Presentation on theme: "Chapter 16. Island Arc Magmatism"— Presentation transcript:
1 Chapter 16. Island Arc Magmatism Arcuate volcanic island chains along subduction zonesDistinctly different from mainly basaltic provinces thus farComposition more diverse and silicicBasalt generally subordinateMore explosiveStrato-volcanoes most common volcanic landform
2 The initial petrologic model: Igneous activity related to convergent plate situations- subduction of one plate beneath anotherThe initial petrologic model:Subducted oceanic crust is partially meltedPartial melts- more silicic than sourceMelts rise through the overriding plate ® volcanoes just behind leading plate edgeUnlimited supply of oceanic crust to meltThis simple elegant model fails to explain many aspects of subduction magmatismPartial melts should be less mafic than their parentUltramafic mantle mafic basaltBasaltic oceanic crust intermediate andesites
3 The Subduction Factory From Tatsumi, Y. (2005) The subduction factory: How it operates in the evolving Earth. GSA Today, 15, 4-10.
4 Active Continental Margin (ACM) Ocean-ocean Island Arc (IA)Ocean-continent Continental Arc orActive Continental Margin (ACM)Figure Principal subduction zones associated with orogenic volcanism and plutonism. Triangles are on the overriding plate. PBS = Papuan-Bismarck-Solomon-New Hebrides arc. After Wilson (1989) Igneous Petrogenesis, Allen Unwin/Kluwer.
5 Subduction Products Characteristic igneous associations Distinctive patterns of metamorphismOrogeny and mountain beltsComplexlyInterrelatedOrogenic and Subduction-related synonyms when referring to the common association of basalts, basaltic andesites, andesites, dacites, and rhyolites produced at subduction zones= “orogenic suite”
6 Structure of an Island Arc Note mantle flow directions (induced drag), isolated wedge, and upwelling to back-arc basin spreading systemBenioff-Wadati seismic zone (x x x x)Volcanic Fronth is relatively constant depth is importantFigure Schematic cross section through a typical island arc after Gill (1981), Orogenic Andesites and Plate Tectonics. Springer-Verlag. HFU= heat flow unit (4.2 x 10-6 joules/cm2/sec)
7 Volcanic Rocks of Island Arcs Complex tectonic situation and broad spectrum of volcanic productsHigh proportion of basaltic andesite and andesiteMost andesites occur in subduction zone settingsBasalts are still very common and important!
8 Major Elements and Magma Series Tholeiitic (MORB, OIT)Alkaline (OIA)Calc-Alkaline (~ restricted to SZ)All three series occur in SZ setting, yet something about SZ is different that CALC-ALKALINECalc-alkaline magma series is used as yet another synonym to orogenic suite by some workersSince other magma series can occur at subduction zones, I recommend that we use the term calc-alkaline strictly to denote a chemical magma series, not a tectonic association
9 Major Elements and Magma Series a. Alkali vs. silicab. AFMc. FeO*/MgO vs. silicadiagrams for 1946 analyses from ~ 30 island and continental arcs with emphasis on the more primitive volcanicsFigure Data compiled by Terry Plank (Plank and Langmuir, 1988) Earth Planet. Sci. Lett., 90,
10 Figure 16. 4 The three andesite series of Gill (1981) Figure 16.4 The three andesite series of Gill (1981). A fourth very high K shoshonite series is rare. Contours represent the concentration of 2500 analyses of andesites stored in the large data file RKOC76 (Carnegie Institute of Washington).
11 Figure a. K2O-SiO2 diagram distinguishing high-K, medium-K and low-K series. Large squares = high-K, stars = med.-K, diamonds = low-K series from Table Smaller symbols are identified in the caption. Differentiation within a series (presumably dominated by fractional crystallization) is indicated by the arrow. Different primary magmas (to the left) are distinguished by vertical variations in K2O at low SiO2. After Gill, 1981, Orogenic Andesites and Plate Tectonics. Springer-Verlag.
12 Figure b. AFM diagram distinguishing tholeiitic and calc-alkaline series. Arrows represent differentiation trends within a series.
13 Figure c. FeO*/MgO vs. SiO2 diagram distinguishing tholeiitic and calc-alkaline series. The gray arrow near the bottom is the progressive fractional melting trend under hydrous conditions of Grove et al. (2003).
14 6 sub-series if combine tholeiite and C-A (some are rare) May choose 3 most common:Low-K tholeiiticMed-K C-AHi-K mixedFigure Combined K2O - FeO*/MgO diagram in which the Low-K to High-K series are combined with the tholeiitic vs. calc-alkaline types, resulting in six andesite series, after Gill (1981) Orogenic Andesites and Plate Tectonics. Springer-Verlag. The points represent the analyses in the appendix of Gill (1981).
15 Tholeiitic vs. Calc-alkaline differentiation These Harkers represent > 1 volcano from each arc, but give the general idea nonethelessC-A shows continually increasing SiO2 and lacks dramatic Fe enrichment (note Guatemala and PNG in FeO*/MgO)TiO2 decr due to Fe-Ti oxideCaO/Al2O3 should increase with Plag FX, but decreases in all types -> Cpx responsible for CaOSiO2 doesn’t show up in Harker, but it increases progressively in CALC-ALKALINE (although ~ constant in Thol)Figure From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.
16 Calc-alkaline differentiation Early crystallization of Fe-Ti oxideProbably related to the high water content of calc- alkaline magmas in arcs, dissolves high fO2High PH2O also depresses plagioclase liquidus more An-richAs hydrous magma rises, DP plagioclase liquidus moves to higher T crystallization of considerable An- rich-SiO2-poor plagioclaseThe crystallization of anorthitic plagioclase and low- silica, high-Fe hornblende may be an alternative mechanism for the observed calc-alkaline differentiation trend
17 Other Trends Spatial Temporal “K-h”: low-K tholeiite near trench C-A alkaline as depth to seismic zone increasesSome along-arc as wellAntilles more alkaline N SAleutians is segmented with C-A prevalent in segments and tholeiite prevalent at endsTemporalEarly tholeiitic later C-A and often latest alkaline is commonMany exceptions to any trend!
18 Trace ElementsREEsSlope within series is similar, but height varies with FX due to removal of Ol, Plag, and Pyx(+) slope of low-K DMSome even more depleted than MORB!Others have more normal slopes® heterogeneous mantle sourcesHREE flat, so no deep garnetFigure REE diagrams for some representative Low-K (tholeiitic), Medium-K (calc-alkaline), and High-K basaltic andesites and andesites. An N-MORB is included for reference (from Sun and McDonough, 1989). After Gill (1981) Orogenic Andesites and Plate Tectonics. Springer-Verlag.
19 MORB-normalized Spider diagrams Intraplate OIB has typical hump Figure Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall. Data from Sun and McDonough (1989) In A. D. Saunders and M. J. Norry (eds.), Magmatism in the Ocean Basins. Geol. Soc. London Spec. Publ., 42. pp
20 MORB-normalized Spider diagrams IA: decoupled HFS - LIL (LIL are hydrophilic)What is it about subduction zone setting that causes fluid-assisted enrichment?Figure Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall. Data from Sun and McDonough (1989) In A. D. Saunders and M. J. Norry (eds.), Magmatism in the Ocean Basins. Geol. Soc. London Spec. Publ., 42. ppFigure 16-11a. MORB-normalized spider diagrams for selected island arc basalts. Using the normalization and ordering scheme of Pearce (1983) with LIL on the left and HFS on the right and compatibility increasing outward from Ba-Th. Data from BVTP. Composite OIB from Fig 14-3 in yellow.
21 IsotopesNew Britain, Marianas, Aleutians, and South Sandwich volcanics plot within a surprisingly limited range of DMFigure Nd-Sr isotopic variation in some island arc volcanics. MORB and mantle array from Figures and After Wilson (1989), Arculus and Powell (1986), Gill (1981), and McCulloch et al. (1994). Atlantic sediment data from White et al. (1985).The principal source of island arc magmas is very similar to MORB source (DM)Trace element data still require enriched componentsThe data for the other arcs extend along 2 enrichment trends, one for the Banda arc and the other for the Lesser Antilles extend beyond the OIB fieldAntilles (Atlantic) and Banda and New Zealand (Pacific) can be explained by partial melting of a MORB-type source + the addition of the type of sediment that exist on the subducting plate (Pacific sediment has 87Sr/86Sr around and 143Nd/144Nd around 0.715)The increasing N-S Antilles enrichment probably related to the increasing proximity of the southern end to the South American sediment source of the Amazon
22 Pb is quite scarce in the mantle Low-Pb mantle-derived melts susceptible to Pb contaminationU, Pb, and Th are concentrated in continental crust (high radiogenic daughter Pb isotopes)204Pb non-radiogenic: 208Pb/204Pb, 207Pb/204Pb, and 206Pb/204Pb increase as U and Th decayOceanic crust also has elevated U and Th content (compared to the mantle)Sediments derived from oceanic and continental crustPb is a sensitive measure of crustal (including sediment) components in mantle isotopic systems93.7% of natural U is 238U, so 206Pb/204Pb will be most sensitive to a crustal-enriched componentMantle-derived melts are susceptible to contamination from U-Th-Pb-rich reservoirs which can add a significant proportion to the total PbU, Pb, and Th are concentrated in sialic reservoirs, such as the continental crust, which develop high concentrations of the radiogenic daughter Pb isotopes204Pb is non-radiogenic, so 208Pb/204Pb, 207Pb/204Pb, and 206Pb/204Pb increase as U and Th decayOceanic crust has elevated U and Th content (compared to the mantle) as will sediments derived from oceanic and continental crustPb is perhaps the most sensitive measure of crustal (including sediment) components in mantle isotopic systemsSince 93.7% of natural U is 238U, equation 9-20 will dominate over equation 9-21, and thus 206Pb/204Pb will be most sensitive to a crustal-enriched componentU 234U 206PbU 207PbTh 208Pb
23 HIMUGeochron = line on which all undisturbed/unreset Pb isotopic systems (such as BSE) plot.~ NONE of the arc volcanics fall on the geochron.Pb in some arcs overlap with the MORB (DM component = major subduction reservoir)Majority of data enriched in radiogenic lead (207Pb and 206Pb), trending toward the appropriate oceanic marine sedimentary reservoirSeveral arcs could = mixing of DM, PREMA, and sedimentary sourcesSunda extends to EMIIAntilles follow mixing line between DM or PREMA and Atlantic sediments, perhaps extending beyond toward HIMU (Fig. 14-6)Pb data clearly indicate a sedimentary component in arc magmasBut WHEN?? Ancient EM component or recent sedimentary additions to the mantle?Nd-Sr-Pb isotopes cannot distinguishFigure Variation in 207Pb/204Pb vs. 206Pb/204Pb for oceanic island arc volcanics. Included are the isotopic reservoirs and the Northern Hemisphere Reference Line (NHRL) proposed in Chapter 14. The geochron represents the mutual evolution of 207Pb/204Pb and 206Pb/204Pb in a single-stage homogeneous reservoir. Data sources listed in Wilson (1989).
24 10Be created by cosmic rays + oxygen and nitrogen in upper atmos. Earth by precipitationReadily clay-rich oceanic sedimentsHalf-life of only 1.5 MaLong enough to be subductedAfter about 10 Ma 10Be is no longer detectableNot a part of main mantle systems10Be/9Be averages about 5000 x in the uppermost oceanic sedimentsIn mantle-derived MORB and OIB magmas, & continental crust, 10Be is below detection limits (<1 x 106 atom/g) and 10Be/9Be is <5 x 10-149Be is a stable natural isotope & used as a normalization factor
25 Boron is a stable element Very brief residence time deep in subduction zonesB in recent sediments is high ( ppm), but has a greater affinity for altered oceanic crust ( ppm)In MORB and OIB it rarely exceeds 2-3 ppmVery brief residence time deep in subduction zone source areas & cycles quickly through to shallow crustal and hydrospheric systemsB concentrations in recent sediments is high ( ppm), but has a greater affinity for altered oceanic crust ( ppm)In MORB and OIB it rarely exceeds 2-3 ppm
26 10Be/Betotal vs. B/Betotal diagram (Betotal 9Be because 10Be so rare) Figure Be/Be(total) vs. B/Be for six arcs. After Morris (1989) Carnegie Inst. of Washington Yearb., 88,Each arc formed unique linear arrayOther known reservoirs: typical mantle (virtually no 10Be or B), hydrated and altered oceanic crust (high B, low 10Be), and young pelagic sediments (low B and 10Be/Be extending off the diagram up to 2000)Simplest explanation: each arc represents a mixing line between a mantle reservoir (near the origin) and a fluid (or melt) reservoir, that is specific for each arc and itself a mixture of slab crust and sedimentHypothetical fields for each arc are illustrated, but the exact location along the extrapolated line is unknown
27 Petrogenesis of Island Arc Magmas Why is subduction zone magmatism a paradox?Paradox: Great quantities of magma are generated in regions where cool lithosphere is being subducted into the mantle and isotherms are depressed, not elevatedNo adequate petrogenetic model can be derived without considering the thermal regime in subduction zones
28 Main variables that can affect the isotherms in subduction zone systems: 1) Rate of subduction2) Age of subduction zone3) Age of subducting slab4) Extent to which subducting slab induces flow in the mantle wedge5) Effects of frictional shear heating along W-B zoneOther factors, such as:Dip of slabEndothermic metamorphic reactionsMetamorphic fluid floware now thought to play only a minor role
29 Typical thermal model for a subduction zone Isotherms will be higher (i.e. the system will be hotter) ifa) Convergence rate is slowerb) Subducted slab is young and near the ridge (warmer)c) Arc is young (< Ma according to Peacock, 1991)yellow curves = mantle flowThis is sufficiently representativeFigure Cross section of a subduction zone showing isotherms (red-after Furukawa, 1993, J. Geophys. Res., 98, ) and mantle flow lines (yellow- after Tatsumi and Eggins, 1995, Subduction Zone Magmatism. Blackwell. Oxford).
30 The principal source components IA magmas 1. Crustal portion of the subducted slab1a Altered oceanic crust (hydrated by circulating seawater, and metamorphosed in large part to greenschist facies)1b Subducted oceanic and forearc sediments1c Seawater trapped in pore spacesFigure Cross section of a subduction zone showing isotherms (red-after Furukawa, 1993, J. Geophys. Res., 98, ) and mantle flow lines (yellow- after Tatsumi and Eggins, 1995, Subduction Zone Magmatism. Blackwell. Oxford).
31 The principal source components IA magmas 2. Mantle wedge between slab and arc crust3. Arc crust4. Lithospheric mantle of subducting plate5. Asthenosphere beneath slabThe last three sources on the list are unlikely to play much of a roleLithospheric mantle of subducting plate (4) is already refractory (extraction of MORB at the ridge) and it heats very little in the upper 200 km of the subduction zoneAsthenosphere beneath the slab (5) flows with it, but does not heat much at all, since the isotherms are essentially parallel to the flow lines at these depthsArc crust: isotherms at the base of the (predominantly andesitic) crust the temperature is too low for melting, even under hydrous conditions(much more important in active continental margins)Figure Cross section of a subduction zone showing isotherms (red-after Furukawa, 1993, J. Geophys. Res., 98, ) and mantle flow lines (yellow- after Tatsumi and Eggins, 1995, Subduction Zone Magmatism. Blackwell. Oxford).
32 Left with the subducted crust and mantle wedge Trace element and isotopic data ® both contribute to arc magmatism.How, and to what extent?Dry peridotite solidus too highLIL/HFS ratios of arc magmas water plays a significant role in arc magmatismSince we know what the general composition of the constituents in Fig are, it is a matter of combining this information with the other information in the figure showing us the pressure-temperature conditions to which the constituents will be subjected as they move through the subduction zone, and considering the consequences
33 Sequence of pressures and temperatures a rock subjected to during burial, subduction, metamorphism, uplift, etc. is called a pressure-temperature-time (P-T-t) pathOceanic crust, as it subducts, will begin to heat at about km depth, and will continue to heat with rising pressure, although slowly due to the depressed isothermsMantle wedge material will follow the path of drag-induced flow, also illustrated in Fig This is less well known, but should follow a path (like the arrows beginning in the center of the right edge of the figure) of initial cooling from about 1100oC to about 800oC at nearly constant pressure, and then heat up toward 1000oC as pressure increases
34 P-T-t paths for subducted crust Based on subduction rate of 3 cm/yr (length of each curve = ~15 Ma)Subducted CrustYellow paths = various arc agesRed paths = different ages of subducted slabThe yellow P-T-t paths represent various arc agesA newly formed arc will not have subducted sufficient cool slab material to fully depress the initial ocean geotherm that existed prior to the initiation of subductionIt will thus follow a path toward higher temperatures at low pressureOnly after Ma will a steady state equilibrium be attained between subduction and heating of the slabFigure Subducted crust pressure-temperature-time (P-T-t) paths for various situations of arc age (yellow curves) and age of subducted lithosphere (red curves, for a mature ca. 50 Ma old arc) assuming a subduction rate of 3 cm/yr (Peacock, 1991, Phil. Trans. Roy. Soc. London, 335, ).
35 Add solidi for dry and water-saturated melting of basalt and dehydration curves of likely hydrous phasesSubducted CrustFigure Subducted crust pressure-temperature-time (P-T-t) paths for various situations of arc age (yellow curves) and age of subducted lithosphere (red curves, for a mature ca. 50 Ma old arc) assuming a subduction rate of 3 cm/yr (Peacock, 1991). Included are some pertinent reaction curves, including the wet and dry basalt solidi (Figure 7-20), the dehydration of hornblende (Lambert and Wyllie, 1968, 1970, 1972), chlorite + quartz (Delaney and Helgeson, 1978). Winter (2001). An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.None of the P-T-t paths intersect the dry solidus -> NO partial melting of dry basaltic crustMelting of wet basalt will depend upon the availability of “free” (non-bonded) waterNumerous metamorphic reactions involve dehydration of the hydrated oceanic crust and/or subducted sedimentsBecause chlorite and hornblende are the two dominant hydrous minerals in metamorphosed hydrated basalt, the breakdown curves for chlorite + quartz and hornblende are includedBiotite and other hydrous minerals will mostly dehydrate between these curves
36 Mature arcs (lithosphere > 25 Ma): Dehydration D releases water in No slab melting!Slab melting M in arcs subducting young lithosphere.Dehydration of chlorite or amphibole releases water above the wet solidus ® (Mg-rich) andesites directly.Subducted CrustWhat happens to intermediate cases, such as the curve for the 5 Ma old arc?If dehydration occurs before the subducted material reaches the wet melting point, will it melt?Depends upon the fate of the water released by dehydrationIf it remains in place, melting will occur when the wet solidus is reachedMany workers believe that the water rises and leaves the system, and the remaining crust will no longer have free water, and no melting will occur at the wet solidusAn anhydrous eclogite soon formsWill it melt, or will the water escape first? This is the subject of great controversy, and models diverge at this point into those that perceive the melting of the oceanic crust (as eclogite plus water) as the dominant source of arc magma, and those that favor the mantle wedge.
38 Slab melting in mature arcs no longer precluded by models. Subducted CrustSlab melting in mature arcs no longer precluded by models.Debate renewed.What happens to intermediate cases, such as the curve for the 5 Ma old arc?If dehydration occurs before the subducted material reaches the wet melting point, will it melt?Depends upon the fate of the water released by dehydrationIf it remains in place, melting will occur when the wet solidus is reachedMany workers believe that the water rises and leaves the system, and the remaining crust will no longer have free water, and no melting will occur at the wet solidusAn anhydrous eclogite soon formsWill it melt, or will the water escape first? This is the subject of great controversy, and models diverge at this point into those that perceive the melting of the oceanic crust (as eclogite plus water) as the dominant source of arc magma, and those that favor the mantle wedge.
39 Flat HREE pattern argues against a garnet- bearing (eclogite) source LIL/HFS trace element data underscore the importance of slab-derived water and a MORB-like mantle wedge sourceFlat HREE pattern argues against a garnet- bearing (eclogite) sourceModern opinion has swung toward the non- melted slab for most casesMcCulloch (1993) suggests that young, warm subducted crust is more likely to melt before it dehydrates. Thus slab melting may be more common where the a subduction zone is close to a ridge, or during the Archean, when heat production was greater
40 Mantle Wedge P-T-t Paths Start at XTemperature decreases followed by pressure increase with only slight temperature increase (clockwise P-T-t)
41 Amphibole-bearing hydrated peridotite should melt at ~ 120 km Phlogopite-bearing hydrated peridotite should melt at ~ 200 km second arc behind first?Figure Calculated P-T-t paths for peridotite in the mantle wedge as it follows paths similar to the flow lines in Fig Included are dehydration curves for serpentine, talc, pargasite, and phlogopite + diopside + orthopyroxene. Also the P-T-t path range for the subducted crust in a mature arc, and the wet and dry solidi for peridotite. Subducted crust dehydrates, and water is transferred to the wedge (labeled arrows). Areas in which the dehydration curves are crossed by the P-T-t paths below the wet solidus for peridotite are stippled and labeled D for dehydration. Areas in which the dehydration curves are crossed above the wet solidus are hatched and labeled M for melting. Note that although the slab crust usually dehydrates, the wedge peridotite melts as pargasite dehydrates (Millhollen et al., 1974) above the wet solidus. An alternative model involves dehydration of serpentine chlorite nearer the wedge tip (lower-case d) with H2O rising into hotter portions of the wedge (gray arrow) until H2O-exess solidus is crossed (lower-case m). A second melting may also occur as phlogopite dehydrates in the presence of two pyroxenes (Sudo, 1988). After Peacock (1991), Tatsumi and Eggins (1995). Winter (2001). An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.Crust and Mantle WedgeAreas in which the dehydration curves are crossed by the P-T-t paths below the wet solidus for peridotite are blue and labeled D for dehydrationAreas in which the dehydration curves are crossed above the wet solidus are purple and labeled M for melting.Note that although the slab crust usually dehydrates, the wedge peridotite melts as amphibole dehydrates above the wet solidusA second melting may also occur as phlogopite dehydrates in the presence of two pyroxenes.
42 Island Arc Petrogenesis Figure A proposed model for subduction zone magmatism with particular reference to island arcs. Dehydration of slab crust causes hydration of the mantle (violet), which undergoes partial melting as amphibole (A) and phlogopite (B) dehydrate. From Tatsumi (1989), J. Geophys. Res., 94, and Tatsumi and Eggins (1995). Subduction Zone Magmatism. Blackwell. Oxford.Altered oceanic crust begins to dehydrate at depths ~ 50 km or less, as chlorite, phengite, and other hydrous phyllosilicates decomposeFurther dehydration takes place at greater depths as other hydrous phases become unstable, including amphibole at about 3 GPa.The slab crust is successively converted to blueschist, amphibolite, and finally anhydrous eclogite as it reaches about km depthIn most (mature) arcs, the temperature in the subducted crust is below the wet solidus for basalt, so the released water cannot cause melting, and most of the water is believed to rise into the overlying mantle wedge, where it reacts with the lherzolite to form a pargasitic amphibole and probably phlogopite (yellowish area)Slightly hydrous mantle immediately above the slab is carried downward by induced convective flow where it heats up, dehydrates, and melts at A (120 km)
43 A multi-stage, multi-source process Mantle wedge ® HFS and other depleted and compatible element characteristicsSlab dehydration (and perhaps melting) ® LIL, 10Be, B, etc. enrichments + enriched Nd, Sr, and Pb isotopic signaturesThese components, plus other dissolved silicate materials, are transferred to the wedge in a fluid phase (or melt in some cases?)Mass-balance calculations suggest that the contribution of the subducted sediments is only a few percent in most arc systems, but the result is the high LIL/HFS patternsThe nearly closed-cell induced flow in the wedge may result in progressive depletion of the wedge as arc magmas are extracted. This provides an explanation of some HREE and more compatible trace element data, which in many cases is more depleted than MORB
44 Phlogopite is stable beyond amphibole breakdown Wedge P-T-t paths reach phlogopite dehydration at ~ 200 km depthIf this occurs above the wet peridotite solidus, a second phase of melting will occur at B a position appropriate for the secondary volcanic chain that exists behind the primary chain in several island arcsThe P-T-t paths are nearly parallel to the solidus, and may be above it or below it. Thus dehydration may or may not be accompanied by melting, so that the development of a second arc will depend critically upon the thermal and flow regime of a particular arcMelting initiated by the breakdown of the potassium-rich mica will probably be more potassic, as is true in most secondary arc occurrences.The K-h relationship probably more complex, reflecting the decreasing quantity of H2O with depth and thus the degree of partial melting, as well as the depth of melting (which becomes more alkaline with depth), and perhaps the vertical length of the rising melt diapir column within the mantle wedge
45 Fractional crystallization takes place at a number of levels In the shallower chambers the calc-alkaline fractionation trend takes place in a hydrous magma with the fractionation of magnetite, hornblende, and/or a highly anorthitic plagioclase, as discussed previouslyThe restriction of calc-alkaline magmas to subduction zones may thus result from the uniquely high water content or the thickened arc crust that causes the primary tholeiites to pond and fractionateThe SZ environment is a complex one, and the generation of arc magmas reflects a number of sources and stagesThe wedge may be heterogeneous and variably depleted/enriched, the subducted material has a variety of crustal and sedimentary constituentsThe thermal and flow patterns are variable, and the nature of the fluid is poorly constrained and probably variable as well
46 From Peacock (2003) Geophysical Monograph 138 Am. Geophys. Union