Peat oxic and anoxic controls of <i>Sphagnum</i> decomposition rates in the Holocene Peatland Model decomposition module estimated from litterbag data

Abstract

The Holocene Peatland Model (HPM) is a widely applied model for understanding and predicting long-term peat accumulation, but it is difficult to test due to its complexity, measurement errors, and lack of data. Instead of testing the complete model, tests of individual modules may avoid some of these problems. In particular, the HPM decomposition module can be tested with litterbag data, but no such test has been conducted yet. Here, we estimate parameter values of the HPM decomposition module from available Sphagnum litterbag experiments included in the Peatland Decomposition Database and with a litterbag decomposition model that considers initial leaching losses. Using either these estimates or the standard parameter values, we test whether the HPM decomposition module fits decomposition rates (k0) in Sphagnum litterbag experiments along a gradient from oxic to anoxic conditions. Both litterbag data and model versions where HPM decomposition module parameters were estimated suggest a less steep gradient of decomposition rates from oxic to anoxic conditions and larger anaerobic decomposition rates for several species than the standard parameter values. This discrepancy may be caused by ignoring the effects of water table fluctuations on aerobic and anaerobic decomposition rates. Moreover, our analysis suggests that maximum possible decomposition rates of individual species (k0,i) vary more than suggested by the standard parameter values of the HPM plant functional types. Based on previous sensitivity analyses of the HPM, the estimated differences from the standard parameter values can cause differences in predicted 5000-year carbon (C) accumulation up to 100 kg m−2. The HPM decomposition module with standard parameter values fits k0 estimated from Sphagnum litterbag data, but model versions where HPM decomposition module parameters were estimated and differ significantly have an equivalent fit. The reason why models with different parameter values have equivalent fit is that errors in remaining masses and the design of available litterbag experiments support a range of initial leaching loss and k0 estimates. Consequently, applications of the HPM and any other peatland model should consider that a broad range of decomposition module parameter values is compatible with available litterbag experiments. Improved litterbag experiments are needed for more accurate tests of any peatland decomposition module and for obtaining parameter estimates accurate enough to allow even only approximate predictions of long-term peat accumulation. The modeling approach used here can be combined with different data sources (for example, measured degree of saturation) and decomposition modules. In light of the large differences in long-term peat accumulation suggested by the parameter estimates, we conclude that it is worth conducting such experiments, not only to improve the decomposition module of the HPM, but also to improve peatland models in general.