Evaluating possible sources of error in tree-ring ¹⁴C data using multiple trees across South America

Abstract A limitation in fine-tuned tree-ring radiocarbon ( 14 C) data is normally associated with overall data uncertainty. Tree-ring 14 C data variance as a result of sample heterogeneity can be reduced by adopting best practices at the time of sample collection and subsequent preparation and analysis. Variance-reduction of 14 C data was achieved by meticulous sample handling during increment core or cross-sectional cuttings, in-laboratory wood reductions, and cellulose fiber homogenization of whole rings. To demonstrate the performance of those procedures to final 14 C results, we took advantage of the replicated data from assigned calendar years of two Pantropical post-1950 AD tree-ring 14 C reconstructions. Two Cedrela fissilis Vell. trees spaced 22.5 km apart, and two trees of this species together with one Peltogyne paniculata Benth tree spaced 0.2 to 5 km apart were sampled in a tropical dry and moist forest, respectively. Replicate 14 C data were then obtained from grouped tree-ring samples from each site. A total of 88% of the replicated 14 C results fell into a remarkably consistent precision/accuracy range of 0.3% or less, even though multiple tree species were used as pairs/sets. This finding illustrates how adopting a few simple strategies, in tandem with already established chemical extraction procedures and high-precision 14 C analysis, can improve 14 C data results of tropical trees.