Abstract Supercontinent amalgamation is described by the end-member kinematic processes of introversion – closure of interior oceans; extroversion – closure of exterior oceans; or orthoversion – amalgamation 90° from the centroid of the previous supercontinent. However, supercontinent formations are often ascribed to contradictory mechanisms; for example, Pangaea has been argued to have formed by introversion from Pannotia/Gondwana, and extroversion from Rodinia. Conflicting interpretations arise partly from attempting to define oceans as interior or exterior based on palaeogeography or the age of the oceanic lithosphere relative to the time of supercontinent breakup. We define interior and exterior oceans relative to the external subduction ring, and associated accretionary orogens that surround amalgamated supercontinents. All oceans within the continental dominated cell and internal to the subduction ring are interior oceans. The exterior ocean is separated from the interior oceans by the subduction ring and bordered by external accretionary orogens. Wilson cycle tectonics dominate the interior continental cell, conversely, subduction of the exterior ocean is doubly vergent and lacks continent–continent collision. For the exterior ocean to close, the subduction ring must collapse upon itself, leading to the collision of external accretionary orogens. Employing this definition, Rodinia formed by extroversion, but all other supercontinents formed by introversion.

Abstract Continental arcs are key sites of granitic magmatism, yet details of the origins of these magmas, including the role and contribution of mafic magma, the timing and location of initial zircon formation and how zircon isotopic signatures relate to granite formation, remain as challenges. Here we use U–Pb dating, trace elements and Hf isotopic systematics of zircon in mafic microgranular enclaves (MMEs), from the convergent plate margin Satkatbong diorite (SKD) in Korea to understand lower arc magmatism and zircon production. The host granitic body and MMEs display similar major element evolutionary trends and similar ranges of Sr, Nd and Hf isotopes, implying a cognatic relationship. Zircons show a large variability in ε Hf (t) (c. 6 units) and age (>30 Ma). We propose that the SKD and MMEs originated from the same, long-lasting, lower crustal mush reservoir, enabling long and variable residence times for zircons. Prolonged zircon ages, combined with the Hf isotope variability within a single pluton (SKD and its MME), indicate that not all zircons were instantaneously crystallized in a rapidly cooling shallow magma chamber but were continuously formed in a long-lasting hot source. A low-melt-fraction mush type reservoir in a deep crustal hot zone provides a viable model for the source setting. Continuous replenishment of mafic magmas acts as the main re-activator of the reservoir, and provide a critical role in spawning zircons that record a long age span, because (1) the magma adds Zr into the reservoir, enabling it to reach zircon saturation and (2) the generated zircon grains are transported upward as antecrysts by flow inside of the reservoir. This means that antecrysts with different ages may mix with each other in the ascending magma body. The significance of this model is that a conclusive time of intrusion cannot be constrained by such zircon ages, as these antecrysts constitute inherited grains.