the role of orbital forcing in the early middle ... · the early middle pleistocene transition...

lable at ScienceDirect

Quaternary International xxx (2015) 1e9

Contents lists avai

Quaternary International

journal homepage: www.elsevier .com/locate/quaint

The role of orbital forcing in the Early Middle Pleistocene Transition

Mark A. Maslin*, Christopher M. BrierleyDepartment of Geography, Pearson Building, University College London, London, WC1E 6BT, UK

a r t i c l e i n f o

Article history:Available online xxx

Keywords:Orbital forcingEarly Middle Pleistocene TransitionMid Pleistocene TransitionPrecessionObliquityEccentricity

* Corresponding author.E-mail address: [email protected] (M.A. Maslin)

http://dx.doi.org/10.1016/j.quaint.2015.01.0471040-6182/© 2015 Elsevier Ltd and INQUA. All rights

Please cite this article in press as: Maslin, MInternational (2015), http://dx.doi.org/10.10

a b s t r a c t

The Early Middle Pleistocene Transition (EMPT) is the term used to describe the prolongation andintensification of glacialeinterglacial climate cycles that initiated after 900,000 years ago. During thetransition glacialeinterglacial cycles shift from lasting 41,000 years to an average of 100,000 years. Thestructure of these glacialeinterglacial cycles shifts from smooth to more abrupt ‘saw-toothed’ liketransitions. Despite eccentricity having by far the weakest influence on insolation received at the Earth'ssurface of any of the orbital parameters; it is often assumed to be the primary driver of the post-EMPT100,000 years climate cycles because of the similarity in duration. The traditional solution to this is to callfor a highly nonlinear response by the global climate system to eccentricity. This ‘eccentricity myth’ isdue to an artefact of spectral analysis which means that the last 8 glacialeinterglacial average out atabout 100,000 years in length despite ranging from 80,000 to 120,000 years. With the realisation thateccentricity is not the major driving force a debate has emerged as to whether precession or obliquitycontrolled the timing of the most recent glacialeinterglacial cycles. Some argue that post-EMPT de-glaciations occurred every four or five precessional cycle while others argue it is every second or thirdobliquity cycle. We review these current theories and suggest that though phase-locking between orbitalforcing and global ice volume may occur the chaotic nature of the climate system response means therelationship is not consistent through the last 900,000 years.

© 2015 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction

The Early Middle Pleistocene Transition (EMPT; previouslyknown as the Mid-Pleistocene Transition or Revolution; Berger andJansen,1994; Head et al., 2008) is the last major ‘event’ or transitionin a secular trend towards more intensive global glaciation thatcharacterizes the late Cenozoic (Zachos et al., 2001). The earliestrecorded onset of significant regional glaciation during the Ceno-zoic was the widespread continental glaciation of Antarctica atabout 34 Ma (e.g., Zachos et al., 2001; Huber and Nof, 2006; Sijpet al., 2009). Perennial sea ice cover in the Arctic has occurredthroughout the past 14 Ma (Darby, 2008; Schepper et al., 2014).Glaciation in the Northern Hemisphere lagged behind, with theearliest recorded glaciation on Greenland occurring before about 6Ma (e.g., Larsen et al., 1994; Thiede et al., 2011). Schepper et al.(2014) have identified a number of key Pliocene glacial eventswhich may have been global and occurred at 4.9e4.8 Ma, ~4.0 Ma,~3.6 Ma and ~3.3 Ma. It is not until the PlioceneePleistocenetransition that the long-term cooling trend culminates in the

.

reserved.

.A., Brierley, C.M., The role of16/j.quaint.2015.01.047

glaciation of Northern Europe and North America around 2.6 Ma(Maslin et al., 1998). The extent of glaciation did not evolvesmoothly after this, but instead was characterized by periodic ad-vances and retreats of ice sheets on a hemispherical scale e the‘glacialeinterglacial cycles’.

The EMPT is the marked prolongation and intensification ofglacialeinterglacial climate cycles initiated sometime between 900and 650 ka (Fig. 1). Before the EMPT, global climate conditionsappear to have responded primarily to the obliquity orbital peri-odicity (Imbrie et al.,1992; Tiedemann et al.,1994; Clark et al., 2006;Elderfield et al., 2012) through glacialeinterglacial cycles with amean periodicity of ~41 kys. After about 900 ka, starting with Ma-rine Oxygen Isotope Stage (MOIS) 22, glacialeinterglacial cyclesstart to occur with a longer duration and a marked increase in theamplitude of global ice volume variations (Elderfield et al., 2012;Rohling et al., 2014). The increase in the contrast between warmand cold periods may also be in part due to the extreme warmth ofmany of the post-EMPT interglacial periods as similar interglacialconditions can only be found at ~1.1Ma, ~1.3Ma and before ~2.2Ma.Fig. 2 shows time-series analysis of the ODP 659 (Tropical EastAtlantic ocean) benthic foraminifera oxygen isotope record span-ning the EMPT (Mudelsee and Stattegger, 1997). The analysis sug-gests the EMPT was a two-step process with the first transition at

orbital forcing in the Early Middle Pleistocene Transition, Quaternary

mailto:[email protected]

www.sciencedirect.com/science/journal/10406182

http://www.elsevier.com/locate/quaint

http://dx.doi.org/10.1016/j.quaint.2015.01.047



Fig. 1. A) Comparison of the benthic foraminiferal oxygen isotope curve from ODP Site 659 (Tiedemann et al., 1994) with the timing between Terminations (Tiedemann et al., 1994;Raymo, 1997). B) Comparison of two alternate records of global sea level. Black: the marginal sea reconstruction of Rohling et al. (2014) for non-sapropel layers (the dotted portionscorrespond to the median estimates from linear interpolation and are included for aesthetic reasons). We convert to Global Eustatic sea level, by applying a multiplicative factor of1.23 chosen for the most recent glacial cycle. Red: Eustatic sea level determined from a deconvolution of combined benthic foraminifera d18O and Mg/Ca temperature measurementfrom the South-west Pacific (Elderfield et al., 2012). We show the mode and 5e95% range from the probabilistic re-interpretaion by Rohling et al. (2014). (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version of this article.)

M.A. Maslin, C.M. Brierley / Quaternary International xxx (2015) 1e92

about 900 ka, when there is a significant increase in global ice vol-ume but the 41 ky climate response remains. This situation persistsuntil the second step, about 700 ka, when the climate system finds athree-state solution and strong quasi-100 ky climate cycles begin(Mudelsee and Stattegger, 1997). This is consistent with the morerecent evidence from ODP Site 1123 in the Southern Pacific ocean,which shows a step like increase in ice volumeduring glacial periodsstarting at MOIS 22 at about 900 ka (Elderfield et al., 2012).

During the EMPT there seems to be a shift from a two stableclimate state system to a system with three quasi-stable climatestates (Fig. 3). These three states roughly correspond to: 1) fullinterglacial conditions, 2) moderate glacial conditions such as MOIS3 that are analogous to the glacial periods prior to the EMPT and 3)maximum glacial conditions for example MOIS 2, the Last GlacialMaximum (LGM). This has also added confused to the definition ofthe EMPT as many of the intermediate climate periods have beenoverlooked such as the weak interglacial at ~740 ka, which does nothave it own defined MOIS, or the double warm peaks during MOIS15, 13, and 7.

2. Climate feedback mechanisms

Central to understanding the EMPT is the appreciation thatorbital variations do not directly cause global climate changes.Rather they induce small changes in the distribution of insolation

Please cite this article in press as: Maslin, M.A., Brierley, C.M., The role ofInternational (2015), http://dx.doi.org/10.1016/j.quaint.2015.01.047

across the globe that can in some instances be enhanced by strongpositive or negative climate feedbacks and ultimately push theglobal climate into or out of a glacial period. The initial suggestionby Milankovitch (1949) was that glacialeinterglacial cycles wereregulated by summer insolation at about 65�N; this was because hereasoned that for an ice sheet to expand additional ice had to sur-vive each successive summer. The focus on the Northern Hemi-sphere is because the capacity for ice growth is much less in theSouthern Hemisphere due to its smaller landmasses combinedwiththe fact that Antarctica is already close to its ice storage limit. Theconventional view of glaciation is that low summer insolation inthe temperate North Hemisphere allows ice to survive the summerand thus build-up on the northern continents. As snow and iceaccumulate the ambient environment is modified. This is primarilyby an increase in albedo that reduces the absorption of incidentsolar radiation, and thus suppresses local temperatures. The coolingpromotes the accumulation of more snow and ice and thus afurther modification of the ambient environment, causing the so-called ‘ice albedo’ feedback. Other climate feedbacks such aschanges in atmospheric circulation, surface and deep water circu-lation and the reduction in atmospheric greenhouse gases thenplay a role in driving the climate into a glacial period (e.g., Berger,1988; Li et al., 1998; Ruddiman, 2004; Brovkin et al., 2012). Thesefeedbacks then operate in reverse when summer insolation startsto increase (Brovkin et al., 2012; Shakun et al., 2012).


Fig. 2. Statistical analysis of the mid-Pleistocene revolution by Mudelsee andStattegger (1997) demonstrating a delay of 200 ky between a significant increase inglobal ice volume and the start of the 100 ky glacialeinterglacial cycles. They show thatglobal ice volume increased significantly between 940 and 890 ka. Whereas evolu-tionary spectral analysis reveals an abrupt increase of 100-ka cycle amplitude muchlater at approximately 650 ka. Probability density function (PDF) exhibits a bifurcationbehaviour at approximately 725 ka. This suggests that the EMPT moved from two statesystem, full glacial period and full interglacial period to a more complicated systemwith multiple states as suggested by Paillard (1998) and Saltzman (2001) and illus-trated in Fig. 3.

Fig. 3. Cartoon illustrating the changes in glacialeinterglacial cycles before (A) andafter the EMPT (B and C). Note that post-EMPT the glacialeinterglacial cycles are notsimple saw-toothed feature (B) but rather a tripartite system (C) which spends rela-tively short periods of time in either full glacial or interglacial conditions.MOIS ¼ Marine Oxygen Isotope Stage.

M.A. Maslin, C.M. Brierley / Quaternary International xxx (2015) 1e9 3

For pre-EMPT it is suggested that there is a linear relationshipbetween obliquity, heat transfer between latitudes and ice growth(Raymo and Nisancioglu, 2003). Two major differences occur Post-EMPT. First, glacial periods become longer indicating that the ice


sheets are able to survive orbitally-induced increases in summerinsolation and having done so ice volume increases markedlyduring the following downturn in summer insolation creatingmoreintense glacial periods such as the LGM. Second, the deglaciationsare much more abrupt. In the case of the Termination I, the lastdeglaciation, the transition from glacial to interglacial state lastedonly 5e6 ky, even including the brief return to glacial conditionscalled the Younger Dryas period (see detailed references in Maslinet al., 2001). Hence an additional rapid climate feedback mecha-nism must be activated, namely sea level. Once ice sheets haveexpanded to their maximum and so impinge on the marine envi-ronment they become vulnerable. With an upturn in summerinsolation in the North the ice sheet start to melt, this causes sealevel to rise. The ice sheets adjacent to the coasts are under cut byrising sea levels accelerating their collapse, which in turns raisessea level. This sea-level feedback mechanism can be extremelyrapid. This rapid deglaciation that has been postulated causes thesaw-tooth climate signal, that is characteristic of glacialeintergla-cial cycles post-EMPR. However, this is a simplification, becausethough 80% of the ice sheet volume melts during this short periodof time the remaining 20% or about 25m of global sea level does notfully disappear until 5000 years later (Woodroffe and Webster,2014), producing a kink in the rapid deglaciation curve (see Fig. 3C).

Despite the pronounced change in Earth system responseshown in palaeoclimatic records across the EMPT, the frequencyand amplitude characteristics of the orbital parameters do not vary(Berger and Loutre, 1991; Berger et al., 1999). This indicates that the



cause of change in response at the EMPT is internal rather thanexternal to the global climate system.

3. The ‘eccentricity myth’

The major problem with understanding the EMPT is how tointerpret the ‘100 ky’ glacialeinterglacial cycles and the role ofeccentricity (Saltzman et al., 1984; Ghil, 1994). There are two pri-mary views (Maslin and Ridgewell, 2005). The first suggests thatthere is non-linear amplification in the climate system of the ec-centricity signal; the second that the other factors drive globalclimate change and eccentricity rather acts as a pacing mechanism.This debate has not received the attention that it should amongstthe wider palaeoclimatic community, and in many cases the lasteight glacialeinterglacial cycles are thought to be synonymouswith ‘eccentricity forcing’. This view or ‘myth’ is fundamentallyflawed and prevents many excellent palaeoclimatic records frombeing interpreted correctly. Below we reiterate why eccentricitycannot be the direct forcing of the 100 ky glacialeinterglacial cycles.

Eccentricity has spectral peaks at 95 ky, 125 ky and 400 ky (Hayset al., 1976; Berger and Loutre, 1991). In contrast, the spectralanalysis of benthic foraminiferal oxygen isotopes (Fig. 4), which area proxy for global ice volume and deep water temperature,consistently reveals a single peak that dominates the spectra lyingvery close to a period of 100 ky (Muller and MacDonald, 1997).If eccentricity were the primary cause of the 100 ky cycle thenone would expect the global ice volume ~100 ky spectral peak tocontain a double peak and that there would be at least some powerat 400 ky (Fig. 4) (Muller and MacDonald, 1997; Berger, 1999).Ruddiman (2003) for instance clearly shows that the maximumof either an interglacial or glacial is missed by a simple 100,000year filter.

This mismatch between the spectral signatures has led to thesearch for alternative drivers for the assumed ‘pure’ 100 ky

Fig. 4. A) general properties of benthic foraminifera oxygen isotope records from thePacific (OJsox96) and the Atlantic (Imb84) for the last 700 ky (adapted from Berger andWefer, 1996). B) fourier spectra of the two series, showing not the strong single peak at~100 ky, but the dominant orbital frequencies of 19, 23, 41, 95, 125 and 400 ky.


response of the climate system, such as the orbital inclinationdriver proposed by Muller and MacDonald (1997) or carbonate-chemistry response time (Toggweiler, 2008). However, a strong100 ky spectral peak in a truncated time series does not imply thepresence of a 100 ky periodicity in the data (see Berger et al., 2005).Indeed, Ridgwell et al., (1999) showed that the observed spectralsignature of ice volume can be reproduced by a simple saw-toothpattern based on the long glaciation period followed by the shortdeglaciation (Fig. 5). Here, the timing of the rapid deglaciationevent is simply assumed to occur synchronous with every fourth orfifth precessional cycle. A similar saw tooth pattern can also beproduced from obliquity using every second or third cycle, but thespectral analysis of this parameter matches the oxygen isotoperecord less precisely (Huybers and Wunsch, 2005). The length ofthe glacialeinterglacial cycles is far from uniform in this analysisand the resultant spectral signature with a dominant ‘100 ky’ peakis thus in effect an artefact of spectral analysis. For example, Fig. 1shows that the time between deglaciations can vary from 120 kyto as little as 77 ky over the last 700 ka. Although the deglacialtransitions are no more than quasi-periodic and do not recur atanything like a regular 100 ky interval, the resultant spectralanalysis gives the appearance of a ~100 ky periodicity present inthe data. This is the ‘eccentricity myth’. This observation has notstopped significant effort by the palaeoclimate community inves-tigating the role of eccentricity in driving Pleistocene climate. Forexample, Lisiecki (2010) uses a high Rayleigh number to justify thatpost-EMPT cycles are synchronised to eccentricity e despite thefact that this is “not necessarily a good indicator of reliable syn-chronisation” Crucifix (2013). Rial et al. (2013) claim the 100 kycycles emerge as a harmonic of the longer 413 ky eccentricity cycleexciting a natural resonance after nearly 5 million years.

If the post-EMPT ‘100 ky’ cycles are in effect an artefact of thespectral analysis of a truncated time series containing a dominantquasi-periodic glacial termination motif, then one must alsoquestion what we really mean by the EMPT. Although there is aclear visible change in the appearance of the ice volume variabilityrevealed in proxy records (e.g. Fig. 1), the EMPT is typically definedthrough spectral analysis. The results of evolutive spectral analysisstudies suggest step changes in the dominant frequency and meanice volume associated with the EMPR, although not necessarilyoccurring synchronously (e.g., Mudelsee and Schulz, 1997;Mudelsee and Stattegger, 1997). The implications are of progres-sive steps to a new mode (oscillation) of the climate system, or anumber of ‘bifurcations’ (Maslin, 2004). However, it is less easy tosee this elegant picture from the raw data.

As an example, we present a wavelet analysis of the global icevolume record of Rohling et al. (2014) shown earlier (Fig. 6;Torrence and Compo, 1998; NCAR's Command Language, 2013). Forinstance, whereas significant power in the 100 ky band emerges atc. 900 ka (MOIS 23e22), it then recedes between ~800 and ~600 ka(MOIS 19e16) and there is an interval of apparent obliquity-dominated (Fig. 6). Even between ~600 ka (MIOS 16) and present,the variability in ice volume at times contains significant power inthe obliquity and precession bands (as seen in the pre-EMPT 41 ky‘world’). It is also clear the significant power is not strictly confinedto 100 ky, but spans 80e125 ky. However, sufficient quasi-periodicrecurrence occurred in the past 600 or 900 ky to allow a strong100 ky peak emerge in a spectral analysis (see Fig. 4).

4. Obliquity versus precession debate

A debate has emerged over whether precession or obliquitycontrolled the timing of the most recent glacialeinterglacial cycles,in light of the observation that eccentricity did not. Huybers andWunsch (2005) and Huybers (2007, 2009) argue that post-EMPT


Fig. 5. A) SPECMAP stacked d18O composite showing marine oxygen isotopes stage with spectral analysis B) insolation for June 21st 65�N showing the quasi-periodic insolationmaxima of unusually low strength (shown by arrows) preceding glacialeinterglacial terminations by one precessional cycle in each case, with spectral analysis and C) sawtoothartificial ice volume signal with spectral analysis with spectral analysis (adapted from Ridgwell et al., 1999).


deglaciations occur every second or third obliquity cycle. Alterna-tively, Ridgwell et al. (1999) andMaslin and Ridgewell (2005) arguethat deglaciation occurred every four or five precessional cycle.

Prior to the EMPT the climate system is dominated by obliquity,the so-called 41 ky world. Raymo and Nisancioglu (2003) suggestthe climate system was sensitive to obliquity-forced latitudinalinsolation gradients, which exert a strong control on summer

Fig. 6. Wavelet power shown in the eustatic sea level reconstruction of Rohling et al. (2interpolation over sapropel layers will lead to some errors at higher frequencies. Cross-haregions are statistically significant at the 5% confidence level. Significant power in the 100merges after 600 ka. The interval between ~800 and ~600 ka is apparently obliquity-domithe obliquity and precession bands. It is also clear the significant power is not strictly confi


atmospheric heat transport. Although the insolation curves for theNorthern Hemisphere are dominated through time by precession,the insolation gradient between high and low latitude is dominatedby obliquity (Berger et al., 2010). This obliquity-driven insolationgradient must therefore be the prime control on glacialeintergla-cial cycles prior to the EMPT. Raymo and Nisancioglu (2003) sug-gest that differential heating between high and low latitudes in

014). The analysis has been performed on the continuous median estimates, whosetching indicates regions influenced by the end of the reconstruction, whilst stippledky band emerges at c. 900 ka, then recedes between ~800 and ~600 ka and then re-nated. After ~600 ka the variability in eustatic sea level contains significant power inned to 100 ky, but spans 80e125 ky.



summer exerts a dominant control on global climate through itsimpact on the atmospheric meridional flux of energy, moisture andlatent heat. As the majority of heat transport between 30� and 70�Nis by the atmosphere, a linear relationship between obliquity,northward heat transport and glacialeinterglacial cycles can beenvisioned. When this summer heat transport was low the icesheets could build-up, and when heat transport increased the icesheets correspondingly shrunk. This is a bi-modal systemresponding approximately linearly to insolation gradients. Thepollen-based reconstruction of the latitudinal temperaturegradient by Davis and Brewer (2008) not only supports this sug-gestion, but also implies a greater sensitivity to insolation gradientsthan initially expected.

Alternate explanations for the role of obliquity include it con-trolling Antarctic snow accumulation (Lee and Poulsen, 2009) andthe fact that obliquity controls annual mean insolation over alllatitudes (Loutre et al., 2004). The limited size of continental icesheets meant they were less susceptible to rapid deglaciation dueto sea level rise, resulting in the relatively gradual deglaciationsobserved at this time. The spectral signature of pre-EMPT ice vol-ume therefore shows a dominance of obliquity over precession. Ithas also been suggested the dominance of obliquity is an artefactarising as the precessional signal is ‘cancelled out’ (Raymo andHuybers, 2008). This is because precession is anti-phase betweenthe Northern and Southern Hemispheres while obliquity changesare in phase. If the temperature and sea level changes between thetwo Hemispheres driven by precession were approximately thesame, they would be out of phase and hence cancel out. One couldenvisage Northern Hemisphere ice sheets varying with precessionleading to a benthic foraminifera d18O response that is countered byan equal and opposite response from the Southern Hemisphere icesheets (Raymo et al., 2006). This suggestion is perhaps refuted byrecent sea level reconstructions (Figs. 1 and 6; Elderfield et al.,2012; Rohling et al., 2014). The Antarctic ice volume variationsrequired to disguise precessional forcing of benthic d18O and sealevel are different, because of differing isotopic fractionation of theNorthern and Southern Hemisphere ice-sheets. For example duringthe last glacial, the Laurentide ice sheet had an average d18O of �28per mil to�34 per mil, whereas the Antarctic was between�40 permil and �60 per mil (Maslin and Swann, 2006). Moreover recon-struction of the waxing and waning of early Pleistocene glacialoutwash plains in North America has been shown to be obliquitypaced, strongly arguing against a precession controlled Laurentideice sheet during this time (Naafs et al., 2012).

Huybers and Wunsch (2005) and Huybers (2007, 2009) arguedthat the obliquity mechanism is intrinsic to the climate system andmust continue after the EMPT. They suggested that post-EMPT de-glaciations occur every second or third obliquity cycle. The problemthat they encountered is the timing of each of the major de-glaciations, since the EMPT, occurs on very different parts of therising arm of the obliquity curve. They also encounter the Termi-nation III problem as this occurs as obliquity is dropping, so theyopted to define this deglaciation not at 240 ka but at the second risein oxygen isotopes at 210 ka.What is clear fromboth oxygen isotopeand global sea level records is that the timing of deglaciations sincethe start of the EMPT all occur on the rising limb of the precessionalcurve (Fig. 5) when there is the greatest rate of change in preces-sional forcing (Maslin and Ridgewell, 2005).

Raymo (1997) suggested that the episodic occurrence of pre-cessionally driven unusually low maxima in Northern Hemispheresummer insolation is the critical factor controlling subsequentdeglaciation. In thismodel, glacial terminationonlyoccurswhen theclimate system has been predisposed by excessive ice sheet growthby a previous low maximum in Northern Hemisphere summerinsolation to some criticalmaximumdegree of glaciation (see Fig. 4).


Of course one such time is the LGM. This critical threshold mightreflect the sinking of the underlying bedrock to an extent sufficientto allow full activation of additional and catastrophicmechanisms ofice sheet collapse once the ice sheet starts to initially retreat underan unfavourable combined insolation and CO2 radiative forcingregime (e.g., Imbrie et al., 1993; Clark and Pollard, 1998; Ganopolskiand Calov, 2011). In otherwords, precession driven summerminimapush the glacial climate system too far and it collapses all the wayback into an interglacial period (Tziperman and Gildor, 2003).Although eccentricity determines the envelope of precessionalamplitude it is ultimately whether the insolation minimum occurson the fourth or fifth precessional cycle that determines when over-extension of the ice sheets can occur which results ultimately in therapid deglaciation (Ridgwell et al., 1999; Maslin and Ridgewell,2005). This theoretical construction is supported by the sea levelevidence from the most recent deglaciation. The loss of ice equiva-lent to 90e100m of global sea level occurred very rapidly while themajor ice sheets were vulnerable to the effects of rising sea level.Once the continental ice sheets had retreat sufficiently not to beunder cut by rising sea level the remaining 20e25 m of equivalentsea level took nearly 5000 years tomelt. The precessional thresholdmodel is also support by recent ice sheet modelling work, whichshows that precessional forcing is the dominant control onglobal icevolume (Abe-Ouchi et al., 2013).

It may be that such an argument is itself unnecessary and thatboth obliquity and precession contribute to the 100 ky cycles seenafter the EMPT. Huybers (2011), for example, used a novel statisticalapproach to find robust influences with both. Using concepts ofgeneral synchronisation from the discipline of dynamical systemstheory, De Saedeleer et al. (2012) reach a similar conclusion.Crucifix (2013) goes further and demonstrates the climate systemforced by the full orbital solution is sensitive to stochastic events tosuch an extent that the question of obliquity or precession may beill posed. This instability will lead to quantum skips of insolationcycles - effectively at random (Crucifix, 2013).

5. Discussion

The EMPT could be thought of not as a transition to a newmodeof glacialeinterglacial cycles per se, but simply the point at which amore intense and prolonged glacial state and associated subse-quent rapid deglaciation becomes possible. An important point inthis view is that whereas from the EMPTonwards it may be possiblefor the climate system to achieve this new glacial climate solution,it need not do so each time. The success or failure to achieve thisstate would be determined by factors such as the exact details ofinsolation regime and carbon cycling. One would also expect anincreasing probability of a ‘100 ky’motif occurring with time, as thelong-term late Cenozoic cooling/CO2 trend presumably continues.

A number of different theories have been forwarded as to whyafter the EMPT ice sheets were able to survive longer and becameincreasing vulnerable to catastrophic collapse and the resultantrapid deglaciation. It has been suggested that long term coolingthrough the Cenozoic instigated a threshold, which allowed the icesheets to become large enough to ignore the 41 ky orbital forcingand to survive between 80 and 100 ka (Abe-Ouchi, 1996; Raymoet al., 1997). In one version of this, Gildor and Triperman (2000)and Tziperman and Gildor (2003) suggest that long term coolingof the deep ocean during the Pleistocene alters the relationshipbetween atmospheric temperature and accumulation rates of snowon continental ice sheets and the growth of sea ice. They envisionthis alteration causing more extensive global coverage of sea ice.The climate could then be affected by a so-called sea-ice switchwhich would produce the rapid asymmetric deglaciation observedafter the EMPT. Alternatively, Northern Hemisphere ice sheets


Fig. 7. Surface ocean aqueous Pco2 (dots with errors bars) calculated as a function of pH, alkalinity, SST, and salinity by H€onisch et al. (2009) compared with the ice core record ofatmospheric Pco2 (solid line). Average glacial and interglacial carbon doixide levels are indictaed by horizontal lines (this study). Note the changes at the Early Middle PleistoceneTransition (EMPT) and the Mid Brunhes Event (MBE).


impinging into the ocean may be stable to some (Gomez et al.,2010), but not all forcing conditions.

It has been suggested that the critical size of the NorthernHemisphere ice sheetsmayhave been influenced bychanging levelsof greenhouse gases (Clark et al., 2006; Crowley and Hyde, 2008;DeConto et al., 2008). The atmospheric carbon dioxide reconstruc-tion by H€onisch et al. (2009) shows significant changes at the EMPTand the Mid-Brunhes Event (MBE) see Fig. 7. Though the data issparse it seems to indicate a drop in both the interglacial level ofcarbon dioxide from ~290 ppm to ~ 260 ppm and the glacial levelfrom 220 ppm to 180 ppm. At the MBE the glacial level remains thesame but the interglacial level rises to it pre-EMPT level of 290 ppm(H€onisch et al., 2009). Hence the cooling trend through the EMPTcould have been driven by the observed reduced concentration ofgreenhouse gases in the atmosphere. This secular decline in theconcentration of CO2 in the atmosphere could have brought theglobal climate to a threshold, allowing it to respond non-linearly toorbital forcing thereafter (Mudelsee and Schulz,1997;Mudelsee andStattegger, 1997; Raymo et al., 1997; Berger et al., 1999). Saltzman(2001) takes this further and proposes that in crossing thisthreshold (‘bifurcation’) an internal instability arises in the globalcarbon cycle that leads to the activation of an internal 100 kyoscillator. Such a threshold occurs in themodel used by CrowleyandHyde (2008), who state that ~100 ky power is “embedded in thephysics of the coupled climate/ice-sheet system”. In this view,orbital forcing plays only a very minor phase-locking role. If atmo-spheric carbon dioxide and not Northern Hemisphere ice volume isthe primary driving force of the ~100 ky glacialeinterglacial cycles(as in the suggestion of Toggweiler, 2008), then an interpretationconsistent with palaeoclimatic proxies suggests that eccentricity,atmospheric carbon dioxide, Vostok (Antarctica) air temperaturesand deepwater temperatures are in phase; whereas ice volume lagsthese other variables, as originally suggested by Shackleton (2000).In this case the EMPT could represent a change in the internalresponse of the global carbon cycle to orbital forcing. This isconsistent with ice sheet modelling results, which show theimportance of atmospheric carbon dioxide levels in the initiationand maintenance of the Eastern Antarctic (Pollard and DeConto,2009) and Greenland (Lunt et al., 2008) ice sheets. Crowley andHyde (2008) even found a further threshold, after which Eurasiawould start to maintain large ice-sheets. They suggested such athreshold would have been crossed in the geological record, hadmankind not altered the climate system's trajectory.

6. Conclusion

There is clear evidence for both a prolongation and intensifica-tion of glacialeinterglacial climate cycles during the Early MiddlePleistocene Transition (EMPT). We suggest the structure of gla-cialeinterglacial cycles shifts from a smooth sinusoidal structure to


a tripartite system (Fig. 3) whose spectral signature is dominated bythe large and rapid deglaciations. We suggest that previous ex-planations of a non-linear response to eccentricity or a linearresponse to either obliquity or precession are too simplistic. It isclear that prior to the EMPT the timing of glaciations and de-glaciations is linked to obliquity and after the EMPT they are alsoinfluenced by precession. The intensification of the glacial periods,however, are linked to the ability of the climate system to ignoresuccessive upturns in obliquity, hence the glacial cycles last onaverage two or three obliquity cycles before deglaciation occurs.This also represents an increased dominance of the Northernhemisphere in driving glacialeinterglacial cycles. Prior to the EMPTthe obliquity forcing would have been in the same direction in bothHemispheres producing a re-enforcement into or out of a glacialperiod. Post EMPT precession provides the deglaciation timing butits influence on the Hemispheres would have been in opposition.We speculate that the EMPT was may be due to a changing internalresponse of the global carbon cycle to orbital forcing, whichallowed increased ice sheet growth in the Northern Hemispherewhich provided enough local climate feedbacks for the increasedheat transport northward driven by obliquity to be ignored. Itseems that the climate system is a stochastic system, which ismodulated by orbital forcing. However the ever changing rela-tionship between the different orbital parameters and their rela-tionship with the internal climate feedbacks means that there is noconsistent phase lock-in (Huybers, 2011; De Saedeleer et al., 2012;Crucifix, 2013). What does seem to be critical post-EMPT is theoccasional generation of ‘full’ glacial conditions which make thewhole system very unstable allowing the climate to rebound outsuddenly in to full interglacial condition.

Acknowledgements

We would like to thank the UCL Department of GeographyDrawing Office in help in preparing the diagrams for this paper. Wewould like to thank the two reviewers, Michel Crucifix and the UCLMonday manuscript group for all their helpful comments. Waveletsoftware was provided by C. Torrence and G. Compo (and is avail-able at the URL: http://paos.colorado.edu/research/wavelets/)although we use the implementation in NCL.

References

Abe-Ouchi, A., 1996. Quaternary transition: a bifurcation in forced ice sheet oscil-lations? Eos Transactions AGU 77 (46). Fall Meeting Supplement, F415.

Abe-Ouchi, A., Saito, F., Kawamura, K., Raymo, M.E., Okuno, J., Takahashi, K.,Blatter, H., 2013. Insolation-driven 100,000-year glacial cycles and hysteresis ofice-sheet volume. Nature 500, 190e193. http://dx.doi.org/10.1038/nature12374.

Berger, A., 1988. Milankovitch theory of climate. Review of Geophysics 26 (4),624e657.

Berger, W.H., 1999. The 100-ky ice-age cycle: internal oscillation or inclinationalforcing? International Journal Earth Science 88, 305e316.


http://paos.colorado.edu/research/wavelets/

http://refhub.elsevier.com/S1040-6182(15)00070-1/sref1


http://dx.doi.org/10.1038/nature12374








Berger, W.H., Jansen, E., 1994. Mid-Pleistocene climate shift: the Nansen connection.In: Johannessen, et al. (Eds.), The Polar Oceans and Their Role in Shaping theGlobal Environment, AGU Geophysical Monograph 85, pp. 295e311.

Berger, A., Loutre, M.F., 1991. Insolation values for the climate of the last 10 millionyears. Quaternary Science Reviews 10, 297e317.

Berger, W.H., Wefer, G., 1996. Expeditions into the Past: Paleoceanographic studiesin the South Atlantic. In: Wefer, et al. (Eds.), The South Atlantic: Present andPast Ciculation. Spinger-Verlag, Berlin, pp. 363e410.

Berger, A., Li, X., Loutre, M.F., 1999. Modelling northern hemisphere ice volume overthe last 3 Ma. Quaternary Science Reviews 18, 1e11.

Berger, A., Melice, J.L., Loutre, M.F., 2005. On the origin of the 100-ky cycles in theastronomical forcing. Paleoceanography 20, PA4019. http://dx.doi.org/10.1029/2005PA001173.

Berger, A., Loutre, M.F., Yin, Q.Z., 2010. Total irradiation during any time intervalof the year using elliptic integrals. Quaternary Science Reviews 29,1968e1982.

Brovkin, V., Ganopolski, A., Archer, D., Munhoven, G., 2012. Glacial CO2 cycle as asuccession of key physical and biogeochemical processes. Climates of the Past 8,251e264.

Clark, P., Pollard, D., 1998. Origin of the Middle Pleistocene transition by ice sheeterosion of regolith. Paleoceanography 13, 1e9.

Clark, P.U., Archer, D., Pollard, D., Blum, J.D., Rial, J.A., Brovkin, V., Mix, A.C., et al.,2006. The Middle Pleistocene transition: characteristics, mechanisms, and im-plications for long-term changes in atmospheric pCO2. Quaternary ScienceReviews 25, 3150e3184.

Crowley, T.J., Hyde, W.T., 2008. Transient nature of late Pleistocene climate vari-ability. Nature 456, 226e230.

Crucifix, M., 2013. Why could ice ages be unpredictable? Climate of the Past 9,2253e2267.

Darby, D.A., 2008. Arctic perennial ice cover over the last 14 million years. Paleo-ceanography 23 (1), PA1S07. http://dx.doi.org/10.1029/2007PA001479.

Davis, B.A.S., Brewer, S., 2008. Orbital forcing and role of the latitudinal insolation/temperature gradient. Climate Dynamics 32 (2e3), 143e165.

De Saedeleer, B., Crucifix, M., Wieczorek, S., 2012. Is the astronomical forcing areliable and unique pacemaker for climate? A conceptual model study. ClimateDynamics 40, 273e294.

DeConto, R.M., Pollard, D., Wilson, P.A., P€alike, H., Lear, C.H., Pagani, M., 2008.Thresholds for Cenozoic bipolar glaciation. Nature 455, 652e656.

Elderfield, H., Ferretti, P., Greaves, M., et al., 2012. Evolution of ocean temperatureand ice volume through the Mid-Pleistocene Climate Transition. Science 337,704e709.

Ganopolski, A., Calov, R., 2011. The role of orbital forcing, carbon dioxide andregolith in 100 ky cycles. Climates of the Past 7, 1415e1425.

Ghil, M., 1994. Cryothermodynamics: the chaotic dynamics of paleoclimate. PhysicaD: Nonlinear Phenomena 77 (1e3), 130e159.

Gildor, H., Triperman, E., 2000. Sea ice as the glacial cycles climate switch. Paleo-ceanography 15, 605e615.

Gomez, N., Mitrovica, J.X., Huybers, P.J., Clark, P.U., 2010. Sea level as a stabilizingfactor for marine-ice-sheet grounding lines. Nature Geoscience 3 (12), 850e853.http://dx.doi.org/10.1038/ngeo1012.

Hays, J.D., Imbrie, J., Shackleton, N.J., 1976. Variations in the Earth's orbit: pacemakerof the Ice Ages. Science 194, 1121e1132.

Head, M.J., Pillans, B., Farquhar, S., 2008. The Early Middle Pleistocene Transition:characterization and proposed guide for the defining boundary. Episodes 31 (2),255e259.

H€onisch, B., Hemming, N.G., Archer, D., Siddall, M., McManus, J., 2009. Atmosphericcarbon dioxide concentration across the mid-Pleistocene transition. Science324, 1551e1554.

Huber, M., Nof, D., 2006. The ocean circulation in the southern hemisphere and itsclimatic impacts in the Eocene. Palaeogeography, Palaeoclimatology, Palae-oecology 231, 9e28.

Huybers, P., 2007. Glacial variability over the last two million years: an extendeddepth-derived agemodel, continuous obliquity pacing, and the Pleistoceneprogression. Quaternary Science Reviews 26 (1e2), 37e55.

Huybers, P., 2009. Pleistocene glacial variability as a chaotic response to obliquityforcing. Climate of the Past 5 (3), 481e488.

Huybers, P.J., 2011. Combined obliquity and precession pacing of late Pleistocenedeglaciations. Nature 480, 229e232.

Huybers, P., Wunsch, C., 2005. Obliquity pacing of the late Pleistocene glacial ter-minations. Nature 434, 491e494.

Imbrie, J., Boyle, E., Clemens, S., Duffy, A., Howard, W., Kukla, G., Kutzbach, J.,Martinson, D., McIntyre, A., Mix, A., Molfino, B., Morley, J., Peterson, L., Pisias, N.,Prell, W., Raymo, M., Shackleton, N., Toggweiler, J.R., 1992. One the structure andorigin of major glaciation cycles. 1. Linear responses to Milankovitch forcing.Paleoceanography 7, 701e738.

Imbrie, J., Berger, A., Boyle, E.A., Clemens, S.C., Duffy, A., Howard, W.A., 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.G., Raymo, M.E., Shackleton, N.J.,Toggweiler, J.R., 1993. On the structure and origin of major glaciation cycles. 2.The 100,000 year cycle. Paleoceanography 8, 699e735.

Larsen, H.C., Saunders, A.D., Clift, P.D., Beg�et, J., Wei, W., Spezzaferri, S., Ali, J.,Cambray, H., Demant, A., Fitton, G., Fram, M.S., Fukuma, K., Gieskes, J.,Holmes, M.A., Hunt, J., Lacasse, C., Larsen, L.M., Lykke-Andersen, H., Meltser, A.,Morrison, M.L., Nemoto, N., Okay, N., Saito, S., Sinton, C., Stax, R., Vallier, T.L.,


Vandamme, D., Werner, R., 1994. Seven million years of glaciation in Greenland.Science 264, 952e955.

Lee, S.-Y., Poulsen, C.J., 2009. Obliquity and precessional forcing of continental snowfall and melt: implications for orbital forcing of Pleistocene ice ages. QuaternaryScience Reviews 28, 2663e2674.

Li, X.S., Berger, A.M., Loutre, M.F., 1998. CO2 and Northern Hemisphere icevolume variations over the middle and late Quaternary. Climate Dynamics14, 537e544.

Lisiecki, L.E., 2010. Links between eccentricity forcing and the 100,000-year glacialcycle. Nature Geoscience 3 (5), 349e352.

Loutre, M.-F., Paillard, D., Vimeux, F., Cortijo, E., 2004. Does mean annual insolationhave the potential to change the climate? Earth and Planetary Science Letters221, 1e14.

Lunt, D.J., Foster, G.L., Haywood, A.M., Stone, E.J., 2008. Late Pliocene Greenlandglaciation controlled by a decline in atmospheric CO2 levels. Nature 454,1102e1105.

Maslin, M.A., 2004. Ecological verses climatic thresholds. Science 306, 2197e2198.Maslin, M.A., Ridgewell, A., 2005. Mid-pleistocene Revolution and the Eccentricity

Myth. In: Special Publication of the Geological Society of London, 247,pp. 19e34.

Maslin, M.A., Swann, G., 2006. Isotopes in marine sediments. In: Leng, M. (Ed.),Isotopes in Palaeoenvironmnetal Research. Springer, Dordrecht, TheNetherlands, pp. 227e290.

Maslin, M.A., Li, X.S., Loutre, M.F., Berger, A., 1998. The contribution of orbital forcingto the progressive intensification of Northern Hemisphere Glaciation. Quater-nary Science Reviews 17, 411e426.

Maslin, M.A., Seidov, D., Lowe, J., 2001. Synthesis of the nature and causes of suddenclimate transitions during the Quaternary. In: Seidov, D., Haupt, B., Maslin, M.A.(Eds.), The Oceans and Rapid Climate Change: Past, Present and Future, AGUGeophysical Monograph Series 126, pp. 9e52.

Milankovitch, M.M., 1949. Kanon der Erdbestrahlung und seine Anwendung aufdas Eiszeitenproblem. In: Royal Serbian Sciences, Spec. Pub. 132. Section ofMathematical and Natural Sciences, 33, Belgrade, p.633 (Canon of Insolationand the Ice Age Problem, English translation by Israel Program for ScientificTranslation and published for the U.S. Department of Commerce and theNational Science Foundation, Washington D.C., 1969).

Mudelsee, M., Schulz, M., 1997. The Mid-Pleistocene climate transition: onset of 100ka cycle lags ice volume build up by 280 ka. Earth and Planetary Science Letters151, 117e123.

Mudelsee, M., Stattegger, K., 1997. Exploring the structure of the mid-Pleistocenerevolution with advance methods of time-series analysis. Geologische Run-dschau 86, 499e511.

Muller, R.A., MacDonald, G.J., 1997. Glacial cycles and astronomical forcing. Science277, 215e218.

Naafs, B.D.A., Hefter, J., Acton, G., Haug, G.H., Martínez-Garcia, Alfredo, Pancost, R.,Stein, R., 2012. Strengthening of North American dust sources during the latePliocene (2.7 Ma). Earth and Planetary Science Letters 317e318, 8e19.

Paillard, D., 1998. The timing of Pleistocene glaciations from a simple multiple-stateclimate model. Nature 391, 378e381.

Pollard, D., DeConto, R.M., 2009. Modelling West Antarctic ice sheet growth andcollapse through the past five million years. Nature 458, 329e332.

Raymo, M.E., 1997. The timing of major climate terminations. Paleoceanography 12,577e585.

Raymo, M.E., Huybers, P.J., 2008. Unlocking the mysteries of the ice ages. Nature451, 284e285.

Raymo, M.E., Nisancioglu, K., 2003. The 41 ky world: Milankovitch's other unsolvedmystery. Paleoceanography 18 (1), 1011. http://dx.doi.org/10.1029/2002PA000791.

Raymo, M., Oppo, D., Curry, W., 1997. The mid-Pleistocene climate transition: a deepsea carbon isotope perspective. Paleoceanography 12, 546e559.

Raymo, M.E., Lisiecki, L.E., Nisancioglu, K.H., 2006. Plio-Pleistocene ice volume,Antarctic climate, and the global d18O record. Science 313, 492e495.

Rial, J.A., Oh, J., Reischmann, E., 2013. Synchronization of the climate system toeccentricity forcing and the 100,000-year problem. Nature Geoscience 6,289e293. http://dx.doi.org/10.1038/ngeo1756.

Ridgwell, A.,Watson, A., Raymo,M.,1999. Is the spectral signature of the 100 kyglacialcycle consistent with a Milankovitch origin. Paleoceanography 14, 437e440.

Rohling, E.J., et al., 2014. Sea-level and deep-sea-temperature variability over thepast 5.3 million years. Nature 508, 477e482.

Ruddiman, W.F., 2003. Orbital insolation, ice volume, and greenhouse gases. Qua-ternary Science Reviews 22, 1597e1629.

Ruddiman, W.F., 2004. The role of greenhouse gases in orbital-scale climaticchanges. EOS 285 (1), 6e7.

Saltzman, B., 2001. Dynamical Paleoclimatology : Generalized Theory of GlobalClimate Change. Academic Press, San Diego, p. 320.

Saltzman, B., Sutera, A., Hansen, A.R., 1984. Earth-orbital eccentricity variations andclimatic change. In: Berger, A.M. (Ed.), Milankovitch and Climate, Part 2. D.Reidel, Norwell, Mass, pp. 615e636.

Schepper, S.D., Gibbard, P.L., Salzmann, U., Ehlers, J., 2014. A global synthesis of themarine and terrestrial evidence for glaciation during the Pliocene Epoch. Earth-Science Reviews 135, 83e102.

Shackleton, N.J., 2000. The 100,000 year Ice Age cycle identified and found to lagtemperature, carbon dioxide and orbital eccentricity. Science 289,1897e1902.
















http://dx.doi.org/10.1029/2005PA001173

http://dx.doi.org/10.1029/2005PA001173
























http://dx.doi.org/10.1029/2007PA001479



























http://dx.doi.org/10.1038/ngeo1012





























































































































http://dx.doi.org/10.1029/2002PA000791

http://dx.doi.org/10.1029/2002PA000791







http://dx.doi.org/10.1038/ngeo1756




























Shakun, J.D., et al., 2012. Global warming preceded by increasing carbon dioxideconcentrations during the last deglaciation. Nature 484, 49. http://dx.doi.org/10.1038/nature10915 (2012).

Sijp, W.P., England, M.H., Toggweiler, J.R., 2009. Effect of ocean gateway changesunder greenhouse warmth. Journal of Climate 22, 6639e6652.

The NCAR Command Language (Version 6.1.2) [Software]. (2013). Boulder, Colorado:UCAR/NCAR/CISL/VETS. http://dx.doi.org/10.5065/D6WD3XH5.

Thiede, J., Jessen, C., Knutz, P., Kuijpers, A., Mikkelsen, N., Norgaard-Pedersen, N.,Spielhagen, R., 2011. Millions of years of Greenland ice sheet history recorded inocean sediments. Polarforschung 80, 141e159.

Tiedemann, R., Sarnthein, M., Shackleton, N.J., 1994. Astronomic timescale for thePliocene Atlantic d18O and dust flux records of ODP Site 659. Paleoceanography9, 619e638.


Toggweiler, J.R., 2008. Origin of the 100,000-year timescale in Antarctic tempera-tures and atmospheric CO2. Paleoceanography 23 (2). http://dx.doi.org/10.1029/2006PA001405.

Torrence, C., Compo, G.P., 1998. A Practical guide to wavelet analysis. Bulletin,American Meteorological Society 79, 61e78.

Tziperman, E., Gildor, H., 2003. On the mid-Pleistocene transition to 100 ky glacialcycle and the asymmetry between glaciation and deglaciation times. Paleo-ceanography 18 (1), 1001. http://dx.doi.org/10.1029/2001PA000627.

Woodroffe, C.D., Webster, J.M., 2014. Coral reefs and sea-level change. Marine Ge-ology 352, 248e267.

Zachos, J.C., Pagani, M., Sloan, L., Thomas, E., Billups, K., 2001. Trends, rhythms andaberrations in global climate 65 Ma to present. Science 292, 686e693.


http://dx.doi.org/10.1038/nature10915 (2012)

http://dx.doi.org/10.1038/nature10915 (2012)




http://dx.doi.org/10.5065/D6WD3XH5










http://dx.doi.org/10.1029/2006PA001405

http://dx.doi.org/10.1029/2006PA001405




http://dx.doi.org/10.1029/2001PA000627







the role of orbital forcing in the early middle ... · the early middle pleistocene transition...

Documents