Many scientists have been considering
the idea that the Earth may have completely
frozen over sometime in the past. Kirschvink
[1992] proposed that a runaway ice-albedo
effect on continents amassed in low latitudes
caused the descent into a global freeze. During
a prolonged glacial period when oceans
and atmosphere were isolated from each
other, volcanic emissions would cause a
buildup of atmospheric carbon dioxide
(CO
2), with no silicate weathering or photosynthesis
to act as a sink. Eventually a reverse
ice-albedo effect would cause a transient
heat wave as the ice cover melted back rapidly.
Hoffman et al. [1998] and Hoffman and
Schrag [2002] proposed that the idea of
global glaciation, ‘Snowball Earth,’ was strongly
supported by the occurrence of distinctive
so-called cap carbonates with negative carbon
isotopic ratios directly overlying glacial
sediments. Global correlation of such negative
carbon isotopic excursions suggested
that they were primary, synchronous reflections
of the chemistry of Neoproterozoic
ocean waters, and later studies have calibrated
Neoproterozoic time [1000-543 million years
ago (Ma)] on this basis.
The importance of Snowball Earth is not
simply that geologists want to satisfy their
curiosity about a murky phase of the planet’s
history. A climatic excursion of this scale
would serve as a benchmark for numerical
climate models at the limit of environmental
change. And such an extreme climate state
would also be expected to have profound
importance for Earth’s fledgling biosphere,
because scientists currently are speculating
on the ways in which life diversified in the
run-up to the great evolutionary radiation of
the Cambrian Period. Previous suggestions
of global ‘Eiszeit’ (global glaciation) have
been made but have not stood the test of
time. What makes the Snowball Earth hypothesis
different is that it strives to integrate disparate
lines of evidence into a unified, though
extraordinary, explanation.
A recent meeting brought together a multidisciplinary
group of 75 climate modelers,
glaciologists, paleobiologists, sedimentologists,
geochemists, stratigraphers, and tectonics
experts to evaluate the current standing
of this controversial idea that the Earth froze
over completely in the deep geological past.
Several reflections from the conference are
summarized below.
Timing, Number, and Duration of Glaciations
and Deglaciations
When plotted without any attempt at filtering,
radiometric age dates constraining
Neoproterozoic glaciations fall in a widely
scattered cloud of data points and error
bars between approximately 760 and 560
Ma. Making sense of this cloud is an urgent
goal, and there has been significant progress
in improving our understanding of the timing
of glaciations during the past few years.
During the meeting, participants agreed that
there are now tightly constrained uranium lead
(U-Pb) zircon dates on glacial deposits
in Namibia and cap carbonates in South
China, strongly suggesting a glaciation terminating
at 635 Ma on two widely separated
continental plates.
Other Neoproterozoic glaciations are not
as well constrained, and new rheniumosmium
(Re-Os) dates presented at the conference
cast doubt on any simple correlation
scheme of glacial successions, even
within a single plate such as Australia. Geochronology
cannot at present constrain the
duration of glaciations, but different data
sets provide interesting though oblique new
insights. Magnetic reversals from cap carbonates
reported at the conference constrain
the minimum duration of deglaciation.
Whereas the prototype Snowball Earth
hypothesis, and the climate physics experiments
that followed, viewed deglaciation as
rapid (2000-10,000 years) and catastrophic,
it was agreed that deglaciation probably
lasted at least 100,000 to one million years.
Plate Configurations, Paleolatitudes, and
Supercontinental Assembly
Although the original Snowball Earth
hypothesis invoked a clustering of continental
plates in low latitudes, the latest results
presented at the meeting indicated that the
reverse is the case, with a state of widely dispersed plates during glacial periods.
Indeed, many researchers link glaciation
with rift basin and passive margin development.
It was suggested that there are large
uncertainties in Neoproterozoic paleogeographical
reconstructions, and that small
changes—for example, in the building of
land bridges or the opening of oceanic gateways—
may have been as important in the
Neoproterozoic as they are known to have
been in more recent times. Critically, new
paleomagnetic studies reported at the meeting
continue to suggest that some glaciations
took place at low paleolatitudes of 15-24°.
There was, therefore, consensus that glaciation
most likely took place at sea level at a range
of paleolatitudes, including the tropics.
Carbon Isotopic Trends in Neoproterozoic
Carbonates
Although negative carbon isotopic signatures
originally were seen as highly diagnostic
of the melting of a Snowball Earth, the causal
link between glaciation and the occurrence
of depleted carbonates was strongly questioned
at the meeting, since the largest carbon isotopic
excursion (from +5‰ to -12‰) in the
Neoproterozoic from the Huqf Supergroup
of Oman occurs where there is no evidence
for contemporary glaciation.
Some meeting participants argued that
few bulk carbonate isotopic ratios are pristine
or indicative of oceanographic values,
and may be more representative of the biology
and chemical transformation of
organic-rich seabed sediment than of glacial
forcing per se. The wide spread of isotopic
values from different components of
framework and matrix of carbonate rocks
prompted participants to question exactly
what the bulk carbonate values mean. If isotopic
ratios are not primary, the question
arises as to whether they can be safely used
for global correlation and the partitioning of
Neoproterozoic time.
Paleoenvironments, Glacial Dynamics, and
Impacts on the Biosphere
It was generally accepted at the meeting
that the sedimentary record contains abundant
evidence for dynamic glaciers, including
great thicknesses of glacial sediments containing
faceted, shaped, and striated clasts
and dropstones. Interleaved sediments show
unequivocal evidence for open water and
high sediment transport rates. One view
expressed was that deposition of thick glacial
sediments took place only during final
melting of Snowball Earth. Another view was
to claim that the glacial sediments reflect a
reduced but effective hydrological cycle
rather like that of Antarctica today. Yet another
proposal was to adopt an alternative model
that retains low-latitude open oceans. The
recognition of marked geochemical variations,
attributed to changes in chemical
weathering of contemporary land surfaces,
suggests pronounced climatic variations
within glacial epochs, not unlike the more
familiar icehouse worlds since the end of
the Neoproterozoic.
An intriguing question addressed by delegates
was the role of extreme climate swings
on the evolution of the Neoproterozoic biosphere.
Different views were expressed at
the meeting. On the one hand, opportunistic
prokaryotes and more advanced eukaryotes
persisted through the major glacial epochs
in a business-as-usual fashion, and organic
compounds fail to register any evolutionary
bottlenecks during glaciations. Major evolutionary
developments such as the emergence
of complex acritarchs and soft-bodied macrobiota
take place considerably later in the
Ediacaran period.
On the other hand, major metazoan
clades were suggested to be rooted in Neoproterozoic
glaciations on the basis of
molecular clock data, and biodiversity collapses
were reported in the aftermath of glaciations. Since the only well-constrained
evidence for glaciation in the Ediacaran
period (the youngest part of the Neoproterozoic)
is the short-lived Squantum-Gaskiers
glaciation of northeastern Laurentia
(dated 582 Ma), conference delegates were
left pondering why major evolutionary
developments follow the most feeble of the
Neoproterozoic glaciations, and why putative
Snowball Earth events are not accompanied
by major mass extinction.
Numerical Climate Modeling
Different models make different predictions
for the atmospheric CO
2 levels necessary
for the onset and exit of glaciation; the
extent to which a runaway ice-albedo effect
would make global glaciation inevitable;
and the timescale, and nature, of glacial
shutdown and postglacial aftermath.
A number of contributions showed that
descent into a Snowball state on a planet
with 6% less solar luminosity than at present
is much easier than recovery. Although
a sophisticated treatment of the critical icealbedo
effect was demonstrated at the
meeting, scientists are in almost uncharted
waters in terms of the effects of the distribution
of continental plates and their past
topography.
It was suggested that minute (1000 square
kilometer) areas of open water, surely present
around volcanoes and hydrothermal systems
even during Snowball glaciations,
would allow essentially free exchange
between atmosphere and ocean, making the
idea of mutual isolation difficult to sustain.
However, new estimates presented at the
meeting of the value of atmospheric CO
2
immediately following glaciation—using
multiple proxies of boron, calcium, carbon,
and oxygen isotopes—suggest exceptionally
high levels.
Current Status
Although important disagreements about
the implications of the sedimentary and
paleobiological record persist—such as the
need for near-complete shutdown of the
water cycle, the extent of open ocean during
glacial episodes, and the strength of the impact
of glaciation on Earth’s marine biota—a
number of convergences became evident at
the meeting.
Participants were impressed by the similar
timing (635 Ma) of the end of at least
one glaciation in widely separated locations,
and by the locally firm evidence for low-latitude
glaciation at sea level, and they noted
the ease with which Earth may become fully
glaciated in climate experiments due to a
runaway ice-albedo effect. What is clear is
that the original pillars that supported the
Snowball Earth hypothesis—such as the
presence of iron formations, a low-latitude
clustering of continental plates, and the
characteristic carbon isotopic composition
of marine carbonates deposited during
deglaciation--have been discarded.
The meeting allowed a critical examination
of the Snowball Earth hypothesis in all
its facets. In particular, it stimulated an
enhanced exchange of views between climate
modelers and geologists, and it has
helped to inform the research agenda on
Neoproterozoic Earth history and climate
dynamics over the coming decade.
The First International Assessment of
Snowball Earth Hypothesis was held 16-21
July 2006 in Ascona, Switzerland.
References
Hoffman, P. F., and D. P. Schrag (2002), The Snowball
Earth hypothesis: Testing the limits of global
change,
Terra Nova, 14, 129-155.
Hoffman, P. F., A. J. Kaufmann, G. P. Halverson, and D. P.
Schrag (1998), A Neoproterozoic Snowball Earth,
Science, 281, 1342-1346.
Kirschvink, J. L. (1992), Late Proterozoic low latitude
glaciation: The Snowball Earth, in The
Proterozoic
Biosphere: A Multidisciplinary Study, edited by J.
W. Schopf and C. Klein, pp. 51-52, Cambridge Univ.
Press, New York.
—Philip A. Allen, Imperial College London, U.K
Replies
I have some quick comments on your meeting report concerning this summary paragraph:
re: " What is clear is that the original pillars that supported the Snowball Earth hypothesis—such
as the presence of iron formations, a low-latitude clustering of continental
plates, and the characteristic carbon isotopic composition of marine
carbonates deposited during deglaciation--have been discarded."
I beg to differ, both with your list of the original pillars and whether they have been discarded.
If you go back and read my original Snowball paper (Kirschvink, 1992,
pdf),
and my paleogeography paper in the same volume (from which the little snowball
note was excised by Bill Shopf,
pdf)
you will see that the "original pillars that supported the snowball hypothesis" were:
- The peculiar abundance of carbonate clasts in the Neoproterozoic diamictites
compared to Gondwanan or Pleistocene deposits.
- Dropstones associated with sediments typical of low-latitudes (carbonates
and evaporates).
- Paleomagnetic data showing that at least one clear glacial unit (the Elatina)
was at Sea level on, or very near, the Equator, and the lack of any clear
data for polar continents.
- The unique association of banded iron stones with the glacials.
Certainly none of these observations have changed, and even the paleogeography
in the PPRG chapter is not too bad (although certainly things like TPW were
not considered then). Clustering of plates in the Tropics was NOT one of
the pillars of this hypothesis - the mere fact that ONE continent on the Equator had bullet-proof data for widespread, sea-level glaciation was the important point that argued strongly for an ice-albedo runaway.
In the same paper, I made some suggestions on how to test this idea beyond the straightforward paleomagnetic work. These were:
- A. Global Glaciations should be synchronous (as noted much earlier by Harland and others).
- B. Sedimentary sections on widely scattered localities should have striking similar variations in lithology, due to the global-scale nature of the climatic signals and fluctuations thereof.
- C. The deep oceans should go anoxic, allowing metal-rich
waters to build up and eventually deposit the Neoproterozoic BIF deposits towards
the end of the glaciations.
Well, "A" is still being debated heavily, but certainly the Chinese/Namibian termination is supporting it, as you point out.
Item "B" is still running strong, with the recognition of the Cap Carbonates themselves (which, by the way, were not appreciated when I wrote the 1992 paper back in 1988!), as well as the tubers, crystal fans, BIFs, etc.
Item "C" is still in the running as well, despite your statement to the contrary.
Do you remember when I stood up in the discussion session at the meeting and
asked point blank if anyone present could explain how to get BIFs with the hydrothermal
REE fingerprints without sealing off the oceans with ice? NO ONE ANSWERED. Saying
that it is simply local rifting (
e.g., Williams) does not work, as we have had
630 myr of rifting since then without similar production of BIFs, much less BIFs
associated with glacial deposits at low latitudes. Remember Paul made the comment
that to get a BIF it is also necessary to stop the input of continental sulfate
to the oceans, as the sulfate reducers turn this into sulfide that knocks out
the iron before it could ever form a BIF. A Phanerozoic-style high-latitude-only
glaciation will not do that because the glacial runoff is loaded with sulfates
oxidized from the glacial flour. In contrast, a snowball does this nicely. The
statement in your article does not reflect that component of the meeting at all.
Hence, may I suggest that the original pillars of the Snowball Hypothesis are
still firmly in place?
Cheers,
Joe [Kirschvink]
Since when was "clustering of continental plates" a
"pillar" of the snowball hypothesis?
Let's see what the three papers you like to cite as
defining the hypothesis actually say about the
continental paleogeography.
(1) Kirschvink (1992)
"...large portions of the continental land masses
probably were within middle to low latitudes during
the late Precambrian glacial episodes, a situation
that has not been encountered at any subsequen time in
earth history."
He then goes on to explain how this preponderance of
continents in middle to low latitudes would raise the
planetary albedo, and also the albedo feedback
associated with glacioeustatic change.
Nothing about continents being clustered (the tropics
are large).
(2) Hoffman et al. (1998)
(p. 1345) "Fragmentation of the Rodinia supercontinent
may have contributed to the CO
2 drawdown [required for
glaciation] by creating many new continental margins,
which are major repositories for organic carbon in the
modern ocean, consistent with the high d13C values
observed before the glaciation. This is also
consistent with the observation that Sturtian and
Varangian glaciations accompanied the opening of the
Pacific and Iapetus oceans, respectively, and might
explain why the only known older examples of similar
carbon isotope excursions and low-latitude glaciations
accompanied the fragmentation of a late Archean
megacontinent."
(3) Hoffman & Schrag (2002)
(p. 148) "The Neoproterozoic snowball era was a period
of continental dispersal, involving the breakup of
supercontinent Rodinia and the aggregation of
megacontinent Gondwana."
If you must put the snowball hypothesis "on trial", is
it too much to ask that the trial be fair?
Better, why don't you line up different theories for
low-latitude Proterozoic glaciation, and ask which
best satisfies the agreed-upon observations?
This is what was done at the famous Maui conference on
the origin of the moon. Going into the conference, the
giant impact hypothesis would surely have failed an
up-or-down vote (a "trial"). But when forced to choose
between the known theories in a straw poll, giant
impact emerged less fatally wounded than the other
options.
I regret that no such poll was taken at Ascona. I have
no idea how the participants would have voted. But I
would not have been afraid to find out.
Yours sincerely,
Paul [Hoffman]
Submitted to EOS (AGU), 20 Dec 06
The Pillars of Snowball Earth
In his report of the Snowball Earth 2006 conference in Ascona, Switzerland, co-convenor
Philip A. Allen (EOS, 7 November 2006) described the "original pillars" of
the hypothesis as having "been discarded".
From my reading of the hypothesis (Kirschvink, 1992a,b; pers. comm. 1989), the
original pillars are the following. Each is more strongly supported now than
when the hypothesis was first proposed.
- That Cryogenian glacigenic deposits occur on virtually every paleocontinent.
- That Cryogenian glaciers flowed directly into the ocean close to the paleoequator
according to combined paleomagnetic and sedimentological evidence.
- That Neoproterozoic paleomagnetic data show that thick carbonate-dominated
successions were also deposited at paleolatitudes less than 35 degrees, indicating
a normal pole-to-equator temperature gradient.
- That the only regionally-extensive sedimentary Fe2O3 and MnO2 ores in the
past 1.9 Ga are intimately associated with Cryogenian glacial marine deposits,
implying exceptional perturbations in seawater chemistry as expected if the oceans
were ice covered for long periods.
- That in low paleolatitudes, Cryogenian glacial strata begin and end abruptly,
consistent with a climatic instability due to ice-albedo and other positive feedbacks,
which resulted in sudden glaciation and deglaciation in those regions.
References
Allen, P.A. (2006), Snowball Earth on trial. EOS, 87(45), pp. 495.
Kirschvink, J.L. (1992a), Late Proterozoic low-latitude glaciation: the snowball
Earth,
in The Proterozoic Biosphere: A Multidisciplinary Study, edited by J.W.
Schopf and C. Klein, pp. 51-52, Cambridge Univ. Press, Cambridge.
Kirschvink, J.L., (1992b), A paleogeographic model for Vendian and Cambrian time,
in The Proterozoic Biosphere: A Multidisciplinary Study, edited by J.W. Schopf
and C. Klein, pp. 569-581, Cambridge Univ. Press, Cambridge.
—
PAUL F. HOFFMAN, Harvard University
Reply to Hoffman’s Comment on Pillars of Snowball Earth
I welcome the signs
of a dialectical engagement in the evaluation of the Snowball Earth hypothesis.
Paul Hoffman proposes that on the basis of Kirschvink’s two-page chapter
[Kirschvink, 1992], there are five main pillars in support of the Snowball Earth
hypothesis. I cannot agree that the five pillars mentioned by Hoffman can be
discerned in Kirschvink’s original piece. A pillar must be a line of argument
and supporting data that the protagonist views as being strongly and uniquely
supportive of the hypothesis. Using the architectural metaphor, a pillar is solid,
supportive, and unchanging. There are no such pillars in Kirschvink’s article.
He admits that his hypothesis is “speculative” (p. 51) or, more precisely,
involves an alternate “equally speculative mechanism” to that of
high obliquity to explain apparently low-latitude glaciation. Kirschvink suggested
that there were several implications of his global snowball model, but no compelling
evidence was presented.
There are no pillars evident in Kirschvink’s article, but papers by Hoffman
and coworkers elaborated the idea considerably [Hoffman et al., 1998; Hoffman
and Schrag, 2000, 2002]. These papers collectively provide the ideas and data
that allowed the hypothesis to be propagated. To take one example, the single
most emphatic line of evidence cited by these authors is the negative carbon
isotopic ratio of carbonates directly overlying and locally underlying glacial
diamictites or correlative unconformities. Hoffman and Schrag [2002, p. 142]
refer to the isotopically light cap carbonates as “…the ‘smoke’,
if not the ‘gun’ of the snowball Earth.” Since the origin of
the isotopic signature of Neoproterozoic carbonates is currently under intense
debate, as described in the meeting report, I would argue that this ‘pillar’ is
in danger of collapse, and it is certainly unsafe to associate all negative
carbon isotopic excursions with global glaciation. Metaphorically, there might
be some
smoke, but we are unsure of what gun or guns it came from.
The so-called pillars mentioned by Hoffman are an airbrushing of the short historiography
of this scientific debate. Rather than argue about the pillars, the protagonists
and antagonists would do well to be clear about what it is they are contesting,
and to agree to a procedure for resolving the contest. In this case, opposing
viewpoints contest whether or not there were a small number of fully global glaciations.
I suggest that our crucial, binding experiment can be tested by asking questions
related to the global synchroneity of glaciation in the Neoproterozoic. There
may be any number of other contradictory matters, but if glaciations were global
and were driven by runaway ice-albedo and super-greenhouse effects, we must be
able to say with some confidence that glaciations were synchronous within the
limits of analytical resolution.
References
Kirschvink, J. L. (1992), Late Proterozoic low-latitude glaciation: The Snowball
Earth, in The Proterozoic Biosphere: A Multidisciplinary Study, edited by
J. W. Schopf and C. Klein, pp. 51–52, Cambridge Univ. Press, New York.
Hoffman, P. F., and D. P. Schrag (2000), Snowball Earth, Sci. Am., 282(1),
68–75.
Hoffman, P. F., and D. P. Schrag (2002), The snowball Earth hypothesis: Testing
the limits of global change, Terra Nova, 14, 129–155.
Hoffman, P. F., A. J. Kaufman, G. Halverson, and D. P. Schrag (1998), A Neoproterozoic
snowball Earth, Science, 281, 1342–1346.
Philip A. Allen, Department of Earth Science and Engineering, Imperial College
London
