The Stable Hydrogen Isotopic Signature: From Source Water to Tree Rings

Abstract

The hydrogen isotopic signature (δ²H) of water in trees contains information on plant functional responses to climatic changes and on the origin of the water. This is also true for the non-exchangeable hydrogen isotopic signature (δ²H_NE) of plant organic matter, which contains additional physiological and biochemical information that can be dated to specific years if extracted from annual rings of trees. Despite this potential for gaining unique insights from δ²H_NE of tree-ring cellulose (δ²H_TRC), it has not been widely used compared to other isotope signals, likely due to challenging methodological constraints and interpretations of these isotopic signals. In this chapter, we first summarize hydrogen isotope (²H-) fractionation that occurs between source water and tree rings and review methods (e.g. nitration, equilibration, position-specific applications) and calculations to determine δ²H_NE in tree material. Building upon a summary of the current state of knowledge, this chapter also provides an exhaustive synthesis of δ²H_TRC papers, applications, and associated data from approximately 180 sites across the globe (paired with modelled precipitation δ²H values and climate data). The data allow us to investigate the hydrological-climatic effects driving δ²H_TRC pattern on a global scale, the relationship of hydrogen with oxygen isotopes in the same tree-ring material, as well as the influence of physiological-biochemical effects (e.g., species differences, tree growth) that appear to be more important on local or temporal scales  than on a large spatial scales. Thus, when local hydro-climatic influences on source water δ²H can be isolated, δ²H_TRC gives novel insights on tree physiological responses to abiotic and biotic stresses. We conclude that the growing constellation of tree-ring metrics, including advancements in ²H-processing (i.e., equilibration techniques allowing rapid determinations of δ²H_NE) and further refinements to the understanding of post-photosynthetic ²H-fractionations will together provide many new opportunities to understand past climates and ecophysiology by using δ²H in tree rings.

1 General Introduction

Hydrogen isotopes in various biomarkers have been widely used by geochemists and paleo-climatologists to reconstruct past hydrological and climatic changes (Feng and Epstein 1994; Sachse et al. 2012). The underlying assumption of these studies has been that the hydrogen isotope ratio (δ²H) in a compound reflects the δ²H of water used during photosynthesis and during biosynthesis of these compounds in various plant organs. Given that δ²H of precipitation water is well known to be under the control of air humidity during evaporation and temperature during condensation, δ²H of plant compounds have primarily been used to reconstruct those two climatic variables (Libby et al. 1976; Voelker et al. 2014a). The more abundant protium (99.98%) is the lighter of the two stable hydrogen isotopes, consisting of one proton and one electron (mass 1). The less abundant deuterium (0.02%) is the heavier of the two, holding an additional neutron (mass 2). The large difference in mass between the two hydrogen isotopes is physically expressed as strong kinetic and equilibrium isotope fractionation that can cause δ²H variation of up to 550‰ in water and organic (plant) samples (Schmidt et al. 2007). Moreover, there are strong differences in ²H-fractionation among plant metabolic pathways leading to δ²H variation between carbohydrates, lipids, and proteins. In recent decades, measurements of δ²H have been increasingly applied on plant-derived lipids such as n-alkanes, that are highly recalcitrant to environmental degradation over timescales of thousands of years. These compounds can be extracted from soil or sediment cores, or other sources that are often dated using radiocarbon or other dating methods (Sachse et al. 2012). Yet, we are just beginning to understand the physiological and hydrological driving factors causing isotopic variations of plant-derived lipids (Kahmen et al. 2013; Newberry et al. 2015). In contrast, analyses of non-exchangeable δ²H of tree-ring cellulose (δ²H_TRC) is still rarely applied, which is mainly caused by what has appeared to be a low sensitivity to local climatic changes and associated difficulties interpreting δ²H signals (Boettger et al. 2014; Lipp et al. 1991), as well as due to methods that hinder the rate of sample processing (Epstein and Yapp 1976; Schimmelmann 1991). However, further developments during the last two decades led to high-throughput methods for δ²H_NE analyses of plant organic material (including δ²H_TRC) and important new interpretations of metabolic controls over δ²H_NE in plant material (Cormier et al. 2018; Sanchez-Bragado et al. 2019) and tree-ring cellulose (Kimak 2015; Lehmann et al. 2021; Mayr 2003). The coming decade promises more rapid developments in this field that should lead to more studies employing δ²H_TRC to understand past climates and plant ecophysiology.

2 The Hydrogen Isotopic Signature of Water in Trees

The hydrogen isotopic compositions of precipitation and atmospheric water represent the primary determinants of δ²H in terrestrial plant tissues including tree rings. Establishment of the temporal and spatial dynamics of the hydrogen isotopic composition of source water (δ²H_S, but see also Chapter 18), via measurements and modelling, is therefore an important prerequisite for optimal interpretation of δ²H values in tree rings (Allen et al. 2019; Dawson et al. 2002; Saurer et al. 2012). As δ²H values in precipitation and water progressively change due to various factors including temperature, elevation, latitude, distance to coast, amount of a rain event, weather events, and atmospheric circulation pattern, the variability of water isotopes across land surfaces are substantial. Ocean water is used as international reference point for isotope measurements (Vienna Standard Mean Ocean Water, VSMOW), with a δ²H value of 0‰. After evaporation from a water surface or from vegetation, vapor is depleted in ²H (lower δ²H) compared to its origin due to fractionation during phase transition from liquid to gas form; that transition favors the preferential evaporation of lighter isotopologues (e.g., ¹H¹⁶O) over heavier ones (e.g., ²H¹⁸O). In contrast, when water condenses in clouds, precipitation is enriched in ²H (higher δ²H) than the water vapor remaining in the air (Fig. 11.1). Therefore, precipitation will become more depleted in ²H with each rain event as an air mass undergoes successive condensation and precipitation while moving inland or to higher elevations. This process, commonly termed “rainout”, depends strongly on temperature conditions during condensation, partially explaining the strong temperature dependency of precipitation δ²H values (Dansgaard 1964; Libby et al. 1976). Given that the isotopic composition of precipitation is not always available, global maps of average water isotope values have been modelled and can be used to infer δ²H values of precipitation in the absence of measurements (Bowen and Revenaugh 2003). The isotopic signals of precipitation can be modified by evaporation effects prior to when trees take up water in the soil, which has often been characterized by an enrichment in ²H from deep to more shallow soil layers (Allison et al. 1983; Lin and Sternberg 1993). On the other hand, δ²H of soil water that is taken up by trees is an admixture of recent and older precipitation that can depend on the size, depth, and geological properties of water catchments and groundwater aquifers, as well as on water turn-over rates in concert with soil characteristics, vegetation rooting depths and other processes modifying the mixing and uptake of water by tree roots (Brooks et al. 2010; Lin and Sternberg 1993). Overall, the isotopic composition of precipitation and source water taken up by trees can lag precipitation inputs by weeks to years, and can vary markedly among individuals and across regions (Allen et al. 2019; Voelker et al. 2019; Volkmann et al. 2016).

Fig. 11.1

① The hydrogen isotope composition in tree rings is dependent on the origin of the rain (water vapor source). The higher the latitude (latitude effect) and altitude (altitude effect), and the more continental the formation area of the clouds is, the more ²H-depleted the precipitation will become. ② Surface water becomes ²H-enriched due to evaporation of isotopically lighter water. ③ Water infiltrating deeper soil layers mixes with water of previous precipitation events, which is often ²H-depleted compared to surface water as it experiences no evaporative effects. ④ The isotopic composition of groundwater can deviate strongly from rainwater as it may have a different catchment area and a different temporal origin. ⑤ Depending on the root depth and transpiration rates, tree xylem water can show a mixed isotopic composition of soil and groundwater, with generally no further ²H-fractionation during uptake. ⑥ Leaf water is ²H-enriched due to transpiration of isotopically lighter water. This effect can be influenced by changes in stomatal aperture, temperature, and air humidity. ⑦ ²H-enriched leaf water and ²H-depleted NADPH derived from photolysis of H₂O are the main isotopic sources of primary assimilates. ⑧ Unresolved ²H-fractionation processes linked to carbon fixation, starch synthesis, re-mobilization, and downstream transport of sugars lead to a ⑨ ²H-enrichment of tree-ring cellulose compared to leaf assimilates. Blue lines refers to water, with dashed lines indicating processes that show little or no ²H-fractionation. Red lines and axis refers only to ²H-fractionations in organic molecules, including NADPH and carbohydrates (i.e. sugars, starch, and cellulose)

Source water (δ²H_S) becomes enriched in heavy isotopes after reaching leaves due to preferential evaporation loss of lighter isotopologues from the leaf water pool during transpiration. The fractionation from δ²H_S to δ²H of leaf water (δ²H_L) has been closely predicted by the Craig-Gordon model (Dongmann et al. 1974), describing the isotopic enrichment at the evaporative site of a water pool as (Craig and Gordon 1965):

$$\delta^2 {\text{H}}_{\text{L}} = \, \delta^2 {\text{H}}_{\text{S}} + \, \varepsilon_{\text{e}} + \, \varepsilon_{\text{k}} + \, (\delta^2 {\text{H}}_{\text{V}} - \, \varepsilon_{\text{k}} - \, \delta^2 {\text{H}}_{\text{S}} )*{\text{e}}_{\text{a}} /{\text{e}}_{\text{i}}$$

(11.1)

where ε_e is the temperature-dependent equilibrium isotope fractionation between vapor and liquid water (Horita and Wesolowski 1994), ε_k is the kinetic isotope fractionation occurring during the diffusion of water molecules through the stomata cavity and the boundary layer (Merlivat 1978), δ²H_V is the hydrogen isotope ratio of atmospheric water vapor, and e_a/e_i is the ratio between the partial pressures of the atmosphere and intracellular water vapor. As implied by Eq. 11.1, δ²H_L values are mainly driven by the leaf-to-air vapor pressure ratio, similar to relative humidity (i.e., which is often used as an estimate for e_a/e_i) or by the vapor pressure deficit (VPD, difference between saturated and ambient pressure) and have found widespread application for reconstruction of environmental and hydrological conditions from plant organic compounds (Sachse et al. 2012; Voelker et al. 2014a; Zech et al. 2013). Importantly, at high relative humidity (ca. > 60%), the isotopic value of water vapor dominates the isotopic signal of leaf water (Gerlein-Safdi et al. 2018; Lehmann et al. 2018; Roden et al. 2000). The basic leaf water model (Eq. 11.1) tends to overestimate δ²H_L values; corrections include representing leaves as two pools as well as Péclet effects, to account for unenriched water pools (e.g., lignified leaf veins) or progressive leaf water enrichment (e.g., from the bottom to the tip or from the main vein to the margin in leaves) (Roden et al. 2015). Variation in δ²H_L strongly influences δ²H of assimilates (Cormier et al. 2018) and thus δ²H_TRC (Roden et al. 2000). The following section will therefore focus on the manifold isotope fractionations influencing δ²H_TRC chronologies.

3 The Hydrogen Isotopic Signature of Tree-Ring Cellulose

²H-fractionations occurring in trees during the biosynthesis of organic compounds from leaf water and CO₂ to sugars and subsequent molecules result from several biochemical processes (Luo and Sternberg 1992; Schmidt et al. 2003; Sternberg et al. 1984; Yakir 1992). The biosynthetic hydrogen isotope fractionation (ε_bio) between leaf water and organic compounds was (and still is) largely treated as a constant in practice in paleo-biogeochemical studies (Sachse et al. 2012). However, ε_bio may vary among species and compounds, as related to plant enzymatic reactions and the specific biochemical origin of H atoms during their biosynthesis (Cormier et al. 2019, 2018; Estep and Hoering 1981; Ziegler et al. 1976). For biosynthesis of compounds in plants, three origins of H are important in this respect (Fig. 11.1):

• (1) The organic precursor molecules in a biosynthetic pathway; for example, the H atoms of ribulose-1,5-bisphosphate that are transferred to the two triose phosphates synthesized in the Calvin cycle.

• (2) Redox cofactors (e.g. the biological reducing agent NADPH) that provide an important part of the H atoms in the biosynthesis of organic compounds.

• (3) The cellular water.

Yakir and DeNiro (1990) suggested that the overall ²H-fractionation factor occurring between leaf water (δ²H_L) and carbohydrates (δ²H_Carbo) can be divided into photosynthetic (F1) and post-photosynthetic ²H-fractionations (F2), and that the relative fraction (f) of these two processes drives the ε_bio according to the following equations:

$$\delta^2 {\text{H}}_{{\text{Carbo}}} = \, \delta^2 {\text{H}}_{\text{L}} + \, \varepsilon_{{\text{bio}}}$$

(11.2)

$$\varepsilon_{{\text{bio}}} = {\text{ f }}*{\text{ F1 }} + \, \left( {{1 }-{\text{ f}}} \right) \, *{\text{ F2}}$$

(11.3)

The photosynthetic ²H-fractionation (F1) occurs in the chloroplast during the light reaction of photosynthesis where ferredoxin-NADP + reductase produces NADPH with reduced H that is strongly ²H-depleted compared with leaf water (Luo et al. 1991). This ²H-depleted H pool in NADPH is subsequently introduced into organic compounds in the Calvin cycle, causing the first building blocks of sugar pools to be ²H-depleted compared with leaf water. Cormier et al. (2018) suggests that this initial process is independent of the rate of photosynthesis within a species and possibly stable for any given species. In turn, this suggests variation in ε_bio of carbohydrates are driven by post-photosynthetic ²H-fractionations.

Post-photosynthetic ²H-fractionation (F2) processes are assumed to have strong ²H-enrichment effects, commencing in sugars produced by the Calvin cycle and reflecting several overlapping processes and enzymatic reactions. Triose phosphates (TP) synthesis in the Calvin cycle allows (partial) exchange of C-bound H atoms with (²H-enriched) cellular water in CH₂ groups adjacent to CO groups via an enolic structure (Knowles and Albery 1977). Moreover, only one out of four C-bound H atoms in TPs is derived from ²H-depleted NADPH from the light reaction of photosynthesis; alternatively, the others derive from precursor molecules that become ²H-enriched after exchange with cellular water, as described earlier. During the fructose1,6-bisphosphate aldolase reaction, two TPs are used to build one hexose phosphate and one out of four C-bound H atoms is lost to the surrounding water (Schmidt et al. 2015). Phosphoglucose isomerase, used to inter-convert glucose 6-phosphate and fructose 6-phosphate, might further ²H-enrich the sugar pool by allowing partial exchange of specific H atoms with the surrounding cellular water (Schleucher et al. 1999). In non-autotrophic plants and plant tissues, NADPH is likely strongly ²H-enriched as it derives from the oxidative pentose phosphate pathway reactions. Overall, δ²H_NE values of carbohydrates typically do not deviate as strongly from those in leaf water as we would expect from the primary ²H-depletion of the NADPH pool that is generated in the light reaction of photosynthesis because post-photosynthetic ²H-enrichment of the sugar pool provides nearly a net balance of isotopic fractionation events affecting tree rings and other plant tissues.

The ²H-fractionations in sugar pools of trees do not only occur at the leaf level. Roden and Ehleringer (1999) experimentally showed that about 36% of the leaf isotopic signature in a given tree-ring was replaced by the source water isotopic signal during cellulose biosynthesis. From their findings, the authors developed a mechanistic model for δ²H_TRC:

$$\delta^2 {\text{H}}_{{\text{TRC}}} = {\text{ f}}*(\delta^{2} {\text{H}}_{\text{S}} + {\text{F2}}) \, + \, \left( {{1} - {\text{f}}} \right)*(\delta^{2} {\text{H}}_{\text{L}} + {\text{F1}})$$

(11.4)

The estimates of ²H-fractionation factors were derived from various experimental systems (Luo and Sternberg 1992; Yakir and DeNiro 1990), with F1 and F2 as −171‰ and + 158‰, respectively. The Roden et al. (2000) model provided a 1:1 relationship between observed and modeled δ²H_TRC under two different relative humidity conditions. Similarly, the Roden model has been applied to paired tree-ring oxygen and hydrogen isotopes for reconstruction of relative humidity over space and time (Voelker et al. 2014a). Although the Roden model has found practical applications for tree-ring studies it is known to neglect that the compensating effects of F1 and F2 simultaneously determine δ²H_Carbo (Eqs. 11.2 and 11.3) and that δ²H_NE of leaf sugars are lower than δ²H_TRC. Moreover, as discussed toward the end of this chapter, the use of heterotrophic ²H-enriched starch might play an important role for δ²H_TRC values of mature trees (Kimak et al. 2015; Lehmann et al. 2021; Mayr 2003). Thus, further improvements to the Roden et al. (2000) model are needed to more accurately understand drivers of plant tissue δ²H and, consequently, better realize the research potential that δ²H_TRC can provide.

4 Methods and Calculations for δ²H Analysis of Tree Carbohydrates

There are two kinds of hydrogen in carbohydrates or hydrocarbons such as sugars and cellulose. Hydrogen linked to carbon atoms (C-H) is called non-exchangeable hydrogen and cannot be easily exchanged without enzymatic catalysis in biochemical pathways. The δ²H ratio of such carbon-bound non-exchangeable hydrogen (often termed δ²H_NE) in carbohydrates holds meaningful information on hydrological, climatic, or biochemical conditions during their synthesis and is therefore of interest in dendrochronological studies. In contrast, hydrogen that is linked to oxygen atoms in hydroxyl (O–H) groups of carbohydrate molecules is called exchangeable hydrogen. Exchangeable hydrogen is in equilibrium with the hydrogen of the surrounding water such as recent sap water or environmental water and thus does not reflect the hydrological, climatic, or biochemical conditions during cellulose synthesis. Moreover, during extraction of carbohydrates such as sugars or cellulose, the exchangeable hydrogen will be exchanged with extraction water or solutions and therefore its original isotopic composition will be again overwritten. Therefore, it is crucial to exclude the exchangeable hydrogen for an accurate determination of δ²H_NE values in carbohydrates. Several methods have been developed to overcome this problem over the last decades to isolate the δ²H_NE values of sugars and cellulose in plants. In Table 11.1, we summarized the different available δ²H_NE methods and in the following section we explain the main principles.

Table 11.1 Overview of methods to determine δ²H_NE in plant carbohydrates. The given references reflect seminal papers that provide detailed descriptions of the methods and demonstrate some applications. “Hot equilibration” is performed in an oven or in a heated operational unit, while “cold equilibration” is conducted at room temperature. “Online” means that individual samples (of a batch) are equilibrated with water vapor and subsequently analyzed by IRMS. “Offline” means that sample equilibration with water vapor is performed independently of the actual IRMS analysis

4.1 Nitration Methods

Nitration is the most commonly employed method (Table 11.1) for determination of δ²H_NE values in cellulose and particularly helpful to establish δ²H_NE reference material (Boettger et al. 2007). The nitration of cellulose is a reaction where the hydroxyl groups are replaced by nitro groups (NO₂) via an electrophilic substitution. The reaction can be performed using a mixture of HNO₃ (90%) and H₂SO₄ (~98%) with a final concentration of HNO₃ of about 25% in a 150 times weight excess at room temperature for ≥ 12 h. Alternatively, cellulose can also be nitrated with a 40 times weight excess of a solution of 28.7 m% P₂O₅ in HNO₃ (90%) at room temperature (RT) for 4 h (modified after Alexander and Mitchell 1949). For both methods, the mixing of the two components is highly exergonic and the solution needs to be cooled down to 4 °C, e.g., in an ice-water bath. The complete dissolving of P₂O₅ can take several hours or overnight, where a slow temperature increase to RT is uncritical. For achieving highly nitrated cellulose, it is recommended to constantly stir the reaction solution and to keep the water content as low as possible. After the nitration, the product can be washed with distilled water and stabilised with methanol. Incompletely nitrated side products can be removed via the dissolution of the highly nitrated product in acetone and centrifugation. Cellulose nitrate with a nitrogen content of > 12.7% is assumed to reflect a nearly complete nitration, but more than 13.5% N are not achievable (de la Ossa et al. 2011). To complete these steps noted in the methods above and produce cellulose nitrate with an acceptably high nitrogen content, there is probably an upper limit of about 50 samples per week that can be processed by one person, which has limited its application in attempts to understand inter-annual variation in δ²H of tree rings. Note that cellulose nitrate is an explosive material and should be stored in a refrigerator to reduce degradation. In addition, nitration methods for sugars have been developed (Dunbar and Schmidt 1984) and applied in food authenticity studies (Doner et al. 1987). However, sugar nitration should be performed very carefully under constant cooling as temperature above 20 °C can cause an extremely vigorous reaction and explosion. Finally, δ²H analysis of cellulose nitrate with a TC/EA-IRMS system might be best performed in a reactor consisting of a mix of chromium and quartz chips to avoid potential isotope fractionation during pyrolysis of nitrogen-rich compounds (Gehre et al. 2015).

4.2 Equilibration Methods

In contrast to the cellulose nitration method, where the exchangeable H on hydroxyl (OH) groups is replaced by a nitro group (NO₂), the equilibration method is based on a physico-chemical principle of exchanging the OH groups with those of standard water. This process of H isotope exchange happens naturally with ambient water or vapor. In order to guarantee stable and robust δ²H_NE values, which makes up 70% of the hydrogen atoms in a cellulose molecule, it is accepted to follow an equilibration with in-house-standard water of known isotopic composition before sample analysis with a TC/EA-IRMS system. Samples are decomposed by pyrolysis at ca. 1400 °C to hydrogen (H₂) and carbon monoxide (CO) gases, with the former being used to determine δ²H values. Given strong methodological improvements over the last decades, the equilibration method has become more popular for determining δ²H_NE in cellulose.

First, offline methods were used to perform the equilibration (Feng et al. 1993; Grinsted and Wilson 1979; Schimmelmann 1991; Wassenaar and Hobson 2000). They followed different temperature regimes but all of them suffered from disadvantages: (1) slow equilibration process at low temperatures, (2) difficult and time-consuming removal of the excess water through sublimation for the method of Feng et al. (1993), (3) isotope fractionation potential between water and exchangeable hydrogen during the different process steps and (4) a rather large amount of sample (3–12 mg) that is required. Wassenaar and Hobson (2000) described a similar method with static offline equilibration that can be applied for cellulose, as well as for complex organic materials, for example, feathers or keratin. More recent developments in offline methods allow equilibration of a batch of cellulose, starch, or sugar samples in a metal chamber under a continuous flow of vapor with a known isotopic composition at high temperature conditions (ca. 70 to 130 °C; “Hot offline equilibration”, Table 11.1) in an oven and to dry them before TC/EA-IRMS analysis (Sauer et al. 2009; Schuler et al. 2022). This system was further optimized, enabling sample equilibration and drying to occur directly in a heated and helium purged autosampler to prevent contamination with humid laboratory air. The autosampler is placed on an TC/EA-IRMS system, so that multiple samples can be equilibrated with injected water vapor, dried, and directly measured (Wassenaar et al. 2015). Equilibrations are mostly performed at high temperatures to vaporize a water source and to accelerate the equilibration process. However, other studies showed that sufficient equilibration of plant material with a water source can also be achieved at room temperature in an evacuated glass desiccator over a longer time period (“Cold offline equilibration”, Table 11.1), followed by drying of the samples in a vacuum oven over several days (Qi and Coplen 2011; Sanchez-Bragado et al. 2019).

In addition to offline equilibration systems, Filot et al. (2006) developed an online equilibration method. An equilibration chamber temperature controlled at 110 °C with a moving piston selecting between three positions to load (sample derived from an autosampler), equilibrate and inject a sample for subsequent isotope ratio analysis. The equilibration chamber is flushed with helium and in-house-standard water that is immediately vaporized when entering the heated equilibration chamber. An analytical batch consists of 94 measurements (i.e. 55 samples and 39 standard materials), allowing the analysis of about 165 samples per week. Samples and standard materials are equilibrated sequentially within 600 s and then pyrolyzed by a TC/EA-IRMS to determine δ²H values. A GC-column held at 70 °C was used to separate H₂ and CO gases to the degree that sequential isotope ratio measurements are feasible on IRMS, enabling to do triple isotope analysis (Loader et al. 2015). The “hot online equilibration” method (see Table 11.1) can actually be used for the quadruple isotope ratio determination by a corresponding adjustment of the GC-column that allows the separation of N₂ from CO.

4.3 Position-Specific Methods to Determine δ²H_NE in Wood Material

Deuterium in plant carbohydrates can also be position-specifically determined by nuclear magnetic resonance (NMR, Table 11.1) spectroscopy. δ²H_NE methods have been developed for starch and sugars from different herb and tree species (Schleucher et al. 1999; Zhang et al. 2002) and for plant and tree-ring cellulose (Betson et al. 2006). The intra-molecular analysis reveals strong δ²H differences for carbon-bound hydrogen atoms within carbohydrate molecules. Thus for interpretation of δ²H_TRC values the underlying δ²H_NE variations on individual positions should be considered (but see Chapter 7 for more information). Moreover, a novel and simple preparation method was developed to measure the δ²H value of lignin methoxyl groups (δ²H_meth, Table 11.1) by gas chromatograph/pyrolysis-IRMS (Greule et al. 2008). δ²H_meth values of wood samples were strongly correlated with the δ²H_P on a global scale. δ²H_meth might therefore reflect a new climate proxy for temperature changes over time if extracted from annual rings of trees (Gori et al. 2013). However, the biosynthetic isotope fractionation between δ²H_meth and δ²H_P of ca. -217 ‰ can vary among different tree species and therefore care should be taken if used for climate reconstructions (Anhauser et al. 2017). δ²H_meth might be of particular interest for paleo-wood samples, which have often been buried under anoxic conditions and have undergone substantial or nearly complete losses in cellulose due to degradation, but may still contain lignin.

4.4 Calculation of Non-exchangeable Hydrogen Isotopic Composition, International Standards, and Referencing

During our review for this chapter, we found that equations regarding δ²H_NE can vary among studies using equilibration methods. To facilitate comparisons among past and future δ²H_TRC values, we propose the following systematics and syntax. The total δ²H values (δ²H_T), which is the result of an isotopic analysis of equilibrated samples, can be calculated as follows:

$$\delta^{2} {\text{H}}_{\text{T}} = \, f{\text{e}}* \, \delta^{2} {\text{H}}_{\text{E}} + \, ({1} - \, f_{\text{e}} ) \, * \, \delta^{2} {\text{H}}_{{\text{NE}}}$$

(11.5)

where ƒ_e is the fraction of total exchangeable hydrogen in the total H pool and δ²H_E the δ²H value of the exchangeable hydrogen atoms. For a precise estimation of δ²H_NE, a comprehensive evaluation of different potential cases for ƒ_e are needed. Theoretically, 30% of the hydrogen in cellulose should be exchanged, but in most cases less were exchanged, most probably due to difficulties in accessing all exchangeable hydrogen and to inhomogeneities of the cellulose preparation steps. Therefore, the range for ƒ_e of cellulose varies between 0 and 0.3 (Schimmelmann 1991; Wassenaar and Hobson 2000). The higher ƒ_e values of a measurement, the better the equilibration processes and the precision and accuracy of the δ²H_NE analysis:

$$f_{\text{e}} = \, (\delta^{2} {\text{H}}_{\text{T}} - {\text{A}} - \, \delta^{2} {\text{H}}_{\text{T}} - {\text{B}}/ \, (\delta^{2} {\text{H}}_{\text{W}} - {\text{A}} - \, \delta^{2} {\text{H}}_{\text{W}} - {\text{B}})*({1} - \, \varepsilon_{\text{e}} /{ 1}000)$$

(11.6)

where δ²H_T-A and δ²H_T-B are δ²H_T values of cellulose that have been equilibrated with in-house-standard water (A or B) and thus with water vapor of a known isotope ratio (δ²H_W-A, δ²H_W-B), respectively. The isotopic differences between the two in-house-standard waters A and B is mostly several 100‰ and allow for the determination of ƒ_e (approximately the slope of relationship between δ²H_T and δ²H_W of both waters). The ²H-fractionation (ε_e) occurring between δ²H_E and δ²H of water vapor has been found to vary between 60‰ and 100‰ for most plant derived compounds. Most methods set ε_e to ca. 80‰, as it has been found to be an average isotope fractionation factor for most cellulose samples (Schimmelmann 1991; Wassenaar and Hobson 2000). Both, ε_e and ƒ_e are temperature dependent, so temperature needs to be tightly controlled during equilibration and also held constant among batches to provide consistent results.

A mass balance allows the determination of δ²H_NE from the δ²H_TA or δ²H_TB where δ²H_T values are normalized to the international VSMOW scale using an international standard, which has no exchangeable hydrogen (e.g., polyethylene foil, IAEA-CH-7) following Filot et al. (2006):

$$\delta^{2} {\text{H}}_{{\text{NE}}} = \, (\delta^{2} {\text{H}}_{{\text{TA}}} - {1}000*f_{\text{e}} *( \, \varepsilon_{\text{e}} - {1}) \, - \, f_{\text{e}} * \, \varepsilon_{\text{e}} * \, \delta^{2} {\text{H}}_{{\text{WA}}} ) \, / \, ({1 } - \, f_{\text{e}} )$$

(11.7)

δ²H_NE values of equilibrated cellulose samples should be checked against cellulose standards with known δ²H_NE values to verify method accuracy. Cellulose nitrate derived from different wood material or chronologies might be ideal for verification of these methods (Filot et al. 2006; Schuler et al. 2022). Recently published international wood standards (USGS 54–56) might also be used for δ²H calibration (Qi et al. 2016). There is also an international cellulose standard available (IAEA-CH-3), however, there is no agreement yet on its actual δ²H_NE value (Loader et al. 2015; Sauer et al. 2009). Equilibration can also be performed on other organic matter, however, the degree of exchangeable hydrogen and the exchangeability of hydrogen atoms itself can differ among substances.

5 Synthesis of δ²H_TRC Data, Applications, and Interpretations

For this book chapter, we reviewed ca. 50 studies and unpublished records showing information on δ²H_NE of tree-ring cellulose (δ²H_TRC; mainly derived via equilibration or nitration methods) from ca. 180 sites. We approximated the mean δ²H_TRC values for each site chronology and plotted the distribution on a global map (Fig. 11.2). Table S1 holds all information on publications, sample preparation and material, tree species, temporal resolutions, location, and tree-ring isotope data. For each site, we estimated the mean annual temperature (MAT) and mean annual sum of precipitation (MAP) using a gridded data product at 5-arc-minute resolution (Fick and Hijmans 2017). We also modelled long-term mean annual δ²H and δ¹⁸O of precipitation (δ²H_P and δ¹⁸O_P) for each site using grids available on Bowen and Revenaugh (2003). For studies that reported site elevation, δ²H_P values were adjusted by a value calculated from multiplying the isotopic lapse rate, assumed to be −2.24‰ δ²H per 100 m (Poage and Chamberlain 2001) for the difference between site elevation and pixel-mean elevation; for studies that did not report site elevation, the gridded values were used without adjustment.

Fig. 11.2

Global map for δ²H_NE of tree-ring cellulose (δ²H_TRC, VSMOW, markers) and δ²H of precipitation (δ²H_P, VSMOW, shaded) for ca. 140 sites derived from literature data (Table S1). Paleo sites (< 1000 years before 2000 AD) are not shown due to climatic and sampling uncertainites

5.1 Global δ²H_TRC Patterns and Hydro-Climatic Effects

For more than four decades, δ²H_NE of wood or tree-ring cellulose has been recognized to closely correlate with MAT and that δ²H_NE is generally sensitive to continental-scale climatic trends (Epstein and Yapp 1976; Gray and Song 1984; Libby et al. 1976; Schiegl 1974). In retrospect, the temperature dependence of δ²H_TRC at these large scales is partially due to the isotopic composition of meteoric water, which also correlates with temperature variations (Dansgaard 1964; Yapp and Epstein 1982). Here we investigated the global relationship of δ²H_TRC with δ²H_P and MAT across various tree species and study sites (Fig. 11.3). Including δ²H_TRC values from paleo study sites (< 1000 years before 2000 AD) only slightly affected the relationship. However, to be conservative, we excluded paleo studies (ca. 40 sites) from this analysis due to large uncertainties related to time-transgressive estimates of paleoclimate temperatures and associated estimates of δ²H_P.

Fig. 11.3

Relationships between δ²H_NE of tree-ring cellulose (δ²H_TRC, VSMOW) with modelled δ²H of precipitation water (δ²H_P) and mean annual temperature (MAT). Mean values of ca. 180 sites derived from ca. 50 publications and unpublished records (Table S1) are given. Paleo sites (< 1000 years before 2000 AD) were excluded from the regression line due to climatic and sampling uncertainites. Unknown = tree species could not be identified

As expected, δ²H_P is indeed significantly related to δ²H_TRC (r^2 =0.69), demonstrating that spatial variability in δ²H_P of precipitation determines more than two thirds of the spatial variation in tree-ring isotope ratios. Interestingly, the relationship between δ²H_P and δ²H_TRC largely envelopes the 1:1 line, albeit with a slightly shallower slope of 0.9 ± 0.05. The overlap of the 1:1 line is somewhat surprising, given that δ²H_TRC is modified by ²H-enrichment of leaf water and by various ²H-fractionation in carbohydrates in leaf and sink tissues. Roden et al. (2000) showed that it is rather unlikely that the leaf water signal is fully lost during tree-ring cellulose synthesis for both oxygen and hydrogen isotopes. We therefore assume that the sum of the ²H-enrichment during leaf water evaporation and ²H-fractionations of carbohydrates (Eqs. 11.2 and 11.3) cancel each other out, so that the mean δ²H_TRC of a chronology approximately reflects the average δ²H_P of a site. While MAP showed no clear relationship with δ²H_TRC (not shown), we found a significant relationship between MAT and δ²H_TRC (r² = 0.47), with a slope-derived change of 5.2 ± 0.5‰/℃ and an intercept of −21.5 ± 4.0‰. The linear relationship was very similar between MAT and δ²H_P (r² = 0.67, not shown), with a slope-derived change of 5.8 ± 0.3‰/℃ and an intercept of −122.2 ± 3.0‰. Our results agree with 5.5‰/℃ that was observed by Gray and Song (1984), demonstrating that the temperature-dependent equilibrium isotope effect in precipitation water is the dominant factor influencing δ²H_TRC on a global scale (however, not on a local scale as discussed later).

Moreover, we additionally investigated the global δ²H-δ¹⁸O relationship for tree-ring cellulose of about 110 sites (70 modern/40 paleo) (Fig. 11.4). Previously, these two types of data had been used to model leaf water isotope across continental climate gradients (Voelker et al. 2014a). However, to our knowledge, a global δ²H-δ¹⁸O relationship for tree-ring cellulose across diverse tree species has not previously been presented. We found that the global δ²H-δ¹⁸O of tree-ring cellulose is shifted by an average δ¹⁸O offset of + 38.6‰ between precipitation and the signal incorporated in tree-ring cellulose. This offset approximates the difference between measured δ¹⁸O values of xylem water and tree-ring cellulose (Roden and Ehleringer 2000). The changes in δ²H_TRC values are low (Fig. 11.4), which can be explained by the close relationship to δ²H_P (Fig. 11.3). The slope of the δ²H-δ¹⁸O relationship for tree-ring cellulose is 8.1 and thus closely mirrors that of the global meteoric water line (the GMWL is here derived from the modelled precipitation isotope data), reflecting the dominance of isotope fractionations driving spatial variation in the signals of δ²H_P and δ¹⁸O_P taken up by trees. Global δ²H-δ¹⁸O relationships of tree-ring cellulose were not clearly influenced by broadleaf or coniferous tree species (Fig. 11.3 & 11.4) and species-effects might not play a significant role at this scale (but see species effects on the local scale below). Perhaps more importantly, variability of how δ²H vs δ¹⁸O diverge from this global relationship spatiotemporally and among and within species certainly has the potential to yield unique insights on climatic variability, ecohydrology and tree ecophysiology that needs further exploration.

Fig. 11.4

Gobal relationships between δ¹⁸O and δ²H values (both VSMOW) for precipitation (modelled) and tree-ring cellulose. Mean values of ca. 180 sites are given for precipitation and of ca. 110 sites for tree-ring cellulose (Table S1). Paleo sites (< 1000 years before 2000 AD) were exludced from the regression line due to climatic and sampling uncertainites. Isotope data of precipitation follow the typical equation of the global meteoric water line (GMWL). Isotope data of tree-ring cellulose follow the same slope as the GMWL. Unknown = tree species could not be identified

5.2 Paleo-Climatic δ²H_TRC Applications

Due to the strong temperature response, δ²H_TRC analyses were undertaken to provide early paleoclimate or “paleo-thermometer” proxies that could be applied to ancient wood samples (Feng and Epstein 1994; Yapp and Epstein 1977). Other studies on ancient wood material combined δ²H_NE and δ¹⁸O for the reconstruction of relative humidity (Edwards and Fritz 1986; Voelker et al. 2014a) and atmospheric circulations patterns (Feng et al. 2007). Some studies used the δ²H_TRC difference between ancient and modern wood to infer climatic changes or monsoon intensity (Boettger et al. 2003; Feng et al. 1999). Moreover, δ²H_TRC can function as a supplementary information, helping to date tree rings in wood of unknown sources (Becker et al. 1991). This is particularly important for dating of ancient wood samples, which are rarer and thus can be more difficult to cross date using conventional dendrochronological methods. The longest published δ²H_TRC record so far (over 8000 years before 1950, at 50-year resolution) was established by Feng and Epstein (1994) from bristlecone pine trees. However, δ²H_TRC was also analyzed from much older individual wood samples from the Late Glacial/Holocene (14 to 8 Ka) (Boettger et al. 2003; Edwards and Fritz 1986; Epstein 1995; Feng et al. 2007), Late Pleistosence (60 to 30 Ka) (Stojakowits et al. 2020), Early Pliocene (~4.5 Ma) (Csank et al. 2011), and Eocene (~45 Ma) (Jahren and Sternberg 2008). Despite this past work, hydrogen isotope analysis of tree rings is rarely conducted compared to the use of hydrogen isotopes from other paleo isotope archives (e.g., ice or sediment cores). Hence, novel approaches that combine δ²H_TRC with carbon and oxygen isotope analyses and other tree-ring properties at inter-annual resolution may provide promising new multi-proxy tools for understanding past climates and their impact on plant ecophysiology.

5.3 Local δ²H_TRC Pattern and Physio-Biochemical Effects

Despite known correlations between climate variables and δ²H_TRC across large spatial scales, temporal variations in local δ²H_TRC chronologies often do not strongly correlate with local climatic changes or hydrological influences (Boettger et al. 2014; Lipp et al. 1991; Loader et al. 2008; Pendall 2000). Oxygen and hydrogen isotopes in an individual tree should theoretically be derived from the same water sources and thus be correlated (as shown in Fig. 11.4), however, only weak relationships between δ²H and δ¹⁸O chronologies of tree-ring cellulose have been observed on local scales (Fig. 11.5). Early research indicates that climatic information in water-derived hydrogen isotopes is potentially overwritten by physiological factors driving changes in δ²H_TRC, including differences in ²H-fractionation between autotrophic and heterotrophic tissues and metabolic pathways (Luo and Sternberg 1992; Yakir and DeNiro 1990). The following text reviews a growing body of evidence documenting strong physio-biochemical isotope effects on δ²H of plant-derived carbohydrates that are eventually incorporated into δ²H_TRC and may explain the difference between global versus local isotopic patterns.

Fig. 11.5

Local relationships between δ¹⁸O and δ²H chronologies of tree-ring late wood cellulose (TRC, both VSMOW) for three European Quercus robur forest sites (Haupt et al. 2011; Hilasvuori and Berninger 2010; Szczepanek et al. 2006). The chronologies are annually resolved for the period 1901–2003. Relationships are low for all sites (r² < 0.1)

(1) Maturation effect: In Germany, δ²H_TRC of oaks was shown to increase by about 20‰ between approximately age 0 and 150, with little change thereafter (Lipp et al. 1993; Mayr 2003). We term this trend over 100 + years a “maturation effect” because the juvenile stage for many oaks, biologically, usually only lasts up to about 40–50 years and sometimes much less if trees are from sprout rather than seed origin. This increase in δ²H_TRC with age is the opposite of that which would be expected if rooting depths increased with development, leading to them tapping deeper soil water sources with more negative δ²H_S values. Changes in δ²H_TRC values with tree age might be related to an age-related increase in the availability or use of non-structural carbohydrates for growth (i.e. fresh assimilates vs. carbon storage) and/or increased phloem transport distance between canopy leaf area and the stem base (i.e. tree height) where wood samples are typically extracted for isotopic analyses. The aforementioned ideas should be tested with species that differ in growth rate and maturation as only a few δ²H_TRC studies have investigated the maturation effect so far (Arosio et al. 2020).

(2) Intra-annual effects: An early δ²H_TRC study investigated intra-annual variation (Epstein and Yapp 1976) and documented that earlywood was ²H-enriched by 7 to 52‰ compared to latewood within the same tree and year, with some exceptions; ²H-enriched earlywood was confirmed by other studies (Kimak 2015; Mayr 2003). Higher-resolution intra-annual measurements in fossil (Metasequoia sp.) or in modern (Quercus crispula, Pinus radiata) wood demonstrated that cellulose is most ²H-depleted in the center of an annual tree-ring, while the start and end are more ²H-enriched (Jahren and Sternberg 2008; Nabeshima et al. 2018; Wilson and Grinsted 1978). Leaf cellulose of two deciduous trees in Switzerland were also observed to be more ²H-enriched in the early than in the late growing season, while the opposite pattern was observed for an evergreen conifer (Kimak et al. 2015). The drivers of these intra-annual effects are not simply explained by changes in source water or environmental conditions. Winter/spring source water from temperate and boreal zones is ²H-depleted, which is in contrast to higher δ²H_TRC values in wood formed at the growing season, assuming much of the water used by trees during this time are from dormant season precipitation recharge of soil moisture. The VPD-driven leaf water ²H-enrichment would also be lesser during the early growing season than in summer, which is again in contrast to the observed intra-annual δ²H_NE pattern in leaf and tree-ring cellulose. It is likely, the intra-annual effects are caused by a higher relative use of ²H-enriched storage pools for the formation of earlywood and consequently ²H-depleted photosynthetic assimilates for the formation of late wood (Lehmann et al. 2021; Vitali et al. 2022). Cleary, the drivers of intra-annual isotope variation needs further exploration to provide better insights on seasonal changes in plant-water relations and physiology.

(3) Growth effects: Change in biomass productivity, as indicated by various metrics of tree-ring growth, have often been used as primary indicator of plant physiological performance under changing climatic conditions. Here we combined growth and δ²H_TRC data of bur oaks (Quercus macrocarpa Michx.) at three North American sites with a range in environmental conditions (Voelker et al. 2014a, b). Using the Roden model (Eq. 11.4), we calculated the expected δ²H_TRC for at least fifteen years at each of those three sites and then subtracted these modeled values from the observed δ²H_TRC values and plotted these residuals versus growth rates as calculated by Voelker et al. (2014b), which yielded a significant negative relationship (r² = 0.30; Fig. 11.6). For the model, we assumed constant δ²H_S values for each year, as the species is known to be very deeply rooted, and δ²H values of vapor to be in equilibrium with those of source water. If the Roden model would have accounted for all δ²H_TRC variation, then there should be no offset and no trend in data shown. There are two possibilities explaining the trend. The first is that the modelled δ²H_S used in the Roden model were increasingly underestimated across regions with slower growth rates, which is unlikely, because the slower growth rates arise from a complex combination of climatic phenomena that would not have provided such shifts in modeled δ²H_S values. The second explanation for this trend is that low growth rates are inherently associated with greater ²H-fractionation, which correspond to how slow-growing oak trees used a greater proportion of stored carbohydrates that were ²H-enriched. To some extent this would be expected, particularly in oak trees, because their earlywood vessels are largely formed early in the spring (i.e. before leaf expansion is complete) from stored carbohydrates and as oaks grow more slowly, a larger proportion of the tree-ring tissue in any given year is earlywood vessels (Voelker et al. 2012). Figure 11.6 demonstrates that correlations were weak within sites, which would be expected given a lack of year-specific knowledge of water δ²H values to plug into the Roden model. Importantly, the significant relationship across sites provides the first tree-ring based support for the contention of Cormier et al. (2018) that constraints on metabolic rates can increase δ²H in plant compounds. Besides, a recent publication suggest the use of carbon storage under stress conditions by observing a negative relationship between δ²H_TRC values and tree-ring width chronologies on four sites in Europe and Asia, where growth is limited by precipitation or light (Lehmann et al. 2021). However, on a site where temperature was the growth-limiting factor, the relationship between δ²H_TRC and TRW was slightly positive. Thus, further studies are needed to determine the driving factors of ²H-fractionaions related to annual growth (Vitali et al. 2022).

Fig. 11.6

The δ²H differences between observed and modeled tree-ring cellulose (δ²H_TRC, VSMOW) following (Voelker et al. 2014a), plotted against adjusted growth rate index values following (Voelker et al. 2014b). Each data point represents a cross-dated tree-ring year within a site pooled across > 10 trees of bur oak (Quercus macrocarpa). The three sites were sampled across continental gradients in temperature and humidity. The relationships within sites are not significant (P > 0.05), which likely reflects a lack of data for changes in precipitation δ²H across years. The significant relationship across sites supports the contention of Cormier et al. (2018) that constraints on metabolic rates can increase δ²H

(4) Herbaceous tissue effects: Tissue-specific δ²H_NE effects represent important physiological processes, but have rarely been analyzed in trees (e.g. stem vs. root cellulose, Roden and Ehleringer 1999). We thus focused on δ²H_NE reports working with herbaceous plant tissues and assume that this knowledge is helpful to understand tree physiological responses. Sanchez-Bragado et al. (2019) documented the lowest δ²H_NE values in plant assimilates of autotrophic flag leaf tissues of wheat, intermediate δ²H_NE values in semi-autotrophic ears, and highest δ²H_NE values in heterotrophic roots and grain tissues. These results indicate the importance of the photosynthetic reactions causing ²H-depleted NADPH pools in autotrophic tissues. The findings are supported by δ²H_NE responses of leaf cellulose to light intensity and CO₂ concentrations (Cormier et al. 2018), δ²H_NE differences in leaves of different morphology between parasitic plants and their hosts (Cormier et al. 2019), leaf blade and vein tissues (Kimak et al. 2015), and photosynthetic pathways (Sternberg and DeNiro 1983). These findings collectively indicate that autotrophic leaf tissues and their compounds are likely more ²H-depleted compared to all other plant tissues of trees (Ruppenthal et al. 2015). It will remain a future challenge to determine δ²H_NE values of compounds in different tree tissues and to better understand the most relevant processes leading to ²H-fractionations that occur during photosynthesis, transport of freshly assimilated sugars, and during cellulose biosynthesis in woody tissues.

(5) Species effects: We analyze the previously described synthesized dataset for species effects because they are under-described in the literature (but see Ziehmer et al. (2018)). Using sub-fossil wood remains from alpine locations in the European Alps, we show that δ²H_TRC is consistently 40‰ lower in the deciduous Larix decidua than in evergreen Pinus cembra across a 9000-year chronology (Fig. 11.7). Such differences might be partly driven by δ²H_S difference, or, on the other hand, they might indicate divergent metabolic ²H-fractionations associated with leaf formation, translocation of assimilates, and tree-ring formation (Arosio et al. 2020). To investigate this further, we corrected the δ²H_TRC value from each site in the literature data with modelled δ²H values of precipitation (Δ²H_TRC). As global-scale isotope variations in precipitation water are taken into account by this correction (see Fig. 11.3), Δ²H_TRC can be compared between tree species from different sites. However, no clear Δ²H_TRC pattern were observed on a global scale (Fig. 11.8), suggesting that Δ²H_TRC variations must be related to some combination of species/genus and regional climatic differences (i.e. MAT, MAP). We further investigated the Δ²H_TRC values across genus and species, with mean values ranging strongly between −57 and 30‰ and −19 and 34.5‰, respectively (Fig. 11.9). Δ²H_TRC values were significantly different for some genera (P < 0.001) and species (P ≤ 0.02), showing that taxon-specific information is likely imprinted on the ²H-fractionations in mature trees. Interestingly, we found that Δ²H_TRC values at the genus level were co-influenced by MAP (P ≤ 0.02, ANCOVA), indicating that regional climate conditions at a given site, likely a combination of precipitation and strongly co-varying relative humidity, also contributes to variation in ²H-fractionations. The influence of water availability on Δ²H_TRC would make sense, since growth sometimes depends on carbon storage and thus on ²H-enriched heterotrophic starch rather than on ²H-depleted sugars from recent photoassimilates, particularly in extreme dry years or during defoliation events (Vitali et al. 2022). Thus, it is likely that the highly negative and positive Δ²H_TRC variations at the species and genus levels mainly reflect plastic ²H-fractionations related to the carbohydrate metabolism in trees. In this sense, broad surveys of δ²H_NE values of non-structural carbohydrates and cellulose in different tree species might be helpful to disentangle the drivers of ²H-fractionations and potential phylogenetic effects.

Fig. 11.7

δ²H_NE of tree-ring cellulose (δ²H_TRC, VSMOW) for deciduous Larix decidua and evergreen Pinus cembra over the course of 9000 years in 5-year resolution. These records were established on subfossil wood remains based on 201 trees of the Eastern Alpine Conifer Chronology (EACC) established by the Department of Geography of the University of Innsbruck, where these samples has been calendar-dated (Nicolussi et al. 2009)

Fig. 11.8

Global map for δ²H of tree-ring cellulose corrected for δ²H of precipitation (Δ²H_TRC, VSMOW, circles) for ca. 140 sites derived from literature data (Table S1). Paleo sites (< 1000 years before 2000 AD) are not shown due to climatic and sampling uncertainites

Fig. 11.9

Variations in δ²H of tree-ring cellulose corrected for δ²H of precipitation (Δ²H_TRC, VSMOW) for species and genus level across different sites on a global scale. For genus, the number of sites strongly varies (n = 1 to 34, given in parenthesis). For species, a minimum of three sites were considered (n = 3 to 13). The color-coding indicates the average mean annual precipitation (MAP) for each species and genus varying between < 500 mm (red) and > 1500 mm (blue). Paleo studies were excluded (< 1000 years before 2000 AD; Table S1). Abbreviations: Junip. = Juniperus, Pseu./Pseudo. = Pseudotsuga, Querc. = Quercus, Rhizo. = Rhizophora, Popu. = Populus, Seq. = Sequoia, Seqd. = Sequoiadendron, ab. = abies, gl. = glauca, lo. = longaeva, ni. = nigra, st. = strobus, sy. = sylvestris, tr. = tremuloides, me. = menziesii, ma. = macrocarpa, pe. = petraea, ro. = robur

6 Conclusions

Our chapter reveals that we are at the beginning of a new phase of understanding the hydrogen isotopic signals in trees. Overall, deriving robust inter-annual climate signals from δ²H_TRC is currently challenging, but δ²H_TRC may retain important low-frequency climate variation and allow for additional and complementary inferences on tree physiology and carbon metabolism when paired with other isotopic signals and tree-ring growth. To maximize this potential the δ²H_TRC signals need further study and calibration. Toward this end, the foremost need is to investigate the ²H-fractionation processes between water, photosynthetic assimilates, carbon storage, and cellulose at the leaf and stem level. More specifically, studies should be carefully designed to address how this varies with abiotic or biotic stressors, among species and genus, and across diel and seasonal cycles. Second, we need to develop new isotope modeling approaches that can better isolate hydrological ²H-fractionations (leaf water, source water) from the proportionally larger biochemical ²H-fractionations. Third, the new knowledge on ²H-fractionation processes need to be applied to δ²H_TRC chronologies to assess the potential for obtaining improved multi-proxy climate reconstructions or improved understanding of changes in ecophysiology in response to climate variability or other environmental influences. Great progress has been made in understanding and calibrating the δ²H_TRC signal. Continuing advancements in ²H-processing that allow for rapid determinations of δ²H_NE, a growing constellation of ancillary tree-ring metrics, and advances in knowledge on post-photosynthetic ²H-fractionations together foretell that applications of δ²H_TRC will be increasingly used in dendroclimatology and forest ecophysiology.