487

Oecologia (2000) 123:481-489

© Springer-Verlag 2000

John S. Roden James R. Ehleringer

Hydrogen and oxygen isotope ratios of tree ring cellulosefor field-grown riparian trees

Received: 7 April 1999 / Accepted: 8 December 1999

Abstract The isotopic composition of tree ring cellu-lose was obtained over a 2-year period from small-di-ameter riparian-zone trees at field sites that differed insource water isotopic composition and humidity. Thesites were located in Utah (cool and low humidity), Ore-gon (cool and high humidity), and Arizona (warm andlow humidity) with source water isotope ratio values of—125/-15%o (6D/6 180), —48/-6%o, and —67/-7%o, re-spectively. Monthly environmental measurements in-cluded temperature and humidity along with measure-ments of the isotope ratios in atmospheric water vapor,stream, stem, and leaf water. Small riparian trees usedonly stream water (both 8D and 6 180 of stem and streamwater did not differ), but 6 values of both atmosphericwater vapor and leaf water varied substantially betweenmonths. Differences in ambient temperature and humidi-ty conditions between sites contributed to substantial dif-ferences in leaf water evaporative enrichment. These leafwater differences resulted in differences in the 6D and6 180 values of tree ring cellulose, indicating that humidi-ty information was recorded in the annual rings of trees.These environmental and isotopic measurements wereused to test a mechanistic model of the factors contribut-ing to 513 and 6 180 values in tree ring cellulose. Themodel was tested in two parts: (a) a leaf water model us-ing environmental. information to predict leaf waterevaporative enrichment and (b) a model describing bio-chemical fractionation events and isotopic exchange withmedium water. The models adequately accounted forfield observations of both leaf water and tree ring cellu-lose, indicating that the model parameterization fromcontrolled experiments was robust even under uncon-trolled and variable field conditions.

J.S. Roden ([1) • J.R. EhleringerStable Isotope Ratio Facility for Environmental Research,Department of Biology, University of Utah,Salt Lake City, UT 84112, USA

J.S. Roden,Department of Biology, Southern Oregon University,Ashland, OR 97520, USA,e-mail: rodenj@sou.edu, Fax: +1-541-5526412

Key words 8D - 8 180 • Cellulose modeling • Humidity •Water source

Introduction

Tree rings have been used extensively to quantify past cli-matic variation through measurements of ring widths (seereview in Fritts 1976) and more recently through stableisotope analyses of cellulose (see review in Switsur andWaterhouse 1998). The 6D and 6 180 values in tree ringcellulose are derived from water and reflect, to a first ap-proximation, precipitation inputs (Yapp and Epstein 1982;Lawrence and White 1984; White et al. 1994). Regres-sions of the variations in 6D and 6 180 values in tree ringcellulose have been used for temperature reconstruction(Schiegl 1974; Gray and Thompson 1976; Epstein andYapp 1977; Feng and Epstein 1994), since the isotopiccomposition of precipitation is a direct function of con-densation temperature (Dansgaard 1964). However, a per-fect 1:1 relationship between the isotopic composition oftree ring cellulose and source water is seldom observed,implying that tree rings may record additional informationrelated to biological processes that respond to local envi-ronmental variation. Many studies have used correlationanalyses to determine which environmental parameters(water source, humidity, temperature) might be recordedin the cellulose of annual growth rings (Schiegl 1974;Gray and Thompson 1976; Epstein and Yapp 1977; Burkand Stuiver 1981). However, substantial uncertainty stillremains as to what biological and environmental informa-tion is contained in the isotopic composition of tree ringcellulose (DeNiro and Cooper 1990; Edwards 1990). Forexample, some studies report that tree rings may containinformation about local humidity (Edwards and Fritz1986; Lipp et al. 1993; Saurer et al. 1997), while othersfind no evidence for a humidity signal (DeNiro andCooper 1989; White et al. 1994; Terwilliger and DeNiro1995). It is important to determine how these contrastingresults can be reconciled if we are to make ecological in-terpretations from tree ring isotope data.

482

Tree rings are not a direct recorder of the isotopiccomposition of precipitation, since there are many stepsalong the path from source water to cellulose. Roden et al.(2000) presented a model to describe the various fraction-ation events at each step along the way. This model hasbeen shown to adequately account for the various resultsin the literature (Roden et al. 2000) as well as those ob-tained from a long-term experimental system where treerings were produced under controlled environmentalconditions (Roden and Ehleringer 1999a).

The 6D and 6 180 values of tree ring cellulose ulti-mately originate from the water taken up by the roots.Although there is no fractionation upon uptake for either6D or 6 180 (White et al. 1985; Dawson 1993; althoughmangroves growing in sea water are an exception: Linand Sternberg 1993), there are numerous points alongthe pathway to a tree ring where the 6 values of thesource water may be altered. The model, described in de-tail by Roden et al. (2000), has two major components.Firstly, a modified version (Flanagan et al. 1991a) of theCraig and Gordon (1965) evaporative enrichment modelwas used to estimate the 6D or 6 180 of leaf water in thechloroplast using environmental information and a givenisotopic input of source water and atmospheric vapor.Secondly, a biochemical model (developed from the gen-eral models of Sternberg et al. 1986; Yakir andDeNiro 1990; Luo and Sternberg 1992) that predicts cellu-lose isotopic composition based on autotrophic (carbohy-drate synthesis associated with photosynthesis) and het-erotrophic (carbohydrate metabolism associated with cel-lulose synthesis in the stem) fractionation factors as wellas the extent of isotopic exchange between substrates andmedium water in the steps leading to cellulose synthesis:

8pcx —fli(Opwx+Efili)+( l fil )(6Dw i+EFIA)

(1)

6'80cx=f0(8"Ow.+60)+(140)(6 '80wi+E0)

(2)

where the subscripts cx, wx, and wl refer to the xylemcellulose, xylem water, and leaf water, respectively.Yakir and DeNiro (1990) calculated the autotrophic frac-tionation factor for hydrogen U HA) to be —171700. Theautotrophic fractionation factor for oxygen (e0) is +27%ofrom the carbonyl-water interaction during biosynthesis(Sternberg and DeNiro 1983). cfm and co are the hetero-trophic fractionation factors for the enzyme-mediated ex-change or addition of either hydrogen or oxygen. For hy-drogen, EHH=+158%0 (Yakir and DeNiro 1990) and sincethe fractionation factor is the same for autotrophic andheterotrophic metabolism for oxygen (Hill et al. 1995),there is no need to distinguish between the two(c0=+27%0). f is the proportion of the carbon-bound hy-drogen or oxygen that undergoes exchange with the me-dium water at the site of cellulose synthesis (see also asli ghtly different formulation using f as a damping factor,as in Saurer et al. 1997). A spreadsheet version of themodel is available at ftp ://ecophys.biology.utah.edu/tree-ring/.

These models were tested and parameterized on sap-lings of three tree species (alder, birch, and cottonwood)

using a hydroponics system that controlled source waterisotopic composition, and greenhouses that controlled thetemperature and humidity environment and therefore leafwater evaporative enrichment (Roden and Ehleringer1999a). However, it is also important to determine if theresults obtained in controlled environments are applica-ble to 'real world' situations in the field. Although a hy-droponics system can provide a substantially largerrange in source water isotopic composition than naturalsystems, field verification of the model parameters iscritical for the application of the model for climate re-construction or ecological comparisons. Therefore, theobjective of this study was to determine if the models ofRoden et al. (2000) can predict the isotopic compositionof leaf water and tree ring cellulose for trees grown inthe field under varied source waters and relative humidi-ties.

Materials and methods

Plant material and field sites

Field sites were located along stream courses in four locations.The locations were (a) the Bill Williams River Wildlife Reservenear Lake Havasu, Arizona (34°16' N 114°02' W) containing cot-tonwood (Populus fremontii Wats), (b) Cascade Head Field Sta-tion near Otis, Oregon (45°02' N 123°55' W) containing red alder(Alnus rubra Bong), (c) the Weber River near Ogden, Utah(41°08' N 111°54' W) containing cottonwood (P. fremontii), and(d) the Red Butte Canyon Nature Reserve near Salt Lake City,Utah (40°47' N 111°48' W) containing both alder (A. incana L.Moench) and water birch (Betula occidentalis Hook). The siteswere chosen for their proximity to Salt Lake City, similar tree spe-cies, differences in growing season humidity and differences instream water isotopic composition with the Utah sites being sub-stantially more depleted than either the Oregon or Arizona sites.All trees (five to seven individuals per site) were streamside andless than 10 cm in diameter to ensure that they were still using thestream water rather than deeper sources (Dawson and Ehleringer1991). The same trees were followed over two growing seasons(1996 and 1997) except when trees were damaged (by beavers) ordied, and then a replacement tree was selected.

Isotope sampling and environmental measurements

At monthly intervals, approximately 5 ml of stream water wassampled at each site. During the growing season, in addition tostream water, monthly samples of twigs and leaves were collectedfor water extraction. Leaf material, with the midvein removed, andapproximately 5-10 cm of suberized twig material were placed in-to separate glass vials, sealed with parafilm and brought back tothe laboratory on dry ice, and then placed into a freezer (-5°C) un-til the water could be extracted for isotopic analysis. At the time ofleaf and stem water collection, the ambient relative humidity andair temperature were measured and the atmospheric water vaporwas sampled using a pump to draw air through a glass trap sub-merged in a mixture of ethanol and dry ice (-78°C). Relative hu-midity was measured with both a sling psychrometer and a Vaisalahumidity sensor, since Vaisala sensors are less accurate at high hu-midity (Oregon) and sling psychrometers are less accurate at verylow humidity (Arizona) (Runde! and Jarrell 1989). At the end ofeach growin g season, cores of the main stems were taken fromeach cardinal direction. I he annual growth ring produced dunngthat year was cut out, dried and ground to pass a 40-mesh screenusing a Wiley mill (Thomas, Philadelphia. Pa.). The ring was notsubdivided due to the relatively large amount of material needed

0

-50

-100

-150

00

0 -50

(1) -100CO

-150

0

-50

-100

-150

483

for cellulose nitration and the fact that these trees were obtainingwater from streams that had little seasonal variation in SD or 6180,reducing some of the differences between earlywood and late-wood.

Sample preparation and analysis

Leaf water was obtained by cryogenic extraction as described byEhleringer and Osmond (1989). The sample was frozen in liquidnitrogen (-190°C) and once evacuated, the system was then isolat-ed from the vacuum pump and immersed in boiling water. The wa-ter from the leaf was then collected in a tube immersed in liquidnitrogen until all water was extracted. The OD of water samplesfrom the streams, leaves and atmospheric vapor were obtained byreducing the H in 2 al of 1-1 20 to H2 using 100 mg of a Zn catalyst(J. Hayes, Indiana University) in a 500°C oven (modification ofColeman et al. 1982). The 5 180 of water samples were obtained byequilibrating 0.5 ml of water with approximately 16 kPa of CO 2 ina 25°C water bath for 48 h (Socki et al. 1992). The CO 2 was ex-tracted cryogenically using liquid nitrogen and dry ice/ethanoltraps. Both the H2 and CO2 were analyzed on a Finnagan MATdelta S isotope ratio mass spectrometer (San Jose, Calif.) with aprecision of ±1 7oo for 6D and ±0.2%c for 6180.

The 5180 recorded in tree rings was obtained from a-cellulose(Leavitt and Danzer 1992) which involves a delipification step(using toluene and ethanol), boiling in water (to remove solublesugars), bleaching with sodium chlorite and acetic acid (to removelignin and proteins), and washing in a strong alkaline solution (toremove hemicellulose). Approximately 1.2 mg of a-cellulose wasthen placed in a silver capsule and converted to CO by pyrolysis(Saurer et al. 1998) in a hot (1,100°C) high-purity alumina com-bustion column (Carla-Erba interface) and separated from othergases in a 1-m mole sieve GC column connected to a FinnaganMAT delta S isotope ratio mass spectrometer. Repeated samplingwas utilized to reduce memory effects and resulted in a precisionof ±0.4%c.

To obtain the 6D recorded in tree ring cellulose, the a-cellu-lose obtained, as described above, was nitrated to remove the ex-changeable hydrogens. The a-cellulose was placed in a flask witha solution of nitric acid and acetic anhydride. The material wasthen washed, dissolved in acetone (to obtain purified trinitratedcellulose) and freeze dried for storage. Approximately 11 mg ofnitrated cellulose was placed in a pyrex tube with 1 g of cupric ox-ide, evacuated, sealed and combusted for 3 h at 520°C. The result-ing gases were separated cryogenically using liquid nitrogen anddry ice/ethanol traps to move the water vapor to a tube containingthe Zn catalyst for hydrogen reduction as described above.

Throughout this paper we use the conventional 'delta' notationwhich expresses the isotopic composition of a material relative tothat of a standard on a per mil (%c) deviation basis:8=(Rsample/Rstandard-1 ) 1," (3)where 5 is referred to as the isotope ratio (6D for hydrogen and5 180 for oxygen). The standard for both hydrogen and oxygen isstandard mean ocean water (SMOW).

Results

Figures 1 and 2 show the 6D and 6 180 values of atmo-spheric water vapor along with stream, leaf, and xylemsap water. The water taken up by the trees was virtuallyidentical isotopically to the stream water, implying thatthe trees were not tapping deeper water sources. The cot-tonwood trees near the Bill Williams River in Arizonadid tap some deeper water sources, since the surfacestream dried up fairly early in the season. However, therewas evidence that the stream was simply flowing belowthe river bed which is common in desert riparian zones.

,- Weber River, cottonwood -.

- .--.--11L'o .--11"-•-•

Ef ta=g=d7 • • • •v,,.--•\ • •.- 0"._,

'' • ' ''/I ' ' ' ' '

, /-•—• 4-.._.a..„.._."-

.

3,.._o__c) 3____.cy_o___0-

y/TIL-Y-

Bit Williams Rivercottonwood

_ Red Butte Canyon, alder _

- •-_.--. N-••

' ' ' ' " ' ll '' '

:1(.4

•

• . • V110 •

•

Cascade Head, alder' ' '

iNT7•

- Red Butte Canyon, birch

- ...._.-.

_

8=0=F=S ..4,.. -

I150 200 250 150 200

1996 1997Day of year

o xylem sap waterleaf water

v stream water♦ atmospheric water vapor

150 200 250 150 2001996 1997

Day of year

Fig. 1 The hydrogen isotope composition of stream, stem, andleaf water along with atmospheric water vapor during the growingseason over a 2-year period. Values are means±SE (where replica-tion was possible)

The similar isotopic values of xylem sap throughout theyear for the Arizona cottonwoods implies that these treeswere simply tapping the below-ground stream flow. Boththe Arizona (Bill Williams River) and Oregon (CascadeHead) sites had substantially more enriched source water(-67/-7%o and —48/-6%o, 6D/6 180, respectively) than theUtah sites (Weber River and Red Butte Canyon Creekwater, —131/-15700 and —125/-15%0, 6D/6 180, respec-tively). The stream water for these sites did not varymuch throughout the winter months (data not shown), in-dicating that the source water used for cellulose synthe-sis in these trees was not subject to large seasonal varia-tions in meteoric water isotopic composition whichcould complicate the interpretation of tree ring results(requiring the annual ring to be subdivided).

As expected, the leaf water 6D and 5 180 values weremore enriched than the source water. However, the leafwater values were much more variable than those ofsource water (Figs. 1, 2). Leaf water varied within a sin-gle season by as much as 50%o and 15%0 in 6D and 6180,respectively. The higher humidities and lower tempera-tures at the Oregon site (Cascade Head; Table 1) reducedthe amount of leaf water evaporative enrichment corn-

-80 Utah

-120

-100

20meteoric water line(SD -,- 8 * 8 180 - 10)

-20 Oregon

0

-40

-60Oregon

Arizona

Arizona

0 stem water

-140 - • leaf water _

-20 -10 0 10 20Water 6180 (%.)

Fig. 3 The relationship between growing season stem and leafwater 6 180 and 61) at three sites

alder 1996alder 1997cottonwood 1996cottonwood 1997birch 1996birch 1997

co0 -100

a)

ca) -125

a)H-150

-175-175 -150 -125 -100 -75 -50

-50

-75 -

0

0GO

Utah

Temperature(°C)

17.434.023.624.5

Relativehumidity(%)

84.436.035.636.3

Site

Cascade Head (Oregon)Bill Williams River (Arizona)Red Butte Canyon (Utah)Weber River (Utah)

pared to the other sites (Figs. 1, 2). A plot of the 6D ver-sus 6 180 for both leaf and xylem sap water (Fig. 3) dem-onstrates that the source water being utilized by thesetrees fell on the same slope as the meteoric water line(OD=86 180-10). However, the leaf water for these treesfell off the meteoric water line by an amount dependenton the site environmental conditions. The warm tempera-t_--_s and low humidity at the Alizona site (Table 1) re-sulted in highly enriched leaves that fell on a line be-tween source water and leaf water with a slightly smallerslope (m=3.2) than the low humidity but cooler condi-

484

Weber River, cottonwood

0_-...___.

9=9 91-.--F .---

' ' • ' ' ' 'II

e--•A \Nei -

Bit ms Rivercottonwood

' . ' ' ' •// • ' ' '- Red Butte Canyon, alder

- e--.)11- 6/\_-••

fA=C47_ _ • ir__,F___.„." v-,,,,,_•_._v

- .--•---7_

Cascade Head, alder. ' • ' ' ' 'll , , , ll ,, ,

_ Red Butte Canyon, birch 150 200 250 150 200 2501996 1997

- •—•/. Day of year

0 xylem sap water

_ leaf water. • • •. v stream water

.414111..- ♦ atmospheric water vapor•""

150 200 250 150 200 2501996 1997Day of year

Fig. 2 The oxygen isotopic composition of stream, stem, and leafwater along with atmospheric water vapor during the growing sea-son over a 2-year period. Values are means±SE (where replicationwas possible)

Table 1 The mean growing season temperatures and relative hu-midities at the four sites in this study. Values were obtained frommonthly measurements during the summers of 1996 and 1997

20

10

-10

-20

-30200100

0

0) -10

-20

3020

10

0

-10

-20

-30

Xylem water, OD (%0)

Fig. 4 The relationship between source water and tree ring cellu-lose 6D for three species at three sites over a 2-year period. Dataare distinguished by species and year. Sites are distinguished byspecies and source water 6D values with the most depleted watersfrom the Utah sites and the more enriched water from the Arizona(cottonwood) and Oregon (alder) sites. Values are means±SD

tions of the Utah sites (m=3.5) and a much smaller slopethan the cool and humid Oregon site (m=4.6: Fig. 3).

The 6 values of atmospheric water vapor were from5 to 75%o and 2 to 13%o more depleted than stream waterin 6D and 6 180, respectively (Figs. 1, 2). Within a singleseason, the isotopic composition of atmospheric watervapor varied by as much as 43%o and 8%o in 6D and6 180, c3pc■-tivciy (FigN. I, 2). Tdowevei, 011 averagc,,large deviations in atmospheric water vapor may havebeen associated with the intensity and origin of stormevents. Whatever the cause, the heavy isotopes in atmo-

alder 1996alder 1997cottonwood 1996

7 cottonwood 1997birch 1996birch 1997 T

-16 -12 -8 -4

0

485

36

0

O,..c° 32

0(I) 2800)

24

Xylem water, 6 180 (°/00)

Fig. 5 The relationship between source water and tree ring cellu-lose 5 180 for three species at three sites over a 2-year period. Dataare distinguished by species and year. Sites are distinguished byspecies and source water 8 180 values with the most depleted wa-ters from the Utah sites and the more enriched water from the Ari-zona (cottonwood) and Oregon (alder) sites. Values are means±SD

spheric water vapor are clearly not constant and may notalways covary with plant source water especially if treesare not dependent on precipitation events for water.

The 6D values of tree ring cellulose were dependenton source water 6D, but not in a simple 1:1 relationship(Fig. 4). The most depleted values were from the Utahsites and the more enriched alder and cottonwood valueswere from the Oregon and Arizona sites, respectively.Even though the Arizona site had more depleted watersources than the Oregon site, the tree ring cellulose wassimilar or even more enriched in OD than at the Oregonsite. Thus, the differences in humidity and temperaturebetween the two sites, which altered leaf water isotopiccomposition (Fig. 1), affected the isotopic compositionof the cellulose laid down. The differences between theOregon and Arizona sites were even more pronouncedwhen the 6 180 values of source water and tree ring cellu-lose were compared (Fig. 5). This is likely due to thegreater effect of evaporative enrichment on leaf water6 180 than 6D, as seen in deviations from the meteoricwater line (Fig. 3).

The mixture of three species with similar water sourc-es at the Utah sites (the most depleted xylem waters;Figs. 4, 5) as well as the data from the Arizona site showsubstantial variation in tree ring cellulose 6 values be-tween years (open versus closed symbols) and this varia-tion can be as large as that between species. Thus, the in-corporation of stable isotopes into the cellulose of treerings may indeed record valuable environmental infor-mation which is sensitive to differences in climate be-tween years.

DiscussionAlthough the lack of a 1:1 relationship between the hy-drogen isotopic composition of tree ring cellulose andsource water (Fig. 4) would seem to differ from manyother studies (Epstein and Yapp 1977; Yapp and Epstein1982; White et al. 1994; Terwilliger and DeNiro 1995),the variation around the 1:1 line is within that found insome of those studies. For example, in our results, thedata obtained from the Oregon site were about 50%omore depleted in deuterium than predicted by a 1:1 rela-tionship. The studies of Epstein and Yapp (1977) andYapp and Epstein (1982) described a 1:1 relationshipover a wide range of environmental water isotopic com-positions (>l80%) with a maximum deviation from the1:1 line of 25-50%0, respectively. The results from theOregon and Arizona sites (Fig. 4) indicated that humidi-ty differences can affect tree ring isotopic composition.Therefore, much of the variation around the 1:1 relation-ship observed in other studies could be related to addi-tional environmental factors unaccounted for by modelsthat predict the 6D of tree ring cellulose as a simplefunction of source water (e.g., Terwilliger and DeNiro1995). Humidity differences would affect the oxygenisotope ratios of tree ring cellulose (Fig. 5) to an evengreater extent, since evaporation causes a greater enrich-ment in leaf water 6 180 than 6D (Fig. 3).

These data collected from natural field situationswere used to test the Roden et al. (2000) models thatwere previously parameterized based on controlled ex-perimental systems (Roden and Ehleringer 1999a). Themonthly observations of leaf water were used to test amodified version (Flanagan et al. 1991a) of the Craigand Gordon (1965) evaporative enrichment model.Roden and Ehleringer (1999b) found that a modifiedCraig-Gordon model was robust over a wide range ofleaf waters, although other studies have found that theCraig-Gordon model predicted somewhat greater thanobserved isotopic enrichment (Allison et al. 1985;Leaney et al. 1985; Flanagan and Ehleringer 1991;Flanagan et al. 1991b; Wang and Yakir 1995). The leafwater model uses vapor pressures in combination withthe 6 values of source water and atmospheric water va-por as primary inputs (which were all measured at eachsite at the time of leaf water collections) as well as kinet-ic and equilibrium fractionation factors from Flanagan etal. (1991a). Leaf water models are sensitive to humidityinputs (Dongmann and Niirnberg 1974; Flanagan et al.1991a; Roden and Ehleringer 1999b) and since canopiescan substantially modify the water vapor microenviron-ment of a leaf, empirical relationships were developedusing humidity measurements made in the open andwithin the canopy to correct humidity observations col-lected in the open. The corrections were generally verysmall (5-7% for 613) but helped bring closer agreementbetween predicted and observed 6D and 6 180 values forleaf water (Figs. 6, 7).

Using Eqs. 1 and 2 and known autotrophic and het-erotrophic fractionation factors, the proportion of the hy-

25-5 0 5 10 15 20

25

020

015

its10

Cu

0

-5

486

20

-73; 0

0":,)

-20CDCuCu -40Cu

a- u -60ci)

-80

-100-100 -80 -60 -40 -20 0 20

Modeled leaf water, 6D (%.)

Fig. 6 The relationship between the SD of modeled and measuredleaf water for field-grown riparian-zone trees. Variations in leaf waterwere due to different stream water 6D values and temperature and hu-midity differences between sites. The line represents a 1:1 relationship

Modeled leaf water, 6 180 (%.)

Fig. 7 The relationship between the 5 180 of modeled and mea-sured leaf water for field-grown riparian-zone trees. Variations inleaf water were due to different stream water 8 180 values and tem-perature and humidity differences between sites. The line repre-sents a 1:1 relationship

drogen or oxygen that undergoes exchange with mediumwater (fH and frs) can he calculated from the closestagreement between field cellulose observations andmodel predictions (Roden et al. 2000). We used thehydrogen isotope biological fractionation factor for auto-

trophic carbohydrate metabolism (E HA=-1 7 1%0) fromYakir and DeNiro (1990) as well as the biochemical frac-tionation factor associated with heterotrophic syntheticreactions (cH=+158700) from Yakir and DeNiro (1990),which is similar to the mean multispecies value of+155%0 from I u p and Sternberg (1992). The near-equiv-alent magnitudes and opposite signs of CA and cH indi-cate that the hydrogen isotope ratio values of tree ringcellulose might be expected to approach that of sourcewater (1:1 relationship) as has been commonly observedby many previous studies (Epstein and Yapp 1977; Burkand Stuiver 1981; Yapp and Epstein 1982; Lawrence andWhite 1984; White et al. 1994). For oxygen, the fraction-ation factor is the same for both heterotrophic and auto-trophic metabolism ( c0=±27%0; Sternberg 1989; Yakirand DeNiro 1990). The mechanism for the 27%0 enrich-ment is the carbonyl-water interaction during biosynthe-sis (Sternberg and DeNiro 1983). We have also shown(J.S. Roden and J.R. Ehleringer, unpublished data) thatthese fractionation factors are not temperature depen-dent.

The cellulose model was constrained to use only the foand fH values that had been derived from the tightly con-trolled greenhouse experiments (Roden and Ehleringer1999a). For oxygen isotope ratios in cellulose, there wasa very strong agreement between the best-fit fo valuefrom field observations and the greenhouse experiments.Both experiments predicted that fo values should be0.42. For the field observations, it is important to recog-nize that the leaf water values associated with cellulosesynthesis may not have been exactly the same as thosevalues collected at midday on a particular day of themonth. Therefore, our modeling analyses considered therange of possible predictions based on the range of leafwaters observed rather than only the average leaf waterobserved. The range of predicted cellulose 6 180 ratios isdepicted in Fig. 8 as the gray band. For hydrogen iso-topes in field cellulose samples, the best-fit fy value was0.31 (Fig. 9), which is slightly lower than had been ob-served in the well-controlled greenhouse experiment(0.36). When we constrain the model to use only thegreenhouse observed fH value of 0.36, the relationshipbetween the model predictions and field observations isshown by the right-hand edge of the gray band in Fig. 9.The left-hand edge of this gray band corresponds to thelower range of possible values, based on known fieldvariations in leaf water values as previously described. Itis important to recognize that the model is inherentlymore sensitive to variations in Ai than fo values, becausefor oxygen, the heterotrophic and autotrophic fraction-ation factors are identical, but they differ for hydrogen(Roden et al., in press). The predicted differences in fHvalues may be real or possibly associated with the lackof constant environmental control in the field observa-tions relative to the precise controls available undergreenhouse conditions

The results of DeNiro and Cooper (1989) for oxygenand Terwilliger and DeNiro (1995) for hydrogen implythat complete isotopic exchange with xylem water oc-

-60

-80

0

-1000

-160

487

40

036

0co

320

a)O 28-oa)

(f)coa) 24

2020 24 28 32 36

40

Modeled cellulose, 8 180 (°/00)

Fig. S The relationship between the 6 180 of modeled and mea-sured tree ring cellulose for field-grown riparian-zone trees. Varia-tions in tree ring cellulose were due to different stream water 5180values and temperature and humidity differences between sites.Values are means±SD. The line represents a 1:1 relationship. Thegray band represents the predictions of the model using the mea-sured range of leaf water values rather than the average leaf watervalue for a sampling date

Modeled cellulose, 6D (°/00)

Fig. 9 The relationship between the SD of modeled and measuredtree ring cellulose for field grown riparian zone trees. Variations intree ring cellulose were due to different stream water SD valuesand temperature and humidity differences between sites. Valuesare means±SD. The line represents a 1:1 relationship. The grayband represents the predictions for cellulose isotope ratios if fHwas constrained to a value of 0.36 and if we include the observedrange of leaf water values

curs during cellulose synthesis (that is, both fH and fo=1)and thus a strong correlation should be observed betweensource water and tree ring cellulose. We observed that fHand fo were less than 1 (fli=0.31, fo=0.42), indicating in-complete isotopic exchange between carbohydrate sub-strate and medium water during cellulose synthesis.These results agree with the observations of Luo andSternberg (1992) that fH and fo are roughly similar inmagnitude, and that when cellulose is derived fromstarch fic----fo,--0.34. Other studies have also reported thatboth fH and fo are substantially less than 1 (0.4-0.5;Sternberg et al. 1986; Yakir and DeNiro 1990; Yakir1992) and are dependent on the carbohydrate substrate(lipids vs starch). Although Saurer et al. (1997) use asimilar term (f, a damping factor) and obtained a similarresult (f=0.4-0.5) for the 6 180 of cellulose in tree ringsof three coniferous species, their model differs from thatpresented here. Their damping factor (f) includes bothleaf water heterogeneity and isotopic exchange of su-crose with medium water, whereas our model distin-guishes the two factors and is thus not strictly compara-ble. The differences between the Oregon and Arizonasite also indicate that humidity information is recordedin tree ring cellulose, which differs from the predictionsof DeNiro and Cooper (1989) and Terwilliger andDeNiro (1995), but agrees with other studies (Burk andStuiver 1981; Edwards and Fritz 1986; Lipp et al. 1993).

Of course, this modeling analysis could have assumedthat fH and fo are well established in the literature (-----0.4;Yakir and DeNiro 1990) and tested the literature esti-mates of EHH , EHA , and Eo. However, as seen above, theestimates of fH and fo and the concept of isotopic re-equilibration during cellulose synthesis is still the mostcritical concept open to debate. The value of E0 has beenwell established (Sternberg and DeNiro 1983; Sternberg1989), and very similar values for EHH have been foundin two studies (Yakir and DeNiro 1990; Luo and Stern-berg 1992). Estimates of EHA are more difficult to deter-mine, and using field observations would not be the mostappropriate system to test both E HH and EHA since Eq. 1would have two unknowns.

The fact that these parameters provided the greatestpredictive power in both our field and experimental stud-ies (Roden and Ehleringer 1999a) indicates that in-creased confidence in the literature estimates of bio-chemical fractionation is warranted.

Model predictions of the isotopic composition of treering cellulose from field-grown trees are complicated bylimited knowledge as to the prevailing environmentalconditions at the time of cellulose synthesis. Monthlymeasurements of source water and atmospheric watervapor OD and 8 180 along with estimates of leaf microcli-mate are likely insufficient to capture all the possible en-vironmental variation for modeling purposes. Our datashow that the isotopic composition of leaf water and at-mospheric water vapor can vary substantially over shorttime periods (Figs. 1, 2). White and Gedzelman (1984)have shown that the 8D of atmospheric vapor can varyby as much as 60%o within a season and by even as much

488

as 30%o over a 24-h period. As such, the isotopic compo-sition of atmospheric vapor may not always be in equi-librium with local environmental water and can varywith the meteorological setting and the degree of mixingwith upper air masses (White and Gedzelman 1984).Seasonal variation in source water OD and 6 180 couldfurther complicate model predictions, which was avoid-ed in this study by using small streamside trees wheresource water did not substantially vary throughout thesampling period (Figs. 1, 2). Most of these environmen-tal uncertainties would be integrated into leaf water iso-topic signatures, and thus our results include a range ofpossible predictions (gray band; Figs. 8, 9) based on therange of leaf waters observed. Techniques for subsam-pling individual tree rings (Loader et al. 1995) to quanti-fy seasonal variations in cellulose 6D and 6 180 alongwith intensive sampling of leaf microclimate and the 6Dand 6 180 of source water, leaf water, and atmosphericwater vapor could greatly enhance our understanding ofthe way environmental variation is incorporated into atree ring and how much information is lost through aver-aging measurements and bulking samples.

These results obtained from field-grown trees en-hance our understanding of the relationships between theisotopic composition of source water and tree ring cellu-lose. Although the hypothesis of complete isotopic ex-change with stem water at the time of cellulose synthesisis attractive in its simplicity, it is not supported by the re-sults of this study. Clearly, the isotopic composition oftree ring cellulose reflects source water. However, theslope may not be equal to 1.0 under all environmentalconditions and other environmental parameters such ashumidity can alter that relationship. Although this com-plicates the straightforward interpretation of the SD and6 180 values in tree ring records as reflecting precipita-tion inputs only, it also makes those records richer in in-formation. The results presented here corroborate thatthe mechanistic model of tree ring isotope ratios (Rodenet al. 2000), developed and tested under controlled envi-ronments, is robust enough to . account for observed 8Dand 6 180 values of cellulose in trees grown under con-trasting field conditions. The results of this study help toclarify inferred relationships between tree ring isotopiccomposition and environmental parameters derived fromdendrogeochemical studies and correlation analysis.Field confirmation of these models will enhance theiruse as a tool for studying climate change, historical pat-terns of seasonal precipitation, long-term reconstructionof the sensitivity of plants to changes in moisture sourceand humidity, and water relation factors that contributeto immigration and emigration of species from a region.

Acknowledgements We thank C. Cook, M. Lott, W. Ike, and C.F.Kitty at the Stable Isotope Ratio Facility for Environmental Re-search for isotope analyses and valuable discussions. A. King, A.Svendsen, and S. Phillips for technical assistance, N. Gilbertsonand B. Raulston at the Bill Williams River Wildlife Refuge, Arizo-na, for site access and winter sampling, M. Sieg and the U.S. For-est Service for access to the Red Butte Canyon Research NaturalArea, Utah, R. Touchton and S. Greene at Oregon State Universityand the U.S. Forest Service for access and winter sampling at theCascade Head Field Station, Oregon. This study was supported bythe National Science Foundation grant no. IBN 95-08671.

ReferencesAllison GB, Gat JR, Leaney FW (1985) The relationship between

deuterium and oxygen-18 delta values in leaf water. ChemGeol 58:145-156

Burk RL, Stuiver M (1981) Oxygen isotope ratios in trees reflectmean annual temperature and humidity. Science 211:1417-1419

Coleman ML, Shepard TJ, Durham JJ, Rouse JE, Moore GR(1982) Reduction of water with zinc for hydrogen isotopeanalysis. Anal Chem 54:993-995

Craig H, Gordon LI (1965) Deuterium and oxygen-18 variations inthe ocean and the marine atmosphere. In: Tongiorgi E (ed) Pro-ceedings of a conference on stable isotopes in oceanographicstudies and paleotemperatures, Spoleto, Italy, pp 9-130

Dansgaard W (1964) Stable isotopes in precipitation. Tellus16:436-468

Dawson TE (1993) Water sources of plants as determined fromxylem-water isotopic composition: perspectives on plant com-petition, distribution, and water relations. In: Ehleringer JR,Hall AE, Farquhar GD (eds) Stable isotopes and plant carbon-water relations. Academic Press, London, pp 465-496

Dawson TE, Ehleringer JR (1991) Streamside trees that do not usestream water. Nature 350:335-337

DeNiro MJ, Cooper LW (1989) Post photosynthetic modificationof oxygen isotope ratios of carbohydrates in the potato: impli-cations for paleoclimatic reconstruction based upon isotopicanalysis of wood cellulose. Geochim Cosmochim Acta 53:2573-2580

DeNiro MJ, Cooper LW (1990) Water is lost from leaves andtrunks of trees by fundamentally different mechanisms. Geoc-him Cosmochim Acta 54:1845-1846

Dongmann G, Nurnberg HW (1974) On the enrichment of H2 180in the leaves of transpiring plants. Radial Environ Biophys11:41-52

Edwards TWID (1990) New contributions to isotope dendroclima-tology from studies of plants. Geochim Cosmochim Acta54:1843-1844

Edwards TWID, Fritz P (1986) Assessing meteoric water compo-sition and relative humidity from 180 and 2H in wood cellu-lose: paleoclimatic implications for southern Ontario. Can JEarth Sci 22:1720-1726

Ehleringer JR, Osmond CB (1989) Stable isotopes. In: Pearcy RW,Ehleringer JR, Mooney HA, Rundel PW (eds) Plant physio-lo gical ecology: field methods—and instrumentation. Chapman& Hall, London. pp 281-300

Epstein S, Yapp C (1977) Isotope tree thermometers. Nature266:477-478

Feng X, Epstein S (1994) Climatic implications of an 8000-yearhydrogen isotope time series from bristlecone pine trees. Sci-ence 265:1079-1081

Flanagan LB, Ehleringer JR (1991) Effects of mild water stressand diurnal changes in temperature and humidity on the stableoxygen and hydrogen isotopic composition of leaf water inCorm's stolonifera L. Plant Phvsiol 97:298-305

Flanagan LB, Comstock JP, Ehleringer JR (1991a) Comparison ofmodeled and observed environmental influences on the stableoxygen and hydrogen isotope composition of leaf water inPhaseolus vulgaris L. Plant Physiol 96:588-596

489

Flanagan LB, Bain JF, Ehleringer JR (1991b) Stable oxygen andhydrogen isotope composition of leaf water in C3 and C4 plantspecies under field conditions. Oecologia 88:394-400

Fritts HC (1976) Tree rings and climate. Academic Press, NewYork

Gray J, Thompson P (1976) Climatic information from 180/ 160 ra-tios of cellulose in tree rings. Nature 262:481-482

Hill SA, Waterhouse JS, Field EM, Switsur VR, Ap Rees T (1995)Rapid recycling of triose phosphates in oak stem tissue. PlantCell Environ 18:931-936

Lawrence JR, White JWC (1984) Growing season precipitationfrom D/H ratios of eastern white pine. Nature 311:558-560

Leaney FW, Osmond CB, Allison GB, Ziegler H (1985) Hydro-gen-isotope composition of leaf water in C3 and C4 plants: itsrelationship to the hydrogen-isotope composition of dry mat-ter. Planta 164:215-220

Leavitt SW, Danzer SR (1992) Methods for batch processingsmall wood samples to holocellulose for stable-carbon isotopeanalysis. Anal Chem 65:87-89

Lin G, Sternberg LSL (1993) Hydrogen isotope fractionation byplant roots during water uptake in coastal wetland plants. In:Ehleringer JR, Hall AE, Farquhar GID (eds) Stable isotopesand plant carbon-water relations. Academic Press, London, pp497-510

Lipp J, Trimborn P, Graff W, Becker B (1993) Climatic influenceof D/H ratios in the cellulose of late wood in tree rings fromspruce (Picea abies L.). In: Isotope techniques in the study ofpast and current environmental change in the hydropshere andatmosphere. IAEA, Vienna, pp 395-405

Loader NJ, Switsur VR, Field EM (1995). High resolution stableisotope analysis of tree rings: implications of `microdendrocli-matology' for paleoenvironmental research. Holocene5:457-460

Luo YH, Sternberg L (1992) Hydrogen and oxygen isotope frac-tionation during heterotrophic cellulose synthesis. J Exp Bot43:47-50

Roden JS, Ehleringer JR (1999) Hydrogen and oxygen isotope ra-tios of tree-ring cellulose for riparian trees grown long-termunder hydroponically controlled environments. Oecologia121:467-477

Roden JS, Ehleringer JR (1999b) Observations of hydrogen andoxygen isotopes in leaf water confirm the Craig-Gordon modelunder wide-ranging environmental conditions. Plant Physiol120:1165-1173

Roden JS, Lin G, Ehleringer JR (2000) A mechanistic model forthe interpretation of hydrogen and oxygen isotope ratios intree-ring cellulose. Geochim Cosmochim Acta 64:21-35

Rundel PW, Jarrell WM (1989) Water in the environment. In:Pearcy RW, Ehleringer JR, Mooney HA, Rundel PW (eds)Plant physiological ecology: field methods and instrumenta-tion. Chapman & Hall, London, pp 29-56

Saurer M, Aellen K, Siegwolf R (1997) Correlating 8 13C and 6180in cellulose of trees. Plant Cell Environ 20:1543-1550

Saurer M, Robertson I, Siegwolf R, Leuenberger M (1998) Oxy-gen isotope analysis of cellulose: an inter-laboratory compari-son. Anal Chem 70:2074-2080

Schiegl WE (1974) Climatic significance of deuterium abundancein growth rings of Picea. Nature 251:582-584

Socki RA, Karlsson HR, Gibson EK (1992) Extraction techniquefor the determination of oxygen-18 in water using pre-evacu-ated glass vials. Anal Chem 64:829-831

Sternberg LSL (1989) Oxygen and hydrogen isotope ratios inplant cellulose: mechanisms and applications. In: Rundel PW,Ehleringer JR, Nagy KA (eds) Stable isotopes in ecological re-search. Springer, New York, pp 24-141

Sternberg LSL, DeNiro MJ (1983) Biogeochemical implicationsof the isotopic equilibrium fractionating factor between oxy-gen atoms of acetone and water. Geochim Cosmochim Acta47:2271-2274

Sternberg LSL, DeNiro MJ, Savidge RA (1986) Oxygen isotope ex-change between metabolites and water during biochemical reac-tions leading to cellulose synthesis. Plant Physiol 82:423-427

Switsur R, Waterhouse J (1998) Stable isotopes in tree ring cellu-lose. In: Griffiths H (ed) Stable isotopes: integration of biolog-ical, ecological and geochemical processes. BIOS, Oxford, pp303-321

Terwilliger VJ, DeNiro MJ (1995) Hydrogen isotope fractionationin wood-producing avocado seedlings: biological constraintsto paleoclimatic interpretation of SD values in tree ring cellu-lose nitrate. Geochim Cosmochim Acta 59:5199-5207

Wang X-F, Yakir D (1995) Temporal and spatial variations in theoxygen-18 content of leaf water in different plant species.Plant Cell Environ 18:1377-1385

White JWC, Gedzelman SID (1984) The isotopic composition ofatmospheric water vapor and the concurrent meteorologicalconditions. J Geophys Res 89:4937-4939

White JWC, Cook ER, Lawrence JR, Broecker WS (1985) TheD/H ratio of sap in trees: implications for water sources andtree ring D/H ratios. Geochim Cosmochim Acta 49:237-246

White JWC, Lawrence JR, Broecker WS (1994) Modeling and in-terpreting D/H ratios in tree rings: a test case of white pine innortheastern United States. Geochim Cosmochim Acta 58:851-862

Yakir D (1992) Variations in the natural abundance of oxygen-18and deuterium in plant carbohydrates. Plant Cell Environ15:1005-1020

Yakir D, DeNiro MJ (1990) Oxygen and hydrogen isotope frac-tionation during cellulose metabolism in Lemna gibba L. PlantPhysiol 93:325-332

Yapp CJ, Epstein S (1982) A reexamination of cellulose carbon-bound hydrogen SD measurements and some factors affectingplant-water D/H relationships. Geochim Cosmochim Acta46:955-965