We disagree with Gail Whiteman and colleagues that there is “likely” to be a large and sudden release of methane from the East Siberian Arctic Shelf (Nature 499, 401–403; 2013).

Such an event would require an almost 1,000-fold regional increase in methane emissions from thawing permafrost, which would be inconsistent with geological evidence: although the concentration of atmospheric methane rose in response to abrupt warming during recent deglaciations, isotopic methane measurements do not indicate that the gas came from marine gas hydrates during these periods (H. Fischer et al. Nature 452, 864–867; 2008).

Sea-floor temperatures determine the stability of methane hydrate reservoirs. To our knowledge, there is no evidence that these temperatures have changed significantly in the past few decades, or that a sudden change is imminent.

Although elevated methane concentrations have been observed in water and the atmosphere above the East Siberian Arctic Shelf, it is not clear whether these have been caused by recent warming or by natural processes linked to glacial–interglacial changes (V. V. Petrenko et al. Science 329, 1146–1147; 2010). The global increase in methane recorded in the past few years does not seem to have been caused by enhanced Arctic emissions.

We welcome studies that quantify the economic impact of global warming, but they need to be accompanied by a realistic assessment of the uncertainties.

Gail Whiteman, Chris Hope and Peter Wadhams respond: Since 2005, accelerating sea-ice retreat in the Arctic has exposed the sea bed to much warmer conditions than those considered in earlier studies that anticipated a slow release of methane. Summer water temperatures off Siberia now climb to several degrees above 0 °C (see, for example, N. R. Bates et al. Biogeosciences 10, 5281–5309; 2013), causing the upper layers of offshore permafrost to melt more rapidly than they did a decade ago.

We have rerun our model with the same total quantity of emitted methane, but released over 50 or 75 years rather than 10 years. The results show no reduction in the total cost to society — in fact, the discounted costs over time would be larger.