Beyond annual budgets: carbon flux at different temporal scales in fire-prone Siberian Scots pine forests

Along four chronosequences of fire-prone Siberian Scots pine forests we compared net primary production (NPP) and two different mass-based estimates of net ecosystem productivity (NEPC and NEPS). NEPC quantifies changes in carbon pools along the chronosequences, whereas NEPS estimates the short-term stand-level carbon balance in intervals between fires. The chronosequences differed in the mean return interval of surface fires (unburned or moderately burned, 40 yr; heavily burned, 25 yr) and site quality (lichen versus Vaccinium type). Of the Vaccinium type (higher site quality) only a moderately burned chronosequence was studied. NEPC was derived from the rate of changes of two major carbon pools along the chronosequence time axes: (1) decomposition of old coarse woody debris (CWD) left from the previous generation after stand-replacing fire, and (2) accumulation of new carbon in biomass, CWD and soil organic layer by the regenerating stand. Young stands of all chronosequences were losing carbon at rates of -4 to -19 mol C m(-2) yr(-1) (-48 to -228 g C m(-2) yr(-1)). Depending on initial CWD pools and site-specific accumulation rates the stands became net carbon sinks after 12 yr (Vaccinium type) to 24 yr (lichen type) following the stand-replacing fire, and offset initial carbon losses after 27 and 70 yr, respectively. Highest NEPC was reached in the unburned chronosequence (10.8 mol C m(-2) yr(-1) or 136 g C m(-2) yr(-1)). Maximum NEPC in the burned chronosequences ranged from 1.8 to 5.1 mol C m(-2) yr(-1) (22 to 61 g C m(-2) yr(-1)) depending on site quality and fire regime. Around a stand age of 200 yrNEP(C) was 1.6 +/- 0.6 mol C m(-2) yr(-1) (19 +/- 7 g C m(-2) yr(-1)) across all chronosequences. NEPS represents the current stand-level carbon accumulation in intervals between recurring surface fires and can be viewed as a mass-based analogue of net ecosystem exchange measured with flux towers. It was estimated based on measurements of current woody NPP, modelled decomposition of measured CWD pools and organic layer accumulation as a function of time since the last surface fire, but ignores carbon dynamics in the mineral soil. In burned mature lichen type stands, NEPS was 6.2 +/- 2.6 mol C m(-2) yr(-1) (74 +/- 31 g C m(-2) yr(-1)) and thus five times higher than NEPC at the respective age (1.2 +/- 0.6 mol C m(-2) yr(-1) or 14 +/-7 g C m(-2) yr(-1)). Comparing NEPS and NEPC of mature stands, we estimate that 48% of NPP are consumed by heterotrophic respiration and additional 35% are consumed by recurrent surface fires. As expected, in unburned stands NEPC and NEPS were of similar magnitude. Exploring a site specific model of CWD production and decomposition we estimated that fire reduces the carbon pool of newly produced CWD by 70%. Direct observation revealed that surface fire events consume 50% of the soil organic layer carbon pool (excluding CWD). We conclude that surface fires strongly reduced NEPC. In ecosystems with frequent fire events direct flux measurements using eddy covariance are likely to record high rates of carbon uptake, since they describe the behaviour of ecosystems recovering from fire without capturing the sporadic but substantial fire- related carbon losses