A 7.5 ka chronology of stable carbon isotopes from tree rings with implications for their use in palaeo-cloud reconstruction

Tree-ring stable isotope chronologies provide very high-resolution palaeoclimatic data, and the number of records is increasing rapidly worldwide. To extend the chronologies back in time, before the period covered by the old living trees, the use of subfossil wood samples is required. Typically, the longest continuous subfossil chronologies consist of regionally collected tree-ring materials, rather than tree rings from a single site, and are likely more sensitive to data heterogeneity. Yet, the characteristics of such datasets remain hitherto unexplored. Here we produce a continuous, decadally resolved chronology of 13C/12C ratio (δ13C) from Finnish Lapland over the past 7.5 ka (5500 BCE to 2010 CE) for which there is replication of at least five Pinus sylvestris trees. Less negative δ13C values were observed as trees age and for western sites (higher in elevation and further from the cold oceanic air flow). The age-related trends in living tree δ13C data were expressed mainly over the first fifty years mimicking the “juvenile effect” whereas the subfossil data showed trend over the trees' lifespan. These findings demonstrated the need to detrend the individual δ13C series before averaging them into the mean chronology. The corresponding biases were removed from the isotopic data using the methods frequently applied for tree-ring width and density proxies, the Regional Curve Standardization (RCS) combined with signal-free approach. While the RCS procedures commonly preserve the long-term variations in the resulting chronology, not all types of them did so as demonstrated for chronologies produced using separate RCS models for the δ13C series with relatively high and low isotopic level (offset from the grand mean). It was shown that these δ13C levels (i.e. the relative isotopic enrichment) result both from the low-frequency climate signal and biogeographical aspects (the site longitude/altitudes). The non-climatic biases were removed from the δ13C series by using separate RCS models for the subsets (western and eastern) of isotopic series. Similar to previous investigations using annually resolved δ13C data from Lapland, our chronology had strong negative correlations to variations in cloud cover. Here, a bootstrapping experiment was used to verify this dendroclimatic association. The resulting palaeo-cloud reconstruction portrayed decadal to multi-millennial variations with centennial anomalies coinciding with the mid and late Holocene events of climate transitions, highlighting the value of subfossil isotope chronologies from tree rings in synthesising climate dynamics from several proxy sources over the present interglacial.