[j. g. fitton, b. g. j. upton] alkaline igneous ro(bookfi.org)

Alkaline Igneous Rocks

Geological Society Special Publications

Series Editor K. C 0 E

G E O L O G I C A L S O C I E T Y S P E C I A L P U B L I C A T I O N NO 30

Alkaline Igneous Rocks

E D I T E D BY

J. G. FITTON & B. G. J. U P T O N Grant Institute of Geology

University of Edinburgh

Edinburgh EH9 3JW

1987

Published for

The Geological Society by

Blackwell Scientific Publications

O X F O R D L O N D O N E D I N B U R G H

B O S T O N P A L O ALTO M E L B O U R N E

Published by Blackwell Scientific Publications Editorial offices: Osney Mead, Oxford OX2 0EL 8 John Street, London WC1N 2ES 23 Ainslie Place, Edinburgh EH3 6AJ 52 Beacon Street, Boston

Massachusetts 02108, USA 667 Lytton Avenue, Palo Alto

California 94301, USA 107 Barry Street, Carlton

Victoria 3053, Australia

9 1987 The Geological Society. Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by The Geological Society for libraries and other users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that a base fee of $02.00 per copy is paid directly to CCC, 27 Congress Street, Salem, MA 01970, USA. 0305-8719/87 $02.00

First published 1987

Typeset, printed and bound in Great Britain by William Clowes Limited, Beccles and London

DISTRIBUTORS

USA and Canada Blackwell Scientific Publications Inc PO Box 50009, Palo Alto California 94303

Australia Blackwell Scientific Publications

(Australia) Pty Ltd 107 Barry Street, Carlton, Victoria 3053

British Library Cataloguing in Publication Data

Alkaline igneous rocks.--(Geological Society special publications, ISSN 0305-8719) 1. Alkalic igneous rocks I. Fitton, J .G. II. Upton, B. G. J. III. Series 552'. 1 QE462.A4

ISBN0-632-01616-7

Library of Congress Cataloging-in-Publication Data

Alkaline igneous rocks.

(Geological Society special publication; no. 30) Bibliography: p. Includes index. 1. Alkalic igneous rocks. I. Fitton, J. G.

II. Upton, B. G.J . III. Geological Society of London. IV. Series. QE462.A4A43 1987 552'. 1 86-26364

ISBN0-632-01616-7

Contents

Preface vii

Introduction: FITTON, J. G. & UPTON, B. G.J . ix

BAILEY, D. K. Mantle metasomatism--perspective and prospect 1

MENZIES, M. Alkaline rocks and their inclusions: a window on the Earth's interior 15

EDGAR, A. D. The genesis of alkaline magmas with emphasis on their source regions: inferences 29 from experimental studies

LE Bhs, M. J. Nephelinites and carbonatites 53

TWYMAN, J. D. & GITTINS, J. Alkalic carbonatite magmas: parental or derivative ? 85

DAWSON, J. B. The kimberlite clan: relationship with olivine and leucite lamproites, and 95 inferences for upper-mantle metasomatism

BERGMAN, S. C. Lamproites and other potassium-rich igneous rocks: a review of their 103

occurrence, mineralogy and geochemistry

ROCK, N. M. S. The nature and origin of lamprophyres: an overview 191

CLAGUE, D. A. Hawaiian alkaline volcanism 227

WEAVER, B. L., WOOD, D. A., TARNEY, J. & JORON, J. L. Geochemistry of ocean island basalts 253 from the South Atlantic: Ascension, Bouvet, St. Helena, Gough and Tristan da Cunha

HARRIS, C. & SHEPPARD, S. M. F. Magma and fluid evolution in the lavas and associated 269

granite xenoliths of Ascension Island

FITTON, J. G. The Cameroon line, West Africa: a comparison between oceanic and continental 273

alkaline volcanism

BAKER, B. H. Outline of the petroIogy of the Kenya rift alkaline province 293

MACDONALD, R. Quaternary peralkaline silicic rocks and caldera volcanoes of Kenya 313

WOOLLEY, A. R. & JONES, G. C. The petrochemistry of the northern part of the Chilwa alkaline 335

province, Malawi

BOWDEN, P., BLACK, R., MARTIN, R. F., IKEI E. C. KINNAIRD, J. A. & BATCHELOR, R . A . 357 Niger-Nigerian alkaline ring complexes: a classic example of African Phanerozoic anorogenic

mid-plate magmatism

LI/~GEOIS, J. P. & BLACK, R. Alkaline magmatism subsequent to collision in the Pan-African 381

belt of the Adrar des Iforas (Mali)

FLETCHER, C. J. N. & BEDDOE-STEPHENS, B. The petrology, chemistry and crystallization 403

history of the Velasco alkaline province, eastern Bolivia

BARKER, D. S. Tertiary alkaline magmatism in Trans-Pecos Texas 415

EBY, G. N. The Monteregian Hills and White Mountain alkaline igneous provinces, eastern 433

North America

vi Contents

UPTON, B. G. J. & EMELEUS, C. H. Mid-Proterozoic alkaline magmatism in southern 449 Greenland: the Gardar province

LARSEN, L. M. & SORENSEN, H. The Ilimaussaq intrusion--progressive crystallization and 473 formation of layering in an agpaitic magma

NIELSEN, T. F. D. Tertiary alkaline magmatism in East Greenland: a review 489

DOWNES, H. Tertiary and Quaternary volcanism in the Massif Central, France 517

KOGARKO, L. N. Alkaline rocks of the eastern part of the Baltic Shield (Kola Peninsula) 531

Index 545

Preface

The papers contained in this volume were presented at a symposium held

in Edinburgh in September 1984, which marked the passage of ten years

since the publication of The Alkaline Rocks edited by Henning Sorensen.

In organizing the symposium and compiling this volume we aimed to

review recent developments in the petrology and geochemistry of alkaline

igneous rocks. We have, for example, paid particular attention to work on

lamprophyres and carbonatites which are rock associations of current

interest not covered in Sorensen's book. Reviews of recent work on some

of the classic alkaline provinces, such as East Africa, southern Greenland

and the Kola Peninsula, are included together with reviews of less well-

known areas. Other papers discuss the impact of experimental, geochem-

ical and isotopic studies on our understanding of the generation and

evolution of alkaline magmas.

We are indebted to the contributors for their collaboration in producing

this volume and it is with sadness that we note the death, on 14 February

1986, of Brian Baker, whose pioneering field studies formed the basis for

much of our knowledge of the tectonic and volcanic evolution of the East

African Rift. An obituary and appreciation of his work is published in the

Journal of Volcanology and Geothermal Research (28, v-vii). We are grateful

to The Geological Society and The Royal Society of Edinburgh for their

generous assistance with the symposium costs, to Lucian Begg and Dodie

James for their help with organizing the symposium and producing this

volume, and to our colleagues for the care and enthusiasm with which

they reviewed the manuscripts. The efforts of Edward Wates and his staff

at Blackwell Scientific Publications are also gratefully acknowledged. To

all these we offer our sincere thanks. J.G.F.

B.G.J.U.

vii

Introduction

J. G. Fitton & B. G. J. Upton

Alkaline igneous rocks may be defined as those which have higher concentrations of alkalis

than can be accommodated in feldspars alone, the excess appearing as feldspathoids, sodic

pyroxenes, sodic amphiboles and other alkali-rich phases. These rocks are, therefore,

deficient in silica and/or alumina with respect to alkalis and will have nepheline and/or

acmite in their norms. In practice the term 'alkaline' is used to encompass a wide range of

igneous rocks, not all of which conform to this rigid definition. Carbonatites, for example,

are certainly silica-deficient but are rarely alkali-rich. True (nepheline-normative) alkali

basalts grade into hypersthene-normative transitional basalts without any obvious change

in mineralogy. Since transitional basalts are often closely associated with alkali basalts in

the field, they are traditionally regarded as alkaline. It is now usual practice to define

alkaline igneous rocks simply in terms of their alkali ( N a 2 0 + K20) and silica contents

(see, for example, Le Bas et al. 1986). We have not attempted to review the classification of

alkaline igneous rocks in this volume as this has been dealt with elsewhere (e.g. Sorensen

1974; Streckeisen 1967, 1980). The only alkaline rocks not covered in previous reviews are

those hydrous mafic to ultramafic hypabyssal rocks known as the lamprophyres. The

present volume includes three papers on this group. A comprehensive overview of

lamprophyres is given by Rock and of the sub-group of lamproites by Bergman. The relationship between lamproites and kimberlites (which arguably belong to the lampro-

phyres) is discussed by Dawson.

Volumetrically, alkaline rocks account for less than one per cent of all igneous rocks.

Despite this, their remarkable mineralogical diversity has brought them repeatedly to the

attention of petrologists and mineralogists, with the result that alkaline rocks account for

about half of all igneous rock names. Sorensen (1974) lists no fewer than 400 alkaline rock

types. This diversity springs largely from an abundance of alkalis and deficiency in silica

which together generate a large number of mineral species not stable in more silica-rich,

alkali-poor magmas. However, a large part of the attention given to alkaline rocks is due to

their characteristic high concentrations of incompatible or large-ion lithophile elements

(LILE). These are often of more than academic interest as most of the world's resources of

niobium, tantalum and the rare-earth elements are found in or around alkaline igneous

rock bodies. The economic importance of alkaline igneous rocks is further enhanced by

their association with economic deposits of apatite (Kogarko) and with diamonds (Dawson;

Bergman). Evidence for continental alkaline magmatism can be found as far back as the late

Archaean. For example, biotites from the Poohbah Lake syenite in north-western Ontario

have been dated at around 2.7 Ga (Mitchell 1976) and a similar age has been reported by

Larsen et al. (1983) for the Tupertalik carbonatite in western Greenland. At the present

time, alkaline magmas are erupted in all tectonic environments with the possible exception

of mid-ocean ridges. Even here, though, mildly alkaline lavas are sometimes erupted from

off-axis volcanoes, as in the Vestmann Islands of Iceland. Alkaline igneous rocks are found

on all the continents and on islands in all the ocean basins. Their occurrences may be

classified on the basis of tectonic setting into three categories; continental rift valley

From: FITTON, J. G. & UPTON, B. G. J. (eds), 1987, Alkaline Igneous Rocks, Geological Society Special Publication No. 30, pp. ix-xiv.

ix

x In t roduct ion

magmatism, oceanic and continental intraplate magmatism without clear tectonic control,

and alkaline magmatism related to subduction processes. In practice, however, this

classification is not always easy to apply.

Continental rift valleys provide, volumetrically, the most important occurrences of

alkaline igneous rocks although continental rifting is not always accompanied by

magmatism. The best known example (arguably the type example) is the East African Rift

which, in the course of its long history, has yielded almost the entire spectrum of alkaline

magmas. Three papers in this volume are devoted to various aspects of magmatism in the

East African Rift. Baker reviews its magmatic associations with respect to tectonic

development and discusses the origin of the magmas, particularly in relation to the

problems presented by the strongly bimodal distribution of basic and salic lava compositions.

He concludes that the salic magmas evolved from basic parental magmas by processes of

crystal fractionation (cf. Bailey). Macdonald focusses attention on the peralkaline silicic

central volcanoes of Kenya and also favours an origin by crystal fractionation for most of

the evolved magmas. There is little evidence for contamination of the evolving magmas

with ancient continental crust except in the case of the Naivasha comendites. It is not

always possible to demonstrate a genetic link between basic and evolved magmas in the

East African Rift, however. The Chilwa alkaline province in Malawi, at the southern end

of the rift, is an essentially intrusive province in which salic rocks predominate. The

scarcity of basic rocks in this part of the rift has led Woolley & Jones to suggest that the

evolved magmas were produced directly by melting of metasomatised mantle and lower

crust. Insight into the processes occurring at depth beneath rift valleys may be gained by

studying ancient and deeply eroded examples. The Proterozoic Gardar province in South-

West Greenland is probably the best studied of these and is reviewed by Upton & Emeleus.

One of the most striking features of the province is the presence of giant dykes, up to 800

metres wide. These are dominantly basic but in places show in situ differentiation into more

salic rocks. Salic magma generated in the wider portions of these dykes migrated upwards

and may ultimately have accumulated to produce central complexes in which basic magma

was subordinate or absent (e.g. the Ilimaussaq intrusion, Larsen & Serensen). The giant

dykes, therefore, play a crucial role in understanding the relationship between basic and

salic magmas in this and possibly other rift systems.

The separation of continents to form ocean basins must always be preceded by a phase

of continental rifting, leaving volcanic and intrusive complexes stranded along passive

continental margins. The vigorous magmatism which accompanies continental separation

is generally tholeiitic in character as, for example, in the Karoo and Deccan flood basalt

provinces. Alkaline magmas, however, may be emplaced along the trailing continental

margin during the waning phase of magmatism, long after the spreading centre is

established off-shore. The Tertiary volcanic rocks exposed along the east coast of Greenland

(described by Nielsen) provide an excellent example of such an alkaline province.

A second major occurrence of alkaline igneous rocks is provided by intraplate magmatic

provinces whose activity and siting are not subject to any obvious tectonic control. In the

ocean basins such magmatism manifests itself as ocean islands which are sometimes

aligned in chains with ages increasing away from the active centres, as in the Hawaiian

islands. In these cases it is possible to relate the magmatism to convective plumes within

the asthenosphere. The Hawaiian islands (reviewed by Clague) show clearly defined

In troduct ion xi

magmatic cycles starting with alkaline magmas (represented by Loihi seamount), passing

through a voluminous tholeiitic shield-building stage and returning to alkaline magmatism

during the waning phases of individual volcanic centres. These cycles seem to result from

movement of the oceanic plate over a rising plume of partially molten asthenosphere in

which the degree of melting increases towards the centre. They may be typical of ocean

islands in general and are broadly analogous to similar cycles seen in flood basalt provinces

preserved on passive continental margins.

Continental intraplate magmatism may also show age progressions as in the Niger-

Nigeria province (Bowden et al.). These progressions are, however, very rare and not so

clearly defined as in ocean island chains, probably because the continental lithosphere is

thicker and less easily penetrated than oceanic lithosphere. The Monteregian Hills and

White Mountain provinces of eastern North America (Eby), for example, show no obvious

progression but their seaward extension, the New England seamounts, show a regular

decrease in age eastwards. Reviews of other continental intraplate alkaline provinces are

given by Fletcher & Beddoe-Stevens (Velasco province, Bolivia) and Kogarko (Kola

Peninsula). The Cameroon line in West Africa (Fitton) includes both continental and

oceanic alkaline volcanic centres. None of these examples shows any clear progression of

ages. Some continental provinces undergo repeated alkaline magmatism in one place over

long periods. For example, the Kola Peninsula (Kogarko) was the site of alkaline magmatism

in the mid Proterozoic and again in the Devonian. Such examples could be the result of

coincidence but are more likely the result of the repeated exploitation of zones of weakness

in the lithosphere.

Destructive plate boundaries provide the third tectonic setting in which alkaline igneous

rocks may occur. During the life of a subduction zone the characteristic calc-alkaline

magmas tend to become more potassic with time and may give way to volcanic rocks of the

shoshonitic association, some members of which may contain leucite. A discussion of

subduction zone processes is beyond the scope of this volume and the reader is referred to

the reviews of Gill (1981) and Ewart (1982). There are, however, two circumstances under

which subduction processes can lead to the generation of more 'normal' alkaline magmas.

Once the descending slab has become dehydrated at depth it loses its capacity to

stimulate the generation of calc-alkaline magmas but it can still cause melting in the

overlying asthenosphere. This can lead to the production of alkaline magmas from the

mantle above the deepest parts of subduction zones. One such example of alkaline magmas

erupted under a compressive regime is provided by the Trans-Pecos province of west Texas

(Barker). The alkaline rocks in this area grade south-westwards into the calc-alkaline rocks

of the Sierra Madre Occidental in Mexico and Barker relates both suites to subduction of

the Farallon Plate.

After the cessation of subduction, relaxation of the former compressive regime often

results in extension and the generation of alkaline magmas. The resulting switch from

subduction-related calc-alkaline to extensional alkaline magmatism appears to be a

common phenomenon. It occurred, for example, in the western U.S.A. about 17 Ma ago

and in parts of Africa and Arabia at the end of the Pan-African metamorphic episode

during the late Precambrian. An example from the Pan-African belt in Mali is discussed

by Liegeois & Black. The origin of alkaline magmas has attracted a great deal of interest among igneous

petrologists over the last ten years or so. This interest has been stimulated by two important

xii Introduction

and characteristic features of alkaline rocks. Firstly, they contain high concentrations of

LILE and yet isotopic evidence suggests that their parental magmas had a mantle source

which had been depleted in these elements for a long time. Thus alkaline igneous rocks

may provide useful information about enrichment and/or melting processes in the mantle.

Secondly, many mafic alkaline volcanic rocks contain xenoliths inferred to have originated

within the mantle (Menzies). These are often enriched in LILE when compared with

concentrations expected of chondritic mantle material and sometimes contain amphiboles

and micas of metasomatic origin (Bailey). Such clear evidence for the existence of

metasomatically enriched mantle, coupled with the problem of extracting LILE-rich

magmas from LILE-poor mantle, has led to hypotheses involving mantle metasomatism as

a precursor to alkaline magmatism. These hypotheses, reviewed by Bailey, have gained

popularity over the past fifteen years and are invoked by several contributors to this

volume. Mantle metasomatism neatly explains many of the features of alkaline magmatism.

For example, the frequent association of alkaline magmatism with areas of large-scale

regional uplift is consistent with the relatively low density of metasomatized mantle.

An essential feature of all models involving a metasomatized mantle source for alkaline

magmas is that this source must lie in the lithosphere. This is the only part of the mantle

where enriched material can remain in one place for long periods without being swept

away by convection. The lithospheric mantle beneath the continents is likely to be

chemically and isotopically different from that beneath the oceans. Continental lithospheric

mantle is old and will have had as complex a metamorphic and magmatic history as the

overlying crust. Oceanic lithosphere, on the other hand, is relatively young and probably

depleted in LILE. These differences should be reflected in the compositions of continental

and oceanic alkaline rocks. However, alkali basalts erupted in continental and oceanic

settings are generally identical both chemically (Fitton) and isotopically (Menzies). Since

enriched mantle xenoliths are only commonly found in continental regions it follows that

the enriched lithospheric mantle represented by these xenoliths is not the source of most

continental and oceanic alkaline magmas. An asthenospheric source is therefore implied.

This is not to say that enriched continental lithospheric mantle is never involved in the

generation of alkaline magmas. There is good evidence (e.g. Edgar) that pockets of ancient

enriched mantle beneath cratonic regions provide the source for LILE-rich mafic and

ultramafic alkaline rocks such as micaceous kimberlite (Dawson), and lamproite and other

potassic igneous rocks (Bergman). It is significant that these rock types are exclusively

continental. More extensive melting may involve the continental lithosphere mantle in the

production of less exotic rock types such as flood tholeiite and mildly alkaline basalt. Upton & Emeleus, for example, argue for a lithospheric mantle source for the Gardar alkaline

magmas.

If most alkaline magmas have an ultimate source in the asthenosphere then they must

share this source with unequivocally asthenosphere-derived rocks such as mid-ocean ridge

basalt (MORB). The consistent isotopic differences between MORB and alkali basalts (and

indeed all intraplate basalts) requires that the asthenosphere be heterogeneous. This

heterogeneity may result from the entrainment of lower mantle material in deep mantle

plumes as suggested for the Hawaiian island chain (Clague). The entire convecting upper

mantle may also be heterogeneous on a small scale and alkaline magmas may be generated

by the selective melting of LILE-enriched streaks while more extensive melting produces

MORB (Fitton). Geochemical studies on ocean island basalts from the South Atlantic

Introduct ion xiii

(Weaver et a/.) suggest that one component of these enriched streaks is provided by

subducted ocean-floor sediment.

Derivation of LILE-rich magmas from an asthenospheric source depleted in these

elements requires either very small degress of partial melting or extensive crystal

fractionation. There can be no doubt that many alkaline rocks are the products of extensive

low pressure crystal fractionation but this cannot be true of those alkaline rocks which host

mantle xenoliths. Even the most magnesian alkaline rocks, which must represent near-

primary magmas, are rich in LILE. Small-degree partial melting (< l~ ) is therefore

required to produce such magmas. McKenzie (1985) has recently shown that the extraction

of melt fractions as small as 0.2~ is not only physically possible but inevitable where the

melt viscosity is low, as it probably is in the case of alkaline magmas. Experimental studies

on alkaline rocks and synthetic analogues (reviewed by Edgar) provide useful constraints

on the feasibility of fractional crystallization and partial melting models and on the

temperatures and pressures involved.

Many lines of evidence suggest that volatile components form a significant part of

alkaline magmas. The development of extensive zones of metasomatised country rock

(fenite) around alkaline plutons, the abundance of chlorine and fluorine in some alkaline

igneous rocks, and the frequently explosive eruption of alkaline magma all point to high

concentrations of volatiles. These volatile components play an important role in the

evolution of alkaline magmas and yet relatively little is known about them. Constraints on

their composition have been provided by fluid inclusion studies (Harris & Sheppard) and

by thermodynamic considerations (Kogarko).

The effects of volatile components on the evolution of alkaline magmas can be seen

clearly in both intrusive and extrusive rocks. Larsen & Serensen, for example, discuss the

crystallization history of the Ilimaussaq intrusion in South-West Greenland and show how

the upward migration of low-density, low-viscosity volatile-rich magma delayed crystalli-

zation under the roof of the intrusion. Silicic alkaline pyroclastic deposits around central

volcanoes in Kenya often show striking variations in the abundance of some incompatible

elements within a single vertical section, implying compositional zonation in the magma

chamber before eruption. Macdonald shows that these variations are too large to be

accounted for by crystal fractionation alone and suggests that some elements have been

transported to the magma chamber roof zones as complex ions in a volatile phase.

Carbonatites provide perhaps the best illustration of all of the influence of volatile

components on the origin and evolution of alkaline rocks. There is now a consensus that

their parental magmas originate by the separation of an immiscible carbonate liquid phase

from a CO2-saturated nephelinite or phonolite magma. There is, however, some

disagreement over the nature and subsequent evolution of this parental carbonate magma.

Le Bas argues that the parental magma is rich in alkalis and similar in composition to the

natrocarbonatite lavas erupted from Oldoinyo Lengai. This magma evolves at low pressure

towards the more common calcite carbonatite (s6vite) by loss of alkalis to the surrounding

country rocks which are metasomatized (fenitized) as a result. Twyman & Gittins offer an

alternative scheme in which s6vite magmas are parental and natrocarbonatite magmas are

derived from them by crystal fractionation.

Most petrologists now believe that evolved alkaline magmas are produced by the

fractional crystallization of basic magma. The more highly undersaturated parental

magmas represented by basanite and nephelinite will produce undersaturated derivatives

xiv I n t r o d u c t i o n

such as phonolite and foyaite. Mildly alkaline and transitional basalt m a g m a is likely to

produce trachyte and, with extreme fractionation, alkali rhyolite. The production of

peralkaline acid rocks by crystal fractionation alone is likely to be an inefficient process.

The operation of this process, however, is clearly demonstrated by their association with

transitional basalt on some ocean islands, such as Ascension (Harris & Sheppard).

Peralkaline acid rocks only occur in abundance in continental environments, however, and

here there is often good evidence that crustal contaminat ion has accompanied crystal

fractionation. Well-documented examples of the operation of crustal contaminat ion in the

evolution of alkaline magmas are presented by several of the contributors to this volume.

Downes, for example, shows that the assimilation of lower crustal granulite has affected the

evolution of alkaline magmas in the French Massif Central and uses isotope data to

estimate the extent of this contamination. Other examples are presented by Bowden et al.

(Niger -Niger ia granite ring complexes), Eby (Monteregian Hills and White Mountain

provinces, Nor th America), Fitton (Cameroon line, West Africa) and Fletcher & Beddoe-

Stevens (Velasco province, Bolivia). Other authors propose the derivation of evolved

alkaline magmas directly from metasomat ized mantle or lower crust (Bailey; Woolley &

Jones). Our understanding of the origin and evolution of alkaline magmas has come a long way

since the publication of Sorensen's book in 1974, largely through the acquisition of a far

larger geochemical and isotopic data base. The contributions to this volume review the

current state of this understanding. Emphasis has shifted from crustal to mantle processes

with the recognition of mantle metasomat ism and its possible role as a precursor to alkaline

magmatism. More recently, though, there has been a swing towards the opposite view, that

mantle metasomat ism is caus ed by alkaline magmat ism. Theoretical and experimental

studies on the migration and segregation of small-degree melts seem destined to accelerate

this swing. Despite these advances, however, many mysteries remain unsolved and alkaline

rocks will still provide a fruitful field of research for many years to come, yielding further

insights into the nature of mantle processes and the evolution of magmas.

References

EWART, A. 1982. The mineralogy and petrology of Tertiary-Recent orogenic volcanic rocks: with special reference to the andesitic-basaltic compositional range. In Thorpe, R. S. (ed.) Andesites. Pp. 25-87. John Wiley & Sons, London.

GILL, J. B. 1981. Orogenic Andesites and Plate Tectonics, 390 pp. Springer-Verlag, Berlin.

LARSEN, L. M., REX, D. C. & SECHER, K. 1983. The age of carbonatites, kimberlites and lamprophyres from southern west Greenland: recurrent alkaline magmatism during 2500 million years. Lithos 16, 215-21.

LE BAS, M. J., LE MAITRE, R. W., STRECKEISEN, A. & ZANETTIN, B. 1986. A chemical classification of

volcanic rocks based on the total alkali--silica diagram. J. Petrol. 27, 745-50.

MCKENZIE, D. 1985. The extraction of magma from the crust and the mantle. Earth planet. Sci. Lett. 74, 81-91.

MITCHELL, R. H. 1976. Potassium-argon geochronology of the Poohbah Lake alkaline complex, northwest- ern Ontario. Can. J. Earth Sci. 13, 1456-9.

SORENSEN, H. (ed.) 1974. The Alkaline Rocks. 622 pp. John Wiley & Sons, London.

STRECKEISEN, A. 1967. Classification and nomenclature of igneous rocks. N. Jb. Miner. Abh. 107, 144-240.

- - , 1980. Classification and nomenclature of volcanic rocks, lamprophyres, carbonatites and melilitic rocks. Geol. Rundschau, 69, 194-207.

J. G. FITTON & B. G. J. UPTON, Grant Institute of Geology, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW, U.K.

Introduction

J. G. Fitton & B. G. J. Upton

Alkaline igneous rocks may be defined as those which have higher concentrations of alkalis

than can be accommodated in feldspars alone, the excess appearing as feldspathoids, sodic

pyroxenes, sodic amphiboles and other alkali-rich phases. These rocks are, therefore,

deficient in silica and/or alumina with respect to alkalis and will have nepheline and/or

acmite in their norms. In practice the term 'alkaline' is used to encompass a wide range of

igneous rocks, not all of which conform to this rigid definition. Carbonatites, for example,

are certainly silica-deficient but are rarely alkali-rich. True (nepheline-normative) alkali

basalts grade into hypersthene-normative transitional basalts without any obvious change

in mineralogy. Since transitional basalts are often closely associated with alkali basalts in

the field, they are traditionally regarded as alkaline. It is now usual practice to define

alkaline igneous rocks simply in terms of their alkali ( N a 2 0 + K20) and silica contents

(see, for example, Le Bas et al. 1986). We have not attempted to review the classification of

alkaline igneous rocks in this volume as this has been dealt with elsewhere (e.g. Sorensen

1974; Streckeisen 1967, 1980). The only alkaline rocks not covered in previous reviews are

those hydrous mafic to ultramafic hypabyssal rocks known as the lamprophyres. The

present volume includes three papers on this group. A comprehensive overview of

lamprophyres is given by Rock and of the sub-group of lamproites by Bergman. The relationship between lamproites and kimberlites (which arguably belong to the lampro-

phyres) is discussed by Dawson.

Volumetrically, alkaline rocks account for less than one per cent of all igneous rocks.

Despite this, their remarkable mineralogical diversity has brought them repeatedly to the

attention of petrologists and mineralogists, with the result that alkaline rocks account for

about half of all igneous rock names. Sorensen (1974) lists no fewer than 400 alkaline rock

types. This diversity springs largely from an abundance of alkalis and deficiency in silica

which together generate a large number of mineral species not stable in more silica-rich,

alkali-poor magmas. However, a large part of the attention given to alkaline rocks is due to

their characteristic high concentrations of incompatible or large-ion lithophile elements

(LILE). These are often of more than academic interest as most of the world's resources of

niobium, tantalum and the rare-earth elements are found in or around alkaline igneous

rock bodies. The economic importance of alkaline igneous rocks is further enhanced by

their association with economic deposits of apatite (Kogarko) and with diamonds (Dawson;

Bergman). Evidence for continental alkaline magmatism can be found as far back as the late

Archaean. For example, biotites from the Poohbah Lake syenite in north-western Ontario

have been dated at around 2.7 Ga (Mitchell 1976) and a similar age has been reported by

Larsen et al. (1983) for the Tupertalik carbonatite in western Greenland. At the present

time, alkaline magmas are erupted in all tectonic environments with the possible exception

of mid-ocean ridges. Even here, though, mildly alkaline lavas are sometimes erupted from

off-axis volcanoes, as in the Vestmann Islands of Iceland. Alkaline igneous rocks are found

on all the continents and on islands in all the ocean basins. Their occurrences may be

classified on the basis of tectonic setting into three categories; continental rift valley

From: FITTON, J. G. & UPTON, B. G. J. (eds), 1987, Alkaline Igneous Rocks, Geological Society Special Publication No. 30, pp. ix-xiv.

ix

x In t roduct ion

magmatism, oceanic and continental intraplate magmatism without clear tectonic control,

and alkaline magmatism related to subduction processes. In practice, however, this

classification is not always easy to apply.

Continental rift valleys provide, volumetrically, the most important occurrences of

alkaline igneous rocks although continental rifting is not always accompanied by

magmatism. The best known example (arguably the type example) is the East African Rift

which, in the course of its long history, has yielded almost the entire spectrum of alkaline

magmas. Three papers in this volume are devoted to various aspects of magmatism in the

East African Rift. Baker reviews its magmatic associations with respect to tectonic

development and discusses the origin of the magmas, particularly in relation to the

problems presented by the strongly bimodal distribution of basic and salic lava compositions.

He concludes that the salic magmas evolved from basic parental magmas by processes of

crystal fractionation (cf. Bailey). Macdonald focusses attention on the peralkaline silicic

central volcanoes of Kenya and also favours an origin by crystal fractionation for most of

the evolved magmas. There is little evidence for contamination of the evolving magmas

with ancient continental crust except in the case of the Naivasha comendites. It is not

always possible to demonstrate a genetic link between basic and evolved magmas in the

East African Rift, however. The Chilwa alkaline province in Malawi, at the southern end

of the rift, is an essentially intrusive province in which salic rocks predominate. The

scarcity of basic rocks in this part of the rift has led Woolley & Jones to suggest that the

evolved magmas were produced directly by melting of metasomatised mantle and lower

crust. Insight into the processes occurring at depth beneath rift valleys may be gained by

studying ancient and deeply eroded examples. The Proterozoic Gardar province in South-

West Greenland is probably the best studied of these and is reviewed by Upton & Emeleus.

One of the most striking features of the province is the presence of giant dykes, up to 800

metres wide. These are dominantly basic but in places show in situ differentiation into more

salic rocks. Salic magma generated in the wider portions of these dykes migrated upwards

and may ultimately have accumulated to produce central complexes in which basic magma

was subordinate or absent (e.g. the Ilimaussaq intrusion, Larsen & Serensen). The giant

dykes, therefore, play a crucial role in understanding the relationship between basic and

salic magmas in this and possibly other rift systems.

The separation of continents to form ocean basins must always be preceded by a phase

of continental rifting, leaving volcanic and intrusive complexes stranded along passive

continental margins. The vigorous magmatism which accompanies continental separation

is generally tholeiitic in character as, for example, in the Karoo and Deccan flood basalt

provinces. Alkaline magmas, however, may be emplaced along the trailing continental

margin during the waning phase of magmatism, long after the spreading centre is

established off-shore. The Tertiary volcanic rocks exposed along the east coast of Greenland

(described by Nielsen) provide an excellent example of such an alkaline province.

A second major occurrence of alkaline igneous rocks is provided by intraplate magmatic

provinces whose activity and siting are not subject to any obvious tectonic control. In the

ocean basins such magmatism manifests itself as ocean islands which are sometimes

aligned in chains with ages increasing away from the active centres, as in the Hawaiian

islands. In these cases it is possible to relate the magmatism to convective plumes within

the asthenosphere. The Hawaiian islands (reviewed by Clague) show clearly defined

In troduct ion xi

magmatic cycles starting with alkaline magmas (represented by Loihi seamount), passing

through a voluminous tholeiitic shield-building stage and returning to alkaline magmatism

during the waning phases of individual volcanic centres. These cycles seem to result from

movement of the oceanic plate over a rising plume of partially molten asthenosphere in

which the degree of melting increases towards the centre. They may be typical of ocean

islands in general and are broadly analogous to similar cycles seen in flood basalt provinces

preserved on passive continental margins.

Continental intraplate magmatism may also show age progressions as in the Niger-

Nigeria province (Bowden et al.). These progressions are, however, very rare and not so

clearly defined as in ocean island chains, probably because the continental lithosphere is

thicker and less easily penetrated than oceanic lithosphere. The Monteregian Hills and

White Mountain provinces of eastern North America (Eby), for example, show no obvious

progression but their seaward extension, the New England seamounts, show a regular

decrease in age eastwards. Reviews of other continental intraplate alkaline provinces are

given by Fletcher & Beddoe-Stevens (Velasco province, Bolivia) and Kogarko (Kola

Peninsula). The Cameroon line in West Africa (Fitton) includes both continental and

oceanic alkaline volcanic centres. None of these examples shows any clear progression of

ages. Some continental provinces undergo repeated alkaline magmatism in one place over

long periods. For example, the Kola Peninsula (Kogarko) was the site of alkaline magmatism

in the mid Proterozoic and again in the Devonian. Such examples could be the result of

coincidence but are more likely the result of the repeated exploitation of zones of weakness

in the lithosphere.

Destructive plate boundaries provide the third tectonic setting in which alkaline igneous

rocks may occur. During the life of a subduction zone the characteristic calc-alkaline

magmas tend to become more potassic with time and may give way to volcanic rocks of the

shoshonitic association, some members of which may contain leucite. A discussion of

subduction zone processes is beyond the scope of this volume and the reader is referred to

the reviews of Gill (1981) and Ewart (1982). There are, however, two circumstances under

which subduction processes can lead to the generation of more 'normal' alkaline magmas.

Once the descending slab has become dehydrated at depth it loses its capacity to

stimulate the generation of calc-alkaline magmas but it can still cause melting in the

overlying asthenosphere. This can lead to the production of alkaline magmas from the

mantle above the deepest parts of subduction zones. One such example of alkaline magmas

erupted under a compressive regime is provided by the Trans-Pecos province of west Texas

(Barker). The alkaline rocks in this area grade south-westwards into the calc-alkaline rocks

of the Sierra Madre Occidental in Mexico and Barker relates both suites to subduction of

the Farallon Plate.

After the cessation of subduction, relaxation of the former compressive regime often

results in extension and the generation of alkaline magmas. The resulting switch from

subduction-related calc-alkaline to extensional alkaline magmatism appears to be a

common phenomenon. It occurred, for example, in the western U.S.A. about 17 Ma ago

and in parts of Africa and Arabia at the end of the Pan-African metamorphic episode

during the late Precambrian. An example from the Pan-African belt in Mali is discussed

by Liegeois & Black. The origin of alkaline magmas has attracted a great deal of interest among igneous

petrologists over the last ten years or so. This interest has been stimulated by two important

xii Introduction

and characteristic features of alkaline rocks. Firstly, they contain high concentrations of

LILE and yet isotopic evidence suggests that their parental magmas had a mantle source

which had been depleted in these elements for a long time. Thus alkaline igneous rocks

may provide useful information about enrichment and/or melting processes in the mantle.

Secondly, many mafic alkaline volcanic rocks contain xenoliths inferred to have originated

within the mantle (Menzies). These are often enriched in LILE when compared with

concentrations expected of chondritic mantle material and sometimes contain amphiboles

and micas of metasomatic origin (Bailey). Such clear evidence for the existence of

metasomatically enriched mantle, coupled with the problem of extracting LILE-rich

magmas from LILE-poor mantle, has led to hypotheses involving mantle metasomatism as

a precursor to alkaline magmatism. These hypotheses, reviewed by Bailey, have gained

popularity over the past fifteen years and are invoked by several contributors to this

volume. Mantle metasomatism neatly explains many of the features of alkaline magmatism.

For example, the frequent association of alkaline magmatism with areas of large-scale

regional uplift is consistent with the relatively low density of metasomatized mantle.

An essential feature of all models involving a metasomatized mantle source for alkaline

magmas is that this source must lie in the lithosphere. This is the only part of the mantle

where enriched material can remain in one place for long periods without being swept

away by convection. The lithospheric mantle beneath the continents is likely to be

chemically and isotopically different from that beneath the oceans. Continental lithospheric

mantle is old and will have had as complex a metamorphic and magmatic history as the

overlying crust. Oceanic lithosphere, on the other hand, is relatively young and probably

depleted in LILE. These differences should be reflected in the compositions of continental

and oceanic alkaline rocks. However, alkali basalts erupted in continental and oceanic

settings are generally identical both chemically (Fitton) and isotopically (Menzies). Since

enriched mantle xenoliths are only commonly found in continental regions it follows that

the enriched lithospheric mantle represented by these xenoliths is not the source of most

continental and oceanic alkaline magmas. An asthenospheric source is therefore implied.

This is not to say that enriched continental lithospheric mantle is never involved in the

generation of alkaline magmas. There is good evidence (e.g. Edgar) that pockets of ancient

enriched mantle beneath cratonic regions provide the source for LILE-rich mafic and

ultramafic alkaline rocks such as micaceous kimberlite (Dawson), and lamproite and other

potassic igneous rocks (Bergman). It is significant that these rock types are exclusively

continental. More extensive melting may involve the continental lithosphere mantle in the

production of less exotic rock types such as flood tholeiite and mildly alkaline basalt. Upton & Emeleus, for example, argue for a lithospheric mantle source for the Gardar alkaline

magmas.

If most alkaline magmas have an ultimate source in the asthenosphere then they must

share this source with unequivocally asthenosphere-derived rocks such as mid-ocean ridge

basalt (MORB). The consistent isotopic differences between MORB and alkali basalts (and

indeed all intraplate basalts) requires that the asthenosphere be heterogeneous. This

heterogeneity may result from the entrainment of lower mantle material in deep mantle

plumes as suggested for the Hawaiian island chain (Clague). The entire convecting upper

mantle may also be heterogeneous on a small scale and alkaline magmas may be generated

by the selective melting of LILE-enriched streaks while more extensive melting produces

MORB (Fitton). Geochemical studies on ocean island basalts from the South Atlantic

Introduct ion xiii

(Weaver et a/.) suggest that one component of these enriched streaks is provided by

subducted ocean-floor sediment.

Derivation of LILE-rich magmas from an asthenospheric source depleted in these

elements requires either very small degress of partial melting or extensive crystal

fractionation. There can be no doubt that many alkaline rocks are the products of extensive

low pressure crystal fractionation but this cannot be true of those alkaline rocks which host

mantle xenoliths. Even the most magnesian alkaline rocks, which must represent near-

primary magmas, are rich in LILE. Small-degree partial melting (< l~ ) is therefore

required to produce such magmas. McKenzie (1985) has recently shown that the extraction

of melt fractions as small as 0.2~ is not only physically possible but inevitable where the

melt viscosity is low, as it probably is in the case of alkaline magmas. Experimental studies

on alkaline rocks and synthetic analogues (reviewed by Edgar) provide useful constraints

on the feasibility of fractional crystallization and partial melting models and on the

temperatures and pressures involved.

Many lines of evidence suggest that volatile components form a significant part of

alkaline magmas. The development of extensive zones of metasomatised country rock

(fenite) around alkaline plutons, the abundance of chlorine and fluorine in some alkaline

igneous rocks, and the frequently explosive eruption of alkaline magma all point to high

concentrations of volatiles. These volatile components play an important role in the

evolution of alkaline magmas and yet relatively little is known about them. Constraints on

their composition have been provided by fluid inclusion studies (Harris & Sheppard) and

by thermodynamic considerations (Kogarko).

The effects of volatile components on the evolution of alkaline magmas can be seen

clearly in both intrusive and extrusive rocks. Larsen & Serensen, for example, discuss the

crystallization history of the Ilimaussaq intrusion in South-West Greenland and show how

the upward migration of low-density, low-viscosity volatile-rich magma delayed crystalli-

zation under the roof of the intrusion. Silicic alkaline pyroclastic deposits around central

volcanoes in Kenya often show striking variations in the abundance of some incompatible

elements within a single vertical section, implying compositional zonation in the magma

chamber before eruption. Macdonald shows that these variations are too large to be

accounted for by crystal fractionation alone and suggests that some elements have been

transported to the magma chamber roof zones as complex ions in a volatile phase.

Carbonatites provide perhaps the best illustration of all of the influence of volatile

components on the origin and evolution of alkaline rocks. There is now a consensus that

their parental magmas originate by the separation of an immiscible carbonate liquid phase

from a CO2-saturated nephelinite or phonolite magma. There is, however, some

disagreement over the nature and subsequent evolution of this parental carbonate magma.

Le Bas argues that the parental magma is rich in alkalis and similar in composition to the

natrocarbonatite lavas erupted from Oldoinyo Lengai. This magma evolves at low pressure

towards the more common calcite carbonatite (s6vite) by loss of alkalis to the surrounding

country rocks which are metasomatized (fenitized) as a result. Twyman & Gittins offer an

alternative scheme in which s6vite magmas are parental and natrocarbonatite magmas are

derived from them by crystal fractionation.

Most petrologists now believe that evolved alkaline magmas are produced by the

fractional crystallization of basic magma. The more highly undersaturated parental

magmas represented by basanite and nephelinite will produce undersaturated derivatives

xiv I n t r o d u c t i o n

such as phonolite and foyaite. Mildly alkaline and transitional basalt m a g m a is likely to

produce trachyte and, with extreme fractionation, alkali rhyolite. The production of

peralkaline acid rocks by crystal fractionation alone is likely to be an inefficient process.

The operation of this process, however, is clearly demonstrated by their association with

transitional basalt on some ocean islands, such as Ascension (Harris & Sheppard).

Peralkaline acid rocks only occur in abundance in continental environments, however, and

here there is often good evidence that crustal contaminat ion has accompanied crystal

fractionation. Well-documented examples of the operation of crustal contaminat ion in the

evolution of alkaline magmas are presented by several of the contributors to this volume.

Downes, for example, shows that the assimilation of lower crustal granulite has affected the

evolution of alkaline magmas in the French Massif Central and uses isotope data to

estimate the extent of this contamination. Other examples are presented by Bowden et al.

(Niger -Niger ia granite ring complexes), Eby (Monteregian Hills and White Mountain

provinces, Nor th America), Fitton (Cameroon line, West Africa) and Fletcher & Beddoe-

Stevens (Velasco province, Bolivia). Other authors propose the derivation of evolved

alkaline magmas directly from metasomat ized mantle or lower crust (Bailey; Woolley &

Jones). Our understanding of the origin and evolution of alkaline magmas has come a long way

since the publication of Sorensen's book in 1974, largely through the acquisition of a far

larger geochemical and isotopic data base. The contributions to this volume review the

current state of this understanding. Emphasis has shifted from crustal to mantle processes

with the recognition of mantle metasomat ism and its possible role as a precursor to alkaline

magmatism. More recently, though, there has been a swing towards the opposite view, that

mantle metasomat ism is caus ed by alkaline magmat ism. Theoretical and experimental

studies on the migration and segregation of small-degree melts seem destined to accelerate

this swing. Despite these advances, however, many mysteries remain unsolved and alkaline

rocks will still provide a fruitful field of research for many years to come, yielding further

insights into the nature of mantle processes and the evolution of magmas.

References

EWART, A. 1982. The mineralogy and petrology of Tertiary-Recent orogenic volcanic rocks: with special reference to the andesitic-basaltic compositional range. In Thorpe, R. S. (ed.) Andesites. Pp. 25-87. John Wiley & Sons, London.

GILL, J. B. 1981. Orogenic Andesites and Plate Tectonics, 390 pp. Springer-Verlag, Berlin.

LARSEN, L. M., REX, D. C. & SECHER, K. 1983. The age of carbonatites, kimberlites and lamprophyres from southern west Greenland: recurrent alkaline magmatism during 2500 million years. Lithos 16, 215-21.

LE BAS, M. J., LE MAITRE, R. W., STRECKEISEN, A. & ZANETTIN, B. 1986. A chemical classification of

volcanic rocks based on the total alkali--silica diagram. J. Petrol. 27, 745-50.

MCKENZIE, D. 1985. The extraction of magma from the crust and the mantle. Earth planet. Sci. Lett. 74, 81-91.

MITCHELL, R. H. 1976. Potassium-argon geochronology of the Poohbah Lake alkaline complex, northwest- ern Ontario. Can. J. Earth Sci. 13, 1456-9.

SORENSEN, H. (ed.) 1974. The Alkaline Rocks. 622 pp. John Wiley & Sons, London.

STRECKEISEN, A. 1967. Classification and nomenclature of igneous rocks. N. Jb. Miner. Abh. 107, 144-240.

- - , 1980. Classification and nomenclature of volcanic rocks, lamprophyres, carbonatites and melilitic rocks. Geol. Rundschau, 69, 194-207.

J. G. FITTON & B. G. J. UPTON, Grant Institute of Geology, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW, U.K.

Mantle metasomatism perspective and prospect

D. K. Bailey

S U M M A R Y : Mantle replenishment in lithophile elements has been discerned in the patterns of trace elements and isotopes in lavas. One replenishment process is identified as metasomatic replacement, seen in ultramafic xenoliths brought up in high-velocity alkaline eruptions. Thus alkaline magmatism provides the best primafacie evidence of metasomatism and open-system conditions in the upper mantle. The list of added lithophile elements includes the following: H, C, F, Na, A1, P, S, C1, K, Ca, Ti, Fe, Rb, Y, Zr, Nb, Ba and rare earths. Some metasomatism may be due to wall-rock alteration near magma bodies, but the evidence for metasomatism prior to melting opens the possibility that the process is a precursor to alkaline magmatism, giving the necessary source enrichment in lithophile elements. In some igneous provinces the metasomafism is widespread, intensive and pervasive; in others it appears as veining of variable intensity. Metasomatism as a large- scale process is best indicated by the widespread distribution of alkaline magmatism in space and time: volatile flux through the lithosphere would then be the necessary precursor of metasomatism and magmatism.

Volatile activity, metasomatism and melt enrichment clearly widen the scope for mafic magma generation in the mantle, but some long-standing problems of the alkaline associations (and indeed the calc-alkaline) also call for re-examination in terms of volatile activity in the lithosphere mantle. These include the diversity of magmas, the generation of large felsic volumes, and composition gaps in magma series. Experiments show that felsic minerals are stable to 30 kb, indicating the possibility of felsic-melt generation in the upper part of the mantle. A combination of volatile flux and melt percolation along geotherms that intersect the solidus at depths of less than 80 km would lead to enrichment and metasomatism, providing distinct mantle sources for felsic magmas. Initial (or residual) melts from such a region, as distinct from those from greater depths, would be constrained by equilibria involving felsic minerals. Thus an igneous cycle could generate a bimodal association, with felsic melts forming in the upper regime and the mafic melts originating at depths below the range of felsic-mineral stabilities. Such a magma system is consistent with the observed eruptive characteristics, explains the typical ultramafic nodule and megacryst suites in the alkali olivine-basalt association and is free of difficulties with relative volumes of melts, with eruption timing and with rapid changes in erupted compositions.

Perspective

From its Greek origins the term metasomatism

should imply a 'change of body' or a change of

substance. It must indicate a chemical change in

a pre-existing rock or mineral, and traditionally

has been used to make the distinction from

metamorphism ('change of form') indicating

reconstitution without chemical introduction.

Unfortunately, early definitions of metasomatism

are not rigorous, but common usage applies to

cases where material has been transferred

through a vapour or fluid without melting. More recently, in reference to the mantle, the usage has

widened considerably, sometimes referring to

simple melt infiltration, and even unspecified

enrichment processes of a pre-existing mantle

composition. Obviously the latter usages are not

justified but there is a boundary problem. When

melt is introduced into a rock it may react with

its surroundings and metasomatism may correctly

describe the alteration process in the wall-rocks;

this borderline condition is perhaps part of the

reason for some of the current ambiguity. There

is at present a need for general agreement about

terminology and it would certainly help to retain

scientific precision if 'metasomatism' could be

restricted to those cases where there is petro-

graphic evidence of replacement of a pre-existing

rock. Certainly the term should not be used to

describe simple melt injection, producing a

hybrid mixed rock, nor when there is no petro-

graphic evidence that there has been replacement

of an earlier mineralogy. In other words, the term

should be applied only when there is unequivocal

evidence of the previous substance--otherwise, to

say that a rock has changed its chemistry is

supposition. If there is a case for chemical

introduction but the process is unclear or un-

known, it would be better, and more straightfor-

ward, to use the term enrichment.

Evidence for metasomatism in the mantle has

From: FITTON, J. G. & UPTON, B. G. J. (eds), 1987, Alkaline Igneous Rocks, Geological Society Special Publication No. 30, pp. 1-13.

2 D . K . B a i l e y

been steadily accumulating during the past 10 years (Harte et al. 1975; Lloyd & Bailey 1975) and the topic has provoked such interest that some surveys and reviews have already appeared (e.g. Boettcher & O'Neil 1980, Introduction and tabulation; Bailey 1982). The cited examples are relatively succinct and in some ways complemen- tary, and may be taken as constituting a comprehensive starting point for the following discussion which aims to sketch an overall perspective to the subject.

Occurrence and distribution

Mantle samples showing evidence of metasomatism are provided in ultramafic nodules carried to the surface by high-speed volcanic eruptions. Some peridotite massifs (e.g. Lherz) contain veins of less refractory minerals, while others also contain evenly distributed minerals such as phlogopite and amphibole (e.g. Finero). These are sometimes cited as examples of 'enriched' and even metasomatized mantle but the evidence that such minerals have been introduced from an external source while the peridotite was part of the mantle has still to be established. Even in volcanically derived mantle fragments there may still be problems in distinguishing reaction products produced after incorporation in the magma; various criteria can be applied, however, and it is clear that in many cases minerals were introduced into the peridotite prior to eruption (Bailey 1982).

Mantle nodules showing signs of metasomatism are brought to the surface exclusively by alkaline activity, especially in ultramafic and mafic eruptions. These range from kimberlites, which are characteristically fragmental, through to magmas in the alkali olivine-basalt association (basanites through to mugearites) which may occasionally carry ultramafic nodules. Hence, in terms of distribution, samples may be found in any parts of the stable plates, both oceanic and continental, where there has been alkaline igneous activity; such activity is usually associated with uplift and dislocation, in other words with some kind of lesion in the lithosphere.

Compositions

Metasomatism is recognized when there is replacement of a pre-existing peridotite mineralogy, new minerals being characterized by their content of mobile and volatile elements, as listed in Table 1. Introduction of the listed minerals into garnet or spinel peridotite (generally taken as representing average mantle composition) signifies mantle enrichment in lithophile ele-

TABLE 1. Lithophile-element-bearing m&erals in peridotite xenoliths which either may show

metasomatic replacement of pre-existing

peridotite mineralogy or are associated with

metasomatism

a (Major) biotite, amphibole, clinopyroxene, carbonate

b (Minor) phosphate, titanates, oxides, sulphides

Possible candidates for higher-level metasomatic minerals

c (< 25 kb) feldspathoids, alkali feldspars

ments. Most descriptions of mantle metasomatism deduce the introduction of some or all of the following elements: H, C, F, Na, A1, P, S, C1, K, Ca, Ti, Fe, Rb, Y, Zr, Nb, Ba, rare earths.

If metasomatism were wholly a replacement process then there should be subtraction of equivalent material, balancing the introduction of the above elements. As yet this aspect has received little attention and this may be because the fabric of the rocks showing metasomatism is characteristically veined, with metasomatism proceeding along grain boundaries and fissures: hence the introduced material could be largely accommodated by an increase in rock volume, with redistribution rather than removal of the displaced elements. A parallel case may be seen in alkaline metasomatism in the crust (fenitiza- tion) which is characteristically marked by the introduction of new minerals along a close network of cracks.

Timing

In many cases the indications are that the metasomatic introduction of new minerals, although it must have preceded eruption, is part of the cycle of igneous activity, and there is isotopic harmony between the nodules and the magmas. In some instances there is evidence of an earlier metasomatic event (Erlank & Shimizu 1977; Menzies & Murthy 1980a) and there are cases of possible complex metasomatism and/or enrichment dating as far back as 3 Ga (Erlank et al. 1980; Menzies & Murthy 1980b). No doubt the picture will become still more complex as new data become available; this may be expected because alkaline activity has often been repeated through the same segment of lithosphere (Bailey 1977) and even if metasomatism were nothing more than a minor part of igneous activity there should be samples showing complex histories.

Conditions of metasomatism

Using a combination of experimentally deter- mined mineral stabilities and solidi it is possible

M a n t l e m e t a s o m a t i s m 3

to put limits on the pressures and temperatures of metasomatism, and hence on the conditions in the source mantle. Figs 1 and 2 summarize the conditions for stability of phlogopite, amphibole, clinopyroxene and felsic minerals in the mantle, and by reference to Fig. 4 the relative stability of carbonates can readily be envisaged. Obviously the diagrams must be generalizations because the individual mineral stabilities will depend on bulk composition (mineral and rock), coexistence with other metasomatic minerals, and the presence or absence of vapour, fluid or melt. In spite of these reservations the diagrams offer useful limits and indicate the following relationships.

1 Amphibole is stable near the peridotite solidus only in the upper part of the mantle, its lower boundary corresponding approximately to that of spinel and showing broad equivalence to the lower stability boundaries of feldspars and feldspathoids and the upper stability boundary of carbonate (see Figs 1, 2 and 4). 2 Phlogopite stability extends to higher temperatures and greater depths than amphibole, its

FIG. 2. P-T relationships of mineral stabilities and solidi, and geotherms. Solidi: as in Fig. 1 with the addition of WE, the solidus in the system KA1SiO4- MgO-SiO2-H20-CO2 (Wendlandt & Eggler 1980). WE is used to provide a measure of phlogopite stability in peridotite mantle; its extrapolation to intersect KS at point P indicates a depth limit for phlogopite in the presence of melt (this is similar to the boundary given by Wyllie (1979)). Geotherms: as in Fig. 1. Mineral stabilities: C, coesite: Ks, kalsilite; AB, albite + nepheline; (AB), albite (jadeite + quartz); OR, K-feldspar; SC, solvus crest for alkali feldspars.

FIG. 1. P-Trelationships of solidi, and geotherms. Solidi : PSD, peridotite vapour-absent solidus; KS, kimberlite solidus (Eggler & Wendlandt 1978); OE, peridotite solidus in the presence of H20 and CO2 (limited) (Olafsson & Eggler 1983). For convenience the stability region of spinel peridotite, with or without amphibole, is shown here rather than in Fig. 2. D is the diamond stability boundary (Kennedy & Kennedy 1976). Geotherms: S and O, shield and ocean (Clark & Ringwood 1964); 180, oceanic lithosphere 180 Ma old (Sclater et al. 1980); 30, oceanic lithosphere 30 Ma old (Oldenburg 1981). If geotherms should converge more rapidly in the depth range around 200 km as proposed by Tozer (1967) the geometry of the melt systems would change but the principles would remain the same.

stability limit at the solidus approximately coin- ciding with the inflexion region in some kimberlite geotherms (about 180km). Phlogopite- bearing nodules or lavas from greater depths would therefore seem to be ruled out (Bailey 1986). 3 Clinopyroxene stability can extend up to the vapour-absent solidus and this mineral is arguably the most important but least appreciated metasomatic mineral.

From the above it may be deduced that carbonate and clinopyroxene may appear in any metasomatic regime, but the geological and experimental evidence suggests that high-pressure, low-temperature conditions (e.g. kimberli- tic) favour carbonates while low-pressure high- temperature conditions (basanitic) favour clinopyroxene (see Fig. 5). Consideration of the magmas and their nodule suites also points to the fact that ultramafic alkaline magmas such as kimberlites, melilitites and nephelinites are characterized by phlogopite (in melts and xenoliths) while other magmas such as some nephelinites, but especially basanites, are characterized by xenoliths containing amphibole. Basanites in particular are likely to carry ultramafic nodules containing both spinel and amphibole, which

4 D . K . Bai ley

may be an indicator of a generally shallower source depth for the nodules compared with ultramafic melts.

Outstanding issues

While the reality of mantle metasomatism seems to have found wide acceptance, some differences may be perceived concerning such questions as the following:

1 Which comes first, the magmatism or the metasomatism? 2 Is metasomatism always a local phenomenon (in the vicinity of alkaline intrusions) or can it develop regionally? 3 What is the source of materials erupted at the surface, e.g. are they from the lithosphere or the asthenosphere ?

The last question seems irresolvable until the nature of the asthenosphere can be unequivocally defined. At the moment, the ascription of a rock to an asthenosphere source (on the basis of selected aspects of its trace-element or isotope chemistry) is a convoluted way of saying that it has some attributes of mid-ocean ridge basalt; it would be wholly inappropriate to digress into this minefield. The first two questions are related, and controversy on these issues seems likely to be fruitless. Obviously, samples showing recogniz- able metasomatism can do so only on a local scale, and because volcanically transported fragments of metasomatized mantle must be small the evidence for metasomatism on a large scale must be sought in other ways. Appropriate lines of evidence, indicating large-scale metasomatism, have been described elsewhere (Bailey 1982) but may be summarized as follows: (a) an ultramafic nodule population dominated by metasomatized fragments throughout an igneous province; (b) a range of samples showing various degrees of metasomatism through to completely transformed nodules (alkali clinopyroxenites); and (c) the chemical equivalence of extensively metasomatized nodules and erupted melt compositions. Much of the cogency of the case for large-scale metasomatism is lost if discussion is allowed to focus just on laboratory data: it is then easy to lose sight of the wider geological perspective, where alkali- and volatile-rich magmatism has to be seen in the context of the geothermal and tectonic conditions of stable plates (Bailey 1983). This larger-scale (geological) evidence must be taken together with that of metasomatism on a small scale if we are to retain a balanced view. Both kinds of evidence exist (O'Reilly & Griffin 1984; Wilshire 1984) and they are not mutually exclusive. No amount of evidence or

argument in favour of one can, of itself, falsify the case for the other. Good petrographic evidence of local metasomatism is obviously immune to rebuttal, and no evidence has yet emerged to reject the case for regional development of the process; until such evidence emerges it is important that the argument should not become polar- ized (as in some older geological disputes) lest we end up with a controversy without roots in nature.

The dilemma of the relationships between melting and metasomatism has been highlighted by Hawkesworth et al. (1984), who distinguish two enrichment processes, one typified by sub- solidus metasomatism in kimberlite nodules and the other typified by injection of small-volume melts (typically associated with basanitic magmatism). These are essentially expressions of the effects of pressure and temperature, and it will be shown that they may be explained by the interplay of geothermal gradient, mantle solidus and mineral stabilities. It will be seen there that sub- solidus metasomatism can also play a vital role in basanitic magmatism. Although it is valuable to identify the chemical characteristics of the two types of enrichment (Hawkesworth et al. 1984) it should be remembered that there is a spectrum of alkaline magmatism between kimberlites and basanites, and a spectrum of enrichment processes may be expected (see Fig. 5 and Bailey 1986).

Geological factors

If all the observed metasomatism were a localized consequence of alkaline igneous activity, then the quantities of metasomatized mantle and the petrological significance of the process would be essentially trivial. It is when the geological picture is seen in total that such a conclusion becomes most suspect.

Association with alkaline magmatism signifies the high activities of alkalis, mobile elements and volatiles, and the high-velocity eruptions that bring the samples to the surface emphasize the role of gases in this type of magmatism. When the tectonic framework, especially the connection between alkaline magmatism and lesions in stable lithosphere (and the repetition through time), is taken into account, it is difficult to escape the conclusion that we are observing a process in which volatiles must play a vital role. Alkali rich means volatile rich and, although there may be room for debate about whether the metasomatism is a precursor to magmatism in any particular instance, there can be little doubt that volatile migration may have far reaching effects without the necessary intervention of magma. One of the most graphic examples is in kimberlite peridotite

Mantle metasomatism 5

nodules containing phlogopite: these were sam- pled from points on the geotherm well below the kimberlite solidus (Bailey 1982, Fig. 3) and hence well outside conditions where any melt could exist. Certainly there are strong grounds for supposing that volatile influx is a natural precursor to both metasomatism and alkaline magmatism.

Prospect

Given an outline of present concepts of mantle metasomatism it becomes fruitful to look beyond and to consider other consequences of volatile movement through the mantle. One aspect, of immediate relevance to a volume on Alkaline Rocks, has been largely neglected in discussion of mantle metasomatism: this concerns the broader spectrum embracing not just mafic and ultramafic compositions but felsic alkaline magmas. At what has been described as the limiting case of cratonic magmatism (Bailey 1980a) there is kimberlite with no felsic associates, and closely related must be the lamproites and melilitites with very limited, if not uncertain, felsic connections. These and the ultramafic lamprophyres share other common features, such as style of eruption, that require special consideration in terms of petrogenesis; they have been discussed elsewhere (Bailey 1986). The more voluminous expressions of alkaline mafic magmatism are essentially basanitic (having some connections with nephelinites) grading through to transitional basalts, and these typically form major volcano complexes and have associated felsic magmas. Such alkali basalt magmas show evidence of mantle enrichment and characteristically may carry mantle nodules, which themselves show signs of metasomatism and/or enrichment. Their mantle source is hardly in doubt, and the possibility that this source may have undergone metasomatism raises the question whether the felsic magmas could be part of the same process. Some of the problems of alkaline felsic magmatism, such as regional magma development and comparative volumes, could be resolved in the context of regional metasomatic processes (Bailey 1972, 1974) and there have been no data on relevant mineral stabilities to exclude the possibility of sources in the upper mantle (Bailey 1976); more recent data lend credence to this possibility and allow a more penetrating look at the question. Before embarking on an examination of the more recent experimental information it is useful to look over the case in geological terms.

The popular view for a long time has been that

felsic alkaline rocks are products of differentiation from mafic parents at relatively shallow (usually crustal) depths. Processes such as fractional crystallization are still currently in vogue in spite of many difficulties (see for instance Yoder 1979; Bailey 1981) and discussions of felsic magmas in particular seem impervious to contrary evidence. When the various pieces of evidence are assembled together, however, the case for low-pressure differentiation as a universal mechanism for producing felsic magmas can be seen to be without foundation.

Essentially, evidence contrary to continuous differentiation at low pressure takes two forms: (1) felsic and intermediate volcanics carry ultramafic nodules; and (2) there are substantial composition and volume discrepancies in the igneous products of a given complex or province. As will be seen later, the first could be considered as the evidence from the high-energy ranges of the magmatic system and the second from the

lower-energy ranges.

Nodules

In addition to the dramatic examples of felsic lavas carrying ultramafic nodules (Wright 1966, 1969), mugearites and hawaiites are commonly reported as the hosts in ultramafic-nodule locali- ties (see Boettcher & O'Neil (1980) for an informative listing). High-speed eruption is essential, and these are unequivocal examples of the existence of felsic and intermediate melts in the mantle: it is clear that low pressure differentiation is not essential for the formation of magmas in the range hawaiite-trachyte-phonolite. In fact, the composition of these melts may be a direct indication of physical and chemical conditions in the mantle source.

Composition gaps

Daly (1925, 1927) first drew attention to the basalt-trachyte bimodality of oceanic lavas, a finding later supported by Barth (Barth et al. 1939) and statistically verified by Chayes (1963) who suggested that this raised doubts about the evolution of trachyte from basalt by continuous differentiation. Quite naturally, this suggestion provoked spirited opposition from advocates of fractional crystallization; the case was reviewed by Yoder (1973) who concluded that the gap was real and that this was not a realistic outcome of continuous fractional crystallization. More recent studies (e.g. Zielinski 1975) still invoke this mechanism, however, without apparent question, and so it is permissible to cite some additional

geological facts.

6 D.K. Bailey

1 In deeply exposed continental sections reveal- ing syenite plutons, comparable alkaline mafic and intermediate plutons are rare or absent. 2 Zoned plutons are rare, and where known are relatively small and generally considered to be composite (e.g. Monteregian Hills). 3 In keeping with 1 and 2, the common nodules in felsic volcanics are typically cognate, e.g. syenite in trachyte, peralkaline granite in com- endite.

To the above it may be added that the typical nodules in basanites are peridotite and gabbro. Hence it is hardly reasonable to argue that the basalt-trachyte bimodality in the ocean basins is an artefact of exposure level, eruption characteristics or sampling bias: the continental plutonic bodies and the nodule suites confirm the scarcity of intermediate magmas.

Perhaps the most vivid evidence of magmatic bimodality lies on Graciosa in the Azores. Here there have been alternating eruptions of felsic and mafic pyroclastics, without detectable time breaks, requiring the simultaneous coexistence of contrasting magmas within the one volcanic system (Maund & Bailey 1982; Maund, 1985). This is simply a graphic example of the contem- poraneous eruption of trachyte and basalt on oceanic islands generally, to which Daly called attention so long ago.

Other composition gaps are present in other magma associations, such as basanite-phonolite and nephelinite-nephelinitic phonolite, but it may be relevant that even among the felsic rocks themselves there seem to be marked distribution maxima, as between phonolites, trachytes and rhyolites (and the plutonic equivalents). These gaps, too, militate against continuous differentiation, tending to favour partial melting or at least multiprocess origins (see Bailey (1976) for a full discussion).

Volume discrepancies

Another major problem for continuous differentiation at low pressure lies in the volume relations of the magmas in many provinces. The usual image of a large alkali basalt centre with small spines and flows of trachyte may be reasonable for some oceanic volcanoes, but it is a dangerous generalization, even for the oceans, e.g. Azores and Canaries. On the continents it is probably the exception rather than the rule: here again we have deep sections providing plutonic evidence.

In major syenite provinces, such as Kola and Malawi, the basic and intermediate rocks required by continuous differentiation are not merely insufficient--they are lacking altogether.

Large syenite plutons (and large monotonic trachyte and phonolite volcanoes, as in the Kenya rift) are further evidence against low-pressure evolution of felsic magmas from a basaltic parent. Indeed, to suppose such differentiation has occurred when there is no sign of the earlier stages is to assume that syenite can form in only one way. In the volcanic regime the volume discrepancy can be seen in the Miocene-Recent activity of the East African rift zone where the calculated volumes of felsic and mafic volcanics are approximately equal (Williams 1972) and there have been regional floods of phonolite and trachyte. The latter are difficult to provide for in any process of magma generation but pose an acute problem for high-level differentiation (Bailey 1974, 1978).

It should be said, too, that detailed studies of particular trachyte and pantellerite suites have not only failed to find any links with basalt (Bailey 1978) but have even revealed distinctive differences between felsic volcanoes erupting contemporaneously in the same province. In the area around Lake Naivasha in Kenya, there are Holocene phonolite, trachyte, pantellerite, com- endite and basalt eruptions from overlapping centres that provide not only a series of major composition gaps but also distinctive major and trace-element patterns that are inexplicable in terms of continuous differentiation (Bailey & Macdonald, 1987). Clearly, there is commonality in that the magmas are part of a cycle of igneous activity, but the need is for a multisource/ multiprocess system of magma genesis.

Yoder's solution (1973) to the dilemma of the Daly gap was the production of two contrasting magmas by fractional melting, along lines indicated by Presnall (1969). This envisages a source composition with two invariant points which by continuous melt extraction during an igneous cycle yields two contrasting magmas. Melt generation from two distinct invariant points would clearly alleviate some of the problems of alkaline magmatism and would explain composition gaps, but the generation of large volumes of felsic magma from a peridotite mantle source remains a difficulty, and a new problem is introduced of irreversibility and timing. Either the source volume would have to keep changing or the liquid from the lower-temperature invariant point could be erupted only at the start of a melting cycle. These difficulties could be solved through replenishment of felsic components by metasomatism or other forms of enrichment. At the very least therefore Yoder's solution requires an open system with melting pulses acting on periodically enriched sources (cycles of melting and enrichment). Even so the volume problem remains, and

Mantle metasomatism 7

an additional factor is introduced by alternating mafic-felsic volcanism, exemplified by Graciosa, where there is effectively simultaneous eruption of felsic-mafic magmas poor in phenocrysts. Such activity implies the need for two (or more) distinctly different source compositions to be melted and tapped during the same igneous cycle. Geological evidence indicates that felsic sources richer than normal peridotite are feasible in the mantle and recent experimental evidence con- firms that appropriate mineral stabilities extend into the upper part of the mantle. When these are taken together with eruption characteristics, depth indicators in the nodule suites and evidence of metasomatism, it becomes possible to propose a scheme of magma genesis that can reconcile all the seemingly conflicting observations and give an integrated pattern to the activity.

Metasomatism and the development

of separate felsic sources

Peridotites containing minor amounts of plagio- clase, trachytes containing peridotite nodules, and sanidine-coesite eclogite nodules are some of the evidence indicating that felsic melts and sub-solidus felsic mineralogies are possible in the mantle, and it has been clear for some years that there is no a pr ior i reason to suppose otherwise (Bailey 1976). In view of the problems with felsic magmatism, it is appropriate to update the evidence on mineral stabilities and then to enquire whether and how two sources could develop, and how magmas could be generated and erupted.

In Figs 1 and 2 some of the relevant stability fields are plotted. Obviously the bulk mantle composition must influence some of these, but broadly speaking felsic minerals could exist to depths of about 100 kin, the exact limit depending also on the prevailing temperature. The special importance of the stability ranges shown in Figs 1 and 2 lies in the fact that the upper mantle within the felsic stability zone has the potential to yield melts of felsic character at a relatively low-temperature invariant point. No such invariant melting point is possible at greater depths, so that the first melts (from a phase assemblage containing no felsic minerals) must of necessity be different, and in all probability distinctly mafic. Thus the possibility of two distinct sources with completely different first melts is evident. Could their potential for producing contrasting magmas be established and enhanced by processes of enrichment and/or metasomatism ?

Starting from the critical observation that high

volatile activity is a hall-mark of alkaline mag- natism, it has been possible to develop a hypoth- esis relating magmatic variations to melting and metasomatism by volatile flux along different geothermal gradients (Bailey 1970, 1980a). Fur- ther consideration of the effects of the processes along low geothermal gradients has been given elsewhere (Bailey 1984, 1986). For the generation of felsic melts, it is necessary to look to activity along steeper geotherms more appropriate to oceanic conditions (or those away from continental craton nuclei). On steeper geotherms, such as those that would cros

[j. g. fitton, b. g. j. upton] alkaline igneous ro(bookfi.org)

Documents

e g e o

t i o n

alkaline rocks

b o u r n e

alkalic igneous rocks

s o c i e t y s p e

geological society of

potassiumrich igneous