identified: a low, Salvinia-free state, and a high state of dense Salvinia biomass. Under some conditions, these two stable states can coexist and population trajectories may be attracted to either state. Under other conditions there might be only a single state that is attractive. With flooding events and stochastic forcing, the system may be bouncing in a complex way between states, making it almost impossible to ascertain the underlying rules or patterns just by looking at the Salvinia or weevil time series.

The model that emerges from Schooler and colleagues’ analysis1 provides a useful tool for understanding the driving forces behind the Salvinia–weevil system — and its alternative stable states — that would otherwise be dif-ficult to identify. With that as background,

the authors discuss how the modelling frame-work helps to suggest practical solutions for biological control. In particular, they argue that it may be possible to take advantage of the sys-tem’s stochastic fluctuations and its associated erratic jumping between alternative states. In the higher state, when Salvinia biomass is at high density, weevil control is least effective. However, augmenting weevil control at those times when the system is attracted to its lower state might possibly trap the system into a sta-ble Salvinia-free state. The model could thus help managers to identify the optimal time to apply biological control.

In all, Schooler and colleagues’ careful attention to data, and their development and implementation of modelling techniques,

set a new standard in ecological time-series analysis. Their approach promises to have many applications in future studies of noisy biological data sets. ■

Lewi Stone is in the Department of Zoology, Faculty of Life Sciences, Tel Aviv University, Ramat Aviv 69978, Israel. e-mail: lewi@post.tau.ac.il

1. Schooler, S. S., Salau, B., Julien, M. H. & Ives, A. R. Nature 470, 86–89 (2011).

2. Scheffer, M., Carpenter, S., Foley, J. A ., Folke, C. & Walker, B. Nature 413, 591–596 (2001).

3. Scheffer, M. et al. Nature 461, 53–59 (2009).4. Ives, A. R. & Carpenter, S. R. Science 317, 58–62

(2007).5. Ives, A. R., Abbott, K. C. & Ziebarth, n. L. Ecology 91,

858–871 (2010).6. Dowd, M. J. Mar. Syst. 68, 439–456 (2007).

radiation (insolation) in summer at northern low latitudes. Insolation reaches its maximum during the part of the precession cycle when the elliptical orbit of Earth takes the planet closest to the Sun during northern summer. The result is a stronger monsoon in Asia and other regions, with more summer precipita-tion, and consequently greater wetland areas and methane production by soil-dwelling microorganisms.

However, the increase of the past 5,000 years departed from this pattern, with an increase in atmospheric methane concentration at a time when northern summer insolation was decreasing. In influential papers5,6, Ruddiman proposed that the increase was due to human, especially agricultural, activity, which over-whelmed the variations in natural sources, even 5,000 years ago. There has been debate

about whether the much smaller human population of that period could really have had such a domi-nant effect. A strength of the hypoth-esis has been that the pattern of the past few millennia differed from that of earlier interglacials, which more closely followed the precessional insolation pattern.

Singarayer and colleagues’ approach2 involved an intensive modelling programme. To provide snapshots roughly every 2,000 years over the last glacial cycle, span-ning 130,000 years, they ‘forced’ the Hadley Centre’s HadCM3 coupled ocean–atmosphere general cir-culation model (GCM) with the appropriate orbital and ice-sheet con-figurations, and with greenhouse-gas concentrations, and ran the model to equilibrium. Other simulations were carried out, in which one or more of these forcings was held constant to isolate the causes of change.

The authors then fed the out-put of each climate simulation through a series of offline models

G l o B a l C h a n G e

Methane and monsoonsThe rising trend in atmospheric concentrations of methane over the past 5,000 years has been attributed to human agency. A modelling study, of a power that has only now become possible, points to another cause. See Letter p.82

e r i C W. W o l f f

Methane is a potent greenhouse gas, and influences the levels of other atmospheric constituents. The huge

increase, of about 150%, to nearly 1,800 parts per billion by volume (p.p.b.v.) in atmospheric concentration over the past two centuries1 is clearly caused by human activities. However, methane concentration also increased significantly, from about 550 to 700 p.p.b.v., over the previous 5,000 years — the later part of the (interglacial) Holocene epoch that began some 10,000 years ago. There has been intense debate about whether this rise was also anthropo-genic or was due to changes in natural sources and sinks. Using models of climate, vegetation and emissions, Singarayer et al.2 show how the increase could have arisen from natural causes (page 82 of this issue).

Detailed ice-core data for methane now cover the past 800,000 years3. They show a characteristic pattern over glacial–interglacial cycles, with higher values during interglacials. From the combined use of methane-concentration and isotopic data, it seems that the main cause of the glacial–interglacial rise was almost certainly an increase in the strength of wetland methane sources4, per-haps allied to a weakening of the atmospheric sink. Rapid fluctuations,

simultaneous with fast, millennial-scale climatic changes in the Northern Hemisphere, are also seen. Finally, the ‘envelope’ of data seems to follow closely the pattern of pre-cession in Earth’s orbit, which has a roughly 20,000-year cycle. The apparent reason for this is that tropical wetland emissions of methane respond to the amount of incoming solar

500

1,000

1,500

02,0004,0006,0008,000

117,000119,000121,000123,000125,000

Age (years before AD 1950)

Met

han

e (p

.p.b

.v.)

Present interglacialLast interglacialModern atmospheric data

Figure 1 | Atmospheric methane concentrations during the present and last interglacials. Almost all of the data come from ice cores1,3. The curves are for the past 10,000 years — the Holocene — and for the equivalent period (125,000–115,000 years ago, in terms of orbital precession) in the last interglacial. The red curve (with modern atmospheric data in blue) shows methane levels during the present interglacial, with a rise commencing 5,000 years ago. The green curve shows the contrasting continual decrease in methane during the last interglacial. Singarayer and colleagues’ modelling study2 can explain the trends in both interglacials in terms of Earth’s orbit, except for the past 200 years, when a marked anthropogenic effect has occurred.

3 F E b R U A R y 2 0 1 1 | V O L 4 7 0 | N A T U R E | 4 9

NEWS & VIEWS RESEaRch

© 2011 Macmillan Publishers Limited. All rights reserved

a r C h a e o l o G y

Trailblazers across ArabiaWhat role did the Arabian peninsula play in the expansion of our species out of Africa? An archaeological site in the United Arab Emirates provides tantalizing new evidence that supports an early human migration from Africa.

M i C h a e l d . P e T r a G l i a

Genetic and fossil information points to Africa as the original homeland of Homo sapiens, but the date and path

of human movements out of Africa remain un resolved. Fossil evidence indicates that H. sapiens had entered the Levant by 130,000 to 120,000 years ago, and that this population sur-vived there until 75,000 years ago1. Convincing human fossils from this period have not been found anywhere else in Eurasia, however, sug-gesting that the early Levantine occupation was geographically limited and temporary. Molecular geneticists have argued that human migrations along the rim of the Indian Ocean occurred rapidly about 65,000 years ago2, fol-lowing a coastal route to avoid the hyper-arid deserts of Arabia. Archaeologists have specu-lated that this coastal migration was accompa-nied by microblade tool industries — cultures associated with the manufacture of small stone blades — carried from southern Africa3.

Reporting in Science, Armitage et al.4 now describe strong evidence for a human presence 125,000 years ago in what is now the United Arab Emirates: stone-tool assemblages buried in sediments dating from that period. Remark-ably, the Arabian tools are similar to those made by anatomically modern humans living in Africa at that time. This implies an early dis-persal of H. sapiens along a route passing from the Horn of Africa to southern Arabia across the Bab al-Mandab Strait (Fig. 1).

The authors’ conclusions are based on sedimentary dating and archaeological finds, with no accompanying human fossils. So how convincing are their evidence and interpreta-tions? Let us first consider the archaeological site. The stone-tool assemblages were under an overhang of rock (known as a rock shelter) situated at the base of a steeply sloping moun-tain range called Jebel Faya in the southeastern part of the Arabian peninsula (Fig. 1). More specifically, the rock shelter is on a piece of land that juts out near the Strait of Hormuz.

Armitage et al. excavated trenches in and around the rock shelter, and discovered three Palaeolithic stone-tool assemblages. They dated the lowest-lying of these, assemblage C, using the optically stimulated luminescence technique. Of the three sedimentary samples taken, two had consistent ages of 127,000 and 123,000 years, but the third was only 95,000 years old. The disparity of the ages may point to some unexplained problems in the stratigraphy and its dating.

The stone artefacts from assemblage C were made using a combination of distinc-tive manu facturing methods (including the Levallois technique for striking prepared flakes from stones), and included a variety of flake tools, such as scraping implements. Small hand-held axes and thick leaf-shaped objects (foliates) were also made by the Jebel Faya inhabitants. The key point argued by Armitage et al. about these tools is that they are similar to those being made by modern humans in East

and calculations to estimate the distribution of vegetation types and methane emissions, and, finally, the resulting methane concentra-tion in the atmosphere. Using a well-resolved GCM for such an experiment is highly novel: the number of simulations was possible only because of recent advances in computing capacity (and because the BBC funded some of the work as research for a television series).

The model results do a good job of follow-ing the upper envelope of methane concentra-tion through the entire period, showing the pattern expected from the orbital precession, similar to that for summer insolation at 30° N. The model diverges from that pattern over the past 5,000 years, however, giving an increase in methane concentration just as shown by the ice-core data. Crucially, no such increase is seen in the model output for the analogous period of the last interglacial, which occurred between about 125,000 and 115,000 years ago.

The model experiments allow the authors to delineate which regions contributed to the global pattern of methane emissions. It appears that the late-Holocene rise was due to an increase in sources from South America south of the Equator that, combined with small increases from other regions, outweighed decreases in Eurasia and East Asia. This South American source reacts to an insolation sig-nal that has a quite different phase from the usually dominant northern one.

At the equivalent time in the last interglacial, larger changes in insolation caused even larger changes in emissions from each region. These were again finely balanced. The experiments in which only orbital forcing was varied do show a small increase in global methane emissions at roughly the equivalent time (in orbital terms) to that in the Holocene. However, the early start of glaciation during the last interglacial induced a further decrease in (high-latitude) Eurasian and North American emissions, preventing an upturn at the global scale.

The trends deduced from the model are the result of a subtle balance between several opposing effects, and it is therefore difficult to know whether the precise result is robust: it will need to be examined using other climate and vegetation models7. Nonetheless, Singa-rayer and colleagues’ study2 provides a sat-isfactory explanation of both the increase in methane concentration in the late Holocene and the decrease during the last interglacial (Fig. 1) — an ‘early anthropogenic’ influence on methane is no longer required, although of course it cannot be ruled out. It is also worth noting that the good fit of the model and data over the last interglacial, including the period when the Arctic was apparently several degrees warmer than at present8, leaves little room for a large influence of additional emissions from methane hydrates in permafrost or marine sediments under such conditions.

Some aspects of the problem remain to be solved. Although this work2 does a good job

of tracing the envelope of observed methane concentrations over the entire glacial cycle, it gives no hint of the large and rapid changes observed at millennial timescales. Finally, in the same paper5 in which he suggested an early human influence on methane, Ruddiman also suggested that the slow rise in carbon diox-ide over the past 8,000 years was a result of human actions. This idea has been strongly rebuffed9,10, but there is not, as yet, any con-vincing study showing why the Holocene behaved differently from the last interglacial for carbon dioxide. ■

Eric W. Wolff is at the British Antarctic Survey, High Cross, Madingley Road,

Cambridge CB3 0ET, UK. e-mail: ewwo@bas.ac.uk

1. MacFarling Meure, C. et al. Geophys. Res. Lett. 33, L14810, doi:10.1029/2006GL026152 (2006).

2. Singarayer, J. S., Valdes, P. J., Friedlingstein, P., nelson, S. & Beerling, D. J. Nature 270, 82–85 (2011).

3. Loulergue, L. et al. Nature 453, 383–386 (2008).4. Fischer, H. et al. Nature 452, 864–867 (2008).5. Ruddiman, W. F. Clim. Change 61, 261–293 (2003).6. Ruddiman, W. F. Rev. Geophys. 45, RG4001,

doi:10.1029/2006RG000207 (2007).7. Weber, S. L., Drury, A. J., Toonen, W. H. J. &

van Weele, M. J. Geophys. Res. 115, D06111, doi:10.1029/2009JD012110 (2010).

8. otto-Bliesner, B. L. et al. Science 311, 1751–1753 (2006).

9. Broecker, W. S. & Stocker, T. F. EOS Trans. 87, 27 (2006).

10. elsig, J. et al. Nature 461, 507–510 (2009).

5 0 | N A T U R E | V O L 4 7 0 | 3 F E b R U A R y 2 0 1 1

NEWS & VIEWSRESEaRch

© 2011 Macmillan Publishers Limited. All rights reserved