Fire interval distributions in landscapes with alternative stable states

KITZBERGER, THOMAS; TIRIBELLI, FLORENCIA; MORALES, JUAN MANUEL

St Andrews

Congreso; International statistical ecology conferece 2018; 2018

University of St Andrews

The relationship between flammability and time since fire modulates feedback mechanisms between fire and vegetation which in turn give ecosystems different resilience to climatic variability and fire regime shifts. Fire frequency analysis (or survival analysis), give us a framework to assess changes the flammability of landscapes along post-fire age. In these analysis, fire intervals are employed to fit distributions from which a flammability function can be derived describing the chance of burning with increasing time since fire. Originally, the flammability function was derived from a Weibull distribution following the idea that flammability would either increase exponentially with time or remain constant. But, there are other distributions that describe relationships among flammability and time since fire that are biologically more plausible. For example, distributions where flammability increases with time since fire up to an asymptote or where it initially increases but then decreases. One important assumption of these analysis is that aging elements are part of the same statistical population, meaning that the landscape is composed of the same vegetation type that changes its flammability as it ages. In heterogeneous landscapes displaying alternative stable states this assumption is not met, as different states are dominated by different vegetation types that may have contrasting age-flammability relationships. Thus, in these landscapes the derived flammability function cannot be attributed to a given vegetation type (or state), it will either represent the most abundant vegetation type, the vegetation that dominates the fire regime, or may be an average of both. Our main objective was to fit flammability functions to landscapes of NW Patagonia composed by alternative stable states: pyrophytic shrublands with heliophyllous fast-growing species which produce a rapid fuel buildup due to their sometimes multi-stemmed architecture, and pyrophobic forests dominated by colonizer, fire-sensitive species that create closed canopies with moist and dark understories as they age.Using a Bayesian approach, we fitted fire interval distributions to landscapes composed of different proportions of shrublands and forests using censored and not censored data. Moreover, to evaluate the ability of survival analysis to detect the flammability functions in landscapes with alternative states, we fitted fire interval distributions to fire histories simulated in landscapes with different proportions of pyrophobic and pyrophytic vegetation types and different fire frequencies with known flammability functions. Our results show that in heterogeneous landscapes the pyrophytic vegetation determines the relationship between flammability and time since fire even when it is not abundant in the landscape. When the landscape had shrublands of heliophilous fast-growing species, flammability increased with time since fire, whereas only when it was almost exclusively dominated by closed-canopied forests flammability increased the first decades after fire and then decreased. This suggests that when landscapes are heterogeneous the pyrophytic vegetation dominates the fire regime regardless of its abundance and, that the pyrophobic vegetation?s flammability is in consequence overestimated.