Controls on zircon age distributions in volcanic, porphyry and plutonic rocks

Abstract. The distribution of zircon crystallisation ages in igneous rocks has been proposed to provide insights into the dynamics of underlying magma reservoirs. However, the ability to interpret magmatic processes from an age distribution is challenged by a complex interplay of factors such as sampling biases, analytical uncertainties and incorporation of extraneous zircon grains. Here, we use a compilation of igneous zircon U–Pb ages measured by chemical abrasion isotope dilution thermal ionisation mass spectrometry (CA-ID-TIMS) to quantify the differences that exist among zircon U–Pb age distributions from different magmatic systems. The compiled dataset was rigorously filtered through a number of processing steps to isolate age distributions least impacted by sampling biases and analytical factors. We also filter the database using a new algorithm to systematically identify and remove old outliers from age distributions. We adopt the Wasserstein distance as a dissimilarity metric to quantify the difference between the shapes of age distributions. Principal component analysis (PCA) of a dissimilarity matrix of pairwise Wasserstein distances between age distributions reveals differences among zircon age distributions found in plutonic, porphyry and volcanic rocks. Volcanic and porphyry zircon populations exhibit a skew towards younger ages in their distributions, whereas plutonic age distributions skew towards older ages. We use a bootstrap forward modelling approach to generate synthetic zircon age distributions, which are cast into the PCA space of the dissimilarity matrix of natural age distributions to allow us to identify the magmatic processes which reproduce distributions found in natural data. We find that the younger skew of porphyry and volcanic zircon age distributions can be reproduced under bootstrap sampling scenarios where zircon crystallisation is truncated (e.g. by volcanic eruption or porphyry dyke emplacement). We also find that sampling underlying zircon age distributions generated under higher magmatic flux can contribute to the younger skew of volcanic and porphyry zircon age distributions, though we emphasise that no difference in flux is required due to the strong effect of truncation. Given the multitude of factors that influence observed zircon age distributions, we urge caution when quantifying the thermal evolution of crustal magma bodies using zircon age distributions integrated with numerical models.