Self-organizing patterns of peat decomposition in mires and implications for greenhouse gas emissions

In many peatlands occurring in both permafrost and temperate terrain, vegetation develops in the form of spatially heterogeneous patterns or pools. Because GHG fluxes and their responses to allogenic forcing from different elements of mire patterns (eg., pools and raised peat strings) can vary, quantifying the role of peatlands in climate change may require using models for the dynamics of patterns. To this end, we investigate patterns of consistently-spaced ≈ 2 m diameter pools that occur in permafrost mires across a 4500 y chronosequence at Espenberg, NW Alaska. Pools first occur in small groups (<5) on 1200 y old mires where peat depth is ≈ 1 m, and pool patterns increase in size thereafter. Dissolved oxygen in pool water and peat stratigraphy in pool walls and floors are consistent with formation by peat decomposition amplified by thaw-derived subsidence of ice-rich organic soil. Natural patterns are reproduced by a model in which new pools principally initiate in ≈ 3 m wide annuli around existing pools, as evaluated using Kolmogorov-Smirnov tests of pool spacing distributions. Models encapsulating spatially random mechanisms for pool formation cannot reproduce natural patterns. Pool replication occurs by suppression of peat accumulation adjacent to 0.2--1.0 m high frost-heaved peat rings by drifting snow that persists into the growing season. Because pool initiation depends on distribution of earlier pools, rates and patterns of peat decomposition in the mire may nonlinearly depend upon, and have intrinsic time-scales exceeding, external influences on decomposition such as climate. Variations in peat accumulation or decomposition, and hence carbon accumulation or release, are localized in pools rather than occurring as a bulk change across a peatland surface. The influence of emergent pattern dynamics in this environment, and others, such as the patterned peatlands of Cape Breton, Nova Scotia, will be discussed.