Beyond the Plume Hypothesis - a student's eye view of the Penrose conference by Marc Davies - 2 October 2003
There are many ways to cook an egg, and it is equally probable that there are many ways to generate intraplate volcanism. It is hardly surprising therefore that the plume hypothesis, in its most rigid form, cannot account for every occurrence of volcanism away from plate boundaries. Should a model as such then be taken as a rigid framework that is discounted as soon as it is seen not to fit? Or should the essence of the model be developed to formulate a more appropriate model?
Picture- the author, suitably braced by scenery, weather and debate. Photo - Marc Davies
This was clearly not the preface of the recent Penrose Conference, Plume IV: Beyond the Plume Hypothesis since the statement of its objectives presented in the opening address was to explore alternatives to the plume model. Again, it is hardly surprising that it is difficult to find a unified theory, as was called for at the end of the meeting, if we are only looking for alternatives.
It is certainly apparent that structural features related to lithospheric architecture and stress can play an important role in controlling intraplate volcanism. O'Connor neatly demonstrated this with reference to the Foundation and Juan Fernandez seamount chains in the South Pacific. It is also conceivable that many such seamount chains that exhibit linear age progressions may be explained by propagating fractures as in the cases of the New England Seamounts (McHone), and the parallel chains in the Pacific, including Puka Puka Ridge (Natland). Similarly, those with non-linear age progressions, such as Louisville, Marshall-Gilbert, Line Island, and Cook-Austral-Marquesas chains, may be interpreted as leaky transforms (Smith). Furthermore, the concept of edge-driven convection may be invoked to explain the voluminous volcanism and melt anomaly tracks associated with the margins of ancient cratons, such as the Columbia River Basalts and the Yellowstone hotspot trail (Christiansen), and possibly the Azores (King).
All these considered, it still remains unclear how such shallow level mechanisms can account for the huge volumes, high melting temperatures and rapid rates of emplacement characteristic of large igneous provinces. Even if we accept that edge-driven convection generated by strong contrasts in lithospheric architecture is capable of producing large volumes of magma, oceanic plateaux and continental flood basalt provinces with no obvious cratonic association still remain unexplained without reference to plumes. If we go as far to accept that changes in lithospheric stress can release huge melt ponds trapped beneath the lithosphere to create large igneous provinces, we have to then come up with some evidence to substantiate the existence of such ponds and explain how such features can be maintained at sub-lithospheric depths for an appreciable length of time.
After three days of heated and often objective discussion on the origins of various seamounts and oceanic island chains, in between the inevitable argumentative discourse on Iceland and Hawaii, I found it disappointing that the response to the presentations on large igneous provinces in the penultimate session of the conference was subdued and largely inconsequential. The role of carbon dioxide in lowering the solidus temperature (Presnall) is not sufficient to discredit the anomalously high temperatures reported for provinces such as the Ontong Java Plateau (Fitton) and the Ethiopian Traps (Davies), and the role of water (Green) doesn?t even come into the equation since these melts are essentially anhydrous. A lower mantle source for plumes may be discounted on the basis that convection in the lower mantle is suppressed by pressure (Hofmeister), and there is growing evidence that primordial noble gas signatures evident in many large igneous provinces are not necessarily indicative of a deep lower mantle contribution. Similarly, osmium isotope data indicate that there is no contribution from the core in the Ethiopian Flood Basalt Province. By discrediting a lower mantle or core we still cannot explain high melting temperatures and anomalous buoyancy fluxes. I went to Iceland looking for something other than a plume to explain the origin of high-Ti flood basalts from the Ethiopian Plateau; I came back without an alternative, but with a more panoramic view of intraplate volcanism than I had before.
In the absence of a unified theory there were elements of commonality at the end of all this discussion. There seemed to be a consensus that the upper mantle is compositionally heterogeneous, and that the geochemical variation we see at ocean ridges (Bonatti, Dick) and in intraplate volcanism is a consequence of this heterogeneiity. The components of the geochemists' alphabet soup should no longer be viewed as representative of discreet mantle reservoirs but as numerical limits, defining compositional mixtures of recycled bits and bobs which themselves constitute upper mantle heterogeneiity. It was not discounted that Earth's rotation is capable of generating a westward drift of the lithosphere relative to the mantle sufficient to influence plate motion, create global asymmetries at rift and subduction zones, and possibly induce shear-melting as a result of decoupling between the lithosphere and underlying asthenosphere (Doglioni). A slower net westward drift of the mantle resulting from same mechanism may also influence patterns of convection in the mantle.
The way forward was encapsulated in a few presentations which considered both non-plume and plume ideas together to explain the observed data. With reference to tomographic sections across the Horn of Africa, Montagner proposed that the Afar upwelling originates at a boundary layer at the 660 km discontinuity. Alongside this he suggested that a series of shallow level parallel N - S trending high velocity regions to the west of Afar are a consequence of edge-driven convection emphasised by the northward movement of the African plate. Similarly, Wilson acknowledged that shear/wrench faulting, as a result of the periodic release of plate stresses, played an important role in controlling the development of the Walvis Ridge and the Rio Grande Rise, previously interpreted as simple hotspot trails.
Still it remains unclear whether or not there is any exchange of heat or material between the upper and the lower mantle, and the boundary layers from which plumes arise remain elusive. With growing evidence that the core-mantle boundary is an unlikely candidate for the origins of many deep mantle plumes, perhaps we should be asking ourselves do we really need a persistent boundary layer to generate a plume? After all, all we need is a compositional difference to generate convection, and there is a multitude of such differences in a heterogeneous mantle.
This great plume debate perhaps needs refocusing. The non-plume protagonists initiated a healthy debate which has encouraged us to consider mechanisms other than boundary generated plumes to explain intraplate volcanism. The issue has now become one of scale! Large igneous provinces are difficult to explain without deep, hot upwellings or plumes, whereas the range of smaller scale features generated by intraplate volcanism from localised and regional volcanic fields to chains of tiny seamounts may be explained within the context of global lithospheric architecture and stress by some variant between mini-plumes, edge effects, propagating fractures and shear melting. Where Iceland and Hawaii fit into this framework is still open to debate, but regardless of whether we are 'top-downists' or 'bottom-upists', there is room within the framework to accommodate our theories. Perhaps then we can ease off defending our preconceptions and concentrate on generating data to test our inferences.
It seems as difficult to attribute all intraplate volcanism to global lithospheric stress patterns, which generally encompass the range of presented alternatives, as it is to explain everything with plumes. With reference to the constellation paradox quoted to illustrate how the plume theory focuses on those features that fit (Anderson), we should be as careful not to focus so much on the stars in between that we lose focus on the constellations themselves. The emphasis throughout the conference on the numerous chains of seamounts and calderas which can be adequately explained by shallow level processes, almost to the exclusion of considering the mechanisms responsible for the generation of large igneous provinces, went some way toward losing this focus. Nevertheless, the convenors of Plume IV: Beyond the Plume Hypothesis should be complimented on staging a meeting that actively encouraged debate. There was a lesson to be learned here! Limiting the presentations to a few provocative keynotes, leaving aside enough time for spontaneous discussion, and limiting numbers so that it was not so daunting to contribute to the proceedings, was altogether a fine recipe for a productive conference. With a finer balance between 'believers' and 'non-believers' it would have been the perfect forum to scramble the egg.
Back to the Plume debate homepage
References
The bracketed names cited in the text refer to presenters at the conference. Their abstracts are published in the conference handbook and may be viewed on the website http://www.mantleplumes.org
Acknowledgements
I should like to thank my supervisors Nick Rogers and Ian Parkinson for their constructive comments on the content and arrangement of the text, and Gillian Foulger and Don Anderson for their thoughtful remarks on the first draft. Also I should like to thank The Open University, The Geological Society of London, The Geological Society of America for their financial assistance, without which I could not have attended the conference.
* Marc Davies (Research Student) Department of Earth Sciences, The Open University, Walton Hall, Milton Keynes MK7 6AA UK
Marc.Davies@open.ac.uk