The arguments presented by Bourdon et al. are threefold: (1) our data set is not representative of the Northern volcanic zone, (2) the presence of high-Mg andesites is unambiguous evidence of slab–mantle wedge interaction, and (3) SiO2 and Sr/Y variations require slab melting. We address each issue in turn below, as well as reemphasizing our main points: (1) that the geodynamic setting of the Northern volcanic zone is not suitable for producing adakites, particularly given geophysical evidence that the slab beneath Ecuador dips 25° (Guillier et al., 2001), and (2) we do not exclude slab melting in the Northern volcanic zone, we simply make the point that identification of slab melts based on geochemical criteria alone is a fallacious and inadequate approach, and that any such signature in the Northern volcanic zone is weak at best.
While our data set is indeed smaller, it includes all data that were published for the Northern volcanic zone at the time of submission, from north to south (i.e., Galeras to Sangay), as well as west to east (i.e., Atacazo to Sumaco). It is of course regrettable that the large data set accumulated over several years available to Bourdon and colleagues is not available to other researchers. We maintain that our abbreviated set is representative of the Northern volcanic zone, as it includes representative volcanoes from what Bourdon et al. consider to be three volcanic chains. Our goal was to construct a geochemical traverse of the Northern volcanic zone, which spans the range of the alleged subducting Carnegie Ridge, and for this purpose our data set is sufficient. Using a larger databank does not make the slab melting argument more convincing; indeed Bourdon et al.'s Figure 1 shows volcanoes at the same latitude with both high- and low-Y concentrations. Slab melting would be expected to produce ubiquitous “adakitic” signatures over a significant length of arc. The temporal constraints for Cayambe volcano do not necessarily reflect subduction of the Carnegie Ridge, particularly since there are no time constraints on when, and even if, the Carnegie Ridge was subducted, a point which was made in our original paper. It is worth pointing out here that the original slab melting hypothesis for the Northern volcanic zone (Gutscher et al., 2000) was based on geochemistry, not geophysics, so using geochemistry as evidence for slab melting is circular. In fact, there is no geophysical evidence for slab melting, and the evidence for the alleged “flat slab” is equivocal at best.
As for high-Mg andesites at some Northern volcanic zone volcanoes, we specifically stated in our paper that none of the published data included high-Mg andesites. As this is now an issue, we point out that of the eight data points (representing three volcanoes: Pinchincha, Antisana, and Sumaco) used to argue for slab melting in the Northern volcanic zone (Bourdon et al., 2003), only three of these have unusual Mg#s, and these range from only 0.32 to 0.40. These are not particularly high compared to the values of 0.45–0.65 for Archean TTGs (tonalite-trondhjemite-granodiorite), which we agree are most likely due to eclogite melting (Drummond and Defant, 1990). In fact, the lack of widespread high-Mg andesites in the Northern volcanic zone argues against models of slab melting, since the Gutscher et al. (2000) and Bourdon et al. (2003) models predict that most andesites should be high-Mg andesites. Experimental literature (Rapp et al., 1999) shows that high-Mg andesites can be produced by slab melts that interact with the mantle wedge. This is true, but does not preclude high-Mg andesites being produced by other mechanisms, such as recycling of lower crust into the mantle wedge (Gao et al., 2003) or hydrous melting (Kushiro, 1972). Another possibility is that mixing of primitive and evolved melts, a likely process during the protracted passage through 60 km of Northern volcanic zone crust, will produce much more Mg- and Ni-rich magmas than crystal fractionation. Furthermore, if there is mantle wedge, the slab is not flat. If the slab is not flat, there is no thermal mechanism by which the slab would melt, a point that is avoided by Bourdon et al. Indeed, high-Mg andesites are in danger of suffering the same fate as adakites (that is high Sr/Y rocks) in that they are used as evidence of slab melting. This argument is based only on geochemical signature and effectively circumvents the issue of the viability of slab melting.
In our paper, we show that the Sr/Y signature that characterizes slab melting can also be produced by crystal fractionation and through melting of hydrated metamorphosed basalt in the lower crust. We do not argue that Sr/Y values above or below 100 exclude slab melting, and in fact we never exclude the possibility of slab melting in the Northern volcanic zone. We simply present a logical case to show that the geochemical evidence for slab melting is highly equivocal, and that the geodynamic setting is not suitable. The arguments presented by Bourdon et al., criticizing our use of Sr/Y and SiO2, underscore this point, that the use of Sr/Y as the sole indication of slab melting is problematic, which is precisely why we presented these data in conjunction with the overall geodynamic setting in the Northern volcanic zone, a point that remains uncontested. Bourdon et al.'s Figure 1 does not show that Y concentrations are lowest above the alleged trace of the Carnegie Ridge, only that they are more variable.
We maintain that the geodynamic setting of the Northern volcanic zone is not conducive to the formation of adakites. The geochemical signature that would be expected from slab melting is weak at best, and could have been produced by many different igneous processes. No evidence has been provided to show that the Carnegie Ridge seamount chain is actually subducting, whether it contributes any anomalous material to the wedge, or that the slab is flat. There remains no geodynamic justification for the steep-flat-steep slab geometry proposed by Bourdon et al. (2003). We reassert the conclusions made in our original paper, that regional geochemical trends in the Northern volcanic zone and their relationship to the subduction-zone architecture are not a priori indications of slab melting and can be fully accounted for by normal arc magmatic processes acting on wedge-derived basaltic magmas.
We appreciate the helpful suggestions from Dennis Geist in reviewing this reply.