1 Introduction

Stable carbon isotope ratios (δ13C) in tree ring—either determined compound-specific in cellulose or in whole wood consisting of various chemical compounds—are widely assessed to obtain retrospective information on tree gas exchange physiology and the related environmental drivers (e.g., Saurer et al. 1997; Schleser et al. 1999; McCarroll and Loader 2004; Marshall and Monserud 2006; Rinne et al. 2010). The tree-ring archive allows on the one hand accurate dating and on the other hand integrates δ13C of the canopy, so that biophysical processes occurring at the leaf scale can be transferred to the whole tree and further to the ecosystem scale. The original signal imprinted in the primary photoassimilates is depending on the δ13C of CO2 and the photosynthetic carbon isotope discrimination. The latter is in a first approximation proportional to the ratio of leaf internal and ambient CO2 concentration ci/ca or—to be more precise—of chloroplastic CO2 concentration (cp) and ca (Farquhar et al. 1982). Thus the δ13C of organic matter can be used to characterise environmental effects and their influences on diffusional versus biochemical controls over photosynthesis. The details on photosynthetic carbon isotope discrimination are given in Chap. 9.

While the understanding of stable isotope fractionation during photosynthesis and the environmental factors affecting it is well developed (Cernusak et al. 2013), there is a larger gap of knowledge about the processes leading to alteration of δ13C in downstream metabolic processes in the leaves and in heterotrophic tissues as well as during transport. Such processes lead to the generally observed pattern that non-photosynthetic tissues are enriched in 13C compared to leaves in C3 plants (Badeck et al. 2005; Cernusak et al. 2009). In this chapter we will thus mainly focus on the path of carbon and its carbon isotopic composition from leaf photosynthate production to wood formation, as summarised in Fig. 13.1. We will discuss here the different post-assimilation processes that might be able to alter δ13C and thus cause a (partial) decoupling between the original leaf isotopic signals and the archived signals in tree rings and aim to provide an update of a recent review also tackling this topic (Gessler et al. 2014). As almost all trees are C3 plants (but see Pearcy and Troughton 1975; Lüttge 2006) we only refer to processes in C3 plants. Tree-ring carbon isotopes are determined on a broad range of temporal scales from very fine scale interannual variations up to millennial time scales (Barbour and Song 2014). Knowledge of the potential alteration of the isotope signal on its way to the tree ring at various time scales allows for a better estimation of the uncertainties when reconstructing the coupling between tree processes and climatic drivers.

Fig. 13.1
figure 1

source of carbon used for wood formation could show seasonal and inter-annual variations. In particular, remobilization of stored carbon is crucial for leaf development and early stem growth in deciduous trees but could also play a key role during stress episodes. See text for further details

Conceptual scheme summarizing the main processes in the way from primary assimilates (triose-P) to the wood that may cause fractionation at time scales relevant to tree-ring archives. The first fractionation processes occur already within the leaf, where various isotopic effects during metabolic pathways and phloem loading cause lipids and lignin to be depleted, and sugars and cellulose to be enriched in 13C. For some species, a basipetal enrichment in phloem sap from twigs to basal stem has been described, potentially associated with the continuous unloading and reloading of sugars during phloem transport. This sugar would be partially exposed to metabolic reactions (e.g. production of lignin) in stem tissues, causing an enrichment of the fraction that is reloaded in the phloem. Wood formation could also be affected by fractionations occurring during stem metabolism. Bark photosynthesis, respiratory fractionation, or PEPC CO2 refixation processes are known to produce fractionations, but their direct effect on tree rings is not clear. Finally, the

2 Post-Carboxylation Fractionation in the Leaves

Metabolically downstream from the CO2 fixation by RubisCO, there is one important reaction imposing post-carboxylation fractionation on the newly-fixed assimilates, which involves the aldolase reaction within the Calvin cycle. Due to its location in the metabolic pathway between triose phosphate and fructose 1,6-bisphosphate, aldolase imposes a diurnally distinct isotope signal on the assimilates available in the leaf cytoplasm. During the light, starch that is derived from fructose-6-phosphate via fructose 1,6-bisphosphate is produced and accumulated in the chloroplasts of leaves. On the other hand, sucrose is produced in the cytoplasm and this synthesis is based on the triose phosphates exported from the chloroplast. The triose-phosphates are 13C depleted, which is a direct consequence of transitory starch synthesis, which favours 13C during production of fructose 1,6-bisphosphate by aldolase, leading to a relative 13C enrichment of starch (Rossmann et al. 1991; Gleixner and Schmidt 1997). During the night, the sucrose synthesized in the cytoplasm of leaf mesophyll cells is originating from the 13C-enriched starch, which is degraded via maltose (Weise et al. 2004). As a consequence of this fractionation step, leaf exported sucrose during the day has been shown to be relatively 13C-depleted while night sucrose was enriched in 13C (Gessler et al. 2008). Observations of diel variation in δ13C of leaf sugars exported into the phloem in different tree species (Pinus sylvestris; Eucalyptus delegatensis) showed day-night differences of up to 1.7‰ (Brandes et al. 2006; Gessler et al. 2007; Kodama et al. 2008). However, there are also indications from other trees species that such diel variations in leaf-exported sugars might not always occur (Fagus sylvatica, Pseudotsuga menziesii) (Bögelein et al. 2019). Moreover, only recently, Lehmann et al. (2019) showed with starch deficient mutants that post-carboxylation carbon isotope fractionation was low in three herbaceous species. Thus, there is still some uncertainty regarding which species and under which environmental conditions the aldolase related isotope effects and thus diel variations in δ13C of leaf sugars are expressed. This might also depend on species specific differences in day vs. night phloem loading even though there is some evidence that sugar transport is not changing during the diel cycle (Peuke et al. 2001).

If, however, occurring, diel variations in δ13C of leaf exported and phloem transported sugars may affect the carbon isotope composition of cellulose in tree rings (Tcherkez et al. 2007), as cellulose is mainly constructed from this carbon source. Cell expansion in plants is known to depend on turgor. Steppe et al. (2015) assumed that all stem growth processes, including cell expansion and deposition of cell wall, occurs during the night, when the tree’s water status and thus cell turgor is most favourable. This is in line with molecular studies that show highest night-time gene expression for enzymes of the lignin (Rogers et al. 2005) and cellulose (Harmer et al. 2000; Solomon et al. 2010) metabolism. If we now assume that mainly sucrose produced from starch during the night is used for tree ring production, δ13C of this tissue might be more than 2‰ enriched compared to the situation where 50% night and 50% day sucrose would be used (Tcherkez et al. 2007). Since phloem transport velocities vary between ca. 0.1 and 1 m h−1 (Jensen et al. 2012) for different species, with conifers at the lower and broadleaf trees at the higher end, not only species but also tree height might influence the proportion of starch-derived (13C enriched) assimilates that are used for wood production at a particular position at the tree trunk. There is, however, indication that even in trees where diel variations in δ13C of leaf-exported sugars occur, the amplitude of variation gets strongly dampened during transport in basipetal direction in the trunk so that aldolase fractionation in the leaves does not likely affect tree ring δ13C. The mechanisms leading to the loss of diel oscillation are discussed below.

Besides temporal variation in δ13C of assimilates, also a relative 13C enrichment of sugars loaded into the phloem - compared to primary assimilates or bulk leaf material - is observed. A process, leading to an apparent enrichment of the leaf sugar fraction that can be exported to the phloem compared to bulk leaf material is related to the different δ13C among distinct chemical compounds (for a comprehensive list of enzyme reactions that are involved in producing compound-specific isotope differences see Hobbie and Werner 2004). Primary assimilates are used in the leaves to produce 13C-depleted compounds, such as lignin and lipids. Consequently, the unreacted sugars that can be loaded into the phloem are assumed to be isotopically heavier than the primary assimilates, due to mass balance reasons (see Fig. 13.1). As woody plants have especially high lipid and lignin contents in the leaves, the carbohydrates that can be loaded to and subsequently allocated in the phloem are more strongly 13C-enriched as compared to non-woody plants (Hobbie and Werner 2004). This assumption of an enrichment of unreacted sugars compared to primary assimilates is in agreement with measurement of Brandes et al. (2006), who showed that this offset amounts to >1.5‰ in Scots pine.

3 Changes in δ13C Related to Phloem Loading and Transport

There are two main alterations of the δ13C directly related to phloem transport: one (a) that acts on the temporal (i.e. diel) scale and one (b) on the spatial scale along the transport pathway. Both changes might, however, be due to one common mechanism.

The dampening of the short-term diel variations (a) in δ13C of sugars along the tree axis in basipetal direction is assumed to be a result of the mixing of different sugar pools with different age and metabolic history during transport between the leaf and the trunk cambium (Brandes et al. 2006; Kodama et al. 2008). Such mixing of pools lead to a reduction of the diel amplitude: while the amplitude amounted to 2.5‰ for sugars transported in the twig phloem, it was reduced to <0.5‰ at the trunk base in Eucalyptus delegatensis, where normally tree rings are sampled (Gessler et al. 2007). Thus, this dampening prevents any fractionation related to leaf starch synthesis to be imprinted in tree ring material. Gessler et al. (2014) proposed that the mixing of different sugar pools during basipetal transport might be explained by intrinsic properties of the dynamic Münch mass flow model (see review by Van Bel 2003). The model proposes the sieve tubes to act as a leaky pipe during transport: a proportion of the sucrose from sieve tubes is always released to the surrounding parenchyma, and to compensate for the loss sugars from the surrounding tissues are reloaded back to the sieve tubes (Minchin and Thorpe 1987; see Fig. 13.1). The loss and retrieval of sugars may allow for continuous mixing of different sugar pools along the transport pathway towards the trunk base, and as a consequence dampen the diel variations in δ13C. Further research is, however, needed to clarify if this mixing effect is universal and occurs in all tree species and under all environmental conditions.

This process might, however, also explain (b) the often observed 13C enrichment of phloem sugars from the twig to the trunk base phloem (Brandes et al. 2006). In a comparable way as for the primary assimilates in the leaves, the sugars leaking out of the phloem will be used as substrates for different metabolic conversions (see Fig. 13.1). Assuming that the produced non-exported compounds (i.e. lignin) are 13C-depleted compared to the original sugar substrates originating from the phloem, the unreacted sugars, reloaded into the phloem, will be 13C-enriched. The continuous unloading, metabolic conversion and reloading of unreacted sugars would consequently lead to an enrichment of phloem transported sucrose in basipetal direction. Such an enrichment is not always observed, sometimes occurs only between branch and trunk phloem but not along the trunk (Bögelein et al. 2019), or is detected only during particular periods of the growing season (Gessler et al. 2004). To understand such differences, a complete isotopic mass balance taking into account the δ13C phloem sugars as well as the δ13C of non-exported compounds over the growing season would be extremely useful.

4 The Hidden Stem Metabolism: Bark Photosynthesis, Stem Respiration, and the Role of Carbon Re-fixation

Tree ring wood or cellulose has been shown to differ in the carbon isotopic composition from the phloem sugars, which are thought to be the source for cellulose formation. In general, an enrichment between 1 and 2‰ has been observed (Gessler et al. 2009a; Wei et al. 2014). Thus, wood formation can be also affected by different fractionations associated to stem metabolism (see Fig. 13.1). Firstly, bark photosynthesis and re-fixation of stem-respired CO2 have a significant effect on total tree carbon balance (Pfanz et al. 2002). In angiosperm trees, it has been estimated that 30–90% of the CO2 respired in the stem could be re-fixed (Pfanz et al. 2002; Hilman et al. 2018), and ca. 25% of refixation has been also reported in the upper stem of Scots pine (Tarvainen et al. 2018). Bark photosynthesis is widely found in young twigs and branches, but in the main stem it seems to be restricted to species with thin (<8 mm) outer bark (Rosell et al. 2015; Tarvainen et al. 2018). During bark photosynthesis, the main source of CO2 is stem respiration, which is 13C-depleted as compared to ambient air, and photosynthesis further discriminates against 13C (Cernusak et al. 2001). CO2 refixation may also occur in non-photosynthetic bark, mainly driven by phosphoenol pyruvate carboxylase (PEPC), thus resulting in 13C-enriched products (Cernusak et al. 2009), although this could be compensated by the use of a 13C-depleted substrate (stem-respired CO2). An additional source of uncertainty is the role of xylem sap as a CO2 carrier, potentially allowing the refixation of root-respired CO2 in the trunk (Bloemen et al. 2013). Nevertheless, given the tight association found between bark photosynthesis and both phloem load and xylem refilling (De Baerdemaeker et al. 2017; Konrad et al. 2018; Liu et al. 2019), it is likely that bark photosynthates are mainly used to force changes in osmotic pressure in the xylem and phloem sap, rather than to build new tissues. Therefore, although significant for the whole-plant carbon balance and potentially also for branch growth (Cernusak and Hutley 2011), so far there is no clear evidence that these processes contribute to wood formation in the main stem.‬‬‬‬‬‬‬‬‬‬

Another metabolic process that may affect the isotope signal in the wood is respiratory isotope fractionation (Ghashghaie et al. 2003). Pyruvate dehydrogenase (PDH) and key enzymes in the Krebs cycle may cause kinetic isotopic effects, but they cannot explain alone the observed respiratory fractionation (Werner et al. 2011). PDH and the Krebs cycle also provide substrates for the shikimate pathway, and thus may affect the isotopic signature of (depleted) lignin precursors (Hobbie and Werner 2004). Depending on the demand of different products, PEPC re-fixation may contribute with malate to the Krebs cycle, in order to restore temporal metabolic imbalances, thus resulting in 13C-enriched respired CO2 (Cernusak et al. 2009; Gessler et al. 2009b). These metabolic changes appear as the best explanation for observed diel cycles in the relative enrichment of stem-respired CO2, which becomes negatively correlated with respiration rates (Kodama et al. 2008). Although a direct connection with wood δ13C is not clear, respiratory fractionation may have an indirect effect on the remaining substrates for wood formation (Eglin et al. 2010; Rinne et al. 2015; Vincent-Barbaroux et al. 2019).

Stem metabolism may also play a role in the seasonal pattern of δ13C. Only recently, Budzinski et al. (2016) reported large seasonal variation in expressed transcripts, proteins and metabolites in the bark of Eucalyptus grandis. For example, they found that during the wet summer the expression of cytosolic 1–6-Fructose bisphosphate aldolase and PEPC increased, together with the accumulation of organic acids (malate, fumarate). This suggests a larger relative contribution of PEPC-derived malate to the Krebs cycle, which contributes to the aforementioned 13C enrichment in respired CO2 (Cernusak et al. 2009; Gessler et al. 2009b). Conversely, during the dry winter, both, sugars (sucrose, fructose, glucose) and secondary metabolites (shikimate, taxifolin) tended to accumulate in the bark. Hence, this is likely a period when isotopic signatures of lignin precursors are formed, and these precursors subsequently accumulate at the onset of cambium initiation (Förster et al. 2000). Evidence so far indicates that the environmental signal stored in lignin is comparable to that in cellulose, but in agreement with these metabolic changes, lignin δ13C seems to be more sensitive to early-season temperature (Loader et al. 2003; Ferrio and Voltas 2005; Gori et al. 2013).

5 Imprint of Storage and Remobilization on the Intra-annual Variation in Tree Rings

As pointed above, starch-derived carbon is 13C-enriched, as compared to primary assimilates (Rossmann et al. 1991; Gleixner and Schmidt 1997). Therefore, seasonal variations in the contribution of stored carbon to growth may lead to intra-annual δ13C patterns that are not linked to leaf-level processes (Helle and Schleser 2004; Eglin et al. 2010; Offermann et al. 2011). Helle and Schleser (2004) first proposed the existence of three phases in the seasonal evolution of tree-ring δ13C. Phase 1 shows the highest seasonal values for δ13C, indicative of the incorporation of starch-derived carbon into the growing tissue (both leaves and wood). This has been consistently observed in a range of deciduous broadleaves (Fig. 13.2b–d; see also Helle and Schleser 2004; Eglin et al. 2010; Schollaen et al. 2014). During this phase, Eglin et al. (2010) further interpreted the initial increase in δ13C as a transition from stored soluble sugars to stored starch. After this peak, Phase 2 shows a decline in wood δ13C, in many cases opposing the expected positive effect on δ13C of increasing water deficit from spring to summer. This has been interpreted as a transition from (enriched) storage to (more depleted) current assimilates as the main source of carbon in the wood (Helle and Schleser 2004, Eglin et al. 2010), and is supported by an increase in phloem sugar concentration (Offermann et al. 2011). According to Helle and Schleser (2004), during Phase 3 the source strength of leaves is reduced due to leaf senescence, and wood formation again relies more on stored carbon, thus becoming enriched. However, this late-season increase is less consistent across studies, and most likely depends on the interaction between wood and leaf phenology. In some cases, leaf senescence begins well after the end of wood growth, and thus cannot be reflected in the tree rings (e.g. Aguilera et al. 2010; Eglin et al. 2010; Offermann et al. 2011). Under such conditions, Phase 3 would be a period of predominant use of current assimilates, during which wood variations would respond mainly to environmental conditions (Eglin et al. 2010; Offermann et al. 2011). Still seasonal variation in photosynthetic discrimination might additionally influence such intra-annual patterns in tree rings.

Fig. 13.2
figure 2

Seasonal trends and offsets in carbon isotope composition (δ13C) form leaves (triangles, dotted lines) to stem phloem (circles, dashed lines) and whole wood (solid lines) in different tree species and locations. Note that the scaling of X axis in panel B is different. a Scots pine (Pinus sylvestris L.) in Hartheim, SW Germany (HA) and Fontainebleu, N France (FO). b holm oak (Quercus ilex L.. Subsp. ballota and the portuguese oak (Quercus faginea Lam.) in Oliola, SE Spain (OL). c Beech (Fagus sylvatica L.) in Freiburg-Zähringen, SW Germany (FR), Tuttlingen, SW Germany (TU) and FO. d Sessile oak (Quercus petraea (Matt.) Liebl.) in FR and FO. SOM, water soluble organic matter. Sampling years and data source for each site: HA, year 2004, (Brandes et al. 2006; Gessler et al. 2009a); TU, year 2007, (Offermann et al. 2011); FR, year 2007, Ferrio, Offermann, Gessler, unpublished; OL, years 2005–2008 (leaf data: Aguilera et al. 2010; whole wood data: Aguilera, Ferrio, Voltas, unpublished); FO, year 2009, (Michelot 2011)

In evergreens, seasonal shifts between storage and current-assimilates are not so evident, neither in conifers (Fig. 13.2a; see also Barbour et al. 2002; Klein et al. 2005; Sarris et al. 2013; Schollaen et al. 2014), nor in the few evergreen angiosperms studied so far (Fig. 13.2b; see also Battipaglia et al. 2010; Schubert and Timmermann 2015; Vincent-Barbaroux et al. 2019). This further supports the interpretation of the early-season δ13C peak in deciduous trees as a starch signature. Whereas deciduous species depend on stored carbon during the initial stages of wood formation (Kagawa et al. 2006; Klein et al. 2016; Andrianantenaina et al. 2019), evergreens would be able to rely more on current assimilates (Ogée et al. 2009; Klein and Hoch 2015). Regarding deciduous conifers (e.g. larch), the few studies performed so far show contradictory results. Kagawa et al. (2006), following an isotope-labelling approach, estimated that about 40% of the carbon used in wood formation was derived from previous-year photoassimilates. Conversely, Rinne et al. (2015) found a good agreement between exported sucrose and tree-rings, concluding that, similar to evergreen conifers, storage dynamics had little effect on the intra-annual patterns.

In summary, intra-annual trends in deciduous broadleaf trees are largely dominated by storage dynamics, whereas evergreen species, and particularly conifers, seem to track the seasonal response of current assimilates. Besides seasonal cycles linked to leaf phenology, the existence of periods of stress (e.g. summer drought) during the growing season might also force a major reliance on stored carbon (Sarris et al. 2013; Klein and Hoch 2015; Hentschel et al. 2016). However, wood growth is also restricted under stress, and therefore severe stress episodes often result in temporal gaps in the wood archive (see e.g. Sarris et al. 2013; Forner et al. 2014). Nevertheless, and regardless of leaf habit, there is increasing evidence of the existence of soluble carbon pools of mixed age within the stem, which could eventually contribute to wood formation (Muhr et al. 2015; Trumbore et al. 2015; Klein et al. 2016).

In the near future, recent technical advances like laser ablation and microdissection are likely to facilitate the construction of high-resolution intra-annual records in tree-rings (Schollaen et al. 2014; Rinne et al. 2015; see also Chap. 7). However, one key limitation in our understanding of the link between leaf and wood seasonal trends remains unsolved, and that is the need of an accurate dating of wood and carbon deposition in the tree ring. Although stem increment can be estimated from dendrometer records, carbon deposition may lag over one month girth growth (Cuny et al. 2015; Andrianantenaina et al. 2019; for further discussion see Chap. 15). So far, repeated sampling of microcores appears as the most reliable way to characterize xylogenesis, but is time-consuming and destructive (Cuny et al. 2015; De Micco et al. 2019). Further research should focus on the characterization of the link between xylogenesis, carbon deposition and stem growth, in order to develop alternative approaches for intra-annual tree dating.

6 Intra-molecular Isotope Distribution in Wood Tissues

Most of the organic carbon in plants is due to the CO2 fixation by RubisCo in the leaves (but see PEPC carbon fixation above). The RubisCo mediated reaction adds one single C-atom from CO2 to the acceptor molecule ribulose-1,5-bisphosphate and due to the cyclic nature of the Calvin cycle the acceptor is regenerated. Since (almost) all organic carbon originates from this process the variations in the fractionation related to this step (i.e. photosynthetic carbon discrimination; see Chap. 9) explain variations in the δ13C of the whole molecules synthesized (i.e., triose-phosphates, sugars) but cannot cause intra-molecular differences in δ13C. Rossmann et al. (1991) showed for the first time evidence of such non-statistical intramolecular 13C isotope distribution in starch and sugars and Gilbert et al. (2012) provided an extensive review on the potential underlying mechanisms. Only recently, intramolecular patterns were also observed in the glucose units of tree-ring cellulose (Wieloch et al. 2018).

Post-photosynthetic carbon isotope fractionation related to enzyme reactions that occur at individual C positions within metabolites are able to imprint position-specific carbon isotope differences. Rossmann et al. (1991) as well as Hobbie and Werner (2004) assumed this intramolecular 13C distribution, with a strong enrichment in the C-4 position of glucose, to be due to the aldolase and transketolase reactions in the Calvin cycle. Tcherkez et al. (2004) developed a model that takes into account the isotope effects of C–C bond-breaking reactions (including aldolase and transketolase) of the Calvin cycle, which consequently lead to a mathematical expression for the isotope ratios in hexoses in the steady state. The assumptions made in this model were verified by assessing day-night differences in leaf and phloem sugars (Gessler et al. 2008). Theoretically, changes in metabolite allocation to a metabolic pathway that includes branching point(s) can change the intramolecular 13C pattern. At such isotope-sensitive metabolic branching point, enzyme-related isotope effects might be more or less expressed depending how much substrate is allocated to the branches of the pathways. Thus, environmental factors that cause shifts in metabolic pathway commitment might affect intramolecular 13C distributions. Thus, any shift in intramolecular 13C distributions in sugar moieties laid down in the tree archives is assumed to reflect such shifts in metabolic branching. In fact, Wieloch et al. (2018) found temporal variability in the intramolecular 13C pattern of glucose units of cellulose across a 34 year-long tree ring series. They could also show that intramolecular 13C abundances can give a purer signal of photosynthetic isotope discrimination and a better estimate of VPD than the whole molecule. This was mainly attributed to the fact that the molecule averaged signal is strongly influenced by both, variation in photosynthetic and post-photosynthetic fractionation, while mainly the C-1 to C-3 position of glucose captures the photosynthetic signal. Moreover, position-specific 13C analysis in tree rings might allow for distinguishing environmental effects on photosynthetic gas exchange (via photosynthetic fractionation) from effects on downstream metabolism (via postphotosynthetic fractionation). At the downside, measurements of intramolecular 13C abundances are time consuming and limited to small sample sets but the analytical advancements in future might make this approach feasible for higher throughput analysis (see Wieloch et al. 2018).

7 Can We Actually Assess Water Use Efficiency from Tree-Ring δ13C?

The link between leaf-level processes and wood δ13C is not straightforward, but with some caution we still can retrieve a relevant environmental/physiological signal. In this regard, δ13C in tree rings has been widely used as a proxy for variations in intrinsic water use efficiency (iWUE) across time, space, or genetic populations (e.g. Saurer et al. 2004; Del Castillo et al. 2013; Fardusi et al. 2016) (for further insight into iWUE see Chap. 17). However, the fractionation processes discussed in this chapter may hinder the physiological interpretation of wood or cellulose δ13C, particularly those linked to carbon partition and storage/remobilization patterns. This is particularly relevant for the interpretation of intra-annual variation, but may also have an imprint on a multi-year scale, partly explaining the drought-legacy often reported after extreme dry events (Sarris et al. 2013; Del Castillo et al. 2015; Hentschel et al. 2016). Evidence so far suggests that deciduous species are more affected by storage-remobilization processes than evergreens, and angiosperms more than gymnosperms (Ogée et al. 2009; Eglin et al. 2010; Rinne et al. 2015; Vincent-Barbaroux et al. 2019). This fits well with our knowledge on their use of carbon resources (Hoch et al. 2003) and early empirical observations of a stronger current-year signal in latewood than in earlywood for deciduous trees (Loader et al. 1995). As a consequence, a single-substrate mechanistic model may be enough to explain seasonal and inter-annual trends in evergreen conifers (Ogée et al. 2009), but more complex models are needed to explain isotope variations in deciduous trees (Eglin et al. 2010; Offermann et al. 2011; Rinne et al. 2015). Overall, although the imprint of photosynthetic discrimination is retained in the isotope signature of tree-rings, physiological information tends to be dampened as compared to the leaves (Roden and Farquhar 2012; Gessler et al. 2014; Fardusi et al. 2016). On top of that, fractionation processes can be modelled with reasonable accuracy, but their magnitude seems to be species- and site- specific (see e.g. Fig. 13.2). Therefore, trends in iWUE can be tracked using δ13C in tree rings, but without external validation it is not advisable to compare absolute estimates of intrinsic water use efficiency based on tree rings, e.g. among contrasting species or sites. Current advances in position-specific isotope analysis may offer in the future an alternative to overcome (or at least to better account for) post-photosynthetic fractionation, allowing to retrieve a ‘clean’ leaf-derived signal (Wieloch et al. 2018).