A statistical analysis of the correlation between large igneous provinces and lower mantle seismic structure

Large igneous provinces (LIPs) lie approximately above the margins of the African and Pacific large low shear velocity provinces (LLSVPs) in the deep mantle. This spatial correlation has been used to argue that plumes are preferentially generated at the margins of LLSVPs. We perform a series of Monte Carlo-based statistical tests to assess the uniqueness of this conclusion. These tests indicate that (1) the reconstructed locations of LIPs are significantly correlated with both slower-than-average shear wave velocity regions, which contain LLSVPs, and the margins of these structures; and (2) these correlations cannot be statistically distinguished. That is, given current constraints, if plumes were generated randomly throughout regions of slower-than-average shear wave velocity in the deep mantle, then statistical tests are expected to show a significant correlation between the locations of LIPs and the margins of LLSVPs. We therefore conclude that it is premature to argue that the margins of LLSVPs represent preferred zones of plume generation. This conclusion is reinforced in our analysis by a demonstration that the expected mean distance of a set of points randomly placed in slower-than-average shear wave velocity regions is consistent with the observed mean distance between LIPs and the margins of LLSVPs. Finally, we also test the correlation between the reconstructed locations of LIPs and the horizontal gradient in deep mantle shear velocity perturbations. We find, given the uncertainty implied by different tomography models, that there is no statistically significant correlation and that being in a slow region (i.e. in the region of LLSVPs) is a stronger geographic requirement for plume generation than being at a specific (high) gradient.