Differential zircon fertility of source terranes and natural bias in the detrital zircon record: Implications for sedimentary provenance analysis

Granitic intrusions emplaced within Laurentia during the Grenville Orogeny (1.15-1.05 Ga) are exceptionally Zr-rich, primarily a function of high modal zircon, relative to Paleozoic granitoids emplaced within eastern Laurentia during phases of Appalachian orogenesis. Erosion of Grenville source rocks would generate disproportionately large volumes of detrital zircon compared to less zircon-fertile source regions. The latter sources are difficult to detect by standard in situ U-Pb dating methods of detrital zircon (SHRIMP or LA-ICP-MS U-Pb analysis of centers of > 100 mu m grains), or could be overlooked during sampling. Grenvillian zircon fertility biased the Neoproterozoic to Recent sedimentary record as a result of two factors: (1) zircon durability and insolubility led to recycling during repeated orogenesis; (2) inertness of zircon below the onset of anatexis means dominantly metamorphosed sedimentary terranes failed to generate significant new zircon corresponding in age to the time of accretion of those terranes to Laurentia. Zircon growth under anatectic conditions generates new zircon, commonly as overgrowths on preexisting zircon, which may be too narrow to easily analyze by laser or ion beam techniques. Accretion of metasedimentary terranes therefore could be rendered all but undetectable via standard U-Pb detrital zircon dating. Grenville age modes dominate detrital zircon age distributions for Laurentian Neoproterozoic rift basins, Appalachian Paleozoic synorogenic elastic sequences, Appalachian metasedimentary terranes, and modem rivers draining these terranes. This is an artifact of the Grenville zircon signal that echoed throughout all Paleozoic orogenies on the eastern Laurentian margin. The natural Grenville bias in the detrital zircon record is further amplified by standard sampling biases (such that large (e.g. 100-300 mu m) crystals are preferentially chosen) and analytical biases (e.g. cores are much more frequently analyzed). (c) 2006 Elsevier B.V. All rights reserved.