Latitudinal differences in the amplitude of the OAE-2 carbon isotopic excursion: pCOi and paleo productivity

. A complete, well-preserved record of the Ceno-manian/Turonian (C/T) Oceanic Anoxic Event 2 (OAE-2) was recovered from Demerara Rise in the southern North Atlantic Ocean (ODP site 1260). Across this interval, we determined changes in the stable carbon isotopic composi­ tion of sulfur-bound phytane (513Cphytane), a biomarker for photosynthetic algae. The á13Cphytane record shows a pos­ itive excursion at the onset of the OAE-2 interval, with an unusually large amplitude (~7 %o) compared to existing C/T proto-North Atlantic á13Cphytane records (3-6 %o). Overall, the amplitude of the excursion of á13Cphytane decreases with latitude. Using reconstructed sea surface temperature (SST) gradients for the proto-North Atlantic, we investigated envi­ ronmental factors influencing the latitudinal 513Cphytaiie gra­ dient. The observed gradient is best explained by high pro­ ductivity at DSDP Site 367 and Tarfaya basin before OAE-2, which changed in overall high productivity throughout the proto-North Atlantic during OAE-2. During OAE-2, produc­ tivity at site 1260 and 603B was thus more comparable to the mid-latitude sites. Using these constraints as well as the SST and 513Cphytaiie-records from Site 1260, we subsequently re­ constructed pC 02 levels across the OAE-2 interval. Ac­ cordingly, pC 02 decreased from ca. 1750 to 900 ppm during OAE-2, consistent with enhanced organic matter burial re­ sulting in lowering pC 02. Whereas the onset of OAE-2 coin­ cided with increased pC 0 2, in line with a volcanic trigger for this event, the observed cooling within OAE-2 probably re­ sulted from CO 2 sequestration in black shales outcompeting CO 2 input into the atmosphere. Together these results show that the ice-free Cretaceous world was sensitive to changes in pC 02 related to perturbations of the global carbon cycle.


Introduction
The Mid-Cretaceous is characterized by an overall warm cli mate (Huber et al., 2002), punctuated by several colder pe riods (e.g., Bornemann et al., 2008: Forster et al., 2007. This overall warm climate probably resulted from elevated atmospheric greenhouse gas concentrations, as atmospheric p C 0 2 levels are estimated to have been 3-8 times higher than pre-industrial values (Schlanger et al., 1987: Wilson et al., 2002: Huber et al., 1999: Berner, 1992: Barclay et al., 2010: Berner and Kothavala, 2001. The most widely accepted ex planations for these high atmospheric p C 0 2 levels are in creased rates of seafloor spreading and enhanced plate mar gin volcanism (e.g. Turgeon and Creaser, 2008: Kerr, 1998: Blättler et al., 2011: Jenkyns, 2010. Superimposed on this period of high atmospheric p C 0 2 levels, several short lived episodes of increased or ganic matter (OM) deposition, so-called ocean anoxic events (OAEs) (Schlanger and Jenkyns, 1976: Arthur et al., 1988: Jenkyns, 1980, occurred. Generally, in ma rine settings this enhanced OM burial during OAEs is thought to be the result of either enhanced bioproductiv ity or increased anoxia or a combination of these two fac tors (Kuypers et al., 2002b). One of the most pronounced and widespread OAEs is OAE-2, which occurred at the Cenomanian/Turonian boundary (C/T: 9 3.5Ma, Gradstein et al., 2004) and which is also known as the Cenomanian Turonian Boundary Event (CTBE). A positive carbon isotopic excur sion accompanying OAE-2 has been observed both in ma rine carbonates and in marine and terrestrial OM (Hasegawa, 1997). This excursion has been attributed to enhanced OM burial (Arthur et al., 1988: Schofle andArthur, 1980), be cause organisms preferentially take up light carbon 12C leav ing the remaining carbon in the ocean-atmosphere reservoir enriched in 13C.
Since phytoplankton fix dissolved inorganic car bon (DIC) during photosynthesis the export of phytoplanktic biomass to deeper water and the subsequent burial act as a biological carbon pump. Due to the exchange of CO2 between atmosphere and ocean, this biological carbon pump effectively removes CO2 from the atmosphere. The strength of this biological pump, therefore, modulates greenhouse climate (Royer et al., 2007;Berner, 1992;Arthur et al., 1988). During OAE-2, the increased burial of OM likely resulted in a more efficient carbon pump, lowering p CO2 levels. Recently a reconstruction based on the stomatal index of land plants demonstrated that during OAE-2, two intervals of enhanced OM burial and associated p CO2 decreases occurred (Barclay et al., 2010). This decrease in pCÖ2 within OAE-2 was initially demonstrated by the observed larger amplitude of the carbon isotopic excursion when measured on organics compared to carbonates, since the rate of isotopic fractionation during carbon fixation by phytoplankton decreases at lower pCOi values (Freeman and Hayes, 1992). A reduction in p CO2 of about 300 ppm during the OAE-2 interval was calculated using the iso topic values of biomarkers for photosynthetic algae and cyanobacteria (Sinninghe Damsté et al., 2008). Such p CO2 reconstructions based on isotopic fractionation rely on assumptions for paleoproductivity, temperature and equilib rium CO2 exchange between ocean water and atmosphere, and, therefore, should be considered as estimates.
A short-lived cooling within OAE-2, called the Plenus Cold Event has been observed at several locations (Forster et al., 2007;Gale and Christensen, 1996;Voigt et al., 2004;Jarvis et al., 2011). The Plenus Cold Event is probably re lated to lower pCÖ2 levels at that time (Sinninghe Damsté et al., 2010), which would be in line with the two stomata based intervals of lower p CO2 coinciding with maxima in the carbon isotope excursion (Barclay et al., 2010). How ever, this carbon isotopic record used by Barclay et al. was measured on bulk organic matter, which might have been af fected by compositional changes as well. Preferably recon structed temperature and p CO2 records should be based on the same sedimentary record.
Here we compare the observed change in the compound specific isotope record of sulfur-bound phytane (513Cphytane) at Demerara Rise with other published á 13Cphytane records from the proto-North Atlantic along a latitudinal gradient. Observed differences in the amplitude of the á 13Cphytane ex cursion during OAE-2 are subsequently discussed in terms of variations in productivity and [C02(aq)]. The boundary conditions from this comparison are used to reconstruct at mospheric pCÖ2 levels across the OAE-2 interval.

Setting and stratigraphy
During the Cenomanian-Turonian, Demerara Rise was situ ated in the tropical region of the proto-North Atlantic off the coast of Suriname (Fig. 1). As Demerara Rise was a subma rine plateau at the time, the Cenomanian-Turonian sediments at ODP Site 1260 were most likely deposited at intermedi ate water depth, probably between 500-1500 m (Erbacher et al., 2004b;Suganuma and Ogg, 2006). The Cenomanian-Coniacian sediments of the studied sequence are mostly dark, laminated, carbonaceous, calcareous mud-to marlstones (black shales), interbedded with occasional thin cal careous layers and foraminiferal packstones (Erbacher et al., 2004a;Nederbragt et al., 2007). OM from Demerara Rise is thermally immature with organic carbon contents of up to 20% and mainly of marine origin (Erbacher et al., 2004a, b;Forster et al., 2004;Meyers et al., 2006). Since Demer ara Rise experienced anoxie bottom water conditions during most of the Cenomanian -Coniacian (Erbacher et al., 2004b;Suganuma and Ogg, 2006;van Bentum et al., 2009)   The exact stratigraphie position of OAE-2 was deter mined using the positive isotope excursion of organic car bon (Fig. 2a;Forster et al., 2007). This carbon isotope excursion accompanying OAE-2 can be divided into three phases (cf. Kuypers et al., 2002a;Forster et al., 2007;Tsikos et al., 2004). Phase Fig. 2) consists of the onset of the excursion, up to the first isotopic maxi mum. Phase B (426.21-425.27 mcd, Fig. 2) starts with a de cline in values of the stable carbon isotopes of bulk organic carbon (5 1 3 C t o c ) , followed by a second increase and ends with an interval of steadily high 5 1 3 C t o c values. Finally, the gradual return to nearly pre-excursion values is part of phase C (425.27-424.85 mcd, Fig. 2). Phase A is equivalent to the "first build-up'' phase of the proposed European refer ence section at Eastbourne (Paul et al., 1999). The decline and second increase in phase B corresponds to the "trough" and "second build-up" phase in Eastbourne, while the high values correspond to the "plateau" (Paul et al., 1999). Fol lowing earlier work (Kolonie et al., 2005;Kuypers et al., 2002b;Tsikos et al., 2004) phases A and B together are re ferred to as OAE-2, while phase C represents the recovery phase after the OAE-2 interval. Estimates for the duration of OAE-2 range from 200 ky to 700 ky (Arthur and Premoli-Silva, 1982;Arthur et al., 1987;Sageman et al., 2006;Frijia and Parente, 2008;Erbacher et al., 2005), however, the recent record of Voigt et al. (2008) suggests a duration of 430-445 kyr for OAE-2.

Materials and methods
Sediments used for this study were collected during Ocean Drilling Program (ODP) Leg 207 at Site 1260 (holes A and B) on Demerara Rise (Erbacher et al., 2004a). Biomark ers were analyzed in sediment samples previously used to determine total organic-carbon content (TOC), carbonate (CaC0 3 ) content, stable carbon isotopes of bulk organic car bon (5 1 3 C t o c , Fig. 2a) and the TEXs6 sea surface temper ature proxy (Forster et al., 2007). Sediment samples (3 to 5 g dty mass) were taken approximately every 10 cm above and below the OAE-2 black shales, while within the OAE-2 section, samples were taken every 2-5 cm. Sed iments were freeze-dried, powdered and subsequently ex tracted with an Accelerated Solvent Extractor (Dionex) using a dichloromethane (DCM) -methanol mixture (9:1, v/v). El emental sulfur was removed from the extracts using activated copper. The extracts were then separated into apolar and po lar fractions using a column of activated alumina by elution with hexane/DCM (9:1, v/v) and DCM/methanol (1:1, v/v), respectively.
Raney Nickel desulfurization and subsequent hydrogena tion (Sinninghe Damsté et al., 1993) were used to release sulfur-bound biomarkers from polar fractions. To ensure suf ficient yield this process was only performed on samples that produced polar fractions weighing >5 mg. The desulfurized fraction was separated further into apolar and polar fractions. The apolar fraction obtained from the desulfurized polar fraction was separated into saturated aliphatic, unsaturated aliphatic and aromatic fractions by column chromatography using AgNCH-impregnated silica as the stationary phase and hexane, hexane/DCM (9:1, v/v) and hexane/DCM (1:1, v/v) as eluents.
All fractions were analyzed on a Elewlett-Packard (TIP) gas chromatograph (GC) fitted with a flame ionization de tector (FID) and a sulfur-selective flame photometric detec tor (FPD). Samples were injected on-column, on a CP-Sil 5CB fused silica column (50 m x 0.32 mm i.d.) with helium as carrier gas set at constant pressure (100 KPa). The oven program started at 70 °C, was then heated by 2 0°C m in to 120 °C and finally by 4 °C/min to 320 °C and kept at this tem perature for at least 15 min. To identify compounds, samples were measured on a GC-MS (Thermo Trace GC Ultra) with a mass range m /z 50-800 using a similar column and heat ing program as for the GC, however, with the carrier gas at constant flow.
Compound specific isotope ratios were measured using a GC isotope-ratio mass spectrometer (HP GC coupled to a Thermo Delta-plus XL). For most GCTRMs measurements a similar column and oven program were used as for the GC and GCMS measurements. Samples were all measured at least in duplicate and 513 C values are reported in the standard delta notation against the VPDB standard. IRM performance was monitored with o fflin e calibrated, co-injected, internal standards, standard mixtures (both in house and Schimmel mann standard mixtures B and C) and through the multiple analyses of samples. Accuracy and precision was around 0 . 3 -0 . 6 % o for phytane based on multiple analyses of sam ples and standards.

The OAE-2 carbon isotope excursion at Demerara Rise
The S-bound phytane carbon isotope (513Cphytane) record at Demerara Rise shows a clear 7 %o excursion across the OAE-2 interval (Fig. 2b). The 513Cphytaiie excursion fol lows the same trend as the bulk S13Cxoc record, except that 5 13C t o c values are about 2 % o higher than the 513Cphytane values ( Fig. 3a). Prior to the OAE-2 interval, 513Cphytane val ues equal approximately -31 %o. At the onset of the OAE-2 interval, 513Cphytaiie values rapidly rise to -2 6 %o, maintain ing this value until 4 2 6 .2 mcd (Fig. 2b). Here a sudden drop of more than 3 % o to -29 %o is observed, although limited to one data point. After this, values increase again to around -2 6% o subsequently dropping to about -28 %o. Due to the low OM content, 513Cphytane could not be measured in the in tercalated carbonate layer between 425.96 and 425.57 mcd. Above this carbonate layer 513Cphytaiie values remain con stant, at values around -24 %o until 425 mcd. Above the OAE-2 interval, within the so-called phase C (cf. Kuypers et al., 2002a;Forster et al., 2007;Tsikos et al., 2004) á 13Cphytane values gradually return to near pre-excursion val ues. At Demerara Rise Site 1260, the bulk OM S13C (S13Cxoc) record shows a positive excursion of 6 .6 %o (Forster et al., 2007) during the OAE-2 interval. The predominantly ma rine source of the OM and its low thermal maturity indicate that the rapid fluctuations observed in the S13Cxoc record during the onset phase of the OAE-2 interval (Fig. 2a) are most likely not caused by changes in OM preservation or thermal maturity but by fluctuating inputs of terrestrial OM or, more likely, changes in the composition of marine OM at this location. Carbohydrates and proteins are for instance typically enriched in 13C relative to lipids. Such variability could overprint the S13Cxoc record (van Kaam-Peters et al., 1998;Sinninghe Damsté et al., 2002). Compound specific isotope records, however, are unaffected by changes in the composition of OM, and provide a more accurate represen tation of the true amplitude of the isotopic excursion during the OAE-2 interval (Fig. 2;Kuypers et al., 2002Kuypers et al., , 2004. At Demerara Rise, most biomarkers were sequestered in the sediment in macromolecular aggregates through incor poration of inorganic sulfur species during early diagene sis (cf. Brasseli et al., 1986;Sinninghe Damsté et al., 1989) as demonstrated by the high yield after desulfurization. To measure the carbon isotopic value of these biomarkers, they were released by Raney Nickel desulfurization. In this way, amongst other compounds S-bound phytane was recovered. Since S-bound phytane is derived from marine photosyn thetic algae and cyanobacteria (Koopmans et al., 1999), its stable carbon isotopic composition (513Cphytane) represents the weighted average of S13C of marine primary producers and is not influenced by fluctuating inputs of terrestrial OM or changes in the composition of marine OM. Preservation of isotopically heavy carbohydrates through sulfurization will result in more positive S13Cxoc than 513Cphytane values (Sin ninghe Damsté et al., 1998). The observed 2 % o offset be tween 5 13C x o c and 5 13Cphytane (Fig. 3) can hence be ex plained by the isotopic heterogeneity of marine OM (Hayes, 1993;Schouten et al., 1998).

Latitudinal variations in the amplitude o f the carbon isotope excursion
High atmospheric pCÖ2 levels during the Cretaceous re sulted in stronger fractionation between DIC and marine OM compared to today (Arthur et al., 1985a). Consequently, Cretaceous OM is overall depleted in 513 C by 4-5 %o com pared to present-day OM. Superimposed on this S13C off set a positive excursion is observed during OAE-2. The observed amplitude of the positive 5 1 3 C p h y ta n e excursion at Site 1260 is large in comparison to other known OAE-2 5 1 3 C p h y ta n e records. Based on carbonate S 1 3 C records, ap proximately 2.5 %o of the OAE-2 excursion has been inter preted as the result of enhanced global OM burial, shifting the 5 13 C of the global carbon reservoir towards more positive values (Kuypers et al., 2002b;Schlanger et al., 1987;Jenkyns et al., 1994;Arthur et al., 1984;Tsikos et al., 2004;Bowman and Bralower, 2005). The remainder of the 6.6 %o excursion at Demerara Rise must, therefore, be due to other processes. Possible causes include changes in [C02(aq)], changes in the S 1 3 C of the local inorganic carbon pool, changes in inorganic carbon spéciation, changes in temperature and changes in marine productivity, which influence phytoplankton growth rate and dimension (Hayes, 1993(Hayes, , 2001Takahashi et al., 1991).
Carbon isotopic values of S-bound phytane from sites at four different latitudes (ODP Site 1260, this study; sites 367 and 603B, Kuypers et al., 2002b; S57 Core, Tarfaya Basin, Tsikos et al., 2004, see Fig. 1 for paleo-locations) show that the amplitude of the 5 1 3 C p h y ta n e OAE-2 excursion in the proto-North Atlantic decreases towards higher lati tudes (Fig. 4). Prior to the OAE-2 interval, 513Cphytaiie values at the different sites are rather similar (Fig. 4), with slightly more positive values at Tarfaya and more negative values at Site 603B. During the OAE-2 excursion, all 5 1 3 C ph y ta n e val ues increased, but all to a different extent. Consequently, ¿>13Cphytaiie decreases by about 3 %o from the equator to 30° N during OAE-2 (Fig. 4). With a global enrichment in the iso topic composition of the DIC reservoir of about 2.5 %o (indi cated by the dark grey arrow in Fig. 4), the latitudinal offset between the different sites has to be explained by one of the possible additional effects mentioned previously, influencing the carbon isotopic fractionation during algal photosynthesis.
Marine plankton in present-day oceans shows a similar, albeit more modest, decrease in S13C values with increas ing latitude (Goericke and Fty, 1994;Rau et al., 1982Rau et al., , 1989 (Fig. 4). To compare present day 5 13C t o c values with OAE-2 S13Cphytaiie values the trend has to be offset by E 5 % o (4 % o more negative for the offset between total or ganic carbon and phytol, Schouten et al., 1998) and 2.5 %o to wards more positive values due to the difference in the global DIC reservoir (Arthur et al., 1985b). The recent gradient in

Reconstruction of changes in [CO2 (aq)] versus time and latitude within the proto-North Atlantic
Based on 513Cphytaiie records [C02(aq)] can be calculated using reconstructed SSTs, estimates for S13C of DIC, and a factor related to primary productivity (b ) (e.g., Sinninghe Damsté et al., 2008;Bice et al., 2006;Freeman and Hayes, 1992;Jasper et al., 1994). Following this approach, a theo retical gradient in [C0 2 (aq)] was here reconstructed by cal culating [C0 2 (aq)] for four different locations in the north ern proto-Atlantic ( (1) prior to OAE-2 with no latitudinal temperature gradient ( Fig. 5b: "pre-OAE-2" -green line), (2) during the Plenus Cold Event, within OAE-2 when there was a temperature gradient ( Fig. 5b: "cooling" -light blue line) and, (3) dur ing the maximum isotopic excursion within the OAE-2 inter val when there was again no temperature gradient ( Fig. 5b: "plateau" -dark blue line). Assuming a general 4 % o offset (AS) between phytol and biomass , 513Cphytane values from the four different sites (Table 1) were used to estimate the iso topic composition of primary photosynthetic carbon (Sp) : The isotopic composition of C02(aq) in the photic zone (5¿) can be calculated from the stable carbon isotopic composi tion of planktonic foraminifera. Since biogenic carbonates were poorly preserved in the OAE-2 sediments, the aver age isotopic value of foraminifera from just below the OAE-2 interval at ODP Site 1260 was used (data from Moriya Table 1. Sum m ary o f data from the four discussed sites. Productivity estim ates based on TO C contents and m ass accum ulation rates o f organic carbon (M ARcorg) in g m -2 y r-1 (Forster et al., 2008;van B entum et al., 2009)  es(fl) = 24.12 -9 8 6 6 /E Based on the 2-2.5 %o global carbon isotope excursion in bulk carbonate (Hayes et al., 1989;Wilson et al., 2002;Tsikos et al., 2004;Jarvis et al., 2006) we assumed DIC to have been enriched by 2 %o during OAE-2. Although 2 %o is only an estimate, a different value for DIC would cause an overall shift of the reconstructions, rather than affecting differences between sites and through time. Calculating sen sitivity of the equations to the input variables temperature, foraminiferal S13C and biomarker S13C demonstrated that foraminiferal S13C values overall do not appreciably impact With e/ being the maximum isotopic fractionation associ ated with the photosynthetic fixation of carbon, which is 2 5 %o in the case of algae (Bidigare et al., 1997). Param eter b is related to productivity and depends on growth rate and cell dimensions (Bidigare et al., 1997;Popp et al., 1998).
Although at high (>3000 ppm) pCC>2 values ep has a lim ited sensitivity, the 6 %o range in sp used for our calculations (gray rectangle Fig. 6) results in a rather robust estimate of pCO2 changes. Hence, we calculated three different scenar ios that could potentially explain the observed latitudinal dif ferences in the 513Cphytane records by changing b and SST.

The "constant" productivity scenario (I)
To assess the impact of productivity changes across the North  (Jefferies, 1962;Voigt et al., 2004;Gale and Christensen, 1996;Forster et al., 2007). At this time, temperatures were cooler in the north (Site 1276) than at the equator and a latitudinal temperature gradient was established (Sinninghe Damsté et al., 2010). During the later warmer episodes of OAE-2, the SSTs gradient was absent again in this part of the proto-North Atlantic. These temper ature gradients can be used be used to estimate SSTs for the two other sites (Fig. 5b) Hetzel et al., 2009). In this scenario (I), we applied this rather high ¿-value to all four sites (Fig. 5a).

Biogeosciences
Using this b value of 1 7 0 % o ¿iM and the reconstructed temperature records [CÜ2 (aq)] was calculated (Fig. 5a-c). The latitudinal [CÜ2 (aq)] shows a strong increase towards higher latitudes during the OAE-2 interval, which is needed to explain the observed latitudinal 513 C offset (both during the plateau and the cooling phase, blue lines) (Fig. 5c) , 2005). Clearly latitudi nal differences in sea surface productivity must have played a major role in shaping the S13C phytane gradients and it is not realistic to assume productivity was similar for all four sites.

The "CO2 equilibrium " scenario (II)
The small differences in [C02(aq)] today across the latitu dinal transect studied (see grey area in Fig. 5c), indicates that we may realistically assume atmospheric pCOi to be in equilibrium with the surface water. This was probably even more so during OAE-2, when surface waters were probably strongly stratified (van Bentum et al., 2009 Tsikos et al., 2004;Site 603B, Sinninghe Damsté et al., 2008). It seems therefore that the offset in 513Cphytane prior to the OAE-2 in terval at Tarfaya (Fig. 4) was mainly due to the higher pro ductivity at that time when compared to the other sites.

The "increasing productivity" scenario (III)
Scenario II assumed that productivity remained constant at Demerara Rise across the OAE-2 interval. However, ev idence from nannofossils (Hardas and Mutterlose, 2007) points to enhanced productivity during OAE-2. Changes in productivity at Demerara Rise would imply that productivity at the other sites must have changed as well in order to ex plain the S13C trends versus time and latitude as observed in Fig. 4. In scenario III (Fig. 5g, h and i), we used the same SSTs as in scenarios I and II but assumed now that factor b at Demer ara Rise increased from 170 to 220 at the onset of the OAE-2 (Fig. 5g). The latitudinal changes in 513Cphytaiie were subse quently used to calculate b at the other 3 sites. This would imply that b increased at most sites during OAE-2 (Fig. 5g).
This scenario is in line with most existing reconstructions, as enhanced productivity is often seen as an important cause of the increased OM burial during OAE-2 (e.g. Kuypers et al., 2002b). Still, while productivity increased during OAE-2 over most of the proto-North Atlantic, this scenario sug gests that productivity at Tarfaya remained similar or even decreased somewhat. In general TOC and the productivity pattern reconstructed using the 13C values agree, the excep tion is Tarfaya, where TOC values increase over the C/T OAE interval (Tsikos et al., 2004). Still, TOC is not a direct proxy for productivity, since TOC is also effected by changes in preservation. An important observation when comparing scenarios II and III is to what degree changes in productivity impact the reconstruction of pCOi-W hereas scenario II, with produc tivity kept constant at Demerara Rise, suggests a decrease in [C02(aq)] between pre-OAE-2 and during the OAE-2 plateau phase of 2 0 p m o ll_1 (Fig. 5f), this difference is only 1 5 p m o ll_1 in scenario III (Fig. 5i), with b increasing from 170 to 220. This provides an uncertainty envelope for calcu lating downcore changes in p CO2 These reconstructions show that before OAE-2 productiv ity was probably higher at Site 367 and at Tarfaya than at Sites 1260 and 603B. During OAE-2 productivity increased in most of the proto-North Atlantic, resulting in more com parable productivity within the proto-North Atlantic. interval and then a steady decline (Fig. 7), with two super imposed spikes of decreased p CO2 concentrations. These results are similar to the p CO2 reconstruction of Barclay et al. (2010), which shows two intervals of enhanced OM burial and associated pCOi decreases during OAE-2. At the on set of the OAE-2 period a temporal offset is observed be tween the atmospheric CO2 increase and the subsequent rise in SSTs (Fig. 8). A similar, but reverse offset is evident for decreasing CO2 concentrations and SSTs during the Plenus Cold Event. Part of this offset may be due to a time-lag be tween the start of magmatic activity versus the onset of OAE-2 and the increased SSTs. Turgeon and Creaser (2008) pos tulated a time-lag of at least 9 up to 23 kyr. between the first evidence for magmatic activity and the onset of OAE-2 at ODP site 1260. Such large time-lags are not consistent with the results of OAE-2 ocean-climate modeling (Flögel et al., 2011) and are much longer than the time-lag observed in to day's ocean-climate system. Probably, some part of the large temporal offset can be explained by the fact that we are com paring aquatic temperatures to atmospheric CO2 concentra tions in this study, while atmospheric temperatures are used in the ocean-climate model of Flögel et al. (2011). In our reconstruction, local SST records, which are prone to be af fected by additional factors other than atmospheric CO2 con centration, are compared with atmospheric global CO2 sig nals. In addition, assumptions that are necessary for any kind of paleo-oceanic climate modeling are difficult to be con strained within the setting of the recent state of the earth's climate system (namely response times of the global oceanclimate system or "climate sensitivity'': see review by Zeebe, 2011). Additionally, currently available climate models may not be suited to resolve time-durations of just a few kyrs asso ciated with abrupt climate changes in the past (Valdes, 2011). Nevertheless our data shows that there are clear indications that higher CO2 levels were inducing higher SSTs (e.g. dur ing volcanic outgassing) and that lower CO2 levels (due to burial of OM) resulted in lower SSTs. The tentative increase in atmospheric/?C02 at the start of the OAE-2 interval (Fig. 8) corresponds with an increase in osmium and zinc concentrations (Turgeon and Creaser, 2008;van Bentum et al., 2009) in the Demerara Rise sedi ments. Turgeon and Creaser (2008) show that the two ob served pulses in Os concentrations coincide with low Os iso tope values. This, together with the similar Os pattern in two different sections, implies that the raised Os concentrations are probably due to enhanced volcanic activity. Increases in osmium and zinc concentrations are probably related to magmatic activity, and are therefore possible evidence for a pulse of magmatic activity at the start of OAE-2 (Turgeon and Creaser, 2008). Enhanced magmatic activity would re sult in an increase in CO2 and since CO2 is a greenhouse gas, this increase could be responsible for the raised SSTs at the start of the OAE-2 interval (Fig. 8). Using marine stron tium isotope ratios Frija and Parente (2008) demonstrated that the increased temperatures at the onset of OAE-2 re sulted in enhanced continental weathering. Although Frija and Parente explain part of the strong positive strontium iso topes shift with increased ocean stratification, a recent study using stable Ca isotopes and modeling (Blättler et al., 2011) demonstrated that a three fold increase in weathering rates at the onset of OAE-2 would explain the total magnitude of the Sr isotope shift. The increased weathering would result in the enhanced drawdown of carbon (Walker et al., 1981), acting as a negative feedback (Fig. 9). At the same time enhanced weathering would increase nutrient input into the ocean, which probably resulted in a more efficient car bon pump. High S-bound isorenieratane concentrations (van Bentum et al., 2009) during this warmer period reveal that stratification was strong, which could be the result of the high SSTs. The increased stratification could also have decreased CO2 outgassing from the deep ocean to the atmosphere (Toggweiler, 1999).

Biogeosciences
After a period of decreased p CO2 , still during the OAE-2 interval, the constant, warm temperatures became cooler and started to fluctuate (late phase A, Fig. 8c). This period of cooling is likely the equivalent of the Plenus cold event ob served in NW Europe (Jefferies, 1962;Voigt et al., 2004;Gale and Christensen, 1996). This cooling has previously been attributed to a drop in atmospheric p CO2 levels, which in turn was caused by enhanced carbon sequestration by OM burial (Hasegawa, 2003;Forster et al., 2007;Gale and Chris tensen, 1996). The cooling lead to a stronger latitudinal temperature gradient in the proto-North Atlantic (Sinninghe Damsté et al., 2010) and as a result, the colder waters at higher latitudes could have taken up more CO2 . A drop in isorenieratane concentrations (van Bentum et al., 2009) (Fig. 8) and a possible benthic foraminifer repopulation event (Friedrich et al., 2006) reveal that stratification at Site 1260 did indeed decrease during this cooler interval. At the start of phase B, another rise in osmium concentra tions is followed by an increase in p CO2 . The higher atmo spheric pCÖ2 again could explain the subsequent tempera ture increase (lower part of phase B, Forster et al., 2007). It appears that the increase in volcanic activity counterbal ances the episode of cooling during OAE-2 (see Turgeon and Creaser, 2008;van Bentum et al., 2009;Snow et al., 2005). Changes in nannofossil assemblages indicate higher produc tivity at Demerara Rise at this time (Hardas and Mutterlose, 2007). This increase in productivity could be related to en hanced continental weathering ( isorenieratane concentrations during phase B does reveal a decrease in stratification at this time (Fig. 8).
The carbonate rich layer found at 425 mcd has been in terpreted as an ash layer (Hetzel et al., 2009). Therefore, it seems likely that an additional, second magmatic pulse oc curred at this time, raising atmosphericpCÖ2 -During this warmer period, the latitudinal temperature gradient was min imized again and higher SSTs apparently intensified oceanic stratification, as both isorenieratane and chlorobactane con centrations increased again at the end of the OAE-2 period (upper part of phase B, van Bentum et al., 2009).
The reconstructed p CO2 record increases after termina tion of the OAE-2 event. This increase is, however, not matched by increasing TEXs6-based SSTs. This could either be due to a local effect, increased upwelling after OAE-2 for example might have prevented SSTs at Demerara Rise from increasing. Another possible explanation for this observa tion could be due to the fact that temperature sensitivity to pCÖ2 decreases at higher levels of p CO2 (Fig. 6). In addi tion, Jarvis et al. (2011) observed that the elevated global sea level during the early Turanian could have decreased conti nental weathering rates, which would lead to a decrease in oceanic nutrient supplies and bioproductivity, resulting in an increase in p CO2 .
Climate during the Cretaceous seems to have responded quite rapidly to disequilibria in carbon cycling. Increased magmatic CO2 outgassing resulted in an overall warmer cli mate, which enhanced oceanic stratification, ocean anoxia and associated OM preservation (Fig. 9). At the same time, increased weathering due to high atmospheric p CO2 re sulted in enhanced nutrient input into the proto-North At lantic, which in turn increased productivity. Furthermore, primary productivity might have removed carbon more ef ficiently from the atmosphere under high pCÖ2 conditions, as suggested by mesocosm experiments showing that under higher p CO2 levels more inorganic carbon was removed us ing less nutrients (Riebesell et al., 2007). The widespread en hanced OM burial during OAE-2 withdrew CO2 from the at mosphere, cooling the Earth in the process (Bralower, 2008;Snow et al., 2005). At the same time, the increased ocean stratification could have decreased CO2 outgassing from the deep ocean to the atmosphere (Toggweiler, 1999). This im plies that OAEs acted as a global negative feedback mech anism in response to massive CO2 inputs (e.g. Turgeon and Creaser, 2008;Barclay et al., 2010).

Conclusions
The observed positive isotope excursion of phytane (~7 %o) at Demerara Rise is unusually large compared to other C/T phytane records (3-6 %o) from locations in the proto-North Atlantic. Using reconstructed SST gradients we demonstrate that before OAE-2 productivity was probably higher at Site 367 and at Tarfaya than at Sites 1260 and 603B. During OAE-2 productivity increased in most of the proto-North At lantic, resulting in more comparable productivity between the sites in the proto-North Atlantic.
Magmatic activity, atmospheric p CO2 and temperature during OAE-2 are linked through both positive and negative feedback mechanisms. Enhanced magmatic episodes seem to have raised pCÖ2 , increasing global temperatures. Higher SSTs and stratification, together with enhanced nutrient input due to more intense weathering, resulted in a more efficient carbon pump as OM burial increased. Organic matter burial lowered p CO2 again, cooling the greenhouse climate. When at the end of OAE-2 carbon burial rates were reduced, p CO2 increased again. This implies that Cretaceous climate was sensitive to small changes in the (internal) feedbacks in the global carbon cycle. A lkenone <513C as a proxy for past C O 2 in surface w aters: R e sults from the Late Q uaternary A ngola C urrent, in: U se o f P rox ies in P aleoceanography: E xam ples From the S outh A tlantic, edited by: Fischer, G. and W efer, G., S pringer N ew York, 4 6 9 -488, 1999. A rthur, M ., Glancy, T., Fiodell, D., Schlanger, S. 0 ., Szak, C., Zachos, J., and A nonym ous: T he C enom anian/ T uronian (M id-Cretaceous) "oceanic anoxic e v en t" as observed in northw estern E urope,