METHANE, METHANE HYDRATES, AND GLOBAL CLIMATE

In the 1980s, a few scientists began to think that methane hydrates, possibly acting together with methane released from permafrost and the bog and swamp wetlands of the far north, could have a significant effect on global climate. First among them was probably Gordon MacDonald, who in 1983 suggested that methane hydrates "may play a significant role in modulating the Earth's climate." In examining the origins of natural gas, he noted that the warming and release of such hydrate could enhance warming, both as a greenhouse gas in itself, and through its oxidation product, carbon dioxide.

Some years later, Euan Nisbet independently came to the same conclusion. In 1989, he reviewed the northern (northern, to Nisbet, a Canadian, was north of about 50°N, about the same latitude as the Canadian-US border, he bemusedly relates) sources of atmospheric methane. Among other sources including hydrate, he observed that the return of beavers after the last ice age greatly enhanced the output of methane from northern waterways, because their dams trapped the organic material from which methanogens produced "swamp gas" (Nisbet, 1989). At this time, Nisbet was clearly (and concernedly) thinking about the climatic consequences of the increasing release of methane in post-glacial times. In short order, however, he began to ruminate about whether methane release could actually have contributed to the end of the ice age itself.

What is known as the Ice Age is actually a series of ice ages (more than a dozen), each lasting about 120,000 to 140,000 years, going back some 1.6 to 1.8 million years, and perhaps as much as 2.4 million years (Balco, 2005). Each major ice age cycle brought ice sheets two or more kilometers (well over a mile) thick across vast stretches of Asia, Europe, and North America. (Current ice sheets on Greenland and Antarctica are more than 3 and 4 kilometers thick at their thickest, respectively.) Prior to the Ice Age, climate had generally been warmer, and no great ice sheets had moved across large land masses and down temperate zone mountainsides, except in the far distant past. Although we presently lack such great continental ice sheets except in Greenland and Antarctica, it is widely accepted that we are still in that period of generally cooler global climate known as the Ice Age.

Geologically, this period is called the Quaternary, which is divided into the Pleistocene (from about 1.8 to 1.6 million years to 10,000 years ago), and the Holocene, from 10,000 to the present. Although the Holocene is not very different climatically from the preceding Pleistocene -- being just another of several interglacial episodes during a long period of planetary cold -- we, of course, consider it special because human beings now dominate the world. But we do have a considerable impact on climate, so it is possible that the ice episode just completed may be the last the planet witnesses for some considerable time.

Our knowledge of the Ice Age has increased enormously since the Swiss naturalist Louis Agassiz announced in 1837 that numerous geological enigmas -- U-shaped valleys (see below), great masses of scraped and polished rock, rocks apparently transported considerable distances from their sources, the large debris piles of gravel and rock we now refer to as moraines -- could be explained if vast ice sheets had once moved across the landscape. We have come to understand that comings and goings of the individual ice ages may be due, to a significant extent, to minute changes in the energy output of the sun (called the solar constant), and to changes in the tilt of Earth's axis, Earth's top-like movement around that axis (called precession), and to variations in Earth's orbit around the sun.

Perhaps most importantly, we have come to recognize that ice ages are not so much the product of a colder climate -- though that too is necessary -- but of shifting patterns of winds, ocean currents, and precipitation. Extreme cold does not provide the snow that accumulates into sheets of ice: it can be too cold to snow. (Antarctica, because of its low precipitation, is considered a desert.) Only warmer, wetter weather permits such snowfalls. It is a cliché among geologists that ice ages occur when the summer's warmth fails to melt the previous winter's snow, and snow thus accumulates from year to year.

As scientists have come to better understand ice ages, they have been surprised, even shocked, by the rapidity with which warming or cooling takes place. The beginning of the end of the most recent ice age started some 14,000 to 12,000 years ago, but that ending was punctuated by a short but abrupt cooling interval referred to as the Younger Dryas. Evidence from Greenland indicates that the sudden cold of the Younger Dryas ended with an average temperature rise of 7°C (more than 12°F) over just 50 years, most of which occurred in just 20 years (Dansgaard, 1989). Similarly, the end of the most recent ice age itself took place in less than 1000, and perhaps as few as several hundred, years (Nisbet, 1990). Indeed, in places as far apart as southwestern Europe and the South China Sea, sea surface temperatures at the end of the ice age may have increased by up to 4°C (7°F) each century for several centuries (Bard, 1989; Broecker, 1988).

Numerous hypotheses have been offered as to the reason for these sudden and substantial warming events. Agassiz himself originally attributed the warmth to the revival of life after the long cold. Others proposed that the release of carbon dioxide from the ocean could have triggered the ice age-ending warmth. In 1990, Nisbet challenged the carbon dioxide proposal. Carbon dioxide, he argued, would have been released too slowly from a gradually warming ocean to account for the abrupt temperature shifts being discovered. Moreover, the release of carbon dioxide could not explain the rapid increase of methane found in Greenland ice cores.

Instead, Nisbet suggested, methane from hydrate in permafrost and possibly the ocean was a better candidate for the abrupt warming agent. It could be released more rapidly; it is a much more powerful greenhouse gas than carbon dioxide (so less is needed to produce the same temperature increase); the sudden shifts in temperature could be explained by its short lifespan (under 10 years) in the atmosphere. Methane release could explain the rise of methane found in the Greenland ice cores, as well as an apparent increase in the carbon available in the biosphere.

The origin of this methane, moreover, could clearly be traced to hydrate. This is because the radioactive carbon isotope (Carbon-14, or ^14C) can be used to determine the age of carbon compounds which have formed in the recent past (up to about 20 to 25,000 years ago). As the methane in the Greenland ice cores contained little ^14C, the post-glacial increase of methane was due to the release of geologically old hydrate (Oeschger, 1987), not to the recent rejuvenation of northern wetlands (Nisbet, 1990).

Noting that the quantity of methane hydrate may have been much greater at the end of the most recent glacial episode than it is now, Nisbet offered the following scenario for methane release. Initially triggered by an earthquake (freeing gas trapped beneath the ice) or a depressurizing drop in sea level, methane would have rapidly warmed the atmosphere (in contrast to carbon dioxide, which would have warmed things much more slowly). Rising sea level (the consequence both of glacial melting and a warmer ocean) then could have allowed relatively warm water (0°C, or 32°F) into low-lying permafrost and under the frigid glacial ice, permitting the release of additional quantities of methane. Eventually the return of forests and wetlands to circumpolar regions would have provided another source of methane. Both this methane and its oxidation product carbon dioxide would have contributed to still further atmospheric warming (Nisbet, 1990).

Interestingly, Nisbet noted that the process of hydrate methane release would be self-terminating. This is because "the depth of warming depends on the square root of the time since the surface warmed." In other words, it takes four times longer to warm hydrate that is buried twice as deep within the sediments. Thus longer and longer periods would be required for the release of deeper methane until the process grinds to a halt (Nisbet, 1990). It should be pointed out, however, that this assumes that temperature does not vary with depth, but, of course, it does, being higher with increasing depth (about 20°C increase per kilometer depth -- or about 58°F per mile; MacDonald, 1983). (Nisbet, 1990, showed this in Figure 4a.) A pulse of heat from surface warming, therefore, would liberate methane at depth more rapidly and for a longer period than Nisbet indicated. Nonetheless, Nisbet's observation that hydrate methane release must eventually be self-limiting is an important one.

While Nisbet recognized that methane hydrate exists in marine as well as onshore environments, he did not invoke marine hydrate in his end of the ice age scenario. In fact, oceanic methane hydrates are mentioned only in passing, though Nisbet noted (Figure 3, Nisbet, 1990) that they are found in continental margins around the world, and can be released as sea level -- and therefore pressure -- falls. Because deglaciation raised sea level in most parts of the ocean (except where glacial rebound occurred), and thereby increased the pressure on hydrates, Nisbet may not have thought about oceanic methane hydrate release in his deglaciation scenario. But the warming that produced deglaciation also warmed the oceans, and this warming could have allowed for at least some marine hydrate dissociation.

MacDonald's and Nisbet's papers were the first to suggest that methane hydrates were not just a permafrost and seafloor curiosity and a potential future source of natural gas, but might also have major environmental consequences. Until they proposed otherwise, hydrates were presumed to be stable, frozen in their peaty and muddy beds in distant corners of the planet. They were not presumed to be available to exchange much (if any) of their carbon with the other carbon repositories (geologically referred to as reservoirs) of the Earth: the atmosphere, the ocean, soils and rock, living and dead organic matter. For years (even today!) compilations and diagrams of "exchangeable" carbon often simply omit the methane hydrate reservoir.

But some scientists did think that methane (though not specifically hydrate methane) could have a serious impact on Earth's climate. In attempting to explain why circumpolar environments of the Early Eocene (roughly from 55 to 50 million years ago) seemed considerably warmer and quite hospitable to mammals, reptiles and even deciduous forests, Lisa Sloan and her colleagues proposed that the cause could have been methane from peatlands. During the Early Eocene, peatlands may have covered more than three times the area they cover today. A slight warming and drying of these lands could have released substantial quantities of methane. Such a release, reasoned Sloan and colleagues, could have produced polar clouds that would have trapped outgoing long-wave radiation and therefore the heat needed to explain the then warmer environment (Sloan, 1992).

In 1995 a paper by Gerald Dickens and three co-authors proposed that the cause of Eocene warming was more likely methane from oceanic hydrate, not methane from peatlands. Dickens hypothesized that there was a major release of methane from hydrate at about 55 million years ago, a release that produced the significant negative carbon isotopic excursion noted in the rocks and fossils from that time. With proper scientific restraint, Dickens stated that the "fate of CH¸4 in oceanic hydrates must be considered in developing models of the climatic and paleoceanographic regimes [overall conditions] that operated [at the time]" (Dickens, 1995).

Fifty-five million years ago marked the event known as the Late Paleocene Thermal Maximum, or LPTM (Zachos, 1993). This event is also referred to as the Latest Paleocene Thermal Maximum, because of uncertainty as to whether it occurred at the exact end of the Paleocene, or merely close to the end. But the terms are frequently used interchangeably, even in the same scientific paper (for example, Dickens, 2000). (The terminology issue has continued. More recently, this event has been referred to as the Paleocene-Eocene Thermal Maximum, or PETM: Zachos, 2003, and as the initial Eocene Thermal Maximum, or IETM: Dickens, 2004, and Svensen, 2004. Despite the changing terminology, the event referred to is the same. Here the original name, Late Paleocene Thermal Maximum, or LPTM, will be used.)

There is also some serious dispute as to when the LPTM (and therefore the end of the Paleocene) actually did take place. Some scientists think it occurred up to five million years earlier than its usually accepted date of about 55 million years ago, which would make it contemporaneous with the first pulse of volcanism in the North Atlantic Igneous Province (Jolley, 2002). Others strongly contest such an early date (Wei, 2003; Thomas, 2003; Srivastava, 2003), in one case objecting that this proposed timeline would cause the Paleocene epoch to be unacceptably shortened from about 10 million to only 3.5 million years (Aubry, 2003).

In any case, the Paleocene was the first temporal subdivision (epoch) of the Tertiary Period, which followed the end-Cretaceous catastrophe. It lasted about ten million years, and was followed by the Eocene epoch. The "thermal maximum" was a time of exceptional warmth, when global temperatures were several degrees higher than at any time since. The warmth is recorded in carbonate found in several, widely separated locations, thus confirming that the warmth was global in extent. The oxygen isotopes in these carbonates provide evidence of the temperatures of the time. Reading this "oxygen thermometer" is not easy, as a number of factors, such as the salinity of the ocean, can affect the mix of oxygen isotopes. But scientists are aware of these confounding factors, and do include their possible effects in estimating ancient temperatures.

The warming was considerable: as much as 4°C (7.2°F) in the deep ocean, and 8°C (14.4°F) in high-latitude (near polar) surface water (Katz, 1999). Tropical sea surface temperatures rose by as much as 4 to 5°C (Zachos, 2003). The deep water temperature, initially about 11°C (about 52°F), rose to about 15°C (about 59°F)(Kennett and Stott, 1991; Zachos, 1993; Zachos, 1994; Thomas and Shackleton, 1995). These unusually high temperatures provided the label, "thermal maximum."

In addition to the temperature changes recorded in the oxygen isotopes, there were also considerable changes in carbon isotopes. Among those deep sea foraminifera which lie on the ocean floor (benthic forams), there was a carbon isotope shift of ­2 to ­3. Among those foraminifera which float freely in the ocean (planktonic forams), the shift was ­4 to ­5. And, in carbonate found in fossilized soils (paleosols), the shift was ­4.5 (Dickens, 2000). Although these shifts vary according to their sources, they all confirm a significant major negative change in the carbon isotope ratio.

These shifts (called excursions) indicated that a considerable amount of light carbon (Carbon-12, or ^12C) had been added to the standard mix of carbon isotopes that was available in atmosphere and ocean. To employ a simple analogy, it was as if additional water had been added to lemonade, making the lemonade less yellow and less tasty. Somehow the standard carbon isotope mix had become diluted, over an extremely short period of time, in less than ten thousand years, a geological instant.

Dickens and his fellow scientists looked at a number of proposed explanations for the carbon isotope excursion. One was that the excursion represented the transfer of isotopically light carbon from the organic reservoir (in organisms, soils, and dissolved in the ocean) to the inorganic reservoir of atmosphere (primarily CO¸2) and ocean (mostly what is known as dissolved inorganic carbon, DIC for short). They found that there was simply not enough light carbon (at about ­25 per mil depletion) in the organic reservoir to account for the isotopic excursion.

In fact, to produce a ­3 per mil carbon isotope shift in today's world (and presumably the Paleocene world was little different), a quantity greater than all the carbon in today's exchangeable organic reservoir would have to transferred to ocean and atmosphere in a geological instant (here, less than 10,000 years). This is impossible, and led Dickens to the conclusion that "the geological record does not support such biomass destruction across the LPTM" (Dickens, 1995).

Dickens and his co-authors' calculations were simple, based upon scientific estimates of the amounts of carbon present in the various "reservoirs," as well as the isotopic composition of this carbon, all figures readily available to those interested in such matters. Indeed, they showed that several hypotheses about the LPTM carbon isotope excursion could be easily dismissed by the use of "simple mass balance equations."

The phrase, in fact, became a kind of refrain in their paper. Using these calculations, they also dismissed the suggestion that the light carbon could have resulted from the volcanic (both terrestrial and marine) outgassing of carbon dioxide (at about ­7 per mil depletion). For such outgassing to have provided enough light carbon, the "rate would [have had to have been] unprecedented in the geological record."

There was only one exchangeable carbon reservoir with sufficient light carbon to explain the observed isotopic excursion, according to their simple mass balance equations: methane from seafloor hydrate. The methane hydrate reservoir may have been twice to three times as great as the organic (organisms, soils, dissolved in the ocean) reservoir, and its carbon was more than twice as light (about ­60, as opposed to ­25). In estimating how much hydrate methane might have been released, Dickens and his co-authors relied on a number of assumptions.

Having no way to determine the amount of methane hydrate that may have been present in the Paleocene continental margins (they did not include permafrost hydrates in their calculations), they assumed that the amount is not greatly different from that of the present day. (For today's methane hydrate carbon, they employ an estimate of from 7.5 to 15 x 10^15 kilograms, or 7500 to 15,000 billion metric tons. A metric ton weighs 10% more than an imperial ton, so they are roughly equivalent.) This assumption may not be correct: Paleocene bottom water was estimated as at about 11°C (about 52°F), which is significantly warmer than today's bottom water temperature of about 0°C (32°F). Warmer bottom water temperatures -- and therefore warmer sediment temperatures -- would have meant less hydrate in the continental margins. Still, there should have been lots of methane hydrate present and available for release at the end of the Paleocene.

Based on a bottom water temperature increase of 4°C, Dickens and his fellow scientists calculated that methane hydrate in continental slopes at about 920 to 1460 meters depth would have been destabilized and released. Their calculations indicated that if methane hydrate were evenly distributed within continental margins wherever temperature and pressure conditions were favorable (another assumption), the amount of hydrate destabilized would have amounted to some 14% of total world hydrate, or between 1.4 to 2.8 x 10^15 kilograms (1400 to 2800 billion metric tons) of methane. The authors did state, however, that this estimate is most likely too low.

Nonetheless, this ballpark figure compares favorably with the amount of light carbon that would have had to have been added to the total ocean/air/land carbon reservoir to result in a ­2 to ­3 change. (1.6 to 2.0 x 10^15 kilograms -- 1600 to 2000 billion metric tons -- of methane would have been needed for a ­2 change; 2.5 to 3.1 x 10^15 kilograms -- 2500 to 3100 billion metric tons -- of methane for a ­3 per mil change.) Over a possible 10,000 year release period, this would amount to about 1.6 to 3.1 x 10^11 kilograms -- 160 to 310 million metric tons -- of methane per year. This is a rate that is -- astonishingly! -- less than that at which human beings are currently pumping methane into the atmosphere via our rice agriculture and domesticated grazing animals.

Methane has quite limited solubility in water, and because it is also a very light gas (lighter than air), it quickly makes its way up through the water column and into the atmosphere. Much methane does get consumed by methanotrophs but -- especially during the rapid release postulated by Dickens for the Paleocene -- much also can make it into the ocean and air. On its way through the ocean, and in the atmosphere itself, methane is rapidly (within ten years) oxidized to carbon dioxide and water. The carbon dioxide combines with water to form carbonic acid, a mild acid, leading to somewhat more acidic conditions in the oceans, in rain, and consequently on land. (The chemical reaction is: CO¸2 + H¸2O Æ H¸2CO¸3.) The extra acid should promote the dissolution of carbonate. In the oceans, therefore, the release of significant quantities of methane should be reflected in deep-sea carbonate dissolution. As Dickens noted, "it is evident that carbonate dissolution indeed occurred during the LPTM," citing two studies of such dissolution as confirmation (Thomas and Shackleton, 1995; Lu and Keller, 1993).

Methane oxidation  Methane readily combines with what is called the hydroxyl ion (OH^­). An ion is an electrically charged particle. Sometimes the term simply refers to an atom with a charge; at others, to a molecule, which is composed of more than one atom. Ions are typically indicated by the electrical charges they possess. Thus, a hydrogen ion has a positive charge, and is indicated as H^+, whereas an oxygen ion has two negative charges, and is represented as O^=. When a hydrogen ion combines with an oxygen ion, the hydrogen's positive charge is neutralized by one of oxygen's negative charges, producing the hydroxyl ion, OH^­. The hydroxyl ion is produced by the breakup of a water molecule:   H¸2O      Æ               H^+           +        OH^­ (water) (yields) (hydrogen ion) + (hydroxyl ion), or by the splitting of a peroxide molecule, H¸2O¸2. Methane is destroyed by chemical combining with hydroxyl ions. The chemical equation for this reaction is:    CH¸4       +         OH^­           Æ            CH¸3^­   +   H¸2O (methane) + (hydroxyl ion) (yields) (methyl ion) + (water) The carbon from the methyl ion, after several intermediate steps, becomes the carbon in the end product carbon dioxide. One of these intermediate steps occurs after the the methyl ion (CH¸3^­) is oxidized to carbon monoxide (CO):        OH^­         +              CO                   Æ               CO¸2            +           H^+ (hydroxyl ion) + (carbon monoxide) (yields) (carbon dioxide) + (hydrogen ion) Note that this reaction, like the previous one, consumes a hydroxyl ion. The overall, simplified equation (deleting the intermediate steps) is:     CH¸4      +    2O¸2         Æ               CO¸2           +  2H¸2O (methane) + (oxygen) (yields) (carbon dioxide) + (water) This chemical process is called oxidation not because oxygen is involved, but rather because the oxygen picks up ("accepts") negative charges (specifically, electrons) from the methane. (In other oxidation reactions, other substances, such as sulfur, may pick up negative charges. Thus, no oxygen need be involved. Sorry, scientific terminology is sometimes inexcusably confusing!) The main, and almost exclusive, process by which methane is destroyed in the atmosphere is via oxidation by hydroxyl. Consequently, as the oxidation of methane proceeds, hydroxyl ions are consumed. As atmospheric methane increases, it therefore slowly depletes the amount of hydroxyl available, because the steady rate of hydroxyl production is outpaced by the methane increase. This has important consequences for atmospheric chemistry. With fewer and fewer hydroxyl ions in the atmosphere, less and less methane can be consumed. That is, in fact, what is currently happening. Additional methane in the atmosphere, of course, leads to even greater consumption of hydroxyl ions. The concentrations of other atmospheric gases are also affected. Hydroxyl ions, as noted above, also remove carbon monoxide, which may be from either natural sources (largely the oxidation of methane) or industrial sources (which are increasing the amount of carbon monoxide). The ions also react with several other gases, both from natural and industrial sources. Among them is hydrogen sulfide (H¸2S), which can chemically combine with hydroxyl to form sulfuric acid (H¸2SO¸4). Fewer hydroxyl ions, therefore, may mean a more acidic atmosphere. And sulfuric acid is not the only acid that may increase as a result of lower hydroxyl ion availability. More methane may result in more chemical reactions with chlorine, producing hydrogen chloride, that is, hydrochloric acid (HCl). Increasing methane concentrations in the atmosphere, consequently, can produce significant changes in atmospheric chemistry, some of which can be projected, others which are as yet unknown, and still others which can impact the level of ozone, our protection against deadly ultraviolet light (Sze, 1977).

An additional effect of methane release into the atmosphere should be enhanced global warming. Methane itself is a significantly more powerful greenhouse gas than carbon dioxide, and though it oxidizes rapidly, it oxidizes to carbon dioxide. Thus both the methane and its successor gas, carbon dioxide, contribute to the warming of the planet, including the oceans. Dickens recognized that this warming could result in further hydrate dissociation, and additional methane release, in a positive feedback cycle, and wondered how the cycle could stop short of what he considers an implausible outcome: the complete depletion of the oceanic hydrate reservoir. (Nisbet, 1990, may have already have provided at least part of the answer: warming takes longer with sediment depth.)

Another issue Dickens examined is how the warming that produced the Latest Paleocene Thermal Maximum began. He found a triggering mechanism for the warming suggested by Thomas and Shackleton (1995) to be "particularly appealing: a rapid emission of CO¸2 associated with a brief interval of explosive volcanism in the North Atlantic," which warmed the planet.

The volcanism Dickens found appealing took place about 55 million years ago. At that time the configuration of the continents was beginning to assume its modern form. The Atlantic had opened but was considerably smaller than it is today; India was approaching its collision with Asia (which uplifted the Himalayas); Australia had detached from Antarctica. In the far north of the North Atlantic, Greenland had started to separate from Norway. This separation, as with the opening of the Atlantic itself, came as a consequence of the formation of the Mid-Atlantic Ridge, which over tens of millions of years was creating an ocean by laying down ocean floor between Africa and Europe on one side and the Americas on the other.

The North Atlantic Igneous Province (NAIP) at two stages of its development. The upper map shows the NAIP at an earlier, older stage; the lower, about 700,000 years later. The dark areas are the NAIP lava flows themselves: by 55.8 million years ago (Ma), those in the western Greenland region (upper left) became inactive. Light gray indicates land areas; white areas are sea. The areas with horizontal stripes have been filled by coastal sedimentation. Numbered dots indicate the locations of boreholes. The NAIP was formed as part of the opening of the North Atlantic Ocean, beginning about 62 million years ago. Its remnants now exist in eastern Greenland and Iceland, and form part of the floor of the North Atlantic. (Knox, 1998)

The opening of the far northern portion of the Atlantic was accompanied by the major volcanic eruptions that continue today, in modified form, in Iceland and on the adjacent ocean floor. These eruptions have produced what is known as the North Atlantic Igneous Province (NAIP), which today extends about 2000 kilometers (about 1240 miles) from eastern Canada to the (Danish) Faeroe Islands, to the north of Great Britain. The major eruptions took place in two phases, the first about 62 million years ago (Saunders, 1997).

But it was the second phase of major igneous activity in this area, beginning about 56 million years ago, that possibly triggered the release of methane hydrates. And while Dickens found the notion of global warming by volcanic carbon dioxide appealing, both that general warming and the direct heating of seafloor sediments by volcanically warmed ocean currents may have each played a role.

Because this part of the North Atlantic was just opening, its connection with the Arctic Ocean, was quite constricted. Nonetheless, this far northern sea was probably also quite cold. The eruption of basalt on the seafloor would have changed that, warming the North Atlantic, then the South Atlantic, and possibly warming and altering thermohaline circulation worldwide (for a discussion of thermohaline circulation, see APPENDIX 3: THERMOHALINE CIRCULATION). The erupted lavas were unusually hot, and they were extruded at a very high rate (Saunders, 1997). They should have been quite sufficient to trigger the temperature increase in ocean bottom water observed at the Latest Paleocene Thermal Maximum. As bottom water temperature increased, methane hydrate would also have been destabilized and released.

These is an alternative (or perhaps additional) mechanism for the release of seafloor methane (Svensen, 2004). This proposal also relies on North Atlantic volcanism as a trigger. As magma rises through sedimentary strata, it frequently pries apart and flows into (intrudes) weaknesses between the sedimentary layers. These magmatic intrusions are called sills: they are thin, flat, and originally horizontal, kind of distortedly-shaped igneous rock pancakes sandwiched between the sedimentary strata.

Obviously, the flow of molten rock directly into ocean floor sediments provides an excellent mechanism for the heating and release of the methane therein. It is also an excellent mechanism for rapid heating, because sills must be emplaced quickly to avoid cooling and solidification. Ocean drilling has revealed such sills deep off the coast of Norway, where some are as much as hundreds of kilometers (over 120 miles) long -- and which must have been emplaced within decades! (according to Svensen, 2004).

Three-dimensional seismic imaging has provided further insight into the nature of the Norwegian sills (Cartwright and Hansen, 2006). The imaging reveals a still structure not unlike a random stack of variously-sized plates, bowls and saucers, each in partial contact with the dish below. For the Norwegian sills, the contact points apparently represent places where magma flooded upward, creating the next sill in the stack. This suggests that "sills can form... efficient conduits for...magma transport." The new study's authors, however, indicate that sill formation could have taken on the order of 1000 to 10,000 years, a considerably longer length of time than suggested by Svensen, 2004, but nonetheless a very short period of time by geological standards (Cartwright and Hansen, 2006). (Moreover, the authors note that similar sill emplacement could have characterized the approximately 183-million-year-old Karoo-Ferrar igneous province as well as that of the 250-million-year-old Siberian Traps [Cartwright and Hansen, 2006]. And indeed, there is evidence of sill intrusion in the thick salt deposits ["evaporites," produced by the evaporation of ancient bodies of water], in the Siberian Tunguska Basin [Svensen, 2006].)

The magma of the Norwegian sills intruded organic-rich mudstones of Cretaceous (144 to 65 million years ago) and Paleocene (65 to 55 million years ago) age. Sonar imaging shows that fluids -- presumably carrying methane -- escaped from these layers, breaking through and distorting the overlying strata. These conduits terminate in hundreds of mounds, craters, and other seep structures -- exactly at the boundary between the Paleocene sediments and those of the overlying Eocene. Clearly then, the fluids emanating from the area of the magmatic sills rose to what was the floor of the ocean at the end of the Paleocene, thus establishing with certainty the time of fluid release. Subsequent sedimentation, during the Eocene and later, has covered and hidden the fluid release structures, which now have been almost magically revealed via the sonar images (Svensen, 2004).

The discoverers of the buried end-Paleocene sills, however, do not believe that the methane that was released by the magmatic heat was of biological origin (Therefore, the methane would not have come from hydrate.) Rather, they indicate that the methane was thermogenic: that it was produced by the heating of the hydrocarbons in the organic-rich mudstones into which the magma intruded (Svensen, 2004, and supported by Storey, 2007). As noted by Dickens (2004), this seems improbable. Because thermogenic methane is far less depleted in the lighter isotope of carbon (^12C, or Carbon-12), it would take almost twice as much thermogenic methane to produce the observed end-Paleocene carbon isotope excursion as methane from hydrate, which is largely biogenic (Dickens, 2004). (The difference in the isotopic composition between the two types of methane is what is used to distinguish them, though Milkov and Dzou, 2007, indicate that "the very first stages" of thermogenic methane release may be much more depleted. But it seems very unlikely that such a level of depletion would have been characteristic of an extended thermogenic methane release.)

Although there was probably some thermogenic methane that would have been produced and released by the intrusion of the magmatic sills, biogenic methane both from the sediments and from hydrate likely contributed to the end-Paleocene methane release, as Dickens has pointed out (2004). Indeed, it would hardly seem possible for it to have been otherwise: the rising, warm fluids mobilized from the sills would have heated and dissociated the overlying hydrate.

(Even those scientists [mentioned previously] who do not accept the 55 million year ago date for the LPTM believe that North Atlantic volcanism may have been the trigger for the initiation of warming, and the release of seafloor methane. Instead of pointing to the second major pulse of North Atlantic volcanism [at about 55 million years ago], however, they believe that the trigger lay in the first pulse, perhaps beginning 60.5 to 60 million years ago [Jolley, 2002]. That is the date they therefore employ for the start of the LPTM.)

The North Atlantic Ocean and surrounding continents, about 55 million years ago. The map shows the relative positions of the Blake Plateau, the Caribbean Plate, and the North Atlantic Igneous Province (NAIP). The gray areas mark 55 million year ago land masses; the white areas are ocean. Black lines delineate the positions of today's land masses. (Ocean Drilling Stratigraphic Network, an initiative of GEOMAR, Research Center for Marine Geosciences/ Kiel and the Geological Institute of the University Bremen: www.odsn.de/odsn/cgi/make_map.pl)

Other scientists (Bralower, 1997) have offered another volcanism-related suggestion for hydrate methane release. They note that ocean floor drill cores from the Caribbean Plate, south and west of the Blake Plateau area, also register significant negative carbon isotopic excursions -- as much as ­12 per mil in one case and ­3 per mil in another -- from the time of the LPTM. A 10 centimeter (4 inch) deep ocean sediment core section indicates that the excursion reached its maximum in about 6000 years, before a recovery period of several tens of thousands of years.

The cores also record a shift from the ordinary disturbance (bioturbation) of sediments caused by such burrowing creatures as worms, crustaceans, and mollusks to laminated (layered) sediments at the start of the LPTM carbon isotope excursion. Laminated sediments are indicative of low oxygen conditions, because the organisms which ordinarily disturb the sediments have been killed off by the lack of oxygen. This lack of oxygen is presumably due both to warmer water, which holds less oxygen, and to the release of methane, which combines with, and therefore depletes, the oxygen in the water column.

In the sediment cores, exactly at the transition between the underlying bioturbated layers and the overlying laminated sediments is a layer of tephra, volcanic ash. This ash is derived from a major pulse of eruptions along the Caribbean Plate. This eruption would have briefly cooled the tropics, possibly leading to cooler and therefore denser tropical waters that could have temporarily replaced cold northern waters locally on the ocean floor, thereby reordering ocean circulation (at least in the Atlantic) and warming and releasing hydrate methane (Bralower, 1997). The direct heating of ocean bottom water (and methane release) by the Caribbean Plate volcanism may constitute an addition or alternative to this scenario.

Geological support for the Dickens hydrate methane release scenario was quick in coming. It came from a surprising source: ocean drilling off the Atlantic coast of southeastern United States (Florida/South Carolina). There the Ocean Drilling Program (ODP), a international scientific program devoted to exploring the oceans and drilling into and obtaining cores (cylindrical samples) from the ocean floor, found fascinating evidence from drilling Site 1051. The cores, from 500 meters (0.3 miles) deep in ocean sediments overlain by 1.5 kilometers (almost a mile) of water, included an unusually thick section from the Latest Paleocene Thermal Maximum (Katz, 1999; Norris and Röhl, 1999). Although this section is now deeply buried in accumulated sediments, at the time of the LPTM, 55 million years ago, it constituted the surface of the seafloor.

The Blake Nose area off the Florida/South Carolina coast. The Blake Nose itself is in the rectangular box. (MacLeod, 2001)

The section, from what is called the Blake Nose, displayed normal oceanic sedimentation on the lower continental slope, then large chunks (clasts) of chalky mud up to about 5 centimeters (2 inches) long. These chunks do not represent normal sedimentation; instead they are debris indicative of a major submarine landslide, or slump. (The word slump is used herein to mean any submarine landslide, though the term is sometimes used by geologists in a more restricted fashion.) Above the mud clasts, much of the carbonate (from foraminifer skeletons) had been dissolved.

The cores provided additional information. They showed that in the latest Paleocene, more than half of the seafloor (benthic) foraminifera became extinct within a period of less than 5000 years. Most (60%) of these forams made their last appearance within the mud clast interval. (This period thus marks what is known as a "Benthic Foraminifera Extinction Event" or BFEE. BFEEs seem to indicate dysoxia or anoxia, that is, conditions of low oxygen or no oxygen.) Within the mud clast interval, faunal diversity plunged (by well over 50%), and organisms which tolerate low oxygen levels are found. Moreover, immediately overlying the mud clast interval was a twenty centimeter (eight inch) thick stratum in which both oxygen and carbon isotope values plummeted. The oxygen isotope drop (measured in foram skeletons) was indicative of an increase in water temperature by over 6°C (Katz, 1999).

The carbon isotope drop was ­3. As Katz and her co-authors (one of whom was Dickens) stated, "Release and oxidation of 1 x 10^18 to 2 x 10^18 g of CH¸4 [1000 to 2000 billion metric tons of methane], and the subsequent propagation of CO¸2 [carbon dioxide] through various carbon reservoirs, is the only known mechanism to explain the sudden, extreme, and global nature of the CIE [Carbon Isotope Excursion]" (1999).

The geological information obtained from the cores allowed scientists to piece together what happened here some 55 million years ago. The lower portion of the cores indicated normal seafloor activity. Forams and other seafloor creatures went about their ordinary business, while a thin rain of sedimentary particles accumulated around them. Then there was a great slump, apparently originating some 15 kilometers (9 miles) upslope. The slump presumably was caused by the dissociation of methane hydrate within the sediments, a consequence of increasing oceanic warmth. (Katz and her fellow scientists attributed the warmth itself to long-term global warming, but this warming obviously may have been supplemented by the eruption of the North Atlantic Igneous Province, or from Caribbean volcanoes.)

The hydrate dissociation destabilized the sediments of the continental slope, sending a load of chunky, chalky mud downslope. Warmed and depressurized methane boiled out of the slump, combining with dissolved oxygen in the water column, and thus reducing oxygen availability to aerobic organisms. The increasing acidity of the water promoted the dissolution of carbonate skeletons, including those which still housed foraminifera, driving many to extinction. The slump scenario was likely repeated elsewhere, especially in the North Atlantic, intermittently over the course of several thousand years. (The Blake Nose slump, incidentally, clearly shows that sediment warming by magmatic sill intrusion, mentioned above, was not the only end-Paleocene methane release mechanism.)

The time interval for the methane release, at least on the Blake Nose, is highly constrained. This is because minute variations in the magnetism of the sediments reflect the Earth's orbital and rotational cycles. One of these cycles (the precessional cycle, the top-like movement of the Earth around its axis which can vary in length from about 19,000 to 23,000 years) averages about 21,000 years long, and the changes in planetary magnetism it produces are recorded in the Blake Nose cores. Because the 20 centimeter stratum which overlies the mud clasts and records the carbon and oxygen isotope excursions represents only a quarter to a third of one of these cycles, Katz (1999) estimated that the ocean warming occurred over a period of from 5000 to 7000 years.

A similar, but even shorter, estimate of the duration of the excursion came from Norris and Röhl (1999), who also examined Blake Nose cores. They found that most of the light carbon release took place "catastrophically over a few thousand years or less." Norris and Röhl's repeated emphasis on the abruptness of the light carbon release and its duration of "no more than a few thousand years" suggests that they in fact thought that the release may have occurred in considerably less than a few thousand years -- possibly only a few centuries or less -- but their data did not allow them to define their time interval more tightly.

A still shorter estimate of the duration of LPTM warming came from ocean floor sediments from near Antarctica. Using oxygen isotopes from foraminifera that lay on the seafloor, Kennett and Stott (1991) found that deep sea temperatures jumped about 8°C (from about 10°C to about 18°C, that is, from 18°F to 32°F) in only 2000 years. This stunningly short duration for a major oxygen isotope excursion provides additional support for catastrophic climate warming and possible reordering of global thermohaline circulation, as well as a confirmation of the Benthic Foraminifer Extinction Event (BFEE) at a site many thousands of kilometers (several thousand miles) away from Blake Nose. Like the BFEE off the coast of the southeastern United States, a substantial number of foram species (40%) disappeared forever, the likely consequence of deep sea anoxia.

LPTM recovery took considerably longer than the several thousand year period of hydrate methane release. Three transition species of free-floating (planktonic) forams (the "transition fauna") appear at Blake Nose immediately above the Benthic Foraminifera Extinction Event, and then are gone. Over some 200,000 years, new species of forams appeared. Oxygen and carbon isotope values returned to roughly starting values over a period of about 140,000 years, first rapidly, then more gradually (Norris and Röhl, 1999). The sediments off Antarctica reveal that warm bottom waters persisted for about 100,000 years before cool temperatures returned (Kennett and Stott, 1991).

When Dickens and co-authors (1997) used computers to test and model his proposal, they found a similar duration for the recovery period. Scientists frequently use what are known as models when they cannot directly observe the things they are interested in (because they are too big, as the Universe; too small, as atoms; too fast, as protein folding or the explosion of supernovae; too complex, as global climate change; too long ago, as events in geological, paleontological, or cosmological history; or too far in the future, as the end of the Universe). Models are often limited in the categories and sizes of their inputs, but they can provide quite useful insights nonetheless. In modeling the consequences of the LPTM hydrate methane release, the model here employed a release of a somewhat smaller amount of hydrate methane than Dickens (1995) thought was actually released at the LPTM. (About 8% of estimated global hydrate was used in the computer model, rather than the 14% in the Dickens, 1995, calculation).

The model Dickens and his coworkers (1997) used indicated that the return to roughly initial carbon isotope values should take about 120,000 years, in close agreement with both with the 120,000 year estimated time for Paleocene carbon burial, and the 140,000 year calculated time for present-day carbon burial (Norris and Röhl, 1999).

What may have taken much longer to return to previous values were global temperatures. According to Dickens' (1997) modeling, greenhouse gas warming from the methane release would have been relatively modest, amounting to only an average 2°C increase in global surface temperatures (the actual temperature increase may have been somewhat higher). But the model indicated that cooling would have taken about 2 million years, considerably longer than the isotopic return to initial values. This seeming discrepancy may have its explanation in that the model indicated that most of the cooling took place quickly, and the oxygen isotope record may have reflected that significant early cooling. The remainder of the cooling -- only a part of a degree -- occurred over a protracted period of time.

While isotopic values have little impact on living things (though it may require slightly more of an organism's energy to process heavier rather than lighter isotopes), temperature does matter. This is particularly the case for marine organisms that are attached to or reside on the ocean floor (or which are attached to other organisms which are), or those (planktonic organisms) which freely float on the ocean's currents. Most of these organisms are adapted to life with little temperature change, and they are devastated when temperatures move even slightly -- that is, by a degree or two -- up or down. Elevated (or depressed) temperatures thus can have significant ecological consequences.

Although the global warmth of the time did result in some extinctions, environmental changes also can favor the survival and spread of organisms which are better adapted to the new circumstances. Thus, one surprising consequence of the Latest Paleocene Thermal Maximum was that it apparently allowed a migration of numerous groups of archaic mammals from their origination areas to elsewhere in the northern continents (Asia, North America, and Europe), across northern polar lands, in several waves (Beard and Dawson, 1999).

Early rodents (the lineage of the modern rat, mouse, squirrel, beaver, porcupine, hamster, guinea pig, chinchilla, muskrat, gerbil, and capybara), which may have evolved in North America (or may have migrated there from Asia), may have had the opportunity to migrate back to Asia and to Europe. On their way, they passed the ancestors of today's primates, as well as artiodactyl mammals (animals with an even number of toes: the lineage of the antelope, pig, deer, sheep, goat, giraffe, camel, and cattle) and perissodactyl mammals (animals with an odd number of toes: the lineage of the horse, rhino, and tapir), and creodonts (now extinct carnivorous mammals) moving into North America and thence to Europe from Asia. The last two of these groups -- perissodactyls and creodonts -- likely originated in Asia; artiodactyls may have come from Africa. Similarly, primates may have originated in Africa, moved on to Asia, and from there to North America during the LPTM, but their fossils are few and their movements obscure (Hooker, 1998; Beard and Dawson, 1999).

These mammals were able to move over lands previously impassable because of their frigidity. Interestingly, this migration may have been made possible just partially by general global warming. Additional warming may have been provided by what are known to meteorologists as type II polar stratospheric clouds. These clouds were previously cited to explain Eocene warming generally (see above; Sloan, 1992), but more recently their contribution has been invoked to provide just enough warmth to have kept circumpolar lands from the deep freeze (Peters and Sloan, 2000).

Polar stratospheric clouds. These clouds come in two types, type I containing water ice and either nitric or sulfuric acid, type II being pure water ice. (Photo: Paul A. Newman, NASA.)

Type II polar stratospheric clouds form in the stratosphere (the part of the atmosphere just above the troposphere, the lowest portion of the atmosphere) only during the long polar night, obviously at extremely low temperatures (below about ­83°C, or ­117°F). The atmosphere in polar regions is usually quite dry, and it has been suggested that it was the oxidation of LPTM methane, to carbon dioxide and water, which provided the moisture that allowed these clouds to form (Peters and Sloan, 2000).

Though extremely cold, these clouds would have trapped the heat below them, allowing critical migration routes to remain at temperatures close to freezing even in winter, rather than dropping prohibitively lower. Two of the main routes, one the more familiar Bering Strait land bridge between Asia (eastern Siberia) and North America (Alaska), the other the Davis Strait land bridge from the islands of the Canadian north through southern Greenland and on to Europe (called the Thulean route), may have also been warmed by coastal currents, possibly maintaining average temperatures above freezing year round (Peters and Sloan, 2000).

With a prohibitively wide and deep marine channel then separating Europe from Asia, it seems likely that mammals from Asian homelands migrated first across the Bering Strait land bridge to North America, and from there via the Thulean route into Europe (Beard, 1998; Beard and Dawson, 1999). Mammals which evolved in North America would have been able to migrate either to Asia or Europe. Whatever indigenous mammals Europe may have had were quickly displaced by the newcomers. The original homelands of many mammals, as well as their dispersal routes, however, have been the subject of ongoing discussion among paleontologists, and that discussion has yet to reach consensus conclusions. Nonetheless, the evidence for rapid mammalian dispersal among the northern continents at the time of the LPTM is unequivocal.

The Blake Nose results (Katz, 1999) reveal one major mechanism by which methane is released from hydrate, at least during methane catastrophes: slumping. In addition, the slumping mechanism provides an additional answer (to that provided by Nisbet) to the problem that Dickens considered in his original paper: how hydrate release stops. Though Dickens assumed, for the purpose of his calculations, that methane hydrate is evenly distributed throughout continental margin sediments as long as they meet the necessary temperature and pressure conditions, this is highly unlikely.

Sedimentation itself is a quite uneven process. Close to shore, sediment accumulation generally is high. Rivers that drain high or growing mountains carry off great quantities of sediment. Areas that have recently been glaciated produce lots of sediment because glacial movement grinds rock into rock flour, which becomes available for transport when the glaciers melt. Other factors being equal, those regions that get more rainfall are likely to generate more sediment. Freeze-and-thaw cycles, rock types, and numerous other factors help determine the availability and amount of sediment that may be deposited in adjacent oceans.

Submarine slumping, like sedimentation itself, is a uneven process. Some parts of the continental slope are presumably at or close to threshold conditions for slumping, meaning that a slight disturbance -- a minor earthquake or a small amount of warming or depressurization -- could trigger a methane release. Other portions of the slope are undoubtedly sufficiently stable that their methane could be released only in extreme circumstances. In addition, slumps are constrained by random initial conditions, including the location of continents and oceans, whether the continental slope is part of an active or passive margin, the strength and warmth of thermohaline circulation and local currents, and numerous other factors. Obviously they may involve large or small amounts of sediment, together with large or small amounts of methane hydrate.

Most importantly, slumping is not a continuous process, but one of starts and stops. Most of the time, methane hydrates slumber in their muddy beds. But, on rare occasion, warmth, depressurization, or earthquakes disturb that slumber, and then there is a slump cascade over tens or hundreds or thousands of years, during which there are one or more submarine landslides, each releasing its store of methane. Then slumping stops, and the long hydrate slumber returns.

The Dickens Achievement

Dickens and his fellow scientists provided a realistic scenario as to how hydrate methane could account for the significant negative carbon isotopic change found at the time of the Paleocene-Eocene boundary. Their calculations also showed that there was no other credible scenario that could produce that change. Subsequently, evidence from ODP drilling showed that underwater slumping was involved in at least one hydrate release event, and that that event involved deep-sea extinctions and a slow process of recovery.

Additional evidence from both models and actual data from submarine sediments indicated that it takes, at most, just a few thousand years to release a substantial amount of methane into the atmosphere, from ten to twenty thousand years for the carbon isotope excursion to peak, and from 120,000 to 200,000 years for the carbon isotope values to return to their approximate starting points.

The main points of the Dickens "LPTM hydrate dissociation hypothesis" are as follows:
1. Mass balance equations required the release of continental margin methane hydrate.
2. Global warming from volcanic CO¸2 served as the trigger for the methane hydrate release.
3. A major seafloor slump or slump cascade allowed the hydrate methane to be released.
4. Additional global warming was the consequence of the methane release.
5. The methane release and recovery cycle (as indicated by the carbon isotope excursion) took about 160,000 years.

Interestingly, Dickens (1997a) refers to the LPTM hydrate dissociation hypothesis as a "default hypothesis." Dickens seems to have come reluctantly to the recognition of the ability of continental margin methane to significantly alter climate and ecological conditions. Numerous other scientists appear to have subsequently reached the same conclusion. As pointed out by Sherlock Holmes: "When you have eliminated the impossible, whatever remains, however improbable, must be the truth" (Conan Doyle, 1890, Ch. 6). The huge carbon reservoir of methane hydrate and the free gas below formerly either was not recognized, or was taken for granted. It was presumed to be static, inert. Dickens showed that it could indeed interact with the biosphere and do so quite dramatically.


CONTINUE TO NEXT SECTION (The Permian-Triassic Boundary)

RETURN TO CONTENTS