Chemosphere 111 (2014) 450–457
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Spectroscopic characterization of dissolved organic matter isolates from sediments and the association with phenanthrene binding affinity Jin Hur a,⇑, Bo-Mi Lee a, Kyung-Hoon Shin b a b
Department of Environment and Energy, Sejong University, Seoul 143-747, South Korea Department of Environmental Marine Sciences, Hanyang University, Keongki-do, Ansan 426-791, South Korea
h i g h l i g h t s Spectroscopic properties of different dissolved organic matter isolates from sediments are compared. Organic sources of sediments are well reflected in alkaline extractable organic matter isolates. Terrestrial humic-like fluorophores are well associated with the extent of phenanthrene binding. Phenanthrene binding affinity is better explained by the degree of biotransformation than by aromatic content.
a r t i c l e
i n f o
Article history: Received 22 January 2014 Received in revised form 21 March 2014 Accepted 8 April 2014
Handling Editor: I. Cousins Keywords: Sediment organic matter Alkaline extractable organic matter Source discrimination Organic carbon-normalized binding coefficient Hydrophobic organic contaminants (HOCs) Parallel factor analysis (PARAFAC)
a b s t r a c t In this study, selected spectroscopic characteristics of sediment organic matter (SOM) were compared and discussed with respect to their different isolation methods, the source discrimination capabilities, and the association with the extent of phenanthrene binding. A total of 16 sediments were collected from three categorized locations including a costal lake, industrial areas, and upper streams, each of which is likely influenced by the organic sources of algal production, industrial effluent, and terrestrial input, respectively. The spectroscopic properties related to aromatic structures and terrestrial humic acids were more pronounced for alkaline extractable organic matter (AEOM) isolates than for the SOM isolates based on water soluble extracts and pore water. The three categorized sampling locations were the most differentiated in the AEOM isolates, suggesting AEOM may be the most representative SOM isolates in terms of describing the chemical properties and the organic sources of SOM. Parallel factor analysis (PARAFAC) based on fluorescence excitation–emission matrix (EEM) showed that a combination of three fluorescent groups could represent all the fluorescence features of SOM. The three categorized sampling locations were well discriminated by the percent distributions of humic-like fluorescent groups of the AEOM isolates. The relative distribution of terrestrial humic-like fluorophores was well correlated with the extent of phenanthrene binding (r = 0.571; p < 0.05), suggesting that the presence of humic acids in SOM may contribute to the enhancement of binding with hydrophobic organic contaminants in sediments. Principal component analysis (PCA) further demonstrated that the extent of SOM’s binding affinity might be affected by the degree of biogeochemical transformation in SOM. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Sediments are considered the sources and the sinks for nutrients and pollutants in aquatic environments, affecting their fate and the distribution. Sediment organic matter (SOM), constituting an important portion of sediments, is a heterogeneous mixture of the classes of the organic substances encompassing carbohydrates, proteins, lignins, organic acids, and various other uncharacterized ⇑ Corresponding author. Tel.: +82 2 3408 3826; fax: +82 2 3408 4320. E-mail address: [email protected] (J. Hur). http://dx.doi.org/10.1016/j.chemosphere.2014.04.018 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
compounds such as humic substances (HS). Owing to its preservative nature, SOM may provide a wealth of information on the biogeochemical changes of organic matters occurring in the sites as well as on their formation and transport (Meyers and Teranes, 2001). Early diagenetic processes on suspended solids such as biodegradation and photo-oxidation make a great contribution to the changes in the quality and the composition of SOM (Jaffé et al., 2006; Osburn et al., 2012). For many lakes and estuaries under human impact, the quality of SOM may be altered by the relative contributions of natural and anthropogenic sources (Wang et al., 2012). Natural sources of SOM can be further divided
J. Hur et al. / Chemosphere 111 (2014) 450–457
into autochthonous and allochthonous origins. The former is derived from carbon excretion and the leachate of algae and aquatic macrophytes, and their microbial degradation products, while the latter is typically referred to the terrigenous inputs from upstream catchments. The fate, the transport, and the bioavailability of hydrophobic organic contaminants (HOCs) in sediments are largely affected by the quantity and the quality of SOM (Chin et al., 1997; Chefetz et al., 2000). In general, organic carbon normalized binding (or partitioning) coefficient (Koc) has been used as an indicator to determine the distributions of HOCs in sediments when the total organic carbon is known (Chin et al., 1997). Many previous studies have presented close relationships between the chemical and the structural nature of dissolved organic matters (DOM) and the Koc value of a specific HOC. For example, hydrophobic nature, aromatic content, and molecular weight of DOM have been suggested as the descriptors to estimate the Koc values (Kopinke et al., 2001; Hur and Schlautman, 2003; Hur et al., 2013). However, it should be noted that the generally accepted relationships may be broken when the subjects are extended into a class of highly heterogeneous organic matters including SOM (Wen et al., 2007; Hur et al., 2009). Recently, fluorescence characterization of SOM has been highlighted as a simple and easy tool for estimating the Koc values independent of the sediment sources (Hur and Kim, 2009). Fluorescence EEM–PARAFAC has been widely used for assessing the quality of DOM in various aquatic systems, and for tracking DOM sources (Stedmon et al., 2003; Fellman et al., 2010). Despite the numerous applications reported to date, little attention has been paid on SOM characterization (Santín et al., 2009; Wang et al., 2013). Recently, Osburn et al. (2012) successfully quantified several fluorescence EEM–PARAFAC components from alkaline extractable organic matter from sediments. They have demonstrated that PARAFAC modeling of SOM in adjunct with DOM possibly distinguished among different organic sources, providing insight into the dynamics of the nutrient supply of the sediments to the overlying water column. Altogether, it can easily be inferred that the chemical and the structural information on SOM derived from fluorescence EEM– PARAFAC could also be used to estimate the extent of HOC binding. To our knowledge, however, there has been no study to explore the potential use of the EEM–PARAFAC for the prediction of the extent of HOC binding. In this study, we collected 16 sediment samples from a costal lake and the surrounding areas, which were further
451
categorized into three different locations including a coastal lake, streams at industrial areas, and upper streams at rural areas. The three categorized sampling locations can be characterized by their major organic sources of in situ primary production of algae, human and industrial activities, and terrestrial input from the catchments covered by forest and agricultural areas, respectively. This study aimed (1) to compare selected spectroscopic properties of SOM and the extent of HOC binding with respect to different SOM isolation methods and three categorized sampling locations, and (2) to suggest the descriptors for the extent of HOC binding primarily based on fluorescence EEM–PARAFAC. 2. Experimental 2.1. Study area and sample collection The study area is an artificial coastal lake, the Shiwha lake (N37°170 , E126°150 ), and the upstream catchments, which are located on the west coast of Korea (Fig. 1). The northern part of the upstream catchments is occupied by an industrial complex, while the upstream main channel or the eastern part of the upstream catchments of the lake is mostly covered by rural and forest areas (Fig. 1). The artificial lake was initially developed in 1994 by the construction of a dike in the boundary between the coast and the open sea. A tidal power plant was constructed in the middle of the dike afterwards so that seawater circulated into the lake. Bottom sediments were collected using a grab sampler (Ekman dredge) in June, 2013, from 16 different sites, which were further categorized into three sampling locations – lake (L1–L7), streams at industrial areas (I1–I5), and streams at rural areas (S1–S4). The samples were kept in vacuum sealed plastic bags and stored in a closed ice box while they were transported from the field to the laboratory for analysis. 2.2. Extraction of SOM from sediments Different extraction procedures were made to obtain three types of SOM isolates from the collected sediments, which include pore water DOM (PDOM), water soluble organic matter (WSOM), and alkaline extractable organic matter (AEOM). WSOM refers to an organic fraction of SOM pool that is easily mobilized into the underlying water column. AEOM includes a majority of the HS
Fig. 1. Map of the lake Shiwha and the surrounding areas. The sediment sampling locations are indicated. The dark grey and the light grey colored letters indicate the geographical names and land uses of the upper catchments, respectively. The letters ‘‘W’’ and ‘‘P’’ indicate the locations of water gate and tidal power plant, respectively. The map is adopted from Phong et al. (2014).
452
J. Hur et al. / Chemosphere 111 (2014) 450–457
fraction contained in SOM, which is rarely isolated by using a mild solution. The three isolates represent different fractions of the total SOM pool. For example, a higher content of aliphatic carbon and a lower portion of oxygen-containing structures are typically observed for the organic matter isolates based on more alkaline solution (Dai et al., 2006). For this study, pore water was extracted by centrifugation of the sediments at 5000 rpm for an hour. The supernatant (or pore water) was carefully collected under an atmosphere of nitrogen and stored in tightly sealed tubes. The remaining sediments were all air-dried, ground, and passed through a 0.18 mm sieve. The WSOM and the AEOM isolates were prepared by mixing the dried sediments at 1:10 of a solid to solution mass ratio with Milli-Q water and 0.1 N NaOH solution, respectively (Hur and Kim, 2009). The alkaline solution was purged with nitrogen before the mixing was conducted overnight at 120 rpm on a reciprocating shaker at room temperature. Particulate matters were removed from the samples by centrifugation at 10 000 rpm for 30 min followed by filtration using a 0.2 lm poresized membrane (cellulose acetate, Advantec). 2.3. Characterization of SOM Dissolved organic carbon (DOC) concentrations of the SOM isolates were quantified by using a Shimadzu V-CPH analyzer. The DOC concentrations ranged from 9.9 to 380 mgC L1. A UV–visible spectrometer (HACH, DR5000) was used with 1 cm-cuvette to determine specific ultraviolet absorbance (SUVA) values, which were calculated by 100-fold ratios of the UV absorbance at 254 nm (UV254) to the corresponding DOC concentration. Fluorescence EEM was measured using a luminescence spectrometer (LS-55, Perkin–Elmer). The emission scan was made from the wavelengths of 280–550 nm with 0.5 nm increments at a stepwise increase of 5 nm for the excitation wavelengths from 250 nm to 500 nm. The scanning speed was set at 1200 nm min1, and the excitation and emission slits were adjusted to 10 nm and 5 nm, respectively. To limit second order Rayleigh scattering, a 290 nm cutoff filter was used for the measurements. All the SOM isolate samples were diluted with Milli-Q water until their UV254 values were below 0.05 cm1, which eliminates the necessity of the inner-filter correction (Hur et al., 2008). The fluorescence responses to the blank (i.e., Milli-Q water) were subtracted from the original EEM data. The obtained fluorescence intensities were then normalized to units of quinine sulfate equivalents (QSE) in lg L1 using the fluorescence of a quinine sulfate dehydrate at an excitation/emission of 350/450 nm. The calculation of humification index (HIX) followed a procedure suggested by Zsolnay et al. (1999), which is based on a ratio of the emission scanning areas at the wavelength range of 300–345 nm to 435–480 nm with an excitation wavelength of 254 nm. Fluorescence index (FI) was estimated using a ratio of the emission intensities at 450–500 nm at a fixed excitation wavelength of 370 nm (McKnight et al., 2001). The pH of all the samples was adjusted to 3.0 prior to the measurements of the optical properties described above. The meanings of SUVA, HIX, and FI values are well explained in the cited references. For this study, the relative order of the optical properties were more considered than the absolute values in interpreting the DOM sources because the different extraction methods involved preferential isolation of a certain SOM fraction and/or potential spectroscopic changes occurring in the different extraction solutions. PARAFAC modeling was applied using MATLAB 7.0 (Mathworks, Natick, MA, USA) with DOM Fluor toolbox (http://www.models. life.ku.dk) to statistically decompose dissimilar fluorescence components from all the EEM data, in which several different fluorophore groups may be overlapped with each other (Stedmon and Bro, 2008). A total of 48 EEMs (16 3) of the SOM samples were incorporated into the PARAFAC modeling process.
2.4. Estimation of phenanthrene Koc values Phenanthrene (Aldrich, 99%) was used as a model HOC for this study. A modified fluorescence quenching technique was adopted to determine the Koc values of the AEOM isolates. This technique has already been proven for reliability, and it has been used in our prior studies examining the variations in the Koc values of different DOM samples (Hur and Schlautman, 2003; Hur and Kim, 2009). The details on this technique are well described elsewhere (Hur and Schlautman, 2003). Briefly, a minimum volume (less than 0.1%) of phenanthrene in methanol was spiked into each AEOM isolate solution that was diluted to a constant DOC concentration (15 mgC L1). The total added phenanthrene concentration was kept to be 0.1 mg L1. The samples were equilibrated in amber vials on a shaker overnight. After the equilibrium, the fluorescence intensities were measured at the excitation/emission wavelengths of 293/365 nm. The slit was set to 2.5 nm for excitation, and 20 nm for emission. Inner-filter correction was made to account for the potential light absorption of the AEOM (Gauthier et al., 1986). From an external standard curve in a blank solution (i.e., Milli-Q water) based the concentrations of a series of diluted phenanthrene versus the corresponding fluorescence intensities, the fraction of the freely dissolved phenanthrene and the AEOM-bound fraction were separately quantified from the total amount initially added in the SOM isolate solutions. The following equation was used to estimate Koc values.
K oc ¼
½Phe AEOM ½Phefree ½AEOM
ð1Þ
where [Phe AEOM] is the AEOM-bound phenanthrene concentration (lg L1), AEOM [Phe]free is the freely dissolved phenanthrene concentration (lg L1), and [AEOM] is the dissolved AEOM concentration (i.e., 15 mgC L1). 2.5. Statistical analyses One-way ANOVA was used to compare the means with respect to the three different SOM isolate types and the three categorized sampling locations. Correlation analyses and multiple regression analyses were performed using selected spectroscopic properties of the AEOM isolates and their Koc values. Principal component analysis (PCA) was carried out in order to better present the relationships between the selected AEOM characteristics and their spatial variations. All the statistics were done using an add-in program of XLSTAT (Addinsoft, New York, USA). 3. Results and discussions 3.1. Comparison of SUVA values for different SOM isolates and for sampling locations SUVA value has been widely used as an index to estimate the aromatic content of DOM (Weishaar et al., 2003). For this study, the SUVA values of the SOM isolates were compared with respect to the different extraction procedures and the categorized sampling locations (Table 1). The SUVA values exhibited a large variation from 0.26 to 1.24 L mgC1 m1, from 0.64 to 2.60 L mgC1 m1, and from 1.01 to 2.77 L mgC1 m1 for the PDOM, the WSOM, and the AEOM isolates, respectively (Table 1). For the lake and the upper stream sediments, PDOM was distinguishable from the other two SOM isolate types (i.e., WSOM and AEOM) by low SUVA values (ANOVA p < 0.05). However, the PDOM isolates from the sediments at the industrial areas did not exhibit any significant differences in the SUVA values in comparison with the other two SOM isolate types (Table 1).
J. Hur et al. / Chemosphere 111 (2014) 450–457 Table 1 Comparison of selected spectroscopic characteristics of sediment organic matter isolates for three different extraction methods (PDOM, WSOM, and AEOM) and the categorized sampling locations (lake, industrial, streams).
SUVA Lake Industrial Streams p-Valuea HIX Lake Industrial Streams p-Valuea FI Lake Industrial Streams p-Valuea a b
PDOM
WSOM
AEOM
p-Valueb
0.55 ± 0.19 (0.26–0.82) 0.72 ± 0.31 (0.34–1.04) 0.79 ± 0.31 (0.60–1.24) 0.339a
1.20 ± 0.15 (1.05–1.46) 1.67 ± 0.86 (0.64–2.60) 1.50 ± 0.39 (1.17–2.03) 0.323a
1.46 ± 0.24 (1.14–1.90) 1.36 ± 0.33 (1.01–1.80) 2.07 ± 0.63 (1.29–2.77) 0.038a
WSOM > PDOM irrespective of the sampling locations. One exception was the SOM isolates from the industrial areas, in which the opposite trend was observed between the WSOM and the AEOM isolates. Considering the difference in the extraction intensity between the two SOM isolation procedures (i.e., WSOM and AEOM), the exceptional result suggests that aromatic moieties of SOM may be loosely bound to the surfaces of the sediments at the industrial areas. 3.2. Comparison of HIX values for different SOM isolates and for sampling locations HIX value can be used as an indicator for the degree of the condensation of aromatic structures in SOM and/or the degree of the conjugation in the unsaturated aliphatic chains (Fuentes et al., 2007). In this study, the average HIX values of PDOM were similar for the three sampling locations with their ranges overlapped with each other (ANOVA p = 0.906) (Table 1). In contrast, the WSOM and the AEOM isolates from the industrial areas were distinguishable from those of the other two categorized locations by comparing their HIX values. The lowest value ranges were observed for the two SOM isolates from the industrial areas among the three categorized locations (ANOVA p < 0.05). This suggests that SOM bound to the sediments may be characterized by relatively low degree of condensed polyaromatic structures, presumably due to the non-aromatic and biodegradable nature of the organic sources exported from the industrial effluents. Similar to the SUVA values, the average HIX values became higher with the intensity of the
FI values are typically used to distinguish between autochthonous and allochthonous origins of DOM, and they are negatively correlated with aromatic content (McKnight et al., 2001). For this study, FI values varied from 1.36 to 2.24, and they discriminated well between the lake and the industrial sediments, and between the upper streams and the industrial areas for the WSOM and the AEOM isolates (ANOVA p < 0.05; Table 1). For example, the lowest range of the FI values was observed for the WSOM isolates from the upper streams, followed by those of the industrial areas. The lake sediments showed the middle range of the FI values between the two other sampling locations. For the AEOM isolates, however, the average FI values tend to decrease in the order of the industrial areas > the upper streams > the lake. The inconsistency in the FI value ranges of the different sampling locations between the two SOM isolation methods may be attributed to the differences in the structural and the chemical composition between the organic fractions loosely and strongly bound to the lake sediments. For example, strongly bound SOM fractions (i.e., AEOM isolates) in the lake may be more associated with lignin-derived origins (i.e., allochthonous), whereas the SOM fractions loosely bound to sediments (i.e., WSOM isolates) are likely to reflect the organic sources of in situ algal production or industrial effluents. This speculation is supported in part by our recent report of Hur et al. (2013), in which hydrophobic fraction was more enriched in leaf litter organic matter than in algal derived organic matter and wastewater effluent. 3.4. Distributions of PARAFAC components in SOM Three PARAFAC components were decomposed from the EEM data set of all the SOM isolates. The number and the spectral shapes of the components were validated by core consistency test and split half analyses (Fig. 2). The core consistency was 90% for three-component model for this study. Component 1 (C1) exhibited two maxima at the excitation wavelengths of 250 nm and 335 nm, and at the emission wavelength of 450 nm. The spectral feature of C1 was similar to those of typical aquatic DOM with terrestrial sources (Holbrook et al., 2006; Santín et al., 2009; Borisover et al., 2012). The primary and the secondary peaks of the component 2 (C2) were blue shifted compared to C1, appearing at 250 (315) nm/400 nm for the excitation/emission wavelengths. A similar PARAFAC component was previously assigned to marine humic-like (Stedmon and Markager, 2005) or microbial humic-like fluorescence (Yamashita and Jaffé, 2008; Shank et al., 2011; Zhang et al., 2011). Component 3 matches well with a typical protein-like fluorescence, displaying the fluorescence peak at the excitation/ emission wavelengths of 280/350 nm (Stedmon and Markager, 2005; Yamashita and Jaffé, 2008). The relative distributions of the PARAFAC components were compared for different SOM extraction methods and for the three categorized sampling locations (Table 2; Fig. 3). Among the three SOM types, PDOM presented the lowest percent distributions in terrestrial humic-like components (i.e., C1) for all the three
454
J. Hur et al. / Chemosphere 111 (2014) 450–457
Fig. 2. Contour plots of the three PARAFAC components (a–c) and the corresponding split-half validation (b, e, and f).
Table 2 Comparison of the percent distributions of PARAFAC components of sediment organic matter isolates for three different extraction methods (PDOM, WSOM, and AEOM) and the three categorized sampling locations (lake, industrial, streams).
%C1 Lake Industrial Streams p-Valuea %C2 Lake Industrial Streams p-Valuea %C3 Lake Industrial Streams p-Valuea a b
PDOM
WSOM
AEOM
p-Valueb
29.6 ± 4.3 (22.4–35.5) 24.6 ± 8.6 (13.4–34.2) 18.5 ± 11.8 (7.8–34.2)
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Spectroscopic characterization of dissolved organic matter isolates from sediments and the association with phenanthrene binding affinity Jin Hur a,⇑, Bo-Mi Lee a, Kyung-Hoon Shin b a b
Department of Environment and Energy, Sejong University, Seoul 143-747, South Korea Department of Environmental Marine Sciences, Hanyang University, Keongki-do, Ansan 426-791, South Korea
h i g h l i g h t s Spectroscopic properties of different dissolved organic matter isolates from sediments are compared. Organic sources of sediments are well reflected in alkaline extractable organic matter isolates. Terrestrial humic-like fluorophores are well associated with the extent of phenanthrene binding. Phenanthrene binding affinity is better explained by the degree of biotransformation than by aromatic content.
a r t i c l e
i n f o
Article history: Received 22 January 2014 Received in revised form 21 March 2014 Accepted 8 April 2014
Handling Editor: I. Cousins Keywords: Sediment organic matter Alkaline extractable organic matter Source discrimination Organic carbon-normalized binding coefficient Hydrophobic organic contaminants (HOCs) Parallel factor analysis (PARAFAC)
a b s t r a c t In this study, selected spectroscopic characteristics of sediment organic matter (SOM) were compared and discussed with respect to their different isolation methods, the source discrimination capabilities, and the association with the extent of phenanthrene binding. A total of 16 sediments were collected from three categorized locations including a costal lake, industrial areas, and upper streams, each of which is likely influenced by the organic sources of algal production, industrial effluent, and terrestrial input, respectively. The spectroscopic properties related to aromatic structures and terrestrial humic acids were more pronounced for alkaline extractable organic matter (AEOM) isolates than for the SOM isolates based on water soluble extracts and pore water. The three categorized sampling locations were the most differentiated in the AEOM isolates, suggesting AEOM may be the most representative SOM isolates in terms of describing the chemical properties and the organic sources of SOM. Parallel factor analysis (PARAFAC) based on fluorescence excitation–emission matrix (EEM) showed that a combination of three fluorescent groups could represent all the fluorescence features of SOM. The three categorized sampling locations were well discriminated by the percent distributions of humic-like fluorescent groups of the AEOM isolates. The relative distribution of terrestrial humic-like fluorophores was well correlated with the extent of phenanthrene binding (r = 0.571; p < 0.05), suggesting that the presence of humic acids in SOM may contribute to the enhancement of binding with hydrophobic organic contaminants in sediments. Principal component analysis (PCA) further demonstrated that the extent of SOM’s binding affinity might be affected by the degree of biogeochemical transformation in SOM. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Sediments are considered the sources and the sinks for nutrients and pollutants in aquatic environments, affecting their fate and the distribution. Sediment organic matter (SOM), constituting an important portion of sediments, is a heterogeneous mixture of the classes of the organic substances encompassing carbohydrates, proteins, lignins, organic acids, and various other uncharacterized ⇑ Corresponding author. Tel.: +82 2 3408 3826; fax: +82 2 3408 4320. E-mail address: [email protected] (J. Hur). http://dx.doi.org/10.1016/j.chemosphere.2014.04.018 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
compounds such as humic substances (HS). Owing to its preservative nature, SOM may provide a wealth of information on the biogeochemical changes of organic matters occurring in the sites as well as on their formation and transport (Meyers and Teranes, 2001). Early diagenetic processes on suspended solids such as biodegradation and photo-oxidation make a great contribution to the changes in the quality and the composition of SOM (Jaffé et al., 2006; Osburn et al., 2012). For many lakes and estuaries under human impact, the quality of SOM may be altered by the relative contributions of natural and anthropogenic sources (Wang et al., 2012). Natural sources of SOM can be further divided
J. Hur et al. / Chemosphere 111 (2014) 450–457
into autochthonous and allochthonous origins. The former is derived from carbon excretion and the leachate of algae and aquatic macrophytes, and their microbial degradation products, while the latter is typically referred to the terrigenous inputs from upstream catchments. The fate, the transport, and the bioavailability of hydrophobic organic contaminants (HOCs) in sediments are largely affected by the quantity and the quality of SOM (Chin et al., 1997; Chefetz et al., 2000). In general, organic carbon normalized binding (or partitioning) coefficient (Koc) has been used as an indicator to determine the distributions of HOCs in sediments when the total organic carbon is known (Chin et al., 1997). Many previous studies have presented close relationships between the chemical and the structural nature of dissolved organic matters (DOM) and the Koc value of a specific HOC. For example, hydrophobic nature, aromatic content, and molecular weight of DOM have been suggested as the descriptors to estimate the Koc values (Kopinke et al., 2001; Hur and Schlautman, 2003; Hur et al., 2013). However, it should be noted that the generally accepted relationships may be broken when the subjects are extended into a class of highly heterogeneous organic matters including SOM (Wen et al., 2007; Hur et al., 2009). Recently, fluorescence characterization of SOM has been highlighted as a simple and easy tool for estimating the Koc values independent of the sediment sources (Hur and Kim, 2009). Fluorescence EEM–PARAFAC has been widely used for assessing the quality of DOM in various aquatic systems, and for tracking DOM sources (Stedmon et al., 2003; Fellman et al., 2010). Despite the numerous applications reported to date, little attention has been paid on SOM characterization (Santín et al., 2009; Wang et al., 2013). Recently, Osburn et al. (2012) successfully quantified several fluorescence EEM–PARAFAC components from alkaline extractable organic matter from sediments. They have demonstrated that PARAFAC modeling of SOM in adjunct with DOM possibly distinguished among different organic sources, providing insight into the dynamics of the nutrient supply of the sediments to the overlying water column. Altogether, it can easily be inferred that the chemical and the structural information on SOM derived from fluorescence EEM– PARAFAC could also be used to estimate the extent of HOC binding. To our knowledge, however, there has been no study to explore the potential use of the EEM–PARAFAC for the prediction of the extent of HOC binding. In this study, we collected 16 sediment samples from a costal lake and the surrounding areas, which were further
451
categorized into three different locations including a coastal lake, streams at industrial areas, and upper streams at rural areas. The three categorized sampling locations can be characterized by their major organic sources of in situ primary production of algae, human and industrial activities, and terrestrial input from the catchments covered by forest and agricultural areas, respectively. This study aimed (1) to compare selected spectroscopic properties of SOM and the extent of HOC binding with respect to different SOM isolation methods and three categorized sampling locations, and (2) to suggest the descriptors for the extent of HOC binding primarily based on fluorescence EEM–PARAFAC. 2. Experimental 2.1. Study area and sample collection The study area is an artificial coastal lake, the Shiwha lake (N37°170 , E126°150 ), and the upstream catchments, which are located on the west coast of Korea (Fig. 1). The northern part of the upstream catchments is occupied by an industrial complex, while the upstream main channel or the eastern part of the upstream catchments of the lake is mostly covered by rural and forest areas (Fig. 1). The artificial lake was initially developed in 1994 by the construction of a dike in the boundary between the coast and the open sea. A tidal power plant was constructed in the middle of the dike afterwards so that seawater circulated into the lake. Bottom sediments were collected using a grab sampler (Ekman dredge) in June, 2013, from 16 different sites, which were further categorized into three sampling locations – lake (L1–L7), streams at industrial areas (I1–I5), and streams at rural areas (S1–S4). The samples were kept in vacuum sealed plastic bags and stored in a closed ice box while they were transported from the field to the laboratory for analysis. 2.2. Extraction of SOM from sediments Different extraction procedures were made to obtain three types of SOM isolates from the collected sediments, which include pore water DOM (PDOM), water soluble organic matter (WSOM), and alkaline extractable organic matter (AEOM). WSOM refers to an organic fraction of SOM pool that is easily mobilized into the underlying water column. AEOM includes a majority of the HS
Fig. 1. Map of the lake Shiwha and the surrounding areas. The sediment sampling locations are indicated. The dark grey and the light grey colored letters indicate the geographical names and land uses of the upper catchments, respectively. The letters ‘‘W’’ and ‘‘P’’ indicate the locations of water gate and tidal power plant, respectively. The map is adopted from Phong et al. (2014).
452
J. Hur et al. / Chemosphere 111 (2014) 450–457
fraction contained in SOM, which is rarely isolated by using a mild solution. The three isolates represent different fractions of the total SOM pool. For example, a higher content of aliphatic carbon and a lower portion of oxygen-containing structures are typically observed for the organic matter isolates based on more alkaline solution (Dai et al., 2006). For this study, pore water was extracted by centrifugation of the sediments at 5000 rpm for an hour. The supernatant (or pore water) was carefully collected under an atmosphere of nitrogen and stored in tightly sealed tubes. The remaining sediments were all air-dried, ground, and passed through a 0.18 mm sieve. The WSOM and the AEOM isolates were prepared by mixing the dried sediments at 1:10 of a solid to solution mass ratio with Milli-Q water and 0.1 N NaOH solution, respectively (Hur and Kim, 2009). The alkaline solution was purged with nitrogen before the mixing was conducted overnight at 120 rpm on a reciprocating shaker at room temperature. Particulate matters were removed from the samples by centrifugation at 10 000 rpm for 30 min followed by filtration using a 0.2 lm poresized membrane (cellulose acetate, Advantec). 2.3. Characterization of SOM Dissolved organic carbon (DOC) concentrations of the SOM isolates were quantified by using a Shimadzu V-CPH analyzer. The DOC concentrations ranged from 9.9 to 380 mgC L1. A UV–visible spectrometer (HACH, DR5000) was used with 1 cm-cuvette to determine specific ultraviolet absorbance (SUVA) values, which were calculated by 100-fold ratios of the UV absorbance at 254 nm (UV254) to the corresponding DOC concentration. Fluorescence EEM was measured using a luminescence spectrometer (LS-55, Perkin–Elmer). The emission scan was made from the wavelengths of 280–550 nm with 0.5 nm increments at a stepwise increase of 5 nm for the excitation wavelengths from 250 nm to 500 nm. The scanning speed was set at 1200 nm min1, and the excitation and emission slits were adjusted to 10 nm and 5 nm, respectively. To limit second order Rayleigh scattering, a 290 nm cutoff filter was used for the measurements. All the SOM isolate samples were diluted with Milli-Q water until their UV254 values were below 0.05 cm1, which eliminates the necessity of the inner-filter correction (Hur et al., 2008). The fluorescence responses to the blank (i.e., Milli-Q water) were subtracted from the original EEM data. The obtained fluorescence intensities were then normalized to units of quinine sulfate equivalents (QSE) in lg L1 using the fluorescence of a quinine sulfate dehydrate at an excitation/emission of 350/450 nm. The calculation of humification index (HIX) followed a procedure suggested by Zsolnay et al. (1999), which is based on a ratio of the emission scanning areas at the wavelength range of 300–345 nm to 435–480 nm with an excitation wavelength of 254 nm. Fluorescence index (FI) was estimated using a ratio of the emission intensities at 450–500 nm at a fixed excitation wavelength of 370 nm (McKnight et al., 2001). The pH of all the samples was adjusted to 3.0 prior to the measurements of the optical properties described above. The meanings of SUVA, HIX, and FI values are well explained in the cited references. For this study, the relative order of the optical properties were more considered than the absolute values in interpreting the DOM sources because the different extraction methods involved preferential isolation of a certain SOM fraction and/or potential spectroscopic changes occurring in the different extraction solutions. PARAFAC modeling was applied using MATLAB 7.0 (Mathworks, Natick, MA, USA) with DOM Fluor toolbox (http://www.models. life.ku.dk) to statistically decompose dissimilar fluorescence components from all the EEM data, in which several different fluorophore groups may be overlapped with each other (Stedmon and Bro, 2008). A total of 48 EEMs (16 3) of the SOM samples were incorporated into the PARAFAC modeling process.
2.4. Estimation of phenanthrene Koc values Phenanthrene (Aldrich, 99%) was used as a model HOC for this study. A modified fluorescence quenching technique was adopted to determine the Koc values of the AEOM isolates. This technique has already been proven for reliability, and it has been used in our prior studies examining the variations in the Koc values of different DOM samples (Hur and Schlautman, 2003; Hur and Kim, 2009). The details on this technique are well described elsewhere (Hur and Schlautman, 2003). Briefly, a minimum volume (less than 0.1%) of phenanthrene in methanol was spiked into each AEOM isolate solution that was diluted to a constant DOC concentration (15 mgC L1). The total added phenanthrene concentration was kept to be 0.1 mg L1. The samples were equilibrated in amber vials on a shaker overnight. After the equilibrium, the fluorescence intensities were measured at the excitation/emission wavelengths of 293/365 nm. The slit was set to 2.5 nm for excitation, and 20 nm for emission. Inner-filter correction was made to account for the potential light absorption of the AEOM (Gauthier et al., 1986). From an external standard curve in a blank solution (i.e., Milli-Q water) based the concentrations of a series of diluted phenanthrene versus the corresponding fluorescence intensities, the fraction of the freely dissolved phenanthrene and the AEOM-bound fraction were separately quantified from the total amount initially added in the SOM isolate solutions. The following equation was used to estimate Koc values.
K oc ¼
½Phe AEOM ½Phefree ½AEOM
ð1Þ
where [Phe AEOM] is the AEOM-bound phenanthrene concentration (lg L1), AEOM [Phe]free is the freely dissolved phenanthrene concentration (lg L1), and [AEOM] is the dissolved AEOM concentration (i.e., 15 mgC L1). 2.5. Statistical analyses One-way ANOVA was used to compare the means with respect to the three different SOM isolate types and the three categorized sampling locations. Correlation analyses and multiple regression analyses were performed using selected spectroscopic properties of the AEOM isolates and their Koc values. Principal component analysis (PCA) was carried out in order to better present the relationships between the selected AEOM characteristics and their spatial variations. All the statistics were done using an add-in program of XLSTAT (Addinsoft, New York, USA). 3. Results and discussions 3.1. Comparison of SUVA values for different SOM isolates and for sampling locations SUVA value has been widely used as an index to estimate the aromatic content of DOM (Weishaar et al., 2003). For this study, the SUVA values of the SOM isolates were compared with respect to the different extraction procedures and the categorized sampling locations (Table 1). The SUVA values exhibited a large variation from 0.26 to 1.24 L mgC1 m1, from 0.64 to 2.60 L mgC1 m1, and from 1.01 to 2.77 L mgC1 m1 for the PDOM, the WSOM, and the AEOM isolates, respectively (Table 1). For the lake and the upper stream sediments, PDOM was distinguishable from the other two SOM isolate types (i.e., WSOM and AEOM) by low SUVA values (ANOVA p < 0.05). However, the PDOM isolates from the sediments at the industrial areas did not exhibit any significant differences in the SUVA values in comparison with the other two SOM isolate types (Table 1).
J. Hur et al. / Chemosphere 111 (2014) 450–457 Table 1 Comparison of selected spectroscopic characteristics of sediment organic matter isolates for three different extraction methods (PDOM, WSOM, and AEOM) and the categorized sampling locations (lake, industrial, streams).
SUVA Lake Industrial Streams p-Valuea HIX Lake Industrial Streams p-Valuea FI Lake Industrial Streams p-Valuea a b
PDOM
WSOM
AEOM
p-Valueb
0.55 ± 0.19 (0.26–0.82) 0.72 ± 0.31 (0.34–1.04) 0.79 ± 0.31 (0.60–1.24) 0.339a
1.20 ± 0.15 (1.05–1.46) 1.67 ± 0.86 (0.64–2.60) 1.50 ± 0.39 (1.17–2.03) 0.323a
1.46 ± 0.24 (1.14–1.90) 1.36 ± 0.33 (1.01–1.80) 2.07 ± 0.63 (1.29–2.77) 0.038a
WSOM > PDOM irrespective of the sampling locations. One exception was the SOM isolates from the industrial areas, in which the opposite trend was observed between the WSOM and the AEOM isolates. Considering the difference in the extraction intensity between the two SOM isolation procedures (i.e., WSOM and AEOM), the exceptional result suggests that aromatic moieties of SOM may be loosely bound to the surfaces of the sediments at the industrial areas. 3.2. Comparison of HIX values for different SOM isolates and for sampling locations HIX value can be used as an indicator for the degree of the condensation of aromatic structures in SOM and/or the degree of the conjugation in the unsaturated aliphatic chains (Fuentes et al., 2007). In this study, the average HIX values of PDOM were similar for the three sampling locations with their ranges overlapped with each other (ANOVA p = 0.906) (Table 1). In contrast, the WSOM and the AEOM isolates from the industrial areas were distinguishable from those of the other two categorized locations by comparing their HIX values. The lowest value ranges were observed for the two SOM isolates from the industrial areas among the three categorized locations (ANOVA p < 0.05). This suggests that SOM bound to the sediments may be characterized by relatively low degree of condensed polyaromatic structures, presumably due to the non-aromatic and biodegradable nature of the organic sources exported from the industrial effluents. Similar to the SUVA values, the average HIX values became higher with the intensity of the
FI values are typically used to distinguish between autochthonous and allochthonous origins of DOM, and they are negatively correlated with aromatic content (McKnight et al., 2001). For this study, FI values varied from 1.36 to 2.24, and they discriminated well between the lake and the industrial sediments, and between the upper streams and the industrial areas for the WSOM and the AEOM isolates (ANOVA p < 0.05; Table 1). For example, the lowest range of the FI values was observed for the WSOM isolates from the upper streams, followed by those of the industrial areas. The lake sediments showed the middle range of the FI values between the two other sampling locations. For the AEOM isolates, however, the average FI values tend to decrease in the order of the industrial areas > the upper streams > the lake. The inconsistency in the FI value ranges of the different sampling locations between the two SOM isolation methods may be attributed to the differences in the structural and the chemical composition between the organic fractions loosely and strongly bound to the lake sediments. For example, strongly bound SOM fractions (i.e., AEOM isolates) in the lake may be more associated with lignin-derived origins (i.e., allochthonous), whereas the SOM fractions loosely bound to sediments (i.e., WSOM isolates) are likely to reflect the organic sources of in situ algal production or industrial effluents. This speculation is supported in part by our recent report of Hur et al. (2013), in which hydrophobic fraction was more enriched in leaf litter organic matter than in algal derived organic matter and wastewater effluent. 3.4. Distributions of PARAFAC components in SOM Three PARAFAC components were decomposed from the EEM data set of all the SOM isolates. The number and the spectral shapes of the components were validated by core consistency test and split half analyses (Fig. 2). The core consistency was 90% for three-component model for this study. Component 1 (C1) exhibited two maxima at the excitation wavelengths of 250 nm and 335 nm, and at the emission wavelength of 450 nm. The spectral feature of C1 was similar to those of typical aquatic DOM with terrestrial sources (Holbrook et al., 2006; Santín et al., 2009; Borisover et al., 2012). The primary and the secondary peaks of the component 2 (C2) were blue shifted compared to C1, appearing at 250 (315) nm/400 nm for the excitation/emission wavelengths. A similar PARAFAC component was previously assigned to marine humic-like (Stedmon and Markager, 2005) or microbial humic-like fluorescence (Yamashita and Jaffé, 2008; Shank et al., 2011; Zhang et al., 2011). Component 3 matches well with a typical protein-like fluorescence, displaying the fluorescence peak at the excitation/ emission wavelengths of 280/350 nm (Stedmon and Markager, 2005; Yamashita and Jaffé, 2008). The relative distributions of the PARAFAC components were compared for different SOM extraction methods and for the three categorized sampling locations (Table 2; Fig. 3). Among the three SOM types, PDOM presented the lowest percent distributions in terrestrial humic-like components (i.e., C1) for all the three
454
J. Hur et al. / Chemosphere 111 (2014) 450–457
Fig. 2. Contour plots of the three PARAFAC components (a–c) and the corresponding split-half validation (b, e, and f).
Table 2 Comparison of the percent distributions of PARAFAC components of sediment organic matter isolates for three different extraction methods (PDOM, WSOM, and AEOM) and the three categorized sampling locations (lake, industrial, streams).
%C1 Lake Industrial Streams p-Valuea %C2 Lake Industrial Streams p-Valuea %C3 Lake Industrial Streams p-Valuea a b
PDOM
WSOM
AEOM
p-Valueb
29.6 ± 4.3 (22.4–35.5) 24.6 ± 8.6 (13.4–34.2) 18.5 ± 11.8 (7.8–34.2)
