Pacific plate reconstructions and uncertainties

Abstract The Pacific plate is positioned into the hotspot reference frame, first directly relative to the Pacific hotspots, and then indirectly relative to the African hotspots by using seafloor spreading data. The reconstructions are repeated for 10, 15, 37, 48, and 61 Ma. The errors associated with each link for both approaches are expressed by covariance matrices which can be combined to yield the overall error in the case of the relative plate motion circuit. Assuming the hotspots are fixed, and that the reconstructions record the true relative positions of the plates, the reconstructions should be equivalent within their uncertainties. In fact, however, there is agreement within the uncertainties only for 10 Ma, with the misfit increasing with age for the earlier reconstructions. These misfits can he accounted for by three alternative explanations: large hotspot location uncertainties, a shift in time of the Pacific group of hotspots relative to the African group, or an additional plate boundary in the relative plate motion circuit, which we assume to be in Antarctica. By examining each of these possibilities in turn, we conclude that the uncertainties in individual hotspot locations required to bring the reconstructions into agreement are unreasonably large, and that the data available at present do not allow us to distinguish which of the remaining two possibilities may be the primary source of reconstruction errors.