The interaction between fire, vegetation, landscape features and climatic conditions is a major determinant of many ecosystems’ spatio-temporal dynamics and function. Empirical studies of wildfire regimes are invaluable but usually limited to spatial and temporal scales much smaller than those on which fire ecosystems develop. Simulation models are therefore widely used for exploring wildfire regimes. Wildfire models have been developed both in ecology and statistical physics. Fire models in ecology cover a wide range of ecosystems, geographic areas, and scales. They are often tailored to specific regions and conditions. In statistical physics fire models are used to reproduce and explain power laws, which are also observed in wildfire size distributions. Thus, most wildfire models from ecology are too complex and those from statistical physics are too simple to maximize payoff in terms of general understanding and prediction at larger scales. Linking these two types of models would allow to integrate the insights gained in both fields and to overcome their mutual limitations. We took two generic and simple ecological wildfire models and compared them with the most widely used model from statistical physics, the Drossel-Schwabl model (DSM). All three models were tested against three patterns known from the field: fire size distributions, fire shapes, and the relation between community diversity and disturbance intensity. We extended the DSM to include succession for this comparison.

Results/Conclusions

All three models are able to reproduce not only one empirical pattern (power laws), but also surprisingly well patterns in fire shapes and diversity-disturbance relationships. We then found a generic formulation of a forest fire model that includes all three specific fire models considered as special cases. Our results show that the DSM and both ecological landscape fire models have the same explanative power, and limits, with regard to patterns in wildfire systems. Although they differ largely in their definition and purpose they share the same core assumptions. They belong to the same class of models in which the analytical tractability of the DSM is combined with the greater wealth of aspects, especially of succession dynamics, considered in ecological models. This unification allows for a more mechanistic understanding of wildfire regimes at large spatio-temporal scales, of the current differences and future changes of wildfire regimes in different regions, and it helps generating several testable predictions.