Straits ofGibraltar

Atlas Mountains

CanaryIslands

Agadir canyon

Agadir basin

Madeira

Madeira abyssal plain

Seineabyssal plain

The seabed of the oceans is prone to failure, generating high-density sediment flows of immense volume that can run for hundreds to thousands of kilometres from the conti-nental margin to the deep oceanic plains. On page 541 of this issue, Talling and co-workers1 report on one such flow, which occurred about 60,000 years ago off northwest Africa.

The authors estimate that, in a period of hours or at most days, 22.5 × 1013 kilograms of sediment was transported into the ocean depths — ten times more than is annually delivered into the oceans by rivers. The sedi-ment flow started as a submarine landslide; it moved downslope as a turbulent suspension as it was funnelled through the narrows of the Agadir canyon (Fig. 1); it was then transformed into a cohesive mass of debris by a slight reduc-tion in the slope of the seabed. The work is significant in documenting the way in which seabed-hugging sediment flows undergo trans-formations in their flow behaviour. It also high-lights the hazard that these huge flows pose to seabed installations, and to coastal communi-ties given the potential of such flows to trigger a tsunami.

It is well over a hundred years since the Swiss limnologist François-Alphonse Forel (1841–1912) detected sediment flows in the Lake of Geneva caused by floods of the River Rhône. Later, in 1929, an earthquake triggered a sedi-ment flow on the Grand Banks off Newfound-land that broke successive telecommunication cables on the seabed, and so demonstrated the existence of fast-moving, long-distance sediment flows across the ocean floor. Today, improved imaging techniques, combined with recovery of cores of sediment that sample hun-dreds of thousands of years of seabed activity2, are transforming our view of the deep ocean floor. The picture is a very dynamic one.

The surface processes on continental margins such as that of northwest Africa are increasingly well understood, and show many signs of seabed instability, from the rock ava-lanches derived from collapse of the volcanic islands of the Canaries3, to the extensive swaths of gigantic debris units, such as the Saharan flow4, to the fretting of the continental margin by channels that focus dilute turbulent suspen-sions on their way to the deeps of the Madeira abyssal plain5 (Fig. 1).

Talling and co-workers1 identified a deposit, or bed, of sediment and debris that is less than 2 metres thick but that can be followed for a staggering distance of more than 1,000 kilo-metres. Over part (250 kilometres) of this dis-tance, the bed contains a cohesive unit of debris flow (debrite) encased in sand (turbidite), attrib-uted to the activity of more dilute turbulent suspensions known as turbidity currents. The onset of debrite deposition is correlated with a small but significant reduction in seabed slope from 0.05° to 0.01°. In other words, the debrite and the encasing turbidite are co-genetic, or ‘linked’6, appearing like a sandwich7, and pro-viding evidence of a downslope transformation of a turbidity current into a cohesive debris flow triggered by a reduction of slope. Conventional thinking, by contrast, is that debris flows most commonly evolve from the disintegration of moving landslide material, and in turn became more dilute and turbulent during downslope transport as they take in water8.

The processes recognized on the northwest-ern margin of Africa are assumed to be typical of the main way in which mass is transferred to the deep ocean. In the past, sedimentologists recognized trends in the vertical superimposi-tion of beds of turbidites and debris flows, but the lateral persistence and variability of beds were largely unknown. One way of bridging this gap in understanding the three-dimen-sional geometry of deep marine deposits is to build a statistical model using the scaling relationships between bed thickness, lateral extent and volume9–11. A lateral transforma-tion of debrite to co-genetic turbidite would be troublesome to these scaling relationships.

Another way to make progress is to docu-ment patterns in the architecture of deep marine deposits using selected examples. The Marnoso arenacea formation — once such a deposit but now evident on land as part of the Apennine mountains in northern Italy — is a celebrated case12. Such documentations offer the possibility of assessing the ‘run-out’ dis-tances of individual flows. In the case of co-genetic flows, it may be possible to answer the intriguing question of whether gigantic run-out distances are aided by flow-transformation processes6.

Deep marine deposits commonly contain valuable resources such as hydrocarbons, so energy companies have been keen to map them in three dimensions. Three-dimensional heterogeneity affects the ability to extract liquid and gaseous resources from these depos-its, and it primarily originates from the inter-bedding of two types of deposit, sandstones and mudstones, left by the main body and dilute tail of turbidity currents respectively. The lateral transformation from a cohesive debrite (which has low porosity) to a tur-bidite (with high porosity) is another source of heterogeneity. If such a transformation can be triggered by a slight reduction in slope, as Talling and co-workers1 suggest, the question arises of whether it is possible to predict such

EARTH SCIENCE

Sediment en route to oblivion Philip A. Allen

Sudden collapses of the sea floor can generate oceanic sediment flows that dwarf the global annual sediment input from rivers. Such flows can travel great distances, and undergo transformation along the way.

Figure 1 | Ocean bathymetry off the northwest coast of Africa. The pathway of the sediment flow responsible for the deposit (called bed 5) investigated by Talling et al.1 is depicted by arrows. The flow started as a submarine landslide, and passed through the Agadir canyon. Part of it was deposited on the Seine abyssal plain; part travelled through the Agadir basin en route for the Madeira abyssal plain. These plains lie more than 4 kilometres beneath the ocean surface, and the Madeira abyssal plain is more than 1,500 kilometres distant from the site of the initial landslide. (Bathymetric image courtesy of T. Le Bas.)

490

NATURE|Vol 450|22 November 2007NEWS & VIEWS

POC

POC +TEP

PresentCO2

DOC

TEP

Sinking Sinking

Thermocline Thermocline

Air

Sea

Air

SeaDOC

TEP

POC

POC + TEP

EnhancedCO2

ba

Aggregation

ExudationExudation

Aggregation

Marine phytoplankton are major players in the carbon cycle, accounting for about 50% of the global biological uptake of carbon dioxide1. Near the ocean surface, these single-celled organisms use light energy to convert CO2 into organic molecules for building cellular struc-tures and driving their metabolism. Some of this organic carbon eventually sinks into the deep ocean, where most of it is either converted back to CO2 or sequestered in sediments. This ‘biological pump’ effectively removes CO2, a greenhouse gas, from the atmosphere for hundreds to millions of years.

On page 545 of this issue, Riebesell et al.2 describe evidence that the biological pump may become stronger at elevated concentrations of CO2 in the atmosphere, and thus provide a neg-ative feedback on increasing atmospheric CO2. According to their calculations, that feedback has accounted for about 10% of the extra CO2 pumped into the atmosphere since pre-indus-trial times (the past 200 years or so).

Since industrialization, atmospheric CO2 has risen from about 280 parts per million (p.p.m.) to more than 385 p.p.m., increasing by some 2 p.p.m. per year during the past decade. Each year, approximately 25–30% of anthropogenic CO2 enters the surface ocean, where it increases both the concentration of dissolved inorganic carbon (DIC) and acidity. The latter has poten-tially adverse consequences for phytoplankton that require calcium carbonate to build their shells. Although the oceans are Earth’s largest reservoir for DIC, only about 1% is in the form of CO2, the molecule required by the photo-synthetic enzyme rubisco. At the low CO2 con-centrations typical of sea water, rubisco operates at rather low efficiency3. So increasing ambient concentrations of CO2 in sea water could boost

photosynthetic efficiency and increase bio-logical uptake of anthropogenic CO2, just as some marine phytoplankton use intracellular carbon-concentrating mechanisms to increase their photosynthetic capacity.

This is the context for Riebesell and col-leagues’ research2 into how phytoplankton might respond to increasing CO2 concentra-tions. They conducted CO2 manipulations in large cylindrical enclosures called mesocosms

Figure 1 | Carbon dioxide in the atmosphere, and organic carbon in the ocean. a, The size, relative to part b, of the different pools and fluxes of organic carbon under present levels of atmospheric CO2 (POC, particulate organic carbon; DOC, dissolved organic carbon; TEP, transparent exopolymeric particles). The thermocline is an abrupt temperature discontinuity that acts as a barrier between the upper mixed ocean and deeper waters. b, According to the results of Riebesell et al.2, uptake of CO2 by phytoplankton increases at enhanced CO2 concentrations (thicker arrow). Exudation of DOC from the pool of POC (primarily phytoplankton) also increases, although the POC pool itself remains unchanged. This extra DOC coalesces to form a larger pool of TEP that facilitates increased POC aggregation and enhances sinking fluxes. Thus, the flux of carbon from the atmosphere to the deep ocean is increased at higher atmospheric concentrations of CO2.

slope-induced changes and their effects on flow behaviour. Given that many deep marine deposits occur in sedimentary basins affected by faulting, however, and that high-density sediment flows commonly enter semi-con-fined basins with lateral slopes, the answer in most cases is probably ‘no’.

The bed-by-bed correlation carried out by sedimentologists on surface exposures such as the Marnoso arenacea, or on cores from below the ocean floor as documented by Talling et al. in their Supplementary Information1, is a painstaking business. The reward for such endeavours is an appreciation of the true

variability of deep marine sediments and of the mechanics of their deposition. The large volumes of sediment en route to oblivion in the dark ocean basins reveal a complexity of seabed processes unimagined even by the visionary Forel, who likened his beloved Swiss lakes to small oceans. ■

Philip A. Allen is in the Department of Earth Science & Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ, UK. e-mail: philip.allen@imperial.ac.uk

1. Talling, P. J. et al. Nature 450, 541–544 (2007).

CARBON CYCLE

Marine manipulations Kevin R. Arrigo

The effect of increasing levels of atmospheric carbon dioxide on carbon uptake in and export from the upper ocean is one of the big questions in environmental science. But it can be tackled experimentally.

that were placed in a fjord in southern Nor-way and extended from the surface to a depth of approximately 9–10 metres. Although this approach is complex and logistically difficult, the advantage is that mesocosms are exposed to the same environmental influences as the surrounding waters, making them reasonable analogues for natural systems. And they can be manipulated experimentally. In Riebesell and colleagues’ study, phytoplankton were grown in different mesocosms with the partial pressure of CO2 adjusted to simulate the present (350 µatm) or projected future (2 × present CO2, 700 µatm, and 3 × present CO2, 1,050 µatm) atmosphere.

What the authors found was intriguing. Uptake of CO2 by phytoplankton (mainly bloom-forming diatoms and coccolithophores) in the 2 × CO2 and 3 × CO2 treatments was 27% and 39% higher, respectively, than in the present-day CO2 treatment. But the additional

2. Weaver, P. P. E., Rothwell, R. G., Ebbing, J., Gunn, D. E. & Hunter, P. M. Mar. Geol. 109, 1–20 (1992).

3. Masson, D. G. et al. Earth Sci. Rev. 57, 1–35 (2002).4. Gee, M. J. R., Masson, D. G., Watts, A. B. & Allen, P. A.

Sedimentology 46, 317–335 (1999). 5. Masson, D. G. Basin Res. 6, 17–33 (1994).6. Haughton, P. D. W., Barker, S. P. & McCaffrey, W. D.

Sedimentology 50, 459–482 (2003).7. McCaffrey, W. & Kneller, B. Bull. Am. Assoc. Petrol. Geol. 85,

971–988 (2001).8. Mohrig, D., Whipple, K. X., Hondzo, M., Ellis, C. & Parker, G.

Geol. Soc. Am. Bull. 110, 387–394 (1998).9. Rothman, D. H., Grotzinger, J. P. & Flemings, P. J. Sediment.

Res. 64, 59–67 (1994).10. Malinverno, A. Basin Res. 9, 263–274 (1997).11. Sinclair, H. D. & Cowie, P. A. J. Geol. 11, 277–299 (2003).12. Amy, L. A. & Talling, P. J. Sedimentology 53, 161–212

(2007).

491

NATURE|Vol 450|22 November 2007 NEWS & VIEWS