Isotopic and Geochemical Investigation of the Chilwa Island ...asimonet/PUBLICATIONS/Simonetti...are followed inward by a ring of calciocarbonatite, ferrocarbonatite, and a central

0022-3530/94 $3.00

Isotopic and Geochemical Investigation of the Chilwa

Island Carbonatite Complex, Malawi: Evidence for a

Depleted Mantle Source Region, Liquid Immiscibility,

and Open-System Behaviour

by A. SIMONETTI AND K. BELL

Ottawa—Carleton Geoscience Centre, Department of Earth Sciences, Carleton University,Ottawa, Ontario K1S 5B6, Canada

(Received 28 September 1993; revised typescript accepted 11 March 1994)

ABSTRACTInitial Nd, Pb, and Sr isotopic data from carbonatites and associated intrusive silica-undersaturated

rocks from the early Jurassic, Chilwa Island complex, located in southern Malawi, central Africa,suggest melt derivation from a Rb/Sr- and Nd/Sm-depleted but Th/Pb- and U/Pb-enriched mantlesource. Initial 143Nd/14*Nd (051265-0-51270) isotope ratios from the Chilwa Island carbonatites arerelatively constant, but their initial 87Sr/86Sr (070319-070361) ratios are variable. The b^O^oy,(9-53—14- 15%o) and (513CpDB ( — 3-27 to — l-50%o) isotope ratios of the carbonates are enriched relativeto the range of mantle values, and there is a negative correlation between

1598 A.SIMONETTI AND K. BELL

continental mantle source, and provide a unique way of monitoring the evolution of themantle throughout much of geological time (Bell et al., 1982; Bell & Blenkinsop, 1987,1989).

The abundant carbonatite-nephelinite centres in the East African Rift Valley System areideal for such studies. Seventeen carbonatite complexes are located in southern Malawi andadjacent Mozambique, at the junction of the north-south-trending fault system of the EastAfrican Rift and east-west-trending fault system of the Zambezi Rift (Fig. 1; Garson, 1965,1966). This region also contains major intrusions of peralkaline granite and syenite, andsomewhat smaller complexes of nepheline syenite and syenite (Woolley & Jones, 1987).These intrusive centres, mainly early Jurassic in age, are unrelated to the modern rift system,and form part of the Chilwa Alkaline Province (CAP). A general account of this province,including its tectonic setting, has been given by Woolley & Garson (1970).

The Chilwa Island carbonatite ring-complex (Fig. 2), located within the CAP (Fig. 1) anddescribed in detail by Garson & Smith (1958), was the first carbonatite to be recognized inAfrica (Dixey et al., 1937). An age of 138±7 Ma (K-Ar, biotite; Snelling, 1965) has beendetermined for the carbonatite, and a titanite fission track age of 126±8 Ma (G. N. Eby,pers. comm., 1993) has been obtained for a microfoyaite.

Using Nd, Pb, Sr, O, and C isotopic ratios, as well as major and trace element data fromcarbonatite, associated intrusive silicate rocks, and host rocks from the Chilwa Islandcarbonatite complex, we attempt to (1) define the nature of the sub-continental mantlesource beneath the southern extremity of the East African Rift System, and (2) evaluate theevolutionary history of the complex, in particular, the relationship between the carbonatiteand the younger, intrusive, silica-undersaturated rocks.

GEOLOGY

The Chilwa Island carbonatite (Fig. 2) intrudes Precambrian basement, which consists ofa series of quartzo-feldspathic gneisses and hornblende granulites, and pre-Chilwa 'older'syenites. Fenitization accompanied carbonatite emplacement. The carbonatite and feniteaureole are separated by either an agglomerate and/or a brecciated potassic fenite, and theseare followed inward by a ring of calciocarbonatite, ferrocarbonatite, and a central core ofsideritic carbonatite (Fig. 2). Detailed petrographic descriptions of carbonatite samplesanalysed in this study are provided in the Appendix. The age of carbonatite intrusive activitydecreases from rim to core (Garson & Smith, 1958), and late-stage carbonatite activity atChilwa Island was accompanied by 'hydrothermal' activity, concentrated primarily in thecore region. Minerals such as fluorite, barite, quartz, galena, pyrite, Mn-rich oxides, anddickite are found in veins and as localized disseminations (Garson & Smith, 1958).

Five per cent of the complex (Fig. 2) exposed at the surface consists of plugs and dykes ofalnoite, camptonite, ijolite, nephelinite, nepheline syenite, and trachyte (Garson & Smith,1958; Garson, 1960). Although the intrusive relationships between the different phases ofcarbonate and silicate rocks at Chilwa Island are difficult to unravel (Garson & Smith, 1958),most of the silicate plugs and dykes cross-cut the main carbonatite, and are therefore youngerthan the carbonatite (Woolley & Jones, 1987).

CHEMICAL RESULTS

Carbonatites

Whole-rock major and trace element analyses for carbonatites and the quartz-fluorite-bearing sample, G 255, are given in Table 1. Four early carbonatite samples (G357, G 369, N1290, and N 1295) from the outer parts of the complex are characterized by lower contents of

THE CHILWA ISLAND CARBONATITE COMPLEX, MALAWI 1599

/ • /

/ /

f // /

/ /

/ /

/ /—

Zambia

Malawi

/ /

/

r

• iV

k *

J

///.>*•Illl

i l l —

B1"%>r

/ HA' ^

/ / ^ <

/ s\L

^ mily\illlldf XAsIIIIr /J

V\

i /

* Tanzania- - • . - . - • " •

Chilwa.Alkalln*

_>^Provlnc«f1 ^

V^lmtoiOcwm

V/y /

1rI

11J l• *•111\I1,5̂

JAAA/ i

y

V

\

///jIy

/

/

/

/

/

/

s

Lake Malawi<>

fI\ \ /XVXT

\

r ' l '

i4V/ /

w\\\.V̂

1I\*/

\0>//A

/

/

CfI•/

/

\ ^

\

S

V'/ // /V // /•

/ /•

V?

v-\1

/ />k= Mil =

^ / /

35°^|

Tl*>/A

•

/ •

/

/

/ ,

u

/uHl =ml ji =gMs

( /

, Pup\B/

<

K\\Y

N•

HMon

hIhf ///

illlllSJIII

/"

//s

/

*>(

>

VrI

/(\III J

MillrllllIII!

///

j£

Y)\\/

\X

X\

= Illl

) /1

/

>

f

<

\/ \

/

/111 =£1111yillp

V(rA in»

\)f

>

l o36°

\\

Illl

Illl i 7

=iw^fiiiF\^— mi j/r

/ °/ 15~

)^>*f' IslandC*Lake \>Chilwan

• J* I

// ( I

^^^JRi 16 —

i i

0 Km 50

FIG. 1. Inset shows location of the Chilwa Alkaline Province (CAP) at the southern extremity of the East African RiftValley System. Boxed area within inset is shown in the main diagram. Main diagram illustrates the regional geologyof Malawi and distribution of carbonatite centres (solid circles) within the CAP. Dashed area, Precambrianbasement; cross-hatched pattern, major intrusions of syenite and nepheline syenite; dark grey shaded area,Permian-Tnassic Karoo sedimentary rocks; thick lines, major fault lineaments; light grey shaded area, alluvium.

[Modified after Bloomfield (1965).]

MgO, A12O3, MnO, and SiO2, and higher CaO contents than the younger, centralcarbonatite phases (Table 1). Such trends are typical of differentiation observed in manyother carbonatites (Woolley & Kempe, 1989). Compared with the magnesio- (MW 166,dolomite-rich) and ferrocarbonatites (MW 168, siderite-rich), the total Fe content of theChilwa Island calciocarbonatites is lower and fairly constant (Table 1). The higher Mn and Siabundances found in the dolomitic and sideritic carbonatites can be attributed to late-stage

1600 A. S I M O N E T T I A N D K. BELL

FIG. 2. Geological map of the Chilwa Island carbonatite (modified after Garson & Smith, 1958). Units:1—Precambnan basement complex, includes pre-Chilwa 'older' syenites; 2—potassic breccia; 3—calciocarbona-tite; 4—magnesiocarbonatite; 5—ferrocarbonatite, 6—various intrusive silicate plugs; 7—various intrusive silicate

dykes; 8—major faults.

formation of secondary minerals such as Mn-rich oxides and quartz that characterize thecentral core of the complex (Garson & Smith, 1958).

The condensed spidergram (Fig. 3) for the Chilwa Island carbonatite samples shows thatthe Chilwa Island carbonatites are enriched in Ba, Nb, Nd, Sm, Sr, Zr, and Y, and depleted inK, Rb, and Ti relative to primitive mantle, chemical features that have been observed in mostcarbonatites (Nelson et al., 1988). The Chilwa Island carbonatites have comparable patternsto average calcio-, magnesio-, and ferrocarbonatites (Woolley & Kempe, 1989), and mantle-normalized values for Nd, Sm, U, Y, and Zr are also similar. The remaining elements shownin Fig. 3, however, plot at significantly lower mantle-normalized values, especially K and Ti.

Silicate rocks

Whole-rock major and trace element data from the olivine nephelinites, camptonites,nepheline syenites, and ijolites are given in Table 2. Petrographic descriptions of these rocks

T H E C H I L W A I S L A N D C A R B O N A T I T E C O M P L E X , M A L A W I 1601

TABLE 1

Major and trace element analyses of Chitwa Island carbonatites and quartz-fluorite-bearingsample (G 255)

SampleType

SiO2TiO2A12O3Fe2O3*MnOMgOCaONa2OK2OP 2O 5Sco2tTotalBaCrZrSrRbYNbZnNiV

G 357C

2-46O030075-890-26037

53-73000001121O01

4O221043

4400

1602 A. SIMONETTI AND K. BELL

10

10

10

10

10

10

1.0

0.1Ba Nb Tl Nd SmK U Rb Sr Zr

FIG. 3. Condensed mantle-normalized spidergram for Chilwa Island carbonatites. In addition, heavy lines representaverage calcio- (open square), magnesio- (solid square), and ferrocarbonatite ( + ) from Woolley & Kempe (1989).

Mantle values from Wood et al. (1979).

involves binary mixing between a primitive (A1-, Si-poor, Mg-rich) magma, such as theChilwa Island nephelinites, and a second component with major element concentrations(A12O3> 180 wt.%; MgO< 10 wt.%; SiO2>50 wt.%; TiO2 wt.% < 10 wt.%) that plot tothe right of the nepheline syenite field, such as lower-crustal granulites. Woolley (1987)attributed the origin of voluminous 'phonolitic' (nepheline syenitic) melts within the CAP todirect partial melting of the mantle, using arguments based on experimental data (Wright,1971), and the presence of lherzolite and olivine nodules in phonolites and trachytes (e.g.,Wright, 1971; Irving & Price, 1981). Compared with the olivine nephelinites, the higherincompatible element contents (e.g., Rb, Zr) in the nepheline syenites can be generated bypartial melting of an incompatible-element-enriched (metasomatized) source region(Woolley & Jones, 1987). Other evidence that might indicate a direct mantle origin for themore 'differentiated' Chilwa Island silicate rocks, such as the nepheline syenites and ijolites,are their low Y/Nb ratios (< 1), which are similar to those from ocean island basalts (Taylor& McLennan, 1985) and some mantle-derived A-type granitoids (Eby, 1990). In addition,the lack of any consistent trends between incompatible element ratios (e.g., K/U, Nb/U,Zr/Nb) and 'differentiation' indices [e.g., SiO2, (Na + K)/Al] are inconsistent with a simplefractionation-controlled origin from a single parental melt for the nepheline syenites.

The relationship between some of the silicate rocks and carbonate rocks at Chilwa Islandcan be evaluated by comparing calculated distribution coefficients for certain trace elementswith those determined by Hamilton et al. (1989) for nephelinite-calcium-rich carbonateliquid pairs. These distribution coefficients were based on experimental results fromconjugate silicate-carbonate liquids produced by liquid immiscibility in the pressure andtemperature range of 1-6 kbar and 1050-1250 °C, respectively. Enrichment of most of thetrace elements into the carbonate liquid is favoured by high pressure, low temperature, andincrease in polymerization of the silicate conjugate melt (Hamilton et al., 1989). The ranges

TABLE 2

Major and trace element analyses from Chilwa Island intrusive silicate rocks

SiO2TiO2A12O3Fe2O3FeO*MnOMgOCaONa2OK2OP 2 O 3STotalmg-no.

BaCrZrSrRbYNbZnNiV

G 352ON

38-215-08

11-438-288-280-259-64

11-560-692-531-07O03

97-8067

2410290317

28705840

1011298328

G 361ON

36172-46

11-055-555-560238-92

14-081-832-931 52O10

91 3074

1470416273

443011638

177107216119

G 399ON

38-554-93

11-488158-150269-50

11-490932-631-06O04

97-7068

1170270314

20007341

1011187933

G 298CP

41-373-37

13-614-629-390196-719-583-582-22066003

95-7056

842112198

12805235619863

160

G396CP

41-093 21

12-894-799-730199-30

10262-671-98091003

97-4063

849237247

1380402473976951

MW 149NS

49-89032

19-323071-73029046339

1O975-43003OOO

95-5032

7229

15101980

17743

205176210

MW 150NS

5064044

18-303-281-840340433-409-745-98OOOOOO

95-1029

7990

22301880

19165

339171

80

G 315NS

5O11039

19-503-841-57028O302-99

10555-61OOOO01

95-8025

3960

13902560

19963

446135

180

G 322NS

46171-57

20443-702-790311-536019195-560680O1

98-7049

18300

4982800

14145

345115

160

MW 167IJ

45-54038

15-182-45061034045

22O07-293-69012O01

98-3057

31118

459449

514066264

119

1152IJ

36-792-448-58

12-783-20043056

22-253-252-65008002

93-9023

10400

2200897

521772946221

1340

MW 164AL

13-712-232-875O0

10150336-52

25-770781-665-79015

761053

21500

8754700

46149621281

58161

HXmn

VIr>zDn>70aOz>HHmoO2•ot -mx

See Table 1 footnotes for elemental uncertainties. ON, olivine nephelinite; CP, camptonite; NS, nepheline syenite; IJ, ijolite; AL, alnoite. * FeO contents arebased on titrated values from analyses of identical samples (Woolley & Jones, 1987).


20 -

10 >-m

20 •

tm• r

10 ,-

n S-

- r

>henrWtM

ownptonlte*

•

•

•

A

B

mphsfne •yvnlto* &

•

Al2O3t%

2

w c«o »•%V MgO w(%

36 40 44 48 52

FIG. 4. Harker variation diagrams of major oxides for intrusive silicate rocks at Chilwa Island (Table 2). (A) A12O3and TiO2 wt.% vs. SiO2 wt.%. (B) CaO and MgO wt.% vs. SiO2 wt.%.

of calculated distribution coefficients (concentration in nephelinite: concentration incarbonatite)forBa (055-1-2), Mn (~ 10), Sm (0-91—1-1), and Zr (071-2-2) from the ChilwaIsland olivine nephelinites and the least altered calciocarbonatites are similar to valuesobtained from the 3-kbar runs of Hamilton et al. (1989). A possible origin for thecarbonatites by liquid immiscibility from an olivine nephelinite, at crustal levels, is consistentwith our data.

Host rocks and fenites

All of the exposed basement granulites and 'older' syenites at Chilwa Island are slightlyfenitized (A. R. Woolley, pers. comm., 1992). The major and trace element composition oftwo (MW 4 and MW 95) of the granulites are given in Table 3.

The degree of fenitization at Chilwa Island increases towards the carbonatite plug, and thefenites can be divided into three main types (Woolley, 1969): quartz fenite (MW 30c, leastfenitized), syenite fenite (M W 65 and MW 27, intermediate), and feldspathic breccia (MW 71and MW 147, referred to as potassic fenites in Tables 3-5 and Figs. 5-7). The feldspathicbreccia (sample MW 71), which is predominantly orthoclase (>95%), reflects the highestdegree of fenitization. Compared with the relatively non-fenitized granulites, the fenites havemuch higher SiO2, K2O, Zr, and Rb contents, and lower TiO2, total FeO, MgO, CaO, andCr contents.

Summary

Variations in major and trace element data from the Chilwa Island carbonatites can belargely attributed to magmatic differentiation with superimposed minor effects (e.g.,


TABLE 3

Major and trace element analyses from fenites and hostgranulites

SampleRock type

SiO2TiO2A12O3Fe2O3*MnOMgOCaONa2OK2OPjO,STotal

BaCrZrSrRbYNbZnNiV

MW4G

50-292-70

14-9514-900-242-707-973-961-531-450-05

1011

1240109482759

217937

20914


Re filament technique, and Pb and U samples were run on single Re filament using silica geland phosphoric acid. All isotope ratios were measured on a Finnigan-Mat 261 multicollectorsolid-source mass spectrometer operated in the static mode. Reproducibility of the Sr and Ndisotope ratios is ±0004%, and for Pb isotope ratios ±0-1%.

ISOTOPE RESULTS

Radiogenic isotopes

Nd, Pb, and Sr isotope results are presented in Tables 4 and 5. Attempts to date thecarbonatite samples using Rb-Sr and Sm-Nd isotope systematics were unsuccessful becauseof the restricted spread in Rb/Sr and Sm/Nd ratios.

Initial Nd and Sr isotope ratios, assuming an age of 130 Ma, are plotted in Fig. 5. With theexception of one ijolite, the data from all of the intrusive rocks plot in the upper left (depleted)quadrant, consistent with the data from other carbonatites of similar age (110 Ma Oka, Wenet al., 1987; 97 Ma Magnet Cove, Tilton et al., 1987; 123 Ma Kangankunde, Ziegler, 1992).The most depleted initial 143Nd/144Nd and 87Sr/86Sr isotopic ratios from the Chilwacarbonatites are similar to those for other young carbonatites associated with the Eastern rift(e.g., Tororo, Bukusu, Sukulu; Bell & Blenkinsop, 1987; Napak, Simonetti & Bell, 1994)some of which were used to define one extreme of the East African Carbonatite Line (EACL,Bell & Blenkinsop, 1987), a Nd-Sr isotopic array based on data mainly from young (


TABLE 4A

Nd and Sr isotopic ratios of Chilwa Island carbonatites and quartz-fluorite-bearing sample

Sample

G357N 1295N 1290G369G410G413MW 166MW 168G255

Type

CCccpADS

QF

Rb(ppm)

0280351-723-932-01O28035

1068021

Sr(ppm)

3208110499

1313217124768211673813242510

87Sr/86Srmeasured

070328070334070349070323070319070334070342070365070364

slSr/S6Srinitial

070328070334070349070322070319070334070342070361070363

Nd(ppm)

186564612220114723

132530643041

Sm(ppm)

1874963217

128187321506

'*3Nd/'**Ndmeasured

051270051274051276051275051275051277051276051272051279

initial

051265051267051268051268051267051268051266051266051270

Isotope results normalized to " " N d / ' ^ N d = 07219 and 86Sr/88Sr = O1194. NBS 987 Sr standard =O71025±O00002; La Jolla Nd standard = 051186±000002; and BCR-1 143Nd/144Nd = O51266±O00002.Uncertainties are given at the 2a level. Uncertainty of Nd and Sr abundances, determined by isotope dilution[details given by Bell et al. (1987)], is ± 0 5 % . Carbonatite types: C, calcitic; P, pyrochlore-rich; A, ankeritic; D,dolomitic; S, sideritic; QF, quartz-fluorite-bearing sample.

TABLE 4B

Nd and Sr isotopic ratios of Chilwa Island intrusive silicates, granulites, andfenites

Sample

G352G361G399G 298G 396MW 164G315G322MW 149MW 150MW 167I 152

M W 4MW95MW 115

MW27MW65MW71MW 147MW30c

Type

ONONONCPCPALNSNSNSNS

II

GGG

SFSFPFPFQF

Rb(ppm)

6486-277-029-015-338-2

163183-6

195252

35-7551

21506104

104-267-2

186-9136-693-8

Sr(ppm)

315041702120

958-7861-3

40702060224020501850477-7947-3

7605576-01821

533-7484-6689-9353-2348-0

s"Sr/B6Srmeasured

"Sr/^Srinitial

Intrusive rocks070376070360070387070374070376070331070410070404070427070422070391070360

070513070576070443

070724070708070527070616071311

070364070348070366070356070366070326070365070383070373070344070351070327

Cranulites070498070525O704I2Fenites070612070634070371070412071167

Nd(ppm)

1351071294O655-3

17253-7

167909611-559-9

87-881159-6

46-9104-9

11534-9721

Sm(ppm)

12 416-320

8-010236-58-6

2213132-5

19-7

18-718 26-9

8-018 66-07-5

106

"'Nd/'^Ndmeasured

051274051273051271051273051272051279051270051275051268051273051257051280

051244051245051106

051252051242051279051271051207

'"Nd/'**Ndinitial

051269051265051263051263051263051268051262051268051260051265051246051263

O51234051233051100

051243051236051252051260051200

See Table 4A footnotes. ON, olivine nephelinite; CP, camptonite; AL, alnoite; NS, nepheline syenite; I, ijolite; G,granulite; SF, syenitic fenitc; PF, potassic fenite; QF, quartz fenite.


granulites, but must have formed from a different protolith. The Nd and Sr isotopic ratiosfrom the fenites (Fig. 5) show that with increasing degree of fenitization (quartz-to-syenitic-to-potassic fenite), the isotopic ratios shift towards the depleted isotope signature of themagmatic rocks.

The present-day 2 0 7Pb/2 0 4Pb-2 0 6Pb/2 0 4Pb ratios from the carbonatites (Fig. 6A) form anear-linear array which, if interpreted as a secondary isochron, would yield an age of 628 Ma,significantly older than the best estimates of the age of the complex based on the K-Ar dates.The present-day 2 0 8Pb/ 2 0 4Pb- 2 0 6Pb/ 2 0 4Pb isotopic ratios shown in Fig. 6B are scattered.The data from one carbonatite sample, G 410, are excluded from Fig. 6B because it has anextremely radiogenic present-day 2 0 8Pb/2 0 4Pb ratio, which may indicate the presence of aTh-rich mineral, such as pyrochlore (Hogarth, 1989).

An interesting feature, however, now emerging from many carbonatites, relates to thelinear arrays between initial 2 0 7Pb/2 0 4Pb and 2 0 6Pb/2 0 4Pb isotopic ratios, which ifinterpreted as secondary isochrons, would yield ages older than the geological age of thecarbonatite (Kwon et al., 1989). Regression lines through the initial2 0 7Pb/ 2 0 4Pb- 2 0 6Pb/ 2 0 4Pb isotopic arrays from young carbonatite complexes are similar inslope to the array defined by mid-ocean ridge basalts (MORBs) and ocean island basalts(OIBs) (Tatsumoto, 1978), and correspond to a secondary isochron age of ~ 1600 Ma. Suchlinear arrays may fit the mean slope of the oceanic regression line at different times in the past(Kwon et al., 1989). The initial 2 0 7Pb/2 0 4Pb and 2 0 6Pb/2 0 4Pb ratios from all but one of theintrusive rocks from Chilwa Island fall to the right of the Stacey-Kramers (1975) Pbevolution curve (Fig. 7), a feature that is also shown by Pb isotopic data from East African

TABLE 5A

Pb isotopic ratios of Chilwa Island carbonatites and rocks (minerals) of hydrothermal origin

Sample

G357

N 1295N 1290

G369

G410

G413

MW166

MW168

G255Galena

Type

CDCCDRLCDPDAD

D-lD

SD

QF

U(ppm)

no*

1-7*1-7*1-8*

21-424-0*

8-0*


TABLE 5B

Whole-rock Pb isotopic ratios of Chilwa Island intrusive silicates, granulites, andfenites

Sample

G352

G 361G399G298G 396G315G322MW149

MW150

MW164I 152MW167

MW4MW95MW115

MW27

MW65MW 71MW147MW30c

Type

ONRL

ONONCPCPNSNS

NSDRL

NSD

ALI

I

GG

G

SFD

SFPF

PFQF

U{ppm)

1-5

3-51-51-6


0.5128 -

0.5126 -

0.5124 -

0.5122 -

0.5120 -

0.7030

I I

I 1

i

O cvboratiln• qu»rtl-fluortt» roc*• intrutlv* •Sola*$ fenitn^ granutttM

CHUR

V—•1 1 1 1 1 1

0.7040 0.705087O /86Sr/ Sr

0.7060 0.7070

FIG. 5. Plot of initial ' " N d / ' ^ N d vs. 87Sr/86Sr isotopic ratios for all samples analysed from Chilwa Island. CHURand Bulk Earth values are corrected for Rb (87Rb/86Sr = 0-083) and Sm ( l 4 7 Sm/ ' "Nd = CH967) decay frompresent-day values of 0512638 and 0-7045, respectively, to 130 Ma ago. Fenite types: Q—quartz fenite; S—syeniticfenite; P—potassic fenite. The East African Carbonatite Line (EACL) is shown (Bell & Blenkinsop, 1987). It shouldbe cautioned that the EACL is based on data from young (0-40 Ma) carbonatites, and only the slope can be used for

comparative purposes.

Kaiserstuhl, Germany; Schleicher et a/., 1990). Compared with the Pb isotopic ratios of theintrusive silicate rocks, the higher 207Pb/204Pb ratios from the carbonatites suggest that thelatter underwent interaction with a different, more radiogenic reservoir, possibly upper crust.Such interaction may have occurred during late-stage hydrothermal activity.

The two granulites with the most depleted 20


15.9 -

18.0 20.0 22.0 24.0

FIG. 6. Plots of present-day Pb isotopic ratios for Chilwa Island samples. Open squares and curve representStacey-Kramers (1975) Pb evolution curve. Time interval of curve is 200 Ma. Other symbols as in Fig. 5. (A)2 0 7Pb/2 0 4Pb vs. 206Pb/20*Pb. Line labelled 130-0 Ma shows the slope generated by radiogenic 2 0 7Pb/2 0 6Pb in a

closed system over the specified time interval. (B) 2 0 8Pb/2 0 4Pb vs. 2 0 6Pb/2 0 4Pb.

greater, by a factor of 1-50, than the 0 1 % estimate we normally attribute to duplicateanalyses. Some silicate rocks (Table 5B) show less variation in their duplicate Pb isotopicratios than those from the carbonatites, and this may reflect the fact that the silicate rockswere not subjected to any late-stage hydrothermal activity, possibly because of their youngeremplacement ages. The fact that the duplicate Pb isotopic analyses from the youngerferrocarbonatites (MW 166 and MW 168) (Table 5A) do not plot on the array defined by thecarbonatite samples, and their similar Pb isotopic ratios to a galena specimen (no. 657) fromthe northern spur region (Table 5A) of Chilwa Island (Garson & Smith, 1958), suggests thatthe fluids associated with the hydrothermal activity had variable Pb isotopic compositions.

Leaching experiments (Tables 5A and 5B) on three samples were undertaken to evaluatethe role, if any, of subsolidus alteration subsequent to 'hydrothermal' activity. Two of thesamples showed highly reproducible Pb isotopic ratios (N 1290 and MW 149). The resultsfrom the leaching experiments given in Tables 5A and 5B show that the Pb isotopiccompositions of the leachates can either be less (N 1290 and M W 149) or more radiogenic (G352) than their corresponding whole rocks. In addition, the leachate and residuecompositions for samples N 1290 and MW 149 are similar to the whole-rock and duplicateratios (Table 5A), suggesting that little or no subsolidus alteration has affected these samples.

1612 A. SIMONETTI A N D K. BELL

15.75

15.65

15.55

oo

•it ^ OT3 S> _

/

•o

ocm

i

•

i i i

18.0 19.0

^ P b / ^ P b

20.0 21.0

FIG. 7. Plot of initial 2O7Pb/2O*Pb vs. 206Pb/2


TABLE 6

Stable carbon and oxygen isotopic compositions fromcarbonatites

Sample

G 357

N 1290

G 369

G410

G413

MW 166

MW 168

Mineral

CalciteDolomiteCalciteDolomiteCalciteDolomiteCalciteDolomiteCalciteduplicateDolomiteduplicateDolomiteduplicateSiderite


processes include exchange with (518O-rich hydrothermal fluids, influx of meteoric water, orisotope exchange at low temperatures (Deines, 1989).

Plots of


similar to the source that produced the Kangankunde carbonatite complex (Malawi), alsolocated within the CAP and also of similar age (Ziegler, 1992). Our finding now extends thedepleted sub-continental mantle below a considerable segment of east and central Africa,stretching from Uganda in the north to Malawi in the south.

Isotopic data from carbonatites of similar age to Chilwa Island are shown in Fig. 10. Nd

0.5130

0.5128

0.5126

0.5124

0.5122

CHUR

mI I I J_

0.703 0.704 0.705

Sr/ Sr

0.706

FIG. 10. Plot ofinitialNd vs. initial Sr for carbonatites ~130Maold.K—Kangankunde (Ziegler, 1992),P—PandaHill (Morisset, 1992), C—Chilwa Island (this work). Complexes outside of Africa: M—Magnet Cove, USA (Tilton

et al., 1987); O—Oka, Canada (Wen et al., 1987); J—Jacupiranga, Brazil (Roden et al., 1985).

and Sr isotopic ratios from the Panda Hill carbonatite, Tanzania (Morisset, 1992), whichmay be slightly younger (116 Ma; Snelling, 1965) than Chilwa Island indicate derivationfrom a mantle source region that was depleted but not as depleted as the one that generatedthe parental melts for Chilwa Island and Kangankunde. Also shown in Fig. 10 are theisotopic data for carbonatites from outside of the African continent; two of these, Oka(Canada) and Magnet Cove (USA), are derived from a depleted mantle. The Jacupirangacarbonatite, Brazil (Roden et al., 1985) is one of the few carbonatites that has Nd and Srisotope signatures characteristic of enriched mantle source regions.

The isotopic ratios from the Chilwa Island carbonatites and silicate rocks show someinteresting differences. The carbonatites have variable initial 87Sr/86Sr isotope ratios andconstant 143Nd/144Nd ratios, unlike the intrusive silicate rocks that show pronouncedvariations in both ratios (Fig. 5). In addition, the fact that the initial Pb isotopic ratiosbroadly cluster into two groups, one consisting of carbonatites and fenites, the other ofintrusive silicates (Fig. 7), suggests independent evolution for the carbonatite and silicatemagmas.

The large spread (0-70319-0-70361) in initial Sr ratios shown by the carbonatites, coupledwith constant Nd isotopic composition, is typical of interaction with groundwater. Thenegative correlation between whole-rock carbonatite initial 87Sr/86Sr and


central carbonatites (MW 166 and MW 168) and quartz-fluorite-bearing sample (G 255)known to have undergone the highest degree of hydrothermal replacement have higher Srisotope ratios than the calciocarbonatites. On the basis of the distinct Nd, Pb, and Sr isotopiccomposition of the quartz-fluorite-bearing sample (G 255) and the fenites, it appears that thehydrothermal fluids and those fluids responsible for fenitization are unrelated. The fact thatthe Pb isotope ratios from the fenites lie close to the values from the carbonatites (see Fig. 7) isconsistent with the derivation of the fenitizing fluids from the carbonatites.

The Nd and Sr isotopic ratios from the intrusive silicate rocks are not uniform and plotalong a negative linear array in a Nd-Sr plot (Fig. 5), suggesting open-system behaviour.Open-system processes that can explain such isotopic variations include combinedassimilation and fractional crystallization (AFC, DePaolo, 1981), and binary mixingbetween either mantle and crustal components or two mantle components. Each of thesemodels is discussed in turn.

Assimilation-fractional crystallization

It is assumed that assimilation and fractional crystallization involved a parental olivinenephelinite magma and several different crustal end-member contaminants. The sameparental olivine nephelinite magma with the following elemental abundances and isotopiccharacteristics (average from three olivine nephelinite samples, Table 2) was used:87Sr/86Sr = 0-70348; Sr = 3146 ppm; 143Nd/144Nd=0-51265; Nd = 124 ppm. The threecontaminants that were chosen were (1) depleted granulite (MW 115), (2) slightly fenitizedgranulite (average of samples MW 4 and MW 95), and (3) quartz fenite (MW 30c, the leastfenitized 'older' syenite). Although the Nd-Sr array of Fig. 5 was approximated whenfenitized granulites were used as the assimilant (at R = 0-4, mass of the assimilant/mass oforiginal magma), the AFC model calculations were incapable of reproducing the Nd and Srelemental variations using appropriate partition coefficients (£>Sr. Dm < 1 0). Other problemsassociated with the AFC model calculations include the lack of documented partitioncoefficients for mineral phases in nephelinites, and the fact that the partition coefficients willprobably change as a result of open-system behaviour (Nielsen, 1989).

Binary mixing

Crust-mantle

Nd, Pb, and Sr isotope results from binary mixing modelling, using the same parentalmagma and contaminants as in the AFC modelling, show that mixing between an olivinenephelinite magma and fenitized mafic granulites better fits the data than in the case of theAFC model. The results of the binary mixing calculations are shown in Figs. 11 and 12.Although binary mixing explains many of the data, the calculations show that a large amountof fenitized granulites (a mixture of 30-40% fenitized granulites and 60-70% of the originalparental magma) is needed to produce the most evolved rock types, the nepheline syenites.Such large amounts of contaminant, however, impose serious constraints on any model thatinvolves bulk assimilation. An alternative model invokes selective assimilation ofincompatible- and REE-rich mineral phases, such as plagioclase, from lower-crustalgranulites. Ponding of the magma (Duda & Schmincke, 1985) at the base of the lower crustmay initiate interaction between nephelinitic melt and granulites, and such a model was alsoproposed to explain the Nd and Sr isotopic ratios from phonolitic lavas (the volcanicequivalents of nepheline syenites) from the Shombole volcano, Kenya (Bell & Peterson,1991). The lack, however, of any correlation between Nd, Pb, and Sr isotopic ratios and any

THE C H I L W A I S L A N D C A R B O N A T I T E C O M P L E X , MALAWI

0.51270

1617

0.5126 -

0.5125 -

0.5124 -

0.51230.703 0.704 0.705

' IFIG. 11. Plot of initial Nd vs. initial Sr isotope ratios from intrusive silicate rocks. Open squares and adjoining curverepresent results of binary mixing between parental olivine-nephelinite magma and fenitized granulites (see text for

discussion). Squares along the binary mixing curve are at 10% intervals.

XIQ.

20.0

19.0

18.0

• •

• • ? •

•-

1 1

^ ^ Q _ _ ^ ,F«nittMd /'

^~~^ C ] . . . - - ' ' '

1 1 1

0.703 0.704 0.705

8 7 S r / 8 6 S r ,

0.706

FIG. 12. Plot of initial 2 0 6Pb/2 0 4Pb vs. initial Sr isotopic ratios from the intrusive silicate rocks at Chilwa Island.Symbols as in Fig. 11.


major [e.g., SiO2 wt.%, (Na + K)/Al] or trace element ratios (e.g., Rb/Sr, Sm/Nd, Nb/U,K/U, Zr/Nb, Y/Nb, and K/Rb) suggests that the silicate rocks were not derived by binarymixing between a parental melt and lower crust. The data are consistent with a model thatinvolves open-system behaviour for individual mantle-derived melts.

Mantle mixingThe Pb isotopic arrays from Chilwa Island shown in Figs. 6 and 7 plot close to the isotopic

data from MORBs and OIBs, but their respective regression lines have different slopes. Thesimilarity of the slope of the Pb isotopic arrays from several young East African carbonatitesto the oceanic regression line has been interpreted as showing mixing between distinct mantlecomponents, such as a large-ion lithophile element (LILE) depleted source and ametasomatic fluid with a high 207Pb/20*Pb (Grunenfelder et al., 1986) or betweenHIMU-like and EM Mike mantle components (Simonetti & Bell, 1994). The different slopeof the Chilwa Island Pb-Pb isotopic array shown in Fig. 7 may be attributed to the mixing oftwo mantle components different from those responsible for the generation of other Africancarbonatite complexes.

Mixing between two mantle components, however, is inconsistent with the similar Nd(0-51262-0-51263) and Sr (0-70327-0-70365) isotopic ratios shown by samples G 315 and I152, the two non-nephelinitic intrusive silicate rocks that plot at the extreme positions of thePb-Pb isotopic array shown in Fig. 7. The fact that the Nd and Sr isotopic data from all of theintrusive rocks plot within the depleted quadrant (Fig. 5) indicates their derivation from asimilar depleted mantle source. The lack of any correlation between major and trace element'differentiation' indices and isotopic ratios suggests that each silicate plug or dyke probablyrepresents a distinct mantle-derived partial melt from the same source. The lesser radiogenicNd and more radiogenic Sr isotopic ratios from the non-nephelinitic silicate rocks may beattributed to crustal contamination during their ascent. The Pb isotopic data for theintrusive silicate rocks shown in Fig. 7 are consistent with the involvement of the magmaswith a depleted component, such as lower-crustal granulites.

CONCLUSIONS

Our findings underline a complicated evolution for the Chilwa Island carbonatitecomplex, involving crustal contamination, groundwater interaction, and magma differentia-tion. We conclude that the many processes which have affected such a fluid-rich system canonly be unravelled using data from several isotope systems.

The new Nd and Sr isotopic data from Chilwa Island show that the parental melt for thecarbonatite was derived from an ancient depleted sub-continental mantle source, similar tothe one that generated the carbonatite melts at Kangankunde (Ziegler, 1992). The new datafrom Chilwa Island extend the known distribution of the depleted sub-continental mantlesource in Africa southward from northern Uganda to Malawi.

On the basis of their Nd and Sr isotopic ratios and trace element chemical data, we proposethat the Chilwa Island carbonatites and olivine nephelinites are related by liquidimmiscibility at low pressures. The carbonate melt then differentiated to produce thedifferent carbonatite phases, as well as the fluids responsible for the extensive fenitizationsurrounding the Chilwa Island complex. Subsequent interaction between late-stagehydrothermal fluids involving meteoric water and the carbonatites is indicated by thenegative correlation between Sr and the O isotopic composition, and the variation in Pb andSr isotopic compositions at constant Nd isotope values. These fluids were markedly differentfrom those responsible for fenitization.


On the basis of the chemical and isotopic data from the Chilwa Island intrusive silicaterocks, we feel that the most likely model is one in which each intrusion represents a distinctmantle-derived (depleted source) melt that subsequently underwent interaction with lower-crustal granulites. Their post-carbonatite time of emplacement, small volumetric propor-tions, and lack of consistent chemical differentiation trends support this interpretation.

In spite of the complex history surrounding the Chilwa Island complex, the similarity ofthe initial Nd and Sr isotopic signatures for the carbonatites to others of similar age suggeststhat the source isotope signatures have been retained.

ACKNOWLEDGEMENTS

We thank A. R. Woolley for kindly providing the samples used in this study. We alsothank A. R. Woolley and G. N. Eby for reviews of the manuscript, invaluable information onfield relationships, and insightful discussions on the alkaline magmatism associated with theChilwa alkaline province. Helpful reviews by Drs. T. Andersen and H. Schleicher improvedthe manuscript.

REFERENCES

Andersen, T., 1984. Secondary processes in carbonatites: petrology of 'rodberg' (hematite-calcite-dolomitecarbonatite) in the Fen central complex, Telemark (South Norway). Lithos 17, 227—45.1987. Mantle and crustal components in a carbonatite complex, and the evolution of carbonatite magma: REE

and isotopic evidence from the Fen complex, southeast Norway. Chem. Geol. 65, 147-66.Bell, K., & Blenkinsop, J., 1987. Nd and Sr isotopic compositions of East African carbonatites: implications for

mantle heterogeneity. Geology 15, 99-102.1989. Neodymium and strontium isotope geochemistry of carbonatites. In: Bell, K. (ed.) Carbonatites:

Genesis and Evolution. London' Unwin Hyman, 278-300.Cole, T. J. S., & Menagh, D. P., 1982. Evidence from Sr isotopes for long-lived heterogeneities in the

upper mantle. Nature 298, 251-3.Kwon, S. T., Tilton, G. R., & Sage, R. P., 1987. Age and radiogenic isotopic systematics of the Borden

carbonatite complex, Ontario, Canada. Can. J. Earth Sci. 24, 24-30.Peterson, T. D., 1991. Nd and Sr isotope systematics of Shombole volcano, East Africa, and the links betweennephelinites, phonolites, and carbonatites. Geology 19, 582-5.

Bloomfield, K., 1965. The geology of the Zomba area. Bull. Geol. Suro. Malawi 16, 193 pp.Carter, S. R., Evensen, N. M., Hamilton, P. J , & O'Nions, R. K., 1978. Neodymium and strontium isotope

evidence for crustal contamination of continental volcanics. Science 202, 743^-6.Cohen, R. S., O'Nions, R. K., & Dawson, J. B., 1984. Isotope geochemistry of xenoliths from East Africa:

implications for development of mantle reservoirs and their interaction. Earth Planet. Sci. Lett. 68, 209-20.Deines, P., 1989. Stable isotope variations in carbonatites. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution.

London: Unwin Hyman, 301-59.DePaolo, D. J., 1981. Trace element and isotopic effects of combined wall-rock assimilation and fractional

crystallization. Earth Planet. Sci. Lett. 53, 189-202.Dixey, F., Smith, W. C , & Bisset, C. B., 1937. The Chilwa Series of southern Nyasaland; a group of alkaline and

other intrusive and extrusive rocks and associated limestones. Bull. Geol. Surv. Nyasaland, 5, 82 pp.Duda, A., & Schmincke, H-U., 1985. Polybancdifferentiation of alkalic basaltic magmas: evidence from green-core

clinopyroxenes (Eifel, FRG). Conlr. Miner. Petrol. 91, 340-54.Eby, G. N., 1990. The A-type granitoids: a review of their occurrence and chemical characteristics and speculations

on their petrogenesis. Lithos 26, 115-34.Friedman, I., & O'Neil, J. R., 1977. Compilation of stable isotope fractionation factors of geochemical interest. In:

Data of Geochemistry, 6th edn. US Geol. Suro. Paper 440-KK, 12 pp.Garson, M. S., 1960. The geology of the Lake Chilwa area. Bull. Geol. Surv. Nyasaland 12, 67 pp.

1965. Carbonatites in southern Malawi. Bull. Geol. Surv. Malawi 15, 128 pp.1966. Carbonatites in Malawi. In: Tuttle, O. F., & Gittins, J. (eds.) Carbonatites. New York: Wiley-Inter-

science, 33-71.Smith, W. C , 1958. Chilwa Island. Mem. Geol. Sure. Nyasaland 1, 127 pp.

Grunenfelder, M. H., Tilton, G. R., Bell, K., & Blenkinsop, J., 1986. Lead and strontium isotope relationships in theOka carbonatite complex, Quebec. Geochim. Cosmochim. Ada 50, 461-8.

Hamilton, D. L., Bedson, P ,& Esson,J., 1989. The behaviour of trace elements in the evolution of carbonatites. In:Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, 405—27.

1620 A. S I M O N E T T I A N D K. BELL

Hoefs, J., 1987. Stable Isotope Geochemistry. Berlin: Springer-Verlag, 241 pp.Hogarth, D. D., 1989. Pyrochlore, apatite and amphibole: distinctive minerals in carbonatite. In: Bell, K. (ed.)

Carbonatites: Genesis and Evolution. London: Unwin Hyman, 105-48.Irving, A. J., & Price, R. C , 1981. Geochemistry and evolution of lherzolite-bearing phonolite lavas from Nigeria,

Australia, East Germany and New Zealand. Geochim. Cosmochim. Ada 45, 1309-20.Keller, J., & Hoefs, J., 1995. Stable isotope characteristics of recent natrocarbonatites from Oldoinyo Lengai. In:

Bell, K., & Keller, J. (eds.) Carbonatite Volcanism: Oldoinyo Lengai and the Petrogenesis of Natrocarbona-tites. IA VCEI Proceedings in Volcanol., 4.

Kwon, S. T., 1986. Pb-Sr-Nd isotope study of the 100 to 2700 Ma old alkalic rock-carbonatite complexes in theCanadian Shield: inferences on the geochemical and structural evolution of the mantle. Unpublished Ph.D.Thesis, University of California, Santa Barbara, 242 pp.Tilton, G. R.,&Grunenfelder, M. H., 1989. Lead isotope relationships in carbonatites and alkalic complexes:an overview. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, 360-87.

McKenzie, D., 1985. The extraction of magma from the crust and the mantle. Earth Planet. Sci. Lett. 74, 81-91.Menzies, M. A., & Murthy, V. R., 1980. Nd andSr isotope geochemistry of hydrous mantle nodules and their host

alkalic basalts: implications for local heterogeneities in metasomatically veined mantle. Ibid. 46, 323-34.Morisset, N., 1992. Stable isotope and radioisotope geochemistry of the Panda Hill carbonatite, Tanzania.

Unpublished M.Sc. Thesis, Carleton University, Ottawa, Ont., 91 pp.Nelson, D. R.,Chivas, A. R., Chappell, B. W., & McCulloch, M. T., 1988. Geochemical and isotopic systematics in

carbonatites and implications for the evolution of ocean-island sources. Geochim. Cosmochim. Ada 52, 1-17.Nielsen, R. L., 1989. Phase equilibria constraints on liquid lines of descent generated by paired assimilation and

fractional crystallization: trace elements and Sr and Nd isotopes. J. Geophys. Res. 94, 787-94.Richardson, S. H., Gurney, J. J., Erlank, A. J., & Harris, J. W., 1984. Origin of diamonds in old enriched mantle.

Nature 310, 198-202.Richter, F. M., 1985. Simple models for trace element fractionation during melt segregation. Earth Planet. Sci. Lett.

77, 333-44.Roden, M. F., Murthy, V. R., & Gaspar, J. C , 1985. Sr and Nd isotopic compositions of the Jacupiranga

carbonatite. J. Geol. 93, 212-20.Roeder, P. L., & Emslie, R. F., 1970. Olivine-liquid equilibrium. Contr. Miner. Petrol. 29, 275-89.Rosenbaum, J., & Sheppard, S. M. F., 1986. An isotopic study of siderites, dolomites and ankerites at high

temperatures. Geochim. Cosmochim. Ada 50, 1147-50.Schleicher, H., Keller, J., & Kramm, U., 1990. Isotope studies on alkaline volcanics and carbonatites from the

Kaiserstuhl, Federal Republic of Germany. Lithos 26, 21-3.Sharma, T., & Clayton, R. N., 1965. Measurement of O1 8 /O1 6 ratios of total oxygen of carbonates. Geochim.

Cosmochim. Ada 29, 1347—53.Simonetti, A., & Bell, K., 1994. Nd, Pb and Sr isotopic data from the Napak carbonatite-nephelinite centre, eastern

Uganda: an example of open-system crystal fractionation. Contr. Miner. Petrol. 115, 356-66Snelling, N. J., 1965. Age determination on three African carbonatites. Nature 205, 491.Stacey, J. S., & Kramers, J. D., 1975. Approximation of terrestrial lead isotope evolution by a two-stage model.

Earth Planet. Sci. Lett. 26, 207-23.Tatsumoto, M., 1978. Isotopic composition of lead in oceanic basalt and its implication to mantle evolution. Ibid.

38,63-87.Taylor, H. P., 1979. Oxygen and hydrogen isotope relationships in hydrothermal mineral deposits. In: Barnes, H. L.

(ed.) Geochemistry of Hydrothermal Ore Deposits 2. New York: John Wiley, 236-77.Taylor.S. R.,& McLennan, S. M., 1985. The Continental Crust • Its Composition and Evolution. Oxford: Blackwell,

312 pp.Tilton, G. R., Kwon, S. T., & Frost, D. M., 1987. Isotopic relationships in Arkansas Cretaceous alkalic complexes.

In: Morris, E. M.,& Pasteris, J. D. (eds.) Mantle Metasomatism and Alkaline Magmatism. Geol. Soc. Am. Spec.Paper 215, 241-8.

Warne, S. St. J., 1962. A quick field or laboratory staining scheme for the differentiation of the major carbonateminerals. J. Sediment. Petrol. 32, 29-38.

Wen, J., Bell, K., & Blenkinsop, J., 1987. Nd and Sr isotope systematics of the Oka complex, Quebec, and theirbearing on the evolution of the sub-continental upper mantle. Contr. Miner. Petrol. 97, 433-7.

Wood, D. A., Joron, J.-L., Treuil, M., Norry, M. J., & Tarney, J., 1979. Elemental and Sr isotope variations in basiclavas from Iceland and surrounding ocean floor. Ibid. 70, 319-39.

Woolley, A. R., 1969. Some aspects of fenitization with particular reference to Chilwa Island and Kangankunde,Malawi. Bull. Br. Mus. Nat. Hist. (Miner.) 2, 189-219.1987. Lithosphere metasomatism and the petrogenesis of the Chilwa Province of alkaline igneous rocks and

carbonatites, Malawi. J. Afr. Earth Sci. 6, 891-8.Garson, M. S., 1970. Petrochemical and tectonic relationship of the Malawi carbonatite-alkaline province

and the Lupata-Lebombo volcanics. In: Clifford, T. N., & Gass, I. G. (eds.) African Magmatism and Tectonics.Edinburgh: Oliver & Boyd, 237-62.Jones, G. C , 1987. The petrochemistry of the northern part of the Chilwa alkaline province, Malawi. In:

Fitton, J. G., & Upton, B. G. J. (eds.) Alkaline Igneous Rocks. Geol. Soc. Spec. Publ. 30, 335-55.


Kempe, D. R. C, 1989. Carbonatites: nomenclature, average chemical compositions, and elementdistribution. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, 1-14.

Wright, J. B., 1971. The phonolite-trachyte spectrum. Lithos 4, 1-5.Ziegler, U. R. F., 1992. Preliminary results of geochemistry, Sm-Ndand Rb-Sr studies of post-Karoo carbonatite

complexes in Southern Africa. Schweiz. Miner. Pelrogr. Mitt. 72, 141-8.

APPENDIX: PETROGRAPHIC OBSERVATIONS

Carbonatites

All carbonatite samples were stained using the technique of Warne (1962).G 357 and G 369: are from the outer parts of the complex, near Kotamu (Fig. 2). G 357 consists of

equigranular, medium-coarse-grained calcite (92%), dolomite (3%), and interstitial areas (5%) infilledwith opaque Fe oxides and quartz. The interstitial material replaces the calcite and dolomite. G 369,similar to G 357, is cut by silica Fe-oxide-rich veinlets (10-15%).

G 410b: medium-grained carbonatite containing pyrochlore (5-10%), acmitic pyroxene (60%),calcite (14-24%), apatite (5-10%) and opaque Fe oxides (6%) (Garson & Smith, 1958). Pyrochlore ishighly fractured and mainly occurs with apatite. The pyroxene is altered, with some grains exhibitingmica-rich rims.

N 1295 and N 1290: N 1290 consists of fine-grained ankeritic (30^M)%) and dolomitic (5%) veinlets,both rich in opaque Fe oxides, hosted by coarse-grained calcite (50-60%) which has been strained andaltered by the ankerite-rich veins. Pyrite (3%) occurs as an important accessory mineral. N 1290 issimilar to N 1295 but contains a greater amount (40%) of dispersed opaque Fe oxides.

G 413: consists of seriate, coarse-grained dolomite (70-80%) surrounded by drusy, fine-grainedcalcite (20-30%), opaque Fe oxides (2%), and trace amount of pyrite.

MW 166: from the central part of the complex, consists of drusy dolomite (60-75%), opaque Feoxides (10-15%), trace amount of pyrite, and quartzo-feldspathic-sideritic veinlets (15-25%).

M W168: composed of coarse-grained siderite, and druses that are lined with calcite and occasionallywith white bastnaesite and pink florencite (Garson & Smith, 1958).

G 255: formed during 'hydrothermal activity', consists of quartz, fluorite, and dickite.

Silicates

G 352, G 361 and G 399: nephelinite dykes composed of augite and olivine phenocrysts in agroundmass of pyroxene, magnetite and trace amounts of biotite (± apatite). Olivine microprobeanalyses (A. Simonetti, unpub. data, 1992) from G 352 show an average composition of Fo7 9 .

MW 164: alnoite which consists of 50% carbonatite xenoliths and phenocrysts of biotite, amphibole,pyroxene, apatite, and an opaque phase (Woolley & Jones, 1987) set in a groundmass of carbonate,melilite pseudomorphs, mica, pyroxene, apatite, and xenocrysts of olivine pseudomorphs (Garson &Smith, 1958).

G 298 and G 396: camptonite dykes which contain olivine, augite, and amphibole phenocrysts set in agroundmass of pyroxene, amphibole, opaque oxides, sodic plagioclase, and unidentified turbid,interstitial material (Garson & Smith, 1958).

G 315, G 322, MW 149 and MW 150: from arcuate and elongate nepheline syenite plugs, northernspur of the island. In general, they are relatively homogeneous, medium-grained, and characterized byslender laths of white orthoclase feldspar (1 -5 cm long) set in a groundmass of nepheline (altered almostentirely to cancrinite) and smaller amounts of aegirine and aegirine-augite (Garson & Smith, 1958).Accessory minerals include biotite, titanite, and apatite.

/ 152 and MW 167: medium-grained, granular ijolites that contain nepheline, wollastonite,aegirine-augite, and irregular patches of melanite. Apatite is an abundant accessory mineral and calciteinfills some interstices.

Documents

Isotopic and Geochemical Investigation of the Chilwa Island ...asimonet/PUBLICATIONS/Simonetti...are followed inward by a ring of calciocarbonatite, ferrocarbonatite, and a central