The work builds on observations seen in recent years. Over the twentieth century, land was abandoned in Spain, leading to afforestation of areas. Fire, it has been proposed, has helped this change by clearing land for oaks and other trees to settle. Fire has effectively been a reset button.
Baudena ane colleagues argue this model is too simplistic to explain current changes in Spain. In particular they note two problems. One is increased fire frequency. Fires are not always clearing old growth with long established seed banks. Instead the fires burn over recently released seeds. Shrubs that have yet to die back from overshading, can leave fresh seeds ready for the next clearing. In contrast, tree seedlings do not leave fresh seeds for regeneration till they reach maturity.
In addition the soil the seedlings will be growing into will be drier, due to higher temperatures and reduced rainfall. Water stress will have an impact on which seedlings survive to maturity and which do not.
To discover what effect the combination of fire and increased aridity had on forest resilience, the team built a model. They set the parameters for difference climates and fire frequencies and then ran the models over short (decades to centuries) terms and long (centuries to millennia) terms.
Under historical conditions, oak forest dominated over scrubland. Adding in random fires prevented a steady state, but oak tended to dominate sooner or later, over the scale of centuries. However, when aridity was added to the model, things changed.
The authors saw more open shrubland, with the fires maintaining the clearance. “For the open shrubland state to occur, aridity needed to affect at least two different factors, e.g. reduction of the resprouting ability and the colonization capacity of oaks…” the authors write. “In contrast, if aridity decreased only oak colonization ability, but no fires occurred, the model would converge to an oak forest for all considered aridity levels. The only effect of aridity in the absence of fire would be a reduction in oak cover (from around 0.90 to 0.77), with grasses coexisting with oaks at the strongest aridity level considered…”
Interestingly a Monte Carlo analysis (running lots of random variations) showed that you could not be certain what the final state of the model would be, from the set-up of scrubland and forest that you gave it to start with.
One of the more startling findings of the paper is that the loss of oak cover is not predictably steady. Thanks to the random element of fire, forests can disappear suddenly and irreversibly. The first clue that the resilience of a forest is not as high as it might appear might be when it disappears for good. “Our study highlights the necessity and urgency of including fire-related functional types and post-fire responses for prediction of fire ecosystems under future scenarios,” conclude Baudena and colleagues.