Detrital zircons as tracers of sedimentary provenance: limiting conditions from statistics and numerical simulation

U-Pb ages of detrital zircons in clastic sediments are potential indicators of sedimentary provenance and crustal evolution. To extract geologically significant information from such data, it is necessary that the zircons analysed in the laboratory reflect the sediment from which they were separated. Among other factors, this depends on the number of grains analysed. The probability that a detrital zircon in a sediment belongs to one of several populations (defined by their age or another well-defined parameter) is described by the binomial probability formula, from which detection limits and expected values for the accuracy and precision of population sizes can be derived. Monte Carlo simulation of sets of data with up to 120 individual analyses drawn from pools with known age probability density distributions shows that quantitative representation of the age distribution of the zircons in the host sediment is unlikely to be achieved in natural cases. Methods that assume such representation (e.g., age spectrum deconvolution, multivariate statistical analysis using probability density scores, and mass balance modelling) are therefore likely to give results of questionable significance. The best way to extract the potential information contained in detrital zircons may be a combination of random and nonrandom selection of grains for analysis; yielding two complementary sets of age data from each sample, the random fraction should comprise 35-70 grains or more, depending on the complexity of the age distribution. The use of trace element and Lu-Hf isotope data from the zircons dated by U-Pb will give additional, valuable information, which may facilitate interpretation of detrital zircon ages. The use of detrital zircon ages to limit the age of deposition of the host sediment is fraught with both geological and statistical sources of error, and is therefore discouraged.