Category: Climate change and palaeoclimatology

Ice extracted from ice sheets by core drilling has provided the most detailed historical information on climate variation at high latitudes and about the varying gas and dust content of the atmosphere.  It provides the best time-resolution currently available, sometimes of the order of 50 years. Cores from the Greenland ice sheet revolutionised ideas about the controls over short-term climate shifts in the northern hemisphere – the millennial-scale Heinrich and Dansgaard-Oeschger events.  It is from those revelations that fears have arisen about the consequences of deep-ocean circulation shut-downs that might arise from current global warming.  The Greenland ice goes back only to cover the last glaciation and part of the interglacial period the preceded it.  Until recently, the Vostok ice core from Antarctica gave the greatest penetration into past climatic events, to around 430 ka that covers the last four glacial epochs.  Again, Vostok revolutionised our understanding of past climate change, principally the differences between climate behaviour in interglacials, and those between the records from northern and southern hemispheres.  North and south have not been in exact harmony, at least as far as high latitudes are concerned.  Ocean-floor sediment cores and those from mid-latitude glaciers do give hints of a global harmonisation of events though.  Since we live in an interglacial period, for the last three of which the previous ice-core records suggest a span around 10 ka, it has seemed likely that ours wouldn’t have lasted much longer than it already has under purely “natural” conditions.  Modelling the possible effects of anthropogenic warming on climate that may be about to change anyway within this millennium, has left climatologists undecided about the future.  That blurring is as much to do with the unknown direction that an unstable climate might take and the limitations of modelling, as with knowledge of past events.  So, the more information on past interglacials, the better the chance of getting a “handle” on the climatic frying pan out of which humanity seems to be on the point of jumping.  The European Project for Ice Coring in Antarctica (EPICA), which involves 57 scientists from 10 European countries, has dramatically expanded the scope for comparison with the past by a 3 kilometre core from one of the deepest parts of the Antarctic ice (EPICA, 2004.  Eight glacial cycles from an Antarctic ice core.  Nature, v. 429, p. 623-628).  The potential information that eventually will flow from the core will dwarf that from any previous climatic research project.  It covers the period when climate settled into a roughly 100 ka rhythm, probably linked with the weakest of the astronomical controls of solar heating, that of orbital eccentricity, and thereby a bit of a mystery even if it twangs the harmonics of purely terrestrial climatic processes.

The first focus, naturally enough, is on the fourth interglacial epoch before the present one, which ended about 400 ka ago.  In terms of overall astronomical forcing, that is the time when insolation patterns were most similar to those during the Holocene.  Vostok only covered the latter stages, but now its entire span is covered.  All the preliminary time-series for it indicate that it was considerably longer than the last three interglacials, around 25 ka rather than 10.  Its initiation following the waning of the preceding full glacial period follows a similar patter to the early Holocene; the warming was interrupted by a sudden, one-off cooling, somewhat like the Younger Dryas around 12 ka ago.  Although the first EPICA report contains preliminary ideas on several important topics, the one that has caused a stir is that duration of the 5^th interglacial.  Maybe out own warm times will be naturally prolonged for several more millennia, in which case fears of instability and a plunge to full glaciation soon could be set aside with some relief.  However, the abstract to the article, concludes ny saying, “…our results may imply that without human intervention, a climate similar to the present one would extend well into the future” [my italics].  But we do intervene, and nobody knows the outcome of that on a climatic pace of change that follows the almost infinitesimally small orbital-obliquity forcing of probable oceanic process that really call the tune.

Smoking gun for end-Palaeocene global warming: an igneous connection

The sudden warming of the Earth at the start of the Eocene 55 Ma ago has been a topic touched on several times in EPN.  It is widely regarded as a consequence of rapid release of methane from sea-floor gas hydrate, a risk that modern anthropogenic warming presents if deep-water temperatures rise much above their present near-freezing temperatures.  However, no evidence gives a direct connection to the “clathrate gun”.  The disturbance in carbon isotopes of marine sediments at the P-E boundary is most easily linked to a massive methane release at the time, but precisely where it began has been unknown.  Many shallow marine basins, such as the North Sea, have a pockmarked modern floor attributed to minor gas release in much more recent times.  The phenomenon can destabilise the sea bed, so more recent releases have been carefully documented where oil-production platforms are situated.  A clue to the much larger release at 55 Ma stems from detailed seismic exploration of western Norway that involved over 150 thousand kilometres of profiling (Svenson, H. et al. 2004.  Release of methane from a volcanic basin as a mechanism for initial Eocene global warming.  Nature, v. 429, p. 542-545).  The surveys revealed that beds immediately beneath the base of Eocene sediments are riddled with hydrothermal vents complexes, which take the form of mounds, craters and eye-shaped structures.  Some are huge, extending to 5 km across. The profiles also show that beneath the vents are pipes of disrupted strata which extend to the depth of a complex of igneous sills of the North Atlantic large igneous complex, itself emplaced at about 55 Ma.  The sills underlie about 80 thousand square kilometres and most of the vents occur within this area.  Biostratigraphic dating of the youngest sediments disrupted by the vents gives ages between 55.0 and 55.8 Ma.  Intrusion of magma into a deep sedimentary sequence unsurprisingly would set hydrothermal circulation going.  If, as they did, the hot fluids reached the sea bed, they would pass through a zone of gas hydrate, destabilise it and release massive amounts of methane to the atmosphere.  In the case of the Norwegian shelf, the intrusions were into deeply buried organic rich rocks, further encouraging methane formation; probably a great deal more than from gas hydrate.  An estimate of 10¹² tonnes of methane generated thermally off Norway is enough to result in a change in carbon isotopes as large as that known from the P-E boundary.  In fact, similar sediments throughout the end-Palaeocene North Atlantic large igneous province are likely to have been “over matured” in this way, and no other explanation for the increase in “greenhouse” gases seems necessary.  The clear connection with large scale magmatism in thick sedimentary basins may help focus ideas about similar methane-related episodes of global warming, such as the C-isotope excursions at the Permian-Triassic and Triassic-Jurassic boundaries, and within Jurassic and Cretaceous sequences.

Earth’s early climate and methane

At the time the Earth accreted, some 4.6 billion years ago, the Sun was less bright than it is now, so that its warming effect was 30% less.  Without some means of retaining in the ancient atmosphere what heat was available, the Earth would have been frigid.  This “faint, young Sun” problem would have persisted into the time when the geological record begins, around 4 billion years ago, slowly increasing in its energy output to its modern level.  Even in the oldest rocks, there is abundant evidence for the dominance of liquid water at the surface in the form of oceans and river transport across continents.  Low solar warming would have made that impossible, and pole-to-pole ice would have made the Earth a highly reflective planet that could never escape glacial condition.  That is, unless the atmosphere contained sufficient “greenhouse” gases to retain far more solar energy than now.  The favoured gas, until recently, has been the same one that dominates fears of global warming today – carbon dioxide – that volcanoes probably emitted throughout Earth’s history.  However, estimates of how much would have been needed to keep the surface free of sea ice and land glaciers, for which there is no evidence until about 2.3 billion years, are extremely high (hundreds of time greater than now).  Levels greater than 8 times present levels encourage the precipitation of iron carbonates in soils, yet soils from the late Archaean and Palaeoproterozoic contain none.  At those times, CO₂ concentrations less than 8 times present ones would not have prevented runaway “ice-house” conditions, so some other gas had to be involved in atmospheric warming.  James Kasting of the University of Michigan, who has been involved in studies of ancient atmosphere and climate for 25 years, summarises the case for methane being the means of keeping Earth free of ice while the sun was fainter in a recent article (Kasting, J.F. 2004.  When methane made climate.  Scientific American, v. 291(1), p. 52-59).  Only about 1000 parts per million of atmospheric methane would have been needed to keep the early Earth ice-free, because its “greenhouse” effect is extremely efficient.  After oxygen rose to become a major atmospheric gas (since 2.2 billion years), heating induced by methane releases has been tempered by its rapid oxidation to CO₂.  At several times in the past, when there were massive methane releases from sea-floor sediments, such as the end of the Palaeocene, that oxidation prevented the opposite problem, a runaway “greenhouse”.  That is “another story”, involving the rise of photosynthesising organisms.  Kasting’s main theme is the role of methane-generating Archaea (once known as archaebacteria) soon after the origin of life.  In the absence of oxygen, rising methane from thriving methanogen communities could itself have produced irreversible heating, were it not for methane’s ability to polymerise to heavier hydrocarbons through photochemical reactions.  That would have produced a “smog” that not only would have acted as a reflector for solar radiation, but would have added chemical “feedstock” to early life.  Kasting gives a fascinating, all-sided summary, but misses what seems to be an obvious point.  Without atmospheric methane, any water on Earth would have frozen soon after it appeared, however that happened, perhaps by outgassing, perhaps delivered by comets.  Without liquid water, life processes cannot develop.  That opens the possibility for a much earlier origin of life, of the methane generating variety, than anyone has dared to speculate on.  Many methanogens metabolise hydrogen and CO₂.  Volcanoes emit small amounts of hydrogen gas, but an even larger source is from sea-floor hydration of ultramafic lavas, common in early times.  Almost certainly the very earliest times would have provided a suitable environment for methanogens to emerge.

Ray Bradbury wrote a seminal political fiction in the 1950s, called Fahrenheit 451.  It is about a repressive regime that tries to snuff out dissent by burning books, the title referring to the temperature (233ºC) at which paper spontaneously bursts into flame in the modern atmosphere.  With no reference to book burning by some future oligarchy, geoscientists have speculated on the possibility of higher atmospheric oxygen contents being able to induce massive conflagration of green vegetation after lightning strikes or meteorite impacts.  One often cited case is at the K/T boundary, where the thin layer that signifies the mass extinction event contains a high proportion of sooty particles.  Late Cretaceous air probably had significantly higher oxygen content than now, generated by pole-to-pole luxuriant vegetation, and the idea of a global wildfire gained much support when first mooted.  During the Carboniferous, there is very good evidence that oxygen levels were as high as 35% compared with 21% today.  It was a time of giant flying insects, whose size is limited by the availability of oxygen. Carboniferous and Permian strata contain much charcoal, which suggests that indeed fires then were a great deal fiercer and more capable of spreading.  They might have destroyed vegetation, despite evidence that the tree-sized plants of the period had developed fire-resistant structures.  Experiments to simulate the effects up to now have used strips of paper in different oxygen levels, and showed a strong correlation between the minimum energy for ignition and oxygen concentration.  US geologists, foresters and engineers have repeated the experiments using a range of natural plant materials as well as paper (Wildman, R.A. et al. 2004.  Burning of forest materials under late Paleozoic high atmospheric oxygen levels.  Geology, v. 32, p. 457-460).  Their results approximately confirm Bradbury’s fictional paper-combustion temperature, but monkey-puzzle (Araucaria) leaves are more easily set alight.  However, the temperature for ignition does not change as oxygen levels increase, although burning is faster.  How natural materials burn depends on their relative proportions of cellulose and lignin, the higher the latter, the greater the temperature for complete combustion.  They behave very differently from paper.  Another finding was that the rate at which burning spreads did not rise as dramatically as expected for Carboniferous conditions.  The limiting factor is moisture content, although that for no-burn does increase with oxygen levels.  This is particularly important for the firing of dead vegetation lying on the surface, which is essential for catastrophic wildfires.  Natural fires are started by lightning, and that occurs during heavy rainfall, when surface debris is thoroughly saturated.  Fires in the canopy would have occurred at higher frequencies and with greater intensities, but the authors consider they would not have seriously threatened plant life.

Six years ago vast areas of Indonesia caught fire after an unusually dry phase in the El Niño – Southern Oscillation (ENSO).  Burning forest and peat deposits swathed a vast area in smoke, but another alarming aspect was the greatest addition of carbon dioxide to the atmosphere in half a century.  Such a wildfire on a global scale is thought to have marked the end of the Mesozoic, perhaps triggered by the K-T impact event and encouraged by higher oxygen content in the atmosphere.  Present oxygen levels seem to be at a balance that staves off spontaneous combustion of green vegetation, but only a few percent more would render vegetation much more prone to bursting into flame.  The end of the Palaeocene involved a sudden global warming that coincides with a decrease in the proportion of ¹³C in marine carbonates.  Since photoynthesis, at the base of the trophic pyramid, favours light ¹²C, such a negative d¹³C “spike” is generally ascribed to an unusually high release of organic carbon to the environment.  The end-Palaeocene warming may have resulted from a massive release of methane from gas-hydrate buried in shallow seafloor sediments (See Methane hydrate – more evidence for the ‘greenhouse’ time bomb and Plankton and the end of the Palaeocene-Eocene global warming August and October 2000 issues of EPN).  However, massive burning of living biomass could also produce the carbon-isotope signal.   Telling the two mechanisms apart requires information from other organic-related cycles.  One key is comparing the carbon- and sulphur-isotopic records that enables the place in which carbon had been stored geologically.  For marine burial, the effect of aerobic bacteria that completely oxidises hydrocarbons back to carbon dioxide and water needs to have been suppressed.  Periods of massive marine carbon burial coincide with oceanic anoxia episodes, when anaerobic bacteria beneath the seafloor reduce dissolved sulphate ions to sulphides, thereby depositing lots of iron sulphide (pyrite) in black organic mudrocks.  This sequesters sulphur that is depleted in ³²S into marine sediments, so that the marine carbon- and sulphur-isotope records fluctuate in a clearly related way.  During the Palaeocene this relationship is absent, while overall the carbon isotopes do signify progressive burial of organic carbon.  The decoupling of the two cycles points to carbon burial on the continents, forming peat and eventually coal deposits.

Playing games on Snowball Earth

For as long as anyone can remember there has been a parade of geoscientific bandwagons in town.  Three of the floats today carry banners saying, “Snowball Earth”, “Climate models” and “continental erosion and CO₂ drawdown”.  Of course there is serious science aboard each, but they are getting overcrowded, especially as separate bands try to jump from one to another.  When it sometimes seems, as now, that the “next Big Thing” is some way off, we get the unseemly spectacle of some bands trying to straddle two or even several of the wagons.  Three is quite a feat, yet the 18 March 2004 issue of Nature contains perhaps not a vast human pyramid, but at least a tetrahedron of the genre (Donnadieu, Y. et al. 2004.  A “snowball Earth” climate triggered by continental break-up through change in runoff.  Nature, v. 428, p. 303-306).  From about 1100 to 750 Ma ago, the bulk of continental lithosphere was gathered in a supercontinent known as Rodinia (from the Russian for “Mother Earth”).  By analogy with modern Eurasia, and the stratigraphic record from the Phanerozoic Pangaea supercontinent, the centre of Rodinia would almost certainly have been dry, being so far from the ocean.  Break-up of that continental mass would also probably have allowed moist maritime air to penetrate over a larger proportion of the fragments.  The hypothesis that Donnadieu and colleagues try to test using linked geochemical and climate models is that such a tectonic change would increase continental weathering and reduce the “greenhouse” effect.  The weak acid formed by solution of carbon dioxide in rain water can provide hydrogen ions to break down silicate minerals.  The reactions contribute bicarbonate and soluble metal ions to surface and subsurface water.  Ultimately, both reach the oceans and contribute to its chemistry.  If conditions are suitable, calcium ions in particular combine with bicarbonate to precipitate calcium carbonate on the ocean floor, either through the action of organisms or inorganically.  The two chemical equilibria involved result in a net burial of one carbon atom out of the two involved in the weathering, thereby drawing down carbon dioxide from the atmosphere.  The climate model used in their cyber-experiment resolves the Neoproterozoic Earth into cells that are 10 x 10 degrees (about 100 thousand km²) and considers Rodinia at 800 Ma and the result of its break-up at 750 Ma, the time of the first good evidence for extensive low-latitude glaciation.  The results, after some tinkering, suggest that increased continental weathering could have reduced CO₂ levels to 250 parts per million.  Taking account of a 6% less energetic Sun at the time, this would have produced sufficient cooling for ice caps to exist to sea level at the equator.  So, taken at face value, the hypothesis seems plausible.  However, there are major snags.  First, in a mere 50 million years their model sees continental dispersion on a scale that has not yet happened to Pangaea in about 200 Ma of Phanerozoic time.  Second, since continental area remains constant, the proportion of rainfall, and therefore weathering and runoff, involving continental crust also stays fixed.  Third, continental weathering refers to the crystalline part of its crust, in which there are unstable minerals, such as feldspars, that can do the chemical trick.  We have little idea how much of the continents at that time was veneered by sediments that are the products of earlier chemical weathering, and contribute nothing to the process.  Exposing such deep crust depends to a large extent on mountain building, which continental extension does not encourage.  Fourth, carbon dioxide is not the only source of hydrogen ions that are involved in weathering, especially as much of it goes on in groundwater – bacterial action and oxidation of iron sulphides create much more acid conditions that rainwater.  Fifth, and most important, where is the complementary geochemical evidence?  Feldspars of the continental crust, on which the hypothesis mainly rests, have high contents of rubidium compared with their oceanic counterparts, and they are old.  Much of Rodinia was underpinned by crust formed as far back as 4 billion years ago.  Prolonged decay of ⁸⁷Rb to radiogenic ⁸⁷Sr makes the strontium isotopes of continental material very different from those of the ocean floor – it has a much higher ⁸⁷Sr/⁸⁶Sr ratio.  Since soluble strontium would be released to runoff by continental weathering, that signature makes its way to the ocean and should pop up in marine carbonates.  Although the ocean strontium isotopes in the Neoproterozoic did rise a little, it did not peak until the very end.  In fact, the details show that the periods around supposed “snowball” conditions involved downturns in radiogenic strontium supply to the oceans.  Whatever the model suggests, all that it amounts to is the equivalent of a table-top train set

The vast reserves of peculiar methane-water ice deposits (gas hydrate or clathrate) in sea-floor sediments are the most likely source of methane releases that could generate sudden warming events, such as that at the end of the Palaeocene, and left traces in polar ice cores during the last few glacial-interglacial episodes.  Methane probably leaks from the sea floor all the time, but is soon oxidised to the lesser “greenhouse” gas CO₂ in the atmosphere, so muting its potential effects to a low background level.  For methane to have a sizeable effect on global warming, lots of it has to blurt out suddenly.  Possibly the only mechanism that can trigger such explosive releases are failures of sea-floor sediments, either by those beneath a steep surface slope collapsing under gravity, or as a result of seismicity.  Geoscientists from University College London and the British Geological Survey have tried to correlate known peaks in atmospheric methane from the recent past (shown by ice cores) with episodes of mass flow on the seabed (Maslin, M. et al. 2004.  Linking continental-slope failures and climate change: Testing the clathrate gun hypothesis.  Geology, v. 32, p. 53-56).  They found that the periods of greatest disturbance of continental-slope sediments over the last 45 ka took place at the tail-end of the last glaciation, between 13 and 15 ka and 8 to 11 ka.  Each correlates with methane highs in the Greenlandic ice cores and with bouts of rapidly rising sea level (the Bølling-Ållerød and Preboreal warming periods).  So they conclude that there is support for a “clathrate gun” model for sudden warming associated with glacial to interglacial transitions.  However, seafloor collapses also correlate with Heinrich events (ice-sheet surges that launched iceberg “armadas” to low latitudes) that punctuated glacial times.  These marked brief periods, repeating every 1000 years or so, which mark cooling when sea-levels were low.  None are associated with upsurges in atmospheric methane., although the following interstadial warmings are.  This lack of correlation rules out a “clathrate gun” influence on millennial-scale climate fluctuations during glaciations.

Super-eruptions and climate

The biggest known, young volcanic crater is that of Toba on Sumatra, which is a caldera complex measuring 30 x 100 km.  Around 74 ka Toba emitted an eruption that dwarfed any in more recent times, and spread a dust cloud around the world – it is present in ice cores from Greenland, and has been linked with a cooling step during the onset of the last glaciation.  It happened around the time that fully modern humans had begun to spread across Asia after migrating from NE Africa – an Acheulean hand-axe has been found in the Toba Tuff – and may have deeply affected those pioneering bands.  There are older ash levels that can also be attributed to Toba eruptions, one found 2500 km away in the sediments of the South China Sea (Lee, M-Y. et al. 2004.  First Toba supereruption revival.  Geology, v. 32, p. 61-64) and at other sites up to 3000 km from Toba.  This gives an age around 800 ka.  Lee and colleagues from Academica Sinica (Taiwan), the National Taiwan University and the University of Rhode Island estimate that almost 1000 km³ of ash was expelled by the eruption.  Unlike the 74 ka ash, this layer falls in the transition from a glaciation to an interglacial period; instead of a possible cooling influence through dust blocking solar heating, there is a warming trend.  Although not quite as big as the 74 ka eruption of Toba, that of 800 ka is still vastly bigger than any other explosive volcanism during the Pleistocene.  So, it suggests that super-eruptions are not significant climate triggers after all.

Since the resurrection of Chamberlin’s idea that the rate of chemical weathering of continental crust helps regulate atmospheric CO₂ by Maureen Raymo, the hypothesis has not yet been supported by convincing geochemical evidence.  There is such a lag between changes in ocean chemistry and evidence for global climate change, that correlations are flimsy.  The need is for a proxy for weathering of the land surface that resides in seawater for a geologically very short period.  Such an element is osmium (Os), which passes from river water through the oceans to sea-floor sediments in about 25 thousand years, so changes in its abundance in sediments ought to match the pace of any climatic shifts.  In principle, there are two main sources for elements in seawater, from sea-floor hydrothermal alteration of oceanic crust, and from continental weathering.  The first can be considered to be more or less constant, except on time scales of tens of million years.  Continental weathering is a response to climate change, and keeps pace with it.  Researchers at the UK Open University and the University of Köln in Germany analysed samples for osmium and carbon isotopes through a sequence of Jurassic mudstones on the NE coast of England (Cohen, A.S. et al. 2004.  Osmium isotope evidence for the regulation of atmospheric CO₂ by continental weathering.  Geology, v. 32, p. 157-160).  The carbon isotopes show a sudden drop in d¹³C within a very hydrocarbon-rich unit famous for it contribution of jet (oil-rich lignite) to Victorian funereal jewellery.  This negative excursion is recognisable world-wide at around 180 Ma.  The most likely explanation is a monstrous blurt of methane from destabilised gas hydrate on the Jurassic sea floor (see Methane hydrate – more evidence for the ‘greenhouse’ time bomb, August 2000 issue of EPN).  The Jet Rock of the Whitby coast therefore preserves a nice example of sudden climatic change, and by the end of its deposition carbon isotopes returned to Jurassic background values.  Methane, a powerful “greenhouse” gas, is rapidly oxidised to CO₂ in the atmosphere, so reducing its initial warming effect, but climate would have been hotter for some time afterwards until the excess CO₂ was drawn down somehow.  Interestingly, the Jet Rock also shows a sudden leap in the abundance of ¹⁸⁷Os, reflected in the ¹⁸⁷Os/¹⁸⁶Os ratio of the samples, and an upward step in the value of the ⁸⁷Sr/⁸⁶Sr ratio – one of the fastest rises known.  The latter is generally assigned to an increase in continental weathering, since continental crust contains more radiogenic ⁸⁷Sr than does oceanic crust.  The implication of the osmium-isotopic shift is odd; it requires an increase in the rate of continental weathering by 4 to 8 times that in the preceding period.  That is a vast change, even if it only lasted for a short period, but it tallies with what is known about the temperature dependence of the dissolved loads of rivers in more recent times.  If the osmium isotope excursion truly reflects massive continental weathering, then it is possible to calculate the drawdown of the excess CO₂ in the atmosphere from a commensurate flux of calcium and magnesium ions from the continents, that would eventually form marine carbonates.  The authors estimate a mere 37-123 ka to get rid of it.  Yet continent-derived radiogenic ⁸⁷Sr remained high for much longer, and the authors’ arguments become tricky.  One interesting aside is that, unlike today, more groundwater found its way to the oceans than surface run-off during the Jurassic, perhaps 6 times more.  It is easy to look on weathering as what happens at the interface between rocks and the weather; the land surface.  Not so.  A great deal of chemistry that releases soluble ions goes on in the subsurface, above and below the water table.  It is by no means as simple as reactions between carbonic acid in rainwater and silicate minerals.  Weathering is the product of hydrogen ions’ (whatever their source) effects on silicates.  Bacteria are extremely important actors in modifying pH below the surface, for example the sulphate-sulphide reducers, and the oxidative dissolution of sulphides produces sulphuric acid.  Even more interesting for the chemistry of groundwater is the curious role of iron hydroxide.  Under oxidising conditions it adsorbs many elements from solution, including platinum-group elements, such as osmium.  Should conditions become reducing, dissolution of goethite skins on sedimentary grains releases the accumulated elements.  A warming trend almost inevitably results in increased precipitation, and rising water tables.  It also should boost biological productivity on land and an increase in the amount of buried organic matter, which create reducing conditions in groundwater.

Whether or not you believe that the Earth was totally encased in ice up to four times during the Neoproterozoic Era, there is convincing evidence that ice sheets did extend to the tropics during such “Snowball” episodes.  How such extremely cold episodes came to prevail for several million years has been the subject of debate for 5 years, since Harland’s notion of global glaciations was resurrected by palaeomagnetic evidence for the low latitudes of Neoproterozoic glaciogenic rocks.  Ice extending almost to the Equator, even if just on the continents, would have driven down global temperatures simply because it would have reflected away solar radiation.  Increased albedo helps explain why frigid conditions lingered, but some other cooling mechanism must first have encouraged the widespread formation of ice sheets.  Essentially, the supply of the “greenhouse” gas CO₂ by volcanic activity must have been outstripped by burial or solution of carbon in some form.  The two usually identified candidates are increased deposition of carbonate sediments and the accumulation of unoxidised organic carbon in sea-floor muds.  It is the first of these that dominates climate control today, by the accumulation of carbonate shells of marine plankton, and that has probably prevailed since foraminifera and coccolithophores began to proliferate in the Mesozoic.  No shelled organisms existed during the Precambrian, so a major factor in damping down climate fluctuations was missing before the start of the Phanerozoic.  This crucial difference between the modern and Precambrian world focussed the attention of Andy Ridgwell, Martin Kennedy (University of California) and Ken Caldeira (Lawrence Livermore National Laboratory) in seeking an explanation for “Snowball” events (Ridgwell, A.J. et al. 2003.  Carbonate deposition, climate stability and Neoproterozoic Ice Ages.  Science, v. 302, p. 859-862).

Carbonate sediments are plentiful in the Precambrian record.  Some formed as a result of organic action (stromatolitic limestones) and others show evidence for direct, inorganic precipitation of carbonates from sea water.  The latter indicate sea water in which calcium and carbonate/bicarbonate ions exceeded the solubility of calcite and the ability of organic activity to remove calcite from solution.  Evidence for such extreme oversaturation is rare, but the cap carbonates that overlie Neoproterozoic glaciogenic rocks are important examples.  The key area of carbonate deposition has always been on shallow continental shelves, the main secreters of carbonates during the Precambrian having been blue-green bacteria that can photosynthesise only in shallow water.  Falls in sea-level or a reduction in the area of shelves during the Phanerozoic reduced this sink for CO₂ in the build-ups of coral and shelly limestones, but plankton of the open oceans continued to accumulate on the deep sea floor.  Because calcite can be dissolved at depth, the deepest sea floor does not contain much carbonate.  However, a fall in sea level,  increases the area suitable for deep-water burial of shelly material, because the carbonate compensation depth or lysocline also falls.  In the absence of shelly plankton, this modern balancing mechanism for ocean chemistry did not exist during the Precambrian.  Superficially, it might seem that a reduction in the area of shelf deposition of carbonates, brought on by a sea-level fall, would allow CO₂ to build up in the atmosphere, driving towards warmer conditions.  However the way in which atmospheric carbon dioxide is related to dissolved carbonate (CO₃^2-) and bicarbonate (HCO₃^–) ions tells a very different story.  This is the equilibrium: CO₂ + CO₃^2- +H₂O = 2HCO₃^–.  Less carbonate accumulation on reduced continental shelves would drive up the carbonate-ion concentration of sea water, and also its pH.  So, according to Le Chatelier’s Principle, the equilibrium proceeds to the right and adds to the more soluble bicarbonate ions in sea water.  This consumes CO­₂, and drives down the “greenhouse” effect.  Ridgwell and colleagues developed a model around this equilibrium, and applied it to conditions of falling sea level when carbonates were only deposited on continental shelves.  Their results show that decreased shelf-carbonate burial during a period of sea-level fall would rapidly drive down the warming effect of atmospheric carbon dioxide.  Combined with the lower solar energy output during the Neoproterozoic, that would be sufficient to create protracted periods of frigidity.  Alkalinity of the oceans would increase through periods of glaciation, so that once sea-level rose, massive carbonate precipitation would form cap carbonates on the newly inundated shelves, thereby reducing the oceanic drawdown of CO₂.

Ridgwell et al’s model is not easy to grasp, and relies on its initiation by falling sea-level.  Either that resulted from build up of continental glaciers because of some other climatic mechanism, or internal processes increased the volume of the ocean basins.  An example of the last is a decrease in sea-floor spreading, when cooling of the lithosphere increases it density so that it sags down.  Periods of accelerated creation of oceanic lithosphere displace sea water upwards, and perhaps that might explain an increase in shelf areas, which would allow warming according to the new model.  The model also needs special pleading to account for the 1 billion-year absence of glaciation before the period of Snowball events.  The authors suggest that it could have been prevented by much wider shelves during earlier times, but without quoting evidence.

Continental erosion and climate

Maureen Raymo suggested in 1988 that long term climate change was modulated by the rise of mountain chains and their erosion and weathering.  This is because chemical weathering of silicate minerals is a net consumer of atmospheric carbon dioxide.  Raymo’s hypothesis, based on T.C. Chamberlin’s theory of glaciation, has set climatically concerned geochemists to analysing the trace element content of river water in many mountainous regions, because those such as strontium are proxies for the amount of weathering going on today.  Others have looked at the flux of elements into seawater through the Phanerozoic in particular, by analysing marine carbonates, to see if the ups and down’s of water composition through time match the record of climate change.  These time series do suggest some matching, but not precise enough for all to agree with the hypothesis.  Measurements of river-water composition have also met set-backs.  Much of the weathering flux from mountains seems to stem from dissolution of carbonate rocks, and that does not lead to long-term loss of CO₂ from the atmosphere.  In a bid to resolve the contributions of carbonates and silicates, Andrew Jacobson and Joel Blum of the University of Michigan have studied the flux from part of the Alps of New Zealand’s South Island (Jacobson, A.D. & Blum, J.D. 2003.  relationship between mechanical erosion and atmospheric consumption in the New Zealand Southern Alps.  Geology, v. 31, p. 865-868).  Their area is a good choice because the New Zealand Alps are actively rising, precipitous and drenched with continual heavy rain and snowfall. Moreover, they offer something that the Andes and Himalaya do not; the rocks are pretty uniform.  What they find will not please Raymo’s followers.  As in many mountain ranges, mechanical erosion favours carbonate weathering over that of the CO₂ sequestering alteration of silicates.  With a low ratio of  silicate:carbonate chemical weathering, mountain building in New Zealand does draw down carbon dioxide, but only by a factor of about 2.  They conclude that more stable areas with lower relief are more likely to affect climate.  Although chemical weathering in them is lower than in mountains, that of silicates is far higher than for carbonates.  Moreover, active mountain ranges are minuscule compared with the extent of more subdued land.  It seems likely from Jacobson and Blum’s findings that the major control of weathering over climate depends to a large degree on where continents are located relative to warm, humid climatic zones.  For much of the early Cenozoic, the dominantly crystalline Precambrian shields of India, Africa, Australia and South America straddled the Equator, and witnessed intense weathering.  Maybe that relationship helped draw down carbon dioxide, and gradually cooled the planet from the hot and humid climate of the late Mesozoic.

Ten years ago the records of climate proxies from the Greenland ice sheet set new benchmarks for understanding how climate has varied over the last 100 thousand years – annual ice layers allowed division of that data to as fine as decades.  Variations in the ice cores helped explain many of the variations found in more blurred data from sea-floor sediment cores in the Northern Hemisphere.  Variations could be correlated with changes in the formation of North Atlantic deep water at high latitudes and the destabilisation of North American and Scandinavian glaciers.  The whole hemisphere behaved in concert, through long-distance connections in climatic processes, but high-latitude processes seemed to dominate.  Development of ²³⁴U/²³⁰Th dating extended high precision to carbonates that have been precipitated from groundwater to form stalagmites or speleothem.  The latest results from speleothem, collected on the Indian Ocean island of Socotra, cover 14 thousand years between 56 and 42 ka, and resolve down to only 8 year intervals (Burns, S.J.  et al. 2003.  Indian Ocean climate and an absolute chronology over Dansgaard/Oeschger events 9 to 13.  Science, v. 301, p. 1365-1367).  They show variations in rainfall on the island, though the d¹⁸O proxy, and thus changes in the strength of the Indian Ocean monsoon.  In terms of shape, the stalagmite record closely resembles d¹⁸O changes in the Greenland ice cores, although the two have opposite senses, because the Greenland proxy is for air temperature above the ice cap.  During the frigid Heinrich events that saw massive southward waves of icebergs, rainfall over Socotra was low.  It became higher as high-latitude conditions warmed in Dansgaard-Oeschger events.  The fine speleothem resolution shows a dramatic change-over that took only 25 years or so.  The explanation is that warmer conditions increased equatorial evaporation from the oceans.  But water vapour is the dominant “greenhouse” gas, and a wetter atmosphere would become warmer.  So the question of whether low- or high latitudes drove the changes is still an open one.  If North Atlantic events were the driver, then the tropical processes would greatly amplify their effects.  One big problem emerges from the joint research by US, Swiss and Yemeni scientists.  The highly reliable U/Th dating gives ages for each event that are about 3000 years older than those interpreted from the ice cores.  The authors are convinced that the ice-core ages need revision, yet there are discrepancies with the event-ages from other similarly dated speleothems.  Commenting on the paper, Frank Sirocko of Johannes Gutenberg University of Mainz in Germany (Sirocko, F. 2003.  What drove past teleconnections.  Science, v. 301, p. 1336-1337) makes the point that maybe the quality and age of ice core records lie behind the widely accepted view that high-latitude process drive climate.  He presents an excellent global image of modern sea-surface temperatures that show the main oceanic shifts of energy – the leakage of cold circum-Antarctic waters northwards, the westward movement of equatorial warm waters to which the El Niño – Southern Oscillation (ENSO) is due, and the unique movement of warm water to Arctic regions in the North Atlantic that is connected to deep water formation.  To that he adds the major effect of continental winter snow cover in central Eurasia, that affects albedo and the size of the winter high-pressure zone there.  Is there a teleconnection between that and events in the North Atlantic?  Nobody knows, because there are no data to compare, yet.  Another uncharted but likely linkage is between the ENSO and processes in the circum-Antarctic current.  Using currently accepted dating of ice cores, records from those in the Antarctic show air temperature changes that precede those from Greenland by several thousand years.  In that respect, the Socotra record possibly has a link with the South Polar climate.  Until the issue of dating is sorted out, it will always be difficult to make concrete statements about global climate change.

Interestingly, in the same issue of Science, sea-floor data (between 9 and 16 ka) from the Cariaco Basin off Venezuela, at about the same latitude as Socotra, mimic the Greenland records to within 30 to 90 years (Lea, D.W. et al, 2003.  Synchroneity of tropical and high-latitude Atlantic temperatures over the last glacial termination.  Science, v.  301, p. 1361-1364).

“Greenhouse” controls challenged

There’s data gathering and there’s theorising.  In palaeoclimate studies the two come into conflict.  Theory suggests that CO₂ is likely to be the principal driver for climatic ups and downs, probably on all time scales.  Atmospheric CO­₂ estimates from the past are based on proxies of different kind, and the various models that they support do not tally vary well.  Worst of all they do not fit climate records through the Phanerozoic at all well, except in the crudest possible way.  Only the long-lived Carboniferous to Permian “icehouse” and Tertiary cooling tally, and then only in Berner’s GeocarbIII model.  One of the best records of major climate shifts, aside from continental tillites, are marine sediments that contain ice-rafted debris, in particular the palaeolatitudes to which they extend.  They record four major cooling episodes: Late Ordovician; Devonian to Late Permian; Late Jurassic to Mid Cretaceous; and those since about 35 Ma ago.  The oxygen isotope record from Phanerozoic fossils, partly correlated with ocean temperatures also suggest 4 global coolings in the last 545 Ma.  Either the CO₂ modelling needs more detail, or the whole issue of the “greenhouse” effect is under question.  That is the conclusion of a study by Nir Shaviv of the Hebrew University of Jerusalem, and Ján Veiser of the Ruhr University and The University of Ottawa (Shaviv, N.J. & Veiser, J.  2003.  Celestial driver of Phanerozoic climate?  GSA Today, Huly 2003, p. 4-10).  Veiser has been analysing the chemistry of carbonates, especially their oxygen isotopes, for his 30 year career, and has amassed more data than any other geochemist on carbonate-related issues.  The two have worked together because their interests fit together extremely well.  Shaviv has reconstructed the variation of cosmic ray flux from studies of the exposure of iron meteorites to them, blended with analysis of how the Solar System moves through the spiral arms of our galaxy.  Cosmic rays are known to affect the Earth’s cloudiness and therefore albedo.  Greater cosmic ray flux should increase the amount of solar energy reflected away by the Earth, thereby causing global cooling.  The degree of fit between the cosmic ray flux and palaeoclimatic records is so good that up to 2/3 of climate variation may be connected with the Earth’s celestial position.  That is, as it passes through the star-rich spiral arms cosmic rays intensities go up.  This happens every 140 Ma or so, which fits very well with the 4 icehouse periods during the Phanerozoic.  They even suggest that the climate-CO₂ relationship may be the opposite of that generally agreed; climate might drive carbon dioxide levels.  A secondary role for “greenhouse” gases wreaks havoc on attempts at modelling climate change feared to result from increasing anthropogenic releases.  Shaviv and Veiser’s work comes at a particularly awkward time for climate modellers, who have just initiated a programme for  running huge simulations by corralling the combined computing power of millions of home PC users, similar to the approach pioneered by the SETI Institute (Allen, M.R.  Possible or probable.  Nature, v. 425, p. 242).  Perhaps the view of Phillip Stott, that climate modelling is a complete waste of time (Stott, P. 2003.  You can’t control the climate.  New Scientist, 20 September 2003, p. 25) might sink in as a result of the possible link between cosmic ray flux and climates of the past.  Stott believes that acting on the output of such models might perhaps even be dangerous, since we clearly do not understand short-term climate change well enough.

Precambrian CO₂ levels

Whether or not fluctuations in the “greenhouse” effect drive climate change, the fact remains that CO₂, methane and water vapour all act to retain solar heat in the Earth system.  Were it nor for their presence in the atmosphere, the Earth would be about 33 degrees colder than it is.  It would be covered by ice.  Theoretical modelling of how stars evolve suggests that the Sun had progressive less energy output going back in Earth’s history.  Only gaseous heat retention could have prevented a sterile, frigid planet.  Yet periods of cooling sufficient to hold large amounts of water in surface ice have occurred only a few times, 4 in the Phanerozoic, a flurry of so-called “Snowball” epochs in the Neoproterozoic and the earliest known glaciation around 2200 Ma ago.  The earliest coincided with the first evidence for free oxygen in the atmosphere, and may have been caused by that.  Methane, a more powerful “greenhouse” gas than water or carbon dioxide and abundantly produced by anaerobic decay, is easily oxidised.  In later time, it has been ephemeral in the atmosphere, unless continuously released, for instance by destabilisation of gas hydrate in sea-floor sediments.  Warming by CO₂ has undoubtedly kept total frigidity at bay since then.  The problem is charting just how much was in the air, because most estimates have been based on studies of palaeosols that give odd and very imprecise results for the early Palaeozoic (see Shaviv and Veiser, 2003; previous item).

Photosynthetic organisms derived their carbon from CO₂, either in the air or dissolved in water through equilibration with the atmosphere.  The extraction favours lighter ¹²C, so biological activity results in their products being depleted in the heavier ¹³C by about 25 parts per thousand (‰) relative to carbon in air and water.  If organic carbon becomes buried, the remaining carbon in the surface environment gets richer in ¹³C, and that signature becomes fixed in contemporaneous carbonates, both organic and inorganic.  It is therefore possible to use the two carbon-isotope signatures to estimate the reservoir of CO₂; its proportion in contemporary air. However, the degree of fractionation depends on the specific carbon metabolism of different organisms, yet most organic carbon in sediments is a mixed product of widely differing life styles.  That severely blurs estimates of atmospheric carbon dioxide content.  What is needed are data from a single source with known metabolism.  Acritarchs are fossil remains of single-celled marine eukaryotes that were, and still are, marine photosynthesisers.  They are made of degraded hydrocarbons.  Advanced ion-microprobe resolution is now sufficient to produce carbon-isotope measurements of individual fossils (about 200 micrometres across).  Sediments from northern China, roughly 1400 Ma old, contain abundant little-altered acritarchs and carbon isotope data from them give good estimates of atmospheric CO₂ levels, that are independent of other methods (Kauffman, A.J. & Xiao, S. 2003.  High CO₂ levels in the Proterozoic atmosphere estimated from analyses of individual microfossils.  Nature, v.  425, p. 279-282).  The estimates suggest between 10 to 200 times higher contents than today, but just about sufficient to keep the Earth above the limit of glacial temperatures when solar luminosity was about 88% of the present.  Acritarchs are present throughout the Neoproterozoic, and it should prove possible to examine the critical periods of “Snowball” conditions using this method.

The main driver for biological activity in the oceans far from land is the availability of iron, and this helps control the burial of organic carbon and hence aspects of global climate.  At low Fe concentrations, as they have been since the oxygenation of the surface environment from 2 billion years ago, iron is cycled in the marine environment in a matter of a few hundred years.  So, ocean water responds very quickly, in geological terms, to changes in the source of any dissolved iron.  There are two main sources, discharge of hydrothermal fluids from the oceanic lithosphere and delivery of river water and dust derived from the continents.  Of the last, riverine sources probably end up in near-shore sediments and only dust contributes significantly to deep ocean water.  The slowly growing nodules and crusts, composed mainly of iron and manganese compounds, on the ocean floor can chart variations in the relative proportions of these sources, because their growth produces zonation.  Measurements of d⁵⁶Fe in various materials show that the two sources are different in isotopic composition (Beard, B.L. et al. 2003.  Iron isotope constrains on Fe cycling and mass balance in oxygenated Earth oceans. Geology, v. 31, p. 629-632).  While continent derived materials exude iron that is essentially the same as that in terrestrial volcanic rocks (d⁵⁶Fe ~0.0‰), ocean-floor hydrothermal activity is significantly depleted in ⁵⁶Fe (‰⁵⁶Fe ~ -0.38‰).  From 6 Ma to 1.7 Ma iron-manganese crusts record iron with a dominant hydrothermal origin, but during the glaciation-dominated period since 1.7 Ma the contribution of continent-derived dusts becomes overwhelming – cooling forces drying on a global scale.  Because hydrothermal contributions probably stay much the same over very long periods, because of the sluggishness of plate tectonics, iron isotopes in deep marine sediments, such as Fe-Mn crusts,  may be important tracers for glacial events in the distant past, such as the glaciations during the Neoproterozoic and Palaeozoic. Interestingly, the largest iron-rich deposits on the planet, the BIFs that peaked during Archaean and Palaeoproterozoic times, record far larger excursions in iron isotopes than any other.  The very low d⁵⁶Fe values of some BIFs (down to – 2.4‰) probably signify the dominance of sea-floor sources, although a non-oxidising atmosphere would have mobilised dissolved iron from the continents too, which explains the range in BIFs up to +1.0‰.

The gas-hydrate “gun”

As fears of anthropogenic climate warming have risen, so more geoscientists have looked in detail at the stratigraphic record for signs of past warming, and funds have become more targeted towards palaeoclimatology.  One of the most important discoveries was that the end of the Palaeocene, about 55 Ma ago, was a time of sudden global warming during the overall cooling that has characterised the Cenozoic.  The first sign that something strange had happened then came from using the oxygen isotope geothermometer on plankton tests from marine drill core that passed through the boundary.  There seemed to have been a 7º C jump in surface seawater temperature.  An explanation for the thermal spike arose after carbon isotopes revealed a coincident spike in the lighter ¹²C.  Periods of low primary biological production can impose such anomalies, because photosynthesis selectively binds light carbon in carbohydrate.  However, some of that light carbon ends up buried in sea-floor sediments, so another explanation for a negative excursion in d¹³C is that organic carbon has somehow been released from sedimentary storage to the atmosphere.  So, either there was a sterile ocean or a massive release of organic carbon at the Palaeocene/Eocene boundary.  Some kind of erosion to achieve the second possibility could not have led to such a speedy shift in carbon isotopes.  The accepted explanation, suggested in 1995, stemmed from organic carbon that had been metabolised by methanogen bacteria in anaerobic sea-floor sediments to form methane.  Given low enough sea-bottom temperatures and sufficient pressure, methane can crystallise with water to form an icy substance, known as gas-hydrate or clathrate, in sea-floor sediments.  Being an unstable compound, gas hydrate can break down rapidly if seafloor temperature rises or sea-level falls.  And, of course, the methane can rush to the surface as bubbles.  Being 4 times more efficient than carbon dioxide at trapping thermal radiation emitted by the Earth’s surface, methane releases are excellent explanations for sudden warmings in the stratigraphic record.  And there is a great deal of methane locked as gas hydrate beneath the sea floor, about 2 teratonnes (2 x 10¹² t).  Quirin Schiermeier reviews the basic concept (Scheiermeier, Q. 2003.  Nature, v. 423, p. 681-682), but poses the question of how methane-induced warming is reversed.  Methane is quickly oxidised to CO₂ in the atmosphere, so lessening its warming effect.  So a “spike” that lasts thousands of years has to be fed by continual releases.  Since warming drives gas hydrate breakdown, something must intervene to stop the releases before the warming becomes a “runaway greenhouse”.  One view, and probably the correct one, is that warmth and more CO₂ drives up biological activity so that the increased atmospheric carbon is “pumped” down by living processes, back to sedimentary burial.  If sufficient nutrients are available, there is no way of stopping this negative feedback until a balance is restored.  Schiermeier reports that new ocean drilling plans to test the hypothesis that the Palaeocene/Eocene warming accelerated continental erosion, which was able to wash the crucial nutrients phosphorus and iron into the oceans.  Experiments have shown that increased iron in ocean-surface water far from land – now pretty sterile because it is iron-deficient  – sparks up photosynthetic plankton.  That is one possible way of artificially drawing down anthropogenic CO₂.  The problem is, if such a process was involved in cooling the Eocene Earth, it took about 100 thousand years.

Red Sea record links to northern hemisphere climate

In his forthcoming book, Out of Eden: the Peopling of the World (Constable and Robinson, July 2003), Stephen Oppenheimer offers the novel suggestion that fully modern humans left Africa by island hopping on log rafts across the Straits of Bab el Mandab, which connects the Red Sea to the Indian Ocean.  The rationale to his suggestion is that sea-level falls during major glaciations would have partially exposed the shelf that lies beneath the Straits, presenting a route to SW Arabia across only 18 km of island-dotted sea. As today, it would have been impossible to trek across the deserts of the Middle East after a northward African migration along the Nile, without chains of wells.  His thesis then sees humans migrating along coasts eventually to reach east Asia at about 70 ka.  Precisely when the Straits of Bab el Mandab became shallow enough would have been determined by global climatic conditions, for only glacial maxima result in sufficient sea-level falls for such island hopping to be possible. 

The shallowing of the shelf across the southern outlet of the Red Sea would have had a profound impact on seawater circulation.  Already having restricted connection to the world’s oceans, Red Sea water has elevated ¹⁸O levels, because evaporation from it favours loss of lighter ¹⁶O.  With more restricted circulation, evaporation would have driven this up further.  Geoscientists from the Universities of Southampton, Tuebingen and Göttingen, and the Geological Survey of Israel have analysed the variation in oxygen isotopes of foraminifera from a Red Sea core to quantify ups and downs in  sea level in more detail than possible from open-ocean cores, which have uncertainties of about ±30m) (Siddall, M. and 6 others 2003.  Sea-level fluctuations during the last glacial cycle.  Nature, v. 423, p. 853-858).  The method that they used models the effects on Red Sea oxygen isotopes of evaporation and changed circulation to estimate how the depth of the Straits of Bab el Mandab changed.  They claim a precision of ±12m.  Through the period from 70 to 20 ka, leading up to the last glacial maximum, their sea-level record tallies nicely with climate records from both Antarctic and Greenland ice cores, including shifts linked to the short-lived Heinrich and Dansgaard-Oeschger cycles. During the last glacial maximum(18-20 ka), sea-level fell by almost 120 m, so that the Straits of Bab el Mandab were on average only 15 m deep.  The first human Exodus out of Africa to populate Eurasia would have been between 120 to 130 ka, as suggested by Oppenheimer, when sea level probably fell a little further.  However, at about 65 ka, sea level dropped to about 100 m below modern levels, perhaps presenting another window of opportunity.

Broecker reviews climate triggers

Wallace Broecker, of the Lamont-Doherty Earth Observatory at Columbia University, was the first to quantify in 1975 the 19^th century prediction of Svante Arrhenius that increasing atmospheric carbon dioxide would drive up global temperatures.  Broecker’s early work lies at the centre of concern about global warming, and his subsequent contributions are enmeshed with the entire study of past climate change.  A review by him of current ideas on palaeoclimates of the recent past is therefore compulsory reading, for all geoscientists (Broecker, W.S. 2003.  Does the trigger for abrupt climate change reside in the ocean or in the atmosphere?  Science, v. 300, p. 1519-1522.  As well as the astronomically connected cyclicity that is apparent in all kinds of climate record through the Pleistocene, those records are punctuated by sudden, short-lived phenomena, whose magnitudes and pace are sufficiently dramatic to focus attention on processes that are probably entirely terrestrial.  Foremost among these during the last glacial interglacial cycle are the astonishing coolings of Heinrich’s iceberg armada events and the possibly catastrophic (in a human as well as an ecological sense) Younger Dryas, which reversed warming from the last Glacial Maximum, and the equally sudden warmings associated with Dansgaard-Oeschger events.  Broecker’s review focuses on the two mechanisms that have been suggested to underlie these overturns.  One links such changes to shifts in whole-ocean water circulation, especially the ons and offs of deep-water circulation beneath the North Atlantic, the other to perturbations of the way in which atmosphere and ocean interact in the tropics.

An entirely plausible scenario for climate-driving changes in North Atlantic water circulation is flushes of freshwater from the surrounding continents, so that formation of sea ice leaves residual water that is not saline or dense enough to sink and drag in water from lower latitudes.  The problem is that the complete thermohaline cycle, which impacts on global atmospheric circulation, has a period longer than the changes that might be induced by its perturbation in the North Atlantic.  Tropical atmosphere-ocean dynamics are the largest elements in global climate, in terms of the energy and mass that are shifted, so they are a natural candidate for a driving mechanism.  Tropical climate shifts abruptly today in well-known ways, most important being the El Niño-La Niña cycle.  There is no ponderous underlying dynamic that would damp down connections between cause and global effect, and prevent sudden climate change.  Yet, some kind of “flywheel” is essential to keep long-term cyclicity going and lock sudden changes into century to millennium-long climate “states”, which should rapidly decay if effect rebounded on cause, as it does in the case of El Niño-La Niña.  Broecker covers all the critical evidence that has borne on both hypotheses up to now.  His conclusion is interesting.  Both hypotheses are very much model led, and in need of as much empirical support as can be had.  Yet, and here is the nub, the crucial data are those bearing on correlating times of events that are recognised all over the place.  Time resolution is of the greatest importance, since climate transitions are fast; faster in fact than we can presently resolve before historical times.  It is entirely likely that suitable resolution of times past may be absolutely impossible.  Both hypotheses have a lot of empirical and theoretical support.  So, what is the problem of combining them in a cunning way?  Partly, that may be because reductionism (controlling a few variables and looking for developments in another simple set) still plagues science.  That is odd in climatology, where all motions and energy changes palpably relate to one another, with no control of a rational kind.  Reductionism demands ever more staggering computing power and speed, to “keep all the eggs in the air”.  There is always the feeling, as Jimmy “Shnozzle” Durante observed in his musical monologue, The Man Who Found The Lost Chord, that if you find a hitherto overlooked connection, then everything goes well; if you can remember it!  Broecker suggests that the missing connection must “transmit” from deep ocean water to tropical atmosphere.

Tropical rainforest in Africa and South America is the most diverse biome on the planet, both as regards plants and animals.  One view of how such luxuriance arose is that the forests have blanketed the humid tropics for as long as 50 or 60 million years, and the fact that they encompass a huge variety of environments created by different levels in the dominant and diverse vegetation.  Thousands of niches and the interactions between organisms that exploit them during lengthy stasis inevitably drives rapid evolution towards all kinds of specialisation.  The other view is that rainforests are by no means static over millions of years, but climate shifts have caused them to retreat and advance, perhaps hundreds of times during the Cenozoic.  Amazonia in particular shows surprising variation in diversity, some patches being far more biologically rich than others, and having regionally distinct assemblages of plants and animals.  This theory suggests that climatic stress, probably drying associated with globally cool episodes, resulted in rainforest shrinking to “refugia”.  In them, populations of plants and animals shrank, thereby reducing the gene pool and giving greater chance for evolution by natural selection; different in different refuge areas.

Tropical soils are continually reworked and their highly oxidising nature destroys organic remains.  So no record of its development exists in rainforest.  However, wind and rivers transport spores, pollen and other biomarkers to seafloor sediments, where a complete record of fluctuations in biomass and diversity becomes preserved.  A test of the popular refugia hypothesis is therefore to analyse organic matter in continuous cores taken from offshore sediment.  Known fluctuations in global climate, from the oxygen isotope record should be matched by changes in the record of terrestrial biomarkers carried to the sea.  Cores from the deep-sea sediment fan off the mouth of the Amazon potentially provide such a test (Kastner, T.P. & Goñi, M.A. 2003.  Constancy in the vegetation of the Amazon Basin during the late Pleistocene: Evidence from the organic matter composition of Amazon deep sea fan sediments.  Geology, v. 31, p. 291-294). Kastner and Goñi, from the University of South Carolina, examined phenols and organic acids in the cores, which can discriminate between grassy plants and trees that would have dominated savannah and rainforest, whose relative cover of the Amazon basin should have changed, according to the refugia hypothesis, as climate shifted from globally cool-dry to warm-humid..  Although their record only spans the last glacial cycle since 70 ka, they detected no significant change in the proportion of grasses and trees in the Amazon catchment.  Moreover, the biomarkers remained similar to those carried by the Amazon today, right through the last glacial maximum, when drying of the tropics would have been most likely to have driven a shrinkage of rainforest area.  It seems unlikely that forest refugia developed during one of the most extreme climate shifts in the last 55 Ma.  Global climate fluctuations were considerably less before 1 million years ago, when the current round of 100 ka cycles began.  So there is little reason to doubt that the Amazon rainforest has had a more or less constant area for much of the Cenozoic.  The same cannot be said for those in Africa and SE Asia, partly because there are no useful data from offshore sediments, but also because those regions have experienced changing topography due to major tectonic activity, whereas eastern South America has remained stable.  To conclude, as the authors do, that the data signify no great fluctuation in rainfall is not so certain.