PALEOCEANOGRAPHY, VOL.???, XXXX, DOI:10.1029/, Supporting Information for Poorly ventilated deep ocean at the Last Glacial Maximum inferred from carbon isotopes: a data-model comparison study L. Menviel 1,2,, J. Yu 3, F. Joos 4, A. Mouchet 5,6,7, K.J. Meissner 1,2, M.H. England 1,2 L. Menviel, l.menviel@unsw.edu.au 1 Climate Change Research Centre, University of New South Wales, Sydney, Australia 2 ARC Centre of Excellence for Climate System Science, Australia 3 Research School of Earth Sciences, The Australian National University, Canberra, ACT 0200, Australia 4 Climate and Environmental Physics, Physics Institute and Oeschger Centre for Climate Change Research, University of Bern, CH-3012 Bern, Switzerland
X - 2 MENVIEL ET AL.: LGM δ 13 C Contents of this file 1. Text S1 to S7 2. Figures S1 to S10 3. Tables S1 to S3 S1. LGM experiment performed with the Bern3D Earth System Model To further test the robustness of our results, we also analyze a LGM simulation performed with the Bern3D Earth System Model of intermediate complexity [Ritz et al., 2011]. The Bern3D model consists of an Energy Balance Model, a physical ocean model, here applied with a horizontal resolution of 36 x 36 grid boxes and 32 unevenly spaced layers, a sea ice model, a terrestrial carbon model, a marine biogeochemical cycle and a sediment model [Parekh et al., 2008; Tschumi et al., 2008]. A transient experiment was 5 Astrophysics, Geophysics and Oceanography Department, Université de Liège, Liège, Belgium 6 Laboratoire des Sciences du Climat et de l Environnement (LSCE), IPSL-CEA-CNRS-UVSQ, Gif-sur-Yvette, France 7 Now at Max-Planck Institute for Meteorology, Hamburg, Germany
MENVIEL ET AL.: LGM δ 13 C X - 3 performed from the last interglacial (125 ka B.P) to the LGM [Menviel et al., 2012] forced with time-varying changes in insolation, ice sheet extent as well as radiative changes due to the varying atmospheric CO 2 and CH 4 content. As the model includes iron as a prognostic tracer, atmospheric iron deposition to the ocean at the LGM is enhanced. The simulation also includes an interglacial-glacial terrestrial carbon release of 340 GtC. This LGM state features a weak and shallow NADW, very weak AABW transport and no formation of NPIW. As can be seen in Figure S1, LGM δ 13 C anomalies simulated in this experiment display positive anomalies in the North Atlantic at intermediate depth and significant negative anomalies in the deep Atlantic and Pacific, in good agreement with benthic δ 13 C data. The results obtained with the Bern3D model are in good agreement with the ones obtained with LOVECLIM as both show similar responses to changes in NADW and AABW [Menviel et al., 2015]. An interesting feature of the Bern3D experiment is that due to the higher vertical resolution in the deep ocean and sediment-ocean interactions, very low δ 13 C values are simulated below 4200 m.
X - 4 MENVIEL ET AL.: LGM δ 13 C S2. Terrestrial carbon change Figure 1 suggests that the best fit between model and paleoproxy is obtained for experiments with an equivalent terrestrial carbon change of 369 to 426 GtC, weak NADW and weak to very weak AABW (V3LNAwSOs and V3LNAwSOwSHWw ). Both experiments display whole ocean means within two σ of the benthic estimate ( =-0.34 and =-0.385 compared to =-0.38 ). The mean Atlantic basin anomaly is slightly underestimated ( =-0.33 and =-0.27 versus =-0.41 ). The mean Indian ocean anomaly amounts to -0.36 ( ) and -0.42 ( ) compared to -0.45 ( ). In the Pacific basin, the simulated South Pacific mean δ 13 C anomaly is in fairly good agreement with the proxy for both experiments ( = -0.31 and =-0.42 versus =-0.39 ), while the North Pacific δ 13 C anomaly is generally overestimated ( =-0.39 and =-0.45 compared to =-0.28 ). Regarding the set of experiments with a 570 GtC lower LGM terrestrial carbon reservoir (V4), Figure 1 suggests that the mean δ 13 C anomalies of the experiment with weak NADW ( ) are in better agreement than for the experiment with weak NADW and very weak AABW ( ). Both experiments lead to mean Atlantic δ 13 C anomalies in good agreement with the proxy ( =-0.44 and =-0.37 compared to =-0.41 ), but the whole ocean mean δ 13 C anomaly is lower than in the data for the experiment with weak NADW and very weak AABW ( =-0.49 versus =-0.38 ), principally due to very low Pacific δ 13 C anomalies ( =-0.55 compared to =-0.34 ). Since atmospheric δ 13 CO 2 is computed prognostically in these experiments, the LGM δ 13 CO 2 value should also be taken into account. For example, LGM δ 13 CO 2 is lower in
MENVIEL ET AL.: LGM δ 13 C X - 5 the experiment with weak NADW V4 than in the one with weak NADW and AABW ( =-6.95, =-6.58 ), thus explaining the relatively higher oceanic δ 13 C value in the experiment with weak NADW ( ) compared to the one with weak NADW and AABW ( ). The results shown in Figure 1 suggest that the deglacial terrestrial carbon uptake was less than 570 GtC. Mean δ 13 C anomalies are not negative enough for the set of experiments V2 (green symbols) everywhere but in the North Pacific, thus suggesting that a terrestrial carbon uptake of 205± 50 GtC is not sufficient to explain the mean change in oceanic δ 13 C. Since the V3 experiments with strong (V3L ) or shutdown (V3LNAoff +) NADW formation or strong AABW (V3LSOs, V3LNAwSOs, V3LNAwGRSOs ) display no significant correlation with the proxy data, we discard them in our estimate of the most probable glacial-interglacial terrestrial carbon change. We thus estimate that the LGM terrestrial carbon was 378±44 GtC (1σ) lower than during the late Holocene. This estimate is also in agreement with a LGM experiment performed with the Bern3D model featuring a 340 GtC lower LGM terrestrial carbon content ( ). S3. δ 13 C as a function of oceanic circulation While the mean LGM δ 13 C globally represents glacial/interglacial changes in terrestrial carbon, the anomaly pattern is dictated by changes in oceanic circulation. The width of the probability density functions (PDF) varies as a function of the oceanic circulation, with a narrow PDF associated with a strong oceanic circulation and a wide PDF associated with a globally weak oceanic circulation (Figure S2).
X - 6 MENVIEL ET AL.: LGM δ 13 C S4. LGM SST The SST anomalies as simulated in the LGM experiments are in good agreement with the MARGO dataset [Waelbroeck et al., 2009] (Figure S5). The mean annual LGM SST is about 1 C lower at low latitudes (20 S-20 N), about 3.5 C lower in the Southern Ocean (40 S-60 S) and 6 C lower in the North Atlantic (40 N-60 N). The main model-data discrepancies occur in the Eastern Equatorial Atlantic and in the Northeastern Pacific, where the cooling is underestimated compared to the proxies. In addition, a few records from the Greenland Iceland Norwegian seas suggest warmer conditions there during the LGM, in contrast to our model results.
MENVIEL ET AL.: LGM δ 13 C X - 7 S5. Enhanced export production in the Southern Ocean The simulated export production in LGM experiments with weak NADW displays lower export at high southern latitudes and in the Eastern part of the Pacific basin, as well as increased export production in the equatorial Atlantic in relative agreement with a compilation of paleoproxy records [Kohfeld et al., 2005]. However, since the impact of iron fertilization is not simulated, export production also decreases north of the current Antarctic polar front contrarily to paleoreconstructions (Figure S6 left). To simulate the impact of iron fertilization, export production was enhanced between 56 S and 36 S in some experiments (e.g. V3LNAwGR, Table 1). Instead of decreasing by 10%, export production increases by 9% in the Southern Ocean (Figure S6 right), therefore leading to a relative 30% increase compared to LGM experiments with weaker NADW (V3LNAw ). The resulting enhanced nutrient utilization over the Southern Ocean has however a relatively weak impact on the oceanic δ 13 C distribution (Figure S7). Increased nutrient utilization over the Southern Ocean increases the export of depleted organic matter to depth, thus leaving the surface ocean enriched in 13 C, while slightly decreasing intermediate/deep Southern Ocean δ 13 C. The positive surface δ 13 C anomalies are advected to low latitudes through the Sub-Antarctic mode waters. This result is in agreement with a previous experiment performed with the Bern3D Earth system model [Menviel et al., 2012], which includes enhanced iron fertilization. Figure S7, E shows that the associated 16% increase in export production over the Southern Ocean in the Bern3D model only leads to a 0.03 δ 13 C decrease in the deep Southern Ocean.
X - 8 MENVIEL ET AL.: LGM δ 13 C S6. Enhanced carbon storage in the deep ocean when the oceanic circulation is weak We estimate the change in deep ocean carbon resulting from the various oceanic circulation states. As seen in Figure S8, weak NADW leads to a mean deep ocean DIC increase everywhere but in the North Atlantic and Southern Ocean. By contrast when both NADW and AABW are weakened, deep ocean DIC is greater everywhere, with a strong increase in the Western Pacific and in the South Atlantic. Deep ocean ( 2600 m depth) carbon is thus 216 GtC greater when NADW weakens (V3LNAw ) and up to 504 GtC greater when both NADW and AABW (V3LNAwSOwSHWw ) are reduced. S7. Dissolved oxygen Figure S9 shows dissolved oxygen anomalies as simulated in different LGM experiments and compared to the control run as well as to a compilation of LGM changes in dissolved oxygen [Jaccard and Galbraith, 2012]. A weakening of NADW to about 15 Sv (V3LNAw ) leads to positive O 2 anomalies in the intermediate North Atlantic (Figure S9,B) due to increased solubility of O 2 in cold waters and reduced export production (Figure S6). If NADW is further weakened (V3LNAwSOwSHWw, Figure S9 C,F) or shut down (V3LNAoff +, Figure S9,A,D), then negative O 2 anomalies develop in the intermediate and deep North Atlantic. Reduced AABW formation lowers the dissolved O 2 content in the deep Southern Ocean (Figure S9,F) in agreement with Southern Ocean proxy records. Positive dissolved oxygen anomalies are simulated in the intermediate Pacific in all the simulations. Colder conditions increase the solubility of O 2 in seawater. A slight increase in North Pacific Deep Water formation also increases the dissolved oxygen content in the
MENVIEL ET AL.: LGM δ 13 C X - 9 North Pacific in experiments with weak NADW, while a large O 2 increase in the North Pacific is simulated in the experiment with NADW off because of the strong NPDW formation. Reduced export production in the Eastern Pacific (Figure S6, left), also contributes to the higher intermediate depth oxygen content. Disagreement between model and proxies occurs mainly in the Arabian Sea and the Angola basin, which could be due to a misrepresentation of these upwelling areas in the model. While the best agreement between simulated δ 13 C and paleoproxy record is obtained for a poorly ventilated ocean, most of the deep ocean is well above anoxia (Figure S10) in agreement with paleoreconstructions [Jaccard et al., 2014]. In the experiment which displays the weakest AABW formation (V3LNAwSOwSHWw, Tables 1 and S1), dissolved oxygen levels are below the anoxic level of 20 µmol/l only in a small area of the Eastern Equatorial Pacific and the North Pacific.
X - 10 MENVIEL ET AL.: LGM δ 13 C Figure S1. LGM δ 13 C ( ) as simulated by the Bern3D model. Zonally averaged over (A) the Atlantic and (B) Pacific oceans. Overlaid is the meridional overturning streamfunction (Sv). 3.5 3 Data AABW strong NADW weak NADW & AABW weak NADW off Normalized occurence 2.5 2 1.5 1 0.5 0 1.5 1 0.5 0 0.5 1 δ 13 C, permil Figure S2. Bands of probability density functions as a function of the oceanic circulation. The mean LGM δ 13 C was substracted from all the experiments as well as from the data to highlight the width of the PDF. From narrow to wide PDF: strong AABW (SOs, red), weak NADW (NAw, blue), weak NADW and AABW (NAwSOw, magenta) and NADW off (orange).
MENVIEL ET AL.: LGM δ 13 C X - 11 Figure S3. δ 13 C anomaly ( ) zonally averaged over the Atlantic basin for LGM experiments with strong NADW formation, from left to right, V1L, V2L and V3L and compared to the pre-industrial control run. The δ 13 C pattern is consistent across the 3 experiments, irrespective of the mean oceanic δ 13 C change.
X - 12 MENVIEL ET AL.: LGM δ 13 C Figure S4. δ 13 C anomaly ( ) zonally averaged over (left) the Atlantic and (right) the Pacific basin for LGM experiments with 514-610 GtC lower LGM terrestrial carbon (V4) and (A) weak NADW (V4LNAw), (B) weak NADW and very weak AABW (V4LNAwSOwSHWw), (C) weak NADW and strong AABW (V4LNAwSOs). The over-estimation of negative δ 13 C anomalies in the deep Pacific and intermediate North Atlantic indicates that the LGM to late Holocene change in terrestrial carbon was most likely less than 500 GtC.
MENVIEL ET AL.: LGM δ 13 C X - 13 Figure S5. LGM SST anomalies as simulated in an experiment with weak NADW and very weak AABW ( ) as well as for the MARGO dataset [Waelbroeck et al., 2009]. The annual LGM 0.1m sea-ice level is overlaid. Figure S6. Export production anomalies (gc/m 2 /yr) for LGM experiments with (left) weak NADW ( ) and (right) weak NADW and enhanced export production over 56 S-36 S ( ) compared to the control pre-industrial experiment and to a compilation of paleoproxy records [Kohfeld et al., 2005]. For the paleoproxy reconstructions, a bright orange star indicates higher and light orange slightly higher export production at the LGM than in the late Holocene; a dark blue and light blue star indicate lower and slightly lower export production at the LGM.
X - 14 MENVIEL ET AL.: LGM δ 13 C Figure S7. δ 13 C anomalies ( ) for experiments with (A, B) very weak AABW ( ); (C, D) 30% enhanced export production over 56-36 S ( ), both compared to an experiment with weak NADW ( ) averaged over (left) the Atlantic basin and (right) the Pacific basin. (E, F) δ 13 C anomalies for a LGM experiment featuring enhanced iron fertilization compared to a LGM experiment without iron fertilization performed with the Bern3D [Menviel et al., 2012].
MENVIEL ET AL.: LGM δ 13 C X - 15 Figure S8. DIC anomalies (µmol/l) averaged below 2600 m depth for experiments with (A) weak NADW (V3LNAw ) and (B) weak NADW and very weak AABW (V3LNAwSOwSHWw ) compared to an experiment with strong NADW (V3L ).
X - 16 MENVIEL ET AL.: LGM δ 13 C Figure S9. Dissolved oxygen anomalies (µmol/l) as simulated in the LGM experiments compared to the preindustrial control run and to dissolved O 2 proxies [Jaccard and Galbraith, 2012], (left) in intermediate waters (700-1000m) and (right) in deep waters (2600-3300m). The experiments shown are (A, D) for NADW off (V3LNAoff +), (B, E) weak NADW (V3LNAw ) and (C, F) weak NADW and very weak AABW (V3LNAwSOwSHWw ). The paleoproxy reconstructions shown here for O 2 are qualitative only: a red star indicates higher O 2 content at the LGM than in the late Holocene, a blue star lower O 2 content and a yellow star an ambiguous record.
MENVIEL ET AL.: LGM δ 13 C X - 17 Figure S10. Dissolved oxygen content (µmol/l) in the deepest layer of the ocean as simulated in a LGM experiment with weak NADW and very weak AABW ( ). The contour shows the 20µmol/L level, below which anoxic conditions are present.
X - 18 MENVIEL ET AL.: LGM δ 13 C Expt. NADW (Sv) AABW (Sv) pco 2 (ppmv) δ 13 CO 2 ( ) 14 CO 2 ( ) Initial Cond. Length (kyr) PI Control 25.5 12 280.5-6.38 0 V1L 26.7 13 202-6.45 394 PI 23 V1LNAoff + 2.3 7.5 191-6.45 440 V1LNAw 2 V1LNAw 10.4 8.2 187-6.46 437 V1L 13 V1LNAwSOw 15.0 7.6 183-6.51 436 V1LNAw 10 V1LNAwSOs 19.1 9.7 192-6.46 415 V1LNAw 4 V1LNAwGR 17.4 8.7 183-6.45 425 V1LNAw 4 V2L 23.6 9.4 207-6.5 408 V1L 5 V2LNAoff + 2.5 7.8 206-6.47 440 V2LNAw 2 V2LNAw 14.0 8.3 202-6.43 430 V2L 4 V2LNAwSOw 16 7.1 202-6.55 438 V2LNAw 2 V2LNAwSOs 19.0 10.5 205-6.52 402 V2LNAw 2 V2LNAwSHWw 13.8 7.3 195-6.37 468 V2LNAw 2 V2LNAwGR 13.0 8.0 184-6.58 440 V2LNAw 2 V3L 22.5 9.5 209-6.81 388.5 V1L 5 V3LNAoff + 2 8.0 206-6.67 429 V3LNAw 2 V3LNAw 14.7 8.3 203-6.71 418 V3L 4 V3LNAwSOw 11.3 6.4 196-6.62 442 V3LNAw 2 V3LNAwSHWw 13.9 7.2 199-6.54 469 V3LNAw 2 V3LNAwSOwSHWw 11.2 5.1 191-6.46 500 V3LNAw 5 V3LSOs * 26 16 215.5-6.78 360 V3L 2 V3LNAwSOs 16 13 212.5-6.78 378 V3LNAw 2 V3LNAwGR 14.8 8.1 200-6.64 300 V3LNAw 2 V3LNAwGRSOs 18 13 201-6.67 381 V3LNAw 2 V4LNAw 14.6 10.5 191-6.95 439 V1L 5 V4LNAwSHWw 13.2 6.9 181-6.65 480 V4LNAw 2 V4LNAwSOwSHWw 13.6 5.8 174-6.58 507 V4LNAw 2 V4LNAwSOs 16 13 194-6.90 393 V4LNAw 2 V4LNAwGR 13 8.3 184-6.87 441 V4LNAw 2 Table S1. Main results of LGM experiments at 20 ka B.P. indicating the maximum overturning streamfunction in the North Atlantic (NADW), the maximum AABW transport in the Indo-Pacific basin (AABW), the LGM atmospheric CO 2, δ 13 CO 2 and 14 CO 2 as well as the initial conditions of each experiments and their length.
MENVIEL ET AL.: LGM δ 13 C X - 19 Core ID Latitude Longitude Depth Ventilation Reference (m) age (years) RAPiD-17-5P 61 29 N 19 32 W 2303 2700 Thornalley et al. [2011] JC89-SHAK-10-10K 37 50 N 09 31 W 1127 441 Freeman et al. [2016] JC89-SHAK-14-4G 37 50 N 09 44 W 2063 1423 Freeman et al. [2016] MD99-2334K 37 48 N 10 10 W 3146 2410 Skinner et al. [2014] JC89-SHAK-06-4K 37 34 N 10 22 W 2642 1973 Freeman et al. [2016] JC89-SHAK-03-6K 37 43 N 10 30 W 3735 2227 Freeman et al. [2016] JC89-SHAK-05-3K 37 36 N 10 42 W 4670 2406 Freeman et al. [2016] KNR140-39GGC 31 67 N 75 42 W 2975 1895 Keigwin and Schlegel [2002]; Freeman et al. [2016] KNR140-26GGC 29 70 N 73 40 W 3845 1750 Keigwin [2004]; Freeman et al. [2016] KNR140-12JPC 29 10 N 72 90 W 4250 2300 Keigwin [2004]; Freeman et al. [2016] KNR140-22JPC 28 02 N 74 41 W 4712 2200 Keigwin [2004]; Freeman et al. [2016] MD03-2707 02 30 N 09 24 E 1295 1233 Weldeab et al. [2016] RC24-08 01 20 S 11 54 W 550 (3885) 825 Cléroux et al. [2011]; Freeman et al. [2016] GS07-150-17/1GC-A 04 13 S 37 04 W 1000 976 Freeman et al. [2015] MD09 3257 04 15 S 36 21 W 2344 1631 Freeman et al. [2016] MD09 3256Q 03 33 S 35 23 W 3257 2026 Freeman et al. [2016] KNR-159-5-36GGC 27 31 S 46 28 W 1268 1398 Sortor and Lund [2011]; Freeman et al. [2016] KNR-159-5-78GGC 27 48 S 46 33 W 1829 1173 Lund et al. [2015] TNO57-21 41 06 S 7 48 E 4981 2383 Barker et al. [2010]; Freeman et al. [2016] MD07-3076Q 44 05 S 14 13 W 3770 3477 Skinner et al. [2010]; Freeman et al. [2016] Deep sea corals 60 S 60 W 819 1697 Burke and Robinson [2012]; Freeman et al. [2016] Deep sea corals 60 S 60 W 1134 1680 Burke and Robinson [2012]; Freeman et al. [2016] Table S2. LGM ventilation ages (years) estimated from Atlantic sector marine sediment cores. Please note that these values are associated with an uncertainty of 200-500 years.
X - 20 MENVIEL ET AL.: LGM δ 13 C Core ID Latitude Longitude Depth Ventilation Reference (m) age (years) EW0408-85JC 59 55 N 144 15 W 682 1594 Davies-Walczak et al. [2014] MD02-2489 54 39 N 148 92 W 3640 3093 Gebhardt et al. [2008]; Sarnthein et al. [2013]; Rae et al. [2014] ODP887 54 37 N 148 45 W 3467 2789 Galbraith et al. [2007]; Okazaki et al. [2010] MD01-2416 51 27 N 167 73 E 2317 4185 Sarnthein et al. [2013] NES25-1 GGC27 49 60 N 150 20 E 995 2427 Keigwin and Schlegel [2002]; Okazaki et al. [2010] B34-91 49 00 N 150 03 E 1227 2508 Keigwin and Schlegel [2002]; Okazaki et al. [2010] NES25-1 GGC20 48 90 N 150 40 E 1510 2472 Keigwin and Schlegel [2002]; Okazaki et al. [2010] NES25-1 GGC18 48 8 N 150 40 E 1700 2598 Keigwin and Schlegel [2002]; Okazaki et al. [2010] NES25-1 GGC15 48 6 N 150 40 E 1980 2773 Keigwin and Schlegel [2002]; Okazaki et al. [2010] GH02-1030 42 23 N 144 21 E 1212 2159 Sagawa and Ikehara [2008]; Okazaki et al. [2010] W8709A-13PC 42 12 N 125 75 W 2712 2706 Okazaki et al. [2010] CH84-14 41 73 N 142 55 E 978 1505 Duplessy et al. [1989]; Okazaki et al. [2010] ODP1019 41 68 N 124 93 W 980 2086 Okazaki et al. [2010] MR01K03-PC4/PC5 41 12 N 142 40 E 1366 2582 Ahagon et al. [2003]; Hoshiba et al. [2006]; Okazaki et al. [2010] ODP893 34 29 N 120 04 W 588 1400 Kennett and Ingram [1995]; Okazaki et al. [2010] KT89-18-P4 32 15 N 133 90 E 2700 2748 Okazaki et al. [2010] MV99-GC31/PC08 23 50 N 111 60 W 705 1893 Marchitto et al. [2007]; Okazaki et al. [2010] 50-37KL 18 90 N 115 77 E 2695 2691 Broecker [1998]; Okazaki et al. [2010] MD01-2386 1 0 N 130 0 E 2800 2597 Broecker et al. [2008]; Okazaki et al. [2010] MD97-2138 1 0 S 146 0 E 1900 2357 Broecker et al. [2004]; Okazaki et al. [2010] VM21-30 1 22 S 89 68 W 617 2526 Stott et al. [2009]; Okazaki et al. [2010] TR163-31 3 62 S 83 97 W 3210 3599 Shackleton et al. [1988]; Okazaki et al. [2010] SO161-SL122 36 S 73 00 W 1000 1436 Pol-Holz et al. [2010] RR0503-JPC64 37 25 S 177 00 E 651 974 Rose et al. [2010] H213 37 2 S 177 10 E 2065 3470 Sikes et al. [2000] MD97-2121 40 23 S 177 60 E 2314 2715 Skinner et al. [2015] MD97-2120 43 32 S 174 56 E 1210 772 Rose et al. [2010] SO213-84-1 45 0 S 175 0 E 972 1574 Ronge et al. [2016] SO213-82-1 46 0 S 176 0 E 2066 4770 Ronge et al. [2016] SO213-76-2 46 50 S 178 0 W 4339 4272 Ronge et al. [2016] MD07-3088 46 0 S 75 41 W 1536 2303 Siani et al. [2013] Table S3. LGM ventilation ages (years) estimated from Pacific sector marine sediment cores. LGM ventilation ages (τ= 1 λ.ln( 14 C atm+1000 )) were calculated with the decay constant λ=1/8223 14 C bf +1000 yr 1, 14 C atm the contemporaneous IntCal 2013 atmospheric 14 C value [Reimer et al., 2013] and 14 C bf the LGM benthic data ( 21 18 ka B.P.). When several measurements were available during this timeframe, the mean value was taken. Please note that these values are associated with an uncertainty of 200-500 years.