Using Two-Dimensional Correlations of 13C NMR and FTIR To

Oct 13, 2010 - Jamie Warren , Susan E. Ziegler ... Xiaoyan Cao , George R. Aiken , Robert G.M. Spencer , Kenna Butler , Jingdong Mao , Klaus Schmidt-R...
0 downloads 0 Views 1MB Size
Download PDF
Environ. Sci. Technol. 2010, 44, 8044–8049

Using Two-Dimensional Correlations of 13C NMR and FTIR To Investigate Changes in the Chemical Composition of Dissolved Organic Matter along an Estuarine Transect H U S S A I N A . N . A B D U L L A , †,§ ELIZABETH C. MINOR,‡ AND P A T R I C K G . H A T C H E R * ,§ Department of Ocean, Earth, and Atmospheric Sciences, Old Dominion University, Norfolk, Virginia 23529, Large Lakes Observatory and Department of Chemistry & Biochemistry, University of Minnesota, Duluth, 109 RLB, 2205 East Fifth Street, Duluth, Minnesota 55812, and Department of Chemistry & Biochemistry, Old Dominion University, Norfolk, Virginia 23529

Received March 19, 2010. Revised manuscript received August 23, 2010. Accepted September 13, 2010.

Applying two-dimensional correlation spectroscopy to 13C NMR and FTIR spectra of the high molecular-weight dissolved organic matter (HMW-DOM) isolated along an Elizabeth River/ Chesapeake Bay salinity transect shows that HMW-DOM consists of three major components that have different biogeochemical reactivities. The first appears to be a heteropolysaccharide (HPS) component and its contribution to carbon increases as we approach the marine offshore. The second appears to be composed of carboxyl-rich compounds (CRC); its carbon percentage decreases. The third component contains the major functional group of amide/amino sugar (AMS) and its carbon percentage stays almost constant along the salinity transect. It seems that the HPS and CRC are present in many aquatic environments at different relative ratios. The 2D-correlation maps reveal that each of these components is composed of dynamic mixtures of compounds that share similar backbone structures but have significant functional group differences. Twodimensional (2D) correlation spectroscopy is a powerful new biogeochemical tool to track the changes in complex organic matter as a function of space, time, or environmental effects.

Introduction Changes in the chemical structural composition of dissolved organic matter (DOM) along estuaries and continental shelf regions of the world influences its ecological and environmental reactivity, e.g. bioavailability, transport of hydrophobic organic contaminants (1-3), and redistribution of trace metals (4). Previous studies attempting to address changes in chemical structures of DOM along estuaries and coastal regions have used optical parameters, e.g., photo* Corresponding author phone: (757)683-6537; fax: (757)683-4807; e-mail: [email protected]. † Department of Ocean, Earth, and Atmospheric Sciences, Old Dominion University. § Department of Chemistry & Biochemistry, Old Dominion University. ‡ University of Minnesota. 8044

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 21, 2010

bleaching of colored dissolved organic matter (CDOM) (5, 6), or mass spectroscopy applied to isolated fractions (7-9). While these approaches have been qualitatively informative, they have been unable to provide quantitative insight into compound-class variations within DOM as a function of spatial and seasonal changes. Both 13C NMR and FTIR can provide a more quantitative evaluation of changes in compound class distributions within DOM. However, the 13C NMR and FTIR spectra are convoluted by the complexity of DOM; the presence of many different compounds containing similar functional groups creates overlap among the different bands and reduces the spectral resolution. DOM studies are also complicated by the widely ranging rates of reactivity for DOM compounds (from hours to centuries) (10) and the addition and removal of different compounds to the DOM pool as DOM transits through the estuary. Because of the extreme heterogeneity of DOM, using traditional 1-dimensional approaches will not enable us to detect significant changes in the FTIR and 13C NMR spectra of the DOM sample. Noda (11) has developed a new way to examine data sets that are not unlike the ones that can be generated from spectroscopic studies of DOM along the above transition. The concept is called generalized perturbation-based twodimensional correlation spectroscopy (2D-correlation spectroscopy). By distributing spectral intensity trends within a data set collected as a function of the perturbation sequence (e.g., time, temperature, pressure change, chemical reaction) over a second dimension, one can observe cross-correlations that define structural relationships. This helps to resolve overlapped peaks (as many of them do not coevolve throughout the data set), provides relations between different peaks, and provides information on the sequential order of the changes in these peaks. The use of 2D spectroscopy in this fashion can potentially resolve many ambiguities especially in a complex mixture of compounds with many overlapping peaks (12, 13). This approach has been applied widely outside the environmental arena and successfully used to understand changes within polymer materials and proteins, to follow chemical reactions, and to investigate several research issues in biological and biomedical sciences (12, 14) but has, to date, been underutilized in environmental and aquatic sciences. Unlike most ordination statistical analyses approaches, which investigate variations in the spectra combined into a single data set to examine differences among samples, 2D correlation spectroscopy exploits intensity changes of the individual chemical bands along a reaction series. Identified correlations among different bands are based on in-phase and out-of-phase changes in their intensities across the suite of spectra. In this work, we investigate the changes in the chemical composition of high molecular weight (>1000 Da) estuarine DOM isolated during two years along a salinity transect from Great Bridge, Virginia, USA, through the Elizabeth River and lower Chesapeake Bay to the coastal Atlantic Ocean. To our knowledge, while 2D correlation spectroscopy is just beginning to be applied to FTIR analyses of environmental samples (15), it has not previously been applied to investigate changes in the chemical composition of natural organic matter samples using 13C NMR or the combination of 13C NMR and FTIR data. Such approaches can provide critical insights into DOM composition (as shown here) and should prove useful in studies of particulate and sedimentary organic matter as well. 10.1021/es100898x

 2010 American Chemical Society

Published on Web 10/13/2010

Methods and Materials Sampling Sites and Sampling. Samples were collected from four sites along the transect from Great Bridge (an upriver site on the southern branch of Elizabeth River, VA) through Elizabeth River/Chesapeake Bay system to the coastal ocean: site GB (Great Bridge, Virginia), site TP (Town Point Park, Norfolk, Virginia), site CBB (Chesapeake Bay Bridge-Tunnel), and site OSC, an offshore coastal site in the Atlantic ocean off the Virginia coast (37° 10.132′ N, 75° 37.891′ W). Surface water samples were collected according to our previous work (16) from the four sites during seven sampling periods (October 2005 to May 2007). The samples from OSC were collected only during four sampling periods due to weather instability or unavailability of the research vessel. Upon arriving in the laboratory, pH was measured with a calibrated pH meter (Thermo Scientific), while salinity was measured by a salinity refractometer with automatic temperature compensation (Fisher Scientific). The refractometer was calibrated with Sargasso Sea surface water (S ) 35). Then, samples were sterile-filtered using cleaned and preconditioned 0.1 µm Whatman Polycap cartridge filters. DOC concentrations were measured (with triplicate injections) using a standard DOC analyzer (Aurora 1030W TOC analyzer, College Station, TX) as descried previously (16). Ultrafiltration. The ultrafiltration processes were conducted according to our previous work (16). In brief, an ultrafiltration system equipped with a polysulfone 1 kDa cartridge that has 25 sq. ft active surface area (Separation Engineering, Inc.) was used to separate both high and low molecular weight fractions (hereafter HMW and LMW, respectively). After cleaning and conditioning the system, 40-160 L of sterile-filtered water was concentrated to approximately 2.3 L at pressures of approximately 30 psi. The salt-containing retentate (HMW) solution was diafiltered with 15-20 L deionized water until the filtrate reached a salinity of zero (measured by refractrometer). The retentate fraction (HMW) from each of the four sites was further diafiltered with 3 L deionized water using a stirred cell (Amicon 8400) equipped with a 1 kDa regenerated cellulose membrane (Millipore). Subsamples from the final retentate were taken for DOC measurements. The resulting retentates were frozen and then freeze-dried. Solid State 13C NMR. All HMW-DOM samples isolated during the seven sampling periods from GB, TP, and CBB sites -with the exception of GB-November 2005, which was lost before 13C NMR analysis- and the four samples from OSC were analyzed by cross-polarization/magic angle spinning (CP/MAS) solid-state 13C NMR using a Bruker Avance II spectrometer operating at 100 MHz for 13C, spun at 14 kHz. The recycle delay (D1) and contact time were 1 s and 1.5 ms, respectively. Exactly 4569 acquisitions were averaged for each sample after verifying that the spectral acquisition parameters are quantitatively representative of the sample compositions. This was accomplished by optimizing the various parameters (17). All the experiments were conducted using a 4-mm triple resonance probe, and the samples were packed into 4-mm NMR rotors with Kel-F caps. 13C NMR spectra were normalized by their total area between 0-220 ppm and multiplied by 1000 before being used for 2D correlation 13C NMR. FTIR Analysis. Samples were prepared as a mixture of exactly 1.0 mg sample and 100 mg KBr and were then ground and homogenized to reduce light scatter (18) using a WigL-Bug ginding mill. A subsample was then compressed between two clean, polished iron anvils in a hydraulic press twice at 20,000 psi to form a KBr window with 90° degree rotation before the second press, to minimize wedging effects (19). FTIR spectra were obtained according to our previous work (16), by collecting 200 scans with a Nicolet 370 FTIR spectrometer. The FTIR spectra were converted into absorbance units, normalized by the summed absorbance from

4000-500 cm-1, and multiplied by 1000. We will only focus on the region between 1800-900 cm-1, as it includes the major bands of amide, carboxylic, ester, and carbohydrate functional group (16). 2D Correlation Spectroscopy. Normalized spectra from 13 C NMR and FTIR (24 spectra each) were analyzed by 2D correlation spectroscopy using an in-house modified version of 2Dshige software (Kwansei-Gakuin University, Japan). The practical computation of 2D correlation spectra was according to Noda and Ozaki (12). Applying synchronous and asynchronous calculations to the entire range of wavenumber/chemical shift regions with initial water-sample salinity as the perturbation variable will produce two square matrices that could be represented by two 2D contour maps, one for the synchronous spectrum and the second for the asynchronous spectrum. Because changes in noise are normally out-of-phase with each other, and noise could cause artifacts and misinterpretation, especially for asynchronous maps (12), we applied the principal component analysis (PCA) noise reduction method proposed by Jung and co-workers (20). The FTIR spectra did not require noise reduction. The synchronous map of the 2D heterospectral correlation analysis of FTIR and 13C -NMR was developed according to Noda and Ozaki (12).

Results Salinity and DOC Concentrations. In general, during all seven sampling periods the salinity ranged along the transect from 10 to 32. While the DOC concentration ranged from 109 to 940 µM-C, the lowest salinity and highest DOC values were at GB and the highest salinity and lowest DOC were at OSC. Ultrafiltration DOC recoveries after the second diafiltration step ranged from 11 to 48%; the percentage recovery decreased approaching the marine end member, which is in agreement with previous studies (21, 22). 2D Correlation 13C NMR. Overlaying the 24 area-normalized 13C NMR spectra of the estuarine/marine HMW-DOM isolates (Figure S1a- Supporting Information) shows major changes in the intensity of most the bands; however, the spectra appear noisy. Since the first two principal components (PC-1 and PC-2) account for a significant amount (88%) of the variation in 13C NMR spectra, we can use them to reconstruct less noisy spectra (i.e., the PCA noise reduction method). The resulting reconstructed spectra retain their major peaks and exhibit a reduction in noise (Figure S1bSupporting Information). One of the drawbacks for using only 88% of the variation in the original spectra is that the approach may also remove some informational features along with the noise. Therefore, we plotted the spectra of the removed noise (Figure S1c- Supporting Information) which exhibits no recognizable features and is clearly background noise. As an additional check we compared synchronous and asynchronous contour maps with and without the noise; the synchronous map shows no noticeable differences, while the unadjusted asynchronous contour map appears a little bit noisy. All the 13C NMR 2D-correlation results presented below were generated from the reduced-noise spectra. 13 C NMR Synchronous Map. The synchronous map shows a symmetric spectrum with respect to the diagonal line (Figure 1); nine major autopeaks were identified along the diagonal line. The nine autopeaks in our data set are those peaks that changed intensity in all the 13C NMR spectra: three of these (20, 26, and 32 ppm) appear in the paraffin carbon region; one autopeak (46 ppm) is in the methoxy and amino group region; a major autopeak (74 ppm) is in the middle of the O-alkyl (HCOH) region; another autopeak (103 ppm) indicates anomeric carbons, while a broad autopeak (132 ppm) occurs in the region of CdC/aromatic groups. The carboxylic acid, ester, and amide region (160-190 ppm) shows two autopeaks, a very small one at the band maximum VOL. 44, NO. 21, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8045

FIGURE 1. The synchronous contour map generated from all the 13 C NMR (n ) 24) of HMW-DOM isolated from the four estuary/ marine sites during the seven sampling periods where the top and the right side are the average 13C NMR spectra (included as reference). Red represents positive correlations and blue represents negative correlations; higher color intensity indicates a stronger positive or negative correlation. (178 ppm) and another larger one at the high chemical shift side of the band (183 ppm). Comparing noise-reduce 13C NMR spectra (Figure S1b- Supporting Information) we see only small changes in the maximum intensity of the 178 ppm band; however, the bandwidth narrows toward higher chemical shift, consistent with the autopeaks results. Comparing autopeaks intensities, the 74 ppm band exhibits the highest spectral variation through the estuarine transect; the 178 ppm shows the lowest spectral variation. Off-diagonal peaks (cross peaks) in the synchronous map illustrate correlated signals. For example, the band centered at 74 ppm is negatively correlated with bands at 32, 46, 132, and 183 ppm as indicated by the blue cross peaks between them and positively correlated with the bands at 20, 26, 103, and 178 ppm (red cross peaks). This indicates that signals at 74 ppm (carbohydrates) are covarying with other signals that follow the same temporal and spatial trends. Although most of the chemical shifts within the 13C NMR spectra show changes either through autopeaks or cross peaks, the region between 54 and 62 ppm, which is a part of the methoxy/ amino region (45-60 ppm), displays no correlated intensity changes regardless of its high intensity in all the original 13C NMR spectra. 13 C NMR Asynchronous Map. The asynchronous map provides information about the sequential order, or timing, of the changes in spectral bands. Although salinity is used as the perturbation parameter to generate the asynchronous map, we cannot ascertain if the indentified DOM compositional changes are due specifically to ionic strength effects or other environmental factors (e.g., light availability, primary productivity) that are often correlated with salinity and can also affect DOM composition. However, the asynchronous map can still be informative, supporting what we observe in the synchronous map and differentiating overlapped band resolution. It is especially useful in investigating whether two bands are totally in-phase with each other or out of phase. The asynchronous map (Figure S2- Supporting Information), unlike the synchronous map, is antisymmetric with the respect to the diagonal line and exhibits no 8046

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 21, 2010

FIGURE 2. The synchronous contour map generated from the FTIR region (between 1800-900 cm-1) (n ) 24) of the HMW-DOM isolated from the four estuary/marine sites during the seven sampling periods. The average FTIR spectrum is plotted along the top and right side as a reference. To aid in visualization of intense peaks such as 1151 cm-1, peaks are contoured rather than presented in red and blue as in Figure 1. Here gray indicates negative correlation. autopeaks. The asynchronous map clearly confirms that the band maximum at 178 ppm consists of two biogeochemically different bands at 178 and 183 ppm. In general, both synchronous and asynchronous maps demonstrate that the DOM along our transect is a complex mixture of organic compounds differing in biogeochemical reactivity. The maps also show that major 13C NMR bands appearing in the original 1-D spectra result from overlapped bands from the common functional groups in DOM. FTIR Synchronous Map. The synchronous map generated from the 24 HMW-DOM FTIR spectra (Figure 2), displays 8 major autopeaks at 1735, 1660, 1580, 1467, 1402, 1151, 1088, and 1043 cm-1. The highest change in intensity was in the band at 1151 cm-1. We assign these bands as follows: the band at 1735 cm-1 to the asymmetric stretching of CdO of ester, the band at 1660 cm-1 to the asymmetric stretching of the CdO band of amide compounds, the band at 1580 cm-1 to the asymmetric stretching of deprotonated carboxylic acid (COO-), the band at 1402 cm-1 to the symmetric stretching of deprotonated carboxylic acid, the band at 1467 cm-1 to CH3 asymmetric deformation, and the band at 1151 cm-1 to the carbohydrate ring vibration, the band at 1088 and 1043 cm-1 to the C-O asymmetric stretching (16). The cross-peaks show a positive correlation between the bands at 1580 and 1402 cm-1 and another positive correlation between the bands at 1151 and 1088 cm-1. This correlation confirms our assignment of these bands as due to carboxylic acid-rich structures and carbohydrate-rich structures, respectively. The cross peaks also indicate that the bands at 1580 and 1402 cm-1 are negativity correlated with the bands at 1151 and 1088 cm-1. We further suggest that the band around 1580 cm-1 is due to asymmetric stretching of highly oxygen substituted aliphatic and deprotonated carboxylic acids. This suggestion is based on the previous extensive study by Hay and Myneni (23) which shows that, although the location of asymmetric stretching of (COO-) is very sensitive to changes in chemical structure within the model compounds, in natural organic matter it always appears as a narrow band centered at 1578 cm-1. This indicates the presence of a very specific chemical

environment for the carboxyl groups in DOM and matches the band position of model carboxylic compounds that have oxygen substitution (hydroxyl, carboxylic, or ester) at the R or β carbon position. 2D Hetero-Correlation of FTIR and 13C NMR Spectra. The 2D hetero-correlation map (Figure S3-Supporting Information), used to examine covariations among the two techniques, shows a positive correlation between the FTIR band at 1580 cm-1 and the 13C NMR band at 183 ppm, indicating that both bands result from the same chemical bond (the carboxylic group in the highly oxygen-substituted aliphatic carboxylic-acid). The 2D hetero-correlation map also shows a positive correlation between the FTIR band at 1151 cm-1 and the 13C NMR band at 74 ppm, which further strengthens our assignment of the FTIR band at 1151 cm-1 to the alkyl-O bond in carbohydrates, highlighting the possibility of using the FTIR band at 1151 cm-1 to indicate changes in carbohydrate content between different DOM samples. These band positions match with previously identified peak positions for carboxylic and carbohydrate bands in both FTIR and 13C NMR (24-26).

Discussion The 2D maps discussed above provide information regarding the way NMR and FTIR peaks of DOM correlate along a salinity gradient in the lower Chesapeake Bay. However, there is additional information that can be gleaned from the correlations-primarily the nature of DOM components. The expectation is that structural components having similar chemical properties will respond uniformly to salinity changes and this response is likely to be different than that experienced by other components having a different structural makeup. Thus, one can gain insights about the individual components that synchronize in response to salinity by identifying their functional group composition relating to the various spectral signatures. From the 2D analysis, it is clear that two or three major components can be readily discerned. Heteropolysaccharides (HPS). Because the band at 74 ppm represents carbohydrate ring structures (HC-OH) and is the most intense band in the 13C NMR spectra, it is a useful tracer to investigate chemical structures that are correlated with it. By extracting a horizontal slice (50-100 ppm) of the synchronous map shown in Figure 1, plotting the entire 13C NMR spectral range on the abscissa (see Figure S4- Supporting Information), we readily observe that the peak at 74 ppm (carbohydrates) correlates in the positive direction with bands at 20, 26, 103, and 178 ppm. These bands have reasonably and previously been proposed, from 13C NMR spectra, as being associated with heteropolysaccharide (HPS) compounds (28). This HPS component of DOM becomes more dominant toward the marine end of the transect and can be expected to correlate well with salinity, especially as it appears to be the major component of HMW-DOM in the surface open ocean (21, 28, 29). However, the correlation intensity of the major HPS bands (20, 26, 103, and 178 ppm), in both synchronous and asynchronous maps, indicates that HPS is not a single class of compounds but a mixture of compound classes that have the same basic backbone structures but slightly different geochemical reactivities. This is consistent with what we know about heteropolysaccharides in general, which can be found as the sugar component in many different compound groups (e.g., lipopolysaccharides, peptidoglycans, mucopolysaccharides, etc.). Carboxylic-Rich Compounds (CRC). Another group of compounds that emerge from 2D correlations is called “carboxylic rich compounds” (CRC), and these are associated with the carboxylic acid band at 183 ppm. The horizontal slice for this region of the synchronous map (Figure S5aSupporting Information) shows a positive correlation be-

tween the band at 183 ppm and bands at 32, 46, and 132 ppm. In addition, a similar slice of the asynchronous map (Figure S5b- Supporting Information) shows no correlation (no cross peak) between the 183 ppm band and 32 ppm band indicating that, these bands are originally in-phase with each other. The 32 ppm band is consistent with the expected signal for a carbon R to carboxyl groups (30). The asynchronous in-phase correlation between 183 ppm and 32 ppm is consistent with this relationship. For example, if the carboxylic group is altered and is subjected to decarboxylation or lost from the DOM, the band originally at 32 ppm will either shift to a lower ppm value or is lost along with the carboxyl group. Based on the above 2D correlation, the chemical structures of the carboxylic rich compounds (CRC) have similar spectral qualities to the proposed carboxylicrich alicyclic molecules (CRAM) (27). However, there is not enough evidence to ascertain the alicyclic nature of these compounds, so we only identify them as carboxylic rich compounds (CRC). Supporting the hypothesis that at least some of the CRC are CRAM, Sleighter and Hatcher (9) identify carboxylic-rich compounds from FT-ICR-MS on C18 extracts of DOM isolated from the same water collected during the November 2006 period (Dismal Swamp to the offshore site). The resulting mass spectra are consistent with the CRAM structures proposed by Hertkorn and co-workers (27). However, they are also consistent with compounds that could derive from lignin. Quantification of the Identified Components Using FTIR and 13C NMR. We modified the integration of 13C NMR spectra used by Dria and co-workers (17) to reflect the biogeochemical reactivities of each of the 13C NMR regions using the synchronous map of the 13C NMR data (see above). Instead of the traditional integration regions that are commonly selected, we subdivide the 13C NMR spectrum into ten regions labeled in the following manner: 1CHn (0-29 ppm), 2CHn (29-40 ppm), 1CHn-O (40-55 ppm), 2CHn-O (55-62 ppm), HC-OH (62-90 ppm), O-C-O (90-115 ppm), CdC/Ar (115-140 ppm), Ar-O (140-160 ppm), COO/CON (160-190 ppm), and the aldehyde and ketone region (190-220 ppm). To deconvolute the carboxylic, ester, and amide components of the COO/CON region (160-190 ppm), we couple the 13C NMR data with the ratio of component areas calculated by FTIR as described in our previous work (16). In brief, the carbon percentages of the carboxylic acids, amides, and esters are estimated by multiplying the ratio of the area of the functional group to total carbonyl response (as determined by FTIR) by the COO/CON relative area from the 13C NMR according to the following equation %Cindividual ) (Aindividual /Atotal)*%CCOO/CON

(Eq.1)

where Aindividual is the FTIR total area of either carboxylic acid, amide, or ester, Atotal is the total carbonyl area in FTIR, and %CCOO/CON is the relative %C of carboxyl groups from 13C NMR spectra. The results of this analysis are shown in Table S1Supporting Information. In general, we observe an increase in DOM’s carbohydrate contents (HC-OH, O-C-O) that is seasonally modulated as we shift toward the marine end member, while the carboxylic carbon percentage is antiphase to this change. The amide content is significant at estuarine/ marine sites along the transect and has an average value of 7.36 ( 0.87%C; the small variability of this signal agrees with the small intensity of the autopeaks at 178 ppm and 1660 cm-1 in the synchronous maps of 13C NMR and FTIR, respectively (Figures 1 and 2). As discussed above, the 2D correlations allow us to subdivide peaks into two or more correlated components. The first group of components appears to be mostly assigned VOL. 44, NO. 21, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8047

FIGURE 3. The percentage of heteropolysaccharide (HPS) (9 -red) carbon, carboxyl-rich (CRC) (O- open) carbon, and amide/ amino sugar (AMS) (2-green) carbon relative to total HMW-DOC as calculated from the modified integration of the 13 C NMR against the salinity of their samples. to HPS. It includes the following NMR chemical shift regions: 1 CHn (0-29 ppm), HC-OH (62-90 ppm), O-C-O (90-115 ppm). The second group is assigned to the CRC. The following regions are associated with this: 2CHn (29-40 ppm), 1CHn-O (40-55 ppm), CdC/Ar (115-140 ppm), Ar-O (140-160 ppm), and the aldehyde and ketone region (190-220 ppm). A third group can be identified containing major functional groups of the amide/amino sugar (AMS) components including the amide and ester carbons as well as the region 2CHn-O (55-62 ppm). We can sum the relative areas of these three component groups, and these are given in Table S1. The above classification scheme is an operational scheme that is likely an oversimplification, as there will be some overlap in the various integration regions. For example, the aromatic signals like those from lignin could overlap and lead to an overestimation of the CRC contribution. If we assign the entire area in the region between the 115-160 ppm to aromatic compounds, these contribute 28 ( 10% and are included as CRC; however, the carbon double bond (CdC) in the unsaturated components of CRC would also contribute significantly to that region (27, 31). Until we can define pure end-members for each of the classification regions of the spectra, the generalized assignments will always exhibit a certain level of uncertainty. In addition, quantifying the amide/amino sugar compound based on the carbon percentage of the amide functional group and the 2CHn-O (55-62 ppm) could underestimate the amount AMS compounds, as these two regions represent only two carbon types of the amide/amino sugar compounds, while the rest of the carbon types could be included in the heteropolysaccharide component (e.g., N-acetyl amino polysaccharides). This applies to the current work as well as other studies attempting to classify structural entities within DOM (27, 32). Nonetheless it has been deemed useful to define compound classes based on observed spectral signatures. What distinguishes our work from that of others (27, 32) is the fact that classification is based on 2D correlations along a salinity gradient and not hypothetical structural makeup, even though we introduce some level of structural makeup to suggest possible class entities to explain correlations influenced by salinity variations. Based on our simplified scheme, we plot the percentage of HPS, CRC, and AMS components from all sites during the seven sampling periods against salinity (Figure 3). The total contribution of HPS in the HMW-DOM shows a wide range from 45-75%. The percentages increase as we shift toward the marine end member. In general, all sites present their lowest percentages in the winter seasons and their highest during the summer season. Conversely, the CRC show a range 8048

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 21, 2010

from 12 to 40% with the highest values during the winter and fall sampling periods and the lowest percentage in the summer season (see Table S1). This is in agreement with the δ13C data of these samples (unpublished data) that show stronger shifts toward 13C-enriched material in the summer than winter seasons, indicating higher phytoplankton primary production and photo-oxidation of DOM during the summer season relative to the winter seasons (33, 34). In contrast, the AMS component exists within a narrow range (12 to 15%) and does not show strong seasonal or spatial trends. This invariance could be indicative of the important role of heterotrophic bacteria in altering the DOM throughout the transect, regardless of season or DOM character/source; heterotrophic bacteria have been estimated to contribute ∼50% of the DON in marine environment (34). Such a relationship was unexpected and needs further investigation (35). Are the AMS compounds (or a significant proportion of them) refractory or are their sinks and sources in steady state relative to the HPS and CRC? We applied the same modified integration to the previously reported 13C NMR and FTIR spectra of HMW-DOM from the Dismal Swamp (DS) isolated on August 2007 (16), recognizing, of course, that the spectrum is partly influenced by the abundance of Fe which may affect signal quantitation. DS DOM represents one of the terrestrial DOM sources to the Chesapeake Bay. In DS, the carbon percentage of the CRC in the HMW-DOM is 60%, while the HPS is only 35%, and the AMS is 5%. If we calculate a ratio of carboxylic acid to the total CRC carbon in the Dismal Swamp, we obtain a value of 1 (carboxylic-carbon): 3 (CRC-carbon). This same ratio for the estuarine and coastal HMW-DOM samples in this study shows a wide range, 1:7 to 1:32. The ratio increases as we approach the marine offshore sample. If the DS DOM is a sole source of CRC to the offshore and CRC is refractory, then we would expect this ratio to be unchanged across the transect. That it is not constant means that CRC in offshore samples is different than CRC in the DS sample. Alternatively, CRC could be undergoing modification by microbial or photochemical processes along the transect. Our data are consistent with previous work (21, 29), where two major components of HMW-DOM are identified along the Mid-Atlantic Bight using 1H NMR. The first component bears some relationship to our HPS components (but was called acyl polysaccharide APS) accounting for 50-80% of the HMW carbon with the highest percentage found in fully marine DOM. The other HMW-DOM component is more abundant in the DOM isolated from estuarine and deep ocean DOM, and has been suggested to resemble humic substances. However, work done recently (27) suggests that the deep water humic substances are mostly CRAM. Our study shows that CRC are also present in estuaries and could potentially come from terrestrial sources as suggested by others (36). However, unlike Hertkorn and co-workers (27), we show that CRC are biogeochemically reactive and not refractory, which raises many questions about the biogeochemical cycle of these compounds and their sources and sinks.

Acknowledgments Thanks to Robert F. Dias (Chemistry Dept., Old Dominion University) for the laboratory space and help during sampling, Junyan Zhong (COSMIC, Old Dominion University) for his help obtaining the solid state CP/MAS 13C NMR, David Burdige (OEAS, Old Dominion University) for his valuable comments on the earlier version of the manuscript and the crew of the R/V Slover. We also thank the three anonymous reviewers for their detailed comments. This work was supported by the National Science Foundation (OCE 0453777 to E.C.M. and OCE 0612712 to P.G.H.).

Supporting Information Available 1) Table S1. Summary of the relative carbon percentages of HMW-DOM functional groups resulting from modified integration of the 13C NMR. 2) Figure S1a-c. Normalized 13C NMR spectra before and after noise reduction method. 3) Figure S2. The asynchronous contour map of 13C NMR. 4) Figure S3. The synchronous contour map of the 2Dheterospectral correlation (13C NMR and FTIR). 5) Slices of the synchronous and asynchronous spectra of 13C NMR (Figures S4-S5). This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Abdulla, H.; Dias, R.; Minor, E. Understanding the enhanced aqueous solubility of styrene by terrestrial dissolved organic matter using stable isotope mass balance and FTIR. Org. Geochem. 2009, 40, 547–552. (2) Chefetz, B.; Deshmukh, A. P.; Hatcher, P. G.; Guthrie, E. A. Pyrene sorption by natural organic matter. Environ. Sci. Technol. 2000, 34, 2925–2930. (3) Chiou, C. T.; Kile, D. E.; Brinton, T. I.; Malcolm, R. L.; Leenheer, J. A.; MacCarthy, P. A comparison of water solubility enhancements of organic solutes by aquatic humic materials and commercial humic acids. Environ. Sci. Technol. 1987, 21, 1231– 1234. (4) Shiller, A. M.; Duan, S.; van Erp, P.; Bianchi, T. S. Photo-oxidation of dissolved organic matter in river water and its effect on trace element speciation. Limnol. Oceanogr. 2006, 51, 1716–1728. (5) Del Vecchio, R.; Blough, N. V. Spatial and seasonal distribution of chromophoric dissolved organic matter and dissolved organic carbon in the Middle Atlantic Bight. Mar. Chem. 2004, 89, 169– 187. (6) Rochelle-Newall, E. J.; Fisher, T. R. Chromophoric dissolved organic matter and dissolved organic carbon in Chesapeake Bay. Mar. Chem. 2002, 77, 23–41. (7) Minor, E. C.; Boon, J. J.; Harvey, H. R.; Mannino, A. Estuarine organic matter composition as probed by direct temperatureresolved mass spectrometry and traditional geochemical techniques. Geochim. Cosmochim. Acta 2001, 65, 2819–2834. (8) Minor, E. C.; Dalzell, B. J.; Stubbins, A.; Mopper, K. Evaluating the photoalteration of estuarine dissolved organic matter using direct temperature-resolved mass spectrometry and UV-visible spectroscopy. Aquat. Sci. 2007, 69, 440–455. (9) Sleighter, R. L.; Hatcher, P. G. Molecular characterization of dissolved organic matter (DOM) along a river to ocean transect of the lower Chesapeake Bay by ultrahigh resolution electrospray ionization Fourier transform ion. Mar. Chem. 2008, 110, 140– 152. (10) Davis, J.; Benner, R. Quantitative estimates of labile and semilabile dissolved organic carbon in the western Arctic Ocean: A molecular approach. Limnol. Oceanogr. 2007, 52, 2434–2444. (11) Noda, I. Generalized two-dimensional correlation method applicable to infrared, Raman, and other types of spectroscopy. Appl. Spectrosc. 1993, 47, 1329–1336. (12) Noda, I.; Ozaki, Y. Two-Dimensional Correlation Spectroscopy: Applications in Vibrational and Optical Spectroscopy; John Wiley & Sons Inc.: San Francisco, 2004. (13) Wang, H. Palmer, R. A. Further insights on 2D-correlation theory. In Two-Dimensional Correlation Spectroscopy; Noda, I., Ozaki, Y., Eds.; American Institute of Physics: AIP conference proceedings No. 53, Kobe-Sanda, Japan, 1999. (14) Ozaki, Y. Two-dimensional Correlation Spectroscopy in Vibrational Spectroscopy. In Handbook of Vibrational Spectroscopy, Sample Characterization and Spectra Data Processing; Chalmers, J. M., Griffiths, P. R., Eds.; John Wiley & Sons, Ltd.: 2002. (15) Mecozzi, M.; Pietrantonio, E.; Pietroletti, M. The roles of carbohydrates, proteins and lipids in the process of aggregation of natural marine organic matter investigated by means of 2D correlation spectroscopy applied to infrared spectra. Spectrochim. Acta, Part A 2009, 71, 1877–1884.

(16) Abdulla, H. A.; Minor, E. C.; Dias, R. F.; Hatcher, P. G. Changes in the compound classes of dissolved organic matter along an estuarine transect: A study using FTIR and 13C NMR. Geochim. Cosmochim. Acta 2010, 74, 3815–3838. (17) Dria, K. J.; Sachleben, J. R.; Hatcher, P. G. Solid-state carbon-13 nuclear magnetic resonance of humic acids at high magnetic field strengths. J. Environ. Qual. 2002, 31, 393–401. (18) Smith, B. C. Fundamentals of Fourier Transform Infrared Spectroscopy; CRC Press: 1996. (19) Hirschfeld, T. Diagnosis and correction of wedging errors in absorbance subtract Fourier transform infrared spectrometry. Anal. Chem. 1979, 51, 495–499. (20) Jung, Y. M.; Shin, H. S.; Kim, S. B.; Noda, I. New approach to generalized two-dimensional correlation spectroscopy. 1: Combination of principal component analysis and two-dimensional correlation spectroscopy. Appl. Spectrosc. 2002, 56, 1562–1567. (21) Repeta, D. J.; Quan, T. M.; Aluwihare, L. I.; Accardi, A. Chemical characterization of high molecular weight dissolved organic matter in fresh and marine waters. Geochim. Cosmochim. Acta 2002, 66, 955–962. (22) Wang, X. C.; Altabet, M. A.; Callahan, J.; Chen, R. F. Stable carbon and nitrogen isotopic compositions of high molecular weight dissolved organic matter from four US estuaries. Geochim. Cosmochim. Acta 2004, 68, 2681–2691. (23) Hay, M. B.; Myneni, S. C. B. Structural environments of carboxyl groups in natural organic molecules from terrestrial systems. Part 1: Infrared spectroscopy. Geochim. Cosmochim. Acta 2007, 71, 3518–3532. (24) Mayo, D. W.; Miller, F. A.; Hannah, R. W. Course Notes on the Interpretation of Infrared and Raman Spectra; Wiley-Interscience 2004. (25) Pavia, D. L.; Lampman, G. M. Introduction to Spectroscopy: A Guide for Students of Organic Chemistry; Saunders Golden Sunburst Series, 1996. (26) Wilson, M. A. N. M. R. Techniques and Applications in Geochemistry and Soil Chemistry; Pergamon Press: 1987. (27) Hertkorn, N.; Benner, R.; Frommberger, M.; Schmitt-Kopplin, P.; Witt, M.; Kaiser, K.; Kettrup, A.; Hedges, J. I. Characterization of a major refractory component of marine dissolved organic matter. Geochim. Cosmochim. Acta 2006, 70, 2990–3010. (28) Aluwihare, L. I.; Repeta, D. J.; Chen, R. F. A major biopolymeric component to dissolved organic carbon in surface sea water. Nature 1997, 387, 166–169. (29) Aluwihare, L. I.; Repeta, D. J.; Chen, R. F. Chemical composition and cycling of dissolved organic matter in the Mid-Atlantic Bight. Deep-Sea Res., Part II 2002, 49, 4421–4437. (30) Jacobsen, N. E. NMR Spectroscopy Explained: Simplified Theory, Applications and Examples for Organic Chemistry and Structural Biology; Wiley-Interscience: 2007. (31) Esteves, V.; Otero, M.; Duarte, A. Comparative characterization of humic substances from the open ocean, estuarine water and fresh water. Org. Geochem. 2009, 40 (9), 942–950. (32) Baldock, J.; Masiello, C.; Gelinas, Y.; Hedges, J. Cycling and composition of organic matter in terrestrial and marine ecosystems. Mar. Chem. 2004, 92, 39–64. (33) Harvey, H. R.; Mannino, A. The chemical composition and cycling of particulate and macromolecular dissolved organic matter in temperate estuaries as revealed by molecular organic tracers. Org. Geochem. 2001, 32, 527–542. (34) Pennock, J. R.; Sharp, J. H. Phytoplankton production in the Delaware Estuary: Temporal and spatial variability. Mar. EcolProg. Ser. 1986, 34, 143–155. (35) Kaiser, K.; Benner, R. Major bacterial contribution to the ocean reservoir of detrital organic carbon and nitrogen. Limnol. Oceanogr. 2008, 53, 99–112. (36) Lam, B.; Baer, A.; Alaee, M.; Lefebvre, B.; Moser, A.; Williams, A.; Simpson, A. J. Major structural components in freshwater dissolved organic matter. Environ. Sci. Technol. 2007, 41, 8240–8247.

ES100898X

VOL. 44, NO. 21, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8049