Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for...

9
Extraction of α-cellulose from mummied wood for stable isotopic analysis Benjamin A. Hook a, , Jochen Halfar b , Jörg Bollmann a , Ze'ev Gedalof c , M. Azizur Rahman b , Julito Reyes d , Daniel J. Schulze b a Department of Earth Sciences, University of Toronto, Toronto, ON M5S 3B1, Canada b Department of Chemical and Physical Sciences, University of Toronto Mississauga, Mississauga, ON L5L 1C6, Canada c Department of Geography, University of Guelph, Guelph, ON N1G 2W1, Canada d Geological Survey of Canada (Calgary), Calgary, AB T2L 2A7, Canada abstract article info Article history: Received 15 August 2014 Received in revised form 16 February 2015 Accepted 3 April 2015 Available online 15 April 2015 Editor: Michael E. Böttcher Keywords: Paleoclimate Sub-fossil Brendel Jayme-Wise Paleocene/Eocene transition Subarctic North America Stable isotope ratios of carbon (δ 13 C) and oxygen (δ 18 O) from tree ring cellulose can provide valuable paleocli- matic information at annual and subannual resolution, from time periods long before instrumental climate re- cords. Recently, mummied (non-permineralized) wood was discovered within Canadian Subarctic kimberlites, inviting paleoclimatic investigations of the time in which the trees grew. In the mummied wood, polysaccharides (hemicellulose, α-cellulose) are frequently preferentially degraded, in the low-silica anaerobic burial environment of the kimberlites, resulting in a lignin-rich material. However, some samples from the Ekati Panda kimberlite pipe (ca. 53.3 Ma) contain remnant cellulose, demonstrating the extraordinary preserva- tion potential of kimberlites. Preservation of α-cellulose is important because it allows for the construction of sta- ble isotopic proxy records of deep-time paleoclimates at annual to subannual resolution. Established α-cellulose extraction methods [i.e., using a 17.5% sodium hydroxide (NaOH) solution] were unsuitable for this material because all holocelluloses were dissolved. Therefore, we tested variants of two cellulose extraction methods [i.e., Brendel et al. (2000) and Jayme-Wise (Leavitt and Danzer 1993, Loader et al. 1997)] to optimize a procedure for extraction of a consistent yield of mummied wood cellulose for stable isotopic analysis. Stable carbon (δ 13 C) and oxygen (δ 18 O) isotopes were measured from cellulose from each extraction method, as well as Extractive- Free Wood and unextracted mummied wood. vitrinite reectance, used to assess thermal alteration of the unextracted material to estimate post-burial temperatures within the kimberlite, suggested low post-burial peak temperatures (T peak = 60 °C). We detected no mineral contaminants (i.e., iron oxides) in the material using Energy-Dispersive X-ray Spectroscopy (EDX). Despite low cellulose yield (b 5%), the Attenuated Total Re- ection Fourier-Transform Infrared (ATR-FTIR) spectra of mummied cellulose samples strongly resembled modern α-cellulose. All cellulose treatments were similar in stable isotope values and ATR-FTIR peaks, but signif- icantly different from unextracted wood and Extractive-Free Wood. As, reported by other sources (Anchukaitis et al. 2008, Brookman and Whittaker 2012), use of the Brendel method may cause cellulose acetylation, introduc- ing an FTIR peak near 1720 cm -1 , thus complicating hemicellulose detection at a peak near 1725 cm -1 . For this reason, the Jayme-Wise method is recommended for detection of hemicelluloses in mummied wood. If hemi- celluloses are not present in holocellulose, due to groundwater hydrolysis, the NaOH step may be omitted or re- duced in concentration and still produce α-cellulose. © 2015 Elsevier B.V. All rights reserved. 1. Introduction Mummied wood is a valuable paleoclimatic resource because it can be preserved for up to tens of millions of years under diagenetic condi- tions which limit permineralization or petrication, within kimberlites (Wolfe et al., 2012; Staccioli et al., 2014), lignite mines (Sahay, 2011), or in anoxic aqueous environments (Lewis et al., 2003). In these rare burial conditions, rather than decomposed or replaced by secondary minerals, wood material is preserved. Mummied wood is subjected to a much different diagenetic pathway than petried wood. For exam- ple, in the petried forests of Yellowstone National Park (USA), silicied woods are abundant in regions downwind from volcanoes that produce copious quantities of silicic volcanic ash (Ammons et al., 1987). Silica dissolves into groundwater, which inltrates wood cells after burial, allowing permineralization to occur (Chadwick and Yamamoto, 1984; Ballhaus et al., 2012). Wood silicication may occur in as quickly as 10100 years in the right conditions, such as Si-rich hot springs Chemical Geology 405 (2015) 1927 Corresponding author. E-mail address: [email protected] (B.A. Hook). http://dx.doi.org/10.1016/j.chemgeo.2015.04.003 0009-2541/© 2015 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Chemical Geology journal homepage: www.elsevier.com/locate/chemgeo

Transcript of Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for...

Page 1: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

Chemical Geology 405 (2015) 19–27

Contents lists available at ScienceDirect

Chemical Geology

j ourna l homepage: www.e lsev ie r .com/ locate /chemgeo

Extraction of α-cellulose from mummified wood for stableisotopic analysis

Benjamin A. Hook a,⁎, Jochen Halfar b, Jörg Bollmann a, Ze'ev Gedalof c, M. Azizur Rahman b,Julito Reyes d, Daniel J. Schulze b

a Department of Earth Sciences, University of Toronto, Toronto, ON M5S 3B1, Canadab Department of Chemical and Physical Sciences, University of Toronto Mississauga, Mississauga, ON L5L 1C6, Canadac Department of Geography, University of Guelph, Guelph, ON N1G 2W1, Canadad Geological Survey of Canada (Calgary), Calgary, AB T2L 2A7, Canada

⁎ Corresponding author.E-mail address: [email protected] (B.A. Hoo

http://dx.doi.org/10.1016/j.chemgeo.2015.04.0030009-2541/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 August 2014Received in revised form 16 February 2015Accepted 3 April 2015Available online 15 April 2015

Editor: Michael E. Böttcher

Keywords:PaleoclimateSub-fossilBrendelJayme-WisePaleocene/Eocene transitionSubarctic North America

Stable isotope ratios of carbon (δ13C) and oxygen (δ18O) from tree ring cellulose can provide valuable paleocli-matic information at annual and subannual resolution, from time periods long before instrumental climate re-cords. Recently, mummified (non-permineralized) wood was discovered within Canadian Subarctickimberlites, inviting paleoclimatic investigations of the time in which the trees grew. In the mummified wood,polysaccharides (hemicellulose, α-cellulose) are frequently preferentially degraded, in the low-silica anaerobicburial environment of the kimberlites, resulting in a lignin-rich material. However, some samples from theEkati Panda kimberlite pipe (ca. 53.3 Ma) contain remnant cellulose, demonstrating the extraordinary preserva-tion potential of kimberlites. Preservation ofα-cellulose is important because it allows for the construction of sta-ble isotopic proxy records of deep-time paleoclimates at annual to subannual resolution. Established α-celluloseextraction methods [i.e., using a 17.5% sodium hydroxide (NaOH) solution] were unsuitable for this materialbecause all holocelluloses were dissolved. Therefore, we tested variants of two cellulose extraction methods[i.e., Brendel et al. (2000) and Jayme-Wise (Leavitt and Danzer 1993, Loader et al. 1997)] to optimize a procedurefor extraction of a consistent yield of mummifiedwood cellulose for stable isotopic analysis. Stable carbon (δ13C)and oxygen (δ18O) isotopes were measured from cellulose from each extraction method, as well as Extractive-Free Wood and unextracted mummified wood. vitrinite reflectance, used to assess thermal alteration of theunextracted material to estimate post-burial temperatures within the kimberlite, suggested low post-burialpeak temperatures (Tpeak = 60 °C). We detected no mineral contaminants (i.e., iron oxides) in the materialusing Energy-Dispersive X-ray Spectroscopy (EDX). Despite low cellulose yield (b5%), the Attenuated Total Re-flection Fourier-Transform Infrared (ATR-FTIR) spectra of mummified cellulose samples strongly resembledmodernα-cellulose. All cellulose treatmentswere similar in stable isotope values and ATR-FTIR peaks, but signif-icantly different from unextracted wood and Extractive-Free Wood. As, reported by other sources (Anchukaitiset al. 2008, Brookman andWhittaker 2012), use of the Brendelmethodmay cause cellulose acetylation, introduc-ing an FTIR peak near 1720 cm−1, thus complicating hemicellulose detection at a peak near 1725 cm−1. For thisreason, the Jayme-Wise method is recommended for detection of hemicelluloses in mummified wood. If hemi-celluloses are not present in holocellulose, due to groundwater hydrolysis, the NaOH step may be omitted or re-duced in concentration and still produce α-cellulose.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

Mummifiedwood is a valuable paleoclimatic resource because it canbe preserved for up to tens of millions of years under diagenetic condi-tions which limit permineralization or petrification, within kimberlites(Wolfe et al., 2012; Staccioli et al., 2014), lignite mines (Sahay, 2011),or in anoxic aqueous environments (Lewis et al., 2003). In these rare

k).

burial conditions, rather than decomposed or replaced by secondaryminerals, wood material is preserved. Mummified wood is subjectedto a much different diagenetic pathway than petrified wood. For exam-ple, in the petrified forests of Yellowstone National Park (USA), silicifiedwoods are abundant in regions downwind from volcanoes that producecopious quantities of silicic volcanic ash (Ammons et al., 1987). Silicadissolves into groundwater, which infiltrates wood cells after burial,allowing permineralization to occur (Chadwick and Yamamoto,1984; Ballhaus et al., 2012). Wood silicification may occur in as quicklyas 10—100 years in the right conditions, such as Si-rich hot springs

Page 2: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

20 B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

(Akahane et al., 2004). However, in anoxic burial conditions that do notcontain enough soluble Si to permineralize thewood, it may remain in awoody state and become mummified.

Mummified wood may be used for paleoclimatic reconstructions ofancient climates on annual to subannual scale resolution, because stableisotopes of carbon and oxygen from tree-ring cellulose have beenshown to record paleoclimatic conditions (e.g., temperature of conden-sation, relative humidity, and amount of sunlight received) (McCarrolland Loader, 2004; Poole and van Bergen, 2006; Jahren and Sternberg,2008; Richter et al., 2008; Csank et al., 2011, 2013). However, mummi-fied wood is physically and chemically altered with respect to modernwood. Different wood constituents (i.e., α-cellulose, hemicellulose,lignin) are degraded at different rates (Rutherford et al., 2005), and somummified wood has an altered proportion of constituents than theoriginal wood. Lignin degrades more slowly than holocellulose, somummified wood commonly has a high concentration of lignin and lit-tle or no holocellulose (Staccioli et al., 2014). However, it has beenshown that some cellulose may exist in this material, allowing forpaleoenvironmental analysis of bulk cellulose (Wolfe et al., 2012),which is white in color, and fibrous in texture. Because mummifiedwoods are commonly low in cellulose content, their color is typicallydarker than modern wood of the same species.

Alpha-cellulose (α-cellulose) has been traditionally defined as theportion of holocellulose (α-cellulose and hemicellulose) that remainsinsoluble in a solution of 17.5% NaOH. Although this NaOH concentra-tion has been recommended for use in the pulp and paper industrysince the early 20th century for use with modern wood, it may not beappropriate for paleoclimatic studies of mummified wood. Hemicellu-loses are typically removed from wood in the laboratory before stableisotope analysis because they may contain exchangeable carbonyl oxy-gen, causing error in the signal (Wright, 2008). However, it is possiblethat one could modify or remove the NaOH and still produce viable α-cellulose. Due to the extreme age of mummified wood, the structurallyweaker hemicelluloses are likely to have been decomposed into sugarmonomers (e.g., glucose, galactose,mannose) via groundwater hydroly-sis (Rissanen et al., 2014), slowly leaching out of the material. Theabsence of hemicellulosewould negate the need for aNaOH step. If rem-nant hemicellulose remains, it may be possible to remove it with a 1–5%concentration of NaOH (Wolfe, pers. comm., 2013).Most peer-reviewedpapers on mummified wood cellulose extraction using NaOH do notspecify the concentration (Richter et al., 2008;Wolfe et al., 2012). How-ever, we could not replicate these results, finding that all holocellulosedissolved in a 17.5% NaOH solution. Due to this problem, we tested sev-eral modifications of standard cellulose extraction procedures to opti-mize a procedure to provide a consistent yield of purified α-cellulosefree of mineral contaminants for a tree-ring study of the early Eoceneat annual- to subannual-resolution (Hook et al., 2014).

1.1. Site description and materials

Mummified wood was recovered from kimberlites in the process ofdiamond mining operations near Lac de Gras, Northwest Territories,Canada, (Fig. 1). The Lac de Gras kimberlites were the result of sudden,violent eruptions of volatile-rich magma that chilled instantaneouslythrough adiabatic expansion of gases (Wilson and Head, 2007). Diavikpipes A-154 and A-418 (both located at 64° 29′ 46′ N, 110° 16′ 24″W) and Ekati Panda kimberlite pipe (64° 42′ 49″ N, 110° 37′ 10″W) were emplaced through this bedrock at various intervals duringthe Paleocene/Eocene transition [55.5, 55.2, and 53.3 (±0.7)Ma respec-tively] (Graham et al., 1999; Heaman et al., 2003; Creaser et al., 2004).After ejecting materials (e.g., including bedrock, trees, etc.) high intothe air, most of thematerials were deposited along the rim of the crater,while some woods became buried in resedimented volcaniclasticbackfall debris (Stasiuk et al., 1999). Following emplacement, the Lacde Gras region remained a swampy lacustrine environment until ca.45 Ma, exhibiting no orogenic alteration or tectonic metamorphism

within the Canadian Shield (Padgham and Fyson, 1992; Stasiuk et al.,2006). Kimberlite eruptions were dated using rubidium/strontium iso-chron radiometric techniques (Heaman et al., 2003; Creaser et al.,2004), allowing the mummified wood to be dated indirectly by theage of the enclosing kimberlite (Wolfe et al., 2012).

Most of the samples of mummified wood were small fragmentsranging from 3 to 10 cm3 in diameter, but large logs (N50 cm3) werealso found in Ekati Panda and Diavik A-418 kimberlites. Morphogeneraof samples were identified as Taxodioxylon Hartig 1848, and PiceoxylonGothan 1905 based on wood-anatomical characteristics (Philippe andBamford, 2008) (Fig. 2). Samples of Taxodioxylon from Diavik A-154 orA-418 did not contain cellulose; the polysaccharides had been preferen-tially removed, resulting in fragile lignin “skeletons” of the wood. How-ever, some Piceoxylon samples from Ekati Panda pipe did containremnant cellulose at low yield (1–5%). Other samples of Taxodioxylonfrom the same kimberlite pipe did not contain cellulose, suggestinginhomogeneous wood preservation within kimberlites. Other speciesof mummified wood containing cellulose from the Panda pipe havebeen previously described, including Taxodioxylon metasequoia Mikiex. Hu & Chang 1948 (Wolfe et al., 2012) and Sequioxylon canadenseBlokhina (Staccioli et al., 2014).

For stable isotopic analysis of tree rings, it is important to isolate asingle wood constituent (i.e., cellulose, lignin) for analysis becauseeach constituent has a unique isotopic signature, and the proportionof each varies within and between tree rings (Barbour et al., 2001;McCarroll and Loader, 2004; Leavitt, 2010). Therefore, a change in con-stituent ratio could be misinterpreted as a paleoenvironmental signal.This is especially true in mummified wood, where diagenetic degrada-tion of cellulose leaves behind a lignin-rich material (Staccioli et al.,2014). Incomplete lignin removal is a concern in α-cellulose analysisbecause lignin is about 3‰ lighter in δ13C, and 13–15‰ lighter in δ18O,than α-cellulose (Sheu and Chiu, 1995; Barbour et al., 2001), whichcould bias the signal. Different researchers do not always employ thesame methods of cellulose extraction, which may limit comparabilityamong laboratories if the extractionmethods significantly affect the iso-topic signal. We tested variants of two commonly used extraction tech-niques (i.e., Brendel and Jayme-Wise) to determine if there were anydifferences among the treatments resulting in differences in stable iso-tope (δ18O and δ13C) compositions of mummified α-cellulose. At ca.53.3 Ma (Creaser et al., 2004), material from Ekati Panda kimberlite isamong oldest verified α-cellulose found to date (see also Wolfe et al.,2012; Staccioli et al., 2014). This study corroborates the findings ofthose studies, and goes a step further to compare cellulose andwood ex-tracts from several variations of extraction methods, using AttenuatedTotal Reflection Fourier Transform Infrared (ATR-FTIR) spectroscopyto investigate the molecular structure of the extracts to determinewhether lignins and hemicelluloses had been removed. Due to the im-mense age of this mummified cellulose, it is important to characterizeits purity before making any paleoclimatic interpretations from stableisotope data. Isotopic errors may result from the presence of mineralcontaminants (e.g., iron oxides; Richter et al., 2008) or thermal alter-ation (Schleser et al., 1999). To determine whether the treatments af-fected stable isotope data, we compared replicates of the samesamples to see if any significant differences existed in stable isotope ra-tios (δ13C and δ18O) among the treatments.

2. Methods

Vitrinite reflectance was used to assess the degree of thermal al-teration in raw samples of mummified wood (3.1). Several variantsof commonly used cellulose extraction methods were performed[i.e., Brendel (Brendel et al., 2000) and Jayme-Wise (Leavitt andDanzer, 1993; Loader et al., 1997)]. We tested different concentrationsof sodium hydroxide (NaOH: 2.5%, 5%, and 17.5%) with the goal of re-moving hemicelluloses from holocellulose, while minimizing sampleloss during chemical processing (Section 3.2). After processing, several

Page 3: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

Ekati Panda

10 km

Kimberlite minesLakes

Enlarged Area

Diavik A-154and A-418Lac de Gras

Fig. 1. Lac deGras kimberlite locations. The study area (white square), near Lac deGras inNorthwest Territories, Canada (insetmap (upper right corner) source data: http://www.vecteezy.com/map-vector/5930-north-america-map-vector; enlarged map source data: Google 2014).

21B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

methods were used to assess sample purity and similarity to moderncellulose. Scanning electron microscopy (SEM) was used to visualizemummified cellulose extracts. Energy-Dispersive X-ray Spectroscopy(EDX) was used to identify elements present in the sample, includingany possible contaminants (i.e., Fe or Mn oxides). We used AttenuatedTotal Reflection Fourier Transform Infrared (ATR-FTIR) spectroscopyto evaluate the molecular bonds of the mummified cellulose, for com-parison with a reference spectrum of modern cellulose (Rutherford

Fig. 2. Mummified wood hosted in Lac de Gras kimberlites. Cross-section of Piceoxylonmummified wood from Ekati Panda kimberlite (ca. 53.3 Ma) containing cellulose. Scalebar = 30 mm.

et al., 2005) (Section 3.3). If the mummified wood sample matchesmodern cellulosewemay conclude that it has not undergone significantalteration, and is thus suitable for paleoclimatic reconstructions. Wesampled from two sections of Piceoxylon wood from Ekati Panda, induplicate, for a total of (n = 4) samples per extraction method(Section 3.2), and unextracted mummified wood. Samples were ana-lyzed for stable isotope ratios of δ13C and δ18O using continuous flowstable isotope mass spectrometry (Section 3.4).

2.1. Vitrinite reflectance

Wood samples from Ekati Panda (n = 3), Diavik A-154 (n = 2),and Diavik A-418 (n = 1) were measured with vitrinite reflectanceto evaluate post-burial thermal alteration. We followed standardprocedures for organic petrology based on Mackowsky (1982). Sam-ples were mounted in a one-inch mold and polished. Random reflec-tance (% Ro) was measured using a Leitz reflected light microscope[50-X oil immersion objective and white (halogen, 546 nm) andfluorescent (HBO 100W) light sources]. Random percent reflectance(% Ro) in oil (Zeiss Immersol 518 F for fluorescence microscopy) wasmeasured with a Leitz MPV II–COMBI photometer system (n = 5 to50) on huminite macerals (i.e., textinite, corpohuminite) of mummi-fied wood. Vitrinite equivalent is calculated using the measured re-flectance of primary bitumen (Jacob, 1989). Estimation of burialheating peak temperature (Tpeak) is based on the relationship be-tween temperature and % Ro, following Eq. (1), after Barker andPawlewicz (1994):

Tpeak ¼ ln %Roð Þ þ 1:68½ �=0:0124: ð1Þ

Page 4: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

22 B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

2.2. Extraction treatments

Two samples of air-dried Piceoxylon wood from Ekati Panda pipeencompassing 4–5 tree rings each (approximately 15 cm3 in volume)were used. The wood was ground using a mortar and pestle (1–20 μmparticle size) within a glass vial to ensure complete homogenization.Wood contains significant intra- and inter-annual isotopic variation(Barbour et al., 2002; Leavitt, 2010), which may be misinterpreted asvariation among the extraction techniques if thematerial is incomplete-ly homogenized. Therefore, we homogenized the material into a finepowder before dividing samples for processing using different extrac-tion methods (Table 1). However, loss of wood powder proved to be aproblem in the JW method, because the borosilicate fiber poucheshave a 250 μm pore size (see Supplementary materials). One maygrind the wood with a Wiley mill, if available, and extract the 40–60 mesh portion (0.25–0.42 mm) to avoid this problem (Staccioliet al., 2014).

Cellulose extraction treatments were applied to powder from eachsample vial of homogenized wood (Table 1). Extractive-Free Wood(EF)was produced using a soxhlet apparatus, using toluene/ethanol sol-vent extraction followed by boiling the sample in water to removewaxes and resins, as in the first steps of the Jayme-Wise (JW) method.The JW treatment delignifies the wood using acidified sodium chlorite,after which we added a 2.5% NaOH step. Standard Brendel 1 (SB1)uses a 30-minute heated (120 °C) acid hydrolysis step (nitric/aceticacids) to delignify and remove extractives in one step. Standard Brendel2 (SB2) uses a 60-minute acid hydrolysis (Brookman and Whittaker,2012). Modified Brendel 1 (MB1) follows SB2 with the addition of a2.5% NaOH solution, Modified Brendel 2 (MB2) follows SB2 with a 5%NaOH solution, and Modified Brendel 3 (MB3) follows SB2 with a17.5% NaOH solution. Two replicates per sample vial were analyzedfor stable isotope ratios of δ13C and δ18O for all treatments (n = 4 pertreatment), and a control group of unextracted mummified woodfrom the sample vials (see Supplementary materials for detailedprotocols).

2.3. Cellulose verification and purity assessment

A Zeiss Supra VP55 Scanning Electron Microscope (SEM) (Universityof Toronto Earth Sciences Department) was used to visually inspect theextracts at high resolution (1.7 nm at 1 kV) using a thermal field emissiontype Schottky emitter. Low voltage (b5 kV) was used to avoid chargingand damage to the sample, and produce a high-magnification image ofthe cellulose extracts. Sputter coating was not used, so that elementalcomposition could bemeasured. An INCA EnergySEM 350 Energy Disper-sive X-ray Microanalysis system (EDX) was used to characterize the ele-mental composition of the extracts and test for mineral contaminants.This system is combined with the SEM, and uses an INCAx-act Analyticalsilicon drift detector 133 eV (nominally 10 mm2) with a Super Atmo-spheric Thin Window (SATW) for detection of B-Pu, with a liquid N freeX-Stream X-ray acquisition system and detector control.

Table 1Extraction treatments used with Piceoxylon mummified wood. Treatment name, treat-ment code, and short description of eachmethod shown. For further details on each of the-se methods, see Supplementary materials.

Treatment name Treatmentcode

Short description

Control CTRL Mummified wood, no chemical treatmentExtractive-Free Wood EF Toluene/ethanol extraction, water boilJayme-Wise JW EF treatment + NaClO2 and NaOH (2.5%)Standard Brendel 1 SB1 Nitric/acetic acid hydrolysis (30 min), 120 °CStandard Brendel 2 SB2 Nitric/acetic acid hydrolysis (60 min), 120 °CModified Brendel 1 MB1 SB2 treatment + NaOH (2.5%)Modified Brendel 2 MB2 SB2 treatment + NaOH (5%)Modified Brendel 3 MB3 SB2 treatment + NaOH (17.5%)

Attenuated Total Reflection Fourier Transform Infrared (ATR-FTIR)spectra were collected from unextracted mummified wood and treat-ments (Section 3.2). We used a Perkin Elmer Spectrum One FTIR withPerkin Elmer Universal ATR Sampling accessory, equipped with a dia-mond crystal at ANALEST lab (University of Toronto). Cellulose andwood samples (0.5–1 mg) were analyzed, averaging 16 scans at 1 cm−1

intervals. All spectra were normalized for comparison of relative absor-bance peaks using Spectrum forWindows (V 6.3.5) FTIR control software.Wavenumbers (cm−1) were analyzed in the mid-infrared spectrum(550–4000 cm−1). Relative absorbance peaks of three scans per treat-ment were averaged. For comparison, reference spectra of commerciallypurified cellulose and pine wood were used (Rutherford et al., 2005).We note that ATR-FTIR peaks are slightly shifted (4–5 cm−1 lower)when compared to those of transmission FTIR (i.e., the KBr method) dueto ATR correction algorithm applied to the spectra (Griffiths and deHaseth, 2007). In this paper we use cellulose wavenumbers from trans-mission FTIR for comparability with other studies.

2.4. Stable isotope analysis

Stable isotope ratios of δ18O and δ13Cweremeasured simultaneouslyfrom carbonmonoxide (CO) gas produced via low-temperature pyroly-sis. Two Piceoxylon samples, processed using various cellulose purifica-tion treatments (described in Section 3.2) were measured in duplicatefor stable isotope analysis. Mass spectrometry was conducted by theStable Isotope Laboratory at the University of Maryland. Samples wereweighed into silver capsules (350 ± 35 μg per sample), and shippedto the independent laboratory, where they were desiccated underultra high purity helium (UHP He, 99.999%) at room temperature over-night in a sealed autosampler. This method has been found to desiccatecellulose better than heating under atmosphere (Evans, pers. comm.,2014). Cellulose was pyrolyzed to CO at 1080 °C over glassy carbonwithin a stream of UHP He. Sample gas was passed through traps forCO2 and H2O, and CO separated from N2 by gas chromatography, beforeisotopic analysis by continuous flow stable isotope mass spectrometry(CF-SIMS) (Werner et al., 1996). Molar masses of 28, 29, and 30 corre-sponding to isotopes of 12C16O, 13C16O, and 12C18O, respectively, weremeasured by CF-SIMS. Isotopic data were corrected for runtime drift,amplitude dependence and scaling usingwidely separatedworking cel-lulose isotopic standards calibrated to international reference materials(Vienna Pee Dee Belemnite, VPDB for δ13C, and Standard Mean OceanWater, SMOW, for δ18O) (Evans, pers. comm. 2014). Both isotope ratiosare calculated under the formula in Eq. (2):

δ ¼ Rsample=Rstandard

� �–1

� ��1000 ð2Þ

where δ is the isotope ratio relative to the standard, R is the molar ratioof the heavier isotope to the lighter isotope, of the sample and standard(denoted by subscripts). The overall precisions for the corrected data,based on replicate standard analyses, are 0.14‰ for δ13C and 0.23‰for δ18O. Groups (treatment, sample) were tested for statistical differ-ences in δ13C, δ18O, and CO gas amplitude using two-way ANOVA andTukey's Honest Significant Difference (HSD) tests.

3. Results

3.1. Vitrinite reflectance

Vitrinite reflectance was 0.23—0.39% Ro suggesting low burialtemperatures (Tpeak = 60 °C) (Barker and Pawlewicz, 1994) amongmummified wood samples from the three Lac de Gras kimberlites inthis study (Table 2). Therefore, these samples are considered huminitebased on % Ro values b0.5 (International Committee for Coal andOrganic Petrology, ICCP, 1998; Sýkorová et al., 2005). Cellulose beginsto degrade at temperatures N180 °C in laboratory conditions, and time

Page 5: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

Table 3Piceoxylon cellulose mean percent yield, mean CO amplitude, and stable isotope ratios(δ13C and δ18O) by extraction treatment. Treatment codes: Control (CTRL), Extractive-FreeWood (EF), Standard Brendel 1 (SB1), Standard Brendel 2 (SB2), Modified Brendel 1(MB1), Modified Brendel 2 (MB2), Modified Brendel 3 (MB3), and Modified Jayme-Wise(JW) (see Section 3 and Supplementary materials for details on thesemethods). Cellulosemean percent yield (±95% confidence interval, CI; n = 6), mean amplitude of isotopemeasurement peaks (±95% CI; n = 4), δ13C mean (±95% CI; n = 4), and δ18O mean(±95% CI; n=4). Lowercase italic letters show results of TukeyHSD tests, grouped by col-umn. *Significant sample loss occurred in some replicates of the JWmethod in this studydue to the finely-ground material escaping from pores of the borosilicate fiber pouches(ANKOM technologies), therefore yield data are incorrect.

Treatment Yield (%) mean(95% CI)

Amplitude mean(95% CI)

δ13C mean(95% CI) ‰

δ18O mean(95% CI) ‰

CTRL N/A 2.3 (0.3)c −23.9 (0.3)b 9.0 (0.2)bEF 27.5 (2.6)a 2.1 (0.3)c −24.0 (1.0)b 7.5 (0.9)cSB1 5.0 (2.5)b 6.2 (0.3)b −20.8 (0.3)a 24.0 (0.5)aSB2 3.6 (1.4)bc 6.2 (0.5)b −20.7 (0.3)a 24.6 (0.4)aMB1 1.8 (1.1)c 5.9 (0.5)b −21.1 (0.4)a 24.1 (0.7)aMB2 1.2 (0.6)c 6.4 (0.2)b −20.8 (0.4)a 25.3 (0.6)aMB3 0 N/A N/A N/AJW* 1.5 (1.0)c 8.5 (0.5)a −20.7 (0.3)a 24.7 (0.5)a

23B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

is a factor as well as temperature; cellulose degradation progresses fasterat higher temperatures (Rutherford et al., 2005; Rutherford andWershaw, 2008). However, many samples with Tpeak b 60 °C did notyield cellulose, including a sample of fusinized gymnosperm wood thatcould not be assigned a morphogenus because of excessive compaction.

3.2. Extraction treatments

Different extraction treatments were tested on samples of kimberlite-hosted mummified wood from all three kimberlites. The only wood thatcontained cellulose was Piceoxylon from the upper 100 m of Ekati Pandapipe (Table 2). The outermost 8–10 tree rings presumed to be in contactwith the kimberlite did not contain any cellulose, but cellulose was pres-ent in all subsequent interior tree rings. Piceoxylon sampleswere lighter incolor than other species that did not contain cellulose (i.e. Taxodioxylon).Aside from MB3, which completely dissolved the holocellulose, treat-ments SB1, SB2, MB1, MB2, and JW produced cellulose. Average percentcellulose yield had a negative association with increased strength ofNaOH among the treatments, ranging from 5.0 ± 2.5% (SB1) to 1.2 ±0.6% (MB2) to (Table 3). Yield measurements for JW are lowered due tosample loss from the borosilicate fiber pouches during the water-boilingstep of the extraction. The color of the extracts ranged from very dark(in EF, resembling CTRL group), to light tan (SB1), off-white (SB2), andwhite (MB1, MB2, and JW).

3.3. Cellulose verification and purity assessment

SEM images of the cellulose extracts appeared fibrous or papery,matching the description of modern cellulose that has been finely pow-dered. No Fe oxides or other mineral contaminants were detected usingEDX, only strong peaks of C and O (n.b., EDX is not capable of detectingelements of atomic mass less than 5, therefore it cannot detect H). ATR-FTIR spectra of unextracted Taxodioxylonwood (fromDiavik A-154 kim-berlite), which did not contain cellulose, showed non-cellulosic peaks(i.e., 1745, 1700, 1600, 1512, and 1265 cm−1; solid lines, wavenumbersin boxes in Fig. 3). Unextracted Piceoxylonwood (from Ekati Panda kim-berlite) showed non-cellulosic and cellulosic peaks (dotted lines,wavenumbers without boxes in Fig. 3), especially the four main diag-nostic peaks for cellulose (i.e., 1029, 1058, 1105, and 1161 cm−1; i.e.,wavenumbers with asterisks in Fig. 3, Rutherford et al., 2005). Mummi-fied cellulose extracts from various treatments showed few differencesin ATR-FTIR spectra, and all methods successfully removed non-cellulosic peaks, suggesting all methods successfully extracted purifiedα-cellulose as shown by the well-defined peaks in the dominant cellu-lose region (wavenumbers with asterisks in Fig. 4). However, a moder-ate peak existed at 1720 cm−1 in the Standard and Modified Brendelmethods (SB1, SB2,MB1,MB2),whichwas not present in themodern cel-lulose FTIR reference spectrum (Rutherford et al., 2005). This peak wasvery slight in the JW samples (Fig. 4). The dominant cellulose peaks ap-peared differently than the standard in the JW treatment, with someATR-FTIR peaks (e.g., 1058 and 1105 cm−1) being represented as shoul-ders rather than isolated peaks as in Brendel methods (Fig. 4). There

Table 2Vitrinite reflectance (% Ro) ofmummifiedwood from Lac de Gras kimberlites. SamplemorphogRo (standard deviation) of textinitemacerals, % Romean (1 σ) of corpohuminite macerals, estimabsence of cellulose, and general appearance (color, compaction in earlywood (EW) or latewo

Morphogenus, kimberlite mine Textinite % Romean (σ)

Corpohuminite %Ro mean (σ)

Piceoxylon, Ekati Panda 0.23 (0.01) 0.26 (0.01)Piceoxylon, Ekati Panda 0.26 (0.01) 0.29 (0.01)Taxodioxylon, Ekati Panda 0.29 (0.02) 0.39 (0.01)Taxodioxylon, Diavik A-418 0.28 (0.01) 0.32 (0.01)Taxodioxylon, Diavik A-154 0.28 (0.02) 0.34 (0.01)Unknown Gymnosperm, Diavik (unknown) 0.24 (0.01) 0.26 (0.00)

were no major differences in ATR-FTIR peak wavelengths of cellulosesamples extracted (using MB1) along a tree-ring transect (Fig. 5).

3.4. Stable isotope analysis

Stable isotope ratios of cellulose extracts (SB1, SB2, MB1, MB2, JW)were significantly heavier than CTRL or EF groups (Two-way ANOVA:δ18O, F = 340.6, p b 0.0001; δ13C, F = 18.9, p b 0.001; Tukey's HSDtests; Fig. 6, Table 3). Mean isotope ratio of all cellulose extracts was24.5‰ δ18O, and −20.8‰ δ13C. δ18O was significantly different amongtreatments (Type 1 SS: F=736.9, p b 0.0001), but not between samples(Type 1 SS: F = 4.5, p = 0.051). δ13C was significantly different amongtreatments (Type 1 SS: F = 39.8, p b 0.0001) and between samples(Type 1 SS: F = 6.4, p = 0.24). Tukey's HSD tests confirmed that EFwood was significantly lighter in δ18O, but not in δ13C, when comparedto the CTRL wood (Fig. 6). These isotopic differences emphasize whyisolation of a single wood constituent (e.g., cellulose, lignin) is necessaryfor reconstructing accurate paleoclimatic signals. Stable isotope valuesof δ18O and δ13C were not significantly different among the celluloseextraction treatments (SB1, SB2, MB1, MB2, JW), suggesting that allmethods successfully produced fully delignified cellulose (Fig. 6). COgas amplitude was significantly different among treatments (F =117.3, p b 0.0001) but not samples (F = 0.024, p = 0.88). Amplitudewas higher in JW cellulose than all Brendel methods (which were sim-ilar). All cellulose extracts had higher amplitude than the lignin-richCTRL and EF treatments.

4. Discussion

This study confirmed the existence of the oldest verified samples ofPiceoxylon mummified wood cellulose found, to date at 53.3 ± 0.6 Ma,

enus (Philippe and Bamford, 2008) and kimberlite mine, mean vitrinite reflectance in oil %ated peak burial temperature (Tpeak after Barker and Pawlewicz, 1994), and the presence/od (LW)) are shown.

Tpeak (°C) Cellulosepresent–absent

Color, physical appearance

17–27 Present Light brown, compacted in EW27–37 Present Light brown, compacted in EW37–60 Absent Dark brown, compacted in EW33–44 Absent Dark brown, compacted in EW33–48 Absent Dark gray, compacted in EW20–27 Absent Black, fusinized, extremely compacted

in EW and LW

Page 6: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

1800 1600 1400 1200 1000 800 600

1640

1430

1370

1315 1205

1161

*11

05*

1058

*10

29*

900 61

566

3

1265

1512

1600

1700

1745

Rel

ativ

e A

bso

rpti

on

Modern Cellulose

ModernPinusWood

Piceoxylon(with cellulose)

Taxodioxylon(no cellulose)

Wavenumber(cm-1)

Fig. 3. ATR-FTIR spectral comparison of mummified woods with and without cellulosewith modern Pinus wood and cellulose. Lac de Gras mummified woods contain manyATR-FTIR peaks diagnostic of modern gymnospermwood, including those relating to cel-lulose (dotted gray lines) and other constituents (solid black lines). Dominant ATR-FTIRbands associated with cellulose (*1029–1161) are absent in Taxodioxylon wood fromDiavik pipe A-154 (ca. 55.5 Ma), but present and low in relative absorption in Piceoxylonwood from Ekati Panda pipe (ca. 53.3 Ma). Reference FTIR spectra for cellulose and Pinuswood after the Figure B1 in Rutherford et al. (2005). 1800 1600 1400 1200 1000 800 600

Rel

ativ

e A

bso

rpti

on

ModernCellulose

JW

MB2

MB1

SB2

SB1

EF

CTRL

Wavenumber (cm-1)

1640

1430

1370

1315

1205

1161

*11

05*

1058

*10

29*

900 61

566

3

1265

1512

1600

1720

Fig. 4. ATR-FTIR spectral comparison of experimental treatments of mummified woodwith modern cellulose. Treatment codes (top to bottom): Control (CTRL), Extractive-Free Wood (EF), Standard Brendel 1 (SB1), Standard Brendel 2 (SB2), Modified Brendel1 (MB1), Modified Brendel 2 (MB2), and Modified Jayme-Wise (JW). Diagnostic peaksfor identification of cellulose shown by dotted gray lines,with numbers of peaks above. Di-agnostic peaks for identification of non-cellulosic wood constituents shown by solid blacklines, with numbers in black boxes above. Reference FTIR spectrum for cellulose afterFigure B1 in Rutherford et al. (2005).

24 B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

from the Ekati Panda kimberlite (see also Wolfe et al., 2012; Staccioliet al., 2014). However, as in Staccioli et al. (2014), not all mummifiedwood samples in this study contained cellulose. Diagenetic conditionsare not homogeneouswithin a kimberlite, as heat and pressure general-ly increase with depth (Stasiuk et al., 1999, 2005). Other samples ofmummified wood found in Ekati Panda and other nearby kimberlite di-amondmines (i.e., Diavik A-154 and A-418) did not contain cellulose. Inthese samples, preferential cellulose decomposition left primordial lig-nin and wood extractives (e.g., resins, waxes). Thermal alteration of allsamples was low (b60 °C), even in samples appearing highly fusinized,similar to charcoal (Table 2) (Barker and Pawlewicz, 1994). AttenuatedTotal Reflection Fourier Transform Infrared (ATR-FTIR) spectra ofmummified wood containing cellulose (Piceoxylon from Ekati PandaPipe) showed many ATR-FTIR peaks diagnostic of lignin (e.g., 1600,1512, and 1265 cm−1) and α-cellulose (e.g., 1029, 1058, 1105, and1161 cm−1), whereas a sample not containing cellulose (Taxodioxylonfrom Diavik Pipe A-154) contained peaks from lignin but not α-cellulose (Fig. 3). Thus, ATR-FTIR may be used to decide which samplesof mummified wood are suitable for α-cellulose extraction, by lookingfor the dominant α-cellulose peaks (wavenumbers with asterisks inFig. 3) in unextracted wood. The color of the powdered wood mayalso be an indication of the presence of cellulose; in this study the sam-ples containing cellulose were a lighter brown color, whereas those de-void of cellulose were darker (Table 2).

Except for treatment MB3, which completely dissolved allholocellulose during the 17.5% NaOH step, all treatments producedsome cellulose at low yield. Studies of mummified (Richter et al.,2008; Staccioli et al., 2014) and subfossil (Stambaugh and Guyette,2009; Savard et al., 2012) wood in other diagenetic environmentshave also shown that subfossil cellulose yields are typically low, Inmod-ern wood, hemicelluloses can easily be leached via hydrolysis underpressurized conditions under heat in a matter of hours (Rissanenet al., 2014), thus after millions of years, most hemicelluloses havebeen removed naturally via hydrolysis (Staccioli et al., 2014). In thisstudy, cellulose yield and color varied by extraction treatment, withthe greatest amounts in SB1 (5.0 ± 2.5%), which was light tan in color,

suggesting incomplete delignification; however, the ATR-FTIR andisotopic data of SB1 closely matched that of the other Brendel extracts(SB2, MB1, MB2) and lacked diagnostic FTIR bands for lignin (e.g., 1600,1510 cm−1, Richard et al., 2014) contradicting this idea. Greater sampleloss occurred in SB2 than in SB1, presumably due to a longer acid hydro-lysis step, but the color in SB2 was whiter than SB1, closer to the color ofmodern cellulose. Modified Brendel methods (i.e., including NaOH: MB1,MB2) were whiter in color than SBmethods, but were also lower in yieldthan SB methods; yield was negatively associated with NaOH % solution(Table 3). Energy-Dispersive X-ray analysis (EDX) detected only strongC and O peaks (H is not detectable using this method), and no mineraloxides were detected. The Brendel extracts appeared fibrous and paperyin texture using SEM. JW extracts were white, and appeared more poly-merized compared to Brendel extracts. ATR-FTIR spectra of all celluloseextracts matched the reference spectrum of cellulose, giving further evi-dence that thermal or other major diagenetic alteration did not occur inthismaterial (Fig. 4). FTIR spectra of thermally altered cellulose and ligninshow significant spectral distortions and shifts to higher wavenumbers,and these shifts occur at a faster rate when experimentally exposed tohigher temperatures (Rutherford et al., 2005).

After extraction, ATR-FTIRmay be used to determinewhether resins,lignins, and hemicelluloses have been removed, leaving purified α-cellulose (Richard et al., 2014). Although many hemicellulose ATR-FTIR peaks overlap with those of α-cellulose, a peak near 1725 cm−1

in galactoglucomannan (in gymnosperms), and one near 1735–1731 cm−1 in xylan (in angiosperms), are not present in α-cellulose(Richard et al., 2014). Therefore, the presence of a peak near

Page 7: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

TR 54

TR 44

TR 34

TR 28

TR 24

TR 20

TR 16

TR 12

Rel

ativ

e A

bso

rpti

on

1800 1600 1400 1200 1000 800 600

Wavenumber (cm-1)

1640 14

30 1370

1315

1205 11

61*

1105

* 1058

*10

29*

900

615

663

1720

Fig. 5. ATR-FTIR spectral comparison of cellulose along a tree-ring transect. ModifiedBrendel 1 was used for all samples here. Diagnostic peaks for identification of celluloseare shown by dotted gray lines (same lines as in Fig. 4). Top spectrum of “TR 12” is of cel-lulose extracted from the 12th tree ring from outer edge (bark side). Successive spectrafollow this convention.

-30

-25

-20

-15

-10

-5

0

0

5

10

15

20

25

30

EF SB1 SB2 MB1 MB2 JW

Treatments

aa a a ab b

δ13C

‰ (

VP

BD

)δ18

O ‰

(S

MO

W)

aa a a a

b c

A

B

CTRL

EF SB1 SB2 MB1 MB2 JW

Treatments

CTRL

Fig. 6. Stable isotopes of oxygen (δ18O, upper graph), and carbon (δ13C, lower graph).Treatment codes: Control (CTRL), Extractive-Free Wood (EF), Standard Brendel 1 (SB1),Standard Brendel 2 (SB2), Modified Brendel 1 (MB1), Modified Brendel 2 (MB2), andModified Jayme-Wise (JW). Data points (horizontal bars) are means of each group±95% confidence interval (n = 4). Lowercase letters are the results of Tukey's HSD testscarried out on each isotope.

25B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

1720 cm−1 in the Standard andModified Brendelmethodsmay indicatethe presence of hemicelluloses as suggested by Richter et al. (2008),who also used amodification of the Brendel method using NaOH. Alter-natively, the peak may have been introduced via chemical alteration(acetylation, acetonation, oxidation) during the cellulose extractionprocess. Cellulose acetate has a prominent FTIR peak near 1730–1745 cm−1 (Jebrane et al., 2011). Therefore, slight cellulose acetylation,due to the use of strong acetic acid (80%) solution,may explain the peakat 1720 cm−1 (Brendel et al., 2000). A peak at 1720 cm−1was attributedto cellulose acetylation by Anchukaitis et al. (2008) who comparedBrendel and Jayme-Wise methods using modern wood. Like thisstudy, Anchukaitis et al. (2008) found that the 1720 cm−1 peak existedin Brendel cellulose, but not in Jayme-Wise cellulose. However, acetyla-tion has not been found to significantly affect δ13C or δ18Omeasurementsof tree-ring cellulose (Anchukaitis et al., 2008). However, Gaudinski et al.(2005) found that the Brendel method added C and N and left residues ofextractives (i.e., waxes, resins) in the cellulose. The strong nitric acid(69%) solution used in the Brendel method may introduce carboxylicacid groups (C = O), contributing to the 1720 cm−1 peak (Pradhan andSandle, 1999). Another possible factor of the 1720 cm−1 peak in theBrendel method is the incomplete removal of acetone, which may causecellulose acetonation (Brookman andWhittaker, 2012).

Stable isotope data revealed that the unextracted mummified wood(CTRL) and Extractive-Free Wood (EF) were both significantly lessenriched than all of the cellulose extraction treatments studied (SB1,SB2, MB1, MB2, JW). Extractives were –1.5‰ in δ18O, and +0.1‰ inδ13C relative to unextracted wood (CTRL). Most of the bulk materialwas dissolved during the delignification step. Cellulose extracts werenot significantly different from each other in terms of δ18O or δ13C,

according to Tukey's HSD tests (Fig. 6, Table 3). Themean of all cellulosetreatments (24.5‰ in δ18O,−20.8‰ in δ13C) was enrichedwith respectto EF by 17‰ in δ18O, and by 3.2‰ in δ13C, and enriched with respect toCTRL by 15.5‰ in δ18O, and by 3.0‰ in δ13C. Similar isotopic resultswere also obtained from SB1, which produced extracts that seemed tobe incompletely delignified based on color, to treatments using NaOH(MB1, MB2, JW; Table 3). These results suggest that any of the methodsstudied here would be adequate for paleoenvironmental analysis ofmummified wood α-cellulose. Additionally, studies using differentmethods should be comparable, as long as an assessment of α-cellulose purity is conducted, such as ATR-FTIR. CO gas amplitudesduring CF-SIMS were significantly higher in JW cellulose than Brendelcellulose treatments, possibly indicating a purer cellulose extractcontaining more oxygen in the JW treatment (Table 3). Cellulose(C6H10O5) has more oxygen than lignin [guaiacyl lignin (C10H12O3)and p-hydroxyphenyl lignin (C9H12O2) are common in gymnosperms],so about twice as much CO gas is formed from pyrolysis of cellulose asan equal mass of lignin. Therefore, amplitudes of CTRL and EF treat-ments were both significantly lower than all cellulose extracts, due tohigh lignin/cellulose ratio in the raw mummified wood.

Considering the above, it seems that the Brendelmethodmay: 1) in-troduce a peak related to acetylation, or 2) incompletely remove hemi-celluloses. Hemicellulose extraction procedures are necessary to fully

Page 8: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

26 B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

address this question. The fact that these factorsmay each contribute toan ATR-FTIR peak near 1720 cm−1 suggests that the Brendel methodshould not be used for the detection of hemicelluloses in holocellulose.Therefore, for future studies of mummified wood we suggest that theJW method be used to produce holocellulose, which should be testedfor the presence of hemicelluloses using ATR-FTIR. If none are detected,the NaOH step may be omitted. If hemicelluloses are detected, onemayexperimentwith low-percentage solutions of NaOH to remove hemicel-luloses, verifying the α-cellulose using ATR-FTIR. The exact solution ofNaOH suitable for each different sample of mummified may vary bystudy, as variability in groundwater and bedrock chemistry may leadto different diagenesis in the wood. One may test inner vs. outer treerings to test for diagenetic differences throughout the sample. If asmall percentage of remnant hemicellulose is found, and slight acetyla-tion not of concern, these may not significantly affect the stable isotopedata. If so, the SB1 treatment may be sufficient to quickly produce anisotopically similar cellulose extract with higher yield than morelabor-intensive methods (SB2, MB1, MB2, JW).

5. Conclusion

Although thermal alteration was low in mummified wood from allthree kimberlites studied here, only the Piceoxylon samples from EkatiPanda Pipe (dated at ca. 53.3 Ma) contained α-cellulose, suggestingthat other factors besides thermal alteration are involved in cellulosedegradation such as post-burial hydrolysis. In this study, the concentra-tion of NaOH used for removal of hemicelluloses in modern wood(17.5%) completely dissolved all holocellulose (method MB3), butlower concentrations of NaOH (2.5–5%) produced cellulose, albeitwith low yield (1–2%) (MB1, MB2). Standard Brendel methods thatdid not include NaOH (SB1, SB2) also produced cellulose very similarin ATR-FTIR as treatments including NaOH. Fourier Transform InfraredAnalysis (ATR-FTIR) confirmed that mummified cellulose matched a ref-erence spectrumofmodernα-cellulose (Rutherford et al., 2005), suggest-ing little if any degradation to remnant α-cellulose. No significantdifferences were observed between stable isotope values (δ13C or δ18O)of cellulose from the treatments tested, suggesting that all methods suc-cessfully extractedα-cellulosewithout significantly influencing the isoto-pic signal. Therefore, inter-study comparisons using versions of methodsdescribed here should be comparable. However, a FTIR peak near1720 cm−1 in the Brendel methods suggests that they should not beused to detect hemicelluloses. Hence, we suggest that the Jayme-Wisemethod should be used for testing for the presence of hemicellulose. Ifhemicelluloses are absent, any cellulose extraction method (SB1, SB2,MB1, MB2, JW) could be used to adequately delignify the wood, enablingreliable and repeatable stable isotope data collection frommummifiedα-cellulose.

Author contributions

B.A.H. designed study, developed protocols, collected and analyzeddata, wrote manuscript, J.H. edited manuscript, Z.G. edited manuscript,J.B. edited manuscript, M.A.R. helped develop protocols, J.R. analyzedsamples for vitrinite reflectance, edited manuscript, D.S. edited manu-script. Authors declare no competing financial interests.

Acknowledgments

Funding was provided by J. Halfar through a National Sciences andEngineering Research Council of Canada Discovery Grant number316003. We thank Ekati Diamond Mine and Diavik Diamond Mine(jointly owned by Dominion Diamond Corporation and Rio TintoGroup, when Panda Pipe samples were donated, Ekati Diamond Minewas owned by BHP Billiton), for access to sample materials; U. Fekl foraccess to equipment; R.E. Plummer, K.J. Steele, and M.N. Evans at theUniversity of Maryland Stable Isotope laboratory for conducting mass

spectrometry; A.P.Wolfe andW.E.Wright for helpful discussions on cel-lulose extraction techniques.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.chemgeo.2015.04.003.

References

Akahane, H., Furuno, T., Miyajima, H., Yoshikawa, T., Yamamoto, S., 2004. Rapid wood si-licification in hot spring water: an explanation of silicification of wood during theEarth's history. Sediment. Geol. 169, 219–228.

Ammons, R., Fritz, W.J., Ammons, R.B., Ammons, A., 1987. Cross-identification of ringstructures in Eocene trees (Sequoia magnifica) from the specimen ridge locality ofthe Yellowstone Fossil Forests. Palaeogeogr. Palaeoclimatol. Palaeoecol. 60, 97–108.

Anchukaitis, K.J., Evans, M.N., Lange, T., Smith, D.R., Leavitt, S.W., Schrag, D.P., 2008. Con-sequences of a rapid cellulose extraction technique for oxygen isotope and radiocar-bon analyses. Anal. Chem. 80, 2035–2041.

Ballhaus, C., Gee, C.T., Bockrath, C., Greef, K., Mansfeldt, T., Rhede, R., 2012. The silicifica-tion of trees in volcanic ash — an experimental study. Geochim. Cosmochim. Acta84, 62–74.

Barbour, M.M., Andrews, T.J., Farquhar, G.D., 2001. Correlations between oxygen isotoperatios of wood constituents of Quercus and Pinus samples from around the world.Australian J. Plant Physiol. 28, 335–348.

Barbour, M.M., Walcroft, A.S., Farquhar, G.D., 2002. Seasonal variation in δ13C and δ18O ofcellulose from growth rings of Pinus radiata. Plant Cell Environ. 25, 1483–1499.

Barker, C.E., Pawlewicz, M.J., 1994. Calculation of vitrinite reflectance from thermal histo-ries and peak temperatures. Chap. 14 in vitrinite reflectance as a maturity parameter.Mukhopadhyay, P.K, and Dow, W.G., Eds. ACS Symp. Ser. 570, 216–229.

Brendel, O., Ianetta, P.P.M., Stewart, D., 2000. A rapid and simple method to isolate purealpha-cellulose. Phytochem. Anal. 11, 7–10.

Brookman, T., Whittaker, T., 2012. Experimental assessment of the purity of α-celluloseproduced by variations of the Brendel method: implications for stable isotope(δ13C, δ18O) dendroclimatology. Geochem. Geophy. Geosys. 13. http://dx.doi.org/10.1029/2012GC004215 (Q0AI01).

Chadwick, A., Yamamoto, T., 1984. A paleoecological analysis of the petrified trees in thespecimen creek area of Yellowstone National Park, Montana, U.S.A. Palaeogeogr.Palaeoclimatol. Palaeoecol. 45, 39–48.

Creaser, R.A., Grütter, H., Carlson, J., Crawford, B., 2004. Macrocrystal phlogopite Rb–Srdates for the Ekati property kimberlites, Slave Province, Canada: evidence for multi-ple intrusive episodes in the Paleocene and Eocene. Lithos 76, 399–414.

Csank, A.Z., Fortier, D., Leavitt, S.W., 2013. Annually resolved temperature reconstructionsfrom a late Pliocene–early Pleistocene polar forest on Bylot Island, Canada.Palaeogeogr. Palaeoclimatol. Palaeoecol. 369, 313–322.

Csank, A.Z., Patterson, W.P., Eglington, B.M., Rybczynski, N., Basinger, J.F., 2011. Climatevariability in the Early Pliocene Arctic: annually resolved evidence from stable iso-tope values of sub-fossil wood, Ellesmere Island, Canada. Palaeogeogr. Palaeoclimatol.Palaeoecol. 308, 339–349.

Evans, M.N., 2014. Personal Communication. Department of Geology. University ofMaryland, College Park, Maryland, USA.

Gaudinski, J.B., Dawson, T.E., Quideau, S., Schuur, E.A.G., Roden, J.S., Trumbore, S.E.,Sandquist, D.R., Oh, S.-W., Wasylishen, R.E., 2005. Comparative analysis of cellulosepreparation techniques for use with 13C, 14C, and 18O isotopic measurements. Anal.Chem. 77, 7212–7224.

Graham, I., Burgess, J.L., Bryan, D., Ravenscroft, P.J., Thomas, E., Doyle, B.J., Hopkings, R.,Armstrong, K.A., 1999. Exploration history and geology of the Diavik kimberlites,Lac de Gras. Northwest Territories, Canada. Seventh International Kimberlite Confer-ence vol. 1, pp. 262–279.

Griffiths, P.R., de Haseth, J.A., 2007. Fourier Transform Infrared Spectrometry. 2nd ed. JohnWiley and Sons (560 pp.).

Heaman, L.M., Kjarsgaard, B.A., Creaser, R.A., 2003. The timing of kimberlite magmatism inNorth America: implications for global kimberlite genesis and diamond exploration.Lithos 71, 153–184.

Hook, B.A., Halfar, J., Gedalof, Z., Bollmann, J., Schulze, D., 2014. Stable isotope paleoclima-tology of the earliest Eocene using kimberlite-hosted mummified wood from the Ca-nadian Subarctic. Biogeosci. Discuss. 11, 16269–16308. http://dx.doi.org/10.5194/bgd-11-16269-2014.

International Committee for Coal and Organic Petrology (ICCP), 1998. The new vitriniteclassification (ICCP system 1994). Fuel 77 pp. 349–358.

Jacob, H., 1989. Classification, structure, genesis and practical importance of natural solidoil bitumen (“migrabitumen”). Int. J. Coal Geol. 11, 65–79.

Jahren, A.H., Sternberg, L.S.L., 2008. Annual patterns within tree rings of the Arctic middleEocene (ca. 45 Ma): isotopic signatures of precipitation, relative humidity, and decid-uousness. Geology 36, 99–102.

Jebrane, M., Harper, D., Labbé, N., Sèbe, G., 2011. Comparative determination of thegrafting distribution and viscoelastic properties of wood blocks acetylated by vinylacetate or acetic anhydride. Carbohydr. Polym. 84, 1314–1320.

Leavitt, S.W., 2010. Tree-ring C–H–O isotopic variability and sampling. Sci. Total Env. 408,5244–5253.

Leavitt, S.W., Danzer, S.R., 1993. Method for batch processing small wood samples toholocellulose for stable-carbon isotope analysis. Anal. Chem. 65, 87–89.

Page 9: Extraction of α-cellulose from mummified wood for … of α-cellulose from mummified wood for stable isotopic analysis ... Alpha-cellulose (α-cellulose) has been traditionally defined

27B.A. Hook et al. / Chemical Geology 405 (2015) 19–27

Lewis, B.A., Wrenn, J.H., Lewis, A.J., Alford, J.J., Alford, D., 2003. Middle Winsonsinan andrecent wet site mummifiedwood, humus, peat, and pollen, Santa Rosa Island. Florida.Rev. Palaeobot. Palynol. 126, 243–266.

Loader, N.J., Robertson, I., Barker, A.C., Switsur, V.R., Waterhouse, J.S., 1997. An improvedtechnique for the batch processing of small wholewood samples to α-cellulose.Chem. Geol. 136, 313–317.

Mackowsky, M.-Th., 1982. Methods and tools of examination. In: Stach, E., Mackowsky,M.-Th., Teichmuller, M., Taylor, G.H., Chandra, D., Teichmuller, R. (Eds.), Stach's Text-book of Coal Petrology, 3rd ed Gerbruder Borntaeger, Berlin, pp. 295–299.

McCarroll, D., Loader, N.J., 2004. Stable isotopes in tree rings. Quat. Sci. Rev. 23, 771–801.Padgham, W.A., Fyson, W.K., 1992. The Slave Province: a distinct Archean Craton. Can.

J. Earth Sci. 29, 2072–2086.Philippe, M., Bamford, M.K., 2008. A key to morphogenera used for Mesozoic conifer-like

woods. Rev. Palaeobot. Palynol. 148, 184–207.Poole, I., van Bergen, P.F., 2006. Physiognomic and chemical characters in wood as

palaeoclimate proxies. Plant Ecol. 182, 175–195.Pradhan, B.K., Sandle, N.K., 1999. Effect of different oxidizing agent treatments on the sur-

face properties of activated carbons. Carbon 37, 1323–1332.Richard, B., Quilés, F., Carteret, C., Brendel, O., 2014. Infrared spectroscopy and multivari-

ate analysis to appraise α-cellulose extracted from wood for stable carbon isotopemeasurements. Chem. Geol. 381, 168–179.

Richter, S.L., Johnson, A.H., Dranoff, M.M., LePage, B.A., Williams, C.J., 2008. Oxygen isotoperatios in fossil wood cellulose: isotopic composition of Eocene- to Holocene-aged cel-lulose. Geochim. Cosmochim. Acta 72, 2744–2753.

Rissanen, J.V., Grénman, H., Xu, C., Willför, S., Murzin, D.Y., Salmi, T., 2014. Obtainingspruce hemicelluloses of desired molar mass by using pressurized hot water extrac-tion. ChemSusChem 7, 2947–2953.

Rutherford, D.W.,Wershaw, R.L., Cox, L.B., 2005. Changes in Composition and Porosity Oc-curring During Thermal Alteration of Wood and Wood Components: U.S. GeologicalSurvey Scientific Investigations Report 2004–5292 (80 pp.).

Rutherford, D.W., Wershaw III, R.L. Reeves, J.B., 2008. Development of Acid FunctionalGroups and Lactones During the Thermal Degradation of Wood and Wood Compo-nents: U.S. Geological Survey Scientific Investigations Report 2007–5013 (43 pp.).

Sahay, V.K., 2011. The hydrocarbon potential, thermal maturity, sequence stratigraphicsetting and depositional palaeoenvironment of carbonaceous shale and lignite suc-cessions of Panandhro, northwestern Kutch Basin Gujarat, Western India. Cent. Eur.J. Geosci. 3, 12–28.

Savard, M.M., Bégin, C., Marion, J., Arseneault, D., Bégin, Y., 2012. Evaluating the integrityof C and O isotopes in sub-fossil wood from boreal lakes. Palaeogeogr. Palaeoclimatol.Palaeoecol. 348–349, 21–31.

Schleser, G.H., Frielingsdorf, J., Blair, A., 1999. Carbon isotope behaviour in wood andcellulose during artificial aging. Chem. Geol. 158, 121–130.

Sheu, D.D., Chiu, C.H., 1995. Evaluation of cellulose extraction procedures for stable car-bon isotopemeasurement in tree-ring research. Int. J. Environ. Anal. Chem. 59, 59–67.

Staccioli, G., Santoni, I., Pizzo, B., 2014. Decay of fossil wood from kimberlite pipes of Lacde Gras in the Canadian sub-Arctic area. Ann. Paléontolog. 100, 87–94.

Stambaugh,M.C., Guyette, R.P., 2009. Progress in constructing a long oak chronology fromthe central United States. Tree-Ring Res. 65, 147–156.

Stasiuk, L.D., Lockhart, G.D., Nassichuk, W.W., Carlson, J.A., 1999. Thermal maturity evolu-tion of dispersed organic matter inclusions from kimberlite pipes, Lac de Gras, North-west Territories, Canada. Int. J. Coal Geol. 40, 1–25.

Stasiuk, L.D., Sweet, A.R., Issler, D.R., 2006. Reconstruction of burial history of eroded Me-sozoic strata using kimberlite shale xenoliths, volcaniclastic and crater facies, North-west Territories, Canada. Int. J. Coal Geol. 65, 129–145.

Stasiuk, L.D., Sweet, A.R., Issler, D.R., Kivi, K., Lockhart, G.D., Dyck, D.R., 2005. Pre- andpost-kimberlite emplacement thermal history of Cretaceous and Tertiary sediments,Lac de Gras, Northwest Territories, Canada. 8th International Kimberlite ConferenceLong Abstracts. 8th International Kimberlite Conference, Victoria Canada.

Sýkorová, I., Pickel, W., Christanis, K., Wolf, M., Taylor, G.H., Flores, D., 2005. Classificationof huminite — ICCP System 1994. Int. J. Coal Geol. 62, 85–106.

Werner, R.A., Kornexl, B.E., Roßmann, A., Schmidt, H.L., 1996. On-line determination ofδ18O-values of organic substances. Anal. Chim. Acta 319, 159–164.

Wilson, L., Head III, J.W., 2007. An integratedmodel of kimberlite ascent and eruption. Na-ture 447, 53–57.

Wolfe, A.P., 2013. Personal Communication. University of Alberta, Alberta, Canada,Department of Biological Sciences.

Wolfe, A.P., Csank, A.Z., Reyes, A.V., McKellar, R.C., Tappert, R., Muehlenbachs, K., 2012. Pris-tine Early Eocene wood buried deeply in kimberlite from northern Canada. PLoS One 7e45537.

Wright, W.E., 2008. Statistical evidence for exchange of oxygen isotopes in holocelluloseduring long-term storage. Chem. Geol. 252, 102–108.