The paper “Global ecosystems and fire: a multi-model assessment of fire-induced tree cover and carbon storage reduction” (, led by Gitta Lasslop (Senckenberg: has come out today. This paper analyses one of the FireMIP modelling experiments, in which fire is completely suppressed and contrasts this experiment with historical simulations when fire occurs.

Effectively, this is a repeat of the famous “world without fire” model experiment carried out by Bond et al. in 2005. However, as is often the case when you repeat an experiment with multiple models — the FireMIP paper is based on results from seven different fire-enabled vegetation models — the results turn out to be more complex than we thought. Our estimate of the global reduction in tree cover and the carbon stored in vegetation caused by fires is smaller (10%) than the 50% reduction shown by Bond et al. This is largely because human-induced changes in land use, included in our simulations but not in the Bond et al. simulation, already reduces the amount of fire.

The picture below shows the percentage tree cover when there is no fire (TC no fire) and the difference in tree cover between the historic simulation with fire and the simulation without fire (TC fire-TC no fire). The biggest changes are in fire-prone regions like the savannas.