Testing the hypothesis of post-volcanic missing rings in temperature sensitive dendrochronological data

The precise, annual dating control, inherent to dendrochronology, has recently been questioned through a combined analysis of tree-growth and coupled climate models (Mann et al. (2012; hereafter MAN12)) suggesting single tree-rings in temperature limited environments are missing following large volcanic events. We test this hypothesis of missing, post-volcanic rings by using a compilation of maximum latewood density (MXD) records that are typically used for reconstructing temperature and the detection of volcanic events, together with a unique set of long instrumental station data from Europe reaching back into the early 18th century. We investigate the temporal coherence between tree-ring MXD and observed summer temperatures before and after the most significant, precisely dated, volcanic event of the past 1000 years, the 1815 Tambora eruption widely known as the cause for the 1816 “year without a summer”. Comparison of existing and newly developed MXD chronologies from cold environments in Northern Scandinavia (r¯North=0.70, N = 3) and the European Alps, including the Pyrenees, (r¯Central=0.46, N = 4) reveals significant interseries correlations over the 1722–1976 common period, suggesting coherence among these independently developed timeseries. Comparisons of these data with observed JJA temperatures – from 1722 to 1976, a 94-year pre-Tambora (1722–1815), and a 94-year post-Tambora (1817–1910) period – reveals significant and temporally stable correlations ranging from 0.32 to 0.68. However, if we assume the 1816 ring is missing in the MXD chronologies (i.e., shift the pre-Tambora data by one year), all proxy/instrumental correlations fall apart approaching zero. Results from an additional experiment, where the long instrumental record is replaced by an annually resolved, 500-year, summer temperature reconstruction derived from documentary evidence, corroborates the findings from the first experiment: significant positive correlations with the unmolested chronologies and zero correlation with the perturbed chronologies back to 1500 AD. These elementary analyses indicate that either the tree-ring chronologies are correctly dated, i.e., no is ring missing in the year without a summer, or that both the long instrumental and documentary records contain dating uncertainties. As the latter is unlikely, we conclude the MAN12 hypothesis on post-volcanic missing rings can be rejected based on simple comparisons of tree-ring, instrumental and documentary data over the past 300–500 years from Central and Northern Europe.