Article pubs.acs.org/est
Molecular Characterization of Dissolved Organic Matter in Glacial Ice: Coupling Natural Abundance 1H NMR and Fluorescence Spectroscopy Brent G. Pautler,† Gwen C. Woods,† Ashley Dubnick,‡ André J. Simpson,† Martin J. Sharp,‡ Sean J. Fitzsimons,§ and Myrna J. Simpson*,† †
Environmental NMR Centre and Department of Chemistry, University of Toronto, Toronto, Ontario, M1C 1A4, Canada Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, Alberta, T6G 2E3, Canada, § Department of Geography, University of Otago, P.O. Box 56, Dunedin, New Zealand ‡
S Supporting Information *
ABSTRACT: Glaciers and ice sheets are the second largest freshwater reservoir in the global hydrologic cycle, and the onset of global climate warming has necessitated an assessment of their contributions to sea-level rise and the potential release of nutrients to nearby aquatic environments. In particular, the release of dissolved organic matter (DOM) from glacier melt could stimulate microbial activity in both glacial ecosystems and adjacent watersheds, but this would largely depend on the composition of the material released. Using fluorescence and 1H NMR spectroscopy, we characterize DOM at its natural abundance in unaltered samples from a number of glaciers that differ in geographic location, thermal regime, and sample depth. Parallel factor analysis (PARAFAC) modeling of DOM fluorophores identifies components in the ice that are predominantly proteinaceous in character, while 1H NMR spectroscopy reveals a mixture of small molecules that likely originate from native microbes. Spectrofluorescence also reveals a terrestrial contribution that was below the detection limits of NMR; however, 1H nuclei from levoglucosan was identified in Arctic glacier ice samples. This study suggests that the bulk of the DOM from these glaciers is a mixture of biologically labile molecules derived from microbes.
■
Antarctica since 2006,5 while an estimated 64 ± 14 Gt yr−1 loss per 1 K increase in air temperature of glacier ice has also been observed in the Canadian Arctic Archipelago in response to warmer summer temperatures.6 Disruption of glacier ice environments may result in several biogeochemical variations in both glacier and surrounding ecosystems. For example, changes in glacier ice volume and/or temperature may increase microbial activity within the glacier or on the ice surface (known as cryoconite holes),7 thereby increasing the metabolic heat input within the glacier,8 inducing heterotrophic microbial community development in newly
INTRODUCTION Recent and on-going changes in atmospheric heat transport, alteration of oceanic and atmospheric circulation (including cloud cover), and the disappearance of snow and sea ice have amplified climatic warming of polar regions, making ecosystems in these regions highly fragile.1 An increase in surface air temperatures may alter the polar hydrologic cycle by reducing the mass of polar glaciers and ice sheets, increasing river discharge to the oceans and ultimately raising the global mean sea level.2 Perturbations to the mass and/or volume of polar glaciers and ice sheets, which are the second largest freshwater reservoir on Earth, may increase as the result of climate warming, leading to increased runoff and sea-level rise,3 which was currently measured to be 1.48 ± 0.26 mm yr−1 from all icecovered regions globally.4 Satellite gravity measurements have shown a significant acceleration in glacier ice mass loss in © 2012 American Chemical Society
Received: Revised: Accepted: Published: 3753
November 4, 2011 February 28, 2012 March 2, 2012 March 2, 2012 dx.doi.org/10.1021/es203942y | Environ. Sci. Technol. 2012, 46, 3753−3761
Environmental Science & Technology
Article
Table 1. Summary of Glacier Ice Samples glacier
location
coordinates
elevation (m)
temp (°C)
Victoria Upper Glacier (VUG)
McMurdo Dry Valleys, Antarctica
79°29′S, 161°53′E
−
−23
Clark Glacier (CG)
McMurdo Dry Valleys, Antarctica Garwood Valley, Antarctica
77°417′S, 162°54′E 78°024′S, 163°80′E 79°32′N, 90°84′W
−
Joyce Glacier (JG) White Glacier (WG)
John Evans Glacier (JEG)
Axel Heiberg Island, Canada
Ellesmere Island, Canada
79°70′N, 74°521′W 79°64′N, 74°40′W
79°70′N, 74°521′W
sample type
name
depth (m)
DOC (mg L‑1)
−15
basal ice glacier ice basal ice
VUG basal ice VUG glacier ice CG basal ice
− − −
0.4 0.3 0.3
−
−17
basal ice
JG basal ice
−
0.3
1275
−15
glacier ice core glacier ice core glacier ice core glacier ice core basal ice
WG core 22 “clear ice” WG core 22 “white ice” WG core 28 “clear ice” WG core 28 “white ice” JEG basal ice
9.6
0.5
9.7
0.7
12.6
0.5
12.7
0.7
−
0.4
glacier core glacier core glacier core glacier core
ice
JEG core 5
1.8
0.5
ice
JEG core 5
4.4
1.1
ice
JEG core 6
0.8
0.6
ice
JEG core 6
7.1
1.0
−
−10.9
800
1100
exposed barren “ancient carbon” upon glacial retreat9 and/or release of previously “frozen” nutrients into adjacent aquatic ecosystems.10,11 Dissolved organic matter (DOM) contained within glacier ice is of particular interest because DOM in other water reservoirs is considered to be an important contributor to the global carbon cycle12 and glacial runoff is hypothesized to be a quantitatively important source of biologically labile DOM to marine ecosystems.13 Therefore, it is possible that climatic alterations to the glaciers could release DOM that would not only alter adjacent aquatic biogeochemical cycling but may also result in an increase in atmospheric CO2 from DOM mineralization. Determining the biogeochemical fate of DOM is imperative for future predictions of carbon-cycle alterations resulting from climate change. The reactivity and release of DOM from aquatic ecosystems to the atmosphere may vary depending on its overall composition.14,15 Therefore, the molecular-level characterization of DOM in glacier ice from a variety of locations is needed to more accurately predict the potential impact of its release to the carbon cycle. Initial molecular-level investigations of DOM from glacier ice meltwater extracted by C18 disks and analyzed by Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) have revealed structural variability between samples collected from different sampling sites and different stages of the melt season;16,17 however, DOM extraction may be biased by particular molecular affinities to the C18 disks, as this technique is known to selectively emphasize hydrophobic constituents of DOM and may be biased against hydrophilic analytes.18 Furthermore, DOM extraction from glacier ice is labor intensive due to its low concentration and logistically challenging due to remote sampling locations. As a result, glacier ice DOM has mostly been characterized by fluorescence spectroscopy because this method does not require extraction and entails the minimal handling of samples.19−21 Although this allows for the potential identification of microbial- and terrestrial-derived components of DOM,22 precise molecular
structures cannot be assigned by this technique alone. Recent technical advancements in FT-ICR-MS and 1H nuclear magnetic resonance (NMR) spectroscopy have allowed for detailed molecular-level analysis of DOM in low, naturally occurring concentrations without any significant pretreatment from both natural waters23−25 and glacier ice.26 In this study we investigated the composition of DOM in a variety of glacier ice samples from the Canadian Arctic and Antarctica at natural abundance by both fluorescence spectroscopy and 1H NMR spectroscopy. Glacier ice samples that formed under different conditions from a variety of geographic locations and at different times (represented by sample depths from the ice surface) were analyzed by these techniques in tandem to determine the relative contributions of allochthonous and autochthonous DOM sources to each and assess its potential for mineralization if released from the ice matrix. The spectrofluorescence of DOM in each ice sample is represented by an excitation−emission matrix (EEM) and the data set is modeled by parallel factor analysis (PARAFAC). PARAFAC modeling decomposes each EEM spectral signature from the data set into loadings associated with individual components that can then be related to individual fluorophores or, as is more likely with DOM, complex fluorescence phenomena.27 1H NMR spectra acquired by the shaped presaturation-water suppression by gradient tailored excitation with an optimized W5 pulse train (SPR-W5-WATERGATE) pulse sequence is applied to nonselectively detect individual DOM constituents from glacier and basal ice samples with extremely low (0.3−1.1 mg L−1) dissolved organic carbon (DOC) without isolation, which was previously thought to be unattainable for 1H NMR spectroscopic analysis of natural samples. This information will provide insight into the potential biogeochemical role of glacial DOM in local and global contributions to carbon cycling in a changing climate. 3754
dx.doi.org/10.1021/es203942y | Environ. Sci. Technol. 2012, 46, 3753−3761
Environmental Science & Technology
Article
Figure 1. SPR-W5-WATERGATE 1H NMR spectra of glacier ice samples from two JEG cores at different glacier depths. Major structural (along with water-suppressed) regions are highlighted, and dominant 1H resonances for lactic acid, acetic acid, formic acid, and methanol are assigned.
■
EXPERIMENTAL SECTION
for long-term storage after which subsamples were cut with a bandsaw.26 The basal ice sampling from CG, JG, and JEG was conducted with an ethanol-bathed and flame-sterilized steel chisel, with sampled ice chunks being caught in an ethanolbathed and flame-sterilized aluminum tray and subsequently transferred to sterile Whirlpak bags. DOC was measured as nonpurgeable organic carbon using high-temperature combustion (680 °C) on a Shimadzu TOC-VCSN/TNM-1 analyzer equipped with a high sensitivity catalyst for trace-level analyses (DOC ≤ 1 mg L−1). An approximate conversion factor of 1.7 can be used to estimate DOM from DOC measurements.30 Sample Preparation and NMR Spectroscopy. Approximately 4 g of ice was allowed to melt into a scintillation vial followed by the immediate addition of NaN3 (∼5 mg) to restrict microbial growth and filtration through a 0.2 μm syringe Teflon filter to remove any fine particulates. Although it has been suggested that filtration of DOM samples may alter the characteristics of DOM,31 sample filtration is necessary to remove fine particulates for high-resolution NMR analysis and to prevent biological degradation that has been observed over time in unfiltered samples.24 Our previous research did not observe any alteration to glacier ice DOM composition with filtration.26 A 800 μL portion of filtered melted ice was transferred to 5 mm NMR tubes and 2.5% D2O (v/v) was added to each sample for the spectrometer lock. Organic-free deionized Milli-Q water was used as a method blank and has previously been shown via 1H NMR to be free from background contamination.26 Natural abundance 1H NMR experiments were performed using the SPR-W5-WATERGATE sequence for water suppression24,26 on a Bruker Avance 500
Glacier Samples. Ice samples were collected from four different glaciers in Antarctica and the Canadian Arctic. Antarctic samples were collected from Victoria Upper Glacier (VUG), Clark Glacier (CG), and Joyce Glacier (JG), which are cold-based polar glaciers with ice temperatures well below the pressure-melting point of water, with values of −23 °C for VUG,20 −15 to −17 °C for CG, and JG basal ice (Table 1). Canadian Arctic samples were collected from the White Glacier (WG) on Axel Heiberg Island and John Evans Glacier (JEG) on Ellesmere Island, which are polythermal glaciers containing a mix of ice at and below the pressure melting point measured to be 1528 and 10.9 °C, respectively (Table 1). Two types of ice were sampled from the WG ice core: “white ice”, formed from the metamorphism of snow with very little melting/refreezing, and “clear ice”, which was visibly devoid of air bubbles and formed by refreezing of meltwater within snow during the process of firnification. The proglacial and ice marginal areas of the Late Holocene JEG are characterized by sparse tundra vegetation, and the glacier has overrun organic matter (OM) from vegetation/soils incorporated into its basal ice;29 basal ice was collected from the wall of an ice tunnel excavated 15 m into the terminus of JEG. All basal ice samples were obtained from within 1.5 m of the ice-bed contact. Glacier and basal ice (containing finely laminated grains) were sampled from the terminal cliffs of VUG while only basal ice was collected from CG and JG. Ice cores were collected using a Kovacs ice-coring drill. Sampling from the ice cliff at VUG was conducted using a chain saw to cut large ice blocks which were wrapped in plastic 3755
dx.doi.org/10.1021/es203942y | Environ. Sci. Technol. 2012, 46, 3753−3761
Environmental Science & Technology
Article
On the basis of the presence of several resolved or semiresolved 1H peaks present in the glacier ice samples from JEG, several amino acids are likely contributors to this mixture and include: alanine (Ala), valine (Val), leucine (Leu), isoleucine (Ile), lysine (Lys), serine (Ser), glycine (Gly), and the glutamic acid derivative pyroglutamic acid (PyroGlu; Figures 2 and S1, Supporting Information). Aromatic 1H
MHz spectrometer equipped with a 5 mm QXI probe with an actively shielded Z-gradient. Experiments were acquired with 30 720 scans, a saturation loop of 2.25 s, and 32 768 time domain points. Spectra were externally calibrated to the trimethylsilyl resonance (0 ppm) of 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt, apodized by multiplication with an exponential decay producing a 1 Hz line broadening in the transformed spectrum with a zero-filling factor of 2. The Bruker Biofluid Reference Compound Database (version 2.0.3, Bruker BioSpin) in conjunction with analysis of standard compounds facilitated the identification of the constituents within the glacial ice.26 The pH of all samples was ∼6, so the DOM mixture constituents may be protonated or deprotonated depending on the pKa; for simplicity, all of the compounds identified are reported in their acidic forms. Fluorescence Spectroscopy. Aliquots of the fresh ice samples (without NaN3) were taken for fluorescence analysis immediately after melting and filtration through a 0.2 μm syringe Teflon filter. EEMs were collected using excitation from 230 to 450 nm (5 nm increments) to generate emission spectra from 280 to 550 nm (2 nm increments) on an Agilent 1200 series fluorescence detector (G1231A) containing a xenon flash lamp and an offline cuvette for EEM acquisition. Instrument bias resulting from lamp fluctuations, wavelength-dependent output, and daily fluctuations as well as inner filter effects were corrected for based on previous procedures.27,32−34 The detector is equipped with a reference diode that adjusts for intensity drift and a quartz diffuser that reduces light. The EEMs were normalized for the wavelength-dependent output which was further verified by the analysis of rhodamine B.35 Daily water blank EEMs were subtracted from the sample EEMs followed by normalization to the area under the water Raman peak.36 Inner filter effects were found to be negligible at environmentally relevant DOM concentrations owing to the narrow cuvette (0.5 mm) on the Agilent fluorescence detector.32 Samples were acquired in duplicate and both peak positions and intensity were found to be reproducible within a standard error of
