It has been suggested that a decrease in atmospheric methane levels triggered the progressive rise of atmospheric oxygen, the so-called Great Oxidation Event, about 2.4 Gyr ago1. Oxidative weathering of terrestrial sulphides, increased oceanic sulphate, and the ecological success of sulphate-reducing microorganisms over methanogens has been proposed as a possible cause for the methane collapse1, but this explanation is difficult to reconcile with the rock record2, 3. Banded iron formations preserve a history of Precambrian oceanic elemental abundance and can provide insights into our understanding of early microbial life and its influence on the evolution of the Earth system4, 5. Here we report a decline in the molar nickel to iron ratio recorded in banded iron formations about 2.7 Gyr ago, which we attribute to a reduced flux of nickel to the oceans, a consequence of cooling upper-mantle temperatures and decreased eruption of nickel-rich ultramafic rocks at the time. We measured nickel partition coefficients between simulated Precambrian sea water and diverse iron hydroxides, and subsequently determined that dissolved nickel concentrations may have reached approx400 nM throughout much of the Archaean eon, but dropped below approx200 nM by 2.5 Gyr ago and to modern day values6 (approx9 nM) by approx550 Myr ago. Nickel is a key metal cofactor in several enzymes of methanogens7 and we propose that its decline would have stifled their activity in the ancient oceans and disrupted the supply of biogenic methane. A decline in biogenic methane production therefore could have occurred before increasing environmental oxygenation and not necessarily be related to it. The enzymatic reliance of methanogens on a diminishing supply of volcanic nickel links mantle evolution to the redox state of the atmosphere.