Relative humidity recorded in tree rings: A study along a precipitation gradient in the Olympic Mountains, Washington, USA

Geochimica et Cosmochimica Acta, Vol. 69, No. 4, pp. 791–799, 2005 Copyright © 2005 Elsevier Ltd Printed in the USA. All rights reserved 0016-7037/05 $30.00 ⫹ .00

doi:10.1016/j.gca.2004.08.013

Relative humidity recorded in tree rings: A study along a precipitation gradient in the Olympic Mountains, Washington, USA YONG SHU,1,* XIAHONG FENG,1 CAREY GAZIS,2 DOUG ANDERSON,2 ANTHONY M. FAIIA,1 KUILIAN TANG,1,3 and GREGORY J. ETTL4 2

1 Department of Earth Sciences, 6105 Fairchild, Dartmouth College, Hanover, NH 03755, USA Department of Geological Sciences, Central Washington University, Ellensburg, WA 98926-7418, USA 3 Ontar Corporation, 9 Village Way, North Andover, MA 01845, USA 4 Warren Wilson College, P.O. Box 9000, Asheville, NC 28815-9000, USA

(Received January 20, 2004; accepted in revised form August 5, 2004)

Abstract—Humidity is one of the fundamental variables controlling the energy balance of the climate, and thus its reconstruction represents a significant aspect of paleoclimate studies. We here report a study of oxygen and hydrogen isotopic compositions of tree-ring cellulose along a precipitation gradient in the Olympic Mountains of Washington, USA, to examine whether and how the relative humidity is recorded in the cellulose. According to the physically-based model described in this paper, the relationship between ␦D and ␦18O of cellulose should have the same slope as that in the relationship between ␦D and ␦18O in the source water (⬃8.8 for the local meteoric water) if there is no systematic variation in the mean relative humidity among study sites and no physiologic differences among trees. However, our isotopic analyses of cellulose yielded a slope of 17.4, significantly greater than the slope of the Local Meteoric Water Line. We show that to produce such a slope, a positive covariation between the relative humidity and the ␦D and ␦18O in the source water is required across the precipitation gradient. This work suggests that the ␦D vs. ␦18O relationship in tree rings can be a useful tool for paleohumidity reconstruction. Copyright © 2005 Elsevier Ltd metabolisms (Dongmann et al., 1974; Farris and Strain, 1978; Yapp and Epstein, 1982; Yakir and DeNiro, 1990; Roden et al., 2000). We use this model to argue that the relative humidity is indeed recorded in the tree-ring cellulose of the study sites.

1. INTRODUCTION

Relative humidity is related to the moisture content in the atmosphere, which plays an important role in the energy balance that determines the planetary climate. During the past two decades, efforts have been made to reconstruct relative humidity signals using oxygen or hydrogen isotope ratios in tree rings (Burk and Stuiver, 1981; Yapp and Epstein, 1982; Edwards et al., 1985; Edwards and Fritz, 1986), carbon isotopic compositions of stem cellulose (Lipp et al., 1996; Edwards et al., 2000), and excess air in groundwater (Anovitz et al., 1999; Aeschbach-Hertig et al., 2002; Beyerle et al., 2003). In tree-rings studies, many results from published studies gave empiric correlations between the isotopic compositions and the relative humidity, which could only be applied to the specific sites studied. Furthermore, few earlier efforts for paleohumidity reconstruction using oxygen and hydrogen isotopes in tree rings did not consider the processes of postphotosynthetic isotope exchange between simple carbohydrates and the stem water (Yapp and Epstein, 1982; Edwards et al., 1985; Edwards and Fritz, 1986; Jahren and Sternberg, 2003). These processes, referred to as heterotrophic carbohydrate metabolism (Yakir and DeNiro, 1990; Luo and Sternberg, 1992), are associated with isotopic fractionation and must be considered when interpreting isotopic data with respect to changes in the relative humidity. In this paper, we report our study of the ␦D and ␦18O in tree rings collected along a precipitation gradient in the Olympic Mountains of Washington, USA. We present a physicallybased model that incorporates processes of transpiration and biochemical fractionation in both autotrophic and heterotrophic

* Author to whom correspondence ([email protected]).

should

be

2. SAMPLES AND METHODS Tree-ring samples were collected along a precipitation gradient (Tang et al., 2000) in the Olympic Mountains, which are located in the Northwest United States (Fig. 1). The mountains intercept moisturerich Pacific air from the west side, creating a rain shadow on the east side; as a result, some areas along the windward slope of the Olympic Peninsula receive the heaviest precipitation in the continental United States. Annual precipitation from west to east varies from 3 m in Forks to 0.64 m in Quilcene (Fig. 1). With the decrease in precipitation, the amount of deuterium and 18O also decreases in the meteoric water (Rozanski et al., 1993; Tang et al., 2000). Five sampling sites were chosen that have a broad range of variation in the amount of annual precipitation. Twin Creek (TC) and Sole-Duck (SD) sites are located on the west side of the mountains, and receive 3.5 m and 2.5 m annual precipitation, respectively (Phillips and Donaldson, 1972). Two sites at Deer Park (DP1 and DP2) represent typical climate on the east side of peninsula; the annual precipitation is less than 1 m and soil is poorly developed. Heart-O-Hill (HH) is intermediate in annual precipitation (1.8 m) and soil development among the five sites. Douglas-fir (Pseudotsuga menziesii) grows in all sites, which is chosen to be the primary species for this study. At higher elevation sites (DP sites) subalpine fir (Abies lasiocarpa) was also collected for comparison. All sites were cross-dated using the standard method of dendrochronology (Fritts, 1976). For isotope analysis, 3– 4 tree cores of 1 cm diameter were taken at each site for each species, and those trees that were radially symmetric were used to reduce within-tree variations (Ramesh et al., 1985). Analyzed tree cores also have relatively wide rings (average ring widths of 1.16 –2.42 mm) so that a sufficient amount of cellulose can be extracted from annual rings. Annual tree rings covering 1962–1985 from one Douglas-fir tree at each site were selected for isotopic analysis. This time period was selected in the earlier work by Tang et al. (2000) to study the relationship between the ␦D values of cellulose nitrate and the ␦D values of the source water or temperature. Here we used the same tree-ring samples

addressed 791

792

Y. Shu et al.

3.2. Within-Species Variations in ␦18O Values

Fig. 1. Geographic map of sampling sites with annual precipitation amounts in parentheses.

The hydrogen isotope ratios of cellulose nitrate (␦DCN) have been reported by Tang et al. (2000), and oxygen isotope ratios of cellulose (␦18OC) obtained in this study are listed in Table 1. The variation in both ␦DCN and ␦18OC are illustrated in Figure 3. Neither ␦DCN nor ␦18OC values of tree rings show significant correlation with the age of tree rings (n ⫽ 23 or 24, p ⬎ 0.12). Two trees of the same species were analyzed at sites TC and DP1. At TC, the average ␦18OC for one Douglas-fir is 28.0‰ (TC-DF-07) and 26.7‰ for the other (TC-DF-13). The two means are significantly different using a standard t-test (p ⬍ 0.001, n ⫽ 10). The same conclusion can be made for DP1, where the means for two Douglas-fir trees differ by 2.3‰ (DP1-DF-01 ⫽ 24.7‰; DP1-DF-05 ⫽ 27.0‰; p ⬍ 0.001, n ⫽ 10). This comparison indicates that local hydrological (e.g., soil moisture) and environmental (canopy and aspect) conditions may be quite different from one tree to another. Such local variations are a major source of noise when comparing between-site variations, and must be considered when drawing conclusions about regional trends. 3.3. Between-Species Variations in ␦18O Values

from which cellulose had been extracted to study the relationship between ␦D and ␦18O of tree rings and the relative humidity. For DP1 and DP2, cores from one subalpine fir per site were also studied to examine the species difference. For two sites, TC and DP1, an additional Douglas-fir tree was analyzed for 10 rings (1965–1974) to examine the within-site variation. After separating individual rings of each core, the wood was ground with a wood mill and cellulose was extracted using the standard technique (Epstein et al., 1976, 1977; DeNiro, 1981). Oxygen isotopic compositions in cellulose were determined using a Temperature Conversion Elemental Analyzer interfaced with an Isotope Ratio Mass Spectrometer (IRMS, Delta Plus XL, Finnigan) (Farquhar et al., 1997). The analytic uncertainty is less than 0.3‰ (1␴). Hydrogen isotopic compositions of these samples were measured on cellulose nitrate and the data have been previously reported by Tang et al. (2000). The analytical uncertainty for ␦D measurements is less than 3‰ (1␴). Both oxygen and hydrogen isotope ratios are reported as the per mil difference relative to the VSMOW standard. Meteoric water samples from TC and HH sites were collected bi-weekly from 1998 to 1999. Before placing the plastic rain gauge at the station, mineral oil was added to it to prevent evaporation. Water was separated from the mineral oil using a filter paper (40 ␮m). We have confirmed that this procedure does not cause any isotopic change of the water using a laboratory water standard. Water samples were analyzed for ␦18O values using the method of CO2 equilibration (Epstein and Mayeda, 1953). The ␦D values of water samples were determined using an H-Device (Finnigan) interfaced with an IRMS. Inside the H-Device, water is reduced by hot (850°C) chromium and the H2 gas is passed to the mass spectrometer. The analytical precision is 0.1‰ (1␴) for ␦18O analysis and 0.5‰ (1␴) for ␦D analysis.

For DP1 and DP2, a subalpine-fir core was analyzed to examine between-species variations in ␦18OC values. At DP2, the mean of the subalpine fir (DP2-SAF-04 ⫽ 25.8‰) is not significantly different from the mean of the Douglas-fir (DP2DF-03 ⫽ 25.9‰, p ⫽ 0.53, n ⫽ 23), while at DP1, the two means (DP1-SAF-24 ⫽ 25.9‰ and DP1-DF-24 ⫽ 26.6‰) are significantly different (p ⫽ 0.002, n ⫽ 24; Fig. 3). In the case of DP2, the between-species difference is no greater than the within-species difference of the two Douglas-fir trees. This observation suggests that local differences in growing conditions have a greater control on ␦18OC variations than the physiologic differences in the different species. Tang et al. (2000) made the same observation for the ␦DCN variations of these trees. In the following discussion, we will combine the

3. RESULTS

3.1. Isotopic Compositions of Meteoric Water The results for analyses of the meteoric water samples are shown in Figure 2. The ␦D values of the surface water obtained by Tang et al. (2000) are also shown in the figure. Unfortunately, those surface water samples are not available for ␦18O analysis. The ␦D vs. ␦18O line has a slope 8.8 ⫾ 0.4 (SE) which we consider to be the slope of the local meteoric water line.

Fig. 2. Isotopic data from meteoric water samples collected at TC and HH sites during 1998 –1999. The ␦D measurements by Tang et al. (2000) are also shown.

Relative humidity recorded in tree rings

793

Table 1. ␦18Oc (‰) values for tree rings from five sampling sites of Olympic Mountains, WA, USA. Year 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 a

DP1-DF-01

25.1 25.2 23.8 24.1 24.5 25.3 24.8 25.4 n.a.a 23.7

DP1-DF-05

DP1-SAF-01

DP2-DF-03

DP2-SAF-04

HH-DF-23

SD-DF-22

TC-DF-13

25.1 25.6 25.0 27.4 26.6 27.3 26.0 26.0 27.7 26.8 27.9 27.7 26.2 26.4 25.1 26.6 27.3 27.0 26.1 27.6 26.1 26.7 26.2 28.4

26.2 25.4 25.2 26.6 26.0 26.0 25.7 25.5 26.2 25.8 27.0 26.6 25.1 25.2 24.7 25.6 25.8 26.3 25.3 26.1 25.8 25.5 25.6 27.3

n.a. 26.5 26.2 26.2 25.4 25.9 26.0 25.1 26.1 25.8 26.2 26.5 25.6 25.6 25.0 26.6 26.1 26.4 25.9 25.7 25.6 24.6 25.6 27.4

25.9 26.4 25.4 26.1 24.9 24.8 25.6 25.8 25.9 24.8 25.9 25.7 28.7 25.7 25.0 25.6 25.7 25.9 25.6 25.8 25.5 24.4 25.8 27.4

26.5 26.1 26.5 27.1 27.1 27.9 27.7 26.6 27.4 27.5 27.8 28.1 26.8 26.7 26.1 n.a. 26.5 27.2 25.8 26.6 30.4 26.8 26.6 28.7

26.9 26.9 26.7 26.7 25.9 27.2 26.5 27.5 28.0 27.4 27.6 27.8 26.3 26.6 26.4 27.0 26.6 26.9 26.5 26.3 26.4 26.6 26.9 28.3

25.9 26.6 26.4 26.8 26.4 25.8 26.7 26.5 26.6 25.9 27.3 27.1 27.5 27.6 27.3 27.9 26.7 27.0 27.8 27.4 27.2 27.8 28.5 28.7

TC-DF-07

28.0 27.8 28.0 27.6 28.0 28.6 27.4 28.5 28.3 27.6

Not available

data for Douglas-fir and subalpine-fir trees, assuming that isotopic variations due to physiologic differences between species are insignificant. 3.4. Between-Site Variations in ␦18O Values and the ␦DCN-␦18O Relationship Figure 3b shows that, with the exception of DP1-5, the mean value of ␦18OC is generally higher in trees on the west side of the Olympic Mountains than those on the east side. This trend is consistent with that of the means of ␦DCN values (Fig. 3a). Tang et al. (2000) interpreted this ␦DCN trend to reflect the variation in the ␦D value of meteoric water. Since ␦18O in precipitation correlates with the ␦D following the meteoric water line (Fig. 2), one would expect the trends for mean ␦18OC and the mean ␦DCN to be the same. There is no significant correlation between ␦18OC and ␦DCN within each tree for all series. There is, however, a weak correlation between the mean ␦18OC and the mean ␦DCN values of each tree with an r2 value of 0.36 (n ⫽ 9, p ⬍ 0.09). A significant correlation exists for the total data set of individual rings (Fig. 4, r2 ⫽ 0.16, p ⬍ 0.001, n ⫽ 179). To obtain the quantitative relationship between ␦18OC and ␦DCN, we need to consider the fact that neither variable can be treated as the independent variable. The two variables covary because they are both controlled by the meteorological, hydrological and biologic conditions. In addition, sampling and analytical error exist for both variables, and we could not assume either of them to be error-free. We, therefore, obtained the relationship between ␦18OCN and ␦DCN using method of the Reduced Major Axis (RMA) (Sokal and Rohlf, 1981). First, taking the ␦18OC as the independent variable, we obtained the slope of 7.0 for the regression of ␦DCN against ␦18OC. Then, we

took ␦DCN as the independent variable and the corresponding regression slope is 43.4. The line whose slope is the geometric average of two regression slopes above is taken to represent the covariation of ␦18OCN and ␦DCN. This line has a slope of 17.4 ⫾ 1.2 (SE) and the r2 value remains the same (0.16). Note that this slope is significantly greater than the slope of the local meteoric water line (8.8 ⫾ 0.41; Fig. 2). The uncertainty of the slope was checked with 5000 replicate bootstraps, which yielded 15.2 to 19.9 as the 95% confidence interval. 4. DISCUSSION

The objective of this work is to search for quantitative relationship between air relative humidity and the oxygen and hydrogen isotopic compositions of tree rings. To interpret the data we first describe a model that considers the effects of the relative humidity and physiologic properties of cellulose synthesis on the ␦18OC and ␦DCN values of tree rings. We then use this model to explain our observations of oxygen and hydrogen isotopes. 4.1. Model The commonly-used model for isotopic variations in tree rings combines the isotopic treatment of leaf water and biologic fractionation in autotrophic processes. Craig and Gordon

1

Conceptually, the slope of the local meteoric water line should also be obtained using the RMA technique. However, since the linear relationship between the precipitation ␦18O and ␦D is very tight, the slope yielded from RMA method (9.1 ⫾ 0.4) is not significantly different from the slope yielded from the conventional linear regression. We here use the conventional linear regression for the precipitation ␦18O and ␦D, which is usually done in literature.

794

Y. Shu et al.

hydrates, and it is the sole source of hydrogen. Both CO2 and H2O contribute oxygen atoms to the photosynthetic products. However, the oxygen from CO2 exchanges completely with H2O before carbohydrate synthesis (DeNiro and Epstein, 1979). Therefore, hydrogen and oxygen isotopic compositions of the photosynthetic products isotopically depend on leaf water. Consequently, the relative humidity information carried by leaf water is transferred to the photosynthetic products. After photosynthesis, the carbohydrate products are transported from leaves to other parts of the tree. Before stem cellulose synthesis, a portion of the photosynthetic product experiences further isotopic exchange with the stem water. Both photosynthesis and the subsequent isotope exchange between carbohydrate and stem water are associated with isotopic fractionation, and these fractionations are referred to as autotrophic and heterotrophic fractionations, respectively (Yakir and DeNiro, 1990). If a fraction, f, of carbohydrate experiences postphotosynthetic isotopic exchange, the ␦18O of cellulose is (Roden et al., 2000), Fig. 3. Variations of (a) ␦DCN (from Tang et al., 2000) and (b) ␦ OC values of tree rings (this work). The tree names are abbreviated from Table 1; -S indicates the species subapline fir, and other trees are Douglas-fir. The central line of each diamond is the mean value of ␦DCN or ␦18OC, the top and bottom of the diamond bracket the 95% confidence interval for the mean, and two lines on either side of the central line bracket the 95% overlap range for between-tree comparisons. Two means are significantly different if their overlap ranges do not overlap. There is a systematic decrease in the mean ␦DCN value along the precipitation gradient from the west to the east side of the Olympic Mountains; except for DP1-5, the mean ␦18OC value shows the same trend. 18

(1965) first developed the classical model for evaporative enrichment of deuterium and 18O of a water body subject to free evaporation. Dongmann et al. (1974) modified this model to describe the variation of the isotopic composition in leaf water during leaf transpiration by assuming that the leaf water is in isotopic steady-state. Dongmann’s model was later refined and used by Farris and Strain (1978) and Yapp and Epstein (1982), and this modified model will be used in this work. Assuming that evaporation is the only process by which water is lost from leaves, the isotopic compositions of leaf water are described by the following equation (Craig and Gordon, 1965; Dongmann et al., 1974; Roden et al., 2000):

␦18OC ⫽ fO(␦18Oi ⫹ ␧OH) ⫹ (1 ⫺ fO)(␦18Olw ⫹ ␧OA)

(3)

and the ␦D of cellulose nitrate is

␦DCN ⫽ fD(␦Di ⫹ ␧DH) ⫹ (1 ⫺ fD)(␦Dlw ⫹ ␧DA)

(4)

where the subscripts O, D, A and H represents oxygen, deuterium, autotrophic and heterotrophic, respectively; ␦18Oi and ␦Di are the isotopic compositions of the source water; ␧ is the biologic isotopic discrimination. The autotrophic biologic discrimination is ⫺170‰ for hydrogen and 27‰ for oxygen, and the heterotrophic biologic discrimination is 150‰ for hydrogen and 27‰ for oxygen (Sternberg, 1989; Yakir and DeNiro, 1990; Luo and Sternberg, 1992). To examine how oxygen and hydrogen isotopic compositions in tree rings are related to the isotopic compositions of the

␦lw ⫽ ␣eq␣k(1⫺h)(1000 ⫹ ␦i)⫹ ␣eqh(1000 ⫹ ␦a)⫺1000 (1) where ␦ can be either ␦18O or ␦D; the subscripts lw, a and i represent leaf water, ambient atmospheric water vapor and stem water (source water), respectively; ␣eq and ␣k are the equilibrium and kinetic fractionation factors between liquid water and water vapor, respectively; h is the relative humidity expressed in this work as a fraction from 0 to 1. Assuming that ␦a is in isotopic equilibrium with the source water, the above equation can be rewritten as:

␦lw ⫽ ␣eq␣k(1 ⫺ h)(1000 ⫹ ␦i) ⫹ h(1000 ⫹ ␦i) ⫺ 1000 (2) Eqn. 2 indicates that the relative humidity signal is expressed in the isotopic signature of leaf water due to leaf transpiration. During photosynthesis, leaf water is used to synthesize carbo-

Fig. 4. Plot of ␦DCN against ␦18OC values for all individual tree rings. The slope (17.4 ⫾ 1.2) obtained using the RMA method (see text for more explanation) is considered as the linear relationship between ␦DCN and ␦18OC (the dark line). Other two gray lines are the regression lines of ␦DCN against ␦18OC (slope ⫽ 7.0) and ␦18OC against ␦DCN (slope ⫽ 43.4), respectively.

Relative humidity recorded in tree rings

795

surface, and because of lack of information, we assume that there is no systematic difference in these conditions among sites. Therefore, if there is no systematic variation in the mean relative humidity and no physiologic differences (f values) among sites, we would expect the ␦DCN-␦18OC relationship to define a straight line with a slope close to Ai (S ⬇ Ai). Figure 5 shows the result of calculations of S as a function of h and f (with fO ⫽ fD) assuming that the source water follows the local meteoric water line with a slope of 8.8. Even at the extreme ranges of variations in f and h, the S value only changes from 8.8 to 9.5. Our experimental data shows, however, that the ␦DCN-␦18OC relationship has a slope 17.4 ⫾ 1.2, which is significantly greater than the slope of the meteoric water line (8.8 ⫾ 0.4). This result suggests that the assumption that h and f are constant among sites is likely to be invalid. 4.2. Effect of Soil Hydrological Processes Fig. 5. Contours of the slope for the linear relationship between ␦DCN and ␦18OC. Each calculation of the slope is made by using assigned h and f (fD ⫽ fO) values. Both parameters vary from 0 to 1. The slope is always close to 8.8, which is the slope of the ␦D against ␦18O relationship in the local meteoric water. The values of model parameters are: D␣eq ⫽ 1.0934, D␣k ⫽ 1.025, 18␣eq ⫽ 1.0109, 18␣k ⫽ 1.0160 (Majoube, 1971; Merlivat, 1978), ␧DA ⫽ ⫺170‰, ␧DH ⫽ 150‰, ␧OA ⫽ 27‰, and ␧OH ⫽ 27‰ (Sternberg, 1989; Yakir and DeNiro, 1990; Luo and Sternberg, 1992).

source water, we consider that the isotopic compositions of source water follow a straight line as,

␦Di ⫽ Ai␦18Oi ⫹ Bi

(5)

As mentioned earlier, Ai is ⬃8.8 and Bi is ⬃18.8 for the local meteoric water in this study. Combining Eqns. 2–5, it can be shown that, if both h and f are held constant while the ␦D and ␦18O changes along the line (5), the following relation exists.

␦DCN ⫽ S␦18OC ⫹ I

(6)

where S ⫽ A(AD ⁄ AO)

and

I ⫽ AD[Bi ⫺ AiBO ⁄ AO] ⫹ BD , (7)

AD ⫽ fD ⫹ h(1 ⫺ fD) ⫹ (1 ⫺ fD)(1 ⫺ h)D␣eqD␣k ,

(8)

The meteoric water is not the source water for trees, because the isotopic composition may change within the soil column due to mixing of water from different storms and soil evaporation. The best way to determine the isotopic composition of source water is to collect and analyze soil water or stem water. Unfortunately, these waters are not available. However, what is important for this study is the maximum slope of the ␦D-␦18O relationship in the source water, which can be estimated with confidence. Mixing of meteoric water from different storms does not change the slope of the ␦D-␦18O relationship in the soil water from that of the local meteoric water line. If the mixing process combined water with two distinct isotopic compositions each from the local meteoric water line, the resulting water would have the isotopic compositions sitting along the same meteoric water line between the two end members. Soil evaporation does change the slope of the ␦D-␦18O relationship in the soil water. However, this process tends to decrease the slope rather than increase it because evaporation causes a greater enrichment in 18 O than in D of the source water compared to equilibrium fractionation. This has been observed in both lake water and soil water (Allison, 1982; Barnes and Allison, 1988; Tang and Feng, 2001) that has been isotopically modified by evaporation. We conclude, therefore, that the high slope observed in tree rings is not caused by soil water evaporation.

BD ⫽ 1000(1 ⫺ fD)(1 ⫺ h)(D␣eqD␣k ⫺ 1) ⫹ fD␧DH ⫹ (1 ⫺ fD)␧DA, (9) AO ⫽ fO ⫹ h(1 ⫺ fO) ⫹ (1 ⫺ fO)(1 ⫺ h)18␣eq18␣k ,

(10)

BO ⫽ 1000(1 ⫺ fO)(1 ⫺ h)(18␣eq18␣k ⫺ 1) ⫹ fO␧OH ⫹ (1 ⫺ fO)␧OA. (11) The model predicts that the slope (S) in Eqn. 6 is close to Ai for fixed values of h, fO, fD, ␣eq, ␣k and the four ␧ parameters. This holds true for the entire range of possible values for h and f, i.e., from 0 to 1 (Fig. 5). The variation of ␣eq is a function of temperature, which varies little among sites. The mean annual temperature varies from 9.6 to 10.7°C among weather stations shown in Figure 1 (Tang et al., 2000). Therefore, we assume constant ␣eq. The parameter ␣k is related to aerodynamic conditions near the leaf

4.3. Effect of Relative Humidity To examine the effect of relative humidity, we here assume that the fraction of postphotosynthetic exchange, f, is constant for both oxygen and hydrogen. We consider two stages of isotopic changes. First, the isotopic composition of stem water (␦18Oi, ␦Di) evolves to that of leaf water (␦18Olw, ␦Dlw), following (2). Second, the leaf water is transferred isotopically to cellulose (␦18OC, ␦DCN), following (3) and (4). We then consider the relationship between ␦18OC and ␦DCN given by (6) through (11). We show these two stages in Figure 6 for two possible situations, 1) the relative humidity is positively correlated with ␦18Oi and ␦Di (Fig. 6a) and 2) the relative humidity is negatively correlated with ␦18Oi and ␦Di (Fig. 6b). Figure 6 indicates a counterclockwise rotation of the ␦DCN-␦18OC slope for case 1) and a clockwise rotation for case 2). In other words,

796

Y. Shu et al.

higher by at least 7‰. More specifically, we assume that the relative humidity is a linear function of the ␦D and ␦18O of the meteoric water. Using the relationships in Figure 7a, we can simulate the observed ␦DCN-␦18OC slope of 17.4 in tree-ring cellulose (Fig. 7b). In this calculation, assumed f to be constant, and a relative humidity variation from 30 to 55% (0.3 to 0.55 in the model) along the precipitation gradient was required. Observations of the growing season (June 1–October 15, averaged from 1970 to 1990) relative humidity from a number of fire stations in Washington State show some variations of daily minimum relative humidity (only minimum h were available) along the precipitation gradient. For example, at Forks on the west side of the Olympic Mountains the minimum relative humidity is from 40 to 50% and at Jefferson Ridge on the east is from 30 to 40% 40 to 50%. This ⬃10% change in the relative humidity is smaller than required range 25% predicted by our model.

Fig. 6. Changes of the isotopic composition, from the source water to leaf water and to cellulose, in the process of cellulose formation. The source water has a slope of 8.8, but the slope of the ␦DCN vs. ␦18OC relationship is no longer 8.8 if the relative humidity systematically covaries with the isotopic compositions of the source water. (a) Counterclockwise rotation is predicted if the relative humidity positively covaries with the ␦D and ␦18O of the source water (h1 ⫽ 40% and h2 ⫽ 55%). (b) Clockwise rotation is predicted if the relative humidity negatively covaries with the ␦D and ␦18O of the source water (h1 ⫽ 40% and h2 ⫽ 30%). In the calculation, fD ⫽ fO ⫽ 0.35 and other model parameters are the same as those used for Figure 5.

if the relative humidity is positively correlated with the ␦18O or ␦D values in meteoric water, the slope of the ␦DCN-␦18OC relationship is greater than the slope of the source water or vice versa. In our study, we observed that the ␦D of meteoric water decreased along the precipitation gradient in the Olympic Mountains, from the wet west to the dry east (Tang et al., 2000). It is reasonable to assume that the relative humidity also decreases from the west to the east. We further hypothesize that the observed counterclockwise rotation in the ␦DCN-␦18OC line is a result of the humidity change along the precipitation gradient. If all sites were as wet as TC, the ␦D vs. ␦18O relationship would start from the current upper right position, moving down to the lower left along a line with a slope close to 8.8. Instead, this line rotated counterclockwise by ⬃20o (Fig. 7b). As a result, ␦18O values of cellulose at the dry sites are

Fig. 7. Model simulation of the observed ␦DCN and ␦18OC variations. (a) Assumed relationships between the relative humidity and the ␦D and ␦18O values in the source water. (b) The simulated slope for the ␦DCN-␦18OC relationship using the lines in a) and holding fO and fD constant. The range of ␦D values in the source water is from ⫺104 to ⫺50‰ and the corresponding ␦18O values are calculated from the local meteoric water line. The data for the meteoric water (solid squares) and tree rings (solid circles) are both shown in the figure. Model parameters are the same as in Figure 6.

Relative humidity recorded in tree rings

4.4. Effects of Postphotosynthetic Isotope Exchange We now examine the effect of the fractions of postphotosynthetic isotope exchange, fD and fO, on the slope of the ␦DCN-␦18OC relationship. Variations of fD and fO are determined by the metabolic processes of cellulose synthesis. To produce cellulose, photosynthetic sucrose is first hydrated to form hexose phosphate (Farquhar et al., 1998). A portion of the hexose phosphate is directly incorporated into cellulose, through which ⬃20% of oxygen undergoes isotopic exchange with xylem water (Farquhar et al., 1998). The rest of hexose phosphate cycles with triose phosphates before the former is incorporated into cellulose (Hill et al., 1995). In this hexose phosphate-triose phosphates cycle, six out of 10 oxygen atoms exchange with the medium water (Farquhar et al., 1998). The overall fraction (fO) of heterotrophic exchange of oxygen is the sum of the exchangeable oxygen atoms in the above two pathways, and it was measured by Roden et al. (2000) to be 0.42 and by Yakir and DeNiro (1990) to be 0.35. There are fewer biochemical studies of postphotosynthetic exchange of hydrogen in cellulose. It is likely that processes or pathways discussed for oxygen also affect the hydrogen isotopic compositions. Empirically, Roden et al. (2000) reported 0.36 for fH and Yakir and DeNiro (1990) reported 0.35. Our model can be used to examine how variations of fD and fO affects the slope of the ␦DCN-␦18OC relationship. We discovered, that when holding the relative humidity constant, it is not possible to account for the rotation of the ␦DCN-␦18OC slope from 8.8 to the observed 17.4 by varying fD and fO values alone. A certain level of relative humidity variation is also required. When the range of fD is chosen to be 0.35– 0.36 and of fO 0.35– 0.40, both increasing with humidity, we could produce the slope of 17.4 when relative humidity changes from 30 to 48% from the east to the west side. This range of humidity change is still higher than the observed range of the minimum relative humidity variations from Forks and Jefferson Ridge (ca. 10% change). It is not clear at this point whether variations in f values play a role in determining the slope of the ␦DCN-␦18OC relationship. Based on our model, it is theoretically possible that both variations in f and h result in the rotation of the slope. We emphasize, however, that for these parameters to affect the slope of the ␦DCN-␦18OC relationship, their variation must be systematically related to the variations in the ␦D or ␦18O of the source water. If, on the other hand, the variation of f or h were independent of that of meteoric water, it would be expressed in tree rings only as noise, causing broadening of the data array in the ␦DCN against ␦18OC plot rather than a rotation of the slope. In this study area, the systematic covariation exists between relative humidity and the ␦D/␦18O of the source water, but the same statement cannot be made for the fraction of postphotosynthetic exchange. More physiologic studies of trees are necessary to resolve this question.

797

⌬E (⌬E ⫽ ␦trans ⫺ ␦stem) and the isotopic enrichment between the whole leaf and the stem water is ⌬L (⌬L ⫽ ␦lw ⫺ ␦stem), then ⌬E ⫽ ⌬L. In addition, this isotopic enrichment is independent of the transpiration rate. It has been reported, however, that the leaf water at the site of transpiration is more enriched in 18O than the whole leaf water (Walker et al., 1989; Yakir et al., 1989, 1994; Farquhar and Lloyd, 1993; Wang and Yakir, 1995; Barbour et al., 2000). Some suggest that this phenomenon results from isotopic heterogeneity in the leaf; some fraction of leaf water is not directly connected to evaporation sites and retains some isotopic signature of the stem water (Yakir et al., 1989; Yakir et al., 1994; Roden and Ehtleringer, 1999). However, this mechanism does not explain the observation that the difference between ⌬E and ⌬L increases with the transpiration rate (Walker et al., 1989; Farquhar and Lloyd, 1993). White (1989) provided a model to explain this relationship, which considers back diffusion of the leaf water at the transpiration site to the whole leaf as a result of an isotopic gradient between the stem water entering the leaf and the enriched water at the transpiration site. The average isotopic enrichment in leaf water at a steady state was given by Farquhar and Lloyd (1993) as, ⌬L ⫽ ⌬E

1 ⫺ e⫺p p

,

(12)

where p ⫽ (EL/CD), where p (dimensionless) is known as Péclet number, E (mol m⫺2 s⫺1) is the transpiration rate, L (m) is the average effective length, C (⬃5.56 ⫻ 104 mol m⫺3) is the concentration of water and D (m2 s⫺1) is the diffusivity of HDO (or H218O) in water. Using Eqn. 12, it is apparent that the difference between ⌬E and ⌬L increases with the Péclet number and thus the transpiration rate. In other words, as the transpiration rate increases, the whole leaf water becomes increasingly less enriched in deuterium and 18O compared to the water at the transpiration site, the isotopic compositions of which are predicted by the standard stomatal boundary layer model. Applying Eqn. 12 to Eqn. 2, we can investigate how the transpiration rate of leaves influences the isotopic enrichment of the whole leaf water, which in turn affects the slope rotation of the ␦DCN-␦18OC relationship in tree-ring cellulose. We assume that the transpiration rate increases with the relative humidity over the precipitation gradient such that the ⌬L/⌬E ratio varies from 0.85 on the west side of the Olympic Mountains and to 0.95 on the east side (White, 1989). Given this transpiration rate gradient and a condition of constant postphotosynthetic isotopic exchange (constant fD and fO), a 12% variation in the relative humidity can explain the observed slope of 17.4 in tree-ring cellulose. This range is very close to the relative humidity difference between Forks and Jefferson, WA (10%). 4.6. Significance

4.5. Effect of Transpiration Rate According to the standard stomatal boundary layer model (Eqn. 2, the isotopic composition of the leaf water at the site of transpiration represents the isotopic composition of the whole leaf water. If the isotopic enrichment at the transpiration site is

Both oxygen and hydrogen isotopic compositions in tree rings have been used to reconstruct the isotopic compositions of the source water that is systematically related to the climatic variables such as temperature (Yapp and Epstein, 1982; Gray and Song, 1984; Feng and Epstein, 1994; Feng and Epstein,

798

Y. Shu et al.

1995). Such a reconstruction using either ␦DCN or ␦18OC values alone may be biased if there exists a covariation between the isotopic composition of source water and the relative humidity. For two populations of trees using the source water having the same ␦Di or ␦18Oi values the ␦DCN or ␦18OC would not be the same between the two populations if they grew under different prevailing relative humidities. Our work demonstrates that a covariation between the ␦Di or ␦18Oi in the source water and the relative humidity causes the slope of the ␦DCN-␦18OC relationship to differ from the slope of the source water (slope of ␦Di-␦18Oi relationship). If the relative humidity is positively associated with the ␦Di or ␦18Oi, the slope of the ␦DCN-␦18OC relationship would increase from that of the source water, i.e., a counterclockwise rotation, and vice versa. The direction of the rotation, clockwise or counterclockwise, provides a way of inferring the change of relative humidity in relation to the change in the isotopic composition of the source water. In this study the relative humidity is positively correlated with the ␦D/␦18O of the source water and the rotation is counterclockwise. Conversely, a clockwise rotation corresponding to a slope smaller than that of the source water would be predicted if the relative humidity is negatively associated with ␦D and ␦18O of the source water. Such a prediction may be tested in coastal areas of East Asia, where the ␦D/␦18O of meteoric water is negatively related to precipitation due to the effect of monsoon circulations (Araguás-Araguás et al., 1998; Wang et al., 2001). In this study, the rotation of the slope in the ␦DCN-␦18OC relationship is observed for spatially distributed locations having different prevailing relative humidity. The same argument can be made for two or more tree-ring populations that are temporally related at the same location, particularly when comparing the modern trees with ancient trees. In this case, the rotation of the slope would represent covariation of humidity with the meteoric ␦D and ␦18O over time.

5. CONCLUSION

We have studied the relationship of ␦18O and ␦D in tree-ring cellulose along a precipitation gradient in Olympic Mountains, Washington, USA, to examine whether the relative humidity conditions along this gradient is systematically recorded in the ␦18O and ␦D values in tree rings. We have found that ␦18OC and ␦DCN for all tree rings are significantly correlated with a slope 17.4 ⫾ 1.2 that is significantly greater than the slope of the local meteoric water line (8.8 ⫾ 0.4). We developed a physically based model that describes the oxygen and hydrogen isotopic change from the source water to leaf water and to stem cellulose. The model predicts that, if both the relative humidity (h) and the fractions of postphotosynthetic exchange (fO and fD) are held constant across all sampling sites, the plot of ␦DCN against ␦18OC would define a straight line with a slope similar to the slope of the source water (8.8). Using the model we show that the increased slope is likely caused by a positive covariation of the relative humidity with ␦D/␦18O in the source water. A decrease in relative humidity affects the isotopic compositions of tree rings in two ways, 1) by increasing the isotopic enrichment in the leaf water at the transpiration site, and 2) by decreasing the transpiration

rate, thus enhancing the isotopic enrichment in the whole leaf water. The effect of the degree of the postphotosynthetic isotopic exchange on the functional relationship between ␦DCN and ␦18OC is not clear from our current knowledge of tree physiology. If the fraction of the exchange is systematically related to the relative humidity, it would affect the slope of the ␦DCN vs. ␦18OC relationship. However, there is not yet any evidence for this covariation. Acknowledgments—This research is partially supported by the National Science Foundation (ATM-9628759) and National Oceanic and Atmospheric Administration of the United States (NA96GP0480). The manuscript has been significantly improved by the comments of Hope Jahren and Michael Evans. Associate editor: D. R. Cole REFERENCES Aeschbach-Hertig W., Beyerle U., and Kipfer R. (2002) Excess air in groundwater as a proxy for paleo-humidity. Geochim. Cosmochim. Acta 66 (15A, Suppl.), A8 –A8. Allison G. B. (1982) The relationship between 18O and deuterium in water in sand columns undergoing evaporation. J. Hydrol. 55, 163–169. Anovitz L. M., Elam J. M., Riciputi L. R., and Cole D. R. (1999) The failure of obsidian hydration dating: Sources, implications and new directions. J. Archaeol. Sci. 26, 735–752. Araguás-Araguás L., Froehlich K., and Rozanski K. (1998) Stable isotope composition of precipitation over Southeast Asia. J. Geophys. Res. 103 (D22), 28721–28742. Barbour M. M., Schurr U., Henry B. K., Wong S. C., and Farquhar G. D. (2000) Variation in the oxygen isotope ratio of phloem sap sucrose from castor bean. Evidence in support of the Péclet effect. Plant Physiol. 123 (2), 671– 679. Barnes C. J. and Allison G. B. (1988) Tracing of water movement in the unsaturated zone using stable isotopes of hydrogen and oxygen. J. Hydrol. 100, 143–176. Beyerle U., Rueedi J., Leuenberger M., Aeschbach-Hertig W., Peeters F., Kipfer R., and Dodo A. (2003) Evidence for periods of wetter and cooler climate in the Sahel between 6 and 40 kyr BP derived from groundwater. Geophys. Res. Lett. 30 (4), 1173. Burk R. L. and Stuiver M. (1981) Oxygen isotope ratios in trees reflect mean annual temperature and humidity. Science 211 (4489), 1417– 1419. Craig H. and Gordon L. I. (1965) Deuterium and oxygen-18 variations in the ocean and the marine atmosphere. In Stable Isotopes in Oceanographic Studies and Paleotemperatures (ed. E. Tongiorgio), pp. 9 –130. DeNiro M. J. (1981) The effects of different methods of preparing cellulose nitrate on the determination of the D/H of non-exchangeable hydrogen of cellulose. Earth Planet. Sci. Lett. 54, 177–185. DeNiro M. J. and Epstein S. (1979) Relationship between the oxygen isotope ratios of terrestrial plant cellulose, carbon dioxide and water. Science 204, 51–53. Dongmann G., Nurnberg H. W., Forstel H., and Wagener K. (1974) On the enrichment of H218O in the leaves of transpiring plants. Radiat. Environ. Biophys. 11, 41–52. Edwards T. W. D., Aravena R. O., Fritz P., and Morgan A. V. (1985) Interpreting paleoclimate from 18O and 2H in plant cellulose: Comparison with evidence from fossil insects and relict permafrost in southwestern Ontaio. Can. J. Earth Sci. 22, 1720 –1726. Edwards T. W. D. and Fritz P. (1986) Assessing meteoric water composition and relative humidity from 18O and 2H in wood cellulose: Paleoclimatic implications for southern Ontario, Canada. Appl. Geochem. 1, 715–723. Edwards T. W. D., Graf W., Trimborn P., Stichler W., Lipp J., and Payer H. D. (2000) d13C response surface resolves humidity and

Relative humidity recorded in tree rings temperature signals in trees. Geochim. Cosmochim. Acta 64 (2), 161–167. Epstein S. and Mayeda T. (1953) Variations of 18O content of waters from natural sources. Geochim. Cosmochim. Acta 4, 213–224. Epstein S., Yapp C. J., and Hall J. H. (1976) The determination of the D/H ratio of non-exchangeable hydrogen in cellulose extracted from aquatic and land plants. Earth Planet. Sci. Lett. 30, 241–251. Epstein S., Thompson P., and Yapp C. J. (1977) Oxygen and hydrogen isotopic ratios in plant cellulose. Science 198 (4323), 1209 –1215. Farquhar G. D. and Lloyd J. (1993) Carbon and oxygen isotope effects in the exchange of carbon dioxide between terrestrial plants and the atmosphere. In Stable Isotopes and Plant Carbon Water Relations (eds. R. Ehleringer, A. E. Hall, and G. D. Farquhar), pp. 47–70. Academic Press. Farquhar G. D., Henry B. K., and Styles J. M. (1997) A rapid on-line technique for determination of oxygen isotope composition of nitrogen-containing organic matter and water. Rapid Commun. Mass. Spectr. 11 (14), 1554 –1560. Farquhar G. D., Barbour M. M. and Henry B. K. (1998) Interpretation of oxygen isotope compositions of leaf materials. In Stable Isotopes (ed. H. Griffiths), pp. 27– 62. BIOS Scientific. Farris F. and Strain B. R. (1978) The effects of water-stress on leaf H218O enrichment. Radiat. Environ. Biophys. 15, 167–202. Feng X. and Epstein S. (1994) Climatic implicaitons of an 8000-year hydrogen isotope time series from bristlecone pine trees. Science 265, 1079 –1081. Feng X. and Epstein S. (1995) Climatic temperature records in ␦D data from tree rings. Geochim. Cosmochim. Acta 59 (14), 3029 –3037. Fritts H. C. (1976) Tree Rings and Climate. Academic Press. Gray J. and Song S. J. (1984) Climatic implications of the natural variations of D/H ratios in tree-ring cellulose. Earth Planet. Sci. Lett. 70, 129 –138. Hill S. A., Waterhouse J. S., Field E. M., Switsur V. R., and ap Rees T. (1995) Rapid recycling of triose phosphates in oak stem tissue. Plant Cell Environ. 18, 931–936. Jahren A. H. and Sternberg L. (2003) Humidity estimate for the middle-Eocene Arctic rainforest. Geology 31 (5), 463– 466. Lipp J., Trimborn P., Edwards T., Waisel Y., and Yakir D. (1996) Climatic effects on the ␦18O and ␦13C of cellulose in the desert tree Tamarix jordanis. Geochim. Cosmochim. Acta 60 (17), 3305–3309. Luo Y. H. and Sternberg L. (1992) Hydrogen and oxygen isotope fractionation during heterotrophic cellulose synthesis. J. Exp. Bot. 43, 47–50. Majoube L. (1971) Fractionnement en oxygene 18 et en deuterium entre 1-eau et sa vapeur. J. Chim. Phys. Phys. Chim. Biol. 69, 1423–1436. Merlivat L. (1978) Molecular diffusivities of H216O, HD16O and H218O in gases. J. Chem. Phys. 69, 2864 –2871. Phillips E. L. and Donaldson W. R. (1972) Washington climate for these counties: Clallam, Grays harbor, Jefferson, Pacific and Wakiakum. In Cooperative Extension Service, Washington State University, p. EM-3708. Washington State University.

799

Ramesh R. R., Bhattacharya S. K., and Gopalan K. (1985) Dendroclimatological implications of isotope coherence in trees from Kashmir Valley, India. Nature 317 (802– 804). Roden J. S. and Ehleringer R. (1999) Observations of hydrogen and oxygen isotopes in leaf water confirm that Craig-Cordon model under wide-ranging environmental conditions. Plant Physiol. 120 (1165). Roden J. S., Lin G., and Ehleringer R. (2000) A mechanistic model for interpretation of hydrogen and oxygen isotope ratios in tree-ring cellulose. Geochim. Cosmochim. Acta 64 (1), 21–35. Rozanski K., Araguas-Araguas L., and Gonfiantini R. (1993) Isotopic patterns in modern global precipitation. In Climate Change in Continental Isotopic Records. Geophys. Monogr. 78, pp. 1–36. AGU Geophysical Monograph. Sokal R. R. and Rohlf F. J. (1981) Biometry. Freeman. Sternberg L. (1989) Oxygen and hydrogen isotope ratios in plant cellulose: Mechanisms and applications. In Stable Isotope in Ecological Reseach (eds. P. W. Rundel, R. Ehleringer and K. A. Nagy), pp. 124 –141. Springer-Verlag. Tang K., Feng X., and Ettl J. (2000) The variations in ␦D of tree rings and the implications for climatic reconstruction. Geochim. Cosmochim. Acta 64 (10), 1663–1673. Tang K. and Feng X. (2001) The effect of soil hydrology on the oxygen and hydrogen isotopic compositions of plants’ source water. Earth Planet. Sci. Lett. 185, 355–367. Walker C. D., Leaney F. W., Dighton J. C., and Allison G. B. (1989) The influence of transpiration on the equilibration of leaf water with atmospheric water vapour. Plant Cell Environ. 12, 221–234. Wang X. F. and Yakir D. (1995) Temporal and spatial variation in the oxygen-18 content of leaf water in different plant species. Plant Cell Environ. 18, 1377–1385. Wang Y. J., Cheng H., Edwards R. L., An Z. S., Wu J. Y., Shen C.-C., and Dorale J. A. (2001) A high-resolution absolute-dated late Pleistocene monsoon record from Hulu Cave, China. Science 294 (5550), 2345–2348. White J. W. (1989) Stable hydrogen isotope ratios in plants: A review of current theory and some potential applications. In Applications of Stable Isotopes in Ecological Research (eds. P. W. Rundel, R. Ehleringer and K. A. Nagy), pp. 142–160. Springer-Verlag. Yakir D. and DeNiro M. J. (1990) Oxygen and hydrogen isotope fractionation during cellulose metabolism in Lemna gibbaL. Plant Physiolgy 93, 325–332. Yakir D., DeNiro M. J., and Rundel P. W. (1989) Isotopic inhomogeneity of leaf water: Evidence and implications for the use of isotopic signals transduced by plants. Geochim. Cosmochim. Acta 53, 2769 –2773. Yakir D., Berry J. A., and Giles L. (1994) Isotopic heterogeneity of water in transpiring leaves-identification of the component that controls the delta-O-18 of atmospheric O2 and CO2. Plant Cell Environ. 17 (1), 73– 80. Yapp C. J. and Epstein S. (1982) A reexamination of cellulose carbonbound hydrogen ␦D measurements and some factors affecting plant-water D/H relationships. Geochim. Cosmochim. Acta 46, 955–965.