What is the natural rhythm of temperate and boreal forest disturbances in the absence of human management?

The vast majority of temperate and much of the boreal forest have been completely transformed by human activities, changing forest composition and disturbance regimes. Whilst our capability to observe this transformed state has improved dramatically in recent years, we have precious little information on the state of these forests in the absence of management. To what extent do our forests currently suffer from a surplus or a deficit of disturbance relative to their natural state? What are the implications of this for carbon turnover? Using a novel fusion of satellite observations of stand-replacing disturbances in 80 protected areas, statistical analysis and dynamic vegetation modelling, we generated wall-to-wall estimates of disturbance frequency across northern hemisphere temperate and boreal forests. Analysis of disturbance events in the protected areas revealed that the probability of disturbances from agents including fire, wind-throw and bark beetles was related to community mean functional traits and climate. We used the LPJ-GUESS dynamic vegetation model, which explicitly simulates plant functional types covering different successional stages, to simulate forest functional composition in the absence of human management. We interactively coupled this simulation to a new disturbance probability module to generate estimates of natural disturbance probability across all northern-hemisphere temperate and boreal forests. Disturbance frequency ranged from ca. one stand-replacing event per hundred years in parts of the boreal to less than one per thousand years in broadleaved temperate forests. In many regions the unmanaged disturbance frequencies differed dramatically from those observed in reality over the last two decades, with both disturbance surplus and deficits being recorded. In addition to providing the first quantitative continental-scale assessment of human impact on forest disturbance regimes, our results also provide a lightweight modelling approach for the inclusion of natural disturbances in large-scale vegetation models. They thus facilitate simulation of forest structure, a crucial driver of ecosystem function, from carbon uptake to biodiversity.