MATERIAL AND METHODS

In order to choose the best treatment adapted to different conditions, several chemical protocols were performed on six samples of known age:

1. a) IAEA-C3 is the cellulose standard, in which the certified F¹⁴C is 1.2941±0.0006 and the δ¹³C is –24.9‰ (Rozanski et al. Reference Rozanski, Stichler, Gonfiantini, Scott, Beukens, Kromer and van der Plicht1992).

2. b) Quercus-1980 is the wood of an oak tree (Quercus pubescens Willd.) from Aix-en-Provence (France), which was cut in winter 2014–2015. Only the latewood of the ring corresponding to the year 1980 was sampled, in order to isolate the C assimilated in the corresponding growth year.

3. c) Larix-1980 is the wood of a larch tree (Larix decidua Mill.) from the Alps of Haute Provence (France). The last ring before the bark is dated to 1994, and only the latewood of the ring corresponding to the year 1980 was isolated.

4. d) COUT203 is the wood of a subfossil pine tree (Pinus sylvestris L.) from the Alps of Haute Provence (France). It has a sequence of 170 yr included in a floating chronology (Sivan et al. Reference Sivan, Miramont and Édouard2006; Miramont and Sivan Reference Miramont and Sivan2008). The group of rings 150–160 from COUT203 tree, previously dated at the Poznan AMS laboratory to 7260±40 BP after the ABA pretreatment, was analyzed.

5. e) Plio is the wood of a fossil tree found in a canyon in Nice (France), corresponding to the Messinian period of the Pliocene (personal communication with Jean Claude Hippolyte). This sample is thus expected to be devoid of ¹⁴C.

6. f) VIRI-K is the wood of a Miocene tree, distributed as a blank in the framework of the Fifth International Radiocarbon Intercomparison (Scott et al. Reference Scott, Cook and Naysmith2010).

The two modern wood samples (b, c) were chosen for the high precision dating comparison associated with the atmospheric ¹⁴C bomb spike, assuming that the latewood grew between July and October 1980 (Hua et al. Reference Hua, Barbetti and Rakowski2013). In addition, the chosen samples enable testing different species of trees, both broadleaf and coniferous, with greater difficulty expected for the delignification of the coniferous wood (Cullen and MacFarlane Reference Cullen and MacFarlane2005). The two wood blanks were selected to quantify the modern carbon contamination. The state of preservation of Plio is not optimal as it shows charcoal consistency, probably due to advanced mineralization process, and it will thus be useful for selecting procedures able to pretreat fragile samples. Finally, COUT203 was chosen because it belongs to the same species and it comes from the same region as the subfossil wood used for our ¹⁴C calibration project (Section II below).

Samples were sliced in small pieces (ca. 2 mm³) and ca. 100 mg were isolated for most of them, with the exception of Plio, whose deteriorated wood required ca. 500 mg of material.

Based on previous experience (Capano et al. Reference Capano, Marzaioli, Sirignano, Altieri, Lubritto, D’Onofrio and Terrasi2010, Reference Capano, Marzaioli, Passariello, Pignatelli, Martinelli, Gigli, Gennarelli, De Cesare and Terrasi2012, Reference Capano, Altieri, Marzaioli, Sirignano, Pignatelli, Martinelli, Passariello, Sabbarese, Ricci, Gigli and Terrasi2013) and review of the literature, we have used the following chemical protocols to process these samples. Figure 1 provides a visual summary of these procedures:

1. a) ABA (acid-base-acid) method (de Vries and Barendsen Reference de Vries and Barendsen1954), with a modified procedure using HCl, NaOH, HCl at 3% concentration for 1 hr each (Passariello et al. Reference Passariello, Marzaioli, Lubritto, Rubino, D’Onofrio, De Cesare, Borriello, Casa, Palmieri, Rogalla, Sabbarese and Terrasi2007).

2. b) BABA (base-acid-base-acid), an ABA method with 4% reagent solutions, preceded by an overnight bath in base solution (Kromer et al. Reference Kromer, Manning, Friedrich, Talamo and Trano2010). It was performed for comparison with the standard ABA method, because high pH (i.e. first long alkaline bath) helps to dissociate the main wood components (Němec et al. Reference Němec, Wacker, Hajdas and Gäggeler2010).

3. c) BABAB (base-acid-base-acid-bleaching) method with solutions at 4% (Němec et al. Reference Němec, Wacker, Hajdas and Gäggeler2010). This method entails an overnight alkaline bath for disrupting wood structure before the following treatment (Němec et al. Reference Němec, Wacker, Hajdas and Gäggeler2010) and a bleaching step for holocellulose isolation.

4. d) BABAB-mod is a BABAB modified method from Němec et al (Reference Němec, Wacker, Hajdas and Gäggeler2010). The difference with respect to the original pretreatment is the reduced duration for the initial bath in base solution to 2 hr, in order to reduce the overall duration of the process.

5. e) Holo-cell is a more classical holocellulose extraction procedure modified from Capano et al. (Reference Capano, Marzaioli, Sirignano, Altieri, Lubritto, D’Onofrio and Terrasi2010). The treatment consists of the following: I) an HCl bath at 4% for 1 hr; II) bleaching (60 g of NaClO₂ in 1 L ultrapure H₂O in acid solution (HCl) at pH 3, repeated twice for 1.5 hr each); III) a NaOH bath at 4% for 1 hr, used to remove lignin residue and a portion of the hemicellulose residues (a step which is missing in the BABAB method); and IV) a final HCl bath at 4% concentration.

6. f) α-cellulose extraction procedure modified from Cullen and MacFarlane (Reference Cullen and MacFarlane2005). In order to reduce procedural time, the soxhlet was substituted with several cycles of ultrasonic baths in organic solvents: five 20-min baths in ethanol-toluene (1:2), five 20-min baths in ethanol, and five 20-min baths in ultrapure water. The following day, the cellulose extraction was performed as in the procedure described above [procedure e)], but with NaOH at 17%, which is the minimum concentration for the isolation of α-cellulose, according to Cross and Bevan (Reference Cross and Bevan1912). This α-cellulose pretreatment lasts for two days and it is performed to test whether a better blank could be achieved.

7. g) BABBA1 (base-acid-bleaching-base-acid) is the combining of the two holocellulose extraction methods (procedures d/ e/): the holo-cell treatment is preceded by a bath in NaOH, reduced to 1 hr. Bleaching is performed in two baths of 1 hr each; while the last NaOH and HCl baths are reduced to 30 min. The reaction time is reduced because the first alkaline bath should already disrupt wood structures and facilitate the cellulose extraction (Němec et al. Reference Němec, Wacker, Hajdas and Gäggeler2010).

8. h) BABBA2 (base-acid-bleaching-base-acid) is the same as BABBA1 (described above), but with 1-hr durations for the last NaOH and HCl baths.

9. i) ABA-B is a classical ABA treatment followed by bleaching. Each bath in HCl, NaOH and HCl 4% solutions lasts for 1 hr. Bleaching was performed with 60 g of NaClO₂ in 1 L of ultrapure water in acid solution (HCl) at pH 3. The bleaching step was performed once for 2 hr and was repeated twice for 1 hr, in cases of difficult samples. This treatment omits the post-bleaching bath in base solution and, for this reason, it leads to a less purified holocellulose fraction than do previously tested methods [procedures e), g), h)].

Figure 1 Schematic description of chemical pretreatments for dating wood; rep=repeatable step.

All chemical treatments were performed in glass tubes on a heated block at 70°C, equipped with an agitation system. Solutions of at least 10 mL were changed using plastic pipettes and neutralized with ultrapure water. Finally, samples were oven dried overnight at 70°C.

Several studies extract the resins with a soxhlet apparatus also in holocellulose extraction procedures. However, some other studies omit this soxhlet step (e.g. Němec et al. Reference Němec, Wacker, Hajdas and Gäggeler2010; Southon and Magana Reference Southon and Magana2010) and Rinne et al. (Reference Rinne, Boettger, Loader, Robertson, Switsur and Waterhouse2005) demonstrated that NaOH is sufficient to remove resins from Pinus sylvestris. Consequently, we also omitted the soxhlet step for the holocellulose extraction because our purpose is to select a simple and time-effective procedure.

Each chemical procedure was repeated at least three times on blank samples (Plio and VIRI-K), in order to quantify the reproducibility of blanks. Indeed, this is the parameter that matters most in the propagation of errors for unknown samples. In order to evaluate the ¹⁴C contamination level before any pretreatments, all samples were additionally analyzed without pretreatment.

We measured the mass yield by weighing the initial dried sample and residue after pretreatment, as well as the C percentage by means of the elemental analyzer (EA, VarioMicroCube Elementar). The BABAB method was performed only to compare its mass yield with that of its modified version (BABAB-mod).

After chemical pretreatment, the dried residual samples were weighed in a tin capsule, combusted by using the EA, and the CO₂ was finally transformed into graphite with the AGE III system. The graphite target was then analyzed for its ¹⁴C/¹²C and ¹³C/¹²C ratios by means of AixMICADAS (Bard et al. Reference Bard, Tuna, Fagault, Bonvalot, Wacker, Fahrni and Synal2015).

Standards (OxA2 NIST SRM4990C) and blanks (phthalic acid) were processed together with samples and used for normalization and blank correction. Blank samples (Plio and VIRI-K) were not corrected for background. Two measurements for each chemical pretreatment were performed for all samples (three replicates for blanks).

An additional uncertainty of 1.6‰ was propagated in the ¹⁴C analytical errors. This additional error is based on the long-term laboratory statistics of measurements of the IAEA-C3 standard, processed with the chosen chemical pretreatment (ABA-B). The ¹⁴C data are reported in terms of conventional ¹⁴C age BP and in terms of F¹⁴C, which is the ¹⁴C/¹²C isotope ratio after normalization and blank correction (Stuiver and Polach Reference Stuiver and Polach1977; Reimer et al. Reference Reimer, Brown and Reimer2004; van der Plicht and Hogg Reference van der Plicht and Hogg2006). The ¹³C/¹²C ratios are reported using the conventional δ notation by using OxA2 NIST standard. All reported values for C% and ¹³C/¹²C ratios are the average of at least two replicates and listed errors are the standard deviations (SD) among replicates.

Results on Mass Yield, C%, and ¹³C/¹²C Ratio

The mass yield after the different pretreatments varies extensively because it depends on the efficiency of the isolation of the wood fraction and the loss of unwanted material (Table 1). The wood mass yield probably also depends on the type of tree, its age and state of preservation. Additional sources of variability are found in the incidental physical loss of the sample at each liquid-solid separation performed by manual pipetting during the pretreatments. As suggested by three replications of blank samples this variability could be on the order of several percent (from 1 to 7%; Table 1).

Table 1 The yield mass (%) of all samples is given after each pretreatment method. Plio and VIRI_K values are followed by the standard deviation of replications of the same pretreatment.

As expected, the least aggressive procedure (ABA) leads to the highest mass recovery for the wood residue, whereas BABA leads to a more thorough extraction and, hence, to lower residual mass. For most samples, the α-cellulose extraction is the most aggressive pretreatment, leading to a smaller wood residue after extraction. In the case of the Plio blank wood, the sample disappeared completely in two α-cellulose extractions, while a very small residue amenable to ¹⁴C dating was recovered in the third replicate. Because of this extremely poor incidence of recovery leading to sub-optimal ¹⁴C results (see below), the α-cellulose treatment is clearly not optimal for our purposes.

An initial bath in base solution may already destroy part of the cellulose, as suggested by the results obtained on the C3 standard made of pure cellulose. Indeed, the mass yield for C3 is lower for the pretreatment procedures, which incorporate this initial step (BABAB-mod and BABBA1), than for those which omit it (Holo-cell and ABA-B).

In general, the ABA-B procedure is among the most thorough protocols leading to rather small residues for the various samples (Table 1). At the same time, this relatively simple method allows preservation of a significant mass of the old and damaged Plio blank wood, enabling its subsequent ¹⁴C dating (see below).

When the BABAB method is compared with its modified version (BABAB-mod), two samples (COUT203 and Plio) show the same % yield, while Larix-1980 and VIRI-K exhibit higher yields after the BABAB-mod. Although still compatible with random scatter, this may indicate that for well-preserved wood, the length of the initial bath in base solution results in the removal of more material as reaction time increases. By contrast, the BABBA1 and BABBA2 procedures lead to similar mass yields (Table 3), indicating that lengthening the final baths in basic and acidic solutions does not improve the extraction.

The molecular percentage of carbon in cellulose is 44.4% (Vertregt and de Vries Reference Vertregt and de Vries1987), but it frequently ranges from 43 to 45% (Cullen and MacFarlane Reference Cullen and MacFarlane2005). This is compatible with the value measured on C3 samples with or without pretreatments (43 to 44 %). All untreated woods show high values (ranging between 47 and 53%), indicating the presence of other compounds such as extractives and lignin that have higher C %. However, wood treated with ABA and BABA procedures show higher C concentration, which is probably linked to removal of the ash that is devoid of C (Table 2). It is noteworthy that the %C results after all pretreatments incorporating a bleaching step are compatible with the cellulose range (43–45%).

Table 2 Percentages of C of all samples are given for each pretreatment method. All values are followed by standard deviation of same sample/treatment replications.

Several authors have shown that different wood fractions are characterized by different δ¹³C ‰, and that the cellulose fraction shows less negative δ¹³C values with respect to bulk wood (Borella et al. Reference Borella, Leuenberger, Saurer and Siegwolf1998; Hoper et al. Reference Hoper, McCormac, Hogg, Higham and Head1998; Cullen and MacFarlane Reference Cullen and MacFarlane2005; Cullen and Grierson Reference Cullen and Grierson2006; Harlow et al. Reference Harlow, Marshall and Robinson2006). Even if the δ¹³C measurements by AMS have a rather low precision (on the order of 1 ‰ for AixMICADAS; Bard et al. Reference Bard, Tuna, Fagault, Bonvalot, Wacker, Fahrni and Synal2015) the results obtained on untreated and pretreated wood clearly confirm this trend (Table 3). Indeed, all procedures incorporating a bleaching step for cellulose isolation, lead to a systematic δ¹³C increase by up to several ‰ (the only exception is for the Quercus-1980 which, after α-cellulose extraction, remains compatible within the large δ¹³C errors, with the untreated wood).

Table 3 The δ¹³C ‰ of all samples is given for each pretreatment method. All values are followed by standard deviation of same sample/treatment replications.

¹⁴C results on Standards and Known-Age Woods

Figures 2–4 illustrate the results of standard C3, Quercus-1980, and Larix-1980 modern samples. Because of the expected difficulty of cellulose extraction in coniferous wood as opposed to broadleaf wood (Cullen and MacFarlane Reference Cullen and MacFarlane2005), pretreatments were tested extensively on Larix-1980 with all variants of our chemical procedures (Figure 4).

Figure 2 F¹⁴C analyses of the IAEA-C3 standard pretreated with different methods (error bars are ±1σ, including counting statistics and sample preparation uncertainties), compared with certified standard value (between dashed lines). Five ¹⁴C measurements, from five samples that underwent different combustions/graphitizations, were performed for each pretreatment, with the exception of ABA-B, measured 22 times (22 different samples) from five replications of the same chemical pretreatment. For each treatment, the average value followed by the SD is indicated. For the ABA-B pretreatment, the weighted mean with its ±1σ weighted error is reported in the small dot close to the series of individual F¹⁴C values.

Figure 3 F¹⁴C content of the latewood of Quercus-1980 (error bars are ±1σ, including counting statistics and sample preparation uncertainties), compared with clean air F¹⁴C values (1σ) of the corresponding period (between solid black lines; Hua et al. Reference Hua, Barbetti and Rakowski2013). Two ¹⁴C measurements, from two samples that underwent different combustions/graphitizations, were performed for each pretreatment. The weighted average of all values obtained from pretreatments is the red solid line, while the corresponding ±1σ weighted error is shown by dashed red lines. (Color refers to online version.)

Figure 4 F¹⁴C content of the latewood of Larix-1980 (error bars are ±1σ, including counting statistics and sample preparation uncertainties), compared with clean air F¹⁴C values (±1σ) of the corresponding period (between solid black lines; Hua et al. Reference Hua, Barbetti and Rakowski2013). Two ¹⁴C measurements, from two samples that underwent different combustions/graphitizations, were performed for each pretreatment. The weighted average of all values obtained from pretreatments is the red solid line, while the corresponding ±1σ weighted error is shown by dashed red lines. (Color refers to online version.)

The results of C3 and Larix-1980 show no difference between samples of the same material analyzed with or without chemical treatment, indicating the absence of contamination in the original materials (Figures 2 and 4). By contrast, the Quercus-1980, measured without chemical pretreatment, exhibits F¹⁴C results clearly lower than those of all pretreated samples, indicating a younger contamination. This could be due either to an unusual carbon translocation between rings or to artifacts acquired during tree sampling, storage or laboratory manipulation.

Overall, the results of these three modern samples show that there is no significant difference among the pretreatments, which all lead to accurate values: the IAEA certified value for C3 and clean air atmospheric values for Quercus-1980 and Larix-1980 (Figures 2–4).

Similarly, the results for the subfossil pine COUT203 measurements show no ¹⁴C age difference among pretreatments, nor even for the absence of pretreatment (Figure 5). This indicates the lack of contamination in the original wood. Furthermore, our results are compatible with a single ¹⁴C measurement performed at the Poznan AMS laboratory (Figure 5).

Figure 5 ¹⁴C ages of the COUT203 sample, pretreated with different methods (error bars are 1σ, including counting statistics and sample preparation uncertainties), compared with previous ¹⁴C measurements performed in the Poznan laboratory (±1σ value between solid black lines; ±2σ between dashed black lines). Two ¹⁴C measurements, from 2 samples that underwent different combustions/graphitizations, were performed for each pretreatment. The weighted average of all values obtained from pretreatments is the red solid line, while the corresponding ±1σ weighted error is shown by dotted red lines. (Color refers to online version.)

In summary, the replication for each pretreatment gives highly reproducible results for most samples (Figures 2–5). The only significant differences are observed for the wood blanks (Figure 6). For the Plio wood, samples measured without pretreatment and with the simplest ABA pretreatment, exhibit the largest scatter, in contrast with the results obtained with other pretreatments. The BABA and cellulose extraction procedures (with the exception of α-cellulose and ABA-B procedure) give similar results with average values ranging from 0.0025 to 0.0020 F¹⁴C (48,300 to 50,000 BP; Figure 6A).

Figure 6 F¹⁴C content of blank samples (A: Plio; B: VIRI-K) pretreated with different methods (error bars are 1σ). Different pretreatments (pretr.) were performed for each method (i.e. 1 pretr., 2 pretr., etc. indicate the specific—first, second, etc.—replication of each pretreatment: ABA, BABA, etc.) and two AMS measurements were performed for each pretreatment, when enough material was available. The only exception is VIRI-K treated with ABA-B, which was measured more than twice for each pretreatment. The total numbers of measurements (# meas.) are given for each method at the beginning of each line of the Figure, as well as the ¹⁴C age average with SD. No background correction has been subtracted.

As anticipated, the α-cellulose extraction is probably too aggressive for this very old and poorly preserved sample. Indeed, this procedure destroyed most of the material in two of the three extractions (see also the discussion of extraction yields, above). Moreover, our single measurement on this small-sized residual cellulose sample show a F¹⁴C of 0.0037, corresponding to an age of 45,000 BP. This is probably linked to an increase of the relative impact of the contamination compared to the small amount of residual carbon for this sample.

The first ABA-B pretreatment of the Plio wood was performed with one final bleaching 2-hr step, during which intense effervescence occurred. The results of three replications of this pretreatment (6 measurements) show an average value of 0.0022 F¹⁴C with an SD of 0.0006 (49,300 with an SD of 2500 yr BP). Three additional pretreatments were performed with two 1-hr bleaching steps instead of one bleaching step lasting 2 hr. The six new results give an average F¹⁴C value of 0.0015, corresponding to an age of 52,300 yr BP with an SD of 150 yr, which is better than the first ABA-B procedure (Figure 6A).

Concerning the VIRI-K blank wood (Figure 6B), untreated samples exhibit ages ranging between 0.0040 and 0.0021 F¹⁴C (corresponding to an age between 44,400 and 49,700 yr BP), which is older than those observed for the Plio untreated samples that range between 0.0066 and 0.0028 F¹⁴C (between 40,400 and 47,100 yr BP). This may be due to heterogeneities in the younger material contaminating these blank samples or may reflect a random scatter. The age results for the samples pretreated with the ABA procedure show high variability, similar to that observed for the Plio wood. The BABA pretreatment applied to VIRI-K leads to higher blank values than for Plio, while the BABAB-mod, Holo-cell and α-cellulose procedures show comparable results of about 0.0020 F¹⁴C (50,000 yr BP) on average. The higher scatter of α-cellulose values might be explained by increased contamination during this long pretreatment, which include many steps.

Results with the BABBA1 procedure gave an average value of 0.0030 F¹⁴C with an SD of 0.0009 (47,000 yr BP with an SD of 2700), a higher and more scattered blank value than for previous cellulose extraction procedures. These results were somewhat surprising, considering that BABBA1 combines the strengths of the BABAB-mod and Holo-cell methods, which both provided better blanks. The procedure was thus strengthened (BABBA2) with longer NaOH and HCl steps. The BABBA2 results exhibit a smaller scatter, but the average blank value is still rather high: 0.0029 F¹⁴C with an SD of 0.0004 (equivalent to 47,100 yr BP with a SD of 1000; Figure 6B) when compared to other results for this wood and to measurements obtained for the Plio wood. An additional source of variability may also be linked to contamination during the rather long and strong BABBA procedures.

Finally, the ABA-B procedure gave excellent results with a low background average value and a rather small scatter (Figure 6B). This confirms a similar observation detected for the Plio wood measured with the same procedure. In order to further test its reproducibility, ABA-B was repeated again 9 times on VIRI-K, leading to a total of 25 measurements which gave an average value of 0.0017 with an SD of 0.0003 (corresponding to an age of 51,300 yr BP and an SD of 1500; Figure 6B).

The ABA-B procedure was thus chosen as the preferred protocol based on its precise and accurate measurements on modern, ancient and blank wood samples. This procedure is also optimal because its overall duration is shorter than most of the other procedures.

Further Validation of the Preferred Wood Pretreatment

After the phase of developments and tests described above, the preferred ABA-B procedure was used to date the wood samples distributed in the frame of the 2013 SIRI Radiocarbon Intercomparison. The chemical extraction was performed twice for each sample and two ¹⁴C measurements were generally made for each pretreatment (Table 4). C3 samples and VIRI-K blank wood, pretreated with ABA-B treatment, were used respectively as control standards and for background correction. The ¹⁴C measurements performed with AixMICADAS followed the same protocol as described in the previous section, with a new convention for the background correction and error propagation: it was performed by using the average blank value obtained during the measurement session and a conservative variability of F¹⁴C=0.0003 (value obtained from the standard deviation of 25 VIRI-K blanks). In rare cases of blank variability larger than F¹⁴C=0.0003 during a particular session, the measured variability during the session is used instead for the error calculation.

Table 4 Results of the Sixth International Radiocarbon Intercomparison (SIRI) 2013: official Intercomparison results (Scott et al Reference Scott, Naysmith and Cook2016) and our results with relative 1σ uncertainty (¹⁴C age and F¹⁴C). The number of measurements determined in Aix-en-Provence is indicated in the last column.

^{a 14}C finite age after blank subtraction.

^b Weighted means of ¹⁴C age without blank subtraction.

^c Weighted means of ¹⁴C age after blank subtraction (the F¹⁴C of mean blank is 0.0023±0.0003 and 0.0016±0.0003, respectively, for the two magazines of graphite targets). These conventional age values are calculated following van der Plicht and Hogg (Reference van der Plicht and Hogg2006), who recommend to using F¹⁴C=F¹⁴C+2 σ when F¹⁴C<2 σ.

Table 4 provides the weighted means and errors for the different samples, in comparison with average results from the SIRI intercomparison (Scott et al. Reference Scott, Naysmith and Cook2016). Our results are in very good agreement with the SIRI results, which further validate our choice of the ABA-B pretreatment.

1. II. Dating a Pine Sequence Belonging to the Younger Dryas Event

STUDY AREA AND CONTEXT

In the region of the upper course of the Durance River in the Southern French Alps, many subfossil woods (Pinus sylvestris L.) were discovered in alluvial sediments dated from Late Glacial to the first part of the Holocene (Miramont et al. Reference Miramont, Sivan, Rosique, Edouard and Jorda2000). Since the end of the last glacial period, the rivers of this region have been characterized by alluvial detrital phases, interrupted by periods of river incision. Pines were buried by floods during these sedimentation phases. Marl soils allowed the very good preservation of these subfossil stems, which were sometimes preserved with their bark. These subfossil pines were subsequently revealed by the modern incision of the rivers.

Eighteen subfossil stems were discovered in the Barbiers River alluvium in the region of Sisteron. Because of the geomorphological stress and the harshness of the mountain climate, which is reinforced by the intensity and irregularity of the Mediterranean rainfall regime, tree growth was characterized by many anomalies, which made it difficult to synchronize all sequences with their dendrochronological patterns (Miramont et al. Reference Miramont, Sivan, Guibal, Kromer, Talamo and Kaiser2011). So far, only a few trees have been grouped into two floating chronologies: BARB-A and BARB-B (Miramont et al. Reference Miramont, Sivan, Guibal, Kromer, Talamo and Kaiser2011). Preliminary ¹⁴C ages were used to tentatively place these trees on the absolute dendrochronological timescale (Kaiser et al. Reference Kaiser, Friedrich, Miramont, Kromer, Sgier, Schaub, Boeren, Remmele, Talamo, Guibal and Sivan2012).

CONCLUSIONS AND PERSPECTIVES

The capacity to date wood has been established in the new ¹⁴C laboratory in Aix-en-Provence equipped with AixMICADAS and an automated graphitization system. In this study, different chemical pretreatments are tested on wood samples of known ages. The tested protocols vary from the simple ABA technique to more elaborate protocols for cellulose purification.

All pretreatments are efficient for dating pure cellulose (IAEA-C3 standard) and various wood samples, ranging from modern woods to a subfossil sample around 7000 yr BP. By contrast, different performances are observed for old woods reaching the ¹⁴C limit (“blank wood”). This is due to varying efficiencies of the contamination removal among protocols. We obtain the oldest and most reproducible ¹⁴C results with holocellulose extraction procedures.

The acid-base-acid-bleaching pretreatment (ABA-B) is selected as an optimal choice based on its limited duration and complexity, and because of the excellent and reproducible analytical results (as shown by ¹⁴C results, ¹³C/¹²C ratios, carbon % and overall mass yield %) that it ensures.

The efficiency of the ABA-B protocol is highlighted through the analysis of wood samples of the Sixth International Radiocarbon Intercomparison (SIRI). Our results are in good agreement with consensus values obtained by the SIRI participants.

The new method is applied to a ca. 240-yr-long tree-ring sequence from two subfossil pines (Barb12-17) collected in the southern French Alps and dated to the Younger Dryas cold period. The new ¹⁴C analyses at high resolution (every third year) agree well with high-precision results obtained previously. Barb12-17 is tentatively dated to the interval between 12,836 and 12,594 cal BP by matching its ¹⁴C pattern to the Kauri and YDB chronologies.

A mean IHG value of 57 yr with a SD of 11 yr is derived from the comparison with the Kauri sequence from New Zealand. The IHG stayed relatively high throughout the studied period, without evidence of a collapse around 12,800 cal BP. The IHG exhibits a transient maximum during the 12,750–12,710 cal BP interval. This period corresponds to a global rise of atmospheric ∆¹⁴C but with an apparent delay in the Southern Hemisphere. Future work will be focused on additional trees from the YD period and on complementary records from other archives for both ¹⁴C and ¹⁰Be.