Extended Data Fig. 10: Sensitivity of our plate model to other possible slab lengths and their implied sinking rates.

a, Our plate model predictions are compared with a spatial density plot of total recorded Andean magmatic ages by latitude from a catalogue of undifferentiated Andean magmatic ages⁷¹ (14,709 total published magmatic ages since 140 Myr ago). The bin size is 4° in latitude and 5 Myr in age. This plot is designed to offer an alternative perspective to Fig. 5a. Differences between this plot and Fig. 5a result from our choice of including in Fig. 5a only magmatism that had an interpreted tectonic origin, for example, arc-related, backarc or intraplate magmatism. For Fig. 5a, published geochemical signatures provide a feature-by-feature comparison for each magmatic record that is absent from this plot. The near-white band 32° S and 45° S is not necessarily a magmatic gap, but is the result of limited reported ages, possibly owing to limited outcrop exposure. In Fig. 5a, we used detrital zircon ages to infer the magmatic activity for 32°–45° S. The thick red line shows our predicted southward propagation of Nazca subduction initiation (same as in Fig. 5a). The error range (±10 Myr) shown by transparent red lines come from alternative choices of slab-edge velocity perturbations (see Methods and Extended Data Fig. 3 for details). Overall, our model prediction fits well with a magmatic minimum around 80 Myr ago followed by increased magmatism, which we interpret to correspond to the initiation of the most recent phase of Nazca subduction. Arbitrarily decreased unfolded slab lengths of 80% (light blue dashed line) and 60% (dark blue dashed line) are also shown. The arbitrarily decreased slab lengths show the sensitivity of our plate model predictions to possible tomographic blurring, which would have exaggerated our measured Nazca slab areas. Comparison to the shorter unfolded slab lengths shows that southward propagation of Nazca subduction is preserved when Nazca slab lengths are decreased, but we argue that the alternative subduction initiation timings compare less well to the increased magmatism between 80 Myr and 60 Myr ago north of 32° S. Green dots show that the choice of a thinner initial lithosphere thickness for slab unfolding (80 km rather than 100 km) would increase slab lengths, which would effectively offset some tomographic blurring. b, Comparison between published lower-mantle slab sinking rates^{2,21,25,26,47,53,90,91} and the implied lower-mantle sinking rates for our preferred slab lengths (red line) and arbitrarily decreased 80% (light blue dashed line) and 60% (dark blue dashed line) slab lengths from a. Shorter slab lengths (less than 60%) were excluded from a on the basis of unreasonably fast slab sinking rates.