Kibaran A-type granitoids and mafic rocks generated by two mantle sources in a late orogenic setting (Burundi)

E L S E V I E R Preeambrian Research 68 (1994) 323-356

Preglmbriup nesenrtn

Kibaran A-type granitoids and mafic rocks generated by two mantle sources in a late orogenic setting (Burundi)

L. Tack" , J .P . Li~geois a, A. D e b l o n d b, J .C. D u c h e s n e b

aDepartment of Geology and Mineralogy, Mus~e Royale de l'Afrique Centrale, B-3080 Tervuren, Belgium bLaboratoires Associ~s de G~ologie, P~trologie et G~ochimie, Universit~ de Liege, B-4000 Liege, Belgium

Received July 21, 1993; revised version accepted March 16, 1994

Abstract

In the Mesoproterozoic Northeastern Kibaran Belt of Burundi (Central Africa) two distinct late Kibaran mag- matic suites coexist, both including A-type granitoids. They are located along the Boundary Zone between the Kibaran mobile belt (Western Internal Domain) and the Archaean Tanzanian craton overlain by Mesoproterozoic foreland deposits (Eastern External Domain). Intense deformation, high-temperature metamorphism and intru- sion of abundant peraluminous anatectic crustal granites occur only in the former domain whereas the Mesopro- terozoic sedimentary cover of the latter is much less or even nearly undeformed nor metamorphosed.

The first late Kibaran magmatism (350 km long Kabanga-Musongati alignment with an emplacement age of 1275 _+ ~ Ma; U-Pb on zircon) is mainly composed of mafic and ultramafic layered rocks with subordinate A-type acidic differentiates moderately enriched in incompatible elements. Initial isotopic ratios (SrlR = 0.708; ~Nd = --8 ) indicate an old continental lithospheric mantle origin. The emplacement of these late Kibaran magmatic rocks was controlled by late lateral shear, possibly contemporaneous with the latest intrusions of the Kibaran peraluminous synkinematic granites of the Western Internal Domain ( ~ 1330-1260 Ma).

The second late Kibaran magmatism (40 km long Gitega-Makebuko and Bukirasazi alignment with an em- placement age of 1249_+ s Ma; U-Pb on zircon) is limited in volume. It is mainly granitic in composition (A-type), can be strongly enriched in incompatible elements, and comprises both syenites and mafic rocks. Initial isotopic ratios (SrlR=0.702; ~Na= +4.5 to - 1 . 4 ) point to an OIB-type asthenospheric/lower continental lithospheric mantle origin, with only slight contamination by the lower crust during differentiation. This group was also in- truded during the late lateral shear.

In both groups liquid lines of descent can be reconstructed, although some of the rocks have been strongly albi- tized. This indicates that the granites are produced by differentiation of less evolved magmas and not by crustal anatexis.

Upwelling of the asthenosphere along the Tanzanian craton can generate by adiabatic pressure release the OIB- type basic melts and provide the heat necessary to melt the continental lithospheric mantle sources. This mecha- nism assigns a major role to a lithosphere-scale late Kibaran shear event occurring at the end of the regional com- pressive deformation between two rheologically contrasted domains. Ascent of the asthenosphere, continental lithospheric mantle delamination and late orogenic extensional collapse of the Western Internal Domain are sug- gested as a possible geodynamic model for the entire Northeastern Kibaran Belt. Additional work is however necessary to test this model.

Finally, our results indicate that in the Northeastern Kibaran Belt the Kibaran orogeny ended at ~ 1250 Ma, despite various reactivation events occurring later (e.g. at ~ 1137 Ma).

0301-9268/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSD10301-9268 (94)00032-M

324 L. Tack et al. / Precarnbrian Research 68 (1994) 323-356

1. Introduction

The Mesoproterozoic Kibaran linear belt is situated along the eastern edge of the Congo cra- ton stabilized at ~ 2 Ga (Fig. 1 ). It extends from Shaba in southern Zaire (with the type area of the Kibara Mountains) to southwestern Uganda (Ankole), through eastern Zaire (Kivu), Bu- rundi, Rwanda and northwestern Tanzania (Karagwe). The belt is the northernmost of a se- ries of Mesoproterozoic roughly parallel do- mains in eastern or southern Africa (Irumide, Malawi-Mof, ambique or Lurio, Natal, Nama- qua).

The Northeastern Kibaran Belt is exposed in SW Uganda, NW Tanzania, Rwanda and Bu- rundi. It has been interpreted in Burundi as an intracontinental belt characterized by abundant peraluminous two-mica granites of crustal origin associated with an extensional process from ~ 1330 Ma to ~ 1260 Ma (Klerkx et al., 1987). Extension was followed by compression and a late shear event with alkaline granitic magma- tism (Klerkx et al., 1987; Tack et al., 1990). These A-type granitoids of Burundi are spatially associated with a major alignment of layered in- trusions composed of various marie and ultra- marie rocks (Tack et al., 1990; Deblond, 1993) (Fig. 1 ).

A-type granitoids have been extensively re- viewed in the recent literature and the problem of their origin remains highly debated. Although various petrogenetic schemes, involving melting of a pre-existent lower or middle crustal compo- nent (residual source model), are often pro- posed (Whalen et al., 1987; Sylvester, 1989; Kleemann and Twist, 1989; Creaser et al., 1991 ), derivation of, at least, some A-type magmas from a mantle source, consistent with extended frac- tionation of a basaltic melt, has also been in- voked (Turner et al., 1992). In this latter case, contamination by crustal material is often con- sidered (Lirgeois and Black, 1987; Rogers and Greenberg, 1990; Eby, 1990; Coleman et al., 1992).

In this study, we present new detailed field and laboratory data on Burundi A-type granitoids as-

sociated with mafic rocks. These data bring new major constraints upon their timing and origin and impose reconsideration of the Northeastern Kibaran Belt evolution.

2. Reconsideration of the Northeastern Kibaran Belt

The Northeastern Kibaran Belt (Fig. 1 ) can be subdivided into a western and an eastern do- main whose boundary zone is marked by a NE- SW-trending alignment of marie and ultramafic rocks, subparallel to the general trend of the belt (Fig. 1 ). The two domains differ in many ways: lithostratigraphy, type of magmatism (both in- trusives and voleanics), structural evolution and intensity of metamorphism.

Sediments of the eastern domain are different in composition and structurally below those of the western domain. In Burundi this was at the origin of the use (in the eastern domain where the sediments overlay the Archaean basement) of the term of lower Group of the Burundi Su- pergroup (or lower "Burundian") as opposed (in the western domain) to middle and upper Groups of the Burundi Supergroup (or middle and upper "Burundian"). As the sedimentary sequences of both domains cannot be intercor- related this subdivision appears to be invalid.

2.1. The Western Internal Domain

To the west, this domain is bounded by the Cenozoic Western Rift including Lakes Edward, Kivu and Tanganyika. To the north, it is delim- ited by the Palaeoproterozoic Buganda-Toro formations and the adjacent Archaean craton of Uganda (Fig. 1).

Several structural-lithological units com- posed of either predominantly granito-gneissic or metasedimentary rocks are tectonically bounded (Fig. 1 ). Thrust and shear fold belt features have been described (Theunissen, 1988, 1989; Fer- nandez-Alonso and Tahon, 1991 ). The domain comprises silicielastic sediments with very rare carbonate rocks. Interbedded volcanics, basic in

L. Tack et aL / Precambrian Research 68 (1994) 323-356 325

"to_

4 ° .

2 9 ° I

;U --_

R ;

! T

A - t y p e e , g r a n i t o i d s

30 ° 31 ° m i

32*

~X~XXXXX~XXXXXXXXXXXX> XXXXXXXXXXXXXXXXXXXXX~ ~ X × X X X X X × X X X X X X X X X × X ~ x x x x x x x × x x x x x x ~ x x x x x

xxxxxxx xxxxxxx× x x x ~ x x x x ~ x x x x x

V I R U N G A

Lake

Victoria

/ / ~ 8

FITLY6

0 501

Fig. 1. Geological outline of the Northeastern Kibaran Belt. This updated map is compiled and synthesized, in addition to our own observations, from the following documents: Carte g6~ologique du Burundi, 1/250,000, 1990; Carte g6ologique du Rwanda, 1/250,000, 1991; Geological sketch map of NW Tanzania, 1/25Q,000, 1976; Geological map of Uganda, 1/250,000, sheets Mbarara (1961) and Kabale (1961); Waleffe, 1965; Deblond, 1993. Legend: l=Archaean craton; 2=Palaeoproterozoic (Bu- ganda-Toro belt in Uganda); 3 = Kibaran peraluminous crustal granites including pre Kibaran Palaeoproterozoic relics and post Kibaran tin-bearing granites; 4 = Mesoproterozoic sediments with regional trend of the Northeastern Kibaran Belt (dashed line) and with trend of a major quartzite horizon (stippled line ); 5 = Boundary Zone located in between the Western Internal Domain (WID) and the Eastern External Domain (EED) of the Northeastern Kibaran Belt; the Boundary Zone comprises Mesoproter- ozoic sediments intruded by an alignment of mafic and ultramafic layered bodies (black) and A-type granitoids (white); for more detailed maps see Fig. 2; 6 = Neoproterozoic rhomb-shaped basin; 7= thrust and/or shear zones; 8 =limit of the Western Rift. K= Kabanga; M= Musongati; UR =Neoproterozoic Upper Ruvubu alkaline plutonic complex (Tack et al., 1984); the rect- angle on the main .figure refers to Fig. 2A. Upper left inset: political boundaries, U=Uganda, R=Rwanda, B=Burundi, T= Tanzania and Z= Zaire. Lower right inset: location of the Kibaran belt within the East African geological units; the rectangle shows the Northeastern Kibaran Belt represented in the main part of Fig. 1 (after Cahen et al., 1984, modified).

326 L. Tack et al. / Precambrian Research 68 (1994) 323-356

composition near the base of the series, become intermediate to acidic higher in the stratigraphic column (Ntungicimpaye and Tack, 1992). This sequence is affected by a regional metamor- phism reaching the amphibolite facies and is in- truded by abundant peraluminous two-mica synkinematic granites considered to be anatectic crustal melts, accompanied by subordinate amounts of matic rocks (bimodal magmatism) (Fig. 1 ). The deposition age of the sediments is poorly constrained: it is older than ~ 1330 Ma (oldest age of intrusive granites in Burundi; see below) and generally considered as post Uben- dian, i.e. younger than ~ 1800-1700 Ma.

Five Rb-Sr isochrons on different massifs (Klerkx et al., 1987) and one U-Pb zircon age on an early pluton (Ledent, 1979) in Burundi, as well as one Rb-Sr isochron in a nearby Tan- zanian pluton (Ikingura et al., 1990), indicate that the peraluminous granite activity occurred in the 1330-1260 Ma time interval. These gran- ites were emplaced during a deformation whose schistosity was always parallel to the bedding of the country rocks and granitic contacts. This de- formation has been attributed to a regional ex- tensional process (Klerkx et al., 1987).

A later deformation, characterized by open, uptight cylindrical folds, NE-SW-oriented in this part of the belt and also accompanied by peralu- minous granites, typically occurred under com- pressional conditions (Klerkx et al., 1987 ). The latest Kibaran structural event was a shear, ap- pearing as local zones of intense deformation and associated with mafic-ultramafic intrusions and with alkaline granitoids (Klerkx et al., 1987; Theunissen, 1989). These three tectonic events are older than ~ 1250 Ma (see below).

In the Western Internal Domain no Archaean basement has been observed but Palaeoprotero- zoic ages have been documented locally. In Rwanda, two granitic bodies contain relics of ~ 2 Ga gneissic protoliths (Cahen et al., 1984). In Burundi, the pre Irdbaran crystalline basement of the Bujumbura region (Nzojibwami, 1987) could be of the same age. Unlike in the Eastern External Domain (see below), a major angular unconformity between the Kibaran and its base- ment has nowhere been observed. Finally, we

must note that the post orogenic tin-bearing granites at ~ 1 Ga (Cahen et al., 1984) are re- stricted to the Western Internal Domain (Pohl, 1987).

2.2. The Eastern External Domain

Shallow-water terrigenous siliciclastic sedi- ments of the Eastern External Domain (foreland deposits) overlay the Mugera-Nyakahura inlier (2-3 ° S, 31 ° E; Fig. 1 ) of Archaean basement be- longing to the Tanzanian craton. To the north, in southwestern Uganda, Mesoproterozoic unme- tamorphosed sediments, including basal con- glomerates, clearly cover unconformably the Pa- laeoproterozoic Buganda-Toro basement (Pohl, 1987). A Neoproterozoic rhomb-shaped basin (Fig. 1 ) bounds the domain to the southeast.

The Eastern External Domain is characterized by the total absence of granites and the fading away towards the east of both deformation and regional metamorphism, the latter not exceeding the greenschist facies in the westernmost part (Fig. 1 ). West of the contact with the Archaean Mugera-Nyakahura inlier, overturned to almost recumbent folds and imbricated thrust faults show ESE-vergent tectonic transport, typical of shallow fold and thrust belts near their contact with basement (van Straaten, 1984). On the other side of the boundary zone between the two domains, similar tectonic transport is evidenced by the map geometry of a major quartzite hori- zon corresponding to the base of the western do- main sedimentary sequence (Fig. 1, stippled line). Tabular sedimentary formations ("Bu- koba Sandstone-Kavumwe" and "Nkoma"), which have been suggested to belong to Meso- proterozoic (Tack et al., 1992 ) foreland deposits on the Tanzanian craton rather than to Neopro- terozoic (Waleffe, 1965 ) sediments, are also in- cluded in the eastern domain (Fig. 1 ).

An intercalation of rhyodacitic vitric tufts near the base of the sedimentary succession has given a Rb-Sr age of 1353+46 Ma (Klerkx et al., 1987), indicating a Mesoproterozoic age for the sedimentation in the neighbourhood of the Mugera-Nyakahura inlier.

L. Tack et al. / Precambrian Research 68 (1994) 323-356 327

2. 3. The Boundary Zone between the western and eastern domains and its plutonic bodies

Ten to thirty-five km wide, the NE-SW-trend- ing Boundary Zone is characterized by the fold- ing and shear events already described in Sect. 2.1. Thrust folding with imbricate structures and relative movements of blocks from west to east are frequently observed and most strongly pro- nounced in the southernmost part of the Bound- ary Zone, where the overthrust slice of the Nyanza-Lac Archaean (Demaiffe and Theunis- sen, 1979) basement occurs (Fig. 1, 4°20'S, 29 o 40' E). The sediments of the Boundary Zone have the same lithostratigraphic characteristics as those of the eastern domain.

The Boundary Zone is marked by a 350 km long alignment of mafic and ultramafic layered intrusive bodies with occasional Ni-Cr-V-Fe- Ti-PGE mineralizations. The alignment, also evidenced by geophysical (mainly aeromag- netic) data, is subparallel to the general trend of the entire Northeastern Kibaran Belt (Fig. 1 ) (van Straaten, 1984). It will be referred to as the Kabanga-Musongati (KM) alignment, after two important massifs located, respectively, in Tan- zania and Burundi (Fig. 1 ). In Burundi, the alignment consists of eight bodies, the main ones being the Rutovu, Musongati, Mukanda-Bu- horo and Nyabikere massifs (Fig. 2A). In the Rutovu and Nyabikere massifs (Figs. 2A and 2D), subordinate A-type granitoids occur in spa- tial association with the predominantly marie and ultramafic rocks. The latter are unaffected by any pervasive deformation, except along thrust faults where the rocks have been mylonitized and re- tromorphosed. This further points to their em- placement at the end of the regional compressive deformation. Their remarkable freshness allows identification on all scales of cumulate struc- tures typical of layered igneous rocks. Strati- graphic units have been identified within the Mukanda-Buhoro-Musongati complex (central part of Fig. 2A) and cover a large interval of evo- lution from dunites (olivine Fo9o) to Fe-rich gabbronorites (opx: Fs54 En44 Woo2). These units comprise the Ultramafic Zone, subdivided into a Peridotitic Subzone and a Pyroxenito-Perido-

titic Subzone, and the Mafic Zone, made up of Noritic and Gabbronoritic Subzones (Deblond, 1993). The KM layered bodies are considered to result from the differentiation of several influxes of upper continental lithospheric mantle origin (see below).

Central Burundi is characterized by a 40 km long alignment of predominant A-type grani- toids (both granites and syenites) (Fig. 1 ) in spatial association with subordinate mafic rocks. From north to south, two massifs of A-type gran- itoids are distinguished: the Gitega-Makebuko massif, outcropping in two units, and the Buki- rasazi massif. This second alignment will be re- ferred to as the GMB alignment.

Results of new geological mapping of the A- type granitoids of both the KM and GMB align- ments in Burundi on a scale 1/50,000 (including some 1100 observations) are presented in Fig. 2. In the Nyabikere massif (KM alignment) only two occurrences of A-type granitoids are known. The main lithologic and petrographic character- istics of the Gitega-Makebuko (GMB), Buki- rasazi (GMB) and Rutovu (KM)granitoids are listed in Table 1. The rocks are generally pinkish and medium- to coarse-grained. In the Gitega- Makebuko and Rutovu massifs a perthitic feld- spar predominates over rarely zoned plagioclase, whereas in the Bukirasazi massif one single coarsely perthitic feldspar is typical. Albitization is lacking in the Rutovu massif, occasional in the Gitega-Makebuko massif and extensive (abun- dant patchy and chessboard albite) in the Buki- rasazi massif. Dark minerals are biotite and/or hornblende.

Contacts with the previously deformed and re- gionally metamorphosed sediments of the Bu- rundi Supergroup are either tectonic or intru- sive. In the latter case, thermal aureoles occur locally, characterized by overprinting of uno- riented minerals typical of contact metamor- phism (e.g. andalusite, sillimanite, cordierite) on already deformed rocks (Tack and Deblond, 1990). Pegmatites are entirely lacking.

After their emplacement, the A-type granitoid massifs as well as the mafic and ultramafic bod- ies of both the GMB and KM alignments have been sliced up into a number of thrust sheets

328 L. Tack et al. / Precambrian Research 68 (1994) 323-356

3°10

3 ° 2 0

3°30

3°40

3 ° 5 d

A ) T

2 9 ° 2 9 ' ~ 0 ' 3 0 ° 3 0 ~ 3 0 "

0 t 0 k m

/ /

/

\ +4 GITEGA e/~", -i

/ /

/ / I /

/

/

//

" J ' ~ M U K A N D A

k

2 9 ° 5 0 '

"7 - ' IK IRASAZI

/

l~y -/, ISONGATI

~ A- type granitoid

~ Mafic and/or ultramafic layered body

I J 3 0 ° 0 0 .

I 30"10"

Fig. 2 (legend p. 331 )

L. Tack et al. / Precambrian Research 68 (1994) 323-356 329

(~ I

/ /

I

, J

I

3020 "

z- \-~ \

f f 1 " ~ 1

k'

\!

3o25 '-

3030 '-

0 ,~km I I I I I I

• Pink syenogranite

~ Pink and greyish syenogrenite

~ Mafic rock

++'~ 3o35 '-

L + M a k e b u k o

84

2'°55', ~ ~ i " 4 30ooo

330 L. Tack et al. / Precambrian Research 68 (1994) 323-356

(Fig. 2). Their stacking up is at the origin of the imbricate structure in and around the massifs. Major faults coexist with abundant minor frac- turing. The igneous rocks locally become cata- clastic to mylonitic. The sense of thrusting is from west to east towards the Tanzanian craton. Highly variable in intensity, this non-pervasive defor- mation occurred in greenschist conditions and has locally converted the primary granitoid min- erals as well as the hornfels aureoles into a gra-

noblastic or grano-lepidoblastic retrograde asso- ciation (Table 1). This local retrograde metamorphism is accompanied, especially in the Bukirasazi massif, by circulation of fluids re- sponsible for extensive albitization of mylonitic rocks and crystallization of fluorite, specular he- matite and pyrite coatings only observed in and along the thrust- and joint-surfaces. Subsequent episodic structural reactivation of some of these

(c) i

• Alkafi feldspar granite ~ Alkali feldspar i syenite

~ Mafic rock ~ Rhyolitic pyroc~t ic rock -~I

0 2j,Skm I I I I I t

A

11 12 34 114

/

/

1 . 1 _ - J ~ y ,112e /

- 3 o 4 0 .

A

_ 2

. 3 o 4 5 ,

i E

2 9 o 5 5 ' 3 0 o 0 0 '

L. Tack et al. / Precambrian Research 68 (1994) 323-356 331

(D) 29.~,r~1 ,I 1 1

/ Y

29*55 '

\ \

\ \

0 5 k m I I I I I [

¢

~ Pinkish granite

~ Mafic layered body

Fig. 2. (A) Location of A-type granitoids in Central Burundi (see inset) belonging to the GMB and KM alignments. For more detailed maps of the Gitcga-Makebuko, Bukirasazi and Rutovu massifs see (B), (C) and (D), respectively. Limits of Mu- kanda-Buhoro-Musongati massif are after Deblond (1993). T= tectonic contacts with thrust movements. (B) Geological out- line of the Gitega ( + 40 km 2)-Makehuko ( + 20 km 2) massif (GMB). Mylonitization is more pronounced towards the east. Legend: T= tectonic contacts with thrust movements. C= intrusive contacts with thermal metamorphic aureole; numbers refer to localities of analyz¢~t samples, some of them being selected in Tables 5, 6, 7 and/or 8; A-B: cross-section. (C) Geological outline of the Bukirasazi ( + 12 km 2) massif (GMB). In its southern part present-day topography is close to roof of intrusion with igneous rocks exposed preferentially in valleys; same legend as for (B). (D) Geological outline of the Rutovu ( + 18 km 2) massif (KM). Same legend as for (B).

faults occurred up to Cenozoic times (Tack et al., 1984, 1992).

3. U-Pb, Rb-Sr and Sm-Nd geochronology

The results of isotopic analyses are listed in Table 2 (U-Pb) , Table 3 (Rb-Sr) and Table 4 (Sm-Nd).

3.1. The GMB alignment

3.1.1. Bukirasazi massif U-Pb. Four zircon fractions determine a dis-

cordia line intercepting Concordia at 1249_+7 s Ma (upper intercept) and 117 + 29 Ma (lower inter- cept) (Fig. 3A). The latter value is either geolog- ically meaningless (continuous lead loss) or linked to Cenozoic rift reactivation.

332 L. Tack et al. / Precambrian Research 68 (1994) 323-356

o

0

, .o

¢%

..¢:

¢.)

. J :

¢=

o 0

o

- ~ - . o

- ° ~ ~ .. ~ ~ :~o ~ _ ~ _ - ~ . ~ ~ - ,

• ~ ~ ~ =.~- =~ ~ _

-= ~ ' - ~ ~ ,~

.= . .~ ~.~ ~_ '-~ = . ~ -~

* ~ *'G. s..

~: -~ . . . . . . -

L. Tack et al. / Precambrian Research 68 (1994) 323-356 333

Rb-Sr. Seven whole rocks (six granites and one syenite) have yielded an age of 1137 + 39 Ma (Sr isotope initial ratio (SrlR) =0.7027_+0.0011, MSWD = 2.0 ) with three points (syenites) lying below the defined isochron (Tack et al., 1990), suggesting slightly different initial ratios for granites and syenites (see also Nd below).

Sm-Nd. At 1249 Ma (zircon age) two granites have ENd values of -- 1.3 and - 1.4, two syenites of - 0 . 6 and +0.2 and four marie rocks yield values between +2 and +4.5. The different Nd initial ratios preclude the calculation of an iso- chron. TDM (depleted mantle model ages) range between 1300 and 1800 Ma. As for Sr isotopes, the Nd isotopic initial ratios are different from one petrographic type to another.

3.1.2. Gitega-Makebuko massif Six (from seven) whole rocks have defined an

isochron: 1068 _+ 78 Ma (SrlR = 0.7109 + 0.0044; MSWD=0.9; Klerkx et al., 1987). The seven samples lead to the following results: 1095 _+ 80 Ma (SrlR = 0.7084 _+ 0.0046; MSWD = 5.9).

3.2. The KM alignment

Sm-Nd. At 1275 Ma (zircon age) four marie rocks have end values between - 4.8 and - 7.8 and ToM model ages of ~ 2400 Ma.

3.2.2. Rutovu massif Rb-Sr. Six granitic and six marie rocks have

been analyzed. The six granitic rocks only yield an errorchron (Fig. 3C): 1268_+32 Ma (SrlR=0.7097_+0.0005; MSWD= 13.1). The marie rocks plot around the lower edge of this errorchron. Relatively high MSWD indicate that the Rb-Sr isotopic system has been disturbed some time after the emplacement of the massif. The consistency of this value ( ~ 1268 Ma) with the Musongati age is a good indication that the errorchron is not a mixing line and that the var- ious rocks had similar SrlR and, thus, were prob- ably comagmatic. It will be shown later that they indeed belong to the same magmatic trend.

Sm-Nd. At 1275 Ma (zircon age of Muson- gati) three granites have ENd values between - 6.9 and - 7.4 and ToM model ages between 2404 and 2477 Ma; a marie granophyre (A114) has an end value of -- 10.2 and a ToM model age of 3345 Ma.

3.2.1. Mukanda-Buhoro--Musongati massif U-Pb. Four zircon fractions from a late differ-

entiate of the Amphibole Norite unit of Mutanga (Musongati submassif; Deblond, 1993) deter-

mine a discordia line intercepting Concordia at 127-+11 *, - 10 ma (upper intercept) and 190 _+ 58 Ma (lower intercept) (Fig. 3B). The interpretation of the lower intercept value is the same as in the Buldrasazi massif. The four points are not per- fectly aligned but the individual tzo7/zo6 ages range from 1229 Ma to 1260 Ma and correlate with diamagnetism (Table 2 ), which is in agree- ment with the upper intercept value. Moreover, because of high 2°epb/2°4pb ratios (3000-7000), the error due to common lead correction is neg- ligible. A layered intrusive complex in northwest Tanzania (Kapalagulu, Wadsworth et al., 1982) has given a K/Ar age on phlogopite of 1239 _+ 50 Ma (Cahen and Snelling, 1966; recalculated in Cahen et al., 1984).

3.3. Interpretation

The samples dated by the zircon method are not affected by any pervasive deformation and/ or greenschist facies retrograde metamorphism. Moreover, it has been demonstrated in many cases that greenschist facies metamorphism would only weakly affect the zircon geochrono- meter (e.g. Lancelot et al., 1983; Li6geois et al., 1991 ). We can thus interpret the U - P b results of 1249+78 Ma as the emplacement age of the Buki- rasazi massif (GMB) and 1275 + ]~ Ma as that of the Musongati massif (KM). The previous interpretation of the Rb-Sr values of Bukirasazi ( 1137 __ 39 Ma) as the emplacement age (with a restriction for the 3 points below the isochron; Tack et al., 1990), was based on preliminary structural data and is no longer valid.

The U-Pb zircon ages indicate that the Buki- rasazi massif is slightly younger than the Muson-

334

Table 2 Zircon U-Pb data

L. Tack et aL / Precambrian Research 68 (1994) 323-356

Sample fraction U Pb* 2°6pb/2°4pb 2°6pb*/23aU 2°7pb*/235U 2°Tpb*/2°6ph* t207/206

Bukirasazigranite(Gi~ -5°M/63-106gm 643 106 2363±21 0.1472 1.6381 0.08068 1214 -6°M/63-106~m 663 117 6263 ± 13 0.1587 1.7760 0.08115 1225 - 7°M/63-106~m 599 109 12485 ±99 0.1640 1.8338 0.08107 1223 -8°M/63-106gm 484 91 2067± 2 0.1650 1.8479 0.08124 1227

Musongatidifferentm~(Amphibo&nonte;DBD -4°M/63-106gm 830 171 3074± 2 0.1823 2.0438 0.08132 1229 - 5°M/63-106gm 825 161 5677± 3 0.1756 1.9824 0.08186 1242 -6°M/63-150gm 775 155 5634± 2 0.1810 2.0564 0.08239 1255 -7°NM/63-106gm 809 159 7105±23 0.1783 2.0311 0.08262 1260

*Pb = radiogenic lead; sample fractions are characterized by their diamagnetism and granulometry; U and Pb in ppm.

Table 3 Rb-Sr whole-rock data

Sample Rb Sr 87Rb/a6Sr 875r/SrSr SrlR ( ± 20)

KMgroup (at 1275 Ma) Rutovu massif (granites): A2 140 96 4.231 0.779464± 9 0.7018 A63 153 83 5.326 0.803609+ 8 0.7053 A76 174 113 4.493 0.795489± 1 t 0.7134 A89 150 107 4.085 0.781330+ 9 0.7067 A90 132 127 3.025 0.769501 + 12 0.7t42

Rutovumassif(maficgranophyre): Al l4 46.0 148 0.901 0.725996± 16 0.7097

Rutovu massif (mafic rocks ): Rul 0 4.0* 139 0.0838 0.710968 ± 29 0.7094 Ru 14 3.6* 175 0.0587 0.709153 + 42 0.7081 Ru 17a 10.4* 333 0.0904 0.709971 ± 24 0.7083 Ru30 8.1" 147 0.1587 0.709763 + 49 0.7069 Ru36 4.2* 143 0.0843 0.709030± 35 0.7075

GMBgroup (at1249Ma) Bukirasazimas~f(maficrocks): 188 64.9 1067 0.1760 0.706665± 14 0.7035 199 77.1 694 0.3214 0.707941 ± 11 0.7022 208 38.6 229 0.4879 0.712710± 9 0.7040 253 74.7 1148 0.1882 0.703751 ± 8 0.7004

Individual SrIR have been modified during the ~ 1137 Ma event--this is more pronounced in the granites (high Rb/Sr ratios) than in the marie rocks. *Measured by isotope dilution; Rb and Sr in ppm.

L. Tack et al. / Precambrian Research 68 (1994) 323-356 335

Table 4 Sm-Nd whole-rock data

Sample Sm Nd 147Sm / 144Nd 143Nd/~Nd (~Nd TDM ( _+ 2a)

GMB group (t= 1249 Ma) Bukirasazi massif (granites): Gi7 21.0 104.0 0.1221 0.511960_+ 12 - 1.3 1797 Gil I 22.0 111.0 0.1198 0.511936_+ 11 - 1.4 1793

Bukirasazi massif (syenites): Gi114 13.0 69.0 0.1139 0.511927_+ 10 - 0 . 6 1700 Gi 122 23.5 145.0 0.0980 0.511839 + 18 + 0.2 1583

Bukirasazi massif (mafic rocks): 188 10.3 52.9 0.1177 0.512122-+ 10 +2.3 1459 199 5.3 24.0 0.1330 0.512217_+ 11 +2 .0 1554 208 5.0 18.8 0.1611 0.512479 + 11 +2.6 1624 253 7.2 42.2 0.1031 0.512102_+ 11 +4.5 1296

K M group (t= 1275 Ma) Rutovu massif (granites): A2 11.0 50.0 0.1330 0.511755-+ 13 -7 .1 2443 A63 15.0 69.0 0.1314 0.511750 + 32 - 6.9 2404 A76 12.0 55.2 0.1314 0.511712+ 10 - 7 . 4 2477

Rutovu massif (mafic granophyre ): A114 9.1 35.9 0.1532 0.511750+_ 10 - 10.2 3345

Mukanda-Buhoro-Musongati massif ( mafic rocks): Bu 17 15.70 64.1 0.1481 0.511987 _+ 09 - 4.8 2465 Bu30 0.73 2.24 0.1970 0.512233_+29 - 7 . 8 Bu78 2.38 8.00 0.1798 0.512193+20 - 5 . 9 4181 Mu39 0.67 3.13 0.1294 0.511759+ 10 - 6 . 2 2328

S m / N d ratios close to the depleted mantle value preclude calculation of a TDM model age (Bu 30) or the achievement of a reliable value (Bu 78); Sm and Nd in ppm.

gati massif, suggesting similar time relationships between the GMB and KM groups. On the other hand, the ~ 1275 Ma age for the KM group is older than the closure of the Rb-Sr isotopic sys- tem in the peraluminous Kibaran crustal gran- ites of the Western Internal Domain ( ~ 1260 Ma, Klerkx et al., 1987). It is thus not excluded that the KM alignment has been emplaced contem- poraneously with some of these granites.

Although analyses were performed as much as possible on the least-deformed samples, the Rb- Sr chronometer appears to have been disturbed some time after the emplacement of the three massifs: errorchron for Rutovu, samples lying below the defined isochrons for Buldrasazi and Gitega-Makebuko. A straightforward interpre- tation for the two younger ages is to link them to the thrust event accompanied by greenschist fa-

336

0.2

0.1

L. Tack et al. / Precambrian Research 68 (1994) 323-356

206Pb*/238U

B u k i r a s a z i g r a n i t e 1249 +8/-7 Ma 4.~ ~f BURUNDI

/ - / / 4 zircon froctions lO0~j~__Lt!3 ,~ ( A 1249 +8 / -7 Ma ~"'X/~--~-M' ~ \~:x/

/ / 7 / / ~ - 5 M . . . . . . . . . .

................ 0.17 1000 / 8 M

/ . / . L L I ~ s i ~ _ 6 M

117 +~f/ 29 M a 015] ,////~"--'~ M ._ , 1.5 1,7 1.9

207Pb*/255U I T i F i i -r r i i i i

0.4 0.8 1.2 1 . 2 2.4 2.8

0.2

0.1

206Pb*/238U --/~.~

Mus ° n g a t ~ u d i : f e r e n t i a t e ~ ~

4 zircon fractions I O 0 ~ / j ~ _ : - ~ M -7M

/ 1 9 0 + / ~ 8 1.'9 ' ~ . . . . . . . . ~ ' 207Pb*/2t5U

0.4 0.8 1.2 1.6 2.4 2.8

082 87Sr/86Sr 6 WR 1268 ±32 Ma

0.78 0.7097 ±0.0005 MSWD = 13.1

0.74 ~ ( C )

• granites [] marie rocks 87Rb/86Sr

0.7 ~ ~ + . . . . ~ . . . . . . . . . . . ~ ~ . . . . . . .

0 1 2 3 4 5 6 Fig. 3. Geochronological data for the GMB and KM alignments. (A) U-Pb zircon data for the Bukirasazi (GMB) granite. (B) U-Pb zircon data for a Musongati (KM) Amphibole Norite differentiate. (C) Rb-Sr data for the Rutovu (KM) massif with errorchron.

L. Tack et aL / Precambr•n Research 68 (1994) 323-356 337

cies retrograde metamorphism and circulation of fluids. It has been shown that these conditions favour Sr isotope mobility (e.g. Walravcn ct al., 1990; Lidgeois ctal., 1991 ) and can give rise to fair rehomogenization isochrons. Our trace ele- ment data support this interpretation (see be- low). The best age for this late thrust event with a Rb-Sr isotopic resetting is given by the Buki- rasazi isochron ( 1137 _+ 39 Ma).

Field data indicate that the Bukirasazi and Musongati massifs postdate the first two Ki- baran tectonic events (extension followed by compression; Klcrkx ctal. , 1987) and arc syn- chronous to the shear, which is the last Kibaran deformation event in Burundi. This means that

1249_7 Ma and ,- ~-~o the two zircon ages ( +s 1~7~+1~ Ma) mark the end of the Kibaran orogeny in the area studied. This is in accordance with the age given by the Rb-Sr isotopic system closure in the peraluminous Kibaran granites ( ~ 1260 Ma). In particular, this implies that the main folding event giving rise to the NE-SW trend in the stud- ied part of the belt is older than ~ 1275 Ma. The Rb-Sr age given by the Kiganda granite (1185-+59 Ma, MSWD=0.8; Klerkx et al., 1987), considered by Klerkx et al. (1987) as synkinematic with the compressional phase, cannot therefore correspond to an emplacement age. As suggested by its high SrlR (0.733), this 1185 + 59 Ma age is likely to have resulted from a Rb-Sr isotopic resetting linked to the thrust event, dated by the same method in Buldrasazi at ~ 1137 Ma. This structural reactivation event as well as the tin granitic magmatism ( ~ I000 Ma) must be considered as post Kibaran events. This implies that the term Kibaran should no longer be used in southern Africa for the younger Mesoproterozoic orogeny ( ~ 1200-1050 Ma; U- Pb on zircon ages, Jacobs et al., 1993 ), also char- acterized (although ~200 Ma later) by late transcurrent shear and A-type granitic magma- tism. The term Grenville might probably be more appropriate.

Pan-African isotopic effects through tectonic reactivation of older structures are possible but poorly known. With the exception of the well documented intrusion age of the Upper Ruvubu feldspathoidal syenites in NW Burundi at 739 -+ 7

Ma (U-Pb on zircon), whose emplacement is controlled by a faulting tectonic episode due to local reactivation of shear zones (Tack et al., 1984), poor geochronological constraints exist: a 815 + 156 Ma Rb-Sr age (MSWD= 1.7) given by three Bukirasazi syenites (Tack et al., 1990) (let us note that their Sr initial ratios between 0.665 and 0.667 at 1137 Ma clearly indicate post 1137 Ma isotopic perturbation); a possible Rb- Sr resetting of Kibaran granites (Rb-Sr reacti- vation age) at 697_+18 Ma (MSWD=0.41) (Li~geois et al., 1982); and 4°Ar/39Ar ages in the range of 945-700 Ma obtained on Kibaran gran- ites in NW Tanzania (Ikingura et al., 1992).

4. Major and trace element geochemistry

4. I. A-type characteristics of both GMB and KM granitoids

In agreement with the petrographic characters described in Table 1, the granites and associated syenites of the Bukirasazi (GMB) massif dis- play various degrees of albitization as evidenced by Na20 contents between 3.77% and 7.96% (Table 5), in contrast with the relatively low Na20 contents in the Gitega-Makebuko (GMB) (2.0-2.4%), Rutovu (KM) and Nyabikere (KM) massifs ( 1.8-3.2%) (Tables 6 and 7).

In the Debon and Le Fort (1988) classifica- tion, the Buldrasazi granites are leucogranites, the Gitega-Makebuko granites show a clear light- coloured subalkaline potassic character and the Rutovu and Nyabikere granites are richer in Fe, Ti, and Mg, and "adamellitic" (monzogranitic) in character. All granites are close to the limit be- tween metaluminous and peraluminous do- mains, some samples from Bukirasazi being mildly peralkaline. According to Whalen et al. (1987) and Sylvester (1989), all granites belong to the A-type group, a character which is also confirmed by low Mg/Mg+ Fe values (0.01-0.2) and high Ga/A1 ratios (3 -4× 10 -4) as well as by high contents in Zr, Hf, Nb, Ta, Y, and REE (Fig. 4). As a consequence, they fall in the within-plate granites (WPG) domain ofPearce et al. (1984).

338 L. Tack et al. / Precambrian Research 68 (1994) 323-356

o

i

8

o o ~ - o

I

8 o o - ~ I

o o N N I

I

- - I

- - t I - -

o

t

~ . ~ o

u~

~ o ~ o o o ~ o o ~

~ d ~ o o o ~ o o ~

I

I

- - I I

I

~ t I

I

- - I I

~ . ~ +

r.~

q

q

L. Tack et al. / Precambrian Research 68 (I 994) 323-356 339

i ' q r ~-

v

!t

,<

+

° 3

o ~

7,

~f

340 L. Tack et al. / Precambrian Research 68 (1994) 323-356

Table 6 Chemical compositions of representative samples of the Gitega-Makebuko (GMB) granites

Gi4 Gi85 T 184 N3 N 100 N 174

(wt%) SiO2 74.58 74.64 75.42 74.52 76.82 72.82 TiO2 0.39 0.30 0.21 0.35 0.29 0.33 AI203 12.20 12.67 11.91 11.67 11.40 12.56 FezO3(t) 4.02 3.38 3.07 4.14 3.33 4.42 MnO 0.07 0.06 0.06 0.08 0.06 0.08 MgO 0.32 0.11 0.01 0.34 0.00 0.08 CaO 1.37 1.34 1.03 1.56 0.99 1.46 Na20 2.38 2.00 2.28 2.19 2.15 2.36 K20 5.09 5.44 6.00 4.98 5.00 5.44 P205 0.29 0.03 0.01 0.01 0.01 0.02 LOI 0.75 0.42 Total 100.71 100.57 100.00 99.84 100.05 99.57

AGP 0.77 0.80 0.86 0.77 0.78 0.78 PERAL 1.02 1.00 0.98 0.98 1.06 1.01 Q 213 219 206 219 239 195 P 7 27 35 7 19 13 A 6 21 - 4 - 3 13 3 B 63 49 41 65 45 61 Mg/Mg+Fe 0.14 0.06 0.01 0.14 0.00 0.03

(ppm) U 2.0 2.9 Th 27 28 Th/U 13 10 Zr 419 369 299 464 378 480 Ta 15 15 Nb 165 106 113 109 93 186 Nb/Ta 11 12 Ga 20 23 23 21 22 26 Rb 201 212 214 159 160 195 Sr 156 160 135 176 155 175 Ba 1659 1768 1999 2130 2t16 2172 Rb/Sr 1.3 1.3 1.6 0.9 1.0 1.1 K/Rb 210 213 233 260 259 232 K/Ba 25 6 25 19 20 21

Zn 78 81 74 98 0 112 Y 67 65 75 52 52 70 La 140 113 158 165 190 223 Ce 274 219 307 307 346 411 Nd 117 99 130 125 145 161 Sm 17.6 15.0 16.5 14.8 15.2 19.4 Eu 2.94 2.99 3.00 3.81 3.47 3.43 Gd 15.7 14.1 15.5 12.3 12.6 15.8 Dy 13.7 12.5 14.2 10.2 10.3 13.9 Er 7.1 6.6 7.9 5.6 5.7 7.4 Yb 6.6 6.1 6.7 4.8 4.5 6.3 Lu 1.02 0.94 0.92 0.76 0.70 0.91 Ce/Yb(n) 10 9 11 15 19 16 Eu/Eu* 0.6 0.6 0.6 0.9 0.8 0.6

L. Tack et al. / Precambrian Research 68 (1994) 323-356 341

Table 7 Chemical compositions of representative samples of the Rutovu ( KM ) and Nyabikere (KM) granitoids

Rutovu (KM)

A2 A63 A73 A76 A89 A100 A90

Nyabikere (KM)

NYP86 NYP47

(wt%) SiO2 74.06 73.92 74.33 74.72 73.83 72.34 73.32 68.13 62.65 TiO2 0.58 0.59 0.63 0.59 0.61 0.64 0.63 0.83 1.67 A!203 11.21 11.22 10.96 10.97 11.16 11.21 10.92 13.80 13.21 Fe203(t) 6.29 6.59 6.17 6.12 6.52 7.97 7.26 8.00 11.84 MnO 0.07 0.08 0.07 0.08 0.07 0.11 0.13 0.10 0.14 MgO 0.39 0.31 0.20 0.33 0.37 0.29 0.42 0.63 1.77 CaO 1.48 1.00 2.16 2.04 1.38 1.58 1.98 2.35 4.95 Na20 2.00 2.68 2.06 2.23 3.21 2.79 2.00 2.24 1.77 K20 3.71 3.61 3.74 3.71 3.69 3.69 3.45 3.37 2.35 P2Os 0.07 0.09 0.08 0.07 0.10 0.08 0.08 0.20 0.30 LOI 0.98 0.15 Total 100.56 100.79 100.40 100.86 100.94 100.70 100.69 99.65 100.64

AGP 0.75 0.74 0.68 0.70 0.83 0.77 PERAL 1.00 0.98 0.96 0.96 0.95 0.98 Q 250 236 241 240 212 215 P - 1 2 - 2 8 - 2 6 - 3 0 - 5 0 - 4 0 A 24 21 - 8 - 8 - 1 2 - 5 B 96 98 90 92 98 115 Mg/Mg+Fe 0.11 0.09 0.06 0.10 0.10 0.07

0.72 0.53 0.41 0.95 1.19 0.91

246 206 182 - 2 7 - 4 3 - 9 6

6 43 - 2 5 109 126 213

0.10 0.13 0.23

(ppm) U 5 5 5 Th 19 18 18 T h / U 3.0 3.0 3.0 Zr 319 331 310 276 336 332 340 Hf 9 8 8 Z r / H f 34 39 33 Ta 1 1 1 Nb 15 16 16 15 15 19 14 Nb/Ta 13 13 13 Ga 18 20 20 20 20 19 20 Rb 140 153 172 174 150 152 132 Sr 96 83 116 113 107 102 127 Ba 745 703 651 601 660 766 635 Rb/Sr 1.5 1.8 1.5 1.5 1.4 1.5 1.0 K /Rb 226 192 180 177 207 201 224 K/Ba 41 43 48 51 46 40 45

4 4 16 13 3.8 3.6

283 262 7 6

36 40 3 3

27 30 9 10

142 98 109 116 786 620

1.3 0.8 197 199

36 31

Zn 134 170 182 337 168 234 111 103 108 Y 58 81 76 67 64 70 63 58 60 La 47 72 62 57 59 54 51 51 41 Ce 104 145 127 114 118 110 102 106 85 Nd 51 69 59 52 52 53 47 49 40 Sm 11.0 15.0 11.1 10.6 9.3 10.1 8.6 9.0 8.0 Eu 1.69 1.55 1.76 1.68 1.58 1.88 1.49 1.60 9.7 Gd 9.5 10.6 10.9 9.3 8.9 9.6 8.4 9.8 9.9 Dy 10.2 11.4 12.1 10.4 10.0 11.3 9.7 9.9 9.8 Er 5.9 6.7 7.3 6.4 6.2 7.0 6.1 5.7 5.7 Yb 5.8 5.8 7.1 6.3 6.1 6.9 6.0 5.8 5.8 Lu 0.79 0.87 1.02 0.99 0.87 1.14 0.85 0.81 0.81 Ce /Yb(n ) 4 6 4 4 5 4 4 5 4 Eu/Eu* 0.5 0.4 0.5 0.5 0.6 0.6 0.6 0.5 0.6

342 L. Tack et al. / Precambrian Research 68 (1994) 323-356

+ o

1.6 ;E +

1~ 1.4- £ ~ 1.2- 0 +

xl , , i i J J "~PERALUMINOUS []

/ /

x

A L K A L I N E a z~

I I I I I 2 4 6 8 10

( ~ + FeOt+ T I 0 2 ) / S | O 2

x +

A - T Y P E S

~ 100 x

~. x

x xx ~at

111 ~ ~"

K~RMAL I

100 1000 10000 Zr +Nb+~+Y

+ Gl tega. Makebuko granites x Buklnlsazl granites Zx Rutovu granites ~7 Nyablkere granltolds

Fig. 4. Discrimination diagrams showing in (A) the alkaline character of the KM and GMB granitoids (after Sylvester, 1989) and in (B) their A-type characteristics (after Whalen et al., 1987; Sylvester, 1989).

Each granitoid massif has a specific trace ele- ment fingerprint. The Rutovu (KM) granite (Table 7) appears very homogeneous, its trace dement content being close to the average A-type granites given by White and Chappell (1983) and Whalen et al. (1987), especially for Nb, Ba and REE. The REE distribution (Fig. 5A) shows a distinct Eu-negative anomaly (Eu/Eu* = 0.55 ), a low Ce/Yb(n) value (4-6) and a fiat distri- bution for the heavy REE. The overall similarity with the Nyabikere (KM) samples is also to be noted (Table 7, Fig. 8B).

The Gitega-Makebuko (GMB) granite (Ta- ble 6) also displays a limited range of variation in trace elements and is distinctly enriched in Nb,

Ba and REE (by factors of 6, 3 and 2, respec- tively) compared to average A-type granites. Its REE distribution (Fig. 5B) shows Ce/Yb(n) values between 9 and 19 with similar slopes be- tween light and heavy REE. Eu/Eu* values are in the range of 0.6 to 0.9, i.e. higher than in Ru- tovu (KM).

The Bukirasazi (GMB) granites and associ- ated syenites (Table 5 ) show large variations of trace elements. Nb is generally high ( ~ 100-300 ppm)- -up to three times higher than in the Gi- tega-Makebuko granites---with normal Nb/Ta values ( 12-15 ). It is also usually depleted in Eu, Sr and Ba, with Eu/Eu* values down to 0.1, large K/Ba (up to 857), and K/Rb (up to 300). The Ce/Yb(n) values vary between 6 and 28. The REE distribution (Fig. 5C) in the most enriched samples lie within the range of the Gitega-Mak- ebuko granite, except for Eu.

We will refer hereafter to moderately enriched A-type granitoids for the KM group as opposed to strongly enriched A-type granitoids for the GMB group.

4.2. The GMB differentiation trend

A remarkable feature of the Bukirasazi massif is the coexistence of granitic rocks with inter- mediate and basic rocks. This character is not uncommon in A-type granitoids, as already mentioned by several authors (e.g. Black and Girod, 1970; Demaiffe et al., 1991; Emslie, 1991; Turner et al., 1992), and the large interval of variation in chemical composition is inconsist- ent with an origin of the parent magmas through crustal anatectic processes at eutectic tempera- ture or close to a temperature minimum. The chemical variation points to a different interpre- tation in which differentiation mechanisms are needed to explain the range of composition. In this context the relationship between granites and more marie rocks is worth being discussed. Is it possible to determine whether the rocks repre- sent cumulates, crystal-laden liquids or simply liquids? If some liquids can be identified, to what extent can a liquid line of descent and a parent magma be reconstructed? Answers to these ques- tions are particularly welcome in view of the iso-

L. Tack et aL / Precambrian Research 68 (1994) 323-356 343

1000

10C

10

RUTOVU

1000-

100

10

I I L I I I I I I I ~ LaCe Nd SmEuGdTbDy Er YbLu LaCe Nd SmEuGdTbDy Er YbLu

1000- 1000

100

10

100

10

LaCe Nd SmEuGdTbDy Er YbLu

RASAZI (Syenites)

F6 t I I ] I I I I I r--T-

LaCe Nd SmEuGdTbDy Er YbLu

Fig. 5. Chondrite-normalized REE distribution for the KM and GMB granitoids. (A) Rutovu massif (KM). (B) Gitega-Mak- ebuko massif ( GMB ). (C) Granites of the Bukirasazi massif (GMB). (D) Liquid ( L )- and cumulus ( C ) -syenites of the Buki- rasazi massif (GMB). Same symbols as in Fig. 4, except for L- and C-syenites (see D ).

344 L. Tack et al. / Precambrian Research 68 (1994) 323-356

topic characteristics of the Bukirasazi massif, which point to a mantle origin (Er~d from +4.5 to -- 1.4). The discussion is however difficult be- cause, in the Bukirasazi massif, albitization has superimposed its effect on primary features. Rather than dealing with major elements which, even when they can be considered immobile, are still affected by dilution processes, we will mainly address the trace elements which are relatively less influenced by this effect. In a first step the Sr, Eu and Ba, which seem to behave coherently, will be discussed in terms of albitization. Then, the immobile or less mobile elements will be used to interpret the variation diagrams and an an- swer suggested to the above questions, at least in a qualitative way.

4.2.1. Albitization and the behaviour of Sr, Eu and Ba

Sr, Ba and Eu are distinctly depleted in the Bu- kirasazi granites. These elements are classically concentrated in K-feldspar and thus, fractional crystallization of this mineral could account for the depletion. However, this process does not appear to play a dominant role here because KzO remains at relatively high values in the granites and because in log-log coordinates, Eu vs. Sr and Eu vs. Ba (Figs. 6B and 6C) do not plot on linear trends, as would be expected in a fractional crys- tallization process. Note that Sr and Ba do plot on a grossly linear trend (Fig. 6A ) which also en- compasses the Rutovu and Gitega-Makebuko granites. The mixture of syenitic rocks with granites in the trend, however, precludes a frac- tionation mechanism because this must always proceed in a single direction. In fact, the rela- tionship between Sr and Ba simply reflects close geochemical properties.

It can be inferred that the depletion in Sr, Ba and Eu results from albitization due to circulat- ing fluids. The process is by no means straight- forward, the negative correlation between Na and Ba for instance being rather low (Fig. 6D). Sr is however well known to be mobilized in albitiza- tion (see e.g. O'Brien et al., 1985; Kinnaird and Bowden, 1987). Although Eu 2÷ has numerous geochemical affinities with Sr, here total Eu (i.e. Eu 2÷ +Eu s÷ ) does not show correlation with Sr

(Fig. 6 ). Its depletion might result from its oxi- dation by the fluids into Eu 3+, which is more dif- ficult to accommodate in the feldspar minerals than Eu 2 ÷ and is therefore leached out. The deep negative Eu anomaly in these rocks can thus re- sult from albitization. Consequently, it can be concluded that the three elements have been mo- bilized in the albitization process and cannot be used further to decipher the liquid line of de- scent (see e.g. Kleemann and Twist, 1989). On the other hand, the mobility of Sr in albitization can favour a partial isotopic rehomogenization when it takes place, as suggested by the ~ 1137 Ma Rb-Sr age.

Interestingly, the enrichment in Nb and Ta in the Bukirasazi granites compared to the Gitega- Makebuko granites also suggests mobility in the albitization process, a feature well in agreement with the occurrence of Nb-Ta orebodies related to Na metasomatism of alkali granites (Kin- naird and Bowden, 1987). Although tempting, this behaviour cannot however be confirmed here. As discussed in the next section, the enrich- ment in Nb can also be interpreted in terms of incompatible element behaviour in fractional crystallization.

4.2.2. Differentiation trends Fig. 7 shows the evolution in the three massifs

for TiO2, Zr and Ce--three elements reputed to be immobile or least mobile in albitization pro- cesses--as well as Nb, an element usually consid- ered immobile. Mafic rocks (Table 8) from the Bukirasazi (GMB) massif, a fine-grained gab- bro from the Buhoro (KM) massif (Deblond, 1993), a doleritic gabbro and a mafic grano- phyric gabbro, both from the Rutovu (KM) massif, have also been plotted on the variation diagrams.

In all four diagrams of Fig. 7, a GMB differ- entiation trend can be distinguished from a KM trend, the latter being low in Zr, Ce and Nb, richer in TiO2 in the acidic rocks, and devoid of any peak values in Ce, Nb and Zr. These culmi- nations ofZr, Nb and Ce contents at around 70% SiO2 in the GMB trend, followed by a rapid de- crease, strongly suggest saturation of the liquid in phases enriched in these elements, i.e. zircon,

L. Tack et al. I Precambrian Research 68 (1994) 323-356 345

1000 Sr

100

10

10 Eu -~

~

0 , ,

X x []

X

X x

X [] X

[3 [] []

q I III I ' " I : : ' 100 1000

I I ',:; ', ','~','~,, : ',

x

x [] x x

x

I IIIIII I I IIIIII I 1OO 1000

IL.

(3

" ' 10 Eu

1.

i 1

10000 Ba

~ t l l l

I I I I l l l l l I I I I I I '

%

X x X

x

x

x q,

I I F 1 1 1 1 1 I 10 100

I t i l l Sr 1000

8 ~ I N a 2 0

E~D

6 - - x

x [ ]

4 ~ x

/ x

a~ ++ 2 ~ v +#

V

10 Ba lO00 8 0 0 1600 B a

Fig. 6. Relationship between Sr, Ba, Eu and Na20: (A) Sr vs. Ba; (B) Eu vs. Sr; (C) Eu vs. Ba; and (D) Na20 vs. Ba. Same symbols as in Fig. 4; squares are for Bukirasazi syenites (undifferentiated).

a REE mineral such as allanite (see Table 1 ) and a Nb-Ta-rich mineral (ilmenite or rutile?). The rapid decrease in REE is also accompanied in the most evolved rocks by a distinct decrease in Ce/ Yb ratios (Table 5, Fig. 5C), a characteristic also shared by many evolved granites (see Cocherie, 1984).

Assuming that the REE distribution has not been modified during albitization except for Eu 2+ and Eu/Eu*, inspection of the trace ele- ment contents of the Buldrasazi (GMB) syen- ires suggests that two groups of rocks can be dis- tinguished: the first one (liquid syenites or L- syenites of Table 5 ) would represent liquids, the second one (cumulus syenites or C-syenites),

liquids laden with various cumulus minerals. In- deed, C-syenites are richer in Zr and show a large Ce/Yb ratio (Fig. 5D) suggesting that they have accumulated zircon and feldspars at the liquidus of SiO2-richer liquids. The small positive Eu anomaly and the high Ba content in sample 236 (C-syenite) belongs to the least albitized sam- ple. The high Nb and REE contents of this sam- ple also suggest accumulation of accessory min- erals. On the other hand, the L-syenites show a slight decrease in Ce/Yb ratios with the evolu- tion due to the possible separation of a feldspar- dominated cumulate. Assuming that the alkali content of the most Zr-rich L-syenite (sample Gi122) has not been (too much) modified by

346 L. Tack et al. / Precambrian Research 68 (1994) 323-356

4 0 0

Nb

300-

200-

100-

1200

Zr

10OO-

8 0 0

6 0 0 -

400 -

:200-

O

G M B / / • / / t

/ / • ~x

,11 -4- \ / / / • ~+

o v K M v ~_ ©

45 50 55 60 65 70 75

I I I i I I

SiO2

~ G M B / '~ /

. +~+

I I I I I I I 45 50 55 60 65 70 75

i

Ce

4 0 0 ¸

3OO

2 0 0 -

100

O-

T i02

3-

2

I -

I 45

i . . . . . . ~ - - I I I

..+,

II// I

G M B / I / / i++ • / x~.

/ / / x

/ / x l

/ / x I I

/ / I

+ i

i

r I I I I I 1- 50 55 60 65 70 75 S iO2

I ~ I - - ~ I - t - - ]

' "'.KM I

Lt '7 " x

\ "- . EL G M B " - 0 ~\ ~

o " - - I =. "'-." ÷~'~+ + I S i 0 2 45 50 55 60 65 70 75 S i 0 2

+Gitega - Makebuko granites xBukirasazi granites ,,Bukirasazi L - syenites ~Bukirasazi C - syenites oBukirasazi gabbros ~Rutovu granites oRutovu dolerite (A117B) ARutovu granophyre (A l14) vNyabikere granitoids oBuhoro gabbro (Bu 17)

Fig. 7. Differentiation trend (Harker diagram ) of KM and GMB rocks illustrated by (A) Nb vs. SiO2, (B) Ce vs. SiO2, (C) Zr vs. SiO2 and (D) TiO2 vs. SiO2.

albitization (4.39% Na20), saturation in zircon following Watson and Harrison (1983) occurs at 915°C, a consistent temperature for A-type granitoids.

In conclusion, the GMB differentiation trend shows a series of liquids that would start at about 63% SiO2 in which incompatible element con- tents rise and culminate between 69% SiO2 and 73% SiO2. Assimilation during fractional crys- taUization as suggested by Sm-Nd and Rb-Sr data (see above) cannot be put forward, thus in- dicating little influence on the bulk chemistry.

4.2.3. Relationships with mafic rocks The connection between the least SiO2-rich

liquid and the matic rocks occurring in the mas- sif (Tables 5 and 8) is worth discussing. Three marie rocks have been analyzed (Table 8) and plotted in Fig. 7, where they appear rather dis- persed. Their REE distributions (Fig. 8A) and the spidergram (Fig. 9 ) show positive Eu anom- alies, a clear enrichment in REE with a positive peak for P in sample 188, and a trough in Zr-Hf in all samples. All these features point to a large cumulus component in the composition. Since no basic liquid can be ascertained, it is not possible

L. Tack et al. / Precambrian Research 68 (1994) 323-356 347

Table 8 Chemical compositions ofmafic rocks from the Bukirasazi (GMB), Rutovu (KM) and Buhoro (KM) massifs

Bukirasazi (GMB) gabbros Rutovu (KM)

188 199 253 dolerite granophyre AII7B Al l4

Buhoro (KM) gabbro Bu 17

(wt%) SiO2 42,23 46.77 47.22 55.20 51.42 TiO2 3,96 1.79 2.59 0.49 2.63 A1203 16.05 17.41 19.17 14.84 12.84 Fe2Oa" 4.00 3.27 3.32 10.34 4.00 FeO 9,48 7.04 7.00 12.21 MnO 0.23 0.18 0.22 0.15 0.23 MgO 6.00 6.69 3.86 6.00 3.01 CaO 12.46 14.30 10.41 9.00 7.83 Na20 2.04 1.69 3.17 1.95 2.00 I(20 1.17 1.03 1.57 1.29 1.17 P205 1.67 0.33 0.72 0.15 1.04 LOI 1.72 Total 9.59 100.50 99.75 99.91 99.98

AGP 0.29 0.22 0.36 0.31 0.35 PERAL 0.59 0.58 0.74 0.71 0.68 Q - 6 12 2 108 102 P -263 -288 -255 - 196 - 180 A -221 -246 -131 -121 -117 B 249 229 170 284 158 Mg/Mg+Fe 0.76 0.80 0.70 0.53 0.60

(ppm) U 0.6 0.3 0.8 0.8 2.0 Th 2.0 1.4 4.0 4.4 5.5 Th/U 4.9 5.0 5.2 5.5 2.8 Zr 91 51 55 106 64 Hf 2 2 3 3 4 Zr/Hf 46 24 18 35 30 Ta 3 2 4 0.8 0.9 Nb 38 18 38 7.5 10 Nb/Ta 11 12 9 9 1 l Rb 64 80 42 51 46 Sr 1085 526 707 216 148 Ba 633 658 2345 331 271 Rb/Sr 0.1 0.2 0.1 0.2 0.3 K/Rb 152 107 310 210 211 K/Ba 15 13 6 32 36

V 266 295 183 163 116 Cr 12 126 108 51 7 Co 48 35 27 34 45 Ni 12 59 44 60 12 Zn 100 71 95 84 132

Y 24 15 10 28 51 La 47 20 48 20 25 Ce 108 43 101 43 58 Nd 53 24 42 22 33 Sm 10.3 5.3 7.2 4.5 7.6 Eu 4.36 2.55 3.70 1.17 1.81 Gd 7.5 4.5 5.3 4.6 8.9 Dy 4.7 3.5 3.7 4.7 8.8 Er i.8 1.5 1.5 2.9 5.0 Yb 1.3 1.3 1.3 2.6 4.5 Lu 0.18 0.19 0.20 0.45 0.65 Ce/Yb(n) 20 8 19 4 3 Eu/Eu* 1.5 1.6 1.8 0.8 0.7

47.81 0.76

1 5 . 6 2

12.21

0.20 10.42 11.05

1 .07 0.16 0.07 1 .35

99.36

0.12 0.71

95 -228 - 126

421 0.63

0.0 0.9

79 2

38 0.4 6.7

17 10

117 99 0.1

133 13

272 630

46 12 75

22 11 26 14 3.1 0.91 3.5 3.6 2.1 2.2 0.32 3 0.8

"Total Fe expressed as Fe203, except when FeO is measured.

1000- 1OOO

100 -

1 0 -

100-

10-

GMB

" ~ ~ GITEGA- ~ ; ~ K E B U K O

KM

348 L. Tack et al. / Precambrian Research 68 (1994) 323-356

I i i J i [ i i i i , 1 i i i i f I I I I

LaCe Nd SmEuGdTbOy Er Yb Lu L'aCe Nd SmEuGdTbDy I~r Yb Lu

Fig. 8. Chondrite-normalized REE distribution of (A) GMB rocks and (B) KM rocks. Hatched areas are for Gitega-Makebuko (see Fig. 5B) and Rutovu (see Fig. 5A) granites, respectively; same symbols as in Fig. 7. Also note the similarity between the two Nyabikere granitoids ( ~7 ) and the Rutovu granites.

1 0 0 0

A Q

m

= 100 Q

' ¢

o 10

g

• - - 188 (GMB)

199 (GMB)

- - * - - 253 (GMB)

I I I I I i I I i I I i I i I 4 I

B a R b Th K N b Ta La C e S r N d P H f Z r S m 11 Y Yb

Fig. 9. Chondrite-normalized trace element abundance diagram (spidergram) for three Bukirasazi (GMB) mafic rocks ( 188, 199 and 253) and one Buhoro (KM) gabbro (Bu 17); normalizing values after Thompson et al. (1984).

L. Tack et al. / Precambrian Research 68 (1994) 323-356 349

on a strict trace element basis to extend the GMB trend towards more mafic rocks. On the same basis, it is also not possible to assess or reject the existence of an unidentified basaltic liquid which could have fractionated the mafic rocks as cu- mulates and given rise to the syenitic liquids, as suggested by the low Sr isotope initial ratios shared by all these rocks (0.702 + 0.002).

+5

~Nd

GMB granites GMB syenites

• GMB maf icrocks KM mafic rocks

* KM granites

4.3. The KM differentiation trend

As already noted, a KM trend (Fig. 7 ) starts from a Buhoro fine-grained (chilled) gabbro and a Rutovu dolerite (liquid?) and passes through a Rutovu mafic granophyre and two Nyabikere granitoids to end up in the well defined group of the Rutovu granites. The REE distribution in the marie rocks (Fig. 8B) shows no significant Eu anomaly and a Ce/Yb(n) value close to the Ru- tovu granite value. The spidergram (Fig. 9 ) does not show significant enrichment in the most in- compatible elements and, thus, does not point to an ocean island basalts (OIB) type source (Thompson et al., 1984). The low values in Ba, Th and K as well as the absence of significant Nb and Ta troughs are not in favour of a crustal con- tamination process. The marked trough in Sr--a common feature of many continental flood ba- salts (CFB)--is probably not due to an early separation of plagioelase, because olivine is the first liquidus mineral in Musongati (Deblond, 1993), but points to the presence of plagioclase in the source region. The troughs in P and in K would indicate the presence of some apatite and phlogopite in the source rocks.

The existence of a liquid line of descent link- ing the Rutovu granites to basic liquids together with the spatial association in the field between granites and marie rocks strongly suggest that they belong to a common differentiation trend, in agreement with their similar Sr and Nd iso- topes initial ratios (Fig. 10). Interestingly, this trend (Fig. 7) begins at rather constant if not slightly decreasing SiO2 content in the range of 50-55% SiO2, a typical feature of a tholeiitic trend, e.g. that of the Skaergaard (Wager and Brown, 1968), and continues without showing any culmination in incompatible elements. This

-5

-10

0.7

. J , , , . I 87Sr /86S r)i

0.705 0.710

Fig. 10. ~Nd versus StiR diagram for the GMB and KM groups. Isotopic characteristics recalculated back to 1249 Ma for GMB group and to 1275 Ma for KM group.

low degree of enrichment, compared to the GMB trend, is possibly due to a lower content in in- compatible elements in the parental magma (note the extremely low content in Nb in sam- ples Bu 17, A 114 and A 117B). Consequently, the liquids would never become sufficiently en- riched in incompatible elements to reach satu- ration in accessory minerals. Indeed, the data of Watson and Harrison (1983) indicate satura- tion in zircon in the Rutovu granite at 600°C, a temperature much too low for an A-type granite.

5. Sr and Nd isotopic initial ratios

Initial Sr and Nd isotopic ratios of the GMB and KM rocks are reported in Fig. 10. The (~Nd values have simply been calculated back to 1249 Ma with the measured Sm/Nd ratios of each sample considering that the ~ 1137 Ma event has not redistributed too much the REE pattern, or at least that the ~ 110 Ma period separating the two events is sufficiently short not to have in- duced major changes in the Nd isotope ratios. On the other hand, it is certainly not the case for the Rb-Sr system, as shown by the geochemical data

350 L. Tack et al. / Precambrian Research 68 (1994) 323-356

and by the Rb-Sr isochron at 1137 Ma. As Sr has probably been mobilized during albitization (~ 1137 Ma), it is not pertinent to use Rb/Sr ratios to calculate SrlR at 1249 Ma. For the more acidic rocks, the only reliable data are the SrlR of the Rutovu 1268 Ma errorchron (0.7097 + 0.0005 ) (Fig. 3C) and that of six Bu- kirasazi granites at 1137 Ma, given by the iso- chron (0.7027___0.0011 ). Three Bukirasazi syenites being below the 1137 Ma isochron, their initial ratios can be considered lower. No more precision can be given because their calculated SrlR at 1137 Ma are below 0.7000. Considering an average Rb/Sr ratio of 0.7 for the entire mas- sif, a minimum value of 0.001 can be assessed for the increase of SrlR due to 87Rb decay during ~ 110 Ma. This thus gives a SrlR at 1249 Ma of 0.7017 + 0.0011. This interval has been plotted on Fig. 10 for both granites and syenites.

The lower Rb/Sr ratios of marie rocks reduce the error of estimation of SrlR and the values of the crude calculation have been plotted on Fig. 10. The four GMB marie rocks show a relative scatter in SrlR (from 0.7004 to 0.7040) proba- bly also due to the ~ 1137 Ma thrust event and circulation of fluids. They are, however, not far from the mantle array whose area on Fig. 10 can be considered as the probable original composi- tion of the GMB marie rocks. Sample 253 with an ~Nd= +4.5 and a ToM= 1296 Ma close to the

( 1275_ ~o Ma) could represent emplacement age + 1 ! a depleted mantle product, barely contaminated by the continental crust.

The results show upper mantle values for the mafic GMB rocks (possibly slightly contami- nated by continental crust) and a crustal con- tamination during differentiation towards syen- ites and granites for the more felsic rocks. This contamination however remains weak, as likely values at ~ 1249 Ma for the granites are ~ 0.702 in Sr and - 1.4 in ~Nd with intermediate values for the syenites. The contaminant seems to be an old lower crust component affecting more the Nd than the Sr isotopes. The nearby Archaean Tan- zanian craton could be a suitable candidate.

The isotopic characteristics of the GMB man- tle source (SrIR= ~0.702; ~Nd= +2 to +4.5) coupled with the strong enrichment in incom-

patible elements of the granites point to an ocean island basalts (OIB) mantle type. This suggests a chemically modified asthenosphere as the likely source of the GMB rocks, being either the ther- mal boundary layer (TBL, i.e. the ductile lower part of the continental lithospheric mantle pro- posed as a major relay source for OIB by Black and Lirgeois, 1993 ), or mixing of asthenosphere and TBL. Indeed, the TBL can be considered as essentially composed of cooled asthenosphere variably mixed with OIB-typical components rising from below.

On the other hand, the KM alignment shows distinct Sr and Nd values (SrIR=~0.708; end = --8 ). Major and trace elements geochem- istry (see KM differentiation trend) as well as olivine and Ca-poor pyroxene compositions (re- spectively Fo up to 90 and mg# number up to 87 ) preclude large crustal contamination in marie and ultramafic rocks (Deblond, 1993). A man- tle source with such isotopic characteristics can- not be obtained in the asthenosphere as shown by mid-ocean ridge basalts (MORB) and ocean island basalts (OIB) (see compilation, e.g. in Faure, 1986). The only mantle source which can acquire such characteristics is the continental lithospheric mantle (CLM) or, more precisely, the mechanical boundary layer (MBL: the rigid upper part of the CLM), inasmuch as it has been enriched in Sm and Rb over a period of time suf- ficiently long to allow the radiogenic isotope ac- cumulation up to the values obtained. TDM model ages on the KM alignment are, indeed, > 2400 Ma. Marie rocks not contaminated by continen- tal crust and having such isotope initial ratios are also currently found in continental flood basalts (CFB), generally considered as generated in the continental lithospheric mantle (e.g. Hawkes- worth and VoUmer, 1979; Nelson, 1983; De Pa- olo, 1983; Hawkesworth et al., 1993; Beard and Johnson, 1993 ).

There are thus two groups of A-type granitoids in Burundi. The first one appears as a major lith- ological facies in the GMB massifs and is of as- thenospheric/lower continental lithospheric mantle origin. It is comparable to the classical alkaline ring complexes of West Africa (Ba et al., 1985; Bowden et al., 1987), strongly enriched in

L. Tack et al. / Precambrian Research 68 (1994) 323-356 351

incompatible elements, and typically corre- sponds to OIB sources. The second one is repre- sented by subordinate fades associated with large amounts of marie and ultramafic rocks (KM alignment), moderately enriched in incompati- ble elements and of upper continental litho- spheric mantle origin, which suggests some links with CFB.

6. Geodynamic model for the A-type granitoids and associated marie rocks in Burundi

A major constraint of the Northeastern Ki- baran Belt evolution in Burundi is the need to generate in a short time span A-type granitoids associated with mafic rocks both from an old

10- O-

5 0 -

100-

"r" I '- I,U 150- r,J

2 0 0 -

250 -

WID

~ o , ~ ~ , rT/

BZ EED +AC

M C r u s t , IMoho

o

, / ~ Pemlumlnoui granites in WID

• Maflc rocks in WID

Fig. 11. Idealized section of the Boundary Zone (BZ) be- tween the Western Internal Domain (WID) and the Eastern External Domain (EEL)) +Archaean craton (AC). MBL = mechanical boundary layer; TBL = thermal boundary layer; CLM= continental lithospheric mantle (for definition see text). Asthenosphere/TBL upwelling along the Tanza- nian craton and under lateral shear regime (BZ) induces par- tial melting in the MBUs (KM group, ~ 1275 Ma) and gives rise to GMB group ( ~ 1249 Ma) as a result of adiabatic pres- sure release in the asthenosphere/TBL. In the WID simulta- neous lower crust melting could generate peraluminous syn- kinematic granites associated with subordinate marie rocks. CLM delamination in time interval ~ 1330-1260 Ma and possibly late orogenic extensional collapse are suggested as general processes.

MBL and from an OIB-type modified asthenos- phere or a mixture of asthenosphere and TBL as proposed by Black and Li6geois (1993). More- over, both groups form alignments in the Bound- ary Zone located in between the Archaean Tan- zanian craton and the Kibaran mobile belt (Fig. 1 ). The emplacement of these groups, which took place at the end of the regional compressive de- formation of the belt, is shear-related and prob- ably controlled by the contrasted rheological be- haviour between mobile belt and craton because of a thicker and more rigid CLM beneath the era- ton (Fairhead and Reeves, 1977; Black and Li6- geois, 1993). The Boundary Zone thus coincides with a major structure at the lithosphere scale. We consider that this structure permitted the late orogenic uprise and emplacement of A-type granitoids and associated mafic rocks and also favoured local structural reactivation accom- panied by thrust movements, migration of fluids and Rb-Sr isotopic resetting at ~ 1137 Ma.

Generating in a short time large quantities of basaltic partial melts (KM group) from a rigid and cold MBL requires a great amount of heat. This could have been made available during the rapid uprise of asthenosphedc material, which produced the GMB type of magma by adiabatic pressure release. In other words, we propose that the two groups of A-type granitoids and associ- ated marie rocks were generated during a unique asthenospheric upwelling along the edge of the thick CLM of the Tanzanian craton during a late Kibaran shear event (Fig. 11 ). The two U-Pb zircon ages suggest that the emplacement of the asthenospheric/TBL magmas (GMB) followed the MBL ones (KM). This implies that early movements along the shear zone were large enough to permit intrusion at shallow level of important amounts of marie magmas (several influxes; Deblond, 1993 ) while later movements were more limited, allowing only differentiation at depth with emplacement of restricted amounts of predominantly felsic liquids (Fig. 11 ).

352 L. Tack et al. / Precambrian Research 68 (1994) 323-356

7. Implications for the Northeastern Kibaran Belt

The study of the Kibaran belt as a whole is not the subject of this paper. However, as the mag- marie regimes studied here occurred at the end of the Kibaran orogeny, some suggestions can be made on a regional scale.

Geochronological data indicate that the late orogenic magrnatism in Burundi could be con- temporaneous with part of the peraluminous synkinematic crustal granites. In that case, the subordinate amounts of marie rocks associated with these granites (bimodal magmatism; Klerkx et al., 1987) may be connected with the KM group, which would in turn imply a strong ge- netic link between the latter and the crustal granites.

As a rift setting does not sufficiently curve iso- therms to generate large partial melts in the MBL or in the lower crust (Platt, 1993), the only likely environment we see at the moment to produce in the same time interval partial melts from the lower crust, the continental lithospheric mantle and the asthenosphere, is a late orogenic collapse following a thickening of the crust. This model has been proposed for the Basin and Range Province (Western USA) (Coney and Harms, 1984; Livaccari, 1991 ) where mafic magmas of lithospheric origin are followed 20-25 Ma later by asthenospheric mantle products (Fitton et al., 1991; Harry et al., 1993). Abundant synchron- ous granites lowered the strength of the crust and facilitated the collapse (Armstrong and Ward, 1991 ) which was accompanied by strike-slip faulting and large rotation of numerous tectono- stratigraphic terranes (Ward, 1991). Similar features are also encountered in the Tibetan Pla- teau (England and Houseman, 1989; McKenna and Walker, 1990; Burchfiel et al., 1992).

In the Basin and Range Province and in the Tibetan Plateau, delamination of the CLM (Fig. 11 ) has been proposed (Bird, 1979; Houseman and McKenzie, 1981 ). Indeed, this mechanism induces a rapid uprise of the asthenosphere, which yields large amounts of heat to the crust and to the delaminated CLM. Under these con- ditions partial melts can be generated simulta-

neously in three reservoirs: the lower crust, the continental lithospheric mantle and the asthen- osphere (Fig. 11 ). Moreover, the heated and CLM-free thickened crust has virtually lost its rigidity, thus greatly favouring late orogenic ex- tensional collapse. Obviously this mechanism is consistent with mafic magma underplating, a classical view previously proposed for the Ki- baran (Klerkx et al., 1987) and can be taken as an intraplate event as it occurs relatively far from plate boundaries.

In any case, additional work is needed to test this model, in particular by further studying Ki- baran crustal granites and their emplacement re- lationships with pre-collapse structures.

8. Conclusions

Two late Kibaran magmatic suites have been defined in Burundi: the first one (KM group; emplacement age of 127 ~o_1o+11 Ma, U-Pb on zir- con) forms a 350 km long alignment and mainly consists of marie and ultramafic layered bodies and also includes subordinate A-type granites; the second one (GMB group; emplacement age of 1249_+7 a Ma, U-Pb on zircon), only 40 km long, is dominated by A-type granitoids and also com- prises some marie rocks. Both groups occur within the Boundary Zone between the Western Internal Domain (Kibaran mobile belt) and the Eastern External Domain (Archaean Tanzanian craton overlain by Mesoproterozoic foreland de- posits) of the Northeastern Kibaran Belt. These magmatic suites mark the end of the orogeny (~ 1250 Ma) in the Kibaran-type belt, which appears so far as a unique feature in the African continent.

In each group the A-type granitoids plot at the end of characteristic liquid lines of descent. The KM group displays a tholeiitic trend without cul- mination in incompatible elements and yields small amounts of A-type granites, moderately enriched in these elements. The GMB group en- compasses minor amounts of marie rocks made up of cumulates. The observed liquid line of de- scent is therefore limited in the 63-73% SiO2 in- terval. This trend generates A-type granites de-

L. Tack et al. / Precambrian Research 68 (1994) 323-356 353

riving from syenitic melts with a large interval of variation in incompatible elements. Both groups have been only slightly affected by crustal con- tamination. Their respective trace elements and isotope signature point to two distinct mantle sources: an old upper continental lithospheric mantle source (MBL; SrIR=0.708; end=--8) for the KM group and an OIB-type mantle source, i.e. a modified asthenosphere or an asthenos- phere/lower continental lithospheric mantle (TBL) mixture (SrIR=0.702; eNa=+4.5 to -- 1.4) for the GMB group.

The proposed model for the emplacement of the A-type granitoids and associated mafic rocks assigns a major role to a lithosphere-scale shear event taking place at the end of the regional com- pressive deformation of the belt. The KM and GMB magmatic suites are preferentially located in the Boundary Zone, i.e. the contact zone be- tween the western edge of the unaffected Tanza- nian craton (protected by its thick CLM) and the adjacent Kibaran mobile belt, characterized by intense deformation, high-temperature meta- morphism and abundant peraluminous crustal granites. In the proposed model a rapid asthen- osphere/TBL uprise, possibly due to CLM de- lamination, caused the partial melting of the MBL (KM group) and, as a result of adiabatic pressure release, of the asthenosphere/TBL it- self (GMB group). This occurred preferentially along the Boundary Zone at the end of the oro- geny (Fig. 11 ). Various reactivation events oc- curred later on, e.g. at ~ 1137 Ma.

A complete geodynamic model of the Kibaran belt should incorporate more constraints, espe- cially on the huge amounts of peraluminous crustal granites associated with subordinate mafic rocks (bimodal magmatism), some of them being probably contemporaneous with the KM group. As a track for future work we point out that a late orogenic collapse mechanism might explain the presently available data. This would imply a previously thickened crust possi- bly followed by a CLM delamination. If the lat- ter is so, the debate on the intraplate character of the Northeastern Kibaran Belt (Klerkx et al., 1987; Theunissen, 1989; Rumvegeri, 1991 ) be- comes less relevant. Indeed, as for the Basin and

Range Province or the Tibetan Plateau the phe- nomena occur relatively far from plate bounda- ries and are thus obviously intraplate, even if their driving forces are located further away on plate edges. Whatever the exact details, to con- strain such a model, the entire Kibaran belt must be taken into account, in particular its Zairian part.

Acknowledgements

Part of the mapping work has been undertaken in the framework of a field project by the follow- ing last-year students of the Department of Earth Sciences of the University of Burundi: L. Nkuri- kiye, N. Ntahindurwa, P. Semitita, Th. Simu- zeye and A. Sinzumusi. P. De Paepe (University of Ghent, Belgium) is thanked for both scientific and laboratory help in an early stage of this re- search project. Geochemical analyses were per- formed at the "Centre Interinstitntionnel de G6ochimie instrumentale ". G. Bologne (ULG) and J. Navez (MRAC) are thanked for the assis- tance in the analytical work. Isotopic analyses were performed at the "Centre beige de G6o- chronologie". J. Klerkx, K. Theunissen and R. Black are kindly thanked for their critical read- ing of an early draft of the manuscript. Careful and detailed review by D. Demaiffe has led to significant improvements of the final version. The project has benefited from the support of the Belgian Fund for Cooperative Research and the "'Administration G6n6rale de la Coop6ration au D6veloppement de Belgique" (BDI/CI 14497/ l l ) .

Appendix--analytical methods

Isotopic measurements have been carded out on a Fisons VG Sector 54 mass spectrometer.

Rb-Sr, Srn-Nd. After acid dissolution of the sample and Sr or Nd separation on ion-exchange resin, Sr isotopic compo- sitions have been measured on Ta simple filament and Nd isotopic compositions on triple Ta-Re-Ta filament. Re- peated measurements of Sr and Nd standards have shown that between-run errors are better than 0.00002. These errors have been chosen in the calculations in the general cases where

354 L. Tack et aL / Precambrian Research 68 (1994) 323-356

the within-run errors are lower. The NBS987 standard has given a value for S7Sr/S6Sr of 0.710240 + 0.000005 (2a on the mean, 35 measurements, normalized to S6Sr/ SSSr=0.1198) and the MERCK Nd standard a value for J43Nd/V~Nd of 0.512740+0.000005 (2e on the mean, 26 measurements, normalized to t~Nd/l'~Nd = 0.5119 ). Rb and Sr concentrations have been measured by XRF or by isotope dilution when concentrations were < 30 ppm. In both cases, the error on the Rb/Sr ratio is <2%. Sm and Nd concentra- tions were measured by ICP-MS. The error on the Sm/Nd ratio is < 2%. The Rb-Sr and Sm-Nd ages have been calcu- lated following Williamson (1968) and all the errors are given at the 2e level. Used disintegration constants are 1.42 × 10- I I a - l (STRb, Steiger and J~iger, 1977) and 6.54×10 - '2 a - 1

(147Sm). U-Pb. The method is derived from that of Krogh (1973 )

and Lancelot (1975). Pb and U are separated on ion-ex- change resin after acid dissolution of about 2 mg of pure and homogeneous zircons. Pb is measured on single Re filament and U on triple Ta-Re-Ta filament, both with silica gel. The fractionation coefficient, known at better than 0.1% is equal to 0.12% per a.m.u. Disintegration constants: 235U=9.8485×10 -~° a-~; 23sU=1.55125×10 -~° a -~ (Steiger and J~ger, 1977). The intercepts with Concordia and errors have been calculated following Ludwig (1980). For more precisions, see Li6geois et al. ( 1991 ).

Major elements and Rb, Sr, Ba, Ga, Nb, Y, Zr, V, Ni, Co and Zn were determined by XRF (CGR-Lambda 2020) on Li borate glass discs and on pressed powder pellets. FeO was obtained by titration. U, Th, Ta and REE were measured by ICP-MS (Fisons VG PlasmaQuad PQ2 Plus) after dissolu- tion in Li borate.

References

Armstrong, R.L. and Ward, P., 1991. Evolving geographic patterns of Cenozoic magrnatism in the North American Cordillera: the temporal and spatial association of mag- matism and metamorphic core complexes. J. Geophys. Res., 96:1320 l - 13224.

Ba, H., Black, R., Benziane, B., Diombana, D., Hascoet- Fender, J., Bonin, B., Fabre, J. and Li6geois, J.P., 1985. La province des complexes annulaires sursatur~s de l'Ad- rar des Iforas, Mali. J. Afr. Earth Sci., 3: 123-142.

Beard, B.L. and Johnson, C.M., 1993. Hf isotope composi- tion of late Cenozoic basaltic rocks from northwestern Colorado, USA: new constraints on mantle enrichment processes. Earth Planet. Sci. Lett., 119: 495-509.

Bird, P., 1979. Continental delamination and the Colorado plateau. J. Geophys. Res., 84: 7561-7571.

Black, R. and Girod, M., 1970. Late Palcozoic and recent ig- neous activity in West Africa and its relationship with basement structure. In: T.N. Clifford and I.G. Gass (Ed- itors), African Magmatism and Tectonics. Oliver and Boyd, Edinburgh, pp. 185-210.

Black, R. and Liegeois, J.P., 1993. Cratons, mobile belts, al- kaline rocks and continental lithospberic mantle: the Pan- African testimony. J. Geol. Soc. London, 150: 89-98.

Bowden, P., Black, R., Martin, R.F., Ike, E., Kinnaird, J.A. and Batcbelor, R.A., 1987. Niger-Nigerian alkaline ilng- complexes: a classic example of African Phanerozoic an- orogenic mid-plate magmatism. In: J.G. Fitton and B.J.G. Upton (Editors), Alkaline Igneous Rocks. Geol. Soc. London, Spec. Pubt., 30: 357-380.

Burchfiel, B.C., Chen Zhiliang, Hodges, K.V., Liu Yuping, Royden, L.H., Deng Changrong and Xu Jiene, 1992. The South Tibetan detachment system, Himalayan orogen: extension contemporaneous with and parallel to shorten- ing in a collisional mountain belt. Geol. Soc. Am., Spec. Pap., 269, 41 pp.

Cahen, L. and SneUing, N.J., 1966. The Geochronology of Equatorial Africa. North-Holland, Amsterdam, 195 pp.

Cahen, L., Snelling, N.J., Delhal, J., Vail, J.R., Bonhomme, M. and Ledent, D., 1984. The Geochronology and Evo- lution of Africa. Clarendon Press, Oxford, 512 pp.

Carte g6ologique du Burundi, l/250,000, 1990. Ministate de l'Energie et des Mines du Burundi, R~publique du Burundi.

Carte g6ologique du Rwanda, l / 250,000, 199 I. Minist~re de l'Industile et de l'Artisanat, R~publique rwandaise.

Cocherie, A., 1984. Interaction manteau-crofite: son r61e darts la gen~se d'associations plutoniques calco-alcalines, con- traintes g6ochimiques (616merits en trace et isotopes du strontium et de l'oxyg~ne). Th6se d'Etat, Universit6 de Rennes I, Doc. BRGM, 90 (1985), 246 pp.

Coleman, R.G., De Bail, S. and Peterman, Z., 1992. A-type granite and the Red Sea opening. Tectonophysics, 204: 27- 40.

Coney, P.J. and Harms, T.A., 1984. Cordilleran metamor- phic core complexes: Cenozoic extensional relics of Me- soproterozoic compression. Geology, 12:550-554.

Creaser, R.A., Price, R.C. and Wormald, R.J., 1991. A-type granites revisited: assessment of a residual-source model. Geology, 19: 163-166.

Deblond, A., 1993. G~ologie et l~trologie des massifs ba- siques et ultrabasiques de la ceinture Kabanga-Muson- gati au Burundi. Th~se de doctorat in~dite, Universit6 de Liege, 235 pp. + annexes.

Debon, F. and Le Fort, P., 1988. A cationic classification of common plutonic rocks and their magmatic associations: principles, methods, applications. Bull. Mineral., 111: 493-510.

Demaiffe, D. and Theunissen, K., 1979. Donn6es g6ochron- ologiques U-Pb et Rb-Sr relatives au complexe arch6en de Kikuka (Burundi). Mus. R. Afr. Cent. Tervuren Belg., Dep. Geol. Mineral., Rap. Annu. 1978, pp. 65-69.

Demaiffe, D., Moreau, C., Brown, W.L. and Weis, D., 1991. Geochemical and isotopic (Sr, Nd and Pb) evidence on the origin of the anorthosite-bearing anorogenic com- plexes oftbe A~r Province, Niger. Earth Planet. Sci. Lett., 105: 28-46.

De Paolo, D.J., 1983. Comments on "Columbia River vol-

L. Tack et al. / Precambrian Research 68 (1994) 323-356 355

canism: the question of mantle heterogeneity or crustal contamination". Geochim. Cosmochim. Acta, 47: 841- 844.

Eby, G.N., 1990. The A-type granitoids: a review of their oc- currence and chemical characteristics and speculations on their petrogenesis. Lithos, 26:115-134.

Emslie, R.F., 1991. Granitoids of rapakivi-anorthosite and related associations. Precambrian Res., 51: 173-192.

England, P. and Houseman, G., 1989. Extension during con- tinental convergence, with application to the Tibetan Pla- teau. J. Geophys. Res., 94: 17561-17579.

Fairhead, J.D. and Reeves, C.V., 1977. Teleseismic delay times, Bougner anomalies and inferred thickness of the African lithosphere. Earth Planet. Sci. Lett., 36: 63-76.

Faure, G., 1986. Principles of Isotope Geology. J. Wiley and Sons, New York, N.Y., 589 pp.

Fernandez-Alonso, M. and Tahon, A., 1991. Lithological dis- crimination and structural trends in W-Rwanda (Africa) on images of airborne radiometric and aeromagnetic sur- veys, coreglstered to a Landsat TM scene. Photogramm. Eng. Remote Sensing, 57(9): 1155-1162.

Fitton, J.G., James, D. and Leeman, W.P., 1991. Basic mag- matism associated with Late Cenozoic extension in the western United States: compositional variations in space and time. J. Geophys. Res., 96:13693-13711.

Geological Map of Uganda, 1961. Sheet Kabale and sheet Mbarara, 1/250,000.

Geological Sketch Map NW Tanzania, 1976. 1/250,000, UNDP Mineral Exploration Project, Bukoba, Tanzania.

Harry, D.L., Sawyer, D.S. and Leeman, W.P., 1993. The me- chanics of continental extension in western North Amer- ica: implications for the magmatic and structural evolu- tion of the Great Basin. Earth Planet. Sci. Lett., 117: 59- 71.

Hawkesworth, C.J. and Vollmer, R., 1979. Crustal contami- nation versus enriched mantle: 143Nd/144Nd and STSr/S6Sr evidence from the Italian volcanoes. Contrib. Mineral. Petrol., 69: 151-165.

Hawkesworth, C.J., Gallagher, K., Bradshaw, T.K. and Hooper, P., 1993. Causes of continental magmatism: plumes vs regional tectonics in the Western U.S.A. Terra Abstracts, EUG VII, Strasbourg, 531 pp.

Houseman, G.A. and McKenzie, D.P., 1981. Convective in- stability of a thickened boundary layer and its relevance for the thermal evolution of continental convergent belts. J. Geophys. Res., 86:6115-6132.

Ikingura, J.R., Bell, K., Watkinson, D.H. and van Straaten, P., 1990. Geochronology and chemical evolution of gran- itic rocks, NE Kibaran (Karagwe-Ankolean) belt, NW Tanzania. In: G. Rocci and M. Deschamps (Editors), Re- cent Data in Earth Sciences. Cifeg Occas. Publ. 1990, 22: 97-99.

Ikingura, J.R., Reynolds, P.H., Watkinson, D.H. and Bell, K., 1992. 4°Ar/39Ar dating of micas from granites of NE Ki- baran Belt (Karagwe-Ankolean), NW Tanzania. J. Aft. Earth Sci., 15(3/4): 501-511.

Jacobs, J., Thomas, R.J. and Weber, K., 1993. Accretion and

indentation tectonics at the southern edge of the Ka- apvaal craton during the Kibaran (Grenville) orogeny. Geology, 21: 203-206.

Kinnaird, J. and Bowden, P., 1987. African anorogenic alka- line magmatism and mineralization--a discussion with reference to the Niger-Nigerian Province. Geol. J., 22: 297-340.

Kleemann, G.J. and Twist, D., 1989. The compositionally zoned sheet-like Granite Pluton of the Bushveld Com- plex: evidence bearing on the nature of A-type magma- tism. J. Petrol., 30(6): 1383-1414.

Klerkx, J., Li6geois, J.P., Lavreau, J. and Claessens, W., 1987. Crustal evolution of the Northeastern Kibaran belt, east- ern and central Africa. In: A. KrOner (Editor), Protero- zoic Lithosperic Evolution. Am. Gcophys. Union, Geo- dyn. Set., 17: 217-233.

Krogh, T.R., 1973. A low contamination method for hydro- thermal decomposition of zircon and extraction of U and Pb for isotopic age determination. Geochim. Cosmo- chim. Acta, 37: 485-494.

Lancelot, J.R., 1975. Les syst6mes U-Pb chronom6tres et traceurs de l'6volution des roches terrestres. Th6se de doctorat in6dite, Universit6 Paris VII, 280 pp.

Lancelot, J.R., Boullier, A.M., Maluski, H. and Ducrot, J., 1983. Deformation and related radiochronology in a late Pan-African mylonite bearing shear zone, Adrar des Iforas, Mali. Contrib. Mineral. Petrol., 82: 312-326.

Ledent, D., 1979. Donn6es g6ochronologiques relatives aux granites kibariens de types A (ou G1 ) et B (ou G2) du Shaba, du Rwanda, du Burundi et du SW Uganda. Mus. R. Afr. Cent. Tervuren Belg., Dep. Geol. Mineral., Rap. Annu. 1978, pp. 101-105.

Li6geois, J.P. and Black, R., 1987. Alkaline magmatism sub- sequent to collision in the Pan-African Belt of the Adrar des Iforas (Mali). In: J.G. Fitton and B.J.G. Upton (Ed- itors), Alkaline Igneous Rocks. Geol. Soc. London, Spec. Publ., 30: 381-401.

Li6geois, J.P., Theunissen, K., Nzojibwami, E. and Klerkx, J., 1982. Granito'/des syncin6matiques kibariens au Bu- rundi: 6tude p6trographique, g6ochimique et g6ochrono- logique pr61iminaire. Ann. Soc. Geol. Belg., 105: 345-356.

Li6geois, J.P., Claessens, W., Camara, D. and Klerkx, J., 1991. Short-lived Eburnian orogeny in southern Mall Geology, tectonics, U-Pb and Rb-Sr geochronology. Precambrian Res., 50: 111-136.

Livaccari, R.F., 1991. Role of crustal thickening and exten- sional collapse in the tectonic evolution of the Sevier- Laramide orogeny, western United States. Geology, 19: 1104-1107.

Ludwig, K.R., 1980. Calculation of uncertainties of U-Pb isotope data. Earth Planet. Sci. Lett., 46:212-220.

McKenna, L.W. and Walker, J.D., 1990. Geochemistry of crustally derived leucocratic igneous rocks from the Ulugh Muztagh area, Northeastern Tibet and their implications for the formation of the Tibetan plateau. J. Geophys. Res., 95: 21483-21502.

Nelson, D.O., 1983. Implications of oxygen-isotope data and

356 L. Tack et al. / Precambrian Research 68 (1994) 323-356

trace-element modelling for a large-scale mixing model for the Columbia River basalt. Geology, 11: 248-251.

Ntungicimpaye, A. and Tack, L., 1992. Les mrtavolcanites interm&liaires ~ acides kibariennes du NW du Burundi. IGCP no. 255 Newslett./Bull., 4: 45-50.

Nzojibwami, E., 1987. Le Pr~ambrien cristaUin de la rrgion de Bujumbura (Burundi). Th~se de doctorat inrdite, Universit6 de Liege, 232 pp. + annexes.

O'Brien, C., Plant, J.A., Simpson, P.R. and Tarney, J., 1985. The geochemistry, metasomatism and petrogenesis of the granites of the English lake District. J. Geol. Soc. Lon- don, 142: 1139-1157.

Pearce, J.A., Harris, N.B.W. and Tindle, A.G., 1984. Trace element discrimination diagrams for the tectonic inter- pretation of granitic rocks. J. Petrol., 25(4): 956-983.

Platt, J.P., 1993. Exhumation of high-pressure rocks: a re- view of concepts and processes. Terra Nova, 5 (2): 119- 133.

Pohl, W., 1987. Metallogeny of the northeastern Kibaran belt, Central Africa. Geol. J., 22:103-119.

Rogers, J.J.W. and Greenberg, J.K., 1990. Late-orogenic, post- orogenic and anorogenic granites: distinction by major- element and trace-element chemistry and possible origins. J. Geol., 98: 291-309.

Rumvegeri, B.T., 1991. Tectonic significance of Kibaran structures in Central and Eastern Africa. J. Afr. Earth Sci., 13(2): 267-276.

Steiger, R.H. and J~iger, E., 1977. Subcommission on geo- chronology: convention on the use of decay constants in geo- and cosmochronology. Earth Planet. Sci. Lett., 36: 359-362.

Sylvester, P.J., 1989. Post-coUisional alkaline granites. J. Geol., 97: 261-280.

Tack, L. and Deblond, A., 1990. Intrusive character of the Late Kibaran magmatism in Burundi. IGCP no. 255 Newslett./Bull., 3: 81-87.

Tack, L., De Paepe, P., Deutsch, S. and Lirgeois, J.P., 1984. The alkaline plutonic complex of the Upper Ruvubu (Bu- rundi): geology, age, isotopic geochemistry and implica- tions for the regional geology of the western Rift. In: J. Klerkx and J. Michot (Editors), Grologie africaine--Af- rican Geology. Mus. R. Afr. Cent Tervuren, pp. 91-114.

Tack, L., De Paepe, P., Lirgeois, J.P., Nimpagaritse, G., Ntungicimpaye, A. and Midende, G., 1990. Late Kibaran magmatism in Burundi. J. Afr. Earth Sci., 10(4): 733- 738.

Tack, L., Sindayihebura, A. and Cimpaye, D., 1992. The Nkoma (SE Burundi): an episodically reactivated lower Burundian (Middle Proterozoic) siliciclastic sequence,

locally overlain by a Matagarasian (Upper Proterozoic) sedimentary breccia. IGCP no. 255 Newslett./Bull., 4: 31- 43.

Theunissen, K., 1988. Kibaran thrust fold belt (D1-2) and sbear belt (D2). IGCP no. 255 Newslett./BuU., 1: 55-64.

Theunissen, K., 1989. On the Rusizian basement rise in the Kibara belt of Northeastern Lake Tanganyika. Collision belt geometry or restraining bend emplaced in the Late Kibaran strike-slip environment. IGCP no. 255 News- lett./Bull., 2: 85-92.

Thompson, R.N., Morrison, M.A., Hendry, G.L. and Parry, S.J., 1984. An assessment of the relative roles of crust and mantle in magma genesis: an elemental approach. Philos. Trans. R. Soc. London, Ser. A, 310: 549-590.

Turner, S.P., Foden, J.D. and Morrison, R.S., 1992. Deriva- tion of some A-type magmas by fractionation of basaltic magma: an example from the Padthaway Ridge, South Australia. Lithos, 28:151-179.

van Straaten, P., 1984. Contributions to the geology of the Kibaran belt in northwest Tanzania. Geology for Devel- opment Newsletters, 3, Unesco, pp. 59-68.

Wadsworth, W.J., Dunham, A.C. and Almohandis, A.A., 1982. Cryptic variation in the Kapalagnlu layered intru- sion, western Tanzania. Mineral. Mag., 45: 227-236.

Wager, L.R. and Brown, G.M., 1968. Layered Igneous Rocks. Oliver and Boyd, London, 588 pp.

Waleffe, A., 1965. Etude grologique du Sud-Est du Burundi (rrgions du Mosso et du Nkoma). Ann. Mus. R. Afr. Cent. Tervuren Belg., Srr. 8 °, Sci. Geol., 48, 312 pp.

Walraven, F., Armstrong, R.A. and Kruger, F.J., 1990. A chronostratigraphic framework for the north-central Ka- apvaal craton, the Bushveld Complex and the Vredefort structure. Tectonophysics, 171: 23-48.

Ward, P.L., 1991. On plate tectonics and the geologic evolu- tion of southwestern North America. J. Geophys. Res., 96: 12479-12496.

Watson, E.B. and Harrison, T.M., 1983. Zircon saturation revisited: temperature and composition effects in a vari- ety of crustal magma types. Earth Planet. Sci. Lett., 64: 295-304.

Whalen, J.B., Currie, K.L. and Chappell, B.W., 1987. A-type granites: geochemical characteristics, discrimination and petrogenesis. Contrib. Mineral Petrol., 95: 407-419.

White, A.J.R. and Chappell, B.W., 1983. Granitoid types and their distribution in the Lachlan Fold Belt, southeastern Australia. Geol. Soc. Am. Mem., 159: 21-34.

Williamson, J.H., 1968. Least squares fitting of a straight line. Can. J. Phys., 46: 1845-1847.