Myron G. Best. Igneous and metamorphic 2003 Blackwell Science

Подождите немного. Документ загружается.

Magmatic Petrotectonic Associations

387

 alkali feldspar (22%)  sodic plagioclase (35%;

10–20

) and either muscovite (9%)  biotite (2%) 

tourmaline or tourmaline  muscovite (e.g., France-

Lanord and Le Fort, 1988). Mn-Fe garnet, zircon,

monazite, and apatite are common accessory minerals.

The two-mica leucogranites commonly experienced

subsolidus metamorphic deformation, whereas the

apparently later, cross-cutting tourmaline-muscovite

granites are not foliated. Leucogranites are under-

lain by a thick sequence of highly metamorphosed sed-

imentary and magmatic rocks (quartz-feldspar-mica-

sillimanite gneisses and migmatites) of Precambrian to

lower Paleozoic age and are overlain by another thick

sequence of metamorphosed sedimentary rocks.

There is general agreement that the peraluminous

leucogranite magmas are near-minimum-T (Figures

5.24 and 5.25) partial melts of metamorphosed mica-

ceous sedimentary rocks. No mantle component is

present. The following justify this consensus:

1. Isotope ratios of granites (

Sr/

Sr  0.73–0.83;

143

Nd/

144

Nd  0.5121–0.5118; 

O  9%o–

14%o) that resemble those in the underlying meta-

sedimentary quartzofeldspathic mica gneisses.

2. Lack of spatially and temporally associated mag-

matic rocks that have a mantle component.

3. Similarity to experimentally generated partial melts

of quartz-feldspar-mica source rocks at water-

undersaturated conditions (dehydration melting;

Table 13.9): Under water-saturated conditions, par-

tial melts dissolve larger concentrations of sodic

plagioclase from source rocks, and these high-

Na/K, Sr/Rb, and Sr/Ba trondhjemitic melts are un-

like Himalayan leucogranites. Nonetheless, some

variations in water fugacity and in source rock com-

position are indicated by the somewhat variable

tourmaline/biotite ratio. Tourmaline is stabilized

under low water concentrations and biotite under

high. As for potential Himalayan source rock, meta-

morphosed graywackes have lower B (16 ppm),

whereas metamorphosed shales have higher B

(100 ppm), compared to 225 ppm B in average

leucogranite. Either could be the source, depending

on bulk distribution coefﬁcients and degree of

melting.

Anatexis could have been triggered by release of

aqueous ﬂuids from heated footwall rocks of the Lesser

Himalayas as the hotter gneisses were emplaced over

them along the Main Central Thrust (Figure 13.33);

this would likely have been water-saturated melting.

On the other hand, if water-undersaturated (dehydra-

tion) partial melting prevailed, it may have been caused

by shear heating along the thrust (e.g., Nabelek and

Liu, 1999) or by decompression of already hot, thick

crust (Section 11.1.1) as the High Himalaya was tec-

tonically unroofed.

13.9 ANOROGENIC A-TYPE

FELSIC ROCKS

The anorogenic tectonic regime in which A-type rocks

occur is noncompressional. Included are oceanic is-

lands, generally near spreading ridges (Réunion, As-

cension, Azores, Canary); apparently stable cratons;

Table 13.9. Average major and trace element compo-

sition of Himalayan S-type leucogranites compared

to range of partial melt compositions generated

experimentally by dehydration melting of quartz 

plagioclase  muscovite  garnet  biotite source

rock at 6–10 kbar and 750–900C

SiO

73.64 73.68–75.60

TiO

0.10 00.06–00.29

14.87 14.95–16.17

0.83

FeO 0.47 00.73–01.08t

MnO 0.03 00.02–00.07

MgO 0.11 00.17–00.39

CaO 0.47 00.42–01.15

O 4.05 03.07–04.92

O 4.56 03.40–05.19

0.13

Total 99.27

B 225

Cl 35

Co 96

S61

Rb 286

Ba 213

W 4.4

Th 6

La 11.5

Ce 19

Sr 76

Nd 8

F 1020

Sm 2

Eu 0.5

Sn 19

Gd 2

Dy 2.4

Li 170

Y 14.5

Er 1

Yb 1

Lu 0.1

1, Average Himalayan leucogranite. Data from France-Lanord and

Le Fort (1988). 2, Range of experimental partial melts. Data from

Patiño Douce and Harris (1998).

and intraplate (within-plate) extensional continental

rifts (discussed further in Section 13.11). Some A-type

granite occurrences are said to be postcollisional or

postorogenic, but these categories refer to a time frame

rather than a speciﬁc tectonic regime. Examples of

A-type rocks include many mid-Proterozoic (1.4–

1.1 Ga) granites all over the world, Permian peralkaline

trachytes (“rhomb porphyries”) in the Oslo graben of

Norway, Jurassic White Mountains syenite-granite in-

trusions (including some deeply eroded caldera ring

complexes) of New Hampshire, and many late Ceno-

zoic silicic volcanic rocks in the western United

States.

A-type felsic rocks were only recognized in the last

two decades of the twentieth century as a distinct class

that is different from other felsic rocks. In addition to

appearing in anorogenic tectonic settings, magmas are

more alkali-rich and so crystallize more abundant alkali

feldspar; some are relatively anhydrous, compared with

the generally wetter, more calcic magmas in continental

arcs (Pitcher, 1997).

13.9.1 Characteristics

Chemically, A-type rocks have many distinctive attrib-

utes. With respect to arc felsic rocks, they have moder-

ately high total alkalies; high K/Na, (K  Na)/Al,

Fe/Mg, Ga/Al; halogens (F, Cl); high-ﬁeld-strength

(HFS) cations (Zr, Nb, Ta, Zn); and REEs (except Eu)

but low Ca, Ba, Cr, Co, Ni, Sc, Eu, and Sr. High con-

centrations of incompatible elements form abundant

accessory minerals, including zircon (in wholly crys-

talline rocks) and, in biotite granites, allanite. Cas-

siterite is found in some rocks. F combines with incom-

patible elements in ﬂuorapatite, pyrochlore, cryolite, ﬂu-

orite, topaz, and astrophyllite. In some occurrences

these minerals are in sufﬁcient quantity to warrant

mining. Disruption of Si-O and Al-O polymers in

the atomic structure of alkali-rich silicate melts allows

for greater solubility of HFS cations through ionic

complexing. Though alkalies are highly concentrated,

so too is silica; consequently, there are no silica-

undersaturated rocks containing feldspathoids. Molec-

ular proportions of K, Na, Ca, and Al are close to the

discriminating ratios for metaluminous, peraluminous,

and peralkaline rocks (Section 2.4.4); consequently,

both peralkaline rocks that contain alkaline amphi-

boles and peraluminous rhyolites that contain topaz

(Al

SiO

(OH,F)

) are A-type felsic rocks (Table 13.10).

Magma temperatures are elevated, commonly

900°C, compared to 800°C for arc magmas. Low

water fugacity is typical, so that hypersolvus perthitic

feldspars (Section 5.5.3) are common in granitoids and

anhydrous maﬁc minerals are more widespread. One

discontinuous reaction series in which crystallization of

hydrous maﬁc minerals is limited to near-solidus residual

melts is fayalitic olivine-hedenbergite (clinopyroxene)-

ferrohastingsite (amphibole)-annite (Fe-biotite). Other

alkaline magmas might crystallize riebeckite-arfved-

sonite (Na-Fe amphiboles) and aegirine (Na-Fe clinopy-

roxene) (Appendix A). Unlike ternary-minimum (Fig-

ures 5.24–5.26) collision and arc granitic rocks, A-type

granitoids include alkali-feldspar-rich syenite, quartz

syenite, and alkali-feldspar granite (Figure 2.8). These

phaneritic rocks are commonly linked in space, time,

and genesis to extrusive rocks of similar composition,

388 Igneous and Metamorphic Petrology

Table 13.10. Chemical Composition of Average

Peralkaline and Topaz Rhyolites

123

SiO

74.0 71.2 75.6

TiO

0.21 0.37 0.14

11.6 9.11 12.8

1.25 2.38 1.12t

FeO 1.88 4.52

MnO 0.08 0.21 0.06

MgO 0.04 0.09 0.15

CaO 0.36 0.45 0.83

O 5.35 6.44 3.73

O 4.46 4.40 5.04

0.02 0.05 0.00

F 0.37 0.30 0.33

Cl 0.24 0.28 0.06

ASI

0.81 0.563 0.98

Cs 2.8 11.3

Rb 140 423

Ba 41

Th 17.5 54.8

U 4.8 21.6

Nb 72 53

Ta 3.33 5.6

La 82 39

Ce 169 88

Sr 28

Nd 68 39

Sm 11.4 6.6

Zr 660 129

Eu 0.13 0.3

Sn 30

Tb 1.46 1.3

Li 50

Yb 6.0 8

Lu 0.77 1.2

Compare with average rhyolite in Table 2.2.

1, Peralkaline rhyolite. (Average comendite, major elements from

Macdonald, 1974; typical trace elements from Mahood, 1981.)

2, Peralkaline rhyolite (average pantellerite). (Data from Macdonald,

1974.)

3, Average topaz rhyolite, Thomas Range, Utah. Data from Chris-

tiansen et al. (1986).

Alumina saturation index (ASI)  molecular Al

/(K

O  Na

 CaO).

Magmatic Petrotectonic Associations

389

especially trachytes and rhyolites. In fact, the plutonic-

volcanic link is as clear as in any magmatic association

(Figure 13.36).

A recurring association of A-type rocks with tholei-

itic and mildly alkaline more maﬁc rocks is petrogenet-

ically signiﬁcant. These maﬁc rock types include basalt,

trachyandesite, hawaiite, mugearite, and benmoreite or

their intrusive phaneritic equivalents. Proportions of

the maﬁc and felsic rock types vary. In some locales, a

distinct compositional gap exists between them, result-

ing in a bimodal association of basalt-rhyolite or basalt-

trachyte. Some intrusive associations include anortho-

site and granite, especially in Proterozoic terranes.

13.9.2 Petrogenesis

A-type magmas appear to be polygenetic: No single

process creates all of them.

Fractionation of calcic, Al-rich plagioclase from

parental, slightly alkaline basaltic magma has been ad-

vocated for production of peralkaline residual magmas.

However, Rb and Cs are not as high as expected in

every case, and crystallization of calcic plagioclase

would be more likely in a hydrous magma than in the

typical low-water-fugacity A-type magmas. The compo-

sitional gap (bimodality) between felsic and maﬁc

rocks in many locales is also difﬁcult to reconcile with

fractionation, which would tend to yield a continuum

of intermediate-composition daughter magmas.

A compilation of Y/Nb and Yb/Ta ratios (Figure

13.34) indicates the presence of both mantle and con-

tinental crust components in A-type felsic rocks. Many

fall in the ﬁeld of oceanic island basalts, and others lie

on a trend toward average continental crust. Still oth-

ers fall between island arc basalt and continental crust.

The mantle component could be derived from

basalt magmas having an oceanic island, plumelike sig-

nature that were lodged in the lower continental crust

or underplated beneath it. During prolonged magma-

tism, still hot basalts intruded early in the activity could

be partially melted by later intruded basalt magmas. In

the Yellowstone Plateau volcanic ﬁeld, Wyoming (Fig-

ure 13.29), about 6000 km

of high-silica rhyolite was

extruded since 2.2 Ma along with only 100 km

tholeiitic basalt lavas. Exposures of Archean rocks pe-

ripheral to the volcanic ﬁeld and presumed to consti-

tute the deep crust beneath it are far more radiogenic

(

Sr/

Sr  0.858) than the rhyolites (0.710), thus

precluding an origin by crustal anatexis (Hildreth et al.,

1991). However, rhyolite magmas may have been gen-

erated as partial melts of voluminous earlier intruded

basalts that were variably fractionated to Fe-rich

daughters (e.g., ferrodiorite) and only slightly contam-

inated with continental crust. Late Cenozoic volcanism

and accompanying crustal attenuation in the Snake

River Plain–Yellowstone volcanic ﬁeld are believed to

reﬂect migration of the North American plate over a

mantle plume, possibly the one that created the Co-

lumbia River ﬂood basalt plateau to the west.

Frost and Frost (1997) point out that the Snake

River Plain–Yellowstone rhyolites closely resemble ra-

pakivi granites that are widespread in mid-Proterozoic

terranes. Both can be considered A-type end members

in having high K

O 5 wt.% at 70 wt.% SiO

, K/Na

 1, and Fe/(Mg  Fe)  0.9. Low water fugacity is in-

dicated by the occurrence of fayalitic olivine  Fe-rich

pyroxenes and near-solidus precipitation of amphibole

and biotite, if any. Oxygen fugacity is signiﬁcantly

lower than that of the typically more oxidized, mag-

netite-bearing arc granites.

Anatectic melts of continental crust heated by intru-

sive maﬁc magma from the mantle are believed by

some petrologists to be an important component in

many A-type magmas. For example, Collins et al.

(1982) postulate magma generation from source rocks

that were subjected to previous dehydration melting,

leaving a residuum containing F- and Cl-enriched

biotite and amphibole. However, other petrologists

believe that the crystalline residue after dehydration

melting is incapable of yielding additional felsic partial

13.34 Y/Nb-Yb/Ta ratios in A-type felsic rocks shown by light shad-

ing. Dark shaded area is Jurassic A-type granitic, syenitic, and

minor more maﬁc rocks in the White Mountains (WM) of

New Hampshire. Compositions of rocks from various tectonic

settings indicated by dashed line ﬁelds. Element ratios, which

are not strongly inﬂuenced by differentiation processes, serve

as indicators of magma source. Oceanic island basalt or its

mantle source appears to be a major component in A-type

rocks along with continental crust. (Redrawn from Eby, 1990.)

felsic

rocks

A-type

Oceanic

island

basalts

Arc

granites

Average

continental

crust

Island arc

basalts

Continent-

continent

collision

0.2

1100.1

390 Igneous and Metamorphic Petrology

melts. They instead propose, as did T.F.W. Barth in the

mid-1900s for the Oslo, Norway, rocks, dehydration

melting of tonalite or granodiorite, which could have

been produced in an immediately preceding, or in an

ancient, episode of arc magmatism. Patiño Douce

(1997) ﬁnds that high-T anatexis in the shallow crust (P

less than or equal to 4 kbar) of arc source rocks con-

taining only limited amounts of biotite and amphibole

yields melts that have major-element compositions sim-

ilar to that of A-type granitoids. Because the residue is

mostly calcic plagioclase and orthopyroxene, partial

melts contain virtually all of the F and Cl from the

source and have high Ga/Al (Ga is incompatible in the

residual minerals), Fe/Mg, K/Na, and (Na  K)/Ca

but low compatible Ca, Sr, and Eu. Low Ni, Cr, and Co

in the melts reﬂect low concentrations in the source.

However, the partial melts of arc rocks would have a

negative Nb-Ta anomaly and Rb/Nb would be high—

neither of which is characteristic of A-type felsic rocks.

13.9.3 Anorogenic Ring Complexes in Nigeria

and Niger

The classic A-type rocks lie within a diffuse north-

south belt of ring complexes in Nigeria and Niger in

central Africa (Figure 13.35). An imperfect southward

migration from Silurian to Jurassic together with local

doming might suggest movement of the African plate

over a mantle plume. However, some geologists work-

ing in Africa (e.g., Bowden et al., 1987) believe that the

belt originated as a result of reactivation of Precam-

brian zones of weakness—faults—in the craton. The

reason for the overall southward migration of activity

in the belt for 250 My and the migration of activity in a

particular string of complexes, as well as the magma

heat source, remains uncertain.

Although basaltic rocks occupy only a small propor-

tion of exposures in the ring complexes, the role of

basalt magmas in the evolution of the bimodal maﬁc-

felsic association cannot be overlooked. There are early

basalt lavas and basalt inclusions in granites—some of

which are pillowlike, indicating mingling of basaltic

and granitic magmas—and minor late basalt dikes. The

amazing 65-km-diameter Meugueur-Meugueur cone

sheet consists of troctolite (olivine  plagioclase gab-

bro) and encompasses a 900-km

anorthosite-leuco-

gabbro body intruded by A-type granites and syenites

(Figure 9.24). The petrogenetic linking of basalt,

anorthosite, and granite/syenite seems strong. In other

ring complexes, most precursory basalt magmas may

have lodged deep in the crust, where they promoted

generation of felsic magmas, some of which ascended

high enough to erupt along ring fractures and vent as

ash ﬂows (Figure 13.36). The collapsed calderas out-

lined by the ring dikes partially ﬁlled with ignimbrite.

Late biotite granite magmas rose into the volcanic

roots.

13.10 GRANITES AND GRANITES

It should be obvious by now that a variety of granitic

rocks develop in several different tectonic settings, in-

cluding the following:

1. Granophyric plagiogranite in ophiolite sequences

(Section 13.6.1); most (all?) sequences appear to

have formed in arc settings.

2. Granophyre in differentiated basalt sills (Section

12.4.1) and layered intrusions (Section 12.4.2),

both likely associated with mantle plumes.

3. Dioritic differentiates in island arcs.

4. Granodiorite and lesser associated gabbro, tonalite,

granite, and dioritic rock in huge batholiths of con-

tinental arcs (Section 13.6.2). Both I- and S-types

are represented.

5. S-type leucogranite in continent-continent collision

zones (Section 13.8).

6. A-type granite and syenite in anorogenic regimes

(13.9); these are especially signiﬁcant in the mid-

Proterozoic.

7. A chieﬂy Archean association of tonalite-trond-

hjemite-granodiorite and generally younger Archean

and Proterozoic granitic rocks (Chapter 19).

The ﬁrst three occurrences have only minor volumes of

granitic rock, but locally large volumes of leucogranite

occur in continent-continent collision zones. The

largest volumes of granitic rock are found in continen-

tal arcs and in the Precambrian occurrences.

Since Read (1957) and other early workers wrote of

“granites and granites,” petrologists have considered

and debated granites from several different perspec-

tives—all intending to cast some light on understand-

ing the origin of these multifaceted, polygenetic rock

masses (e.g., Clarke, 1992; Pitcher, 1997). In addition

to tectonic setting, granites have been categorized with

respect to

1. Time of emplacement relative to regional deforma-

tion or tectonism (pre-, syn-, and posttectonic)

2. Several chemical and isotopic attributes (e.g., per-

alkaline, peraluminous, and metaluminous; radio-

genic and nonradiogenic; high and low 

3. Modal composition (leucogranite, granite, granodi-

orite, and so on)

4. Fabric related to degree of water saturation (hyper-

solvus and subsolvus) or other special conditions of

magma evolution (porphyry, rapakivi)

5. Mechanism of ascent through the crust (diapir ver-

sus dike)

6. Depth of ﬁnal crystallization (for example, stabi-

lized epidote at high P)

7. Source (S- and I-type) and degree of crustal con-

tamination

13.35 Generalized geologic map of the Nigeria-Niger anorogenic intrusive ring complexes and the Cameroon volcanic line in Africa. The ring

complexes are generally younger toward the south, as shown by ages in millions of years. (Redrawn from Rahaman et al., 1984). Dashed

lines in center of map indicate the “grain,” essentially old fault zones, in Precambrian host rocks. There is no monotonic age progression

of magmatic activity along the Cenozoic Cameroon line, where ages in millions of years are shown for intrusive complexes (open trian-

gles) related to extrusive rocks (stippled). The Cameroon line is the only known alkaline province straddling continental and oceanic

lithosphere; alkaline basaltic rocks are no different in chemical or isotopic composition along the line, suggesting derivation from as-

thenosphere underlying the contrasting lithospheres, whereas evolved rocks are peralkaline rhyolite on the continent and phonolite in

the ocean islands. The Benue trough (ﬁlled with Cretaceous sedimentary rocks) and Cameroon line are considered to be a failed rift arm

related to opening of the Atlantic Ocean along the actively rifting Mid-Atlantic Ridge. (Redrawn from Fitton, 1987.)

5°

20°

10° 15° Ε

15°

10°

5°

0°

CAMEROON

BENUE

VOLCANIC

TROUGH

Atlantic

Ocean

Africa

258

213

191

186

151

147

141

162

296

Nigeria

Fig. 13.36

NIGERIA-NIGER ANOROGENIC RING COMPLEXES

323

407

426

407

0 400

2000

miles

Lake

Chad

LINE

Cameroon

Fig. 9.24

Niger

Many petrologists have attempted to characterize

magmatic rocks, particularly granites, in known tec-

tonic setting with respect to their trace element con-

centrations (Figure 13.37). The rationale behind this

characterization is that granitic magmas that have dif-

ferent petrotectonic associations should have contrast-

ing evolutionary paths involving different sources and

melting conditions and subsequent ascent and differ-

entiation histories that together yield distinctive com-

positional signatures. However, certain components of

these complex evolutionary paths may not be unique to

a particular tectonic setting; for example, source rocks

of much the same composition can be partially melted

in continental arcs, rifts, and collision zones. Hence,

the products of evolutionary paths in different tectonic

settings tend to be compositionally “smeared” in most

variation diagrams. This lack of natural discontinuity in

most variation diagrams of magmatic rocks was em-

phasized in Section 2.3.2; discriminating boundary

lines can only delineate most, and seldom all, rocks in

artiﬁcial subdivisions. Nonetheless, such discrimina-

tion diagrams are useful in trying to understand the ori-

gin of granitic magmas and trying to infer tectonic set-

tings for rocks produced before the “modern” period

of plate activity.

13.11 CONTINENTAL RIFT

ASSOCIATIONS: BIMODAL

AND ALKALINE ROCKS

Ancient and active modern continental rifts, of which

there are an estimated 100 worldwide, are generally

elongate sectors of the crust that have experienced ex-

tensional normal faulting. Grabens typically lie in a re-

392 Igneous and Metamorphic Petrology

13.36 The Ningi-Burra ring complex of eroded calderas, Nigeria. See Figure 13.35 for location of this complex, which was emplaced at about

183 Ma. The generalized geologic map (top) and cross section (bottom) show six overlapping intrusive-extrusive centers that migrated

from east to west. They are delineated by ring dikes of peralkaline fayalite  hedenbergite  riebeckite granite porphyry. Numbers on

cross section indicate the center where the pre- and intracaldera deposits of chieﬂy peralkaline rhyolite ignimbrite were derived. (Re-

drawn from Turner and Bowden, 1979.)

Granite porphyry

(fayalite ± heden-

bergite ± riebeckite)

Granophyre

Peralkaline riebeckite

± arfvedsonite

granite

Peraluminous

biotite granite

On map

Syenite

Basalt

and trachyte

On cross-

section

Rhyolite, chiefly ignimbrite:

Intracaldera

Pre-caldera

11°00

9°30

rocks

older

Mostly

0 mile 8

0km12

Mostly

older

rocks

9°00

2 km

gional uplift where the crust and mantle lithosphere are

unusually thin and surface heat ﬂow is high. These at-

tributes are probably the consequence of upwelling hot

mantle, a plume in some instances. Whether upwelling,

plume-related asthenosphere is the cause and diverging

plate motion the consequence (so-called active rifting),

or vice versa (passive rifting), is difﬁcult to ascertain in

every instance. Spreading rates of oceanic rifts (gener-

ally 1–10 cm/y) are one to two orders of magnitude

faster than those of continental rifts: Rhinegraben,

western Germany, 0.01 cm/y since the late Creta-

ceous; East African rift, Ethiopia, 0.06 cm/y; Malawi,

0.03 cm/y; Basin and Range, western North America,

0.5 cm/y since the Miocene.

Some continental rifts evolve into oceanic rifts

accompanying separation of the landmass (Section

13.3.3), generally along two arms of a three-arm rift

geometry. (This three-arm rift geometry is clearly evi-

dent where the Arabian Peninsula has rifted from

northeastern Africa; see Section 13.11.2 and Figure

13.38.) Rifting along the third arm, including some as

old as Precambrian (Wilson, 1989, Figure 11.1), slowly

dies, forming failed rifts. An example of a currently ac-

tive rift destined to “fail” is the East African system de-

scribed later. Others are the northeast-trending Benue

trough in Africa and the paralleling Cameroon Vol-

canic Line—a chain of mostly alkaline Cenozoic volca-

noes extending 1600 km from the Atlantic Ocean into

the African interior (Figure 13.35; see also Fitton,

1987). This crustal extension and chain of alkaline

magmatism is not related to passage of the African

plate over a mantle plume because there is no system-

atic age progression along the chain, as in the Hawaiian

island chain.

Magmatic rocks in continental rifts range as widely

as in any tectonic regime. Maﬁc rocks include subalka-

line basalt and more often silica-undersaturated alka-

line basalt, basanite, and nephelinite. Large volumes

of associated felsic rocks, including rhyolite, trachyte,

and phonolite, that occur with the maﬁc rock types de-

ﬁne bimodal associations; intermediate compositions

are subordinate or absent. Some rifts, especially the

slower-spreading ones, have highly alkaline rock types

that contain abundant feldspathoids, locally to the

exclusion of both plagioclase and alkali feldspar.

Silica-poor, feldspar-free carbonatite is a rare, extreme

rock type associated with feldspathoidal rocks. In

some deeply eroded rifts, A-type granitic rocks are

Magmatic Petrotectonic Associations

393

13.37 Tectonic discrimination diagram for granitic rocks based on concentrations of Rb versus Y  Nb. Symbols show modern rock suites used

to subdivide the diagram. Arc granites broadly correspond to I-type and collision granites to S-type. (Redrawn from Pearce et al., 1984.)

110

100 1000

Y + Nb (ppm)

100

1000

(ppm)

Continent-

continent

collision

Anorogenic

Oceanic

ridge

Continental

and island

arc

prominent, as described earlier, in others, such as the

Precambrian Gardar province in southernmost Green-

land, plutonic complexes of nepheline syenite-syenite-

alkali granite are exposed. Many highly alkaline rocks

are unusually enriched in incompatible elements; thus,

Zr, Nb, Ta, Rb, Th, and others, can be major elements

stabilizing major rock-forming minerals such as zircon,

tantalite, and thorite. Associated complex pegmatites

may be economically feasible resources of these exotic

metals.

13.11.1 Transitions from Continental Arc to Rift

Associations in Western North America

Long-lived, contractional continental arcs commonly

evolve in their more mature stages into a regime of

crustal extension. Although reasons for this tectonic

transition vary and may never be certain, there is no

question that the associated magmatic rocks experi-

ence a change in composition accompanying the shift

in tectonism. Two examples sufﬁce to demonstrate the

variation in this transition.

Basin and Range Province. Since about 30 Ma in the

middle Tertiary, subduction along the North American

plate margin in the western United States has been pro-

gressively supplanted by transform plate motion. In-

land of this growing transform system, which includes

the San Andreas fault, east-west crustal extension has

been forming the Basin and Range Province (Figure

13.29). Although the exact cause of the extension re-

mains controversial, its inception has been associated

with a transition from typical continental arc rocks to a

variety of rift-related volcanic rocks.

In the northern Basin and Range, early and middle

Tertiary subduction-related volcanism was character-

ized by extrusion of high-K arc lavas that are mostly

andesite and, in the middle Tertiary, of rhyolite-dacite

ignimbrite (Section 13.6.1). After about 24 Ma, in-

creasing proportions of true basalt (IUGS classiﬁca-

tion) and rhyolite magmas were extruded, forming a

bimodal association. The absence of intermediate-

composition magmas in this association is convention-

ally interpreted to reﬂect the “ease” of extrusion of

mantle-derived basalt magmas in an extensional stress

regime (Figure 9.12) and the generation of anatectic

rhyolite magmas from continental crust and/or basalt

lodged earlier in the crust. Many Basin and Range

rocks became more sodic, as manifested especially in

local peralkaline rhyolites, whereas other rhyolites con-

tain topaz. Both have A-type afﬁnities.

In a study of 750 maﬁc lava samples from the south-

western United States, Fitton et al. (1991) found that

all basaltic lavas with MgO 4 wt.% in the central

Basin and Range Province 5 Ma have a distinct trace

element signature like that of oceanic island basalt

(OIB) generated by an asthenospheric source. Other

basalt ﬁelds, as well as all ﬁelds 5 Ma, appear to have

been generated in lithospheric mantle enriched by

subduction-derived ﬂuids, both of recent (early Ceno-

zoic) and ancient (Proterozoic) vintage. This space-

time contrast in source suggests that the lithosphere

beneath the Basin and Range Province had been ther-

mally eroded by upwelling, hotter asthenosphere, and

in some parts removed completely, by 5 Ma. Sparse

basaltic lavas lacking an arc signature (i.e., having no

negative Nb anomaly) but having OIB afﬁnities had

appeared by about 20 Ma and rather primitive OIB-

like basalt by about 16 Ma in the Basin and Range

Province. Apparently, some unmodiﬁed astheno-

spheric magmas had begun to make their way to the sur-

face long before the lithosphere had been eliminated, or

at least made ineffective, in magma generation.

Western Mexican Volcanic Belt. The 1100-km-long

Mexican Volcanic Belt trends east-west across Mexico

opposite the actively subducting Rivera and Cocos

394 Igneous and Metamorphic Petrology

13.38 East African continental rift system and its union in northern

Ethiopia with the Red Sea and Gulf of Aden oceanic rifts.

These rifts lie within a chain of coalesced broad domical up-

warps (light shaded). High-angle normal faults are shown by

dashed lines. Double lines denote oceanic rifts cut by trans-

form faults. Dotted lines outline countries mentioned in text.

Note several lakes (dark shaded) in the grabens. (Redrawn

from Gass, 1972.)

Mediterranean

Sea

22° E

60° E40° E

20° N

20° S

0°

Tanzania

Malawi

Mozambique

Indian Ocean

0mi500

0km800

Kenya

Gulf

of Aden

Arabian

Peninsula

Red Sea

Ethiopia

Uganda

oceanic plates (Figure 13.29). Since the early Pliocene

(about 5 Ma), and especially during the Quaternary,

minor volumes of magma not generally found in arcs

have been extruded from small rifts within the volcanic

belt. These atypical extrusions are intimately associated

with those of calc-alkaline arc afﬁnity apparently dif-

ferentiated from primitive, Hy-normative basalt mag-

mas. The atypical rocks, similar to those found in many

continental rifts, include OIB-type (Section 13.2.1)

Ne-normative mugearite, benmoreite, trachybasalt, tra-

chyte, and peralkaline rhyolite as well as leucitite and

fresh lava ﬂows of lamprophyre. (Leucitite is a glassy to

aphanitic rock composed of leucite, clinopyroxene,

and variable amounts of olivine; lamprophyre is de-

scribed later.) The lamprophyres have an arclike trace

element signature (depleted Ta and Nb) and high con-

centrations of incompatible elements (K, P, Ba, Sr, light

REEs) typical of alkaline rocks. Luhr (1997) believes

these attributes resulted from generation of magma

in incompatible-element-enriched mantle veined by

metasomatic phlogopite, amphibole, and apatite. Ex-

tensional fractures in the rifted crust allowed these low-

degree partial melts of enriched composition to ascend

to the surface, whereas larger-degree partial melts di-

luted in the vein component yield the more voluminous

calc-alkaline magmas.

13.11.2 Magmatism in the East African Rift System

Broadly concurrent magmatism, crustal uplift, and ex-

tensional faulting are well expressed in the East African

rift system, which extends some 3700 km from Mo-

zambique in central eastern Africa northward through

Ethiopia, where it splits into the Gulf of Aden and Red

Sea oceanic rifts (Figure 13.38). The rate, amount, and

time since inception of extension decrease, scissorlike,

southward. In its central part, the system bifurcates

into eastern and western branches. Grabens within

these continental rifts have dropped as much as 3 km.

Major crustal upwarp occurred in the early Tertiary,

probably in relation to an underlying plume head,

forming the broad Ethiopia dome and ﬂooding nearly

in Ethiopia and southwest Arabia with alka-

line basaltic lavas to depths of as much as 3 km (Table

13.4). Breakup of the continental crust created the

Gulf of Aden in the Miocene and the Red Sea in

Pliocene time. Typical MORB has been since extruded

from these oceanic rifts.

In Ethiopia, Kenya, and northern Tanzania, Miocene

and younger continental extension has been accompa-

nied by extrusion of vast ﬂoods of transitional alkaline-

tholeiitic basaltic lavas as well as peralkaline rhyolite,

trachyte, and phonolite lava and pyroclastic ﬂows.

Individual phonolite ﬂood lavas have volumes of as

much as 300 km

, and their aggregate volume is about

50,000 km

. The origin of such a vast volume of com-

positionally uniform ﬂood lavas poses a petrogenetic

dilemma because of the brevity of activity during the

Miocene and the absence of intermediate rock types

extruded after the earlier-erupted alkaline basalt lavas.

However, experiments by Hay and Wendlandt (1995)

show that, under lower crustal pressures, crystalline

phases at the liquidus of ﬂood phonolite match the

near-solidus assemblage of alkaline basalt; this sug-

gests that the phonolite magmas may represent partial

melts of earlier basalts underplating the crust below

the rift.

The region between western Uganda and Zaire in

the western branch of the rift has been famous for many

decades because of its highly alkaline, ultrapotassic

O  3 wt.%), ultramaﬁc, silica-undersaturated vol-

canic rocks, known as kamafugites. The label for these

globally very rare lavas and pyroclastic deposits of post-

Pliocene age has been coined from the three dominant

rock-type names—katungite, mafurite (Table 13.11),

and ugandite. These consist of a Si-poor alkalic mineral

that is predominantly melilite, kalsilite, or leucite, re-

spectively, in addition to olivine, clinopyroxene, Fe-Mg

mica, Ti-rich magnetite, and perovskite. Kalsilite, virtu-

ally unknown outside Italy and Uganda, is essentially

KAlSiO

and can only crystallize in the most extremely

Si-Na-poor, K-rich magmas. Partial melting of mantle

rock veined with phlogopite is probably the source of

parental kamafugite magmas (Edgar, 1996).

Carbonatite-Nephelinite Association. The southern part

of the African rift system in Tanzania and Malawi and

including parts of the western rift (Figure 13.38) har-

bors volcanic and small shallow intrusive bodies of car-

bonatite. About one-half of the known 330 carbonatite

occurrences worldwide (Bell, 1989) are on the African

plate, including the only two on ocean islands (Canary

and Cape Verde just west of the African coast). Most

carbonatites occur in continental rifts and upwarps; the

Paraná-Etendeka (Figure 13.22) is another region

where they occur. Though very small in total world-

wide area (a few 100 km

), carbonatites are economi-

cally valuable and a great petrologic curiosity. No fewer

than ﬁve books dealing wholly with carbonatite have

been published.

Carbonatite contains 50% carbonate minerals,

usually calcite. However, since 1960 the nephelinitic

Oldoinyo Lengai volcano in Tanzania has erupted al-

kali carbonate lavas and pyroclastics. Earlier arguments

whether carbonatite is truly a magmatic rock were

put to rest by this eruption and discovery of other car-

bonatite volcanic rocks. Table 13.11 (columns 1 and

2) reveals the extreme composition of carbonatites; in

alkali carbonatite, there is 0.2 wt.% SiO

 Al

but major concentrations of SrO, BaO, SO

, Cl, F,

and, of course, CO

. Relative to the mantle and conti-

nental crust, carbonatites are also strongly enriched in

REEs, especially light ones, Y, Pb, Th, U, and Nb, and

depleted in Sc, V, Cr, Co, Ni, Rb, Zr, and T (Barker,

Magmatic Petrotectonic Associations

395

1996a). Carbonatites host no fewer than about 275

documented minerals in addition to carbonate miner-

als; other common minerals include apatite, magnetite,

ﬂuorite, pyrochlore (an Nb oxide that contains substi-

tuting Ta, Ti, Zr, Th, U, Pb, Ce, Ca, Sr, Ba), and alkalic

amphibole.

Most carbonatites are intimately associated with

phonolite and more commonly with olivine-poor

nephelinite (Table 13.11, column 3; nephelinites of

oceanic islands, such as Hawaii, Table 13.2, are

olivine-rich). Although young carbonatite-nephelinite

associations make up composite volcanoes of lavas and

396 Igneous and Metamorphic Petrology

Table 13.11. Chemical Composition of Carbonatites and Highly Alkaline Rocks

1234567 8910

SiO

0.16 13.53 42.20 39.06 48.36 47.4 53.64 42.31 35.09 30.0

TiO

0.02 1.94 2.48 4.36 0.70 1.82 6.29 3.75 1.06 1.8

0.0 2.40 12.04 8.18 16.80 9.3 8.13 3.92 2.55 2.6

0.28t 12.96t 13.97t 4.61 2.55 9.82t 7.78t 9.4t

FeO 4.98 4.40 6.78t 8.27t

MnO 0.38 0.45 0.27 0.26 0.13 0.14 0.15 0.2

MgO 0.38 8.45 6.22 17.66 6.57 16.4 7.82 24.42 29.02 29.4

CaO 14.02 35.33 15.17 10.40 9.85 9.32 3.23 5.00 3.49 10.9

O 37.22 0.87 3.93 0.18 1.30 2.14 0.49 0.50 0.18 0.3

O 8.38 0.07 2.54 6.98 8.33 2.41 9.60 4.01 2.91 1.2

0.85 3.27 0.88 0.61 0.61 1.27 1.23 1.59 0.68 1.6

0.56 5.83 2.64 6.07 7.3

31.55 11.64 5.4

3.72 0.28

LOI 4.76 11.76

F 25,000 3100

CI 34,000 127 110

Cr 801 316 1022 373 1014 1852 1398

Ni 26 29 515 74 699 343 968 1253 1018

Rb 178 45 154 636 132 457 471 159 66

Ba 14,900 3400 1211 2407 1202 621 10,607 10,584 2442 915

Th 3.77 12 46 8.7 30 57 30 30

U 10.6 10.2 3.1 3.8 2.4 3 5

Nb 28 483 111 166 14 26 147 186 97 168

Ta 0.0 0.58 1.8 7.9 9.9 9 11

La 545 92 170 88 58 348 242 168 100

Ce 645 158 202 127 629 414 324 239

Sr 12,000 8000 910 1559 1812 678 1296 1325 1127 1145

Nd 102 61 90 70 212 146 115

Sm 7.80 17.5 16.8 25 18 12.6 10

Zr 10 845 193 202 266 682 1401 1167 214 308

Eu 1.62 3.2 5.33 5.8 4.3 3.00 3

Tb 0.32 1.5 2 1.8 1.4 0.82 0.8

Y 7 25 17 26 47 21 16

Yb 0.46 2.2 2.26 1.6 1.5 0.99 1

Lu 0.058 0.28 0.19 0.17 0.13 0.13

1, Alkali carbonate lava, Oldoinyo Lengai volcano, Tanzania. Composed of alkali carbonate and minor MnS. Data from Barker (1996). 2, Cal-

cic carbonatite lava, Fort Portal ﬁeld, Uganda. Phenocrysts (and/or xenocrysts) of phlogopite, magnetite, clinopyroxene; groundmass of cal-

cite, spurrite, apatite, periclase, perovskite, barite, pyrrhotite. Data from Barker (1996). 3, Olivine-poor nephelinite, Mt. Elgon, border of

Uganda and Kenya. Data from Le Bas (1987). 4, Mafurite, Uganda. Data from Edgar (1996). 5, Leucite phonotephrite, Italy. Data from Con-

ticelli and Peccerillo (1992). 6, Mica-rich minette lamprophyre, Wasatch Plateau, Utah. Data from Tingey et al. (1991). 7, Phlogopite-rich lam-

proite, average of 228 samples, West Kimberly, Western Australia. Contains <0.5 wt.% CO

and modal leucite but 8.8% normative quartz!

Data from Mitchell and Bergman (1991). 8, Olivine-rich lamproite, average of 105 samples, West Kimberly, Western Australia. Data from

Mitchell and Bergman (1991). 9, Orangeite (micaceous, or group II kimberlite), Sover, South Africa. Data from Mitchell (1995). 10, Arche-

typal (Group) kimberlite, average of 30 dike, sill, and pipe intrusions in Kimberly area, South Africa. Data from Scott-Smith (1996). *Mostly

O  CO