to HS in
Environ. Sci.: Processes Impacts, 2013, 15, 485–493 | 485
View Article Online
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Environmental Science: Processes & Impacts drinking water works can produce carcinogenic by-products. Dissolved HS may also be subject to heterotrophic uptake, but the process is dependent on the sub-components of HS and on nutrient availability.12,13 For large freshwater ecosystems like globally large rivers, a slight change in the composition of dissolved HS as a result of photochemical and biological processes will impose considerable impact on carbon cycling on a large geographical scale.4,5,14 Numerous analytical efforts have been devoted to tracing the generation and removal of dissolved HS in freshwater ecosystems. To this end, uorescence spectroscopy has shown to be a promising tool in that it offers low instrumental maintenance and high detection sensitivity and is easy to measure; more importantly, modern uorescence instrumentation can be readily incorporated in on-line monitoring systems for water environment monitoring on a regular basis.15 In addition, uorescence excitation emission matrix (EEM) coupled with a well-established chemometrics tool (parallel factor analysis, PARAFAC) has provided a new dimension to the conventional uorescence analysis, allowing for retrieval of a great deal of information regarding the composition of the organic carbon pool and separation of HS-like uorescence signatures from non-HS signatures (e.g., protein) in natural water.7,13,16 Although uorescence spectroscopy cannot directly provide a value for the concentration of HS, it does offer a potentially sensitive approach to evaluate the relative abundance of HS in natural water. Numerous studies have reported multiple HS uorescence components over a range of natural water samples, but the origin of HS components still remains controversial. Most of the studies identied the HS origin by comparing the emission and excitation spectra of the HS components with previously reported spectra associated with “known” origins. For example, HS uorescence components referred to as “peak A”, “peak M” and “peak C” showing distinct uorescence excitation/emission centers were rst reported by Coble17 in a study on seawater. In contrast to other peaks, peak M was considered to be generated in a marine environment;17 as such a peak or similar has also been observed in other DOM studies on seawater and brackish water,18–20 with microbial origin proven or suggested to be involved for peak M or its analogues in those studies. However, some studies21 indicated that peak M or similar may also have a terrestrial origin in addition to the traditionally accepted marine microbial origin. Moreover, similar peaks have also been observed in freshwater ecosystems, and are attributed to aquatic microbial origins.12,13,22 It should be noted that spectral similarities between a peak M in a new study and a reported peak M in previous studies does not necessarily mean that the new peak M should have the same origin. This uncertainty is in part due to the inability of linking uorescence spectral features to specic molecular structures. Therefore, it still remains uncertain whether peak M or similar in some freshwater watersheds could be primarily attributed to soil origins rather than microbial origins. Given the diffuse nature of soil sources along a river, it might be expected that an accumulation trend for HS may be observed down the river ow path, and such accumulation has been 486 | Environ. Sci.: Processes Impacts, 2013, 15, 485–493
Paper observed in some river DOM studies,4 but the literature studies do not look into the distribution of HS uorescent components at different sampling depths. Fluorescence data at different depths is supposed to give more insight into HS origins. For example, if peak M preferentially exists in the river surface water rather than in the middle and bottom water, this may indicate that photochemical processes are likely involved in peak M generation. Moreover, auxiliary water quality parameters (e.g. water temperature, algal density, chlorophyll a concentration) measured in the eld may also add new dimensions to the uorescence data interpretation, as these parameters may directly tell whether the water environment in the eld is heterogeneous with respect to sampling depths and sites. Provided that a given water column is heterogeneous in terms of these parameters, a relatively uniform distribution of HS at different depths may provide further support to the hypothesis of a diffuse soil source rather than microbial origin, because the latter origin is susceptible to local heterogeneous water environment. The objectives of the present work were to: (1) identify the primary uorescent HS components in the Daning River (China) watershed by applying EEMs-PARAFAC to water samples collected at different depths and sites, (2) evaluate the spatial distribution of the HS uorescence signatures in relation to other non-HS uorescence signatures and selected water quality parameters, and (3) evaluate whether diffusive soil sources may play a role in accounting for HS uorescence signatures in the catchment.
Experimental details Sampling The Daning River was selected for this study due to several reasons related to the Yangtze River, which is the world's thirdlargest23 and China's longest river (ca. 6397 km in length), having the largest drainage basin (ca. 1.8 million km2 in area), the highest annually averaged discharge of 960 billion cubic meters, and providing the most important freshwater sources to approximately one third of the population in China. The operation of the Three Gorges Dam (TGD), the world's largest hydropower station (http://en.wikipedia.org/wiki/ Three_Gorges_Dam), in the Yangtze River will change the hydrology of the watershed. The increased water residence time upstream of the TGD may be as high as 77 days or even more in backwater areas,24 thereby presumably inuencing the natural leaching of HS into the water body. Increasing attention has been devoted to evaluating the effects of the TGD on DOM dynamics in various regions of the Yangtze River watershed.25,26 The Daning River (162 km in length, 136 m3 s1 in mean discharge) is a tributary of the Yangtze River, located in the upstream region of the TGD and owing through a series of valleys which are collectively called the Three Little Gorges and are a scenic spot of increasing popularity. The Daning River serves as a back-up drinking water source for a nearby city called Wushan City which normally uses mountain springs and streams to produce drinking water. Due to the operation of the TGD, the water level has increased in the Daning River resulting
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Paper
Environmental Science: Processes & Impacts
in low ow velocity (ca. 0.00–0.05 m s1) and high water retention time, while the water level may decrease by nearly 30 meters in seasons when the TGD is fully opened to facilitate fast dissipation of the incoming water from upstream to mitigate ooding. Ever since the operation of the TGD, its potential adverse impacts on the ecosystem have been brought to attention. In this regard, the monitoring of dissolved HS in the Daning River watershed is of interest and has been incorporated into a long-term environment monitoring programme that aims to evaluate how the whole ecosystem in the Yangtze River watershed is varying in response to TGD operation and global warming. A eld sampling campaign was conducted on 17 July 2009. During July, the TGD was opened to maintain a relatively low water level at approximately 150 m relative to sea level. As illustrated in Fig. 1, eight sites were surveyed for water sampling. Water from three layers (surface, middle and bottom) at each sampling site (2–8) was collected. Site 1 was located most upstream relative to other sites and was shallow (3 m; Table 1) during our sampling course; therefore, only the surface and bottom water in this site was collected. We therefore had 23 water samples from eight sites for study. Sites 1 and 2 were far from Wushan City and therefore were presumably under the weakest anthropogenic inuences. Sites 3–6 were located in valleys where forested landscape dominated. Sites 7 and 8 were near the conuence point of the Yangtze River and the Daning River, where Wushan City is located and wastewater effluent is discharged. As the sampling was conducted in a dry season without rainfall, the surface runoff of different types of DOM end members from nonpoint sources (e.g., agricultural land and forested land) during the sampling was minimized, if any occurred, and this sampling strategy was different from those
Fig. 1
employed in previous studies, e.g. Stedmon and Markager20 that intentionally collected samples under various weather conditions. Additionally, we did not use an event-based sampling strategy (e.g., during algal bloom) during our sampling period like other literature studies did, and thus the uorescence signature in this study would represent largely the background uorescence signatures of DOM in the watershed. The water samples were collected with Niskin water samplers (HYDRO-BIOS, Waterloos Inc., China). Water depth, pH, temperature, turbidity, conductivity, concentration of dissolved oxygen (DO) and chlorophyll a as well as blue-green algae (cyanobacteria) density were measured in situ by a multi-probe system YSI 6600 installed with several probes. A YSI 6131 probe was deployed to detect the blue-green algae density by observing the uorescence from a uorescent pigment (phycocyanin) that only exists in cyanobacteria,11 while a YSI 6025 probe measured the concentration of chlorophyll a by observing its uorescence emitted from all types of algae present that contained chlorophyll a. Therefore, blue-green algae density may not necessarily show correlation with chlorophyll a concentration.
Laboratory methods The water samples were immediately transported on ice back to the laboratory where the waters were passed through 0.45 mm nylon membrane lters (Millipore) that had been rinsed twice with high purity water (2 200 mL; resistivity > 18 MU, DOC # 0.06 mg L1) prior to ltration. The ltrate was stored at 4 C in the dark before optical and DOC measurements were performed within one day. EEMs were measured on a Hitachi 4500 uorometer. Excitation and emission wavelengths ranged from 250 to 580 nm and 280 to 600 nm with 5 nm and 2 nm intervals, respectively.
Sampling stations in the Daning River watershed. The arrows designate the water flow direction.
This journal is ª The Royal Society of Chemistry 2013
Environ. Sci.: Processes Impacts, 2013, 15, 485–493 | 487
View Article Online
Environmental Science: Processes & Impacts Table 1
Water quality parameters measured in situ at different water layers (S for surface; M for middle; B for bottom) for 8 sampling sites
Site
Layer
Water temperature ( C)
Dissolved oxygen (mg L1)
Sampling depth (m)
Conductivity (ms cm1)
pH
Algae density (cells per L)
Chlorophyll a (mg L1)
Turbidity (NTU)
1
S B S M B S M B S M B S M B S M B S M B S M B
19.83 19.77 26.24 19.65 18.76 30.67 23.87 19.25 30.9 24.2 19.34 29.95 21.24 18.64 26.31 24.65 24.48 25.07 25.09 25.1 25.09 25.04 24.89
7.74 6.99 7.57 6.95 0.4 9.06 5.69 2.47 9.08 5.45 7.04 10.4 5.61 7.82 6.88 5.26 4.8 5.93 5.15 5.08 6.02 5.06 3.25
0 3 0 10 20 0 10 20 0 13 25 0 15 30 0 13 25 0 20 35 0 40 80
238.7 239.6 255.8 216.8 203.9 257.8 257.6 270.9 251.3 276.3 226.3 262.7 271.5 212.6 264.8 265.2 268.4 206.8 206.8 261.6 260.4 260.7 470.2
8.0 8.1 8.4 8.0 7.9 8.7 7.7 7.8 8.9 7.8 7.9 8.7 7.8 7.9 7.9 7.8 7.8 7.7 7.7 7.8 7.8 7.8 7.7
27989 30132 26267 32482 20428 33514 30037 56696 42247 26621 19301 42567 28152 25016 28758 38063 45284 54228 61116 73591 64987 79980 19907
0.42 0.50 3.10 0.64 0.21 8.98 0.29 16.23 12.90 1.01 0.18 17.97 0.96 0.59 2.60 1.24 1.00 1.32 1.26 1.30 1.23 1.40 2.53
52.2 90.5 7.2 63 2.8 8.6 49.2 20.2 10.3 41.2 5.2 19.8 48.8 29.1 58.1 84.7 88.5 216.8 271.3 328.4 300.4 411 3.2
2
3
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Paper
4
5
6
7
8
The bandwidths of the excitation and emission light were set to 5 nm. The day-to-day lamp intensity uctuation was corrected by comparing with the intensity of a pure water Raman scattering peak at 397 nm generated by a lamp light at 350 nm. Excitation light intensity incident on the uorescence cuvette as a function of lamp output wavelength, and uorescence detector sensitivity as a function of uorescence wavelength, were both corrected as recommended by the uorometer user's manual. The scattering light signals in the EEMs were eliminated from the analysis according to an approach employed previously,7,27,28 which consisted in replacing the data region containing scattering light signals with “NaN” (not a number) using a home-written program implemented in a Matlab 6.5 soware package (Mathworks, Natick, US). The UV-Vis absorbance at wavelengths of 250 through 600 nm with 1 nm intervals was measured in 1 cm quartz cuvettes on a Shimadzu 1700 spectrophotometer in double-beam mode. Freshly prepared pure water (DOC < 0.06 mg L1) was used as absorption reference. The uorescence intensities were corrected for inner lter effects (IFEs) according to a method proposed by MacDonald et al.29 and later improved by Gu and Kenny.30 This method has proven efficient in IFE correction for uorescence data in previous PARAFAC studies.25,27,31,32 The concentrations of DOC were determined by high temperature combustion on a Shimadzu TOC-VCPH instrument with a detection limit of 0.06 mg L1. A non-purgeable DOC measurement mode was employed, in which the sample was acidied to pH 2 by aliquots of concentrated HCl and then purged with high purity (>99.995%) oxygen gas to remove inorganic carbon before determination of DOC concentration. The non-purgeable mode is suitable for natural water that
488 | Environ. Sci.: Processes Impacts, 2013, 15, 485–493
contains high concentrations of dissolved inorganic carbon relative to DOC.33,34 It is well-known that the natural water in the Yangtze River and its tributaries ows through carbonate minerals and contains high concentrations of dissolved inorganic carbon.35,36 Average DOC concentrations in triplicate measurements are reported in this study.
Modeling and statistics The corrected EEMs (23 sample 161 emission 67 excitation) were modeled by PARAFAC following the procedures employed elsewhere.7,25 A core consistency (CC) diagnosis25,37 was employed to judge whether an appropriate number of components were correctly chosen to model the EEMs. If the CC value is relatively high (>80) for an n-component model, while much lower (99%) explained variance percentage should be observed for an appropriate PARAFAC modeling in most cases.31 Fluorescence intensities at peak excitation/emission wavelength pairs of a given component, referred to as Fmax, were computed according to the component's spectral features (loading) and relative abundance (score) reported by PARAFAC models.27 Following Stedmon et al.,38 Fmax was divided by the integrated under-peak intensity of the Raman bands of pure water generated by an excitation of 350 nm on the same uorometer as for EEMs measurements, and was reported in Raman units (nm1) in this study. Linear regressions between any two components' Fmax values were undertaken with use of an R soware version 2.10.1. A linear regression can be considered statistically signicant
This journal is ª The Royal Society of Chemistry 2013
View Article Online
Paper
Environmental Science: Processes & Impacts
when the p-value of the slope is 0.05, the intercept may be considered to not differ signicantly from zero, and as such the linear equations can be forced to have a zero intercept.
Published on 02 January 2013. Downloaded by McMaster University on 09/12/2013 20:04:11.
Fluorescence-based diagnostics Fluorescence index (FI) was calculated as the uorescence intensity ratio at 450 nm and 500 nm at an excitation of 370 nm.39 The humication index (HIX) was dened as the ratio
of the integrated uorescence intensity during emission at a wavelength of 435–480 nm to the sum of the integrated uorescence intensity between wavelengths of 435–480 nm and 300–345 nm.40,41 Both FI and HIX were computed based upon the corrected uorescence data.
Results The water quality parameters measured in situ are presented in Table 1. Note that relatively low DO concentrations (
