Matthew G. Jackson

The formerly continuous Vitiaz Arc broke into its Vanuatu and Fijian portions during a reversal of subduction polarity in the Miocene. Basaltic volcanism in Fiji that accompanied the breakup ranged from shoshonitic to low‐K and boninitic with increasing distance from the broken edge of the arc that, presumably, marks the broken edge of the slab. The Sr‐Pb‐Nd isotope ratios of the slab‐derived component in the breakup basalts most closely match those of the isotopically most depleted part of the Samoan seamount chain on the Pacific Plate that was adjacent to the site of breakup at 4–8 Ma, and differ from those of subsequent basalts in spreading segments of the surrounding backarc North Fiji and Lau Basins. Subduction of the Samoan Chain along the Vitiaz Trench Lineament may have controlled the limit of polarity reversal and, hence, where the double saloon doors (Martin, 2013) opened. Prior to breakup, Fijian volcanics were more similar isotopically to the Louisville Seamount Chain.

Oceanic hotspots with extreme enriched mantle radiogenic isotopic signatures—including low 143 Nd/ 144 Nd indicative of subducted continental crust—are linked to plume conduits sampling the southern hemispheric mantle. However, the mechanisms responsible for concentrating subducted continental crust in the austral mantle are unknown. We show that subduction of sediments and subduction eroded material, and lower continental crust delamination, cannot generate this spatially coherent austral geochemical domain. However, continental collisions—associated with the assembly of Gondwana‐Pangea—were positioned predominantly in the southern hemisphere during the late Neoproterozoic appearance of widespread continental ultra‐high‐pressure metamorphic terranes, which marked the onset of deep subduction of upper continental crust. We propose that deep subduction of upper continental crust at ancient rifted‐passive margins during ca. 650‐300 Ma austral supercontinent assembly resulted in enhanced upper continental crust delivery into the southern hemisphere mantle. Similarly enriched mantle domains are absent in the boreal mantle plume source, for two reasons. First, continental crust subducted after 300 Ma—when the continents drifted into the northern hemisphere—has had insufficient time to return to the surface in plumes sampling the northern hemisphere mantle. Second, before the first known appearance of continental ultra‐high‐pressure rocks at 650 Ma, deep subduction of upper continental crust was uncommon, limiting its subduction into the northern (and southern) hemisphere mantle earlier in Earth history. Our model implies a recent formation of the austral enriched mantle domain, explains the geochemical dichotomy between austral and boreal plume sources, and may explain why there are twice as many austral hotspots as boreal hotspots.