Climate fails to predict wood decomposition at regional scales

Decomposition of organic matter strongly influences ecosystem carbon storage(1). In Earth-system models, climate is a predominant control on the decomposition rates of organic matter(2-5). This assumption is based on the mean response of decomposition to climate, yet there is a growing appreciation in other areas of global change science that projections based on mean responses can be irrelevant and misleading(6,7). We test whether climate controls on the decomposition rate of dead wood-a carbon stock estimated to represent 73 +/- 6 Pg carbon globally(8)-are sensitive to the spatial scale from which they are inferred. We show that the common assumption that climate is a predominant control on decomposition is supported only when local-scale variation is aggregated into mean values. Disaggregated data instead reveal that local-scale factors explain 73% of the variation in wood decomposition, and climate only 28%. Further, the temperature sensitivity of decomposition estimated from local versus mean analyses is 1.3-times greater. Fundamental issues with mean correlations were highlighted decades ago(9,10), yet mean climate-decomposition relationships are used to generate simulations that inform management and adaptation under environmental change. Our results suggest that to predict accurately how decomposition will respond to climate change, models must account for local-scale factors that control regional dynamics.