Downloaded from http://jgs.lyellcollection.

org/ by guest on May 7, 2014

Journal of the Geological Society

The geochemical characteristics of granitoids in

contrasting arcs and comments on magma sources
G. C. Brown, R. S. Thorpe and P. C. Webb

Journal of the Geological Society 1984, v.141; p413-426.

doi: 10.1144/gsjgs.141.3.0413

Email alerting click here to receive free e-mail alerts when new
service articles cite this article
Permission click here to seek permission to re-use all or part
request of this article

Subscribe click here to subscribe to Journal of the

Geological Society or the Lyell Collection

Notes

© The Geological Society of London 2014

 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
J . geol. Soc. London, Vol. 141, 1984, pp. 413-426, 8 figs, 1 table. Printed in Northern Ireland.

Thegeochemicalcharacteristics of granitoidsincontrastingarcsand
comments on magmasources

G . C. Brown, R. S. Thorpe and P. C. Webb

SUMMARY: Geochemical variations in modern subduction-related igneous suites with respect
to arc ‘maturity’ in time and space are illustrated using data for both volcanic suites (basalt,
andesite and dacite) and intrusivegranitoidsuites (diorite, tonaliteigranodiorite and granite)
fromcircum-Pacific arcs. Usingtraceelementdata wesuggest that twogroupsofprocesses
controlthenaturalvariation in theparentalmagmas of bothsuites: (a) subduction-zone
enrichment of lithospheric mantle, locally coupled with crustal assimilation allied with fractional
crystallization (AFC) in zonesofthickened crust, all of whichyieldmagmas with enhanced
concentrations of the large-ion lithophile (LIL) elements K, Rb, Th, U, LREEs, etc; (b) with
increasingdistancefromtheactive trench,contributionsfromwithin-platesub-continental
lithosphereproducingmantle-derivedmagmaswithenhancedlevels ofhigh-fieldstrength
(HFS) elements, among which Nb, Ta, Hf and Y are particularly distinctive. Thus, even for the
evolved granitoids of intrusive arc series, ratios of HFSiLIL elements not significantly affected
by crystal fractionation, such as (Ta, Nb)/(K, Rb,La), may throw some light on the origin of
mafic-intermediateprecursormagmas.Interms of theseelements wesuggestthefollowing
groupings for interpreting the tectonic associations of granitoid suites. 1. Primitive, calcic arc
granitoids with low LIL and HFS element abundances. 2. Normal, calc-alkaline continental arc
granitoids with enhanced LILelementabundances and lowHFSiLILratios. 3. Mature
alkali-calcicarcgranitoidswith high levels ofLILand HFS elements and higherHFS/LIL
ratios. 4. Back-adanorogenic alkaline granitoids with the highest levels of HFS elements.

Thebroad geochemical/petrological differences be- light is to be thrown on the petrogenesis of these suites
tween (a) calc-alkaline, met- to peraluminous subduc- and on thesetting of ancient analogues, it is important
tion-related,(b) alkali-calcic, peraluminous collision- to compare and classify granitoids from a variety of
related and (c) alkali-calcic, peralkaline tension- modern arc-related settings. Although there is debate
related igneous suites of intermediate and acid about detailed space-time relationships (Bussell et al.
compositions are well documented in terms of tectonic 1976; Thorpe & Francis 1979) much can also be learnt
environment and plausible magma sources. Yet by comparing data from both extrusive and intrusive
geochemical studies of modern subduction-related suites, for it is clear that there are common problems
igneousrockshave identified somecharacteristics of of magma genesis. In this contribution a petrological-
groups (b) and (c) in suites of type (a). In simple terms geochemical classification of subduction-related grani-
this is because the petrogenesis of all subduction- toids, similar to that for extrusive rocks, is proposed
related
magmas results from common
a set of and used to comment on possible mantle and crustal
processes, each of which affects individual suites to a components involved in their origin. The classification
greateror lesser extent.These processes include is preliminary because certain trace element compari-
partialmelting of heterogeneousuppermantle,de- sons could only becarried out using the evolved
hydration andthe incorporation into magmas of graniticend-members of intrusive seriesfor which
materials derived from basaltic and sedimentary ocean trace element dataare available fromseveral arc
crust within subduction zones, and partial melting of types-further work will be necessary to validate the
heterogeneous crust. Such crustalmaterial may be classification over a wider range of less evolved and,
contributedthrough assimilation accompanied by therefore, potentially less fractionated compositions.
fractional crystallization (AFC), a process which may
be fundamental in the generation of acid-basic igneous
suites above subduction zones. The diversity of granite
Whilst the respective signatures of different source magma types
materials are being identified from recent trace
element and isotopic studies of erupted basalt-andesite Various authors (Brown 1982; Harris et al. 1984;
liquid compositions(Hawkesworth 1982; Hole et al. Pitcher 1983) have recognized arange of different
1984; Pearce1983), for intrusive complexes there is granitoids, many of which are the product, directly or
the additional complication that cumulates and liquids indirectly, of subduction processes. In this context we
cannot easily be distinguished. Moreover,thereare use theterm ‘granitoid’ to describe intrusive com-
relatively few reported data
on intrusive suites, plexes ranging in composition from
diorite and
particularly for critical trace elements.Yet if some tonalite through granodiorite to granite (sensu stricto)
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
414 G. C. Brown et al.
in mineral composition. The problem is that siliceous and Malaya: Vidal et al. 1982; Scharer & Allegre 1983;
liquids of diverse origins tend to evolve,through Beckinsale 1979)-the so-called S-type granitoids of
fractionalmelting and fractionalcrystallization,to- Chappell & White (1974). At the opposite extreme are
wards rather similar end products in mineralogical and the calcic or metaluminous gabbro-quartzdiorite
major element terms. Nevertheless, systematic major granitoids of oceanic island arcs and young continental
element studies (cf. Petro et al. 1979) have provided a cordillera [e.g.,the Western pacific arcs (source
basis for recognizing certainperalkaline acid-basic references quoted in Brown 1982) and the early
suites as being connected with tension-related(non- intrusions of the Caribbean andCentral American arcs
subduction) processes: these aretheA-type(anoro- (Kesler 1978; Pichler & Weyl 1975)l; these arethe
genic) granitoids of Pitcher (1983). M-type granitoids of White (1979, quoted in Pitcher
Granitoid types produced by melting within and 1983). Lying between these two extremes in terms of
abovesubductionzones or within collision belts majorelement compositions arethe major I-type
overlap in composition. There are notable examples of calc-alkaline to alkali-calcic, metaluminous to peralu-
leucocratic two-mica monzogranites, with alkali-calcic minous batholith-forming suites that comprise a range
and peraluminous characteristics, occurring in collision of rock types including diorite, tonalite, granodiorite,
belts, for which there is good isotopic evidence of a monzogranite and granite or syenogranite. Figure 1
major crustal contribution (e.g. in the High Himalayas illustrates variations in the major element chemistry of

~ ~

60 50 70
SO2%

FIG. 1. Plot of log (CaO/(Na,O + K,O) against SiO, for intrusive suites from Mesozoic and Cenozoic magmatic
arcs compared with the range for normal calc-alkaline andesites (stipple; from Brown 1982). The alkali-lime index is
theSiOzconcentrationwhere CaO = KZO+ Na,O. The labelling is as follows: solid lines, NB = New
Britain-Solomon Islands arc; NM = New Guinea mobile belt; NC = New Guinea continental arc; dot-dash lines,
E P = early Panama diorites; CA = N Caribbean arc; LP = late Panama granodiorites; broken lines, PCB = Peru
Coastal Batholith (McCourt 1981); SN = Sierra Nevada batholith, AR = Alaska Range batholith. The two stars,
labelled T, are average Thailand granites (from Brown 1982, fig. 10).
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
Granitoids of contrasting arcs 415
these suites. Chronological and geochemical studies of in other inlandcordillera, might be influenced by
Mesozoic-Cenozoic subduction-related continental heterogeneousmantle chemistry instead of, or in
margin batholiths around the Pacific margins (discus- addition to, more extensive sialic crustal melting and
sed by Brown1982) show that thesesuites become contamination (see later).
more alkali-calcic, silicic (e.g., EP and LP in Fig. 1) Further Nstill, the Cretaceoustin-bearinggrani-
and peraluminous with decreasing age in one location, toids of the Seward Peninsula in Alaska include
and with increasing distance from the active subduc- another high-K suite which, according to Hudson &
tion margin (e.g., NB-NM-NC in Fig. 1). At a single Arth (1983), had a complex multi-stage origin involv-
location,
early intrusive phases may have I-type ing large-scale melting of heterogenous
a source,
compositionswhereas later phases may have greater which they identified as sialic crust, followed by
affinities with S-type compositions, and it is here that fractional crystallization from which residual magmas
the I-S terminology, which was intended to simplify continued to evolve intothe high-K suite; these
the genetic classification of granitoids, has sometimes authors stated that ‘theapplication of the I and S-type
proved difficult to apply. criteria to magma origins could be overly simplistic
As originally conceived, I-type magmas formed by and misleading’ (Hudson & Arth, 1983, p. 788).
partial melting of basaltic or andesitic igneous material Elsewhere in the world less detailed analysis is
in or abovesubductionzones whilst S-type magmas available, yet similar trends towardsmore alkalic,
werederived by anatexis of pelitic metasedimentary peraluminous intrusive rock-types towards the inland
material. Yet mixtures between these two end mem- parts of continental arcs have been noted. Perhaps the
bers seem to exist and, in most cases,they reflect greatestrange occurs across Papua New Guinea
mixed magma sources, though hydrothermally altered (Mason & MacDonald 1978) where,onthe Pacific
I-type granites with secondary muscovite may, in some side,the Solomon Islands-New Britain island arc
instances,have features normally consideredcharac- contains calcic (M-type) diorite-monzonite stocks
teristic of S-types. which give way to a normal low-K calc-alkaline I-type
Concentrating on regional trends across intrusive basic-acid suite in the Papua New Guinea ‘Mobile
cordillera;
the well-known California Peninsular Belt’ and ultimately to a normal to high-K suite across
Ranges batholith,forexample, comprises a low-K the ‘Continental Arc’ of southern Papua New Guinea
gabbro-quartzdiorite association in the W (initial (Fig. 1). Parallel belts of largely I-type and apparently
s7Sr/s6Sr= 0.703) givingway gradationally eastwards S-type Mesozoic batholiths occur throughout eastern
to a slightly younger I-S sequence of tonalites, Asia from the Soviet Far East down to S Korea, China
granodiorites and quartz monzonites with higher and Japan (Takahashi et al. 1980;Hong 1983). The
s7Sr/%r initial ratios (c. 0.708; Taylor & Silver 1978). picture here is complicated because some of the
Similarly, further N, the inlandparts of the Sierra two-mica granites are associated with outer arc
Nevadabatholith andthe WesternCanadian Coast accretionary zones (e.g., in the greenschist belt of SE
Range plutonic complex carry muscovite-bearing gra- Japan) rather than always with magmatism behind the
nites that are strongly peraluminous and have the high main arc, although the latter is the general rule.
initial ”Sr/@Sr ratios (0.708-0.734) that are commonly Finally, we turntothe Mesozoic-Cenozoic bath-
associated with S-type granitesformed by crustal oliths of S America (Fig. 2) which, in terms of the
anatexis in collision zones. In their summary of these Colombian Cordillera, the Peru Coastal batholith and
inner cordilleran granitoids, Miller & Bradfish (1980) the S Chile Batholith, are dominantly I-type and, from
noted that a simple S-type origin by partial melting of such isotopic data as are available, were derived from
metasediment is not required, althougha greater sources with mantle chemical characteristics (see
contribution from ancient crustal material than for the summary by Pitcher1983; also Saunders et al. 1980;
I-typemarginalbatholiths is (in their view) implied. Saunders and Tarney 1982).
However, there
are
DePaolo(1981~) useda wide range of Ndand Sr indications of the spatial evolutionary trends described
isotopic dataforthe SierraNevada and Peninsular above in theform of thelate Tertiary Cordillera
Ranges batholiths to conclude that the magmas could Blanca batholith of Peru, which contains more evolved
simply beproduced by two-component mixing (i.e., peraluminousbiotitegranites, occasionally two-mica
mixing between ‘island arc’ and‘meta-sedimentary’ granites, with slightly higher initial strontium isotope
components) in which the crustal component increases ratios than to the W (data quoted in Pitcher 1983).
from W to E. However, an illuminating isotopic study Similarly, an eastwards trend fromtonalitic to more
of diverse crustal and mantle xenoliths taken from the potassic monzonitic rocks occurs in northernand
SierraNevadabatholith by Domenick et al. (1983) central Chile (Aguirre et al. 1974; McNutt et al. 1975),
suggests that direct derivation of the different magma whilst in the thickest and most remote part of the
compositions from isotopically heterogeneous upper active Andes, 300 km from the subduction margin, are
mantle and lower crust is equally plausible. This raises tin-bearing two-mica granitesand quartz latitepor-
the possibility that the chemistry of apparently more phyry stocks in Bolivia. The intrusive rocks of the
evolved magmas in the eastern Sierras, and by analogy Mesozoic Cordillera Real in northern Bolivia and the
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
416 G. C. Brown et al.

10 10

DN
O N

0
0

10
10
Peru C o a
Batholith
\
\
LChiar
1 KkOllt 20
20

30 30

40 40

O S
' S

50 50
80 70 V 80 70
(a) (b)
FIG.2. Map of western S America showing active volcanism and Mesozoic-Cenozoic intrusive rocks in relation to
the plate tectonic framework, crustal age and thickness. (a) Distribution of active calc-alkaline basalt-andesite-dacite
volcanoes (solid triangles) and some alkaline volcanoes (open triangles). The open circles correspond to locations
mentioned in the text. The stippled areas are underlain by continental crust of Palaeozoic or younger age, and other
areas may be underlain by Precambrian crust. The location of crust exceeding 50 km in thickness is shown (from
Thorpe et al. 1984, Fig. 1). (b) Distribution of Mesozoic-Cenozoic calc-alkaline (I-type) gabbro-diorite-
tonalite-graniteintrusions (solid ornament) and someback-ardanorogenic(A-type)alkaline intrusive complexes
(stars). The open circles represent Cretaceous-Cenozoic possible M-type plutons in western Colombia. The broken
line encloses the eastern Cordillera igneous province of western Bolivia. The dot-dash line represents the eastern
limit of the Andean mobile belt (based on Pitcher 1983; Stewart 1971; Grant et al. 1979; Darbyshire & Fletcher
1979; Fletcher et al. 1981; Fletcher & Litherland 1981; and our unpublished data).
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
Granitoids of contrasting arcs 417
Miocene Kari-Kari batholith further S (Francis et al. al. 1982; Marriner & Millward 1984) through Ecuador,
1981) contain up to 11% K (Grant et al. 1979). EasternPeru and Bolivia down to NW Argentina
Bearing in mind that low-Ca peraluminousand (Dkruelle 1982). Their origin may be related to partial
peralkaline liquids need only have small differences in melting of adifferent kind of 'within-plate' mantle
chemical composition, these highly alkaline rocks may source in a back-arc environment, but Kroonenberg et
represent atrend away from peraluminoustowards al. (1982) suggested that'deep faults that enable
peralkaline magmatism, such as identified in the form magma to ascend are related somehow to the
of Neogenefeldspathoidal volcanic suites in eastern subduction process'.
Peru and Ecuador (Stewart 1971). This brings us full Mantle-normalized trace element data for NVZ and
circle, back toA-type granitoids,forif, as Stewart SVZ basaltic andesites are compared in Fig. 3a with
proposed, these young 'back-arc' volcanic centres alkali basalts from SW Bolivia and NW Argentina,
were to have intrusiveperalkalinesyenogranite or
nepheline syenite cores with within-plate characteris-
tics, then the apparent I-S trend across the cordilleran
batholiths could reflect derivation of liquids from
heterogeneous mantle that changes characteristics
from subduction type to within-platetype with
increasing distance from the trench. The question now
arises as towhether critical trace elements might
discriminate between crustally-derived (true S-type)
magmas, subduction zone mantle-derived magmas
(I-types) that have undergone crustal contamination,
acquiring some S-type characteristics, and within-plate
mantle-derived magmas that also may or may not have
undergone crustal contamination. Some light has been
thrown on this problem by recent geochemical data on
active Andean volcanic suites which we now summa- Rb Ea
Th U K T a Nb La Ce Sr Nd P HI Zr Sm Ti Tb Y Yb
rize before returning to the more complex question of
(a)
intrusive magmagenesis.
0 L013

Andean subduction-related W L034

volcanism
Active Andean calc-alkaline volcanism occurs in a
northern volcanic zone (NVZ; 5"N-2"S), acentral
volcanic zone (CVZ; 16-28"s) and a southern volcanic
zone (SVZ; 33-52"s) (Thorpe & Francis 1979; Thorpe
et al. 1984). Lavas of the NVZ and SVZ are
dominantly basaltic andesites and
andesites (SiOz
range 50-65%) thoughtto have beenproduced by
fractional crystallization of basaltic parental magmas
that formed by partialmelting of mantlematerial
enriched in subduction zone components. In contrast, Rb Ba Th U K Ta Nb La C e Sr Nd P Hf Zr Sm TI Tb Y Yb
lavas from theCVZare dominantlyandesites and
dacites with siliceous ignimbritesheets (SiOz range (b)
56-75%); isotopic data indicate involvement of an FIG. 3. Primordial mantle-normalizedtraceele-
enriched mantle or crustal source in their magmagene- ment patterns for Andean lavas (normalizing
sis. Bearing in mind that the continental crust is much values from Wood 1979). (a) Range for NVZ
thicker (c. 50-70 km: Cummings & Schiller 1971) (stipple; basaltic andesites and andesites n = 5,
beneath the CVZ than the NVZ and SVZ, and that Si02 = 56-58%) andSVZ (basalts and basaltic
isotopically-mature basement underlies only the CVZ andesites, n = 12, SiO, = 50-55%). B-l = alkali
basalt, SW Bolivia; 19P = alkali basalt, NW
(Fig. 2a), this isotopic enrichment is usually taken to be Argentina.(b)
Cerro
Galan basaltic andesites
of crustal origin (cf. Francis et al. 1980; DePaolo (L013 and L034); the vertical barsindicatethe
1981b). To completethepicture, Pliocene-Recent range for twelve samples. The stippled and lined
alkaline basalt-trachyte volcanic centres occur 30-300 areas correspond respectively to the NVZ and SVZ
km to the E of the active calc-alkaline zone along the lavas plotted in (a). (From Thorpe et al. 1984, fig.
length of the Andes from Colombia (Kroonenberg et 3.)
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
418 G. C. Brown et al.
and in Fig. 3b with basaltic andesites from Cerro variation
between the respective suites,e.g. the
Galan (NW Argentina, Thorpe et al. 1984). The NVZ Th-Ta plot in Fig. 4 (cf. Pearce 1982; Thorpe et al.
andSVZ lavas,thoughtfrom trace elementand 1984). Starting from primordial mantle (bottom left),
isotopic datato have beenproducedfrommantle bothelements are relatively enriched in within-plate
containingasubduction-zonecomponent (Hawkes- lavas (Fig. 3a) from which we infer a within-plate
worth 1982), are enriched in Rb, Ba, Th,K, Ce and Sr sourcevariation in which the levels of Th and Ta
relative to Ta, Nb, Hf, Zr and Ti. Whilst the alkali remain roughly equal.Incontrast,the subduction
basalts have similar levels of Rb, Ba and Th, they are process selectively enriches Th but not Ta in the
relatively rich in most of the elements from Ta to Tb overlying mantlesource of calc-alkaline volcanics.
in Fig. 3, particularly in Ta and Nb, consistent with Figure 4 shows that the SVZand NVZ lavas may have
their derivation from within-plate mantle east of the formed from a source region that had undergone some
main calc-alkaline volcanic source region (see below). limited Th and Ta enrichment relative to primordial
The main CVZ lavas (for data see Thorpe el al. 1984, mantle, and more important, that a subduction zone
Fig. 3b), which may containacontinental crustal component probably was introduced into both source
component, tend to have higher levels of Rb, Ba, K, regions. The Bolivia-Argentine alkaline lavas plot
La, Ce, Sm and Tb than the NVZ and SVZ lavas, but close to the most Th-Ta-rich end of the within-plate
show no enrichment in the elements believed to be sourcevariation line but they have similar composi-
characteristic of magmas formed in within-plate set- tions, with only a little less Th, to the Cerro Galan
tings such as Ta, Nb and Ti (Pearce 1983). Thus, apart basaltic andesites. Thereare two possible explana-
from normal subduction-zone components derived by tions: first, the magma source region might have had
partial melting of the volatile and LIL-element fluxed similar chemistry to that of the NVZiSVZ lavas, in
mantle wedge, isit possible to recognize trace which case the liquids may have thenundergone
elements sensitive to additional partial melt contribu- fractional crystallization possibly coupled withassi-
tions from (i)within-plate, or alkalic mantlesources milation (AFC), a process which enriches both Th and
selectively enriched in Ta and Nb with respect to Ta in the liquid and drives it along a path (in Fig. 4)
normal mantle and subduction-zone components, and between the sourcevariation lines for within-plate
(ii) continental crust which carries higher concentra- mantle and subduction-relatedmantledepending on
tions of the most LIL elements than the subduction- the assimilatiodcrystallization rate. Alternatively, the
zone components. The origin of within-plate enrich- Cerro Galan source may have been richer in Th and
ments relative to normal mantle is ascribed by most Tathanthe NVZISVZ source, in which case less
authorsto mantlemetasomatism in the presence of
C02-dominated fluids which mobilize many of the
HFS elements (see discussion by Pearce 1983; also 10-
Weaver & Tarney 1983). Thus,the mantle more
remote from subductioq
a zone is considered to
contribute ahigher'within-plate'HFSiLILelement
Th(ppm)
ratio to partial melts than that immediately over!ying
the subduction zone. S.Chile ?
Against this background wherebysubductiow
related mantle, within-platemantle and continental 1-
crustal contributions to magmas may be distinguished,
itis interesting thatthe lavas from CerroGalan,
200 km E of the main volcanic zone(Francis et al.
Subduction-
1982), appear to include both the latter two compo- r e l a t e ds o u r c e
nents (see Fig. 3b). These lavas have higher concentra- varialion

tions of Ta and Nb (the within-plate mantle compo- ~ _*- Assirnilation-

nent) andparticularly of Rb, K, La andCe (tk 0.1 -
fractional
Crystallization
continental component, also identified from isotopic
i
data,Thorpe et al. 1984) than theNVZ and SVZ I

lavas. Moreover, xenolithscontaining glass with the 0.1 1 Ta(ppm) 10

composition of quartz-feldspar cotectic mixtures found

within the Cerro Galan basaltic andesites (Thorpe et FIG.4. Plot of Th against Ta for Andean lavas.
al. 1984) provideevidence that crustalmeltswere Key: crosses = SVZ lavas (S Chile); open squares
= NVZ lavas (Ecuador); open circles = CVZ lavas
available to 'contaminate' the rising, cooling magmas (N Chile); open triangles = Cerro Galan lavas;
(cf. the assimilation-fractional crystallization process filled circles = alkaline lavas (B-l and 19P); filled
described by DePaolo 1981b). square UC = upper crust; filled square LC = lower
These variations may besummarized using trace crust (crustal data from Taylor & McLennan 1981).
elementdatathat exhibit the greatest degree of (From Thorpe et al. 1984, fig. 6 )
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
Granitoids of contrasting arcs 419
assimilation and fractional crystallization is required in was used outside the limits 66-75% SiO,; as noted
the crust. The position of the CVZ lavas in Fig. 4 is earlier, this is notentirelydesirablebut reflects the
equally ambiguous in that the relative importances of availability of data.
an enriched mantlesource
and AFC cannot be Bearing in mind the provisional nature of Fig. 5 , the
determined for these relatively evolved compositions, information it contains may be used to discuss trends
though both may have been significant. Aproblem with increasing arc maturity (in the sense of categories
posed by Fig. 4 is thatthree groups of processes- i-iii above)for many of the traceelements plotted,
within-plate mantle source
variation,
subduction- trends which are denoted by arrows above the figure.
relatedsourcevariation and AFC-are described in Because we are dealing with silica-rich rocks that may
terms of two elements. Clearly, the lack of a uniquely have undergone considerable crystal fractionation, the
‘crustal’ trace
element
means that
the different interpretation of this diagram is more complicated
components of these processes cannot be modelled thanfor Fig. 3; indeed,the particularly marked
quantitatively, especially where evolved magma com- depletion of Ba,Sr, P and Ti in the alkali-calcic
positions are concerned. Noting the analogous tectonic granitoids of the most mature arcs may be explained
settings, mineralogy and chemical compositions of the by agreaterdegree of fractionation of respectively,
volcanic and intrusive rocks in the Andes and alkali and plagioclase feldspars, apatite and Fe-Ti
elsewhere, we now proceed to considerthesetrace oxide from these liquids. However, in addition to this
element characteristics for some of the intrusive rocks. effect, there is a general increase, with arc maturity, of
elements that are incompatible in both major and most
early crystallizing minor mineralphases at their
Trace elements insubduction- expected abundance levels in granitoids (e.g., Y does
related plutonic suites not become compatible with crystallizing phases until
apatite exceeds c. 2.5% of theseminerals).These
Whilst it has proved relatively straight-forward to include Rb (DRb= c. 0.4 in alkali feldspar), Th, U,
compare the petrological and major element geoche- Ta,Nb,Hf, and Y which clearly increase with arc
mica1 datafor differentsubduction-relatedplutonic maturity, while Zr remains constant or falls slightly,
suites (as earlier in this paper), similar comparisons for possibly as a result of zircon fractionation. This might
trace elements are hampered by a lack of comparable reflect increasing AFC as magmas pass through thicker
information, ranging fromincompletetraceelement crust, but noting the wide range in some incompatible
analyses for many suites to a total lack of published elementconcentrations (e.g.,Th, U, Ta,Nb),the
data for others. Although this problem is slowly being gradualincrease of (Ta,Nb)/(K,Rb,La,etc.) with
resolved, in ordertomake provisional comparisons increase in arc maturityand the approach of some
betweendifferentsuites in Fig. 5 , data were taken incompatible elements (e.g., Th, U) to levels in back
from several circum-Pacific batholiths,notablythose ardanorogenic granitoids, we suggest that the progres-
referenced in Fig. 1, but including new trace element sive increases in some incompatible element concen-
data indicated in the caption to Fig. 5 . In compiling trations (e.g., Ta, Nb) may reflect a contribution from
analytical information, three types of granitoid- a within-plate mantle source rich in these elements. In
bearing arc were distinguished: (i) primitive island and particular, the high (Ta,Nb)/Rb ratio of mature
continentalarcs,
broadly the M-type category of relative to primitive arcscannotbeproduced by
calcic, metaluminous granitoids; (ii) normal continen- assimilation or fractionation (Rb fractionation in
tal arcs-these are
abundant I-type calc-alkaline biotite being relatively rare) and requires either that
metaluminous to peraluminoussuites; (iii) mature there is less or no subduction-related LIL-enrichment
continental arcs, which are alkali-calcic, peraluminous, in the mature arc source, or that there is an additional
often two-mica granitoids that are clearly associated Ta-Nb-rich component.
with subduction zones, and often termed S-type (see There is little doubt (cf. Saunders et al. 1980;
earlier discussion). Adequate trace element data are Saunders & Tarney 1982) that the first three strongly
available from young circum-Pacific batholiths to lithophile elements (Rb, Th and U) which, like
distinguish fields into which granites of categories (i) potassium, show strongtransverse variations across
and (ii) tendto fall. However, as there is little arcs with only calcic and calc-alkaline batholiths, are
published information for category (iii) granitoids, at enriched in magmas relative to most other elements to
this stage we quote only an average of selected data as the right of Fig. 5 by addition of a subduction-zone
a single representative line in Fig. 5. Also, for component and probably also, in some cases, also by a
comparison, a typical anorogenic alkaline granite has continental crust component (cf. earlier discussion of
been added to Fig. 5. Finally, because marked changes subduction-enrichment of theseelements in volcanic
in trace element levels result from fractional melting sourceregions). Similar enrichment of LREEs (La,
and crystallization (Hanson 1978; McCarthy & Hasty Ce), which was noted for volcanic rocks, is not clearly
1976), all the analyses used to compile Fig. 5 were observed in the granitoid data (Fig. S ) , perhaps
taken from suites as near to 70% as possible, and none because REE are being fractionated in minor mineral
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
420 G. C . Brown et al.
Rb Ba Th U TaSr Nb P Hf Zr

1QQQ-
Major trends with increasing arc ' maturity'

100-

10-

P
..........
.......... Primitive islandand contijental arcs
1.0-

- I
...............
...............
...............
...............
..............................
..............................
...............
...............
...............
...............
...............
...............
Normal continental
arcs!
I.

Mature continental arcs

(average)
c--Ailsa Craig

0.1- I 1 I I I I I I I I I I I I I I 1
IFit1 Ba Th U K Ta Nb La Ce Sr Nd P Hf Zr Sm Ti Tb Y
FIG.5. Primordial mantle-normalized trace element patterns for granitoids from Mesozoic and Cenozoic magmatic
arcs. Using the terminology of Fig. 1 with additions,the following contribute to the maturearcdata set:NC,
Alaskan tin granites (Hudson & Arth 1983), Korean Jurassic granites (Hong 1983), n = 8 for all elements, 28 for a
few elements; normal continental arcs include data from SN, PCB, NM, Mexico (Bagby er al. 1981) and Korean
Cretaceous granites (Hong 1983), n = 29 for all elements, 65 for a few elements; primitive arcs are based mainly on
data from NB, n = 20. All data are for rocks in the Si02 range 66-75%. Ailsa Craig is a peralkaline anorogenic
granite (data from Potts et al. 1981; Thompson 1982).
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
Granitoids of contrasting
arcs 421
phases. However, ratios such as (Ta, Nb)/La are much 10
higher in the mature arc rocks. Again by analogy with +Mature continental arcs
0 Normal continentalarcs ++ +t
volcanic magma source regions progressive increases X Primitive island arcs
and continental
++ + +
in theHFSelements,Ta,Nb, Hf and Y with arc arcs 0 0
maturity, may reflect acontributionfromenriched Rb/ Zr o + o +
within-plate mantle sources [cf. discussion by Tarney I o *
X
& Saunders (1979), who argued that theseelements
are locked in stable minor mineral phases during the
dehydration of the downgoing slab that leads to X
subductionenrichment]. The levels of these HFS x x
elements in mature continentalarcgranitoids fre-
quently exceed those of average upper crust (Taylor &
X
McLennan 1981) by up to an order of magnitude and
it is difficult to conceive of melting-fractionation 0.1 I 1 I
1 10 10010 11 I
processes involving crustalrocksalone that would Nb Y
produce such resultant high values of elements which
tend to remain in refractorymineralphases during FIG. 6. Plot of RbiZr against Nb and Y for
partial melting under hydrousconditions. The lower granitesfrom Mesozoic and Cenozoic magmatic
Ta, Nb, Hf, Y abundances in primitive arc granitoids arcs. Sources of data as in Fig. 5.
are unlikely to reflect the fractionation of amphiboles,
Fe-Ti oxides etc. unless thesame processes have source variations; both occur throughout the maturing
affected the co-existing basalt-andesite lavas that have process. Bearing in mind the earlier discussions of (i)
similar LIL/HFS element patterns (Fig. 3); the effects crustal contamination in the most thickened and
of such processes are observed in the P, Zr, Ti datain evolved volcanic arcs and (ii) the isotopic evidence for
Fig. 5 . acrustalcomponent in evolved plutonicsuites, it is
In an attempt to make apreliminary distinction probable that subduction-enrichment may occur in the
between plutonic magma types, various plots of Rb, magma source regions of primitive arcs and that
Th and U against Ta,Nb, Hf and Y may be crustal contamination may also occur in themore
interpreted in the same way as for Fig. 4. To discuss mature arcs. However, the crucial point for which we
the effects of fractional crystallization andsource argue using certain HFS element abundances is that as
composition we use a plot of RbiZr against Nb and Y. arcs becomemore mature and magma generation
Of the likely fractionating phases Rb is incompatible migrates towards regions more remote from an active
in plagioclase, pyroxene and amphibole (DRb < 0.1) trench,
the
greater becomes the influence of a
and alkali feldspar (DRb = c. 0.4) and compatible only within-plate mantle source. Thus,the space-time
in biotite (DRb= c. 2). The increase in Rb in Fig. 5 trends of subduction-related batholiths discussed ear-
partially represents fractional crystallization of these lier and summarized inTable l are, we suggest, a
phases; by contrast, fractional crystallization of zircon reflection of both changes in the mantle source region
may account for the constant value or decrease of Zr andthe previously recognized changes in the input
with arc maturity shown in Fig. 5. Thus,theNb from subduction-enriched mantle and crustal sources.
concentration at a given Rb/Zr might reflect the source Perhaps the most complete cross-section showing
contributionfrom 'within-plate' mantle.Figure 6 thesevariations occurs across the N Chile-Bolivia
shows trends of increasing RbIZr,Nb and Y (for section of theAndes whereplutonic rocks of Cre-
mature arcs), and increasing Nb at given RbiZr with taceous and Tertiary age include the medium-K
increasing arc maturity. The absoluteincrease may calc-alkaline monzonite-dominated batholith in the W
imply asourcedifference(Nbvariation implies a (Aguirre et al. 1974) followed by the high-K alkali-
difference in degree of perfect fractional crystallization calcic peraluminous two-mica granitoids of central
exceeding a factor of 10), and the tendency for the Nb Bolivia (Grant et al. 1979) which, in turn, give way to
concentration at a given Rb/Zrto increase with back-arc
a zone comprising peralkalinenepheline
maturity suggests the involvement of 'within-plate' syenites andriebeckitegranites with associated car-
mantle. It is clear that for certain ranges of values, this bonatites of the Velasco Province in eastern Bolivia
diagram could assist in determining the affinity of a (Darbyshire & Fletcher 1979; Fletcher et al. 1981;
batholithsuitefrom an unknowntectonicenviron- Fletcher & Litherland, 1981). Whilst the central zone
ment. of peraluminousgranitoids may carry the greatest
Notsurprisingly, given the complex mixture of crustal component, we would argue that all these
different processes that have led to the observed trace Cretaceous-Tertiarysuites may be influenced by a
elementcompositions of plutonicrocks, with the W-E change from subduction-enriched
a mantle
present data it is difficult to separate the contributions source to a within-plate mantle source. The addition of
of subductionkrustal-enrichment and within-plate back-ardanorogenic granitoids extends the list of
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
422 G. C . Brown et al.
TABLE1. Majorpetrologicalandgeochemicalvarieties of subduction-relatedgranitoidsuites:apreliminary
classification

Trace element geochemistry

Alkali-lime relationship
Chappell
andWhite and possible nature of
Rock types and location saturation
alumina
definition*
basic-intermediate
and precursors

(1) Gabbro-quartz Calcic, metaluminous M Low LIL

and HFS element abund-
diorite stocks of ances; mantle-derived with limited
primitive island and subduction-zone enrichment
continental arcs
(ii) Diorite-tonalite- Calc-alkaline to alkali- I Intermediate
Rb,
Th, U values due
monzogranite-granite calcic, metaluminous to mainly to subduction-zone enrich-
batholiths of normal
peraluminous ment; some Nb, Ta, Hf, Y enrich-
continental arcs ment relative to normal mantle
and subduction-related magmas
where within-plate mantle compo-
nents are present; mainly mantle-
derived precursors
(iii) Granodiorite
and two- Alkali-calcic, peralumi- ‘S’? High Th,
Rb, U values probably
mica granites or nous, becoming peral- reflecting some crustal comtamina-
quartz latites of kaline in low-Ca, high-K but
tion, largely thought
thebe to
product mature
suites continental of within-plate
mantle
arcs sources from which high levels of
Nb, Ta, Hf and Y were also
derived
(iv) Nepheline
syenite to Alkaline to alkali calcic, Dominated
A by within-plate
mantle
alkali granite (with enriched
signature,
peralkaline in HFS ele-
s riebeckite) aegirine, with suites above
of back arc zones

* See also definitions in Pitcher (1983).

subduction-related intrusive suites to include the four District intrusions at Shap and Skiddaw) are plotted in
varieties summarized in Table 1. Fig. 7a and of three granitoids from the northern
Caledonides (the closely associated eastern Highlands
intrusions at Cairngorm,Ballaterand Mt Battock)
Trace elements in some late appear in Fig. 7b. It is clear that the southern
British Caledonian granites intrusions are, on average, less mature (in the sense
used earlier) than their northern counterparts. In
An interesting application of the
trace element particular, the Shap and Loch Doon intrusions follow
approach to subduction-relatedgranite classification the trend of normal continental arcs (Fig. 5) whilst the
arises from similar examination of the Siluro- Skiddaw data (admittedly from slightly more evolved
Devonianintrusions of N andcentralBritain(de- silica-rich compositions) lie on the transition to mature
scribed in detail by Brown et al. 1981; Pankhurst & continentalarcs. Incontrast,the eastern Highlands
Sutherland 1982). Whilst it is agreed that these intrusions, especially BallaterandCairngorm,more
granites represent the final stage of magmatism closely resemble the mature continental arc types with
associated with closure of theIapetusOcean across enhanced levels of both LIL and HFS elements (e.g.,
SouthernScotland,their geochemistry hasbeen dif- the highest levels of Nb, Ta, Hf and Y among the
cult to classify using I and S-type models (cf. Plant et main Caledoniangranitoids occur in the Cairngorm
al. 1983) because the most alkali-calcic, peraluminous intrusion).
granites in the Caledonian area, with the greatest Aplot of Rb/Zr against HFSelements (Fig. 8)
enrichments of LIL elements, were unlikely to have makes the distinction between northern and southern
been generated by crustal anatexis some 80 Ma after suites even more clear and, by reference to Fig. 6,
the radiometrically-dated climax of crustal meta- these granites lie across the boundary from normal to
morphism. Using the terminology in Table l (cf. Figs S mature continental arcs. We suggest, therefore, that
and 6), matters become more simple. separate mantle source regions may have been
Trace element dataon fresh (mainly borehole) involved in the origin of these two groups, a mainly
samples of three granitoidsemplaced in the central subduction-enriched(type iiin Table 1) source for
(Loch Doon)andsouthern Caledonides(the Lake some of the southern intrusions and a source with a
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
Granitoids of contrasting arcs 423
1000.

'I I
o SKIDDAW
X SHAP

v
100 LOCH DOON
W
l-
4

2
5
B0 10

z 0.1
1010 Nb 10'
n Ta
,
Y h MTBATTOCK A CAIRNGORM 8 SKIDDAW
0
A BALLATER d SHAP 0 LOCH DOON
8
1.o FIG. 8. Plot of Rb/Zr against Ta, Nb and Y for
averagecomposition of freshboreholematerial
(except Loch Doon)from Caledoniangranites
from Britain (cf. Fig. 7). The broken line separates
granites from the northern province (triangles) and
the southern province (circles; unpublished data).
0 .1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Ba Th U K To Nb La C. Sr Nd P Hf Zr Sm TI Tb Y greater within-in plate component (transitional to type
(a) iii in Table 1) for the eastern Highland intrusions. A
more detailed analysis (our unpublished geochemical
1000 data) has indicated some N-S increases in HFS
element concentrations across the southern area, but
the data all fall in the normal continental arc area of
YT.BATTOCK
Fig. 6. We suggest that the overall contrast in mantle
source regions inferred from Fig. 8 is more important
0 BALLATER
in determining the whole-rock geochemistry of these
l00
h X CAIRNGORM late Caledonian granitoids than the contrasting nature
2 of the crust into which they were emplaced and which
z was available for assimilationkontamination. (The
2 easternHighlands area is underlain by Archaean
4
5 lithosphere of presumed Lewisian age, whereas there
a0 10 is no evidence for such ancient lithosphere beneath the
z southernprovince; cf. Brown et al. 1981). By
E, comparison with modern intrusive complexes, the
Y inferences suggested above about magma source
X
8 regions are consistent with the greater distance of the
1.o Eastern Highlands granites from the former subduc-
tion zone (c. 200 km)compared with thesouthern
intrusions in Figs 7 and 8, which lie less than 100 km
from the Caledonian suture. Unlike Pitcher (1983) and
Harris et al. (1984) we are not convinced of the need
for a special late post-orogenic category of I-
0. l
(Caledonian) type granitoidsbut suggest that their
geochemical characteristics may reflect a within-plate
(b) mantle component resulting from their generation in a
FIG. 7. Primordialmantle-normalizedtraceele- mature arc location which has a present day analogue
ment patterns for Caledoniangranitoids
from in circum-Pacific subduction-related arcs (cf. mature
Britain. (a) Skiddaw, Shap (northern England) and arcs in Figs 5 and 6).
Loch Doon (southern Scotland); (b) Mt Battock,
Ballater and Cairngorm (eastern
Highlands of Conclusions
Scotland,unpublished data). Average of three
representative analyses selectedfromlarger data 1. The geochemical variation shown by modern
sets for each intrusion; SiOz = 66-75%. subduction-related volcanic suites which range from
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
424 G. C. Brown et al.
tholeiitic (primitive arcs) to calc-alkaline (normal alkalic, peraluminousgranitoids of inner arc zones
island and continental arcs) to alkaline
(back-arc which have been termed 'S-type' may owe their origin,
zones) is reflected in the more complex geochemical and much of their distinctive trace element geochemis-
evolutionandmaturity of calcic, calc-alkaline and try to initial partial melting of mantle with some
low-Ca alkalic (peraluminous to peralkaline) intrusive within-plate characteristics, followed by combined
arcs. This parallel variation exists both with decreasing assimilation and fractional crystallization (AFC) dur-
age of magmatism in one location and, more particu- ing liquid ascent through the continental crust. Their
larly, with increasing distance from the active trench, primary magmas may resemble mixtures between
and it suggests that there are common factors in the normal calc-alkaline liquids and the back-arc peralka-
generation of both intrusiveand extrusive magmas line liquids which are observed in both intrusive and
from laterally heterogeneous mantle. eruptive centres behind modern arcs.
2 . Concentrating on intrusive rocks: even the more 4. The petrological and geochemical variations
silicic magmas of primitive arc suites carry relatively discussed here may be applied to problems of
low levels of LIL (e.g., Rb, Th,U) and HFS (e.g., Ta, magma genesis in more ancient suites. For example, it
Nb,Hf, Y ) elements; those which have undergone is possible that the late (c. 400 Ma old)Caledonian
subduction-zone enrichment andior crustal contamina- granitoids in the eastern Highlands of Scotland may
tion appear to have enhanced levels of LIL elements carry a within-plate mantle source component whilst
despite the effects of crystal fractionation, whereas those of similar age from northern England are more
those in which within-plate mantle sources appear to typical calc-alkaline intrusions, analogous to those of
have been significant carry the highest levels of Ta, normal continental arcs.
Nb, Hf, Y. Thus, as arcs become more mature and/or
magmas are generated beneath regions remote from
an active trench,themore obvious becomes the ACKNOWLEDGMENTS. Theauthors would like to thank A.
influence of within-plate source regions as seen Tindle for unpublished data on the Loch Doon intrusion, M.
through increases in ratios such as (Ta, Nb)/(K, Rb, Thirlwall for a helpful review, J. Taylor and H. Boxall for
La) which are little affected by fractionation in cartographic
work, and
A. Budd and M. Leggett for
typographic assistance. Core material from other Caledonian
mafic-intermediateprecursor magmas. Thesetrans- intrusions was collected and analysed under sub-contract to
verse geochemical variations across intrusive arcs may the British Geological Survey aspart of geothermal
a
reflect the lateralchangeat depth from oceanic project commissioned by the Department of Energy and the
subduction-zone enriched to subcontinentallitho- Commission of theEuropean Community; permission to
spheric mantle of within-plate character. publish the result of this work is herebyacknowledged.
3. This raises the serious possibility that many Fieldwork in the Andes was supported by the NERC.

References
AGUIRRE, L., CHARRIER, R . , DAVIDSON, J . , MPODOZIS,A . , long-standing subvolcanic regime. Can. J . Earth Sci. 13,
RIVANO,S., THIELE,R.,TIDY,E.,VERGARA,M. & 1020-30.
VINCENTE, J-C. 1974. Andean magmatism: its CHAPPELL, B. W. & WHITE,A. J . R. 1974. Two contrasting
palaeogeographic and structuralsetting in thecentral granite types. Pacific Geol. 8, 173-4.
part (30-35") of the southernAndes. Pacific Geol. 8, CUMMINGS, D. & SCHILLER, G. I. 1971. Isopach map of the
1-38. Earth's crust. Earth Sci. Rev. 7, 97-125.
BAGBY,W. C.,CAMERON, K. L. & CAMERON,M. 1981. DARBYSHIRE, D. P.F. & FLETCHER,C. J. N. 1979. A
Contrasting evolution of calc-alkalic volcanic and pluto- Mesozoic alkaline province in eastern Bolivia. Geology
nic rocks of Western Chihuahua, Mexico. J . Geophys. 7,545-8.
Res. 86, 10402-10. DEPAOLO,D. J . 1981a. A neodymium and strontium isotopic
BECKINSALE, R. D. 1979. Granite magmatism in the tin belt study of the Mesozoic calc-alkaline granitic batholiths of
of south-east Asia. In: ATHERTON M. P. and TARNEYJ . the Sierra Nevada and Peninsula Ranges, California. J .
(eds). Origin of Granite Batholiths, pp 34-44. Shiva Pub. Geophys. Res. 86, 10470-88.
Ltd., Orpington, Kent. -19816. Trace element and isotopic effects of combined
BROWN,G . C. 1982. Calc-alkaline intrusive rocks: their wall-rock assimilation and fractional crystallization.
diversity, evolution, and relationto volcanic arcs. In: Earth Planet. Sci. Lett. 53, 189-202.
THORPE,R. S . (ed.). Orogenic Andesites and Related DERUELLE,B. 1982. Petrology of the Plio-Quaternary
Rocks, pp. 437-61. Wiley, London, volcanism of the south-central and meridional Andes. J .
-, CASSIDY, J . , LOCKE, C. A . , PLANT,J. A. &k SIMPSON, Volc. geothermal Res. 14, 77-124.
P. R. 1981. Caledonian plutonism in Britain: a summary. DOMENICK,M.A., KISTLER,R. W., DODGE, F. C.W. &
J . Geophys. Res. 86, 10502-14. TATSUMOTO, M. 1983. Nd and Sr isotopic study of
BUSSELL,M. A., PITCHER,W. S. & WILSON,P. A. 1976. crustal and mantle inclusions from the Sierra Nevada
Ringcomplexes of thePeruviancoastalbatholith:a and implications for batholithpetrogenesis. Bull. geol.
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
Granitoids
arcs
of contrasting 425
Soc. A m . 94, 713-9. 407-20.
FLETCHER,
C. J. N., APPLETON,J . D . & WEBB,B. C. 1981. M c N u n , R . H., CROCKET,J. H., CLARK, A. H.,CAELLES,
Mineralization in the Cerro Manomo carbonatite com- J. C., FARRAR,E. & HAYNES,S. J. 1975. Initial "Sri8'Sr
plex, eastern Bolivia. Trans.Instn. Min. Metall. 90, ratios of plutonic and volcanic rocks from the central
B37-50. Andes between latitudes 26" and 29" South. Earth
- & LITHERLAND,M. 1981. The geology and tectonic Planet. Sci. Lett. 27, 305-13.
setting of the Velasco Alkaline Province, eastern MILLER,C. F. & BRADFISH, L. J., 1980. An inner cordilleran
Bolivia. J. geol. Soc. London, 138, 541-8. belt of muscovite-bearing plutons. Geology, 8, 412-6.
FRANCIS, P. W., BAKER, M. C. W. & HALLS,C. 1981. The PANKHURST, R. J. & SUTHERLAND, S. D. 1982. Caledonian
Kari Kari caldera, Bolivia, and the Cetro Rico Stock. J . granites and diorite of Scotland and Ireland. In:
Volc. geothermal Res. 10, 113-24. SUTHERLAND, S. D. (ed). IgneousRocks of theBritish
-, O'CALLAGHAN, L., KRETZSCHMAR, G., THORPE,R. S . , Isles, pp. 149-90. Wiley, London.
SPARKS,R. J. S . , PAGE,N., DE BARRIO,R. E., GILLOU, PEARCE, J. A. 1982. Trace element characteristics of lavas
G. & GONZALES,0 . E. 1982. The Cerro Galan fromdestructiveplateboundaries. In: THORPE,R. S.
ignimbrite. Nature (London), 301, 51-3. (ed). Orogenic Andesites andRelated Rocks, pp. 525-48.
-, THORPE,R. S . , MOORBATH,S., KRETZSCHMAR, G. & Wiley, London.
HAMMILL,M. 1980. Strontium isotope
evidencefor -1983. Role of the sub-continental lithosphere in magma
crustal contamination of calc-alkaline rocks from Cerro genesis at active continental margins. In: HAWKES-
Galan, northwest Argentina. Earth Planet Sci. Lett. 48, WORTH, C. J. & NORRY, M.J. (eds). Continental Basalts
257-67. and MantleXenoliths, pp. 230-49. Shiva, Nantwich,
GRANT,J. N., HALLS,C., SALINAS, W. A. & SNELLING, N.J. Cheshire.
1979. K-Ar ages of igneous rocks and mineralization in PETRO, W. L., VOGEL, T. A. & WILBAND,J. T. 1979.
part of the Bolivian tin belt. Econ. Geol. 74, 838-51. Major-element chemistry of plutonic rock suites from
HANSON, G.N. 1978. The application of traceelements to compressional and extensional plate boundaries. Chem.
thepetrogenesis of igneousrocks of granitic composi- Geol. 26, 217-35.
tion. Earth Planet. Sci. Lett. 38, 26-43. PICHLER,H. & WEYL, R. 1975. Magmatism and crustal
HARRIS, N. W. B., PEARCE,J. A. & TINDLE, A. G. 1984. evolution in Costa Rica (Central America). Geol.
Geochemical characteristics of collision magmatism. In: Rundschau. 64, 457-75.
SHACKLETON, R.M., RIES, A. C. & COWARD,M.P, PITCHER,W. S. 1983. Granite Type and TectonicEnviron-
(eds). Collision tectonics. Geol. Soc. London Spec. Publ. ment. In: Hsu, K. (ed.). Mountain Building Process, pp.
(in press). 19-40. Academic Press, London.
HAWKESWORTH, C. J. 1982. Isotope characteristics of PLANT,J. A., SIMPSON, P. R., GREEN, P. M,,WATSON,J. V.
magmas erupted alongdestructiveplate margins. In: & FOWLER, M. B. 1983. Metalliferousand mineralized
THORPE,R . S. (ed). OrogenicAndesites andRelated Caledonian granites in relation to regional metamorph-
Rocks, pp. 549-71. Wiley, London. ism and fracture systems in northern Soctland. Trans.
HOLE,M,, SAUNDERS, A. D . , MARRINER, G. F. & TARNEY, Inst. Min. Metall. 92, B33-42.
J. 1984. Subduction of pelagic sediments: implications P o n s , P. J., THORPE, 0. W. & WATSON,J. S. 1981.
for the origin of Ce-anomalous basalts from the Mariana Determination of the rare-earth element abundances in
Islands. J. geol. Soc. London, 141, 453-72. 29 international rock standards by instrumental neutron
HONG,Y. K. 1983. Petrology and geochemistry of Jurassic activation analysis: a critical appraisal of calibration
andCretaceousgranites, South Korea. Unpublished errors. Chem. Geol. 34, 331-52.
PhD Thesis, University of London, 330pp. SAUNDERS, A. D. & TARNEY, J. 1982. Igneous activity in the
HUDSON,T. & ARTH,J. G . 1983. Tin granites of the Seward SouthernAndesandNorthern AntarcticPeninsula:a
Pensinula Alaska. Bull. geol. Soc. A m . 94, 768-90. review. J . geol. Soc. London, 139, 691-700.
KESLER,S. E. 1978. Metallogenesis in the Caribbean region. -, -& WEAVER,S. D. 1980. Transverse geochemical
J . geol. Soc. London, 135, 429-41. variations across the Antarctic Peninsula: implications
KROONENBERG, S. B., PICHLER, H . & DIEDERIX, H. 1982. for the genesis of calc-alkaline magmas. EarthPlanet.
Cenozoicalkalibasaltic to untrabasic volcanism in the Sci. Lett. 46, 344-60.
uppermostMagdalenaValley, Southern Huila Depart- SCHARER,U. & ALL~GRE, C.J. 1983. The Palung granite
ment, Colombia. Geol. Norandina ( 5 May), 19-26. (Himalaya);high-resolution U-Pb systematics in zircon
MARRINER, G. F. & MILLWARD, D. 1984. The petrology and and monazite. Earth Planet. Sci. Lett. 63, 423-32.
chemistry of Cretaceous Recent
to volcanism in STEWART,J. W. 1971. Neogeneparalkaline igneous activity
Colombia: the magmatic history of an accretionary plate in eastern Peru. Bull. geol. Soc. A m . 82, 2307-12.
margin. J . geol. Soc. London, 141, 473-86. TAKAHASHI, M,.
ARAMAKI, S. & ISHIHARA,S . 1980.
MASON,D. R. & MCDONALD,J. A. 1978. Intrusive rocks Magnetiteseriesiilmeniteseries vs. I-typeis-type grani-
and porphyry copper occurrences of thePapua New toids. In: ISHIHARA,S. & TAKENOUCHI (eds). Mining
Guinea-Solomon Islands region: a reconnaissance study. Geol. Japan Spec. Iss. 8, 13-28.
Econ. Geol. 73, 857-77. TARNEY, J. & SAUNDERS,A . D. 1979. Trace element
MCCARTHY,T. S. & HASTY, R.A. 1976. Trace element constraints on the origin of cordilleranbatholiths. In:
distribution patterns and their
relationship to
the ATHERTON, M. P. & TARNEY, J. (eds). Origin of Granite
crystallization of graniticmelts. Geochim.Cosmochim. Batholiths-Geochernical Evidence, pp. 90-105. Shiva
Acta. 40, 1351-8. Publ. Ltd., Orpington, Kent.
MCCOURT,W. J. 1981. The geochemistry and petrography of TAYLOR, H.P. & SILVER,L.T. 1978. Oxygen isotope
the Coastal Batholith of Peru. J . geol. Soc. London, 138, relationships in plutonic igneous rocks of the Peninsular
 Downloaded from http://jgs.lyellcollection.org/ by guest on May 7, 2014
426 G. C. Brown et al.
Ranges batholith, Southern and Baja California. USGS rocks. Phil. Tram. R. Soc. Lond. (in press).
Open-file report No. 78-701. VIDAL, P.H.,
COCHERIE, A. & LE FORT, P. 1982.
TAYLOR, S . R. & MCLENNAN, S. M. 1981. The composition Geochemical investigations of the origin of the Manaslu
and evolution of the continental
crust: rare
earth leucogranite(Himalaya,Nepal). Geochim.Cosrnochim.
element evidencefromsedimentaryrocks. Phil. Tram. Acta. 46, 2279-92.
R. Soc. Lond. 301A, 381-99. WEAVER,B. L. & TARNEY,J. 1983. Chemistry of the
THOMPSON, R. N. 1982. Magmatism of the British Tertiary sub-continentalmantle:inferences from Archaean and
Volcanic Province. Scott. J. Geol. 18, 49-107. Proterozoicdykesandcontinental flood basalts. In:
THORPE,R. S. & FRANCIS,P. W. 1979. Variations in Andean HAWKESWORTH, C. J. & NORRY, M.J. (eds). Continen-
andesite
compositions and their
petrogenetic signi- talBaraltsand MantleXenoliths, pp. 209-29. Shiva,
ficance. Tectonophysics, 57, 53-70. Nantwich, Cheshire.
- - & O'CALLAGHAN,L. 1984. Relative roles of WOOD, D. A. 1979. A variably-veined sub-oceanic upper
source composition, fractional crystallization and crustal mantle: genetic significance for mid-ocean ridge basalts
contaminationinthe petrogenesis of Andean volcanic from geochemical evidence. Geology, 7, 499-503.

Received 11 September 1983; revised typescript accepted 6 January 1984.

G . C. BROWN,R. S . THORPE& P. C. WEBB, Departmentof Earth Sciences, The
Open University, Milton Keynes MK7 6AA.