Available online atwww.sciencedirect.com ScienceDirect GeochimicaetCosmochimicaActa199(2017)351–369 www.elsevier.com/locate/gca Evaluating the use of amber in palaeoatmospheric reconstructions: The carbon-isotope variability of modern and Cretaceous conifer resins Jacopo Dal Corsoa,b,⇑, Alexander R. Schmidtc, Leyla J. Seyfullahc, Nereo Pretoa, Eugenio Ragazzid, Hugh C. Jenkynse, Xavier Delclo`sf, Didier Ne´raudeaug, Guido Roghih aDipartimentodiGeoscienze,Universita` degliStudidiPadova,ViaGradenigo6,35131Padova,Italy bDipartimentodiFisicaeScienzedellaTerra,Universita` degliStudidiFerrara,viaSaragat1,44122Ferrara,Italy cAbteilungGeobiologie,Georg-August-Universita¨tGo¨ttingen,Goldschmidtstraße3,37077Go¨ttingen,Germany dDipartimentodiScienzedelFarmaco,Universita` degliStudidiPadova,L.goMeneghetti2,35131Padova,Italy eDepartmentofEarthSciences,UniversityofOxford,SouthParksRoad,OxfordOX13AN,UK fDepartamentof‘‘Cie`nciesdelaTerraidel’Ocea`”,FacultatdeCie`nciesdelaTerra,UniversitatdeBarcelona,MartiiFranquess/n, 08028Barcelona,Spain gUniversite´ Rennes1,Ge´osciences,CNRSUMR6118,CampusdeBeaulieu,Bat.15,263AvenueduGeneralLeclerc,RennesCedex,France hIstitutodiGeoscienzeeGeorisorse(IGG-CNR),viaGradenigo6,35131Padova,Italy Received24May2016;acceptedinrevisedform16November2016;Availableonline22November2016 Abstract Stable carbon-isotope geochemistry of fossilized tree resin (amber) potentially could be a very useful tool to infer the composition of past atmospheres. To test the reliability of amber as a proxy for the atmosphere, we studied the variabil- ity of modern resin d13C at both local and global scales. An amber d13C curve was then built for the Cretaceous, a per- iod of abundant resin production, and interpreted in light of data from modern resins. Our data show that hardening changes the pristine d13C value by causing a 13C-depletion in solid resin when compared to fresh liquid–viscous resin, probably due to the loss of 13C-enriched volatiles. Modern resin d13C values vary as a function of physiological and envi- ronmental parameters in ways that are similar to those described for leaves and wood. Resin d13C varies between plant species and localities, within the same tree and between different plant tissues by up to 6‰, and in general increases with increasing altitudes of the plant-growing site. We show that, as is the case with modern resin, Cretaceous amber d13C has a high variability, generally higher than that of other fossil material. Despite the high natural variability, amber shows a negative 2.5–3‰ d13C trend from the middle Early Cretaceous to the Maastrichtian that parallels published terrestrial d13C records. This trend mirrors changes in the atmospheric d13C calculated from the d13C and d18O of benthic forami- niferal tests, although the magnitude of the shift is larger in plant material than in the atmosphere. Increasing mean annual precipitation and pO could have enhanced plant carbon-isotope fractionation during the Late Cretaceous, 2 whereas changing pCO levels seem to have had no effect on plant carbon-isotope fractionation. The results of this study 2 suggest that amber is a powerful fossil plant material for palaeoenvironmental and palaeoclimatic reconstructions. Improvement of the resolution of the existing data coupled with more detailed information about botanical source ⇑ Correspondingauthorat:Hanse-Wissenschaftskolleg(HWK),Lehmkuhlenbusch4,27753Delmenhorst,Germany,andLeibnizCenter forTropicalMarineEcology(ZMT),Fahrenheitstraße6,28359Bremen,Germany. E-mailaddress:[email protected](J.DalCorso). http://dx.doi.org/10.1016/j.gca.2016.11.025 0016-7037/(cid:1)2016ElsevierLtd.Allrightsreserved. 352 J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 and environmental growing conditions of the fossil plant material will probably allow a more faithful interpretation of amber d13C records and a wider understanding of the composition of the past atmosphere. (cid:1)2016Elsevier Ltd. All rightsreserved. Keywords: Coniferresin;Amber;Carbonisotopes;Palaeoclimate;Cretaceous 1. INTRODUCTION resin exudation, for example changes in plant carbon- isotope discrimination linked to environmental stresses SinceC3plantstakeupatmosphericCO duringphoto- such as insect infestation or water availability (Murray 2 synthesis and record its carbon-isotope signature, fossil et al., 1994, 1998; Nissenbaum and Yakir, 1995; McKellar plant remains (such as leaves and wood) can be used to et al., 2008, 2011; Dal Corso et al., 2011, 2013; Tappert reconstruct the palaeoatmosphere (e.g. Arens et al., 2000; et al., 2013). However, contrary to that of wood and leaf, Gro¨cke, 2002; Bechtel et al., 2008; Diefendorf et al., less research has focused on developing amber as a 2010).Plantsdiscriminateagainst13Cduringphotosynthe- palaeoatmosphereproxy(Tappertetal.,2013).Inaddition, sis,andthedegreeof13Cfractionation(D13C )dependsnot the lack of sufficient data on the carbon-isotope geochem- P onlyonplantphysiologybutalsoonanumberofenviron- istry of modern resin, i.e. on the variation of d13C RESIN mental factors and post-photosynthetic D13C processes, (resin d13C) under different environmental conditions, ren- P which determine the d13C of plant tissues (Arens et al., dersinterpretation of d13C problematic. AMBER 2000; Diefendorf et al., 2010; Schubert and Jahren, 2012). Here,asatestcase,weexplorethevalueofd13C AMBER The pristine carbon-isotope composition of wood and as a proxy for the Cretaceous atmosphere. We studied the leaves, the most commonly used tissues in chemostrati- variability of modern d13C using samples produced RESIN graphicanalysis,isfurtherchangedbydiagenesis, through byextantconifersfromdifferenttemperatetotropicalenvi- whichmoietieswithdifferentd13Csignaturesareselectively ronments in order to understand whether resin carbon- removed (e.g. van Bergen and Poole, 2002; Bechtel et al., isotope behaviour is similar to that of plant tissues. We 2002). Natural variability and diagenesis make the recon- do not aim to explain the biological and biochemical rea- structionsofpastatmosphericcarbon-isotopecomposition sons behind the observed behaviour of modern d13C RESIN based on plant d13C analysis difficult because, particularly but rather to highlight the patterns and variability that in deep-time studies, it is often impossible to separate could hamper palaeoclimatic reconstructions and physiological and environmental effects from atmospheric chemostratigraphy.NewCretaceousd13C data,cou- AMBER signals and evaluate the diagenetic effect (Diefendorf pledwitha compilationofpublished data,were compared etal.,2010).Thed13Coffossilwoodandleaves(d13C in order to combine terrestrial and marine d13C records, WOOD and d13C ) has been successfully used to infer changes and interpreted in light of both present-day isotopic vari- LEAF in the carbon-isotope composition of past atmosphere– ability andthe Cretaceous climate. ocean systems, an approach supported by the evidence of d13C excursions synchronously recorded in both terrestrial 2. MATERIAL ANDMETHODS organic matter and marine carbonates (e.g. Gro¨cke, 2002; Strauss and Peters-Kottig, 2003; Dal Corso et al., 2011). 2.1.Methodicalbackground:Factorscontrollingthed13Cof Recordsofd13C andd13C parallelthoseofmar- modern C3 plants WOOD LEAF inecarbonatesandcanrecordgloballong-andshort-term perturbationsofthecarboncycle,suchastheMiddle–early Thed13CofC3plants(d13C )wascalculatedaccording P LateTriassic3‰positived13Clong-termtrend(DalCorso tothemodelofFarquharetal.(1989)[Eq.(1)]anddepends et al., 2011), and the Jurassic and Cretaceous positive and onthed13C (d13Coftheatmosphere)andontheratio ATM negative shifts associated with oceanic anoxic events between the pCO inside the leaves and the atmospheric 2 (OAEs;e.g. Gro¨cke, 2002;Hesselbo etal., 2007). pCO (c/c ): 2 i a (fosAsisltarebeiroecshinem)iiscaelxppercotdeducttoorfectoerrdresthtreiaslampleandt1s3,Casmhbifetsr d13CP¼d13CATM(cid:2)a(cid:2)ðb(cid:2)aÞ(cid:3)ci=ca ð1Þ recorded by other plant compounds and tissues. Amber is where a is the fractionation during diffusion of the CO 2 an extraordinary medium for the preservation of animals, from the atmosphere into leaves and is fixed at 4.4‰; b is plantsandfungithatareotherwiserareinthefossilrecord. the fractionation during ribulose-1,5-bisphosphate car- It is resistant to diagenesis and can maintain its original boxylase/oxygenase(RuBisCO)carboxylationandhasval- chemical and isotopic composition and for this reason is ues of 26–30‰ (e.g. Farquhar et al., 1989; Arens et al., thought to be a very powerful tool for reconstruction of 2000; Schubert and Jahren, 2012); and c/c is the ratio i a the palaeoatmosphere and the palaeoenvironment betweenintercellularandatmosphericpCO .Thec/c usu- 2 i a (Murray et al., 1998; McKellar et al., 2011; Dal Corso ally varies between 0.65 and 0.8 with a maximum range et al., 2011, 2013; Aquilina et al., 2013; Tappert et al., between 0.3 and 0.9 (e.g. Farquhar et al., 1989; Arens 2013).Ithasbeenshownthatamberd13C(d13C )falls et al.,2000). AMBER in the range of typical modern C3 plants and may reveal Post-photosyntheticfractionationalsooccursduringthe information about climate and environment at the time of biosynthesisofplantcompounds,whichconsequentlyhave J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 353 different d13C signatures (Badeck et al., 2005). In general, Modern Resins non-photosynthetic tissues are more 13C-enriched by 1– 3‰ than photosynthetic tissues such as leaves (Cernusak et al.,2009). Studies suggested that the d13C is primarily controlled North P Atlantic byd13C (Arens etal.,2000;Jahrenetal.,2008).How- Pacific Ocean ATM Ocean ever, additional strong dependence of D13C upon other P factors complicates the d13CP – d13CATM relationship (e.g. Indian Nordt et al., 2016). The c/c ratio in [Eq. (1)] is regulated South Ocean i a Atlantic bystomatalconductance,whichisgovernedbytheclosing Ocean or opening of the stomata. Stomatal conductance can be influenced by many environmental factors, particularly water availability and pCO . For example, Diefendorf 2 etal.(2010)andKohn(2010)showedthatD13C isstrongly P Cretaceous (Albian - Cenomanian) resins correlated to mean annual precipitation (MAP) and plant functional types. Both these studies found an increase of D13C with increase of MAP. This dependence was mod- P elled and tested in the fossil record (Diefendorf et al., 2015; Kohn, 2016). Schubert and Jahren (2012) gave evi- North dence that D13C by C3 plants grown in environmentally OPacceiafinc Atlantic P controlled chambers hyperbolically increases with increas- Tethys Ocean ing ambient pCO levels. The model of Schubert and 2 Jahren (2012) was tested against ice-core records and used to reconstruct pCO during the Palaeocene–Eocene Ther- South 2 Atlantic mal Maximum (PETM) (Schubert and Jahren, 2013, 2015). In contrast to this model, recent studies found no or negligible pCO dependence over long time scales Fig. 1. Location of the sampling sites of modern resins (A) and 2 Cretaceous amber (B) analysed for carbon isotopes. Red (Kohn, 2016), keeping open the question as to whether MAP or pCO predominantly control D13C . Moreover, dots=this study; orange dots=previous studies (Nissenbaum 2 P andYakir,1995;DalCorsoetal.,2013;Tappertetal.,2013).(For Berneretal.(2000)andBeerlingetal.(2002)experimentally interpretationofthereferencestocolourinthisfigurelegend,the demonstrated an increase of D13C with increase in pO P 2 readerisreferredtothewebversionofthisarticle.) levels. Subsequently, Tappert et al. (2013) proposed the useoffossilplantd13Ctoreconstructpalaeo-pO ,assuming 2 that,inambientair,D13C isproportionaltopO andthat in Canada (Fig. 1). Their origin and ages are summarized P 2 physiological adaptations did not occur through time. inSupplementaryTable2andSupplementaryFig.1.Span- Although thereis noagreement as to which factor ismost ish amber samples were collected by X. Delclo`s and are important in determining the carbon-isotope composition stored at the University of Barcelona, Spain. They derive of modern plants, all these factors must be taken into fromseveralAptian–MaastrichtiandepositsfromtheCen- accounttocorrectlyinterpretthecarbon-isotopeshiftsreg- tral Asturian Depression, the West and East areas of the istered by fossil plant material in the geological record Basque-Cantabrian Basin, the Maestrazgo (=Maestrat) (Diefendorf et al., 2010; Schubert and Jahren, 2012; BasinandtheCastilianPlatform.Theageofthesedeposits Kohn, 2016). is mainly constrained by pollen and spores and includes uncertaintiesfrom(cid:4)1upto(cid:4)18Myrs(seeSupplementary 2.2.Modern andCretaceousresin samples Table 2 and Supplementary Fig. 1; Pen˜alver and Delclo`s, 2010;Barro´netal.,2015).SamplesofFrenchambercome Modernresinsamples fromtheUSAandNewCaledo- from the collection ofD.Ne´raudeaustoredatthe Univer- niawerecollectedbyA.R.SchmidtandL.J.Seyfullahin site´deRennes,France.TheLateAlbian-EarlyCenomanian 2005, 2010 and 2011 (USA) and in 2005 and 2011 (New andSantonianagesofambersamplesfromdifferentlocal- Caledonia). J. Dal Corso, G. Roghi and E. Ragazzi col- ities in France are well constrained by pollen, spores, lected resinsfromdifferentconifers growingattheBotani- dinoflagellates, foraminifers, ostracods and rudists (Peyrot cal Gardens of the University of Padova in 2010. In et al., 2005; Batten et al., 2010), with uncertainties of (cid:4)1 Padova,Araucariaheterophyllaresin,leavesandwoodwere up to (cid:4)3Myrs (see Supplementary Table 2). Grassy Lake alsocollectedatdifferentheightsfromthebaseofthetree. amberwascollectedandprovidedbyA.Wolfe(University Bothliquid-viscousandsolidresinsweresampled.Carbon- of Alberta) and is Campanian in age, according to isotope data, plant species, altitude and geographic prove- McKellar et al. (2008) and Tappert et al. (2013). d13- nance are summarized in Fig. 1 and Supplementary C dataofCretaceousamberwerecoupledwithpre- AMBER Table1. viously published d13C data from Nissenbaum and RESIN The Cretaceous amber analysed for this study derives Yakir (1995), Dal Corso et al. (2013) and Tappert et al. from different deposits in Spain (25 samples), France (10 (2013). The ages of some of these deposits, namely the samples), and from the Grassy Lake deposit (5 samples) LevantineamberfromIsraelandLebanonandtheSanJust 354 J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 amber from Spain, have been revised according to recent AnalyzercoupledtoaSERCONGeo20/20IRMSrunning stratigraphic data. The Lebanese amber-bearing deposits in continuous flow mode with a He carrier gas (flow rate with bioinclusions (entrapped fossilized organisms) are 100ml per min). The reproducibility of the analyses was from the Lower Cretaceous: Ante-Jezzinian (Maksoud estimated using an internal standard (alanine) routinely et al., 2014), i.e., ante Lower Bedoulian (Bedoulian being checked against international standards IAEA-CH-6 and Upper Barremian–Lower Aptian). Deposits of Cretaceous IAEA-CH-7 and traceable back to the VPDB standard. Lebanese amber with bioinclusions are situated in the All resultsare accurateto better than±0.15‰(1r). Chouf Sandstone Formation (= Gre`s de Base or C1 in older usages), under the recently defined Jezzinian 2.4. Meta-analysisof terrestrial andmarinecarbon-isotope Regional-Stage (uppermost Barremian–lower Aptian). data The lower boundary of the Jezzinian is probably within the uppermost Barremian (Maksoud et al.,2014). Accord- The amber data generated in this study have been cou- ing to new biostratigraphical data, the oldest Lebanese pled with the published d13C data from Tappert AMBER amber deposits with bioinclusions are Early Barremian et al. (2013), Nissenbaum and Yakir (1995) and Dal and the youngest are intra-Barremian (Maksoud et al., Corsoetal.(2013),allowingimprovedresolutionofthed13- 2016).TheSanJustamberoutcropislocatedintheMaes- C record. To compare the variability of d13C AMBER AMBER trazgo Basin and is included in the Escucha Fm. of the with other Cretaceous C3 plant material we used the Utrillas Group (SSS – Superior Sedimentary Succession; recently compiled ISOORG database (Nordt et al., 2016). Rodr´ıguez-Lo´pez et al., 2009). It was dated as Middle- ISOORGcomprisesd13Cdata(d13C )ofplantmate- ISOORG Upper Albian by Villanueva-Amadoz et al. (2010), based rialincludingwood,leaf,charcoal,coal,andbulkterrestrial onthepalynologicalfossilrecordanditisnowconstrained organic matter from various geographical locations. We totheUpperAlbianbycomparisonwithsimilardepositsin also built a low-resolution wood d13C (d13C ) record WOOD the Basque–Cantabrian Basin with similar fossil content coupling the Lower Cretaceous wood data extracted from (Barro´net al.,2015). ISOORG (Nordt et al., 2016) with the Maastrichtian data of Salazar-Jaramillo et al. (2016). Before processing data 2.3.d13C analysis we excluded from ISOORG all d13C data; all of RESIN AMBER Cleansub-millimetricfragmentsofthecollectedmodern resinswereseparatedunderthemicroscopetoperformd13C Table1 analysis. Close attention was paid in order to select clear Plant species from which resin was collected, resin mean carbon- resin portions to avoid the presence of microscopic inclu- isotope values for each species (mean±standard deviation (SD) sions.Thed13CanalysiswasperformedonaThermoScien- andsamplesize. tific Delta V Advantage Isotope Ratio Mass Spectrometer Species d13C±SD(‰) n. in continuous flow mode, coupled with a Flash 2000 Ele- Abiesconcolor (cid:2)23.7 1 mental Analyser and a ConFlo IV interface. 0.03–0.05mg Abiesmagnifica (cid:2)23.6±1.1 2 of resin were weighed in a tin capsule and fed to the Ele- Agathislanceolata (cid:2)24.5±0.2 2 mental Analyser. The Mass Spectrometer analysed CO2 Agathismoorei (cid:2)25.9±0.4 2 gas resulted from high temperature combustion. On the Agathisovata (cid:2)25.7 1 basisofalong-termmeanof>30tin-capanalyses,ablank Araucaria(allspecieslistedbelow) (cid:2)27.1±1.4 13 correctionwasappliedtotherawdataandtheresultswere Araucariacolumnaris (cid:2)26.3±1.3 4 calibrated against repeated analyses of IAEA-CH6 and Araucariaexcelsa (cid:2)28.8±0.8 4 IAEA-CH7 international standards, whose d13C is respec- Araucariahumboldtensis (cid:2)26.3±0.7 4 tively (cid:2)10.449‰ and (cid:2)32.151‰ (Coplen et al., 2006). Araucariarulei (cid:2)27.6 1 Cedrusdeodara (cid:2)24.9±0.5 2 The long-term internal reproducibility was estimated on Cupressusarizonica (cid:2)29.8±0.1 2 repeatedanalysesofaninternalstandard(C3plantsucrose) Falcatifoliumtaxoides (cid:2)24.4 1 andis better than0.15‰(1r). Juniperusoccidentalis (cid:2)27.4 1 SamplesofCretaceousamberwerefirstcrushedwithan Piceaabies (cid:2)27.7 1 agatemortartoobtainafinepowder.Repeatedanalyseson Pinus(allspecieslistedbelow) (cid:2)26.9±1.7 55 different portions of single Cretaceous amber specimens Pinusbalfouriana (cid:2)24.3±0.6 2 have shown that the d13C is remarkably homogeneous Pinuscoulteri (cid:2)25.8±0.32 3 withinthesamepiece(DalCorsoetal.,2013).Wethuscon- Pinusedulis (cid:2)24±0.7 3 sider the measured d13C as representative of the entire Pinuselliottii (cid:2)28.3±0.2 3 amber sample. The amber powder was placed in a Pinusjeffreyi (cid:2)28.2±1.1 10 Pinuslambertiana (cid:2)26.5±1 5 polypropylene tube and treated with 3M HCl to remove Pinuslongaeva (cid:2)24.6 1 possible residual carbonates from the sediments where the Pinusmonophylla (cid:2)24.9±0.9 5 amberhadbeenembedded.Sampleswerethenrinsedwith Pinusmonticola (cid:2)28±2 2 deionizedwateruntilneutralitywasreachedandwereoven- Pinusmuricata (cid:2)28.2±0.9 8 driedat50(cid:3)C.1.5–2mgofamberpowderwereweighedin Pinusponderosa (cid:2)27.2±1.3 9 tincapsulesandfedintotheElementalAnalyzer.d13Canal- Pinusradiata (cid:2)26.5±0.8 2 ysiswasperformedusingaCarloErbaNA1108Elemental Pinussabiniana (cid:2)27.2±0.7 2 J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 355 Table2 Standarddeviation(SD),interquartilerange(IQR)andsamplesize(n.)ofd13Cdataofmodernresinandleaf,andCretaceousamber,wood andothermixedC3plantmaterial(fromISOORG;TOM=bulkterrestrialorganicmatter).FortheCretaceous,SDandIQRhavebeenalso calculatedpereach5Myrsagebinfrom65Mato145Ma(seetextandNordtetal.,2016forexplanations).Binsnotlistedinthetablecontain nodata.ISOORGandwooddataaretakenfromNordtetal.(2016)andSalazar-Jaramilloetal.(2016). Age Type Mean(‰) SD(‰) IQR(‰) n. Modern Resin (cid:2)26.7 1.77 2.7 85 Leaf (cid:2)28.4 2.52 3.6 513 Cretaceous Amber (cid:2)22.3 1.95 2.5 201 C3Plant (cid:2)24.2 1.31 1.4 1384 Charcoal (cid:2)22.9 1.55 2 192 Coal (cid:2)24.2 1.43 1.23 95 Leafandcuticle (cid:2)24.7 1.72 2.14 16 TOM (cid:2)24.4 0.98 1.1 874 Wood (cid:2)23.1 1.31 1.47 207 AGEbin Type SD(‰) IQR(‰) n. 65Ma C3Plant(TOM,leaf,coal) (cid:2)24.5 0.88 1.08 677 70Ma Amber (cid:2)23.7 1.79 2.60 40 C3Plant(TOM,leaf,coal) (cid:2)25.9 1.17 1.60 43 Wood (cid:2)25.2 1,32 1,80 27 75Ma Amber (cid:2)23.5 1.39 1.75 51 C3Plant(TOM) (cid:2)24.7 0.30 0.38 5 85Ma C3Plant(TOM) (cid:2)24 0.31 0.42 14 90Ma Amber (cid:2)22.1 1.10 1.60 36 C3Plant(TOM) (cid:2)24.1 0.51 0.50 9 95Ma C3Plant(TOM) (cid:2)23.6 0.73 1.12 167 100Ma Amber (cid:2)21.1 0.98 1.33 11 105Ma Amber (cid:2)22 1.33 1.85 14 110Ma C3Plant(leaf) (cid:2)24.8 2.03 1.88 8 125Ma Wood (cid:2)22.2 1.15 1.78 19 130Ma Amber (cid:2)20.8 1.20 2.10 25 Wood (cid:2)22.8 0.93 1.29 89 135Ma C3Plant(coal,charcoal) (cid:2)22.9 1.49 1.78 231 Wood (cid:2)23.3 1.04 1.36 35 140Ma C3Plant(coal,wood) (cid:2)23.6 1.56 2.18 87 Wood (cid:2)22.9 0.87 1,05 37 145Ma Wood (cid:2)23.6 0.45 0.69 12 25 Modern C3 plants 20 n si e y r 15 c n e u q Fre 10 Mean resin = -26.7‰ 5 0 -36 -34 -32 -30 -28 -26 -24 -22 -20 -18 δ13C(‰) Fig.2. Distributionofmodernresind13C(orangehistogram)andrangeofvariabilityofmodernC3plantsfromCerlingandHarris(1999) andTippleandPagani(2007).(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionof thisarticle.) 356 J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 which were already included in our compilation. Some of <0.001; Solid resin of Pinus and Araucaria, R=0.634, p the amber deposits described here have age uncertainties value <0.001). Similar correlation was also observed of several millions of years, especially in the case of the between d13C and altitude (R=0.59; Ko¨rner et al., LEAF mid-Cretaceous Spanish ambers (Supplementary Table 2 1988).Thed13C valuesforthe mostrepresented gen- RESIN and Supplementary Fig. 1). These uncertainties depend on era show that Pinus resin is statistically indistinguishable thefactthatCretaceousamberiscommonlyfoundincon- from Araucaria resin (p value=0.6, Students’ t-test; tinental(fluvialsediments,coaldeposits)orcoastal(brack- Table 1). ish estuarine/lagoonal) deposits that lack age-significant fossils. For this reason and to allow comparison with the 3.2. Cretaceousamber d13C ISOORG database, d13C data were placed into AMBER 5Myrs-age bins following the criteria used by Nordt The d13C of amber from Spain varies between (cid:2)17‰ et al. (2016). Amber with age uncertainty larger than the and(cid:2)24.2‰(mean=(cid:2)20.1±1.8‰)intherangeexpected bin was excluded. The same procedure was used also for forC3plantresins(Table2).Similarly,d13Cofamberfrom d13C from Salazar-Jaramillo et al. (2016). Box-and- France and Canada ranges from (cid:2)18.5‰ to (cid:2)23.5‰ WOOD whiskers plots for d13C , d13C and d13C (mean=(cid:2)21.1±1.9‰) and from (cid:2)21.4‰ to (cid:2)23.4‰ AMBER WOOD ISOORG data were built for each age bin with a sample size of at (mean=(cid:2)22.7±0.8‰), respectively (Table 2). The com- least5(KrzywinskiandAltman,2014).Tocomparetheter- piled Cretaceous amber and ISOORG d13C values show a restriald13Csignaltothemarined13Csignalwetookmar- normal distribution (Fig. 5A). On average, amber is more inecarbonatedatafromthedatabasecompiledbyProkoph 13C-enriched (mean=(cid:2)22.3‰±1.9‰) than Cretaceous etal.(2008)andBodinetal.(2015).Weusedd13Candd18O C3 plant material (mean=(cid:2)24.2‰±1.3‰) and wood datafrombenthicandplanktonicforaminiferaandbelem- (mean=(cid:2)23.1‰±1.3‰) (Fig. 5B). Cretaceous nites. Terrestrial d13C data were also compared to the d13- d13C data are more dispersed than d13C values of AMBER C , which was estimated from the d13C and d18O of C3 plant material: the box-and-whisker plots (Fig. 5B) ATM benthic foraminifera using the equations proposed by show the interquartile range (IQR) of d13C to be AMBER Tipple et al. (2010). A third-degree polynomial curve was much larger (2.5‰, Table 2) than d13C (1.4‰). ISOORG fittedtothedatatocomparetheterrestrialandmarinecar- F-test for the equality of variances indicates that the vari- bonate d13C records. Prediction intervals for individual ancesofd13C andd13C aresignificantlydiffer- AMBER ISOORG observationsholdabout95%ofdata.Polynomialcurvefit- ent(p<0.0001).SDandIQRwerecalculatedforeachage ting by the least-squares method and prediction intervals bin (Table 2) and show that d13C is generally more AMBER wereobtainedwithJMPsoftware,version10(SASInstitute dispersed then the d13C and d13C . Amber, ISOORG WOOD Inc., Cary,NC,USA). wood and ISOORG data show that latest Cretaceous (Maastrichtian) d13C values are more 13C-depleted than 3. RESULTS those of the Early Cretaceous (Hauterivian–Barremian) by 2.5–3‰ (Fig. 6). The marine d13C record from whole 3.1.Modern resind13C rock, belemnite, and foraminifera (Prokoph et al., 2008; Bodin et al., 2015) shows a pattern that only partially The d13C of all the analysed modern resins varies from matchestheterrestrialrecords(Fig.6).Thed13C calcu- ATM (cid:2)31.6‰ to (cid:2)22.8‰ (mean±SD=(cid:2)26.7±1.8‰, n=84; lated from benthic foraminifera shows a decrease of Table 1). d13C values obtained in this study show a approx.1‰fromtheAptiantotheMaastrichtianthatmir- RESIN normal distribution with a mean of (cid:2)26.7‰ (Fig. 2). The rors the decrease (2.5–3‰) shown by terrestrial plants meand13C ismore13C-enrichedthanthemeanglobal (Fig. 7). RESIN leaf d13C ((cid:2)28.5‰ calculated from data of Diefendorf et al., 2010; Fig. 3A). A statistically significant difference 4. DISCUSSION (p value=0.001) exists between liquid–viscous resin and solid resin, the former having more 13C-enriched values 4.1. Modernresin (mean (cid:2)25.9‰) than the latter (mean (cid:2)27.1‰) (Fig. 3A). Resin has d13C values systematically more 13C-enriched 4.1.1.Effect ofresin hardeningonthe d13C RESIN by 1–2.3‰ than those of bulk leaf and wood samples col- Toobtainreliableinformationfromthecarbon-isotope lected from the same branch at the same tree height in geochemistry of resin on the physiology of plants and the Araucariaheterophylla,PiceaabiesandCupressusarizonica environmentalconditionsunderwhichtheygrow,itisnec- (Fig. 3B). Resin, wood and leaves collected from a single essarytounderstandwhetherthemeasuredd13C val- RESIN treeofAraucariaheterophyllaatdifferentheightsalsopos- ues actually represent the pristine composition at the time sessvariabled13Csignatures(Fig.3C).Differencesofupto of resin biosynthesis. After exudation, resin is composed 6‰ exist between the mean d13C of resins from different of up to 50% of a volatile fraction (mainly monoterpenes plant species (Table 1). The d13C from different tree and sesquiterpenes). The volatile fraction is lost rapidly RESIN genera growing at the same altitude in the same locality on exposure of resin to air and sunlight, whereas the non- (Padova,Italy)differsbyabout2–5‰(Fig.3B).Liquid–vis- volatile fraction (mainly diterpene acids in conifer resin) cous and solid resin d13C of Pinus and Araucaria signifi- undergoespolymerization(cross-linkingandisomerization) cantly increases with increasing altitude of the sampling with the formation of high-molecular-weight polymers site (Fig. 4; liquid–viscous resin, R=0.632, p value (Langenheim, 1990; Scalarone et al., 2003; Lambert et al., J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 357 A -35 -34 -33 -32 -31 -30 -29 -28 -27 -26 -25 -24 -23 -22 -21 al.) et AForf Ed Ln e ef Di ( Solid (n=57) y) Nd RESIhis stu Loss of volatiles (t Liquid–Viscous (n=28) -35 -34 -33 -32 -31 -30 -29 -28 -27 -26 -25 -24 -23 -22 -21 δ13C (‰) B C 8 s) etre 7 m Leaf e ( 6 e Wood e tr Resin of th 5 e as 4 b -33 -32 -31 -30 -29 -28 -27 -26 -25 -24 he m t 3 δ13C (‰) o ht fr 2 g ei H 1 Araucaria heterophylla Araucaria heterophylla Picea abies resin wood leaf A. heterophylla Cupressus arizonica Cedrus deodara 0 -31 -30 -29 -28 δ13C (‰) Fig.3. (A)Carbon-isotopecompositionofliquid–viscousvssolidmodernresin(Students’t-testpvalue=0.001).Resindataarecomparedto acompilationofleafdatatakenfromDiefendorfetal.(2010).Dataarerepresentedasbox-and-whiskersplotsinordertohighlightdifferences in distribution. The bars represent the first and fourth quartile, the box represents the second and third quartile, and the mid-line is the median.Allinvestigatedspeciesareconsidered.Variationofthecarbon-isotopecomposition(B)ofsolidresin,woodandleavesfromdifferent trees(twotreesofAraucariaheterophylla,aPiceaabiestreeandaCupressusarizonicatree)and(C)ofsolidresincollectedfromasingletreeof Araucariaheterophyllaatdifferentheights.Samplesinboxes(B)and(C)werecollectedintheBotanicalGardenoftheUniversityofPadova. 2008;RagazziandSchmidt,2011).Thisselectiveremovalof comprises both liquid–viscous resins sampled shortly after moietiespointstoapossiblechangeofthebulkd13C exudation and solid resins that already had hardened at RESIN duringresinhardening(DalCorsoetal.,2011).Ourdataset the site of exudation. A statistically significant difference 358 J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 0 500 1000 1500 2000 2500 3000 3500 -21 y=-28.295+0.0012243x R=0.632 A -23 -25 ) ‰ ( C 3 1 δ -27 -29 Liquid–viscous resin -31 -21 y=-28.14+0.0010493x R=0.634 B -23 -25 ) ‰ ( C 3 1 δ -27 -29 Pinus resin Araucaria resin -31 0 500 1000 1500 2000 2500 3000 3500 Altitude (m) Fig.4. Carbon-isotopedatafrommodernPinusandAraucariaresinplottedagainstaltitudeoftheplant-growingsite.(A)liquid–viscousresin samples,and(B)allresinsamples. (p=0.001) is observed between liquid–viscous probably allow correction of the measured d13C of solid (mean=(cid:2)25.9‰) and solid resin (mean = (cid:2)27.1‰), with resinbacktothepristinesignatureatthetimeofexudation an overall 1.2‰ 13C-enriched values in the former in order tofaithfully interpret the data. (Fig.3A).Weconcludethatvolatilemono-andsesquiterpe- nes released by resin during hardening are more 4.1.2.Differences betweenthe d13C andthe d13Cof RESIN 13C-enriched than the non-volatile diterpenoid and other plant material triterpenoid acids. Consequently, changes in the pristine Post-photosynthetic fractionation in plants results in d13C occursoonafterresinexudation.Futureorganic differences in the d13C of plant tissues (Badeck et al., RESIN geochemical studies should precisely determine the magni- 2005). In general, non-photosynthetic tissue tends to be tude of these isotopic changes in different resin types by more 13C-enriched than photosynthetic tissue: leaves were studyingthe pattern of volatile lossduring hardening, and found to have isotopically lighter values than wood and thespecificcarbon-isotopesignatureandtherelativeabun- roots, and above-ground organs are more 13C-depleted danceofthedifferentresincompounds.Suchastudywould than below-ground material (Badeck et al., 2005; J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 359 Cernusak et al., 2009). Several biochemical causes have and are still a topic of debate (review by Cernusak et al., beeninvokedtoexplainthiswidespreadisotopicbehaviour 2009). D13C duringresinbiosynthesisisevidentfromourdata P whencomparingthed13C withthed13Cofotherplant RESIN A material. Our dataset shows that the mean d13CRESIN of -29 -27 -25 -23 -21 -19 -17 -15 fresh liquid–viscous resin ((cid:2)25.9‰) is more 13C-enriched 800 by2.6‰thanthemeand13C fromapublishedcompi- Cretaceous C3 plant LEAF 700 lation of data of C3 leaves ((cid:2)28.5‰; Fig. 3A, Diefendorf et al., 2010), as expected from a non-photosynthetic plant nts 600 compounds. Solid resin ((cid:2)27.1‰) is more 13C-depleted oi than liquid–viscous resin, but remains more 13C-enriched p 500 a than mean leaf d13C, so that carbon-isotope changes due at d 400 to hardening do not overshadow post-photosynthetic of D13C betweenresinandleaf.Thisdifferenceisalsoevident ber 300 fromPsamples taken from the same trees and branches. m nu 200 Solid resin of Araucaria heterophylla, Picea abies and Cupressus arizonica sampled in the Botanical Garden in 100 Padova (Italy) has higher d13C values than leaves from the same branch by approx. 1–2‰ (Fig. 3B and C). This 0 difference should be corrected for the loss of volatiles and 50 was likely larger by 1–2‰ at the time of resin exudation Cretaceous Amber (see Section 4.1.1). Similar differences exist also between resin and wood 40 d13Csignatures.InPiceaabiesandAraucariaheterophylla, s nt resinismore13C-enrichedthanwood,which,inturn,shows oi a p 30 verysmalld13Cdifferencescomparedtoleafcarbon-isotope at signatures(Fig.3BandC).Asshownbythetreessampled d of for this study, fractionation during resin biosynthesis does mber 20 toucrceur(bayndaprepsruolxts. i2n–4a‰v)erwyh1e3nC-ecnormicphaerdedd1t3oCRthEeSINd1s3iCgnoa-f u n otherorgansfromthesameplantbranch.Onthecontrary, 10 d13C and d13C show little difference (<1‰) WOOD LEAF withinthesamebranch(Fig.3BandC).Otherstudiesshow 0 thatonaveragestemwoodandrootsaremore13C-enriched -30 -28 -26 -24 -22 -20 -18 -16 by 1–1.9‰ than is leaf material (Badeck et al., 2005). Our δ13C (‰) resultssuggestthatthepost-photosyntheticD13CPislarger forresinthanforotherbulkplanttissues.Suchpatternsare visiblealsoafterresinhardening,meaningthatbothvolatile B -36 -34 -32 -30 -28 -26 -24 -22 -20 -18 -16 (monoterpenesandsesquiterpenes),andnon-volatile(diter- pene acids) are affected. 4.1.3.Environmental andphysiological effectson d13C N Leaf RESIN R Our results show that the carbon-isotope signature of E resin records the environmental and physiological effects D O of C3 plant 13C discrimination, as does other plant M Resin material. The d13C varies by up to 2‰ within the RESIN same tree (Fig. 3C) and on average d13C differs by RESIN up to 6‰ between plant species and genera, and between ISOORG different localities (Table 1 and Supplementary Table 1). S This high variability in the resin carbon-isotope composi- U O E C Wood 3 A T E R Fig.5. (A)Comparisonofthed13CdatadistributionofCretaceous C Amber C3plantmaterialfromtheISOORGdatabase(Nordtetal.,2016) with amber. (B) Box-and-whiskers plots of d13C data of modern leaves (Diefendorf et al., 2010) and resin (this study), and Cretaceous ISOORG plants (Nordt et al., 2016), wood (Nordt -36 -34 -32 -30 -28 -26 -24 -22 -20 -18 -16 etal.,2016;Salazar-Jaramilloetal.,2016)andamber(thisstudy; NissenbaumandYakir,1995;DalCorsoetal.,2013;Tappertetal., δ13C (‰) 2013). 360 J.DalCorsoetal./GeochimicaetCosmochimicaActa199(2017)351–369 150 145 140 135 130 125 120 115 110 105 100 95 90 85 80 75 70 65 60 -18 (A) AMBER -19 -20 -21 RG -22 CO -23 13 -24 δ -25 -26 -27 -28 -20 (B) WOOD -21 -22 G R -23 O C -24 3 δ1 -25 -26 -27 -18 (C) ISOORG -19 -20 -21 RG -22 CO -23 13 -24 δ -25 -26 -27 -28 (D) Planktonic foram 5 Benthic foram Belemnite 4 3 B AR 2 C C 1 3 1 δ 0 -1 -2 -3 RASSIC Berriasian Valanginian Hauterivian Barremian Aptian Albian Cenomanian Turonian ConiacianSantonian Campanian Maastrichtian EOGENE U L J Early Late A P CRETACEOUS 150 145 140 135 130 125 120 115 110 105 100 95 90 85 80 75 70 65 60 Age (Ma) Fig.6. Carbon-isotope(d13C)curvesfromCretaceousamber,terrestrialorganicmatter,andmarinecarbonate.d13Cdatafromplantmaterial weregroupedin5MyrsbinsfollowingthemethodusedbyNordtetal.(2016)andathird-degreepolynomialcurvewasfittotheplantdatato highlightthemaintrendsshownthroughtheCretaceous(seetextforfurtherexplanation).(A)CompilationofCretaceousambercarbon- isotopedatafromthisstudyandNissenbaumandYakir(1995),DalCorsoetal.(2013)andTappertetal.(2013).(B)Woodd13Cdatafrom Nordtetal.(2016)andSalazar-Jaramilloetal.(2016).(C)ISOORGd13CdatafromNordtetal.(2016).ISOORGdatabasecomprisesisotopic data from wood, leaf, charcoal, coal, palaeosols, and bulk terrestrial organic matter. (D) Marine carbonate carbon-isotope data from planktonicandbenthicforaminifera,andbelemnites(Prokophetal.,2008andBodinetal.,2015).Whole-rockgeneralcarbonatecurve(black line)replottedfromErba(2004).Thearrowsrepresentthemaincarbonated13CtrendduringtheCretaceous.TimescaleafterGradsteinetal. (2012).