Cold climate during the closest Stage 11 analog to recent Millennia

Transcription

1 Quaternary Science Reviews ] (]]]]) ]]] ]]] Cold climate during the closest analog to recent Millennia William F. Ruddiman Department of Environmental Sciences, McCormick Road, University of Virginia, Clark Hall, Charlottesville, VA 2294, USA Received 23January 24; accepted 8 October 24 Abstract The early part of marine isotopic near 4, years ago provides the closest analog to Holocene insolation levels of any interglaciation during the era of strong 1,-year climatic cycles. The CH 4 concentration measured in Vostok ice fell to 45 ppb, and CO 2 values to 25 ppm. These natural decreases contrast with the increases in recent millennia and support the early anthropogenic hypothesis of major gas emissions from late-holocene farming. During the same interval, dd values fell from typical interglacial to nearly glacial values, indicating a major cooling in Antarctica early in. Other evidence suggests that new ice was accumulating during the closest insolation analog to the present day: a major increase in d 18 O atm at Vostok, a similar increase in marine d 18 O values, and re-initiation of ice rafting in the Nordic Sea. The evidence permits extended (42, year) intervals of interglacial warmth in the Antarctic and North Atlantic, yet it also requires that this warmth ended and a new glacial era began when insolation was most similar to recent millennia. The Holocene CO 2 anomaly was produced only in part by direct anthropogenic emissions; over half of the anomaly resulted from the failure of CO 2 values to fall as they had during previous interglaciations because of natural responses, including a sea-ice advance in the Antarctic and ice-sheet growth in the northern hemisphere. r 24 Elsevier Ltd. All rights reserved. 1. Introduction Ruddiman (23a) proposed that early farming produced enough CO 2 and CH 4 during the millennia prior to the industrial era to boost gas concentrations in the atmosphere and counter the natural decreases that would otherwise have occurred. Natural CO 2 and CH 4 trends for the late Holocene were estimated from those in early interglacial marine isotopic Stages 5, 7, and 9, when concentrations in Vostok ice reached late-deglacial maxima and then fell for more than 1, years early in each interglaciation (Petit et al., 1999). During the most recent deglaciation, CH 4 and CO 2 concentrations also reached a late-deglacial peak and began similar decreases, but gas levels then rose rather than falling during recent millennia (Fig. 1). The total gas anomalies reached just before the start of the industrial era the Tel.: ; fax: address: wfr5c@virginia.edu (W.F. Ruddiman). difference between the observed rises and the predicted drops were estimated at 4 ppm for CO 2 and 25 ppb for methane. The CH 4 anomaly was attributed mainly to irrigation for rice, and the CO 2 anomaly to clearance of forests. Another part of the early anthropogenic hypothesis was the inference that these early emissions averted a natural global cooling that would otherwise have initiated a new glacial cycle in the northern hemisphere (Fig. 2). This conclusion was based in part on an energy-balance model study showing that northeast Canada has been close to a state of glacial inception in recent centuries (Williams, 1978)and that a relatively small cooling would be needed to start a new glacial cycle in North America or Eurasia. Additional support came from ice-volume trends simulated by a numerical model that estimated time-dependent d 18 O responses to summer insolation changes in the northern hemisphere (Imbrie and Imbrie, 198) /$ - see front matter r 24 Elsevier Ltd. All rights reserved. doi:1.116/j.quascirev

2 2 ARTICLE IN PRESS W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] Solar Radiation (Watts/m 2 ) (a) 29 Solar Radiation Observed CH 4 Natural 1, 5, CH 4 trend 25 ppb Methane (ppb) 65 N July Insolation (W/m 2 ) Years Future 1, 1, T max P max Stage 1 T max P max T min P min T min P min CO 2 (ppm) Natural CO 2 peak Observed CO 2 Colder Warmer , 5 Glaciation threshold Natural CO 2 trend 4 ppm 41, 4, 39, Fig. 3. Mid-July 651N insolation trends during the Holocene and in isotopic. The closest analog to modern values occurred 397, years ago (Berger and Loutre, 23). (b) Fig. 1. Observed late-holocene increases in concentrations of (a) CH 4 (Blunier et al. 1995) and (b) CO 2 (Indermuhle et al., 1999) compared to predicted natural decreases from the early anthropogenic hypothesis (Ruddiman, 23a). The estimated sizes of the anthropogenic CH 4 and CO 2 anomalies are the differences between the observed and the predicted trends. Natural Solar radiation and Greenhouse-gas Trend Observed climate Anthropogenic Greenhouse Gases Fig. 2. According to the early anthropogenic hypothesis, a large natural Holocene cooling would have initiated a glaciation within the last several thousand years, but anthropogenic greenhouse-gas emissions countered enough of the cooling to prevent a glacial inception. The choice of the first part of marine isotopic Stages 5, 7, and 9 as analogs to the Holocene can be questioned. Eccentricity modulates precession at the 413,-year period, and insolation variations during the current interglaciation have been smaller in amplitude than those during the three previous climatic cycles, although moving in the same direction. As a result, the early anthropogenic hypothesis can be challenged: the reduced amplitude of forcing during the Holocene might have resulted in natural climatic responses smaller than those during the last three interglaciations. Isotopic is a better analog to the Holocene. Because of comparably low eccentricity at that time, insolation variations were smaller in amplitude and closer to modern values (Fig. 3). The analog is not perfect: the tilt and July precession maxima were out of phase on the Stage 12/11 deglaciation (termination V) but were very nearly in phase on the Stage 2/1 deglaciation, while the subsequent insolation minima at tilt and precession were nearly in phase during Stage 11 but are out of phase now. Still, is the closest available analog to Holocene insolation levels within the interval of dominant 1,-year climatic cycles. One method used to compare Stages 1 and 11 is to determine the length of the interglaciation and use it to forecast the duration of the current interglaciation. This approach has led to a widespread view that was an unusually long interglaciation, and that by implication the current interglaciation could last many thousand of years into the future (Burckle, 1993; EPICA Community Members, 24). Some simulations of ice volume with zonal climatic models even show an ice-free interval lasting for tens of thousands of years (Loutre and Berger, 2). Although these results seem to refute the overdue-glaciation hypothesis, Sections 2 5 of this paper will show that such a conclusion is unjustified. The first issue considered here is the response of the climatic system during the part of that provides the closest insolation analog to the recent millennia of

3 W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] 3 the Holocene (Berger and Loutre, 23). During this closest-analog interval, spanning the millennia just prior to 397, years ago, July 651N summer insolation values fell to a minimum very near the present-day level (Fig. 3). Incoming July radiation was 14 W/m 2 below the modern level, compared to values 2 35 W/m 2 lower during the most similar parts of interglacial isotopic Stages 5, 7, and 9. The behavior of the climate system during this closestanalog portion of provides a stringent test of the early anthropogenic hypothesis. If the trends during this time match those during Stage 1, it will be difficult to argue that the Holocene trends are anomalous compared to the natural behavior of the climate system, and the early anthropogenic hypothesis will be refuted. But if the trends show major reductions in greenhouse gases and a substantial climatic cooling as predicted in Figs. 1 and 2, the hypothesis that the Holocene trends are anthropogenic rather than natural in origin will be supported. CH 4 (ppb) (a) Temperature ( C) 1, , 2-2 Years ago Stage I Pre-industrial Predicted 45, 4, Years ago Stage 1 Years ago 1, Stage 1 Pre-industrial Predicted 45, 4, (b) Years ago 1, Stage CO 2 (ppm).25 δ 18 atm (% ) 2. Comparison of Stages 11 and 1: Vostok ice The occurrence of typical full-interglacial values of CH 4,CO 2, dd, and d 18 O atm in the lower tens of meters of Vostok ice indicates that it reaches deep into. Petit et al. (1999) pinned the lower portion of their GT4 time scale at Vostok by correlating a feature near 3265 m ice depth to marine isotopic substage of Bassinot et al. (1994), who had refined the earlier SPECMAP time scale of Imbrie et al. (1984). This choice linked the first major shift of climatic proxy signals in Vostok ice from full-interglacial (substage 11.3) values to partly glacial (substage 11.24) values with the equivalent transition in marine d 18 O records. The Petit et al. (1999) choice was later verified by two independent methods. Shackleton (2) tuned the Vostok d 18 O atm record to northern hemisphere autumn (early September) insolation, and Bender (22) correlated O 2 /N 2 variations in Vostok ice to changes in summer (December 21) insolation at 781S. These time scales confirmed that the interval of Vostok ice showing the first major shift of proxy signals away from full interglacial values dates from 4, to 39, years ago, the interval that includes the closest insolation analog to the late Holocene and the near future (Fig. 3). The time scale used to plot the Vostok records in Fig. 4 uses two minor refinements of these time scales based on separating the proxy climatic signals in Vostok ice into early and late responders. A key early responder is atmospheric methane, which varies in phase with July insolation at the precession cycle, as shown by the 1,5-year age of the last CH 4 maximum in annually layered GRIP ice (Blunier et al., 1995). This July phasing was used to tune the CH 4 time (c) , 4, Years ago , 4, 395, (d) Years ago Fig. 4. Climatic proxy trends in Vostok ice (Petit et al., 1999) during the Holocene and the analogous interval early in isotopic. scale at Vostok back to 34, years ago by Ruddiman and Raymo (23). Using this same rationale, the prominent early methane minimum during in Vostok ice falls at 397, years ago, the age of a significant precession insolation minimum. Other proxies like CO 2 and dd that have long been considered early responders (Lorius, et al., 1985; Jouzel et al., 1987; Waelbroeck et al., 1995) also reach minimum values at or near the level of this CH 4 minimum. The classic late responder in the climate system is northern hemisphere ice volume, which lags 5 years behind northern hemisphere summer insolation (Milankovitch, 1941; Hays et al., 1976; Imbrie et al., 1984). The proxy in Antarctic ice cores tied most closely to northern hemisphere ice volume is d 18 O atm, which records changes in average seawater d 18 O (a measure of ice volume) and in monsoon-driven biomass (Bender et al., 1994). During climatic transitions, the slow ice-volume component embedded in the d 18 O atm signal should cause it to lag behind the faster-responding parts of the climate system. The transition from the substage 11.3 interglacial peak into substage at Vostok (Petit et al., 1999) shows the expected lag: maximum d 18 O atm values are reached a few 1 years later than minima in CH 4,CO 2 and dd.

4 4 ARTICLE IN PRESS W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] Bassinot et al. (1994) dated the substage d 18 O maximum to 39, years ago, consistent with the SPECMAP time scale (Imbrie et al., 1984). Petit et al. (1999) then used this age estimate to anchor the GT4 time scale for Vostok ice. Later studies have suggested that the SPECMAP time scale is too old by an average of 15 2 years (Pisias et al., 199; Shackleton, 2; Ruddiman and Raymo, 23). Allowing for this adjustment, the likely age of the substage feature is probably nearer 392, years. Based on the above considerations, the time scale used to create Fig. 4 places the CH 4 minimum at 397, years ago and the d 18 O atm maximum at 392, years ago. Ages are linearly interpolated between these levels, and ages older than 397, years ago are derived by aligning the lowermost region of high CH 4 values at Vostok with the precession insolation maximum 48, years ago, using the method of Ruddiman and Raymo (23). For the key interval between 4, and 392, years ago, these various time scales disagree by no more than a few thousand years. The greenhouse-gas concentrations at 397, years ago, the time of the closest insolation analog to modern levels, confirm the predictions of the early anthropogenic hypothesis from Fig. 1: CH 4 values fell to 445 ppb (Fig. 4a), just below the predicted value of 45 ppb, and CO 2 values fell to 25 ppm (Fig. 4b), just above the predicted range of ppm. Compared to the measured concentrations actually reached prior to the industrial era (282 ppm for CO 2 and 68 ppb for CH 4 ), the values indicate that late-holocene concentrations were anomalously high by more than 3 ppm for CO 2 and 235 ppb for methane, close to the size of the anomalies predicted by the hypothesis. Other proxy climatic signals in Vostok ice record major shifts toward glacial conditions at high southern latitudes. The dd signal decreased from typically interglacial values to 75% of full-glacial values, and estimates of Antarctic temperature derived from the dd signal dropped to near-glacial levels (Fig. 4c). In addition, the Na + (sea-salt ion) concentration increased halfway from interglacial to glacial values (Petit et al., 1999). These observations indicate that shortly after 4, years ago the Antarctic region had shifted toward a cold glacial state in an insolation regime similar to that today and at greenhouse-gas levels very close to those that should prevail today in the absence of anthropogenic overprints. In contrast, the Holocene dd (and Na + ) signals at Vostok drifted only slightly toward glacial values (Fig. 4c). The failure of these proxies to shift farther toward glacial values during the Holocene suggests that today s Antarctic climate is anomalously warm compared to the natural trends observed during the most analogous part of. The d 18 O trend in air from Vostok ice can be used to test the prediction that a late-holocene glacial cycle is overdue in the northern hemisphere (Fig. 2). As noted earlier, the d 18 O atm signal in ice is controlled both by sea-water d 18 O (ice volume) and biomass changes linked to tropical monsoons (Bender et al., 1994). The average glacial interglacial amplitude of the d 18 O atm signal for the last several climatic cycles at Vostok is 1.5%. An estimated 1.5% of the 1.5% change at the Stage 2/1 deglaciation can be attributed to changes in global ice volume (Duplessy et al., 22; Schrag et al., 22), leaving the remaining.45% as a monsoonrelated biomass contribution. Between substages 11.3 and 11.24, d 18 O atm values increased by.65% (Fig. 4d). Subtracting the (estimated maximum) biomass component of.45% from.65% leaves a residual d 18 O atm increase of.2% that can plausibly be attributed to ice growth. This result suggests that new ice was growing in the northern hemisphere during and after the closest analog to the late Holocene. If the Dole effect were assumed to have been at less than maximum strength during the substage 11.3/11.24 transition, a larger ice-volume residual would be required. 3. Comparison of Stages 11 and 1: EDC ice EPICA Community Members (24) published preliminary results from the EDC ice core extending back 74, years. The dd and dust-concentration analyses span all of, while the published CO 2 and CH 4 analyses cover only the Stage 12/11 transition and the early part of. The EPICA group concluded that warmer interglacial temperatures in Antarctica during lasted for 28, years. Because only 12, years of similar interglacial conditions have elapsed in Stage 1, they concluded that the current interglacial warmth will persist for 16, years more, in disagreement with the overdue-glaciation hypothesis. On the other hand, the major drop in dd values in the EDC record that marks the end of this interval of Antarctic warmth began before 4, years ago and was nearly complete by 39, years ago, a timing that is close to that observed at Vostok (Fig. 4c), given the considerable dating uncertainties at EDC. As a result, the EDC dd record supports the evidence in Section 2 that peak interglacial warmth had ended by 397, years ago, the time of the closest insolation analog to the present day. Both of these predictions obviously cannot be correct. Does the current interglaciation still have 16, years left to run, or should it have ended during the last few millennia? The resolution of this dilemma lies in the alignment used by EPICA to compare Stages 11 and 1. Their choice (Fig. 5 of their paper) places the present-day dd

5 W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] 5 65 N July Insolation (W/m 2 ) P min P max 41, Years Future 1, Stage 1 4, Fig. 5. The alignment of dd trends in EDC ice used by EPICA Community Members (24) positions the present-day 651N insolation minimum with a insolation maximum 49, years ago. value opposite the Stage-11 dd value at 49, years ago, but that selection results in the alignment of the modern-day 651N insolation minimum with a 651N insolation maximum in (as shown in Fig. 5 of this paper). This alignment is clearly incorrect by a very large amount. As Berger and Loutre (23) noted, the closest analog to the modern-day insolation minimum occurred 397, years ago (Fig. 3), well after the alignment chosen by EPICA. This error on EPICA s part resulted from their decision to align terminations V and I and to assume that the two interglaciations then developed in complete parallel (Fig. 6, top). This approach carries considerable risk. The timing of terminations is thought to be critically dependent on the close alignment of insolation maxima from both precession (which dominates the 651N insolation signal) and from obliquity (Imbrie et al., 1992; Raymo, 1997). Termination I, centered at 13, years ago, resulted from the nearly coincident insolation maxima for precession (11, years ago) and obliquity (1, years ago). In contrast, for the interval near termination V, the insolation maximum from the obliquity signal fell at 418, years ago, midway between insolation maxima from the June 21 precession signal at 429, and 48, years ago. SPECMAP (Imbrie et al., 1984) placed marine d 18 O termination V at 415, years ago near the obliquity insolation maximum, a choice that still remains uncertain. Because the relative forcing from obliquity and precession were so different on these two terminations, aligning the two interglacial stages at the terminations is risky. EPICA s attempt to do so produced the implausible misalignment of insolation trends shown in Fig. 5. The correct alignment of Stage 1 against the latter (rather than the earlier) part of (Fig. 6, bottom) allows for an unusually long (42, year) interval of warmth in Antarctica during, but at the same (A) EPICA (24) Ruddiman (25) (B) Stage I Interglaciation Stage II Interglaciation 26, years Stage II Interglaciation will last 16, more years Stage I Interglaciation has ended Fig. 6. Comparison of the Stages 1 and 11 alignments chosen by (A) EPICA Community Members (24) and (B) this paper (as well as Berger and Loutre, 23). The alignment in B is consistent with the 651N summer insolation trends (see Fig. 3); the alignment chosen by EPICA is not (see Fig. 5). time allows that warmth to come to an end during the time that was the closest insolation analog to the present day. This alignment supports the hypothesis that the Stage 1 peak of interglacial warmth in Antarctica should now be over, with a major cooling well underway. For the rest of this paper, the prominent cooling at the substage 11.3/11.24 transition is used as the boundary between peak-interglacial conditions in early and the late portion of that lasted until the /1 transition near 352, years ago (Imbrie et al., 1984). 4. Comparison of Stages 11 and 1: marine sediments The conclusion from the d 18 O atm evidence at Vostok that new ice sheets began growing at the substage 11.3/ transition is at odds with the interpretation that was a uniquely long interglaciation in which no new ice grew in North America and Eurasia (Burckle, 1993). In apparent support of this interpretation, a highresolution study of in the subpolar North Atlantic showed that the ocean surface remained warm for at least 3, years after termination V (McManus et al., 23). However, closer scrutiny shows that ice was not absent from North America and Eurasia for this entire interval. McManus et al. (23) actually stated that that the relatively ice-free portion of lasted longer than other peak interglacials. This careful wording permits a relatively early onset of new ice in. Another problem with the interpretation of a long icefree interglaciation is that North Atlantic sea-surface temperatures cannot be used as a sensitive indicator of northern hemisphere ice volume. Lagging warmth (and

6 6 ARTICLE IN PRESS W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] the absence of ice-rafted debris) often characterized the North Atlantic south of Iceland during intervals when ice sheets were growing on nearby landmasses (Ruddiman and McIntyre, 1979). During isotopic substage 5a, the subpolar North Atlantic was at or close to fullinterglacial warmth (Sancetta et al., 1972), even though sea level was 15 2 m below its modern position (Chapell and Shackleton, 1986; Bard et al., 199). The volume of ice required to drop sea level by 15 2 m was equivalent to the glacial-maximum Scandinavian ice sheet, yet the Atlantic was warm. Several marine proxies further support the Vostok d 18 O atm evidence that new ice appeared on land during the substage 11.3/11.24 transition. One line of evidence comes from marine d 18 O signals, which record changes in ice volume and variations in local ocean temperature and salinity. The maximum size of the temperature - salinity (T S) overprint at any location can be estimated from changes that occurred during the most recent deglaciation. As noted earlier, the ice-volume contribution to the d 18 O shift across termination I is estimated at 1.5% (Duplessy et al., 22; Schrag et al., 22). Any additional isotopic change on termination I beyond this amount can be attributed to the combined effects of local temperature and salinity, and this excess d 18 O change on termination I can be used to estimate the maximum T S overprint present elsewhere in the record, including isotopic. The tacit assumption in this method is that the temperature salinity overprint added to the d 18 O signal in is unlikely to have been larger than the amount removed during a full deglacial transition. Marine cores in the equatorial Atlantic (Tiedemann et al., 1994, Bickert et al., 1997) and tropical Indian Ocean (Bassinot et al., 1994) record d 18 O increases as large as 1.% during the substage 11.3/11.24 transition. Because d 18 O shifts during the last deglaciation in these regions were also very large (2%), indicating T S overprints of about 1%, it is just barely possible to avoid a requirement for ice growth by invoking the maximum T S overprint. Observed d 18 O changes in the eastern tropical Pacific do, however, require ice growth (Table 1). Benthic foraminiferal d 18 O signals from ODP sites 846 and 849 (Mix et al., 1995a, b) register a d 18 O change of 1.61% from Stage 2 to Stage 1, implying a maximum T S overprint of.567.1%. During the transition into substage 11.24, d 18 O values increased by.72.74%. Subtracting the maximum T/S overprint of.56% leaves a residual of.16.18% as the estimated ice-volume effect. This independent estimate is close to the.2% ice-volume residual estimated from the d 18 O atm trend at Vostok. If each.1% change in d 18 O represents 11 m of sea-level change, the growth of ice during substage would have amounted to 2 m of sealevel fall. This amount of ice growth would have accumulated by the substage peak at 392, years ago, a time equivalent to roughly 5 years from now according to the insolation trends in Fig. 3. More pertinent to the early anthropogenic hypothesis is the question of whether or not ice was already growing 397, years ago, the time when the insolation configuration was most analogous to the present day. Several lines of evidence indicate that it was. First, the d 18 O atm trend at Vostok had already increased by more than.55% by the time of the minima in CH 4, CO 2, and dd 397, years ago (Fig. 4). At least.1% of this d 18 O atm increase (the amount in excess of the assumed.45% biomass overprint) should represent increased ice volume. Second, Bauch et al. (2) dated the renewal of icerafted deposition in Nordic Sea sediments to 398, years ago. In order to calve icebergs to the ocean by that time, ice growth must have been underway in the north for several 1 years. Third, a sensitivity-test experiment with the GENESIS 2 climate model showed that lowering greenhouse-gas concentrations to the levels reached during the insolation analog to the present day resulted in year-round snow cover in parts of Baffin Island, Canada (Ruddiman et al., 25). Finally, it seems unlikely that 2 m of sea-levelequivalent ice volume could have accumulated by 392, years ago without any of the ice having grown by 397, years ago. In summary, marine evidence requires the growth of ice sheets equivalent to 2 m of sea level drop by Table 1 Estimate of the ice-volume component of the measured d 18 O increases (%) from substage 11.3to substage 11.24, based on subtracting the maximum temperature salinity (T S) overprint on the ice-volume component of the Stage 2/1 deglacial transition. Stage 2/1 transition Substage 11.3/11.24 transition ODP Measured Max. T S Measured Minimum site d 18 O increase contribution a d 18 O increase ice contribution ODP 846 b ODP 849 c a Assumed ice-volume contribution of 1.5% (Duplessy et al., 22; Schrag et al., 22). b Mix et al. (1995a). c Mix et al. (1995b).

7 W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] 7 substage (392, years ago). The evidence further indicates that some of this ice was growing by 397, years ago, the closest insolation analog to the present day. 5. Comparison of Stages 11 and 1: ice-volume simulations During the current July 651N insolation minimum, solar radiation values lie 2/3of the way toward the glacial extreme of the orbital range over the last several hundred thousand years, yet no new ice has appeared in response to this forcing. McManus et al. (23) termed this absence of new ice a skipped beat in orbital forcing of ice volume. This skipped beat creates a dilemma for modeling efforts that attempt to simulate ice-sheet growth and decay in response to orbital forcing. The dilemma is that 651N summer insolation will not fall as low as it is now for the next 55, years (Fig. 7a). With no ice today, and with higher summer insolation levels far into the future, there seems to be no reason to expect ice to grow for a very long time. Some simulations with timedependent ice-sheet models project no ice growth for that entire interval (Loutre and Berger, 2; Berger and Loutre, 23). Added to the 6 years of the Holocene without significant ice on North America and Eurasia, the current interglaciation would then last more than 6, years. But a 6,-year interglaciation would be without precedent in all 2.75 million years of the northern hemisphere ice age. Even during the smaller ice-volume fluctuations between 2.75 and.9 million years ago, ice sheets grew and melted mainly at a 41,-year cycle, with only about 2, years between glaciations. During the 1,-year cycles of the last.9 million years, ice sheets have been present in the northern hemisphere for 9% of the time, with interglaciations lasting an average of just 1, years (Shackleton, 1987). No interglaciation has ever lasted 6, years, and no obvious reason exists to think that the natural climate system has just entered such an unprecedented ice-free era. points the way to an answer to this dilemma. Insolation trends late in were very similar to those during the Holocene and the future (Fig. 7a). Not surprisingly, the same model simulations of Berger and Loutre (23) that produced no ice growth late in Stage 1 and for 55, years into the future also simulated no ice-sheet growth until the /1 boundary near 35, years ago (Fig. 7b). Yet the evidence from changes in Vostok d 18 O atm (Section 2) and in marine d 18 O and Nordic Sea ice rafting (Section 4) indicates that new ice was already growing before 392, years ago, in disagreement with the simulation shown in Fig. 7b. 65 N July Insolation (W/m 2 ) (a) Simulated Ice volume (1 6 Km 3 ) (b) Years in the Future 1, 2, 4, 6, 4, Stage II 38, 36, 34, Years in the Future 1, 2, 4, 6, 4, Stage I Stage I Stage II δ 18 evidence Stage II Ice Growth 38, 36, 34, Fig. 7. Model simulations of insolation forcing of ice volume. (a) Insolation trends at 651N during are similar to those during the Holocene and future. (b) Model simulations that do not initiate ice growth in the late Holocene also fail to simulate ice growth early in (Berger and Loutre, 23). Yet evidence cited in the text indicates that ice sheets were growing earlier in. On the other hand, ice-volume simulations that include CO 2 forcing based on the Vostok ice record have been able to match the evidence for ice growth earlier in, thus avoiding an unrealistically long ice-free interval (Berger and Loutre, 23). These results, combined with evidence in Sections 2 and 4, suggest a solution to the dilemma of the long ice-free Holocene and future. If the large anthropogenic overprints of CO 2 and CH 4 during the late Holocene were removed, thereby allowing natural gas concentrations to fall to levels like those early in, the resulting model simulations should produce results similar to, with early growth of ice and no implausibly long ice-free interval. This nucleus of late-holocene ice would be important to the progression of the next (current) glacial cycle. Growing ice sheets create positive feedbacks that favor additional glaciation, including albedo-temperature feedback, elevation-temperature feedback, orographic precipitation along south-facing slopes, and slow radial ice flow across unglaciated terrain (Andrews and Mahaffy, 1976). After ice first appeared during the transition into substage 11.24, ice growth appears to

8 8 ARTICLE IN PRESS W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] have been nearly continuous for the rest of and all of Stage 1, in part because of ice feedbacks on the climate system. The same process would presumably be underway now if natural forcing had prevailed. 6. Discussion Comparison of late-holocene climatic trends with those during the closest insolation analog supports the major predictions of the early anthropogenic hypothesis: CH 4 and CO 2 levels dropped to levels close to those predicted (Fig. 4a and b); the early responding Southern Hemisphere cooled dramatically (Fig. 4c); and ice sheets began to grow in the northern hemisphere (Fig. 4d). In contrast, during the middle and late Holocene, gas concentrations rose, south-polar climate cooled only slightly, and northern ice sheets did not grow. The cause of these very different Holocene trends is not likely to lie in natural factors. Not only were insolation trends similar during the two intervals (Fig. 3), but key terrestrial boundary conditions early in were nearly identical to those in the Holocene: northern ice sheets had melted, sea level was high, and vegetation had reached a full-interglacial state. Any differences in insolation and in terrestrial boundary conditions seem too small to account for the fundamentally different direction of the Holocene trends. If natural forcing can be ruled out, the anomalous Holocene gas increases have to be anthropogenic in origin. This conclusion is seemingly at odds with the analysis of Joos et al. (24), who concluded that direct anthropogenic emissions of CO 2 from forest clearance are insufficient to explain the proposed 4-ppm anthropogenic CO 2 anomaly. They estimated that 55 7 Gt C of direct CO 2 emissions from forest burning would be needed to explain a 4-ppm anomaly, well above the GtC calculated by Ruddiman (23a) based on Indermuhle et al. (1999), and even farther above their own estimate of 6 8 Gt C based on DeFries et al. (1999). Joos et al. (24) also noted that the small amplitude of the negative d 13 C trend in late- Holocene CO 2 found by Indermuhle et al. (1999) appears to limit terrestrial carbon emissions to amounts smaller than those needed to account for a 4-ppm CO 2 anomaly. These estimates of Gt C emissions are all based on carbon-cycle models that include both fast exchanges among surface carbon reservoirs as well as the slow uptake of CO 2 by the deep ocean and subsequent dissolution of seafloor CaCO 3. These findings seem irreconcilable. How can the Holocene CO 2 anomaly be entirely anthropogenic in origin if direct anthropogenic carbon emissions from humans are too small to account for it? The resolution to this impasse may lie in processes that contribute to the size of the CO 2 anomaly without requiring direct emissions of CO 2. Ruddiman (23a) defined the Holocene anthropogenic CO 2 anomaly as the difference between the observed CO 2 rise and the predicted CO 2 fall (Fig. 1b). The predicted drop in CO 2 was based on natural trends measured during the early parts of previous interglaciations, when concentrations fell by 3 5 ppm (Petit et al., 1999). The trend shown in Fig. 4b shows a natural drop from 282 to 25 ppm. Implicit in this definition of the Holocene CO 2 anomaly are two distinct components: (1) direct anthropogenic carbon emissions that caused CO 2 values to rise, and (2) natural factors that prevented CO 2 values from falling as they had in previous interglaciations. The resolution of the impasse noted above could lie in the second group of mechanisms. If the natural processes that drove CO 2 lower during previous interglaciations were overridden during the Holocene, the prevention of a natural drop in CO 2 would have added to the size of the CO 2 anomaly, but would not have required direct emissions of carbon. Such an explanation would be consistent with the claim that the entire CO 2 anomaly is anthropogenic in origin, yet it would not violate the carbon-budget and d 13 C constraints. In the concept proposed here (Fig. 8), direct emissions from human activities account for most of the 25-ppb CH 4 anomaly (Ruddiman and Thomson, 21), although the magnitudes are difficult to quantify. In the case of CO 2, direct emissions only account for a fraction of the estimated anomaly of 32 4 ppm. The part of the Holocene CO 2 anomaly that obviously requires direct anthropogenic emissions is the 14-ppm increase between the natural CO 2 peak of 268 ppm that occurred 1,5 years ago and the 282 ppm value reached just before the industrial era (Fig. 1b). For a full anthropogenic anomaly of 32 4 ppm, the 14-ppm increase from direct emissions would then represent 35 45% of the total anomaly. CH 4 and CO 2 Emissions DIRECT ANTHROPOGENIC EFFECT Warming Counters Most of Natural Cooling Trend No Northern Ice Sheets; Smaller Southern Sea-ice Advance Natural Drop in CO 2 Prevented INDIRECT ANTHROPOGENIC EFFECT Fig. 8. Direct anthropogenic emissions explain most of the Holocene CH 4 anomaly but only part of the CO 2 anomaly. The rest of the CO 2 anomaly may result from the prevention of a natural CO 2 drop that occurred in previous interglaciations but was overridden in the Holocene by anthropogenic warming.

9 W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] 9 These direct emissions of CH 4 and CO 2 would have produced an anomalous (anthropogenic) warming effect that prevented Holocene climate from following the major cooling trend that had occurred during previous interglaciations, including early. Suppression of this natural cooling would have stifled the natural internal responses that would have driven CO 2 values still lower and added to the size of the anthropogenic CO 2 anomaly (Fig. 8). Two climate-system responses that could have produced a natural CO 2 drop are evident in the Vostok trends from early in (Fig. 4). Stephens and Keeling (2) proposed that advances of Antarctic sea ice are a plausible mechanism for driving down atmospheric CO 2 values by reducing carbon exchanges between Southern Ocean surface water and the atmosphere. The deuterium-based temperature trends in the Antarctic show a major cooling to near-glacial levels early in, but little cooling during the Holocene (Fig. 4c). A cooling to near-glacial levels during must have caused a major advance of sea ice in the Southern Ocean, but any Holocene advance has been far smaller. If anthropogenic emissions averted most of the natural Holocene sea-ice advance, the natural reduction in atmospheric CO 2 from this process must have been largely overridden. This interpretation fits a long history of Southern Ocean studies. The southern hemisphere ocean has been regarded as a locus of early responses to orbital forcing since the work of Hays et al. (1976), as well as later studies (CLIMAP, 1984; Waelbroeck et al., 1995; Brathauer and Abelmann, 1999). Antarctic temperatures reconstructed from deuterium signals are also regarded as a part of this early response (Jouzel et al., 1987; Waelbroeck et al., 1995). A different mechanism may have been at work in the Northern Hemisphere on the slower time scale of response of northern ice sheets. Imbrie et al. (1992) suggested that one of many ways in which ice sheets enhance their own growth is by reducing the CO 2 content of the atmosphere. Ruddiman (23b) found that the phases of the 41,-year component of the Vostok CO 2 signal and the 41,-year component of d 18 O agree, indicating that the CO 2 signal at the obliquity cycle is a fast feedback on ice volume, rather than a driver. Plausible mechanisms by which ice sheets might alter CO 2 include: (1) control of North Atlantic Deep Water flow and its fast-alkalinity response (Broecker and Peng, 1989); and (2) control of eolian dust fluxes that fertilize planktic algae (Martin, 199). Because proxies for deep-water circulation (Raymo et al., 199) and dust fluxes (Clemens and Prell, 199; Kukla et al., 199) have phases close to the 41,-year d 18 O (ice-volume) signal, they are potential mechanisms for explaining positive CO 2 feedback to ice sheets. If ice grew early in, but was prevented from doing so in the Holocene (Fig. 4d), its failure to appear in recent millennia would have eliminated any ice-driven CO 2 decrease. If these two mechanisms prevented Holocene CO 2 levels from falling as they did in the analogous part of, the result would have been an indirect anthropogenic contribution to the size of the Holocene CO 2 anomaly (Fig. 8). Assuming that direct CO 2 emissions explain 14 ppm out of a total anomaly of 32 4 ppm, this indirect contribution would then be ppm. Given that CO 2 values fell by 3 5 ppm for entirely natural reasons during similar intervals in the four previous interglaciations, an indirect effect of ppm seems plausible. Based on model results from Joos et al. (24), a 14-ppm anomaly would require slightly less than 2 Gt C of direct emissions, a value midway between the estimates of Ruddiman (23a) and Joos et al. (24). In summary, the use of as an analog for the natural behavior of the climate system during the Holocene supports the major predictions of the early anthropogenic hypothesis. During the last several millennia, CO 2 and CH 4 levels should have fallen, climate should have cooled, and ice sheets should have begun to grow. The failure of these changes to occur is best explained by early anthropogenic intervention in the operation of the global climate system. This intervention included both direct greenhouse-gas emissions from human activities and indirect climate-system feedbacks that resulted from the direct emissions. Acknowledgements I thank Robert Smith for the graphics. Discussions with Maureen Raymo and Richard Alley broadened the scope of this paper, as did reviews by Jerry McManus and a semi-anonymous person. This research was supported by NSF grant ATM References Andrews, J.T., Mahaffy, M.A.W., Growth rate of the Laurentide ice sheet and sea level lowering (with emphasis on the 115, BP sea level lows). Quaternary Research 6, Bard, E., Hamelin, B., Fairbanks, R.F., 199. U/Th ages obtained by mass spectrometry in corals from Barbados: sea level during the past 13, years. Nature 346, Bassinot, F.C., Labeyrie, L.D., Vincent, E., Quidelleur, X., Shackleton, N.J., Lancelot, Y., The astronomical theory of climate and the age of the Brunhes/Matuyama magnetic reversal. Earth Planetary Science Letters 126, Bauch, H.A., Erlenkeuser, H., Helmke, J.P., Struck, U., 2. A paleoclimatic evaluation of marine isotopic in the highlatitude northern Atlantic (Nordic seas). Global and Planetary Change 24, Bender, M.L., 22. Orbital tuning chronology for the Vostok climate record supported by trapped gas comparison. Earth Planetary Science Letters 24,

10 1 ARTICLE IN PRESS W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] Bender, M., Sowers, T., Labeyrie, L., The Dole effect and its variations during the last 13, years as measured in the Vostok ice core. Global Biogeochemical Cycles 8, Berger, A., Loutre, M.-F., 23. Climate 4, years ago, a key to the future? American Geophysical Union Geophysical Monograph 137, Bickert, T., Curry, W.B., Wefer, G., Late Pliocene to Holocene (2.6 Ma) western equatorial Atlantic deep water circulation: inferences from benthic stable isotopes. Proceedings Ocean Drilling Program. Scientific Results 154, Blunier, T., Chappellaz, J., Schwander, J., Stauffer, J., Raynaud, D., Variations in atmospheric methane concentrations during the Holocene epoch. Nature 374, Brathauer, U., Abelmann, A., Late Quaternary variations in sea surface temperatures and their relationship to orbital forcing recorded in the Southern Ocean (Atlantic sector). Paleoceanography 14, Broecker, W.S., Peng, T.-H., The cause of the glacial to interglacial atmospheric CO 2 change: a polar alkalinity hypothesis. Global Biogeochemical Cycles 3, Burckle, L.H., Late quaternary interglacial stages warmer than present. Quaternary Science Reviews 12, Chapell, J., Shackleton, N.J., Oxygen isotopes and sea level. Nature 324, Clemens, S.C., Prell, W.L., 199. Late Pleistocene variability of Arabian Sea summer monsoon winds and continental aridity: eolian records from the lithogenic component of deep-sea sediments. Paleoceanography 5, CLIMAP Project Members, The last interglacial Ocean. Quaternary Research 21, DeFries, R.S., Field, C.B., Fung, I., Collatz, G.J., Bounoua, L., Combining satellite data and biogeochemical models to estimate global effects of human-induced land cover change on carbon emissions and primary productivity. Global Biogeochemical Cycles 13, Duplessy, J.-C., Labeyrie, L., Waelbrock, C., 22. Constraints on the ocean oxygen isotopic enrichment between the last glacial maximum and the Holocene: paleoceanographic implications. Quaternary Science Reviews 21, EPICA Community Members, 24. Eight glacial cycles from an Antarctic ice core. Nature 429, Hays, J.D., Imbrie, J.I., Shackleton, N.J., Variations in the earth s orbit: pacemaker of the ice ages. Science 194, Imbrie, J., Imbrie, J.Z., 198. Modeling the climatic response to orbital variations. Science 27, Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G., Prell, W.L., Shackleton, N.J., The orbital theory of Pleistocene climate: support from a revised chronology of the marine d 18 O record. In: Berger, A.L., et al. (Eds.), Milankovitch and Climate, Part I. D. Reidel Publ., London, pp Imbrie, J., Boyle, E.A., Clemens, S.C., Duffy, A., Howard, W.R., Kukla, G., Kutzbach, J., Martinson, D.G., McIntyre, A., Mix, A.C., Molfino, B., Morley, J.J., Peterson, L.C., Pisias, N.G., Prell, W.L., Raymo, M.E., Shackleton, N.J., Toggwiler, J.R., On the structure and origin of major glaciation cycles. I. Linear responses to Milankovitch forcing. Paleoceanography 7, Indermuhle, A., Stocker, T.F., Joos, F., Fischer, H., Smith, H.J., Wahlen, M., Deck, B., Masttroianni, D., Blunier, T., Meyer, R., Stauffer, B., Holocene carbon-cycle dynamics based on CO 2 trapped in ice at Taylor Dome, Antarctica. Nature 398, Joos, F., Gerber, S., Prentice, I.C., Otto-Bleisner, B.L., Valdes, P., 24. Transient simulations of Holocene atmospheric carbon dioxide and terrestrial carbon since the last glacial maximum. Global Geochemical Cycles 18, GB /23GB2156. Jouzel, J., Lorius, C., Petit, J.R., Genthon, C., Barkov, N.I., Kotlyakov, V.M., Petrov, M.V., Vostok ice core: a continuous isotope temperature record over the last climatic cycle (16, years). Nature 329, Kukla, G., An, S., Melice, J.L., Gavin, J., Xiao, J.L., 199. Magnetic susceptibility record of Chinese loess. Transactions Royal Society Edinburgh Earth Science 81, Lorius, C., Jouzel, J., Ritz, C., Merlivat, L., Barkov, N.I., Korotkevtich, V.M., A 15,-year climatic record from Antarctic ice. Nature 316, Loutre, M.-F., Berger, A., 2. Future climatic changes: are we entering an exceptionally long interglacial? Climatic Change 46, Martin, J.H., 199. Glacial-interglacial CO 2 changes: the iron hypothesis. Paleoceanography 5, McManus, J., Oppo, D., Cullen, J., Healey, S., 23. Marine isotope (MIS 11): analog for Holocene and future climate? American Geophysical Union Monograph 137, Milankovitch, M.M., Canon of insolation and the ice-age problem. Beograd: Koniglich Serbische Akademie. (English translation by the Israel program for scientific translations). US Department of Commerce, and National Science Foundation, Washington, DC. Mix, A.C., Le, J., Shackleton, N.J., 1995a. Benthic foraminiferal stable isotope stratigraphy of site 846: 1.8 Ma. Proceedings Ocean Drilling Program, Scientific Results 138, Mix, A.C., Pisias, N.G., Rugh, W., Wilson, J., Morey, A., Hagelberg, T.K., 1995b. Benthic foraminifer stable isotope record from Site 849 ( 5 Ma): local and global climate changes. Proceedings Ocean Drilling Program, Scientific Results 138, Petit, J.R., Jouzel, J., Raynaud, D., Barkov, N.I., Barnola, J.-M., Basile, I., Bender, M., Chappellaz, J., Davis, M., Delaygue, G., Delmotte, M., Kotlyakov, V.M., Lipenkov, V., Lorius, C., Pepin, L., Ritz, C., Saltzman, E., Stievenard, M., Climate and atmospheric history of the last 42, years from the Vostok ice core. Nature 399, Pisias, N.G., Mix, A.C., Zahn, R., 199. Nonlinear response in the global climate system: evidence from benthic oxygen isotopic record in core RC Paleoceanography 5, Raymo, M.E., The timing of major climatic terminations. Paleoceanography 12, Raymo, M.E., Ruddiman, W.F., Shackleton, N.J., Oppo, D.W., 199. Evolution of Atlantic Pacific d 13 C gradients over the last 2.5 m.y. Earth and Planetary Science Letters 97, Ruddiman, W.F., 23a. The Anthropogenic Greenhouse Era began thousands of years ago. Climatic Change 61, Ruddiman, W.F., 23b. Orbital insolation, ice volume, and greenhouse gases. Quaternary Science Reviews 22, Ruddiman, W.F., McIntyre, A., Warmth of the subpolar North Atlantic ocean during northern hemisphere ice-sheet growth. Science 24, Ruddiman, W.F., Thomson, J.S., 21. The case for human causes of increased atmospheric CH 4 over the last 5 years. Quaternary Science Reviews 2, Ruddiman, W.F., Raymo, M.E., 23. A methane-based time scale for Vostok ice: climatic implications. Quaternary Science Reviews 22, Ruddiman, W.F., Vavrus, S.J., Kutzbach, J.E., 25. A test of the overdue-glaciation hypothesis. Quaternary Science Reviews 24, 1 1. Sancetta, C., Imbrie, J., Kipp, N.G., McIntyre, A., Ruddiman, W.F., Climatic record in North Atlantic deep-sea core V23-82: comparison of the last and present in interglacials based on quantitative time series. Quaternary Research 2, Schrag, D.P., Adkins, J.F., McIntyre, K., Alexander, J.L., Hodell, D.A., Charles, C.D., McManus, J.F., 22. The oxygen isotopic

11 W.F. Ruddiman / Quaternary Science Reviews ] (]]]]) ]]] ]]] 11 composition of seawater during the last glacial maximum. Quaternary Science Reviews 21, Shackleton, N.J., Oxygen isotopes, ice volume and sea level. Quaternary Science Reviews 6, Shackleton, N.J., 2. The 1,-year ice-age cycle identified and found to lag temperature, carbon dioxide and orbital eccentricity. Science 289, Stephens, B.B., Keeling, R.F., 2. The influence of Antarctic sea ice on glacial interglacial CO 2 variations. Nature 44, Tiedemann, R., Sarnthein, M., Shackleton, N.J., Astronomic timescale for the Pliocene: Atlantic d 18 O and dust flux records of ODP Site 659. Paleoceanography 9, Waelbroeck, C., Jouzel, J., Labeyrie, L., Lorius, C., Labracherie, M., Stievenard, M., Barkov, N.I., A comparison of the Vostok ice deuterium record and series from Southern ocean core MD over the last two glacial cycles. Climate Dynamics 12, Williams, L.D., The little ice age glaciation level on Baffin Island, Arctic Canada. Palaeogeography, Palaeoclimatology, Palaeoecology 25,