1 Introduction

The roots of stable-isotope dendrochronology extend back to the middle of the 20th Century during the convergence of the maturing discipline dendrochronology with development of powerful isotope analytical tools. By that time, methods and principles of tree-ring studies (see Chap. 2) had already been developed and refined for about 50 years, largely based on the year-to-year variability in size of rings, but to which an arsenal of other tree-ring measurements has progressively been added over the next 50 years (e.g., wood density, cell size, and dendrochemistry including isotopes). Additionally, measurement of isotope ratios by mass spectrometers became established by the middle of the century, providing a new tool for quantitative research. A large (and rapidly growing) number of stable-isotope dendrochronology investigations have been conducted since the commingling of the two scientific fields, as related to reconstruction and understanding of weather/climate, ecology, and physiology. We can view any stable-isotope series derived from tree rings as a rich and layered story that has been transcribed by physiological, physicochemical and biochemical processes in response to a wide range of influences from climate to atmospheric chemistry and pollution to ecology (Fig. 1.1).

Fig. 1.1
figure 1

Schematic representation of interplay between environmental factors and ecophysiology that produces tree-ring C-H-O isotope composition. Environment can influence the ecophysiological, biochemical, and physicochemical mechanisms responsible for producing the isotope composition and consequently isotope dendrochronology can be used to infer environment and ecological events affecting ecophysiology (e.g., insect outbreaks). Tree rings allow identification of changes through time, but if a network of sites is sampled, the tree-ring isotope results can also reveal changes in space (isoscapes)

The dramatic growth of this field from the 1980s into the 2000s is illustrated in the burgeoning number of papers presented at international tree-ring conferences (Fig. 1.2). Several useful overviews are available in the literature related to tree-ring isotope methodology, theory, and applications, such as McCarroll and Loader (2004), Robertson et al. (2008) and Managave and Ramesh (2012). This chapter describes the historical trajectory of studies and their findings from measurements of stable-isotope composition of tree rings. It is not exhaustive, but fortunately further details and more recent advancements and applications are covered in many of the other chapters in this volume.

Fig. 1.2
figure 2

Growth in numbers of light-stable-isotope tree-ring papers (oral and poster) presented at major international tree-ring conferences. Other major meetings, such as AGU, EGU, INQUA, AAG, and regional tree-ring meetings in Asia, Europe and the Americas often also feature tree-ring isotope papers in various sessions

2 Origins

The development of the isotope-ratio mass spectrometer (IRMS) opened the door to isotope measurements on tree rings (and many other materials). Key to the development of IRMS technology was Alfred O.C. Nier at the University of Minnesota, who in the late 1940s‒early 1950s produced the “double-focusing” mass spectrometer, which used both electrostatic and magnetic focusing of ion beams containing the isotopes of interest, an advance that is the basis of most modern instrumentation (Prohaska 2015). Harold C. Urey at the University of Chicago then added a dual inlet system into the design for alternately admitting aliquots of reference gas and unknown gas into the mass spectrometer (Brand et al. 2015). His newly configured mass spectrometer was applied to measurement of stable isotope composition of various materials, in collaboration with renowned geochemists in the early stages of their careers, such as Samuel Epstein and Harmon Craig. Urey was able to discern the importance of temperature effects on isotope fractionation in carbonate-water equilibria, which established the potential of isotope “thermometers” as a novel tool in paleoclimatology (Epstein et al. 1951; Fairbridge and Gornitz 2009).

Craig’s Ph.D. research sought to characterize natural carbon isotope variability in the carbon system (Craig 1953) through extensive δ13C analyses of hot-spring gases, carbon in sedimentary, igneous and metamorphic rocks, terrestrial and marine organic carbon, and a variety of freshwater and marine carbonate rocks. Among the organic carbon samples, he analyzed 22 “modern” wood samples (AD 1892–1950) from four continents and 16 “fossil” wood samples primarily from N. America (15 of which were radiocarbon-dated by Willard Libby of the University of Chicago at 2300 to >25,000 years old). In the modern wood, no patterns in isotope composition related to species, age, geographical location or elevation were detected. The δ13C of the fossil wood fell within the range of modern wood, –22.5 to –27.3‰. Overall, Craig was cognizant that changes in δ13C of CO2 could influence the isotope composition of wood but thought any such changes were being “randomly masked” by other factors, probably local environment effects. He was also aware of interannual variability in isotopic composition among rings within individual trees.

In his Science paper the following year, Craig (1954) interpreted the δ13C data in Craig (1953) as indicating that δ13C of atmospheric CO2 has not varied by more than 2‰ over the last 25,000 years and may have actually been fairly constant over millions of years. However, the real novelty of the new paper was that it communicated the first-ever δ13C analyses of annual growth rings of a calendar-dated giant sequoia from the Sierra Nevada mountains of California. One or two tree rings in each century were analyzed from the period from 1072 B.C. to A.D. 1649. The δ13C values steadily increased ca. 2‰ in the first 150 years of the record, with various upward and downward deviations thereafter. Coincidentally, in long-lived kauri trees in New Zealand, Jansen (1962) found a similar increase in tree-ring δ13C over the first 200 years of the tree’s 800-year life. The sequoia isotope variability did not seem to be related to precipitation (although the tree may have been growing in a landscape position unlikely to be limited by moisture), and variability of atmospheric δ13C of CO2 was discounted. Presaging advancements to come later in the 20th Century, Craig concluded that effects of environmental conditions (e.g., light, temperature, precipitation) on fractionation during photosynthesis and respiration were the primary contributors to the isotope variation observed.

This pioneering tree-ring isotope work in the 1950s exclusively involved stable-carbon isotopes, presumably because the C isotope composition could be readily measured on CO2 gas after the wood samples were combusted. Analysis of H and O isotopes in tree-ring cellulose would have to wait until the 1970s when pretreatment and off-line vacuum methodologies were developed for analyzing only non-exchangeable H (by nitration of cellulose) and procedures for analyzing O isotopes in organic matter (by mercury chloride or nickel pyrolysis) were applied to tree rings (see Epstein et al. 1977).

3 Advances

3.1 20th Century Spin Up

Over the next several decades of isotope dendrochronology, efforts were largely focused on the use of isotopes in tree rings as paleothermometers, along the line of the great success with isotopic composition of marine forams to understand glacial-interglacial temperature history over the last few million years (e.g., Emiliani 1955; Emiliani and Geiss 1959). These tree-ring isotope efforts began in earnest in the 1970s with a surge of papers attempting to quantify temperature coefficients. For carbon isotopes, the early studies by Libby and Pandolfi (1974), Pearman et al. (1976), Wilson and Grinsted (1977), Tans (1978), Farmer (1979), and Harkness and Miller (1980) found both positive and negative temperature coefficients with tree-ring δ13C. Additionally, the 1970s saw attempts to use tree-ring δ13C as a measure of changes in δ13C of atmospheric CO2 (Freyer and Wiesberg 1973; Stuiver 1978; Freyer 1979a, 1981), in an effort to extend the record of regular direct measurements at Mauna Loa and the South Pole, which had just begun at Mauna Loa in the late 1950s (Keeling et al. 1979). Such records are critical to understanding alteration of the global carbon cycle by anthropogenic inputs to the atmosphere of carbon from fossil fuels and land-use change. The above studies did not analyze a common standard wood component nor did they establish a minimum number of trees to be sampled to ensure the δ13C records were representative, which probably also contributes to some of the variability among published results. Variability of isotopes within and between trees and among species was later addressed by Ramesh et al. (1985), Leavitt and Long (1986), and Leavitt (2010). The impetus for developing tree-ring δ13C records as a proxy for long-term changes in δ13C of atmospheric CO2 faded by the late 1980s as reliable isotopic measurements on atmospheric gas trapped in ice cores became established (e.g., Friedli et al. 1986) and the role of environmental influences complicating interpretation of tree-ring δ13C relative to atmospheric δ13C was more fully appreciated.

The development of a plant carbon-isotope fractionation model in the early 1980s (Vogel 1980; Farquhar 1982; see also Chap. 9) provided both the theoretical background with which to better explore the potential of isotope dendrochronology and new insights into the source of the isotope-environment relationships (Francey and Farquhar 1982). Consequently, the next wave of tree-ring δ13C papers usually refined their analysis and interpretations with respect to this model and the reality that a multitude of environmental parameters could actually influence plant δ13C (i.e., δ13 of atmospheric CO2, temperature, light, moisture, humidity) (e.g., Freyer and Belacy 1983; Peng et al. 1983; Stuiver et al. 1984; Stuiver and Braziunas 1987; Leavitt and Long 1988, 1989). The model provides a basis for understanding differences between δ13C records as well as for selection of tree locations to best capture variability of an environmental parameter of interest.

Meanwhile, H and O isotopes of water in tree rings were also being explored in greater depth in the 1970s. For example, Schiegl (1974) was the first to identify a positive correlation between temperature and δ2H in tree rings (of spruce), which was predicted based on the established positive relationship between temperature and δ2H and δ18O of precipitation (Dansgaard 1964), i.e., source water for photosynthates formed in leaves. The whole-wood analysis of Schiegl (1974) left some room for uncertainty because of the suite of compounds in whole wood as well as the fact that analysis of whole wood would include both non-exchangeable and exchangeable hydrogen, the latter unlikely to represent the original composition when the wood was formed. Epstein and Yapp (1976) and Epstein et al. (1976) circumvented these problems by analyzing the cellulose component of wood and removing the exchangeable H atoms by replacing them with nitrate. Tree-ring δ18O studies in these early decades also found positive correlations of δ18O with temperature, but again no specific wood constituent was being analyzed by all studies (e.g., Gray and Thompson 1976; Libby and Pandolfi 1979; Burk and Stuiver 1981). These various early studies identified climate relationships by either spatial gradient analysis of isotope composition of tree rings and mean climate of different sites, or by isotope analysis of tree-ring series compared to inter-annual climate variation, and the results were not always the same. For both δ2H and δ18O, some studies were also identifying significant correlations with direct water-related parameters such as precipitation and humidity (e.g., Burk and Stuiver 1981; Edwards et al. 1985; Krishnamurthy and Epstein 1985; Ramesh et al. 1986).

With expanded awareness of likely environmental and physiological influences on the isotopic composition of meteoric water and water being fixed in plants, including temperature, evaporation, humidity, and biochemical fractionation by enzymes, increasingly sophisticated models for plant δ2H and δ18O slowly began to emerge in the 1980s (e.g., Burk and Stuiver 1981; Edwards et al. 1985; Edwards and Fritz 1986; Saurer et al. 1997). A critical advance in model refinement was supported by the experimental work of Roden and Ehleringer (1999a, b, 2000) in the 1990s examining the acquisition of leaf and tree-ring isotope values under controlled environmental conditions including source water isotope composition and humidity. This culminated in the advanced mechanistic model for tree-ring δ2H and δ18O developed by Roden et al. (2000) (see also Chaps. 10 and 11). This work exposed the primary contributions of both isotopic composition of source water (taken up by the roots) and evaporation in the leaves (along with the biochemical fractionation) toward influencing the composition of the rings. This supported results of previous empirical δ2H and δ18O tree-ring studies, which depending on dominant influence, could be alternately reflecting (a) source water isotope composition (and thus temperature related to atmospheric condensation processes) or (b) moisture variables such as evaporation, relative humidity, vapor pressure deficit, rainfall. Further refinements have been made to account for the Péclet effect, the consequence of net convective and diffusive water movement in the leaf (Barbour et al. 2004), and to improve understanding of post-photosynthetic processes influencing isotopes (Gessler et al. 2014).

3.2 21st Century Expansion

With these models established for stable C, H and O isotopes in tree rings, climate‒tree-ring isotope studies in the 2000s could now be planned and interpreted based on refined understanding of the environmental controls on fractionation processes. The wave of these studies was fortuitously aided by the timely development of continuous flow-through technology for gases produced by elemental analyzers and streamed with He carrier gas into the isotope-ratio mass spectrometers (Brenna et al. 1997; see also Chaps. 6 and 7), which resulted in faster analysis. The elemental analyzers contributed to this reduction in analysis time by producing the gases needed for analysis in minutes compared to the previous need for lengthy production and purification of the gases on separate (“off-line”) vacuum-line systems. This new instrumentation also allowed analysis on samples of as little as tens of micrograms instead of several milligrams commonly needed with the previous generation of mass spectrometers.

In the last two decades, tree-ring isotopes (particularly of oxygen and carbon) have been used to explore, identify and reconstruct various elements of temporal climate variability. For example, variability and impact of large-scale climate modes, such as ENSO (El Nino-Southern Oscillation) have been studied with δ18O in the Amazon basin, N. America, and Asia, (e.g., Li et al. 2011; Brienen et al. 2012; Sano et al. 2012; Xu et al. 2013a, b; Labotka et al. 2015; Liu et al. 2017). Many of these studies were based on precipitation/evaporation links to δ18O, but others utilized δ18O in tree rings as related to rainfall and drought (e.g., Treydte et al. 2006; Rinne et al. 2013; Young et al. 2015), humidity (e.g., Wright and Leavitt 2006; An et al. 2013; Labuhn et al. 2016), monsoon variability (e.g., Grießinger et al. 2011; Szejner et al. 2016), and even tropical cyclones (e.g., Miller et al. 2006). Tree-ring δ18O itself has likewise been used to better explore ecophysiological aspects of the models (e.g., Gessler et al. 2013).

Stable-carbon isotopes have likewise been an important component of a sweeping range of tree-ring projects. Some of these studies have sought climatological information, inferring temperature and precipitation (e.g., Barber et al. 2004; Liu et al. 2008, 2014) and even sunlight/sunshine (e.g., Ogle et al. 2005; Young et al. 2010; Gagen et al. 2011; Loader et al. 2013; Hafner et al. 2014) because according to the plant carbon isotope fractionation models, the amount of sunlight can influence rate of photosynthesis, which in turn would affect the ratio of intercellular to ambient CO2 (ci/ca) and thus δ13C. Tree-ring carbon isotopes have been used in ecological studies related to destructive insect outbreaks (e.g., Haavik et al. 2008; Hultine et al. 2013), some of which also use δ18O (Kress et al. 2009; Weidner et al. 2010). Many other studies are more ecophysiological in nature, inferring water-use efficiency and stomatal response (e.g., Hietz et al. 2005; Peñuelas et al. 2008; Rowell et al. 2009; Wang et al. 2012; Xu et al. 2013a, b, 2018; Saurer et al. 2014; Frank et al. 2015; van der Sleen et al. 2015) and productivity (Belmecheri et al. 2014). Also, McDowell et al. (2010) identified trees most susceptible to mortality in drought with carbon isotopes that showed their reduced ability to regulate the difference between atmospheric and intercellular leaf CO2 during drought. Tree-ring δ13C has been used to infer past atmospheric CO2 concentration (Zhao et al. 2006), and atmospheric CO2 concentration itself has been considered as a subtle influence in accurately modeling tree-ring δ13C (McCarroll et al. 2009). Wang et al. (2019) found evidence from tree-ring isotopes that rising atmospheric CO2 is improving water-use efficiency and thereby decreasing strength of relationships of ring width with moisture on the Tibetan Plateau.

Some studies are more focused on plant physiology and biochemistry related to assimilation and translocation processes (e.g., Kagawa et al. 2005; Eglin et al. 2008; Eilmann et al. 2010; Bryukhanova et al. 2011; Rinne et al. 2015). Tree-ring δ13C and δ18O have also helped to identify growth rings in tropical trees where they may not be clearly visible (e.g., Evans and Schrag 2004; Anchukaitis et al. 2008; Ohashi et al. 2009), and tree-ring δ13C studies have even reconstructed snow (Liu et al. 2011) and sea level (Yu et al. 2004). Finally, other applications not involving environment or ecophysiology include assessing tree-ring isotopes as a means of crossdating (e.g., Leavitt et al. 1985; Roden 2008) and as an aid in determining provenience of wood (Kagawa and Leavitt 2010).

Tree-ring δ13C has also been used to investigate aspects of plant nutrition (e.g., Bukata and Kyser 2008; Walia et al. 2010; Silva et al. 2015) and particularly pollution (e.g., Battipaglia et al. 2010; Rinne et al. 2010), which has been a fertile and growing application of tree-ring isotopes. Pollutants may impact processes such as stomatal conductance or photosynthesis, and the C-H-O isotopic composition may reflect that alteration, often accompanied by tree-ring width decline (see also Chap. 24). One of the earliest investigations was of SO2 pollution from a coal-fired foundry by Freyer (1979b), who found altered tree-ring δ13C. Reduced ring size and less negative δ13C were also found in tree rings in the vicinity of a SO2-emitting copper smelter in Utah (Martin and Sutherland 1990). Likewise, δ13C in tree rings was elevated up to ca. 100 km downwind from a copper smelter, and the anomaly originated at the time the smelter opened and was presumed to be a consequence of activation of stomatal closure (Savard et al. 2004); tree-ring δ2H also seem to be altered by this pollution (Savard et al. 2005). δ15N in tree rings has been found shifted as a consequence of automobile NOx pollution (Doucet et al. 2012) and δ15N shifts along with δ18O and δ13C in tree rings, which indicate increased iWUE near an oil refinery, are consistent with NOx pollution (Guerrieri et al. 2010). Choi et al. (2005) also found such an increase in iWUE in an area of elevated NOx in S. Korea. Tree-ring isotope evidence for ozone pollution has also been found (Novak et al. 2007), and pollution effects have been identified as reducing climate sensitivity of tree-ring isotopes (e.g., Leonelli et al. 2012; Boettger et al. 2014).

4 Emerging Directions

Several directions in tree-ring isotopes have seen growing interest over the last couple of decades and are promising for future investigations. For example, seasonal variations of isotope composition in tree rings (e.g., Helle and Schleser 2004; Monson et al. 2018; see also Chaps. 7, 14 and 15) can be related to both environmental conditions during the growing season as well as late in the previous season when photosynthates may be stored. The resolution of analysis may vary from 2 to 3 earlywood-latewood subdivisions (e.g., Szejner et al. 2016) to numerous microtomed subdivisions a few tens of microns wide (e.g., Evans and Schrag 2004; Helle and Schleser 2004). Computer-controlled milling (Dodd et al. 2008) and laser dissection have been used to separate small subdivisions in lieu of microtoming (e.g., Schollaen et al. 2014) for isotope analysis, and laser ablation holds hope for more rapid analysis of these seasonal isotope variations (e.g., Vaganov et al. 2009; Loader et al. 2017) with the products of ablation admitted to a combustion interface connected to the mass spectrometer.

Other analytical advances applied to tree rings include compound-specific isotope analysis and simultaneous analysis of C and O isotopes (dual-isotope analysis, see Chap. 7). Additionally, interpreting different aspects of the environment with multiple isotopes has been suggested and applied. For example, Scheidegger et al. (2000) described how under some circumstances δ18O and δ13C from the same tree ring may separately reflect humidity and rates of carbon fixation, respectively, although complete independence is not likely (Roden and Siegwolf 2012; see also Chap. 16). Isotopomers (isotope abundance in different intramolecular positions of the glucose repeat units in cellulose, see also Chap. 7) are also regarded as carrying different environmental signals. Sternberg et al. (2006) concluded that different biosynthetic pathways contribute to heterogeneous distribution of isotope composition, and identified O bonded to the carbon-2 position as carrying the isotopic composition of the original source water. O on other positions then carry cumulative influence of source water composition and humidity, from which source water composition could then be subtracted to attain a more accurate proxy of humidity (Sternberg 2009). Augusti and Schleucher (2007) similarly found H isotope differences depending on position in repeating glucose molecule in cellulose so that physiology and climate could be independently inferred from position-specific isotope analysis. Large, position-specific differences in C-isotope composition were recently identified in cellulose, interpreted as post-photosynthetic shifts in metabolic branching, which may be influenced by environment (Wieloch et al. 2018).

It may be possible to fold the plant isotope fractionation models (Farquhar et al. 1982; Roden et al. 2000) into mechanistic computer models of tree-ring formation, which use physiological processes and environmental conditions during the growing season (e.g., moisture, temperature, and sunlight) to produce photosynthates, activate cambium, and then divide, expand and mature cells in order to replicate observed growth rings (see Chap. 26). Some models operate at the cell level with short time steps and dozens of tunable input parameters, such as the ‘TREERING’ model (Fritts et al. 1999) and the ‘Vaganov-Shashkin’ (VS) model (Vaganov et al. 2011). The model of Hölttä et al. (2010) also operates on very short time steps to simulate whole-tree growth. The simpler ‘VS-Lite’ model (Tolwinski-Ward et al. 2011) has fewer tunable input parameters and operates on monthly climate data to produce wood increments rather than cell growth, and it has gained greater traction for routine applications. An early effort to add isotope algorithms into such models was made by Hemming et al. (2001), who inserted carbon isotope fractionation (of Farquhar et al. 1982) into the TREERING model. Gessler et al. (2014) reviews the extensive catalog of processes (e.g., physicochemical, metabolic, fractionation, transport, storage) within tree tissues, which can contribute to the final carbon and oxygen isotope composition in tree rings and must be considered in interpretation and modeling of tree-ring isotopes. Babst et al. (2018) describe the state of mechanistic tree-ring growth models, including scaling from tree-ring series to whole tree, and potential integration with dynamic global vegetation models (DGVMs) (see also Chap. 26). Because DGVMs are an important tool for improved understanding of current and future global ecology, carbon and water cycles, forest productivity, carbon sequestration, etc., implementation of accurate tree-ring isotope subroutines would provide another layer of control and validation of many aspects of the Earth system.

Finally, spatial mapping of isotope composition in what are known as “isoscapes” is of growing interest as related to fields of ecology, hydrology, climate, geology, forensics, etc. (Bowen 2010). Tree-ring isotopes can play a particularly important role in developing these isoscapes because tree rings can add the temporal component to identify changes in isoscapes through time. A number of tree-ring isoscapes related to climate have been developed in the southwestern USA, Europe, Siberia, and particularly notable in China where a rapid expansion of tree-ring isotope chronologies has been ongoing (Saurer et al. 2002; Leavitt et al. 2010; del Castillo et al. 2013). Gori et al. (2018) have recently explored δ18O and δ2H isoscapes as a tool for provenancing wood cut during logging operations.

5 Conclusions

Isotope dendrochronology has been around for ca. 70 years, but the explosion in the number and diversity of studies over the last 30 years has been breathtaking. This has been driven by advancements in analytical equipment as well as growing recognition of problems for which isotope composition of tree rings may provide resolution, particularly when knowing changes through time is also important. Tree-ring isotopes provide insights into physiological and biochemical processes from leaves to site of wood formation, and changes in isotope values can be used to infer changes in environmental conditions, such as climate parameters.

The history, advances and findings presented here are intended to provide a solid and useful chronology of events, but this summary is not exhaustive. The contents of this book can fill in more of those gaps.