Summer litter decomposition is moderated by scale-dependent microenvironmental variation in tundra ecosystems

Tundra soils are one of the world's largest organic carbon stores, yet this carbon is vulnerable to accelerated decomposition as climate warming progresses. The landscape-scale controls of litter decomposition are poorly understood in tundra ecosystems, which hinders our understanding of the global carbon cycle. We examined the extent to which the thermal sum of surface air temperature, soil moisture and permafrost thaw depth influenced litter mass loss and decomposition rates (k), and at which spatial thresholds an environmental variable becomes a reliable predictor of decomposition, using the Tea Bag Index protocol across a heterogeneous tundra landscape on Qikiqtaruk - Herschel Island, Yukon, Canada. We found greater green tea litter mass loss and faster decomposition rates (k) in wetter areas within the landscape, and to a lesser extent in areas with deeper permafrost active layer thickness and higher surface thermal sums. We also found higher decomposition rates (k) on north-facing relative to south-facing aspects at microsites that were wetter rather than warmer. Spatially heterogeneous belowground conditions (soil moisture and active layer depth) explained variation in decomposition metrics at local scales (< 50 m2) better than thermal sum. Surprisingly, there was no strong control of elevation or slope on litter decomposition. Our results reveal that there is considerable scale dependency in the environmental controls of tundra litter decomposition, with moisture playing a greater role than the thermal sum at < 50 m2 scales. Our findings highlight the importance and complexity of microenvironmental controls on litter decomposition in estimates of carbon cycling in a rapidly warming tundra biome.We used the 'brms' package (Bürkner, 2017) and weakly informative priors (half Student-t priors with 3 degrees of freedom) for all models, with two chains of 8000 iterations each and a warmup value of 2000. We conducted all analyses in R version 3.6.3. The code and data used for this study can be found at the following repository: https://github.com/ShrubHub/MicroTeaHub/ Large data: https://zenodo.org/record/6411321Funding provided by: Natural Environment Research CouncilCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100000270Award Number