Abstract
<jats:p>Background. Higher spatio-temporal variation in fire regime attributes in fire-dominated landscapes (pyrodiversity) is thought to promote higher levels of biodiversity – the pyrodiversity-biodiversity hypothesis – yet this effect likely varies across taxa and systems. As a result, ecosystem-based management in heterogenous, fire-dominated forests requires understanding the strength and shape of local pyrodiversity-biodiversity relationships (PBRs). Landscape simulation models can be an effective tool for quantifying the PBR, but they require evaluation of key assumptions. Using LandR, a forest landscape model, evaluated against a dendroecological dataset of historical fire regimes and age structure, we investigated the PBR across forest types of the Montane subregion of the southwestern Alberta Foothills, Canada. We used n-dimensional hypervolumes as a multivariate measurement of pyrodiversity and forest diversity and examined how the PBR changed under two modelling assumptions about fire-caused tree mortality – partial mortality (mixed-severity model) and complete mortality (stand-replacing model).Results. Comparisons against dendroecological data showed that the mixed-severity model provided better estimates of fire intervals, and of old growth and forest diversity under the observed short fire intervals. However, it overestimated Pseudotsuga menziesii dominance and age across the landscape. Both models showed little support for positive PBRs in this area and revealed that PBR strength and direction varied between models, scales and forest types. These results highlight the context-dependency of this relationship.Conclusions. Our findings underscore how fire severity modelling assumptions that support land management frameworks can have serious consequences for forest diversity and old growth. We also caution against extrapolating observed PBRs across systems and fire regimes, especially in heterogeneous landscapes.</jats:p>