No rapid soil carbon loss after a windthrow event in the High Tatra

Windthrows are among the most important disturbances of forest ecosystems in Europe, with expected increasing frequency due to climate change. However, surprisingly little is known about soil carbon dynamics after windthrow mainly due to missing field assessments. After a large windthrow event in the High Tatra Mountains in 2004 three soil monitoring plots were established, one at a non-harvested windthrow left for natural succession, one at a harvested windthrow and one at a reference forest site which remained unaffected by the storm event. No loss in soil organic carbon (SOC) stocks was detected at the two windthrow sites with three inventories over the 3.5 years after the storm event. However, shifts within the organic layers and the mineral soil toward more decomposed organic matter were found. Increasing C/N ratios at the harvested windthrow site indicate that newly established herbaceous vegetation compensated the decline in tree litter input. At the non-harvested windthrow site a flush of needle litter from broken trees helped to sustain SOC stocks. In contrast, SOC stocks at the reference forest increased by 2.2 Mg ha−1 year−1 with major SOC stock accumulation in the forest floor. An assessment of the sample size required to detect future SOC changes revealed that at the windthrow sites a similar sample size is required as in the undisturbed reference forest. Small scale heterogeneity was at such a level that paired sampling did not significantly reduce the number of required samples. However, the separation of forest floor layers and mineral soil was a major obstacle for efficient forest soil carbon monitoring. The required number of soil samples could be decreased by 45% with a simultaneous sampling of forest floor and upper mineral soil, leading to more reliable SOC inventories.

► Repeated soil carbon inventories after windthrow with two post-disturbance treatments.
► Forest floor carbon at the Tatra sites was less vulnerable than previously assumed.
► The sampling design was strong with more than 30 sample points per site.
► Reliable monitoring of forest floor C requires also sampling of the upper mineral soil.