Jukuri, open repository of the Natural Resources Institute Finland (Luke) All material supplied via Jukuri is protected by copyright and other intellectual property rights. Duplication or sale, in electronic or print form, of any part of the repository collections is prohibited. Making electronic or print copies of the material is permitted only for your own personal use or for educational purposes. For other purposes, this article may be used in accordance with the publisher’s terms. There may be differences between this version and the publisher’s version. You are advised to cite the publisher’s version. This is an electronic reprint of the original article. This reprint may differ from the original in pagination and typographic detail. Author(s): V. Vitali, E. Martínez-Sancho, K. Treydte, L. Andreu-Hayles, I. Dorado-Liñán, E. Gutierrez, G. Helle, M. Leuenberger, N. Loader, K.T. Rinne-Garmston, G.H. Schleser, S. Allen, J.S. Waterhouse, M. Saurer and M.M. Lehmann Title: The unknown third – Hydrogen isotopes in tree-ring cellulose across Europe Year: 2022 Version: Preprint version Copyright: The Author(s) 2022 Rights: CC BY-NC-ND 4.0 Rights url: http://creativecommons.org/licenses/by-nc-nd/4.0/ Please cite the original version: Vitali V., Martínez-Sancho E., Treydte K., Andreu-Hayles L., Dorado-Liñán I., Gutierrez E., Helle G., Leuenberger M., Loader N., Rinne-Garmston K.T., Schleser G.H., Allen S., Waterhouse J.S., Saurer M., Lehmann M.M. (2022). The unknown third – Hydrogen isotopes in tree-ring cellulose across Europe. Science of The Total Environment 813, 152281. https://doi.org/10.1016/j.scitotenv.2021.152281. Science of the Total Environment 813 (2022) 152281 Contents lists available at ScienceDirect Science of the Total Environment j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv The unknown third – Hydrogen isotopes in tree-ring cellulose across Europe V. Vitali a,⁎, E. Martínez-Sancho b, K. Treydte b, L. Andreu-Hayles c,d,e, I. Dorado-Liñán f, E. Gutierrez g, G. Helle h, M. Leuenberger i, N. Loader j, K.T. Rinne-Garmston k, G.H. Schleser l, S. Allen m, J.S. Waterhouse n, M. Saurer a,1, M.M. Lehmann a,1 a Stable Isotope Research Center (SIRC), Ecosystem Ecology, Forest Dynamics, Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Forest Dynamics, CH-8903 Birmensdorf, Switzerland b Dendrosciences, Forest Dynamics, Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Forest Dynamics, CH-8903 Birmensdorf, Switzerland c Tree-Ring Laboratory, Lamont–Doherty Earth Observatory of Columbia University, Palisades, NY, USA d CREAF, Bellaterra (Cerdanyola del Vall.s), Barcelona, Spain e ICREA, Pg. Llu.s Companys 23, Barcelona, Spain f Department of Systems and Natural Resources, Universidad Politécnica de Madrid, Madrid, Spain. g Departament de Biologia Evolutiva, Ecologia i Ciències Ambientals, Universitat de Barcelona, Barcelona, Spain h German Research Centre for Geosciences, Section 4.3 Climate Dynamics and Landscape Evolution, Telegrafenberg, 14473 Potsdam, Germany i Climate and Environmental Physics Division and Oeschger Centre for Climate Change Research, University of Bern, Sidlerstrasse 5, 3012 Bern, Switzerland j Department of Geography, Swansea University, Swansea, UK k Natural Resources Institute Finland (Luke), Helsinki, Finland l FZJ Research Center Jülich, Institute of Bio- and Geosciences, Agrosphere (IBG-3), 52425 Jülich, Germany m Department of Natural Resources and Environmental Science, University of Nevada Reno, 1664 N Virginia St., Reno, NV 89557, USA n School of Life Sciences, Anglia Ruskin University, Cambridge, UK H I G H L I G H T S G R A P H I C A L A B S T R A C T • The climate information in hydrogen isotope ratios of tree rings (δ2Hc) is uncertain. • We present the first European-wide century-long study of δ2Hc. • δ2Hc is a weaker climate proxy compared to δ13Cc and δ18Oc. • The climate δ2Hc signal is stronger for Pinus than for Quercus. • δ2Hc records a mixture of hydrological, climatic, and physiological signals. ⁎ Corresponding author. E-mail address: valentina.vitali@wsl.ch (V. Vitali). 1 Joint senior author. http://dx.doi.org/10.1016/j.scitotenv.2021.152281 0048-9697/©2021TheAuthors. Published by Elsevie 0/). A B S T R A C T A R T I C L E I N F O Article history: Received 8 September 2021 Received in revised form 24 November 2021 Accepted 5 December 2021 Available online 21 December 2021 Editor: Christian Herrera This is the first Europe-wide comprehensive assessment of the climatological and physiological information recorded by hydrogen isotope ratios in tree-ring cellulose (δ2Hc) based on a unique collection of annually resolved 100-year tree- ring records of two genera (Pinus and Quercus) from 17 sites (36°N to 68°N). We observed that the high-frequency cli- mate signals in the δ2Hc chronologies were weaker than those recorded in carbon (δ13Cc) and oxygen isotope signals (δ18Oc) but similar to the tree-ring width ones (TRW). The δ2Hc climate signal strength varied across the continent and was stronger and more consistent for Pinus than for Quercus. For both genera, years with extremely dry summer con- ditions caused a significant 2H-enrichment in tree-ring cellulose. r B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4. http://crossmark.crossref.org/dialog/?doi=10.1016/j.scitotenv.2021.152281&domain=pdf http://dx.doi.org/10.1016/j.scitotenv.2021.152281 valentina.vitali@wsl.ch http://dx.doi.org/10.1016/j.scitotenv.2021.152281 http://creativecommons.org/licenses/by-nc-nd/4.0/ http://creativecommons.org/licenses/by-nc-nd/4.0/ http://www.sciencedirect.com/science/journal/ www.elsevier.com/locate/scitotenv V. Vitali et al. Science of the Total Environment 813 (2022) 152281 The δ2Hc inter-annual variability was strongly site-specific, as a result of the imprinting of climate and hydrology, but also physiological mechanisms and tree growth. To differentiate between environmental and physiological signals in δ2Hc, we investigated its relationships with δ18Oc and TRW.We found significant negative relationships between δ2Hc and TRW (7 sites), and positive ones between δ2Hc and δ18Oc (10 sites). The strength of these relationships was nonlinearly related to temperature and precipitation. Mechanistic δ2Hc models performed well for both genera at con- tinental scale simulating average values, but they failed on capturing year-to-year δ2Hc variations. Our results suggest that the information recorded by δ2Hc is significantly different from that of δ18Oc, and has a stronger physiological component independent from climate, possibly related to the use of carbohydrate reserves for growth. Advancements in the understanding of 2H-fractionations and their relationships with climate, physiology, and species-specific traits are needed to improve the modelling and interpretation accuracy of δ2Hc. Such advancements could lead to new in- sights into trees' carbon allocation mechanisms, and responses to abiotic and biotic stress conditions. Keywords: Climate change Dendroecology Deuterium European forests Isotope fractionation Mechanistic modelling Stable isotopes Tree physiology 1. Introduction Tree-ring cellulose chronologies of stable carbon (δ13Cc) and oxygen (δ18Oc) isotopes, together with tree-ring width (TRW), have been used ex- tensively to investigate the effects of past climatic conditions on tree growth (e.g. Andreu-Hayles et al., 2017; Barbour et al., 2002; Loader et al., 2007; Loader et al., 2020; Saurer et al., 1995; Saurer et al., 1997a; Shestakova and Martínez-Sancho, 2021) and physiological performance (e.g. Andreu- Hayles et al., 2011; Frank et al., 2015; Guerrieri et al., 2019; Klesse et al., 2018; Levesque et al., 2019; Martínez-Sancho et al., 2018). In particular, δ13Cc and δ18Oc have been recognized as strong continental-scale climate proxies for long-term trends in the ISONET network (Loader et al., 2013; Saurer et al., 2014; Shestakova et al., 2019; Treydte et al., 2007), as well as in the high-frequency in other European-scale studies (Vitali et al., 2021). In contrast, the environmental, climatic, and physiological informa- tion recorded by the third component of tree-ring cellulose (C6H10O5)n, the non-exchangeable carbon-bound hydrogen (δ2Hc), has been investigated far less. Some studies on δ2Hc and its relationship with climate have been conducted for single sites (Etien et al., 2009; Hafner et al., 2011; Haupt et al., 2011; Hilasvuori and Berninger, 2010; Lipp et al., 1991; Loader et al., 2008; Szczepanek et al., 2006), while continental-scale assessments of δ2Hc are few (Nakatsuka et al., 2020a; Voelker et al., 2014) and still lack- ing in Europe. The usability and interpretation of δ2Hc chronologies has im- proved recently as a result of methodological developments that have increased sample processing power (Filot et al., 2006; Sauer et al., 2009), thus advancing knowledge on how 2H-fractionation processes relate to physiology and biochemical pathways (Cormier et al., 2018; Sanchez- Bragado et al., 2019) and expanding δ2H analysis to various plant com- pounds (e.g. lipids and lignin: (Anhäuser et al., 2020; Gori et al., 2013; Riechelmann et al., 2017; Sachse et al., 2012)). These new advancements can facilitate the use of δ2Hc in ecological research; however, further knowledge on the fundamental drivers and spatiotemporal patterns of δ2Hc is needed to guide the interpretation of δ2Hc variability. Previous research has revealed uncertainties regarding the climate in- formation that can be inferred from δ2Hc chronologies (Boettger et al., 2014; Loader et al., 2008; Pendall, 2000; Waterhouse et al., 2002). While consistent temperature signals in δ2Hc were observed in some earlier stud- ies (Feng and Epstein, 1994; Gray and Song, 1984), more recent studies led to additional interpretations of the factors driving differences between sites and species. In Poland, strong correlations of δ2Hc with summer tempera- ture were found for Quercus, and with summer precipitation and winter temperature for Pinus (Szczepanek et al., 2006). In Austria, a Quercus δ2Hc chronology reflected both summer relative humidity and temperature (Haupt et al., 2011). In Finland, at the northernmost limit of Quercus' European distribution, precipitation and relative humidity, but not temper- ature, showed a strong correlation with δ2Hc (Hilasvuori and Berninger, 2010). These apparently contrasting results suggest that the climate signal stored in δ2Hc tree-ring chronologies is driven by a complex interaction be- tween climatic and environmental processes that varies across geographic regions (Lehmann et al., 2021b) and is amended by species-specific differ- ences (Arosio et al., 2020b). Controlled experiments support the hypothesis that physiological differences among plant species interfere with δ2Hc cli- mate signals, as demonstrated by the tight coupling between a plant's 2 metabolism, 2H-fractionations, and the resulting δ2H values (Cormier et al., 2018; Sanchez-Bragado et al., 2019). A recent study suggested that δ2Hc record the use of old carbohydrate reserves versus fresh photosyn- thates for wood formation (Lehmann et al., 2021b). These findings indicate to the potential of δ2Hc chronologies as physiological proxies for carbon al- location processes that are not captured by other tree-ring parameters. To make full use of the information captured in δ2Hc records, a better understanding of the 2H-fractionation pathways, from the uptake of H2O from the soil and CO2 from the atmosphere to cellulose formation, and the influence of environmental conditions and tree internal processes is needed. The hydrogen isotopic composition of environmental water (e.g. precipitation, soil water, and atmospheric water vapour) is closely linked to and reflects the primary factors affecting δ2Hc variation (Craig, 1961; Joussaume and Jouzel, 1993). Soil can contain water that reflects isotope ratios from several previous precipitation events, resulting in soil water iso- tope ratios potentially deviating considerably from annual precipitation (Allen et al., 2019). At the soil–tree interface, it is often claimed that no iso- tope fractionation occurs during root water uptake and transport (White et al., 1985), although an unexpected 2H-fractionation effect was recently observed at the root level (Barbeta et al., 2020). At the leaf level, strong iso- topic fractionation is induced by leaf and twig evapotranspiration (Cernusak et al., 2016; Treydte et al., 2014), and by the mixing with atmo- spheric water vapour (Lehmann et al., 2018), which leads to a enrichment of leaf water compared with source water (Cernusak et al., 2016; Dongmann et al., 1974). Given that oxygen isotopes share the same hydro- logical pathways as δ2Hc, a strong connection between the two isotopes would be expected (Brooks et al., 2010; Dansgaard, 1964; Edwards and Fritz, 1986), as shown in wet regions where the two isotopes show strong positive correlations (An et al., 2014). However, biochemical processes ap- pear to shape oxygen and hydrogen isotope ratios differently under distinct climatic conditions, and 2H-fractionations before and during cellulose syn- thesis are more variable than 18O-fractionations (Luo and Sternberg, 1992; Yakir and DeNiro, 1990). Photosynthetic isotope fractionation induces a relatively constant 18O-enrichment of sugars compared with leaf water (Lehmann et al., 2021b). On the contrary, there is a not yet fully quantified 2H-depletion in plant sugars compared with leaf water (Dunbar and Schmidt, 1984), which varies with environmental conditions and species (Lehmann et al., 2021a). In particular, for mature trees, post- photosynthetic isotope fractionations during the transport of sugars to sink tissues have been partly quantified for δ18Oc (Gessler et al., 2014; Treydte et al., 2014), but are still not fully resolved for δ2Hc. Recent litera- ture suggests that the use of carbon reserves causes an additional 2H- enrichment in leaf and tree-ring cellulose (Cormier et al., 2018; Kimak et al., 2015; Lehmann et al., 2021b). It remains uncertain to what extent δ2Hc acts as a hydrological and climatic indicator, as observed for δ18Oc (Treydte et al., 2007), andwhether δ2Hc stores information on plant metab- olism, physiology, and carbon allocation (Cormier et al., 2018; Lehmann et al., 2021b; Sanchez-Bragado et al., 2019). Mechanistic models are an important tool for interpreting of tree-ring iso- tope data. They have been developed and applied for the simulation of δ18Oc records in relation to hydrological (Sargeant et al., 2019), climatological sig- nals (Saurer et al., 2016), as well as for δ13Cc simulations that provided phys- iological information (Guerrieri et al., 2019; Lavergne et al., 2020), while it V. Vitali et al. Science of the Total Environment 813 (2022) 152281 has been seldomly applied for themodelling of δ2Hc in tree rings (Nabeshima et al., 2018; Nakatsuka et al., 2020a, 2020b; Voelker et al., 2014). Specifi- cally, the model of Roden and Ehleringer (2000); hereafter RE-model, is currently the applied model used to estimate δ2Hc in tree rings by taking into account the isotopic variation from hydrological sources (i.e. source water, water vapour, leaf water), as well as photosynthetic and post- photosynthetic isotope fractionations (autotrophic and heterotrophic), before and during stem cellulose biosynthesis (Roden and Ehleringer, 2000). Some of the RE-model parameters, which were previously considered as constants, have been found to have temporal variability, as for example in the isotope exchange rates between xylem and leaf water during cellulose formation (f), which exhibited variations among species and tissues (Cernusak et al., 2005; Song et al., 2014), and in the dependence on environmental conditions (Cheesman and Cernusak, 2017; Sternberg and Ellsworth, 2011). However, these studies focused on the variation of δ18Oc and included only on few spe- cies mostly in controlled experiments, and therefore, the conclusions are not fully utilisable for the modelling of δ2Hc in natural conditions. Furthermore, recent findings suggest that important processes causing hydrogen isotope fractionations in plants are not yet integrated into the RE-model, which may lead to miss-estimations as, for example the isotope fractionations of en- zymatic reactions connected to (i) the use of carbohydrate reserves (Kimak and Leuenberger, 2015; Nakatsuka et al., 2020a), (ii) environmental inputs (i.e. light and CO2), (iii) species-specific mechanisms (Arosio et al., 2020a; Cormier et al., 2018; Sanchez-Bragado et al., 2019). However, so far only few attempts have been made to address δ2Hc variability at a continental scale, showing considerable within-site variability (Nakatsuka et al., 2020a; Voelker et al., 2014). The extent to which large-scale or inter-annual varia- tions can be captured by the RE-model remains largely unknown, and the po- tential mis-estimations have yet to be quantified. Therefore, the interpretations of the RE-model results should be considered carefully in light of its limitations. The European isotope network ISONET offers the opportunity to evalu- ate tree-ring cellulose isotope chronologies (i.e. δ13Cc, δ18Oc, δ2Hc) and tree-ring-width chronologies (TRW) for two genera (Pinus, Quercus) over the past 100 years for 17 sites fromFinland to Spain, and Poland. High com- mon variance of δ13Cc and δ18Oc was shown by Treydte et al. (2007), and several site-specific analyses have been carried out during the past decade (see publication list Table S.1). The present study focuses on assessing spa- tiotemporal δ2Hc patterns across Europe for the first time, leveraging the unique and extensive ISONET δ2Hc network. Here, we systematically assessed the performance of δ2Hc as a potential climatic and physiological indicator for the different functional genera. Specifically, we investigated (i) the low- and high-frequency variability of the δ2Hc time series; (ii) the long- and short-term climate signals captured by δ2Hc; (iii) how δ2Hc relates to δ18Oc and TRW; and finally (iv) the ability of mechanistic models to pre- dict δ2Hc values at a continental scale. Accordingly, we hypothesized that: 1. The climatic information recorded in δ2Hc high-frequency chronologies: a. shows lower strength and lower large-scale agreement comparedwith the δ13Cc and δ18Oc chronologies. b. is specific to each genus, due to different physiological mechanisms. c. shows genus-specific responses for extreme climatic conditions. 2. The δ2Hc relationships with TRW are stronger than those with δ18Oc, in- dicating a larger share of physiological information than hydrological signals. 3. TheRE-model successfully estimates the continental-scale climatic effect on δ2Hc, but not the year-to-year variation. 2. Methods 2.1. Tree-ring chronologies and isotope analyses 2.1.1. European tree-ring network and climate data ISONET is a European tree-ring isotope network that includes Mediter- ranean, humid-temperate, continental, and subarctic climatic regions. The 3 sites are distributed along a latitudinal (37°81′ to 68°93′ N), longitudinal (−5°25′ to 30°93′ E), and elevational gradient (5 to 2′100 m a.s.l.), with high-elevation sites concentrated at lower latitudes. The selected sites are old-growth forests (mean ± SD of age = 454 ± 196 years) with two main genera (Quercus and Pinus, Table 1). Spatiotemporal patterns of the TRW, δ18Oc, and δ13Cc chronologies have been explored across the entire network (Balting et al., 2021; Shestakova et al., 2019; Treydte et al., 2007) and with 18 site-level studies (see citation list in Table S.1). How- ever, although the δ2Hc chronologies were measured and analysed with an annual resolution (1905–2002) at 17 sites of the ISONET network (Table 1), results from only 5 sites have been published individually so far (Etien et al., 2009; Haupt et al., 2011; Hilasvuori and Berninger, 2010; Loader et al., 2008; Szczepanek et al., 2006). For each study site, monthly climate data were extracted from the Cli- matic Research Unit, CRU TS4.03 for the period 1905–2002 (Harris et al., 2020), including mean temperature, precipitation and atmospheric water vapour. Monthly vapour pressure deficit (VPD)was calculated as the differ- ence between measured water vapour and the temperature-dependent sat- urated humidity value (esat): esat ¼ 6:1078 aT − 273:16 T − bð Þ where T is the averagemonthly temperature and where a=21.87 and b= 7.66 at T> 0 and a=17.27 and b=35.86 at T> 0, as described byMurray (1967). Across all sites for the June–July–August period, the mean temper- ature (MTJJA) ranged from 11 °C to 23 °C, total precipitation (MPJJA) from 35 mm to 590 mm, and VPD (VPDJJA) from 2.4 to 14.2 (hPa). 2.1.2. Study species The network contains data from two functionally different genera: Pinus (Pinus sylvestris L., Pinus uncinata Ram. and Pinus nigra Arn.) and Quercus (Quercus petraea (Matt.) Liebl. and Quercus robur L.) (Table 1). Pinus sites are mostly located in boreal and high-elevation Mediterranean zones, while Quercus sites dominate western and central European lower elevations. Pinus and Quercus species are widely distributed over Europe and are both ecologically and economically important. They have contrasting eco- physiological characteristics (Bréda et al., 2006; Bréda and Badeau, 2008; Merlin et al., 2015; Michelot et al., 2012a) and use opposite strategies to cope with periods of severe drought (Hochberg et al., 2018; Tyree and Cochard, 1996; Zweifel et al., 2009). Pinus is an evergreen coniferous genus with isohydric characteristics, which implies high stomatal control through a tight and continuous water potential homeostasis (Irvine et al., 1998; Leo et al., 2014; Salmon et al., 2015) and generally a shallow root sys- tem (Grossiord et al., 2014; Laitakari, 1927). In contrast, the deciduous broadleaf Quercus genus, exhibits an anisohydric behaviours keeping sto- mata open and high photosynthetic rates for long time periods (Aranda et al., 2000; Bréda et al., 1993; Klein, 2014), and typically has a deep root system with a large taproot and strong lateral roots (Zapater et al., 2011). 2.1.3. Tree-ring width and stable isotope measurements Increment cores were extracted at breast height from dominant old- growth trees at all sites (on average 46 trees per site). Following standard dendrochronological procedures (Cook and Kairiukstis, 1990), tree rings were visually crossdated and tree-ring widths (TRW) were measured at 0.01 mm precision. Cross-dating validation was carried out following stan- dard procedures (Holmes, 1983). From at least four trees per site, dated tree rings were cut and annually pooled, and cellulose was extracted using stan- dard extraction and purificationmethods (Boettger et al., 2007).Whole tree ringswere pooled for the isotope analysis of Pinus. ForQuercus only the late- wood was used, as the δ2H-values of early- and latewood inQuercus species have been shown to have a clear offset, although strongly correlated (r2 = 0.63), (Kimak, 2015), and in order to avoid undesired signals from previous-year carbohydrate reserves (Helle and Schleser, 2004). For the Swiss Cavergno site (CAV) no separation between early- and latewood was possible because the rings were too narrow. Table 1 Site details on species composition, geography, and climate ordered by latitude (North to South). Climatic variables represent the means of the common study period (1905–2002), obtained from CRU TS4.03. MTJJA = mean air temperature, MPJJA = mean precipitation, and VPDJJA = mean vapour pressure deficit, all for the months of June, July, and August; TRW= mean tree-ring width; isotope ratios of tree-ring cellulose (δ13Cc, δ18Oc, δ2Hc); δ2Hsw = modelled hydrogen source water. Site code Site name Country Species Lat (°N) Lon (°E) Elev. (m a.s.l.) MTJJA (°C) MPJJA (mm) VPDJJA (hPa) TRW (mm) δ13Cc (‰) δ18Oc (‰) δ2Hc (‰) δ2Hsw (‰) INA Kessi/Inari Finland P. sylvestris 68.93 28.42 150 11.7 169 3.9 0.52 −24.4 26.4 −95.6 −106 ILO Ilomantsi Finland P. sylvestris 62.98 30.98 200 14.8 192 4.6 0.29 −23.7 27.1 −106.1 −91.1 GUT Gutulia Norway P. sylvestris 62.00 12.18 800 11.4 234 3.5 0.49 −23.3 27.6 −94.3 −99.6 BRO Bromarv Finland Q. robur 60.00 23.08 5 15.3 189 4.1 1.85 −24.8 25.6 −93.3 −82.0 LCH Lochwood Scotland Q. robur 55.27 −3.43 175 12.9 325 2.4 1.16 −25.2 27.9 −46.3 −60.9 PAN Panemunes Silas Lithuania P. sylvestris 54.88 23.97 45 16.7 226 4.9 0.79 −22.9 28.7 −67.9 −67.6 SUW Suwalki Poland P. sylvestris 54.10 22.93 160 16.7 235 4.6 1.02 −23.1 28.5 −76.1 −73.4 WOB Woburn UK Q. robur 51.98 −0.59 50 15.7 169 3.9 1.37 −23.3 29.1 −54.1 −52.8 DRA Dransfeld Germany Q. petraea 51.5 9.78 320 15.8 227 7.5 1.41 −23.5 28.7 −19.6 −56.3 WIN Windsor UK P. sylvestris 51.41 −0.59 10 15.7 155 3.8 0.47 −22.9 30.4 −31.7 −48.5 GIB Niopolomice Gibiel Poland Q. robur 50.12 20.38 190 17.2 277 4.8 1.87 −25.7 27.8 −76.6 −62.6 POE Poellau Austria P. nigra 47.95 16.06 500 17.5 289 5.3 0.62 −24.2 27 −59.8 −65.5 VIG Vigera Switzerland P. sylvestris 46.5 8.77 1400 12.2 589 3.2 0.47 −23.1 30.8 −45.7 −110.0 CAV Cavergno Switzerland Q. petraea 46.35 8.60 900 12.2 589 3.2 1.11 −23.3 29.2 −51.2 −86.6 LIL Pinar de Lillo Spain P. sylvestris 43.07 −5.25 1600 16.3 143 5.6 0.47 −22.2 30.9 −46.1 −65.2 PED Pedraforca Spain P. uncinata 42.24 1.70 2100 16.1 214 6.1 0.49 −21.9 30.8 −32.1 −45.1 CAZ Cazorla Spain P. nigra 37.81 −2.96 1816 23.2 36.3 14.2 0.471 −21.1 33.6 −26.4 −48.7 V. Vitali et al. Science of the Total Environment 813 (2022) 152281 The extracted cellulose was then split in three parts for the analysis of the three isotope ratios. The details of δ18Oc and δ13Cc analyses are de- scribed in Treydte et al. (2007) and in the site-specific publications listed in Table S.1. For the δ2Hc analysis the cellulose was subjected to two differ- ent procedures: nitration and equilibration. In the first case, a complete re- moval of all exchangeable OH-groups was achieved by nitration (Boettger et al., 2007; Green, 1963), while in the second case, the exchangeable H was set at a known isotope value through the equilibration with water va- pour, which enabled the δ2H-values of the non-exchangeable H to be calcu- lated by mathematical correction (Schimmelmann, 1991; Wassenaar and Hobson, 2003). Both procedures produce comparable results as shown by Filot et al. (2006). The proceedings conducted at each lab and the details are given in Table S.2. For both methods, samples were subsequently con- verted to H2 by high-temperature pyrolysis and analysed by IRMS with a precision of ca. ±2‰. The δ2Hc values of all sites are referenced to the Vienna Standard Mean Ocean Water (VSMOW). The low-frequency domain of the 17 δ2Hc chronologies was ex- plored for the common period 1905–2002 by applying a 100-year smoothing spline (Cook and Kairiukstis, 1990), whereas the high- frequency domain (year-to-year variation) was obtained by calculating the first-order differences (FDiff) for TRW, δ13Cc, δ18Oc, and δ2Hc a sta- tionary mean was obtained to avoid autocorrelation, and to highlight interannual variability. The agreement among sites in the low and high frequency domains of the δ2Hc chronologies was assessed with pairwise correlations using the ‘corr’ function in the base package in R. All computations were performed using the R version 4.0.3 (R Core Team, 2020), and graphics were produced using the R package ggplot2 (Wickham, 2016). 2.2. Climate sensitivity analyses For comparisons with the FDiff chronologies, the same detrending was applied to the climate data: monthly temperature (T), precipitation (P), and vapour pressure deficit (VPD) series. Bootstrapped Pearson's correla- tions were performed between FDiff chronologies and FDiff climate series for the common period 1905–2002 using the R package treeclim (Zang and Biondi, 2015). All correlations were calculated for single months and for 3-month windows of the current year (April–May–June, June–July– August, August–September–October). Because, the June–July-August period yielded the highest correlations for all variables (Fig. S.1), it was the selected period for further climate analyses. Pearson's correlation 4 coefficients between the high-frequency variation (FDiff) of TRW and iso- tope chronologies with climate variables are indicated as X-signal (e.g. δ2Hc-signal). First, the spatial patterns of the obtained climate correlations across the entire network were assessed. To evaluate the differences in TRW-signal, δ13Cc-signal, δ18Oc-signal, and δ2Hc-signal for each species and climate variable, analyses of variance (ANOVA), Tukey's post hoc tests, and Bonferroni cor- rections were performed on absolute values of the climate correlations. Sig- nificance level was set at P < 0.05. The sensitivity of FDiff δ2Hc chronologies to extreme climatic conditions was also assessed. Years with positive (wet) and negative (dry) extreme values in summer P-PET (precipitationminus potential evapotranspiration) were selected at the site level by setting a threshold of +1.5 (wet) or−1.5 (dry) standard deviation from the average value. Summer potential evapo- transpiration (PET) was calculated with the Thornthwaite method (Thornthwaite, 1948) using the R package SPEI (Vicente-Serrano et al., 2010). Significant differences between genera, extreme events, and their in- teractions were evaluated through analyses of variance (two-way ANOVA), Tukey's post hoc tests, and Bonferroni corrections. For an assessment of the potential physiological and hydrological infor- mation in δ2Hc, the relationships between the δ2Hc and TRW chronologies, and between the δ2Hc and δ18Oc chronologies (both measured and FDiff) were investigated at site level using linear models. Further, the slopes from these relationships were assessed against mean precipitation (MPJJA) and mean temperature (MTJJA) in June–July–August (Table 1) and fitted with a polynomial function and a quadratic fit. 2.3. Mechanistic modelling of tree-ring δ2Hc Physiological mechanisms linked to the geographical patterns and tem- poral variability of our isotope chronologies were investigated by mecha- nistically modelling δ2H values for leaf water and tree-ring cellulose. Leaf water δ2H values (δ2Hlw) were estimated using the Craig Gordon Model (CG-Model) at the evaporative site (Craig and Gordon, 1965; Dongmann et al., 1974): δ2Hlw ¼ δ2Hsw þ εk þ εe þ δ2Hv − δ2Hsw − εk � � ∗ ea=ei (1) where ea/ei reflects the water vapour partial pressures outside and inside the leaf, εe and εk are temperature-dependent equilibrium and kinetic frac- tionation factors (Table 2), respectively, δ2Hsw is the hydrogen isotope ratio V. Vitali et al. Science of the Total Environment 813 (2022) 152281 of the source water, and δ2Hv is the atmospheric water vapour. Assuming equilibrium between δ2Hsw and δ2Hv, and that ea/ei is the mean annual rel- ative humidity (RH) (Cernusak et al., 2016), Eq. (1) was transformed to: δ2Hlw ¼ δ2Hsw þ εk þ εe þ −εe − εkð ÞÞ ∗ RHð (2) δ2Hlw was calculated annually for each site using mean summer tempera- tures and constant δ2Hsw estimates for each site (Table 1), extrapolated from the gridded data as defined by Bowen (2008). A generalized model was used and no further corrections for unenriched leaf water pools (Péclet effect) were applied (Roden et al., 2015) due to the lack of species-specific correction factors (Arosio et al., 2020b; Voelker et al., 2014). These dilution effects have been shown to be difficult to estimate accurately and to have large differences between subspecies and over time (Song et al., 2014). They require extensive physiological measurements that are not available for these species and timescales and have therefore not been included in the model to avoid unquantifiable errors. The CG-Model can be extended to organic material, such as tree rings, using the RE-model (Roden and Ehleringer, 2000). δ2HcRE ‰ð Þ ¼ f ∗ δ2Hsw þ εh � �þ 1 − fð Þ ∗ δ2Hlw þ εa � � ð3Þ where εa and εh are the specific autotrophic and heterotrophic 2H- biosynthetic fractionation factors, respectively, and f is the specific fraction before tree-ring cellulose synthesis (Table 2). 3. Results 3.1. High- and low-frequency variability in the δ2Hc site chronologies The δ2Hc chronologies showed a 75‰ difference in absolute values be- tween the northern-most sites (INA, ILO, and GUT) with the lowest values and the Spanish site (CAZ) with the highest values (Fig. 1a, b). Over the course of the 20th century there was no clear common trend among the site δ2Hc mean values, except for the last 10 years when a positive trend oc- curred at most Quercus sites (Fig. 1b) and for a few Pinus sites especially at higher latitudes (Fig. 1a). The long-term trends showed strong site-to-site variability and sparse significant correlations (P < 0.05) between sites (Fig. S.2a), with clusters consisting of the three Spanish sites, the Polish and British sites, and the two Swiss sites. The year-to-year variability retained in the site FDiff δ2Hc chronologies were of a similar range within and between sites (Fig. 1c, d), with smaller variance for Quercus than Pinus. The FDiff δ2Hc chronologies of some geo- graphically close sites were significantly correlated (r = 0.4, CAV–VIG, ILO–GUT; r=0.3, PAN–SUW, LIL–PED–CAZ), while this agreementwas re- duced among sites that were farther apart (Fig. S.2b). Table 2 Summary of fractionation factors for the modelling of δ2Hc and corresponding literature CG-model Kinetic fractionation factor εk H 25‰ Equilibrium fractionation factor εe H ee ¼ exp 24:844 273:16þTmeanð Þ2 � � ∗ 1000 � � − 76:248 273:16þTmean � � þ 0:052612 � �� RE-model Fraction of exchange before cellulose synthesis f H 0.36 Biosynthetic fractionation factors εa H −171‰ εh H 158‰ 5 3.2. Climate correlations 3.2.1. Spatial patterns and species-specific differences In general, δ2Hc-signal was negative regarding summer precipitation and positive regarding summer temperature and VPD across the whole network (Fig. 2a). No clear latitudinal pattern was observed for the δ2Hc-signal and any climatic parameter. However, at the fringes of the network, the strength in the δ2Hc-signal of the northernmost sites was lower than at the southernmost ones, especially for Pinus. The TRW-signal was highly site- specific (Figs. 3, S.3), with no clear latitudinal gradient. At sites with a high TRW-signal, we observed a low δ2Hc-signal (r ~ 0, INA), and vice versa (r ~ 0.5, CAV), although this negative relationship between TRW-signal and δ2Hc-signal was not consistent across the network. The δ13Cc-signal and δ18Oc-signal were generally high (up to r = 0.7) and were negative regarding summer precipitation and positive regarding tem- perature and VPD (Figs. 3, S.3). Contrary to the δ2Hc-signal, temperature and VPD-δ13Cc-signal showed a distinct latitudinal gradient, with strong correla- tions at higher latitudes and the British sites. In comparison, no distinct lat- itudinal trend was observed for the δ18Oc-signal, although the northern sites generally showed lower correlations with summer temperature and VPD than the southern sites. We detected significant differences in climate sensitivity between gen- era, as the summer temperature δ2Hc-signal in Pinuswas significantly higher than in Quercus, resulting in significant differences between their climate signals (Table S.3). Specifically, for Pinus, the absolute δ2Hc-signal was con- sistent for all climatic parameters (r ~ 0.3), but it was significantly lower than the absolute δ13Cc-signal for all climate parameters, and significantly lower than δ18Oc-signal for precipitation and VPD (Fig. 3). For Quercus, the absolute δ2Hc-signal was significantly lower than the absolute δ18Oc-signal and δ13Cc-signal for all climate parameters (Fig. 3). Interestingly, no signifi- cant differences were found between δ2Hc-signal and TRW-signal of the two genera for any climatic variable (Fig. 3). Regarding changes along the gra- dients of precipitation and temperature covered by the sites, no significant effect was found in the climate signal strength recorded by Pinus δ2Hc (Fig. 4). Conversely, for Quercus the strength of the temperature-δ2Hc-signal increased significantly along the precipitation gradient and decreased along the temperature gradient (Fig. 4c, d), while the strength of the precipitation-δ2Hc-signal did not vary significantly along the gradients, indi- cating a stronger impact of temperature on the δ2H-variations particularly at cool and wet sites. 3.2.2. Comparison of the climate signals recorded in δ2Hc and δ18Oc When examining the relationship between δ2Hc-signal and δ18Oc-signal (H- O-signal) across the whole network, we observed a divergence between the two genera with decreasing latitude (Fig. 5). Pinus showed a negative H- O-signal for all three climate variables (significant for precipitation and VPD, but not for temperature). On the contrary, Quercus showed a positive sources. (Cernusak et al., 2016; Merlivat, 1978) − 1 � ∗ 1000 Temperature dependent (Cernusak et al., 2016; Majoube, 1971) (Roden and Ehleringer, 1999) Autotrophic fractionation (Yakir and DeNiro, 1990) Heterotrophic fractionation (Yakir and DeNiro, 1990) Fig. 1. The δ2Hc site chronologies for Pinus (a) and Quercus (b) for the common period (1905–2002), and their low-frequency variability illustrated by a 100-year smoothing spline (dashed lines). The corresponding high-frequency site chronologies calculated asfirst differences (FDiff) are given for Pinus (c) andQuercus (d). Site colours are ordered by latitude (North to South in the legend). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 2.Mapping of Pearson's correlation coefficients of the FDiff δ2H (δ2Hc-signal) with summer (June–July–August) climate variables (precipitation, temperature, and vapour pressure deficit (VPD), also as FDiff). Significant correlations (P < 0.05) are indicated by asterisks. V. Vitali et al. Science of the Total Environment 813 (2022) 152281 6 Fig. 3. Differences in the absolute δ13Cc-signal (C), δ18Oc-signal (O), δ2Hc-signal (H), and TRW-signal (TRW) for summer (June–July–August) climate variables (precipitation, temperature, and VPD) for the genera Pinus (top panel) and Quercus (bottom panel) (Table S.3). Significant differences between pairs are indicated by asterisks (P < 0.05 = *, P < 0.01 = **). The colour of the points indicates the sites, as shown in Fig. 1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) V. Vitali et al. Science of the Total Environment 813 (2022) 152281 H-O-signal for temperature and VPD, and a negative one for precipitation (although none of these trends were significant). At northern sites the two genera showed similar values for δ18Oc-signal and δ2Hc-signal regarding temperature and VPD (δ18Oc-signal r = 0.5; δ2Hc-signal r = 0.2). On the contrary, at southern sites where δ18Oc-signal was around 0 for both genera, δ2Hc-signal increased for Pinus (r = 0.5) and decreased for Quercus (r = 0.1). For example, in Finland (Pinus) the VPD δ18Oc-signal was 0.6 and the VPD δ2Hc-signal was 0.01. On the 7 contrary, in Spain (Pinus) the VPD δ18Oc-signal was −0.2 and the VPD δ2Hc-signal was 0.5. Similarly, in the UK (Quercus), where δ18Oc-signal showed strong correlations with VPD, δ2Hc-signal showed only weak correlations. 3.2.3. Responses of δ2Hc to summer climate extremes We investigated the influence of particularly dry and wet summers on δ2Hc by correlations to P-PET. Climatic conditions had a significant Fig. 4. Relationships between average summer precipitation (Pjja) and the δ2Hc-signal (Pearson's correlation coefficients) for precipitation (a) and temperature (c), and between average summer temperature (Tjja) and the δ2Hc-signal for precipitation (b) and temperature (d). Pinus and Quercus are shown separately. Points reflect single site's δ2Hc-signal values. Significant differences between pairs are indicated by asterisks (P < 0.05 = *, P < 0.01 = **). The size of the boxes is dependent on the genera's ranges on the x axis. V. Vitali et al. Science of the Total Environment 813 (2022) 152281 effect on the FDiff δ2Hc for both genera, while the δ2Hc differences between the two genera were not significant (Table S.4). On average, dry summers (−1.5 SD from the mean P-PET for the June–July–August period) had sig- nificantly higher values of δ2Hc, by 3‰ for both genera, compared with years with normal summer conditions (Fig. 6). The δ2Hc values in years with wet summers (+1 SD)were not significantly different from the values in normal years for either genus. Pinus showed continuously increasing δ2Hc from wet to dry years (Fig. 6a), whereas the δ2Hc values of Quercus were not significantly different in wet years than in dry and normal years (Fig. 6b). 3.3. Relationships between δ2Hc and δ18Oc and between δ2Hc and TRW We found a significant positive relationship between the δ2Hc and δ18Oc for 10 of the 17 sites (Fig. 7a), while only 2 sites showed a significant neg- ative relationship (Table S.5). The explained variance (r2) ranged from 0.003 to 0.2, and the slopes from−4.2 to 3.6 (Table S.5). Regarding the re- lationship between the FDiff δ2Hc and FDiff δ18Oc, 11 of the 17 sites showed a significant positive correlation, whereas only one site showed a signifi- cant negative correlation, with the slopes ranging from −1.8 to 4.1, and the r2 from 0.01 to 0.28 (Fig. S.5, Table S.5). When analysing the effect of mean climate conditions on the FDiff δ2Hc and FDiff δ18O relationships, 8 we observed a non-linear relationship along the MTJJA gradient, with the highest slopes at 5 °C (r2 = 0.4). On the contrary, a linear relationship be- tween δ2Hc-δ18Oc slopes and MPJJA was found (r2 = 0.5) (Fig. S.8). The linear relationships between the δ2Hc and TRW records were highly variable and site-specific, with both positive (4 of 7 sites significant) and negative slopes (7 of 10 sites significant) (Fig. 7b), r2 ranged from 0.002 to 0.33, and slopes ranged from −22 to 32 (Table S.5). For the FDiff standardized data the linear site-level models showed lower r2 values (0.001 to 0.32), and only 9 of the 17 sites showed a significant relationship (3 positive and 4 negatives; Fig. S.5, Table S.5). The slopes of the FDiff δ2Hc and TRW relationships had lower r2 values (0.2) than those of the δ2Hc and δ18O relationships, and in both cases the slopes decreased with increasing temperature and precipitation (Fig. S.5, Table S.5). 3.4. Mechanistic modelling of the tree-ring isotope ratios The RE-model captured the continental-scale mean δ2Hc levels but did not capture the year-to-year site variability. The alignment of the fitted lin- ear models between observed and predicted δ2Hc values across several sites in Europe (Fig. 8 black dotted line) showed r2 values of 0.46 and 0.33 and slopes of 0.76 and 0.91 for Pinus and Quercus, respectively. However, we observed a consistent overestimation of the modelled compared with the Fig. 5. Linear relationships between δ2Hc-signal and δ18Oc-signal (H-O-signal) regarding precipitation, temperature, and VPD per genus (Pinus and Quercus). Site colours are ordered by latitude (North to South). Linear models are fitted for each climatic variable and genus, and significant relationships (P < 0.05) are indicated by asterisks. Pinus regressions are indicated by solid lines and Quercus regressions by dashed lines. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) V. Vitali et al. Science of the Total Environment 813 (2022) 152281 measured δ2Hc values by an average of 43‰ for both species (Fig. 8, Table S.6). For Quercus, the fitted linear model had a slope of 0.91, indicat- ing an almost constant model performance across all sites. For Pinus, the fitted linear model had a slope of 0.7, indicating better model perfor- mance for high-latitude sites. The Swiss sites VIG (Pinus) and CAV (Quercus) showed the largest offset between measured and modelled values, i.e. 99.5‰ and 70.7‰, respectively. These two study sites have the same climate data, being in the same interpolation cell, although they differ in elevation by 500 m. For this reason, a second linear model exclud- ing these two sites was calculated, resulting in an improvement of the gen- eral model fit for both genera (r2 = 0.83 and 0.69, slope= 1.2 and 1.7 for Pinus and Quercus, respectively), resulting in an improved fir for higher lat- itudes sites (Fig. 8). Independently of the dataset used, the year-to-year var- iability within the sites was not captured by the model (the average values of the FDiff model are r2=0.04 and 0.02, slope=0.6 and 0.4 for Pinus and Quercus, respectively; Fig. S.7). 4. Discussion 4.1. Low- and high-frequency variations in the δ2Hc chronologies This is the first European-scale study exploring 100-year δ2Hc chronolo- gies in both low- and high-frequency domains (Fig. 1). The 2H-depletion with increasing latitude follows the naturally occurring isotopic variations in precipitation (i.e. Global Meteoric Water Line, Fig. S.9; (Allen et al., 2019; Craig, 1961; Craig and Gordon, 1965)), independent of the two gen- era. The comparison of the long-term trends among sites resulted in 9 relatively low, sporadic significant correlations (Fig. S.3a). In the last de- cade, a relative 2H-enrichment was apparent at most of the sites, particu- larly those with Quercus (Fig. S.1), although the strength of this increase was site-specific. The between-site agreement decreased when year-to- year variability was assessed. In this case, only relatively geographically close sites remained correlated (Fig. S.3b). The fact that these low- and high-frequency δ2Hc patterns are very site specific, pose issues for further extrapolations of the δ2Hc-signal (e.g. for climate reconstructions: Christiansen and Ljungqvist, 2017). Therefore, both frequency domains should be considered in future studies exploring the potential of δ2Hc for dendroclimatological purposes. 4.2. The climate sensitivity of δ2Hc across Europe compared with TRW, δ13Cc, and δ18Oc In temperate forests, the strength of δ18Oc-signal and δ13Cc-signal across different climates has already been clearly shown in European networks (Shestakova and Martínez-Sancho, 2021; Treydte et al., 2007; Vitali et al., 2021), although biases connected to long-term trends have been re- ported (Esper et al., 2010) and a complete understanding of the related fractionation processes is still missing (Gessler et al., 2014). Nonethe- less, these signals were clearly more consistent than the δ2Hc-signal. Our results confirm our first hypothesis (Hp1a), showing that the δ2Hc-signal was on average weaker than the δ13Cc-signal and δ18Oc-signal, but similar to the TRW-signal (Figs. 1, 2, S.2), with large variation at the continental scale. At the centre of our network, we observed the low- est temperature-δ2Hc-signal. However, significant correlations with Fig. 6. The FDiff δ2Hc differences among years with climatic characteristics different than the site P-PET mean (+1.5 SD = wet summer conditions and −1.5 SD = dry summer conditions; Fig. S.5) for the two studied genera (Table S.4). Significant differences between groups are indicated by asterisks (P < 0.01 = **; P < 0.001 = ***). V. Vitali et al. Science of the Total Environment 813 (2022) 152281 summer temperature (Szczepanek et al., 2006) and summer relative air humidity (Haupt et al., 2011) have been reported in some studies in Poland and Austria respectively. The interplay of biological processes, at stand level (e.g. competition for light and nutrients: Giuggiola et al., 2016) and at tree physiological level (e.g. leaf gas exchange: Guerrieri et al., 2019, pollution: Boettger et al., 2014; Savard, 2010), can potentially mask the recorded δ2Hc-signal. The loss of climatic information in the δ2Hc records could be due to (i) biochemical 2H-fractionations at the leaf level (Yakir and DeNiro, 1990), (ii) kinetic isotope effects in biochemical reactions involved in the fixation of hydrogen in different positions of the glucose molecule (Augusti et al., 2006; Waterhouse et al., 2002), (iii) isotope fraction- ations and H-exchange with water during the biosynthesis of carbohy- drates (Cormier et al., 2018). Thus, the δ2Hc-signal can be assumed to be the result of the complex interaction between climatic and physiolog- ical processes, explaining why it is difficult to find one major climate driver at the continental scale or across climatic areas (Shestakova et al., 2019). 4.3. Genus-dependent differences in the climate sensitivity of δ2Hc We observed a significant genus dependency of the climate sensitiv- ity of δ2Hc (Figs. 3, 4), supporting our hypothesis (Hp1b). When com- paring the functionally distant genera Pinus and Quercus, differences in the δ2Hc-signal were evident, similar to δ13Cc-signal and δ18Oc-signal (Martínez-Sancho et al., 2018). Pinus' δ2Hc climate sensitivity was stron- ger than that of Quercus for all climatic variables across the whole network (Fig. 3). A decoupling of H-O-signal between the two genera 10 was observed at sites with a low temperature- and VPD-δ18Oc-signal, where Pinus showed a strong δ2Hc-signal, while Quercus showed a weak δ2Hc-signal (Fig. 5). These contrasting signals are likely the result of dif- ferent physiological mechanisms affecting isotope fractionation and water uptake dynamics, as explored below. 4.3.1. Use of fresh and stored photosynthates for cellulose formation Hydrogen isotope patterns of leaf (Kimak et al., 2015) and tree-ring cel- lulose (Epstein and Yapp, 1976; Mayr et al., 2003) suggest that stored pho- tosynthates in heterotrophic tissues (branches, stem, roots) are likely more 2H-enriched than fresh photosynthates and that considerable use of carbohydrate reserves for growth could, lead to the observed bias in the δ2Hc-signal (Lehmann et al., 2021b). Deciduous ring-porous species like Quercus partly rely on stored photo- synthates for earlywood growth (Pilcher and Frenzel, 1995), resulting in mixed climatic information from previous years and the current growing season (Reynolds-Henne et al., 2009). In our study, we reduced the poten- tial effect of storage remobilization by sampling only the latewood of Quercus (Waterhouse et al., 2002). This strategy was successful for the δ18Oc-signal, which showed high correlation values (Fig. 3), especially at the British sites (Rinne et al., 2013), but it did not seem effective for the δ2Hc-signal, which exhibited low ones. It should be considered that storage use could still play a role under stressful conditions even for late-wood growth (Sarris et al., 2013) However, other yet to be characterized isotope fractionations may likely interfere autonomously with the current-year δ2Hc-signal in Quercus (e.g. growth release: Lehmann et al., 2021b). On the contrary, Pinus relies largely on fresh (current-year) photosyn- thates as the main source for tree-ring production (Dickmann and Kozlowski, 1970) in temperate sites and under non-drought conditions. A large photosynthetic demand and high transpiration rate ensure a fast turn- over of leaf-water in Pinus and, therefore, a reliable recording of the sea- sonal climate signals (Dickmann and Kozlowski, 1970; Glerum, 1980), as confirmed by our results (Fig. 3). 4.3.2. Root system and source water interactions Water from different soil layers systematically varies in its isotopic com- position. The topsoil is directly dependent on precipitation events and sub- ject to evaporative isotopic enrichment, thus recording current-year climatic signals. On the contrary, the isotopic composition of deeper soil layers is also dependent on winter precipitation and snow melt, and there- fore previous years' climate (Allen et al., 2019). Since no isotopic fractiona- tion typically occurs during root-water uptake (Dawson and Siegwolf, 2007; Geris et al., 2015; Tang et al., 2000), the xylem water (also known as source water) integrated during tree-ring cellulose formation (Augusti et al., 2006; Roden and Ehleringer, 2000) can reflect the uptake depth of soil water. The typical genus-specific rooting depth therefore influences the δ2Hc-signal. Shallow-rooted Pinus generally carries a stronger current- year climate signal, because its sourcewater mostly integrates precipitation events (Fig. 3), as previously shown for other shallow-rooted species (Tang and Feng, 2001). On the contrary, Quercus xylem water has been shown to carry a deep-soil signal or a mixture between deep and shallow soil (Barbeta et al., 2020), hence diluting its precipitation-induced δ2Hc-signal. However, Quercus was reported to use current-year precipitation in clay- rich soil, (e.g. WOB site: Rinne et al., 2013), stressing once again the site- specificity of δ2Hc-signal. While the different rooting depths can explain the stronger δ2Hc-signal for Pinus than for Quercus, the decoupling of H-O-signal at the drier and warmer sites remains unexplained (Fig. 5). A growing number of studies have sug- gested that isotope fractionations can occur during root water uptake (Evaristo et al., 2017; Oerter et al., 2019), as δ2H in stemwater has been ob- served to be progressively enriched with increasing transpiration while δ18O still reflected the soil water isotopic signal (Barbeta et al., 2020; Ellsworth and Williams, 2007). This mismatch between δ2H and δ18O in xylem water could further depend on multiple interacting factors, such as (i) the heterogeneity within the soil matrix (Oerter et al., 2014); (ii) isotope separation between bound and mobile soil water (Tang and Feng, 2001); Fig. 7. Linear relationships between (a) δ2Hc and δ18Oc, and (b) δ2Hc and TRW for each site. See Fig. S.7 for the same relationships but with FDiff data. The fitted linearmodel equations, explained variance (r2), and significance (p-values) are given in Table S.5. Pearson's correlation coefficients (r) calculated for δ2Hc and δ18Oc and for δ2Hc and TRW are given in Fig. S.6. V. Vitali et al. Science of the Total Environment 813 (2022) 152281 (iii) root interactions withwater pools, and isotope compartmentalization (Zhao et al., 2016); and (iv) methodological artefacts (Chen et al., 2020), which could explain the decoupling of H-O-signal between the two genera. Fig. 8. Relationship between measured (δ2Hc) and modelled (δ2Hc RE) values of δ2Hc c (black dotted lines), and for all sites without the outlier sites VIG and CAV (red solid lin a black solid line. Results of genera-specific regressions are given in Table S.6. See F references to colour in this figure legend, the reader is referred to the web version of th 11 Based on our results, we conclude that the δ2Hc of Pinus is a more sensi- tive indicator of environmental changes than the δ2Hc of Quercus, although further research is needed to investigate the driving factors of these species' differences. hronologies for Pinus (a) and Quercus (b). Linear regressions were fitted for all sites es). Significance is indicated by asterisks * (P < 0.05). The 1:1 line is indicated by ig. S.7 for the same relationships but with FDiff data. (For interpretation of the is article.) V. Vitali et al. Science of the Total Environment 813 (2022) 152281 4.3.3. Influence of extreme wet and dry summer conditions We hypothesized that years with extreme wet or dry summer condi- tions, ergo with contrasting VPD and precipitation conditions, lead to dis- tinct ecophysiological responses that shape δ2Hc-signal (Hp1c). Climatic conditions induce changes in the isotopic composition typically leading to higher δ2H values of leaf and source water in dry years compared with nor- mal years, and lower values in wet years. These patterns should subse- quently be imprinted on δ2Hc (Cernusak et al., 2016; Roden and Ehleringer, 2000). We indeed observed a significant 2H-enrichment in tree rings of both genera in dry years compared with normal years, but not a clear 2H-depletion in wet years (Fig. 6). This was unforeseen because the climatic signal transfer from the water isotopes to the tree rings was ex- pected to be more consistent under wet conditions than dry conditions be- cause of an increased translocation of recent photosynthates towards the stem cambium (Michelot et al., 2012b; Pflug et al., 2015; Simard et al., 2013). The general 2H-enrichment in tree-ring cellulose under dry condi- tions thus does not necessarily reflect a climatic signal derived from changes in the isotopic composition of water. It might rather be derived from drought-driven changes in metabolic pathways (Cormier et al., 2018) or from greater use of (potentially 2H-enriched) carbohydrate re- serves for wood formation under stress conditions (Lehmann et al., 2021a). Therefore, in contrast to our hypothesis (Hp1c) that expected differ- ences between the two genera based on their contrasting ecophysiological traits which affect leaf and source water isotope modification in response to drought (Klein, 2014; Martín-Gómez et al., 2017), both Quercus and Pinus showed 2H-enrichment in dry years compared with normal years. Our results provide novel evidence that δ2Hc valuesmight function as an in- dicator for extreme drought conditions, potentially due to specific 2H- fractionations that are only triggered under reduced water availability. However, additional δ2Hc studies focusing on drought effects are needed to further understand the potential of δ2Hc as a drought proxy. 4.4. Disentangling the hydrological and physiological information in δ2Hc To fully understand the nature of the δ2Hc information recorded in tree rings, it is necessary to quantify the individual impacts of climatic, hydro- logical, and physiological factors. Therefore, we evaluated the relationship of δ2Hc chronologies with those of δ18Oc and TRW, and the performance of the mechanistic model in estimating δ2Hc in tree rings. 4.4.1. Relationship between δ2Hc and δ18Oc and between δ2Hc and TRW The hydrological and temperature signals are recorded by δ2Hc and δ18Oc at a continental scale (Gray and Song, 1984; Saurer et al., 1997b), as shown by their distribution along the global meteoric water line (Fig. S.9). Our results indicate that δ18O transfers the source water signal to the tree rings better than δ2H, resulting in a stronger climate signal (Fig. 3). However, the well-known relationship between the two water iso- topes in hydrological cycles barely hold for the year-to-year variation in tree-ring cellulose, as shown by the weak relationships between the δ2Hc and δ18Oc chronologies for most of our studied sites, regardless of species and geographical location (Fig. 7a). This result is likely caused by temporal variability in fractionations and additional factors that influence 2H- fractionations more than 18O-fractionations. However, the climatic and hy- drological information that dominates δ18Oc is likely also present, albeit weakly, in δ2Hc. Themagnitude of δ2Hc may depend on site-specific precip- itation and temperature conditions, as indicated by the positive influence of MPJJA and negative influence of MTJJA on the δ2Hc–δ18Oc relationships (Fig. S.8) (Sprenger et al., 2016). Trees' physiological strategies in response to the local environment are recorded partly by TRW (Hartl-Meier et al., 2015), and thus the δ2Hc–TRW relationships should reflect shared physiological informa- tion. Indeed, our δ2Hc–TRW relationships were stronger than δ2Hc–δ18Oc relationships at many sites (Fig. 7b). This supports that at many sites the δ2Hc recorded a stronger physiological signal rather than a climatic or hydrological one as proposed in our second hypothe- sis (Hp2). Whereas in most cases a significant negative TRW-δ2Hc 12 relationship was found in our study, a few significant positive relation- ships also appeared (Fig. S.6). Although neither positive nor negative re- lationships clustered in a particular geographical area, or in distinct climatic regions, we observed a negative influence of MTJJA on the slope of the δ2Hc–TRW relationship (Fig. S.8). Lehmann et al. (2021b) reported that negative δ2Hc-TRW relationships indicate greater use of carbohydrate storage under stress conditions at four sites in Europe and Asia, where growth was limited by precipitation or light. On the contrary, at a site where temperature was the growth-limiting factor, the δ2Hc-TRW relationship became positive. Nonetheless, this uneven share of negative (7) and positive (4) relationships between δ2Hc and TRW suggests that stronger site-specific variables, like stand density or soil depth, are needed to advance our understanding of δ2Hc variations. 4.4.2. Performance of the mechanistic RE-model The mechanistic RE-model was previously found to successfully explain the δ2Hc of young broadleaf trees (Alnus incana, Betula occidentalis and Populus fremontii) under controlled experimental conditions (Roden et al., 2000). However, its applicability across a variety of species and ecological conditions was not originally tested. Our δ2Hc simulations produced using the RE-model support our last hypothesis (Hp3). We observed a reasonable δ2Hc simulation values for both genera on a continental scale, albeit with an overall consistent overestimation (Fig. 8). This suggests that the RE-model successfully takes into account the large-scale isotopic variations in precip- itation, while the constant overestimation may be attributed to a lack of di- lution effects (e.g. Péclet effect or two-pool estimations) in the CG-model to estimate δ2Hlw (Roden et al., 2015). Therefore, species-specific and tempo- rally variable dilution effects should be considered in future studies (Voelker et al., 2014). As hypothesized (Hp3), themodelled δ2Hc data did not capture the site- specific inter-annual variability, as shown by both δ2Hc and FDiff δ2Hc data (Figs. 8, S.7). Several sources of uncertainty should be considered to evalu- ate the performance of the RE-model. First, the low spatial resolution of the gridded climate data available for the last century. For example, the two Swiss sites located at different elevations (CAV and VIG) share the same cli- matic information, leading to a larger overestimation of the δ2Hc values at the higher-elevation Pinus site (VIG). Second, the lack of annual site-level δ2H precipitation data (used as source water) and unknown mixing effects of the precipitation water with ground water also play a role (Waterhouse et al., 2002). Third, the model does not account for the stochasticity of en- vironmental parameters, such as temperature, that may influence biochem- ical isotope fractionation (Zhou et al., 2011). Finally, the species differences in 2H-fractionation (Arosio et al., 2020b) are not considered in mechanistic models. We observed thatmodelled δ2Hc has a betterfit withmeasured Pinus δ2Hc thanmeasuredQuercus δ2Hc at continental scale (Fig. 8). These results corrob- orate previous findings that the isotopic signature of precipitation is better captured in the δ2Hc of shallow-rooted (Pinus) than deep-rooted (Quercus) trees, due to their ability to access deeper groundwater (Voelker et al., 2014;Waterhouse et al., 2002). Our results clearly demonstrate the necessity to take into account species-specific or functional-trait-dependent (Arosio et al., 2020b; Cormier et al., 2018; Sanchez-Bragado et al., 2019), as well as age-dependent weighting factors (Arosio et al., 2020a). These additions would account for the non-controlled conditions of natural forests and should be integrated into future mechanistic models. 4.5. Potential and limitations of δ2Hc chronologies In this study we demonstrate that δ2Hc chronologies record a mix of hydrological and climatic signals, while the physiological information is still not readily accessible, due to remaining gaps in the understanding of 2H-fractionations and their interactions with C partitioning mecha- nisms, and with climatological, hydrological processes (Waterhouse et al., 2002). New high-throughput methods will make δ2Hc measure- ments increasingly accessible and economically viable (Filot et al., 2006; Sauer et al., 2009; Wassenaar et al., 2015), which will likely V. Vitali et al. Science of the Total Environment 813 (2022) 152281 boost the knowledge on 2H-fractionations in plants in the next years. Thus, further targeted research to disentangle the interference of climatic and metabolic signals is needed, carried out in well-monitored forest sites or in growth chambers with controlled conditions. This untapped physio- logical signal in δ2Hc could lead to further insight into carbohydrate storage use, drought stress, and carbon allocation processes, information that is useful not only from a retrospective view, but also for improving growth models and thus predicting future tree performance. 5. Conclusions The results presented here set a baseline in the understanding of the information stored in δ2Hc. We show that the climate information recorded by δ2Hc is weaker than for δ13Cc and δ18Oc, and that it is site and genera dependent, with Pinus showing a stronger climatic signal than Quercus. Thus, information recorded by δ2Hc is different from that of δ13Cc and δ18Oc, with a stronger physiological component independent from climate, although these effects cannot yet be disentangled. As shown by our results, these physiological effects are significant, but not constant across years, sites, or genera. Nonethe- less, years with dry summer conditions showed a consistent 2H- enrichemnt for both genera. Therefore, combination with other tree- ring-derived parameters is highly advisable to provide complementary information on tree performance. Finally, we showed that the current mechanistic δ2Hc models performed well at a continental but not at a temporal scale. Clearly, improved climate and source water data, as well as further work on mechanistic isotope models is need to improve the spatio-temporal estimation of δ2Hc. This will eventually result in tree rings δ2Hc to be a highly valuable archive of how trees' physiol- ogy and biochemistry has been influenced by past environmental changes. Funding This study was supported by the ISONET EU-project (EVK-CT-2002- 00147): ML: SNF projects 103829, 107554, 111699, and 116540 for the work at the University of Bern. VV, MS: SNF project 200020_182092. EMS and KT: SNF project TRoxy (No. 200021_175888). ID-L: Fundació La Caixa through the Junior Leader Program (LCF/BQ/ LR18/11640004). NJL: NERC (NE/P011527/1) and SSHRC (895-2019-1015). KRG: ERC (755865) and Academy of Finland (295319). EG: Spanish Ministerio de Educación y Ciencia project REN2002- 11476-E/CLI. MML: SNF Ambizione project TreeCarbo (No. 179978). CRedit authorship contribution statement VV completed the data analyses and wrote the first manuscript draft with MS and MML. The manuscript was further developed with the help of EMS and KT. Data were provided by MS, KT, KRG, ML, NJL, and STA. All authors contributed to the writing of the manuscript and agreed upon the final version. Declaration of competing interest The authors declare that they have no known competing financial inter- ests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We acknowledge Nyfeler P. and Moret H., as well as Filot M., for devel- oping the online equilibration system and performing measurements at the 13 University of Bern, Switzerland. We thank Rolfe J. and Hall M. (Godwin Laboratory, UK) for technical support. We are grateful to Muntán E., Bosch O., and Planells O., from the Department of Ecology, UB. Spain, who helped us with field and laboratory work. We thank Grinsted M.J. and Wilson A.T. for inspiring model ideas. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2021.152281. References Allen, S.T., Kirchner, J.W., Braun, S., Siegwolf, R.T.W., Goldsmith, G.R., 2019. Seasonal ori- gins of soil water used by trees. Hydrol. Earth Syst. Sci. 23 (2), 1199–1210. An, W., Liu, X., Leavitt, S.W., Xu, G., Zeng, X., Wang, W., et al., 2014. Relative humidity his- tory on the batang-litang plateau of western China since 1755 reconstructed from tree- ring δ18O and δD. Clim. Dyn. 42 (9–10), 2639–2654. Andreu-Hayles, L., Planells, O., Gutiérrez, E., Muntan, E., Helle, G., Anchukaitis, K.J., et al., 2011. Long tree-ring chronologies reveal 20th century increases in water-use efficiency but no enhancement of tree growth at five iberian pine forests. Glob. Chang. Biol. 17 (6), 2095–2112. Andreu-Hayles, L., Ummenhofer, C.C., Barriendos, M., Schleser, G.H., Helle, G., Leuenberger, M., et al., 2017. 400 years of summer hydroclimate from stable isotopes in iberian trees. Clim. Dyn. 49 (1–2), 143–161. Anhäuser, T., Sehls, B., Thomas, W., Hartl, C., Greule, M., Scholz, D., et al., 2020. Tree-ring δ2H values from lignin methoxyl groups indicate sensitivity to european-scale tempera- ture changes. Palaeogeogr. Palaeoclimatol. Palaeoecol. 546, 109665. Aranda, I., Gil, L., Pardos, J.A., 2000. Water relations and gas exchange in Fagus sylvatica L. and Quercus petraea (Mattuschka) liebl. in a mixed stand at their southern limit of distri- bution in Europe. Trees 14 (6), 344–352. Arosio, T., Ziehmer-Wenz, M.M., Nicolussi, K., Schlüchter, C., Leuenberger, M., 2020a. Cambial-age related correlations of stable isotopes and tree-ring widths in wood samples of tree-line conifers. Biogeosci. Discuss. https://doi.org/10.5194/bg-2020-406 [preprint]. Arosio, T., Ziehmer-Wenz, M.M., Nicolussi, K., Schlüchter, C., Leuenberger, M., 2020. Larch cellulose shows significantly depleted hydrogen isotope values with respect to evergreen conifers in contrast to oxygen and carbon isotopes. Front. Earth Sci. 8. Augusti, A., Betson, T.R., Schleucher, J., 2006. Hydrogen exchange during cellulose synthesis distinguishes climatic and biochemical isotope fractionations in tree rings. New Phytol. 172 (3), 490–499. Balting, D.F., Ionita, M., Wegmann, M., Helle, G., Schleser, G.H., Rimbu, N., et al., 2021. Large-scale climate signals of a european oxygen isotope network from tree rings. Clim. Past 17 (3), 1005–1023. Barbeta, A., Gimeno, T.E., Clavé, L., Fréjaville, B., Jones, S.P., Delvigne, C., et al., 2020. An ex- planation for the isotopic offset between soil and stem water in a temperate tree species. New Phytol. 227 (3), 766–779. Barbour, M.M., Walcroft, A.S., Farquhar, G.D., 2002. Seasonal variation in δ13C and δ18O of cellulose from growth rings of Pinus radiata. Plant Cell Environ. 25 (11), 1483–1499. Boettger, T., Haupt, M., Knöller, K., Weise, S.M., Waterhouse, J.S., Rinne, K.T., et al., 2007. Wood cellulose preparation methods and mass spectrometric analyses of δ13C and δ18O, and nonexchangeable δ2H values in cellulose, sugar, and starch: an interlaboratory comparison. Anal. Chem. 79 (12), 4603–4612. Boettger, T., Haupt, M., Friedrich, M., Waterhouse, J.S., 2014. Reduced climate sensitivity of carbon, oxygen and hydrogen stable isotope ratios in tree-ring cellulose of silver fir (Abies alba Mill.) influenced by background SO2 in Franconia (Germany, Central Europe). Envi- ron. Pollut. (Barking, Essex 1987) 185, 281–294. Bowen, G.J., 2008. Spatial analysis of the intra-annual variation of precipitation isotope ratios and its climatological corollaries. J. Geophys. Res. 113 (D5), 1–10. Bréda, N., Badeau, V., 2008. Forest tree responses to extreme drought and some biotic events: towards a selection according to hazard tolerance? Compt. Rendus Geosci. 340 (9–10), 651–662. Bréda, N., Cochard, H., Dreyer, E., Granier, A., 1993. Field comparison of transpiration, sto- matal conductance and vulnerability to cavitation of Quercus petraea and Quercus robur under water stress. Ann. For. Sci. 50 (6), 571–582. Bréda, N., Huc, R., Granier, A., Dreyer, E., 2006. Temperate forest trees and stands under se- vere drought: a review of ecophysiological responses, adaptation processes and long-term consequences. Ann. For. Sci. 63 (6), 625–644. Brooks, R.J., Barnard, H.R., Coulombe, R., McDonnell, J.J., 2010. Ecohydrologic separa- tion of water between trees and streams in a Mediterranean climate. Nat. Geosci. 3 (2), 100–104. Cernusak, L.A., Farquhar, G.D., Pate, J.S., 2005. Environmental and physiological controls over oxygen and carbon isotope composition of Tasmanian blue gum Eucalyptus globulus. Tree Physiol. 25 (2), 129–146. Cernusak, L.A., Barbour, M.M., Arndt, S.K., Cheesman, A.W., English, N.B., Feild, T.S., et al., 2016. Stable isotopes in leaf water of terrestrial plants. Plant Cell Environ. 39 (5), 1087–1102. Cheesman, A.W., Cernusak, L.A., 2017. Infidelity in the outback: climate signal recorded in Δ18O of leaf but not branch cellulose of eucalypts across an australian aridity gradient. Tree Physiol. 37 (5), 554–564. Chen, Y., Helliker, B.R., Tang, X., Li, F., Zhou, Y., Song, X., 2020. Stemwater cryogenic extrac- tion biases estimation in deuterium isotope composition of plant source water. Proc. Natl. Acad. Sci. U. S. A. 117 (52), 33345–33350. https://doi.org/10.1016/j.scitotenv.2021.152281 https://doi.org/10.1016/j.scitotenv.2021.152281 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092116233531 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092116233531 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092116283491 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092116283491 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092116283491 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092124111479 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092124111479 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092124111479 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092116350574 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092116350574 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092119166295 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092119166295 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092119166295 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092119337642 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092119337642 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092119337642 https://doi.org/10.5194/bg-2020-406 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121415718 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121415718 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121415718 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121424616 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121424616 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121424616 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121431810 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121431810 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121438541 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121438541 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121438541 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121448270 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121448270 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121453352 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121453352 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121453352 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092105595199 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092105595199 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092105595199 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092105595199 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092039100679 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092039100679 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121462925 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121462925 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121462925 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121476764 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121476764 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121476764 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121526429 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121526429 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121526429 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092039111784 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092039111784 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092039111784 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092041390760 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092041390760 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092041390760 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121594201 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121594201 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122028290 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122028290 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122028290 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121554745 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121554745 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121554745 V. Vitali et al. Science of the Total Environment 813 (2022) 152281 Christiansen, B., Ljungqvist, F.C., 2017. Challenges and perspectives for large-scale tempera- ture reconstructions of the past two millennia. Rev. Geophys. 55 (1), 40–96. Cook, E.R., Kairiukstis, L.A., 1990. Methods of Dendrochronology: Applications in the Envi- ronmental Sciences. Springer Netherlands, Dordrecht. Cormier, M.-A., Werner, R.A., Sauer, P.E., Gröcke, D.R., Leuenberger, M.C., Wieloch, T., et al., 2018. 2H-fractionations during the biosynthesis of carbohydrates and lipids imprint a met- abolic signal on the δ2H values of plant organic compounds. NewPhytol. 218 (2), 479–491. Craig, H., 1961. Isotopic variations in meteoric waters. Science (New York, N.Y.) 133 (3465), 1702–1703. Craig, H., Gordon, I., 1965. I. Deuterium and oxygen 18 variations in the ocean and the ma- rine atmosphere. Stable Isotopes in Oceanographic Studies and Paleotemperatures: Consiglio nazionale delle richerche, Laboratorio di geologia nucleare. Dansgaard, W., 1964. Stable isotopes in precipitation. Tellus 16 (4), 436–468. Dawson, T.E., Siegwolf, R.T.W. (Eds.), 2007. Stable Isotopes as Indicators of Ecological Change. Academic, Oxford. Dickmann, D.I., Kozlowski, T.T., 1970. Mobilization and incorporation of photoassimilated C by growing vegetative and reproductive tissues of adult Pinus resinosa ait. trees. Plant Physiol. 45 (3), 284–288. Dongmann, G., Nürnberg, H.W., Förstel, H., Wagener, K., 1974. On the enrichment of H2 18O in the leaves of transpiring plants. Radiat. Environ. Biophys. 11 (1), 41–52. Dunbar, J., Schmidt, H.-L., 1984. Measurement of the 2H/1H ratios of the carbon bound hy- drogen atoms in sugars. Fresenius’ Z. Anal. Chem. 317 (857), 853. Edwards, T., Fritz, P., 1986. Assessing meteoric water composition and relative humidity from18O and2H in wood cellulose: paleoclimatic implications for southern Ontario, Canada. Appl. Geochem. 1 (6), 715–723. Ellsworth, P.Z., Williams, D.G., 2007. Hydrogen isotope fractionation during water uptake by woody xerophytes. Plant Soil 291 (1–2), 93–107. Epstein, S., Yapp, C.J., 1976. Climatic implications of the D/H ratio of hydrogen in C-H groups in tree cellulose. Earth Planet. Sci. Lett. 30, 252–261. Esper, J., Frank, D.C., Battipaglia, G., Büntgen, U., Holert, C., Treydte, K., et al., 2010. Low- frequency noise in δ13C and δ18O tree ring data: a case study of Pinus uncinata in the spanish Pyrenees. Glob. Biogeochem. Cycles 24 (4) (n/a-n/a). Etien, N., Daux, V., Masson-Delmotte, V., Mestre, O., Stievenard, M., Guillemin, M.T., et al., 2009. Summer maximum temperature in northern France over the past century: instru- mental data versus multiple proxies (tree-ring isotopes, grape harvest dates and forest fires). Clim. Chang. 94 (3–4), 429–456. Evaristo, J., McDonnell, J.J., Clemens, J., 2017. Plant source water apportionment using sta- ble isotopes: a comparison of simple linear, two-compartment mixing model approaches. Hydrol. Process. 31 (21), 3750–3758. Feng, X., Epstein, S., 1994. Climatic implications of an 8000-year hydrogen isotope time series from bristlecone pine trees. Science (New York, N.Y.) 265 (5175), 1079–1081. Filot, M.S., Leuenberger, M., Pazdur, A., Boettger, T., 2006. Rapid online equilibrationmethod to determine the D/H ratios of non-exchangeable hydrogen in cellulose. Rapid Commun. Mass Spectrom. 20 (22), 3337–3344. Frank, D.C., Poulter, B., Saurer, M., Esper, J., Huntingford, C., Helle, G., et al., 2015. Water-use efficiency and transpiration across european forests during the anthropocene. Nature Clim Change 5 (6), 579–583. Geris, J., Tetzlaff, D., McDonnell, J., Anderson, J., Paton, G., Soulsby, C., 2015. Ecohydrological separation in wet, low energy northern environments? A preliminary as- sessment using different soil water extraction techniques. Hydrol. Process. 29 (25), 5139–5152. Gessler, A., Ferrio, J.P., Hommel, R., Treydte, K., Werner, R.A., Monson, R.K., 2014. Stable isotopes in tree rings: towards a mechanistic understanding of isotope fractionation and mixing processes from the leaves to the wood. Tree Physiol. 34 (8), 796–818. Giuggiola, A., Ogée, J., Rigling, A., Gessler, A., Bugmann, H., Treydte, K., 2016. Improvement of water and light availability after thinning at a xeric site: which matters more? A dual isotope approach. New Phytol. 210 (1), 108–121. Glerum, C., 1980. Food sinks and food reserves of trees in temperate climates. N. Z. J. For. Sci. 176–185. Gori, Y., Wehrens, R., Greule, M., Keppler, F., Ziller, L., La Porta, N., et al., 2013. Carbon, hy- drogen and oxygen stable isotope ratios of whole wood, cellulose and lignin methoxyl groups of Picea abies as climate proxies. Rapid Commun. Mass Spectrom. 27 (1), 265–275. Gray, J., Song, S.J., 1984. Climatic implications of the natural variations of D/H ratios in tree ring cellulose. Earth Planet. Sci. Lett. 70 (1), 129–138. Green, J.W., 1963. Wood cellulose. In: Whistler, R.L. (Ed.), Methods in Carbohydrate Chemistry. Grossiord, C., Gessler, A., Granier, A., Berger, S., Bréchet, C., Hentschel, R., et al., 2014. Im- pact of interspecific interactions on the soil water uptake depth in a young temperate mixed species plantation. J. Hydrol. 519, 3511–3519. Guerrieri, R., Belmecheri, S., Ollinger, S.V., Asbjornsen, H., Jennings, K., Xiao, J., et al., 2019. Disentangling the role of photosynthesis and stomatal conductance on rising forest water- use efficiency. Proc. Natl. Acad. Sci. U. S. A. 116 (34), 16909–16914. Hafner, P., Robertson, I., McCarroll, D., Loader, N.J., Gagen, M., Bale, R.J., et al., 2011. Cli- mate signals in the ring widths and stable carbon, hydrogen and oxygen isotopic compo- sition of Larix decidua growing at the forest limit in the southeastern european Alps. Trees 25 (6), 1141–1154. Harris, I., Osborn, T.J., Jones, P., Lister, D., 2020. Version 4 of the CRU TS monthly high- resolution gridded multivariate climate dataset. Sci. Data 7 (1), 109. Hartl-Meier, C., Zang, C., Büntgen, U., Esper, J., Rothe, A., Göttlein, A., et al., 2015. Uniform climate sensitivity in tree-ring stable isotopes across species and sites in a mid-latitude temperate forest. Tree Physiol. 35 (1), 4–15. Haupt, M., Weigl, M., Grabner, M., Boettger, T.A., 2011. 400-year reconstruction of july rela- tive air humidity for the Vienna region (eastern Austria) based on carbon and oxygen sta- ble isotope ratios in tree-ring latewood cellulose of oaks (Quercus petraea Matt. Liebl.). Clim. Chang. 105 (1–2), 243–262. 14 Helle, G., Schleser, G.H., 2004. Beyond CO2-fixation by rubisco - an interpretation of 13C/ 12C variations in tree rings from novel intra-seasonal studies on broad-leaf trees. Plant Cell Environ. 27 (3), 367–380. Hilasvuori, E., Berninger, F., 2010. Dependence of tree ring stable isotope abundances and ring width on climate in finnish oak. Tree Physiol. 30 (5), 636–647. Hochberg, U., Rockwell, F.E., Holbrook, N.M., Cochard, H., 2018. Iso/Anisohydry: a plant- environment interaction rather than a simple hydraulic trait. Trends Plant Sci. 23 (2), 112–120. Holmes, R., 1983. Computer-assisted quality control in tree-ring dating and measurement. Tree-Ring Bull. 43, 69–78. Irvine, J., Perks, M.P., Magnani, F., Grace, J., 1998. The response of Pinus sylvestris to drought: stomatal control of transpiration and hydraulic conductance. Tree Physiol. 18 (6), 393–402. Joussaume, S., Jouzel, J., 1993. Paleoclimatic tracers: an investigation using an atmospheric general circulation model under ice age conditions: 2. Water isotopes. J. Geophys. Res. 98 (D2), 2807–2830. Kimak, A., 2015. Tracing Physiological Processes of Long Living Tree Species and Their Re- sponse on Climate Change Using Triple Isotope Analyses. Kimak, A., Leuenberger, M., 2015. Are carbohydrate storage strategies of trees traceable by early–latewood carbon isotope differences? Trees 29 (3), 859–870. Kimak, A., Kern, Z., Leuenberger, M., 2015. Qualitative distinction of autotrophic and hetero- trophic processes at the leaf level by means of triple stable isotope (C-O-H) patterns. Front. Plant Sci. 6, 1008. Klein, T., 2014. The variability of stomatal sensitivity to leaf water potential across tree spe- cies indicates a continuum between isohydric and anisohydric behaviours. Funct. Ecol. 28 (6), 1313–1320. Klesse, S., Weigt, R., Treydte, K.S., Saurer, M., Schmid, L., Siegwolf, R.T., et al., 2018. Oxygen isotopes in tree rings are less sensitive to changes in tree size and relative canopy position than carbon isotopes. Plant Cell Environ. 41 (12), 2899–2914. Laitakari, E., 1927. Morphological study of scots pine root system. Acta For. Fenn. 33 (1). Lavergne, A., Voelker, S., Csank, A., Graven, H., de Boer, H.J., Daux, V., et al., 2020. Historical changes in the stomatal limitation of photosynthesis: empirical support for an optimality principle. New Phytol. 225 (6), 2484–2497. Lehmann, M.M., Goldsmith, G.R., Schmid, L., Gessler, A., Saurer, M., Siegwolf, R.T.W., 2018. The effect of 18 O-labelled water vapour on the oxygen isotope ratio of water and assim- ilates in plants at high humidity. New Phytol. 217 (1), 105–116. Lehmann, M.M., Schuler, P., Cormier, M.-A., Allen, S.T., Leuenberger, M., Voelker, S., 2021a. Chapter: the stable hydrogen isotopic signature: from source water to tree rings. Stable Isotopes in Tree Rings: Inferring Physiological, Climatic and Environmental Responses in print. Lehmann, M.M., Vitali, V., Schuler, P., Leuenberger, M., Saurer, M., 2021b. More than cli- mate: hydrogen isotope ratios in tree rings as novel plant physiological indicator for stress conditions. Dendrochronologia 65, 125788. Leo, M., Oberhuber, W., Schuster, R., Grams, T.E.E., Matyssek, R., Wieser, G., 2014. Evaluat- ing the effect of plant water availability on inner alpine coniferous trees based on sap flow measurements. Eur. J. For. Res. 133 (4), 691–698. Levesque, M., Andreu-Hayles, L., Smith, W.K., Williams, A.P., Hobi, M.L., Allred, B.W., et al., 2019. Tree-ring isotopes capture interannual vegetation productivity dynamics at the biome scale. Nat. Commun. 10 (1), 742. Lipp, J., Trimborn, P., Fritz, P., Moser, H., Becker, B., Frenzel, B., 1991. Stable isotopes in tree ring cellulose and climatic change. Tellus Ser. B Chem. Phys. Meteorol. 43 (3), 322–330. Loader, N.J., Gagen, M., Robertson, I., Jalkanen, R., McCarroll, D., 2007. Extracting climatic information from stable isotopes in tree rings. In: Dawson, T.E., Siegwolf, R.T.W. (Eds.), Stable Isotopes as Indicators of Ecological Change. Academic, Oxford, pp. 25–48. Loader, N.J., Santillo, P.M., Woodman-Ralph, J.P., Rolfe, J.E., Hall, M.A., Gagen, M., et al., 2008. Multiple stable isotopes from oak trees in southwestern Scotland and the potential for stable isotope dendroclimatology in maritime climatic regions. Chem. Geol. 252 (1–2), 62–71. Loader, N.J., Young, G., Grudd, H., McCarroll, D., 2013. Stable carbon isotopes from Torneträsk, northern Sweden provide a millennial length reconstruction of summer sun- shine and its relationship to Arctic circulation. Quat. Sci. Rev. 62, 97–113. Loader, N.J., Young, G.H.F., McCarroll, D., Davies, D., Miles, D., Bronk, Ramsey C., 2020. Summer precipitation for the England and Wales region, 1201–2000 ce from stable oxy- gen isotopes in oak tree rings. J. Quat. Sci. 35 (6), 731–736. Luo, Y., Sternberg, L.D.S.L., 1992. Hydrogen and oxygen isotopic fractionation during hetero- trophic cellulose synthesis. J. Exp. Bot. 43 (1), 47–50. Majoube, M., 1971. Fractionnement en oxygène 18 et en deutérium entre l’eau et sa vapeur. J. Chim. Phys. 68, 1423–1436. Martínez-Sancho, E., Dorado-Liñán, I., Gutiérrez Merino, E., Matiu, M., Helle, G., Heinrich, I., et al., 2018. Increased water-use efficiency translates into contrasting growth patterns of scots pine and sessile oak at their southern distribution limits. Glob. Chang. Biol. 24 (3), 1012–1028. Martín-Gómez, P., Aguilera,M., Pemán, J., Gil-Pelegrín, E., Ferrio, J.P., 2017. Contrasting ecophys- iological strategies related to drought: the case of amixed stand of scots pine (Pinus sylvestris) and a submediterranean oak (Quercus subpyrenaica). Tree Physiol. 37 (11), 1478–1492. Mayr, C., Frenzel, B., Friedrich, M., Spurk, M., Stichler, W., Trimborn, P., 2003. Stable carbon- and hydrogen-isotope ratios of subfossil oaks in southern Germany: methodology and ap- plication to a composite record for the holocene. The Holocene 13 (3), 393–402. Merlin, M., Perot, T., Perret, S., Korboulewsky, N., Vallet, P., 2015. Effects of stand composi- tion and tree size on resistance and resilience to drought in sessile oak and scots pine. For. Ecol. Manag. 339, 22–33. Merlivat, L., 1978. Molecular diffusivities of H2 16O, HD16O, and H2 18O in gases. J. Chem. Phys. 69 (6), 2864. Michelot, A., Bréda, N., Damesin, C., Dufrêne, E., 2012a. Differing growth responses to cli- matic variations and soil water deficits of Fagus sylvatica, Quercus petraea and Pinus sylvestris in a temperate forest. For. Ecol. Manag. 265, 161–171. http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121594982 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121594982 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092041576416 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092041576416 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121551140 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092121551140 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122066277 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122066277 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106144219 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106144219 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106144219 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092041577102 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092042001586 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092042001586 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122137659 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122137659 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122137659 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092042017566 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092042017566 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092042363168 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092042363168 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092044246119 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092044246119 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092044246119 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122144460 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122144460 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122069090 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122069090 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122068933 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122068933 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122068933 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122181580 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122181580 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122181580 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122186948 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122186948 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122186948 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122187036 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122187036 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122187280 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122187280 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092122187280 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092124104452 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092124104452 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092124104452 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106159058 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106159058 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106159058 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106165772 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106165772 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106165772 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021092210 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021092210 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021092210 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021106778 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021106778 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106172742 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106172742 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106172742 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106172742 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021116459 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021116459 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092044411122 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092044411122 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106179833 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106179833 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106179833 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106187255 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106187255 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106194088 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106194088 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106194088 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106194088 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106339963 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106339963 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106347934 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106347934 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106347934 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021131900 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021131900 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021131900 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021131900 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106354353 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106354353 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106354353 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106363488 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106363488 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106369798 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106369798 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106369798 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021142455 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021142455 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021167130 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021167130 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021167130 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021439957 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021439957 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092021439957 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025103190 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025103190 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106376056 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106376056 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025117576 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025117576 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025117576 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106383245 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106383245 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106383245 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025126877 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025126877 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025126877 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025191989 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025202564 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025202564 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092025202564 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106390817 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106390817 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092050367913 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092050367913 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092050367913 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106397149 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106397149 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106397149 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106549426 http://refhub.elsevier.com/S0048-9697(21)07357-5/rf202112092106549426 http://refhub.elsevier.com/S0048-9697(