Potentially only three examples (Easter-eastern Mid Pacific Mountains, Louisville-Ontong Java, Marquesas-Hess/Shatsky) possess the plateau-island chain sequence expected from the plume head-tail model [Clouard & Bonneville, 2001]. Non-linear age progressions are found in many island chains, including the Cook-Austral-Marquesas, Marshall-Gilberts, and Line Islands. Even the Hawaiian-Emperor chain, generally considered the archetypical example of plume volcanism, does not have the features expected in the plume model. The volcanism lacks an associated plateau, eruption rates have increased rather than decreased over time, and paleomagnetic evidence [Sager, 2002] indicates any hotspot could only have been stationary for half its history. Consideration of other models is therefore long overdue.

Correlations between volcanic output along the ocean island chains can now be correlated with basin-wide plate re-organisations such as at 25 and 5 Ma [Kamp, 1991; Smith, 2003] suggesting the model of Jackson & Shaw [1975] was essentially correct. Reconstruction of the evolution of the Pacific basin demonstrates that oceanic plateaus were generated in zones of tension in the wake of retreating triple junctions, and that ocean island chains may be divided on the basis of propagating- or leaky-fracture origin [Smith, 2003]. The latter, including the Louisville, Marshall-Gilbert, Line Island, and Cook-Austral-Marquesas chains are those characterised by non-linear age progressions. Such volcanism followed pre-existing north-northwest/south-southeast trending fracture zones such as the Kashima-Eltanin and Emperor-Easter megatrends [Smoot, 1999]. The fracture zones form part of a pattern of orthogonally intersecting lineaments which are suggested to have been initiated by transform faulting along ridge systems during the early history of the Pacific plate.

Volcanism attributed to propagating fractures includes the Sala y Gomez, Juan Fernandez, and Caroline chains which extrapolate to breaks in nearby subducting slabs suggesting stressing of the plate by convergent margin geometry [Favela & Anderson, 1999; Smith, 2003].

The Emperor chain is unique in the stress field model in representing volcanism along a propagating fracture induced at a divergent margin. The location and orientation of this chain is attributed to the geometry of the Kula-Pacific ridge following plate re-organisations at 82 Ma which prematurely halted triple-junction volcanism on Meiji seamount. Subsequent volcanism along the Hawaiian chain can be explained by re-orientation of the stress field to control by convergent margin geometry following abandonment of the Pacific-Kula ridge, and does not require a change in Pacific plate motion at the time of the Hawaiian-Emperor bend (43 Ma).

The distribution of Pacific intraplate volcanism is therefore more regular than would be expected than in the plume model. It can be explained as a result of shallow volatile-bearing sources tapped under developing hotcell conditions. The change from oceanic plateau to island chain volcanism reflects changes in the stress field as the Pacific plate changed from having an intra-oceanic setting bordered by ocean ridge systems, to subducting beneath the basin margins [Smith, 2003].