Abrupt State Change in Spatially-Patterned Subalpine Forests in Northern Colorado During the Medieval Climate Anomaly

Spatial patterns in many ecosystems arise from feedbacks associated with the potential for critical transitions and multiple stable states. Such systems may be susceptible to abrupt change, which could be indicated by early-warning signals, such as critical slowing down (increasingly long recovery from perturbation as a threshold approaches). Paleoecological data from ribbon forests, a type of subalpine parkland found in the Rocky Mountains, offer an opportunity to test these hypotheses. The forests consist of alternating strips of forest and meadow that form because bands of Picea and Abies trees act as snow fences with large snowdrifts forming on their lee sides. Drifts provide moisture for the adjacent trees, but also increase seedling mortality and shorten the growing season where drifts accumulate. The feedbacks between forest growth and snow accumulation maintain the ribbon forest-meadow pattern, and raise the potential for abrupt change if the feedbacks breakdown in response to factors like drought or fire.

Our fossil pollen data from Summit Lake, located on the Continental Divide in the Park Range, northern Colorado, indicate that a closed forest transitioned rapidly to a ribbon forest state at ca. 1000 BP. Artemisia pollen increased (20 to 35%) and Picea and Abies pollen decreased (25 to 15%) within a century or less after a pair of charcoal peaks. Decreased charcoal influx (from 0.6 to 0.4 pieces/cm²/yr) and fire frequency (from 4.5 to 1.5 fires/ka) coincided with the pollen assemblage changes, and is consistent with decreased landscape biomass and fuel connectivity.

Initial analyses show evidence of critical slowing down before the state change. After eight of eleven fires recorded by peaks in charcoal accumulation, Artemisia pollen percentages rise to a peak consistent with brief opening of the initially forested landscape. After 2000 BP, the magnitude and duration of the post-fire changes increases until no recovery is recorded after the shift at 1000 BP. The data provide the first evidence of leading indicators before state change in long-lived systems, such as forests, and suggests that current theory about patterning and multiple stable states may be accurate for such systems.