emissions, yet they are poorly constrained 2,13,14 . There are large uncertainties not only 57 in the current magnitude of these fluxes, but also in the factors that regulate them 2,13 . 58In particular, there is substantial uncertainty in the parameterisation of the 59 temperature dependence of natural CH 4 emissions in process-based biogeochemistry 60 models [15][16][17][18] , which greatly hinders our ability to predict the response of this key 61 component of the carbon cycle to global warming. For example, temperature 62 sensitivities for ecosystem-level CH 4 emissions have reported apparent activation 63 energies that range from 0.2 to 2.5 eV 6,[19][20][21] (1 eV = 96 kJ mol -1 ). 64In a bid to reduce this uncertainty, which is fundamental to improving 65 projections of future carbon cycle-climate change feedbacks 15-18 , we quantified 66 variation in the temperature dependence of CH 4 fluxes for three different types of 67 experiments -i.e. methanogenic cultures, anaerobic sediment slurries, and seasonal 68 field surveys of CH 4 emissions -that correspond to three distinct levels of biological 69 organisation -i.e. population, community, and ecosystem, respectively. In particular, 70 we assess whether ecosystem-level CH 4 emissions exhibit a temperature dependence 71 similar to that of the underlying methanogenic process, and quantify the magnitude of 72 between site deviations from this physiological response. To do this, we first establish 73 the magnitude and variability of the temperature dependence of key metabolic rate 74 processes (i.e. methanogenesis, growth) for populations of methanogens in culture, as 75 well as the temperature dependence of CH 4 production for anaerobic microbial 76 communities in slurries. We then assess whether these temperature dependencies 77 4 differ from those observed in an ecosystem-level analysis of the seasonal temperature 78 dependence of natural CH 4 emissions from aquatic, wetland and rice paddy 79 ecosystems (see S1 of the Supplementary Information). Our ecosystem analysis 80 includes both new and previously published data that together encompass 1553 paired 81 estimates of CH 4 emission and temperature taken from 126 field sites. 82To directly characterise the physiological temperature dependence of key 83 metabolic rate processes for methanogens, we compiled data on rates of 84 methanogenesis and growth from laboratory cultures of methanogen populations as 85 well as rates of CH 4 production from microbial communities in anaerobic sediment 86 slurries (see S1 of the Supplementary Information). We then separately fit the data 87 compiled for each type of experiment to a Boltzmann-Arrhenius function, which 88 characterises the exponential relationship between metabolic rate and temperature 89 assuming a single enzyme catalysed reaction is rate-limiting 22 , using a linear mixed-90 effects model (see S2 of the Supplementary Information) of the form 23 91(1) 92 where is the natural logarithm of the measured rate of CH 4 production or 93 growth rate at absolute tem...