Welcome to Earth Science Week, everyone! Why not start off with a bang?

At the end of last week, there was some buzz in the geoblogosphere and Twitter about a new paper in Nature that claims to have solved a riddle that has plagued igneous petrology for quite some time. That question is "why are volcanic arc above subduction zones so narrow and what controls their location relative to the trench." (OK, so, that is really two questions.) "Melting above the anhydrous solidus controls the location of volcanic arcs" by Philip England and Richard Katz in the latest issue of Nature tries to solve these conundrums.

However, before we get into those two questions, maybe some brief review on subduction zones. Subduction zones are convergent tectonic boundaries where one tectonic plate is shoved under another. The downgoing plate being subducted is always oceanic due to its denser, thinner nature. The overriding plate can be oceanic or continental. The current theory is that as the downgoing slab is subducted, it heats up as it is shoved into the mantle, so at a certain depth, the slab starts to dehydrate. All the water contained in the sediment and hydrous minerals (like amphibole, serpentine, talc and more that make up altered oceanic basalts) is released, adding water to the overlying mantle wedge (see below), causing the melting point of that mantle to go down - and causing it to melt and form magmas (flux melting). These magmas rise into the lithosphere and crust and eventually erupt on the surface as arc-related magmas. Most arc magmas are what we would classify as calc-alkaline magma (where the magma's CaO content is less than total K2O and Na2O). So, subducting slab releases water as the slab goes down, this prompts melting in the overlying mantle and magmas rise into the overriding plate.

A cartoon of the important geophysical and magmatic components of subduction zones. Figure from van Keken (2003)

So, back to the questions: The first is a question of geometry and process - volcanic arcs don't tend to be very wide as you move from the trench across the arc to the backarc - they tend to be tens of kilometers of volcanism, rather than hundreds or thousands. This is a little perplexing because in a subduction zone, the dominant theory is that melting is caused by water being added to the mantle (see image above) from the downgoing slab - the oceanic crust being shoved underneath the continental or oceanic overriding plate. One could argue that water is being released over a wide area of the slab, so a focussed zone of volcanism over all this dehydration is odd. The second question gets at what might be controlling the location of magma accumulation under the arc - why are most arcs located where they are relative to the downgoing slab and the trench. There have been a number of papers that have tried to tackle the other question - why are volcanoes sometimes spaced so regularly along an arc (Sherrod and Smith, 1990) - but this new study in Nature is more interested in why are arcs where they are and so narrowly focussed.

So, what do England and Katz have to say? Well, they take mathematical models of the thermal dynamics in subduction zone - mostly in the area above the downgoing slab and in the mantle wedge - and try to figure out what is (a) happening to the material in that region thermally and (b) what might be controls on where magma being generated might be apparent. Now, I am no expert on mathematical models (where are you, Magali Billen?), but they argue that their models suggest that the location of slab dehydration to cause hydrous melting does not control where the arc is located. Rather, it is the place in the mantle wedge where it gets hot enough to melt dry mantle (anhydrous - no additional water needed) that sets the stage for the location of the arc. Let's step back for a minute. As I said, the addition of water to mantle can help it melt by lowering the melting point (think about how salt does this to ice in the winter - it is very similar). However, increasing the temperature (with a little help from lowering of pressure from upwelling) will cause the mantle to melt, too - it causes the rock to cross its solidus and start to generate melt (magma). England and Katz models suggest that in the mantle wedge, there is a very restricted zone where this type of melting - decompression melting like we find at the midocean ridges - is allowed and it is this zone that dictates the location of the arc (see Figure 4 below).

Figure 4b from England and Katz (2010) showing the pathway created by the thermal erosion of anhydrous melts. This becomes a "path of least resistance" that other hydrous and anhydrous melts follow below the active volcanic arc.

Now, if I follow them, it isn't just the melting that does it. Rather, that new melt generated without water will then begin to rise buoyantly (remember, magma is less dense than the surrounding mantle, so it rises) and thermally erode (melt) its way through the overlying mantle creating a pathway for other melts - including hydrous melts being generated by the water from the slab - to follow. At the top of the wedge, in fact, where the mantle encounters the overriding plate, the anhydrous melts will begin to crystallize (they have a much higher melting point than hydrous melts), forming a barrier - a channel if you will - that sets up the confined and preferred route to the surface for all melts in the mantle wedge. The anhydrous melts are the vanguard, setting up the pathways to define where all the melts will rise and collect under the volcanic arc.

So, that is the long and short of it: hydrous melts might be generated in the mantle wedge by water from the subducting slab, but it is the very small amounts of dry melting that occurs in the mantle wedge that defines where the magmas under the arc migrate. Remember, this model is based almost solely on mathematical models of what we think the rheology and composition of the mantle wedge is (see figure below), which is a good question in itself. Their models are controlled by factors such as the dip of the downgoing slab, the rate of subduction, the size of the mantle wedge and the thermal diffusivity of the mantle (how well heat travels in the mantle). They claim that other ideas of the controls on the location of volcanic arcs - such as the temperature and pressure-dependence of water release from the slab - just don't cut the mustard and would produce very wide arcs, not the narrow ones they claim. They also examined a number of arcs worldwide and found some correlation - but a number of arcs don't fit their model calculations.

Figure 1 from England and Katz (2003) showing their idealized view of the mantle wedge.

This leads me to some questions, especially concerning how we can prove a radical model like this with real data, such as compositional data from, you know, the rocks themselves. First off, the England and Katz model implies that you have two types of melts forming: anhydrous melts - likely what we call tholeiitic basalt - less Na2O and K2O, more CaO (and iron) that is common at midocean ridges - and hydrous melts - the calcalkaline basalts mentioned above. These magmas can be differentiated by examining melt inclusions in minerals - the pockets of melt trapped in a crystallizing mineral that records the melt surrounding it at the time. If a mineral (such as olivine) is shown to be in equilibrium with those deep crystallizing conditions at the bottom of the crust, the melt inclusions in that olivine might record the presence of the tholeiitic basalt. You might also look for mantle xenoliths - chunks of rocks coughed up in an eruption - that could be evidence of crystallized tholeiitic basalt like the England and Katz model invoke at the bottom of the crust. Volcanic fields like the Big Pine in California have deep-sourced xenoliths that might reflected that boundary between the wedge and the overriding plate.

Secondly, there is no indication in the paper about the ratio and timing of these two magmas - hydrous and anhydrous. One would guess seeing we see little compositional evidence in the magmas erupted that the anhydrous basalt is far overweighed by the hydrous. Secondly, does the anhydrous melting occur forever, or is it truly a vanguard that then stops after the pathways are set. It is thought that the bottom of the crust under subduction zones is a MASH zone (Hildreth and Moorbath, 1988; see below) - which stands for Melting Assimilation Storage and Homogenization - where mantle-derived magma is modified by interactions with the lower crust. How might this new model change the MASH zone - or do we just have another player in the magma mixing game in the lower crust.

Cartoon showing the location of the MASH zone under a volcanic arc.

Thirdly, they mention that arc width is very tight - invoking a measure of "a few kilometers width" for most arcs. Now, that seems unduly narrow in my mind, just thinking about how the volcanoes plot along the Cascades or Andes. One thing that popped into my head is that many arcs are discontinuous and don't follow the same alignment their whole length. At Aucanquilcha in Chile, the active arc takes a pronounced step to the east - almost 30 km - as you head south starting at Ollagüe (Klemetti and Grunder, 2008). Does this mean that the location of anhydrous melting has changed under the arc itself? Their model - at least in the space presented in Nature - doesn't seem to tackle issues such as this.

Overall, I find the findings of England and Katz intriguing. Geodynamically, they show that melting without water is expected under arcs and those tholeiitic magmas will behave differently than hydrous melts. Their model also does do a good job is creating a focussing method of magmas under an arc - which explains the two perplexing questions about why arc location and width are so regular. However, without compositional or seismic data that can support the existence of the anhydrous melts (tholeiitic basalts), I remain cautiously skeptical. Mathematical models are great at giving us potential scenarios and hypothesize to prove, but without the physical data to support it, it cannot stand alone. 

References

England, P. C. and R. F. Katz (2010). "Melting above the anhydrous solidus controls the location of volcanic arcs." Nature 467: 700-704.

Hildreth, W. and S. Moorbath (1988). "Crustal contributions to arc magmatism in the Andes of Central Chile." Contributions to Mineralogy and Petrology 98: 455-489.

Klemetti, E. W. and A. L. Grunder (2008). "Volcanic evolution of Volcan Aucanquilcha: a long-lived dacite volcano in the Central Andes of northern Chile." Bulletin of Volcanology 70(5): 633-650.

Sherrod, D. R. and J. G. Smith (1990). "Quaternary extrusion rates of the Cascade Range, Northwestern United States and Southern British Columbia." Journal of Geophysical Research B 95(B12): 19,465-19,474.

van Keken, P. E. (2003). "The structure and dynamics of the mantle wedge." Earth and Planetary Science Letters 215: 323-338.

Top left: Aucanquilcha in Chile as seen from the mining town of Amincha. Image by Erik Klemetti, November 2000.