Annual gaseous carbon budgets of forest-to-bog restoration sites are strongly determined by vegetation composition.

Large areas of naturally open peatland in western Europe were drained and planted with non-native conifers in the twentieth century. Efforts are currently underway to restore many of these sites. Ultimately, forest-to-bog restoration aims to bring back functional peatlands that can sequester carbon but there is a lack of empirical evidence for whether current approaches are effective. Using a chronosequence design, we compared the annual gaseous carbon balance of two forest-to-bog restoration areas with an open area not subject to afforestation. A closed chamber method was used to determine gas fluxes (Net Ecosystem Respiration, Gross Primary Productivity, Net Ecosystem Exchange (NEE) and methane (CH4)) over a twelve-month period for locations spanning the range of peatland microtopography and vegetation communities. Relationships between gas fluxes, vegetation/cover and environmental factors were analysed and regression models used to estimate annual CO2 and CH4 budgets. During the study period, NEE estimates (total gaseous C expressed as CO2-eq) showed a net sink for the unafforested (-102 g C m-2 yr-1) and oldest (-131 g C m-2 yr-1) restoration area (17 years post-restoration 'RES 17 YRS'), whilst the youngest restoration area (6 years post-restoration 'RES 6YRS'), was a net source (35 g C m-2 yr-1). We observed significantly higher CH4 emissions from restoration areas dominated by Eriophorum angustifolium compared with other peatland vegetation types. Sampling points with higher cover of Sphagnum were found to be most effective for C sequestration. Overall, vegetation composition/cover was observed to be an important factor determining C emissions from forest-to-bog restoration areas. These results suggest that restoration is effective in returning the carbon sink function of peatlands damaged by commercial forestry and - depending on restoration techniques - timescales of >10 years may be required.