A tale of two basins...
The Illilouette Creek Basin and Sugarloaf Creek Basin have been staples in the Thompson Lab for years now (as they were the subject of an NSF CAREER project, and then several PhDs and postdocs' work - shout out to Katya, Gabby and Tavia). One of the strange things we learned about these basins is that even though both of them had experienced restored fire regimes since the 1970s, there had been massive vegetation and hydrological changes in Illilouette Creek Basin, while Sugarloaf had... just rearranged a few vegetation patches, without macroscopic change. Why? We looked into this using simple models.
Tavia Crompton developed a very simple dynamical systems model, which made it easy to calibrate against the vegetation change data and fire data we had for the two basins. The model predicts biomass of two kinds of vegetation - a canopy and an understory, which grow at different rates and are affected by fire in different ways. We explored what happened if the restored fire regime were maintained long-term, or reversed - and we identified the best sets of model parameters to represent the forest and fire dynamics we observed. The model was calibrated for Illilouette Creek where we had the best data.
Then we took the calibrated model (shown below in green) and explored what would happen if we increased or reduced the fire return interval, increased the burn severity or increased climate dryness. The main observation was that if we reduced the return interval seen at Illilouette Creek to match that we'd seen at Sugarloaf Creek, we no longer saw the reduction in forest cover we'd observed in Illilouette Creek Basin. So, the differences between the basins seem likely to be explained by the differences in fire return interval in the basins.
One nifty thing about simple models is that they make it really easy to explore cause and effect - allowing us to check how sensitive the forest cover (as modeled here) was to changes in severity, return interval of the fire, climate and growth rates of the plants.