Modelling carbon dynamics and response to environmental change along a boreal fen nutrient gradient

As the global climate changes so will the balance of carbon dioxide (CO2) sequestration and release from peatlands. The changes occurring within peatlands will be complex, making their future carbon (C) responses difficult to predict. We studied three boreal fen peatlands along a nutrient gradient in northern Ontario, Canada. Specifically, we investigated how abiotic and biotic responses of the fens interacted with seasonal and annual hydrologic variability. We further assessed the role that a local disturbance, such as inundation due to beaver dams, has in confounding fen C dynamics. We conducted these assessments by applying the process-based model Wetland-DNDC, which we parameterized, calibrated and corroborated with field data, to examine how C dynamics were affected by fen type (rich, intermediate and poor), historical and projected climate and local flooding disturbance. We found inter-annual variability in greenhouse gas (CO2 and methane, CH4) dynamics to be large among fens, with larger coefficients of variation for heterotrophic respiration and CH4 effluxes (a range of 40–85% and 31–58%, respectively) than gross primary production (a range of 9–26%). This annual variability in C effluxes correlated to meteorological conditions and water table, although the specific drivers differed among fens. Scenarios of prolonged flooding and drought had opposite effects on C effluxes. Drought increased CO2 efflux by 43%, 175% and 65% for rich, intermediate and poor fens, respectively, and decreased CH4 efflux by 11% and 14% for the intermediate and poor, but increased CH4 efflux by 19% for the rich fen. In contrast, shallow flooding decreased CO2 by 40%, 72% and 57% in the rich, intermediate and poor fen, and increased CH4 effluxes by 87%, 202% and 269%, respectively. Flooding by beaver dams during a prolonged drought moderately decreased CO2 (29%, 66% and 53%) but strongly increased CH4 efflux by 121%, 261% and 333%, respectively, than the flooding only. We conclude that the interacting controls on greenhouse gas dynamics differ among fens types, contributing to the magnitude of effluxes and their response to changing conditions. We propose that effects of water table level, peat nutrient status, and vegetation composition, and their interactions, on C processes require enhanced understanding to define natural ranges of spatial and temporal variability. Only then will it be possible to adequately predict future peatland C responses to climate and other environmental changes. As such, short-term experimental results or field studies should not be used to predict future peatland C dynamics without an understanding of the natural range of variation in the environmental drivers and the response of C processes to those drivers.