Late Proterozoic arc–continent and continent–continent collision in the pan-African trans-Saharan belt of Mali

Late Proterozoic arc - continent and continent - continent collision in the pan-African trans-Saharan belt of Mali1 RENAUD CABY AND URANIE ANDREOPOULOS-RENAUD Centre ge'ologique et gdophysique, Centre national de la recherche scient$que, Universitd des sciences et techniques du Languedoc, Place Eugbne Bataillon, 34060 Montpellier, France AND CHRISTIAN PIN Ddpapartement de la ge'ologie, Unite' associde n O 10 du Centre national de la recherche scientifique, Universitd Blaise Pascal, 5 , rue Kessler, 63038 Clermont-Ferrand, France Received December 23, 1987 Revision accepted October 14, 1988 The Tilemsi magmatic arc, preserved along the suture zone of the pan-African trans-Saharan belt of northern Mali, crops out as a series of northeast- to north-northeast-trending strips along the Tilemsi Mesozoic trough and is about 100 krn in width. The volcanic arc series includes pillowed metabadts o f thdeiitic character and associated with rhyodacites. Overlying sedimentary rocks are turbiditic volcanic greywackes. They are progressively recrystallized into grey gneiss in the vicinity of gabbro-noritic and dioritic intrusions. UIPb zircon dating of a crosscutting metaquartz diorite gives a nearly concordant age of 726': Ma, while that of a plagiogranite mobilizate associated with the grey gneiss is 71022 Ma. Initial Nd and Sr isotopic compositions of two metaquartz diorites (eNd,,, = +6.6, +6.3; (87Sr186Sr)i = 0.7024) are in a agreement with a depleted mantle source similar to modem intraoceanic arcs. Isotopic compositions of two Tilemsi metagreywackes (eNd,,, = +5.8, +4.3; 87Sr186Sr = 0.7027) exclude any significant derivation from an older sialic source and support the ensimatic origin of the magmatic arc. A UIPb zircon age of 635 k 5 Ma has also been obtained on a pretectonic granodiorite batholith at the eastern margin of the arc. Isotopic composition of this intrusion (eNd, = -6.0, -6.4; (87SrlS6Sr)i = 0.7046) illustrates the lack of a genetic link between the 730 Ma old, mantle-derived magmas and these granitoids, which originated from a crustal reservoir. This change in magmatic source is interpreted as the result of accretion of the ensimatic arc along the eastern continent, preceding continent -continent collision during the pan-African event. L'arc magmatique du Tilemsi souligne la zone de suture de la chaine pan-africaine trans-saharienne dans le nord du Mali. Large d'une centaine de kilomttres et de direction nod-est ?I nord-nod-est, il affleure sporadiquement ?i la bordure est du fossC mtsozoique du Tilemsi. La sBrie volcanique est representee par des mktabasaltes tholeiitiques en coussins Ctroitement associts ?I des rhyodacites. Les stdiments qui la surmontent sont des grauwackes volcaniques turbiditiques progressivement recristallisBs en gneiss gris 21 proximitide vastes intrusions gabbro-n&tiques et dioritiques. La datation UlPb sur zircon d'une mktadiorite quartzique caractBristique de cette famille donne un Lge de 726'; Ma et celle d'un mobilisat de plagiogranite associC aux gneiss gris un Lge de 710'2 Ma. Les compositions isotopiques initiales en Nd et Sr de deux Bchantillons de mktadiorites quartziques (eNd,,, = +6,6, +6,3; (87Sr186Sr)i = 0,7024) ttmoignent d'une source mantellique appauvrie analogue ?i celle des arcs intra-ocCaniques modemes. Celles de deux mCtagrauwackes du Tilemsi (eNd,,, = +5,8, +4,3; 87Sr/86Sr = 0,7027) excluent une participation importante de mattriaux dCrivCs d'un socle sialique ancien et confortent l'origine ensimatique de l'arc. Un Lge de 635 k 5 Ma est obtenu par la mCthode UIPb sur zircon sur une granodiorite pktectonique situBe 21 la bordure orientale de l'arc. Les donnCes isotopiques sur ce massif (eNdi = -6,0, -6.4; (87Sr/86Sr)i = 0,7046) soulignent l'absence de lien entre les roches magmatiques mantelliques 2I 730 Ma et ces granitoides dominks par une composante crustale. Ce changement de source magmatique est interprBtt comrne le rksultat de l'accrttion de l'arc ensimatique au continent oriental avant la collision continent-continent au pan-Africain. Can. J. Earth Sci. 26, 1136- 1146 (1989) Introduction The pan-African trans-Saharan belt formed during the colli- sion of two continental masses at ca. 600 Ma (Caby et al. 1981 ; Fabre 1982; Fabre et al. 1982). The suture zone of this collisional orogen runs from western Hoggar to the Gulf of Benin (Fig. 1, inset) and can be defined, on geological grounds, as the sharp contact between two major geological domains (Caby 1987), namely, (i) Late Proterozoic metasedi- mentary units deposited on the passive continental margin of the west African craton, to the west, and (ii) pan-African gneisses derived from both plutonovolcanic assemblages and reworked basement, to the east. The western domain includes very high-pressure metamorphic allochthons (white schists and eclogitic rocks) that are exposed 350 km south of the study area in Gourma (Caby 1987). The eastern domain includes the 'Contribution to IGCP 215: Proterozoic Fold Belts. Printed in Canada 1 Imprimd au Canada study area that has been considered as a Late Proterozoic mag- matic arc preserved at a rather shallow crustal level (Caby 1978, 1987). Between these two major geological domains, remnants of an oceanic domain represented by metabasalts with blue amphibole, slices of Cr-rich mantle harzburgites, gabbros, and a chaotic melange are locally preserved at Time- trine. On geophysical grounds the suture zone is defined well by a string of positive gravity anomalies, related to exposed or shallowly burried, metabasic massifs (Fig. 1). Modeling of these gravity highs (Bayer and Lesquer 1978; Ly et al. 1980) suggests that they represent dense east-dipping masses (D = 3.0 g/cm3), up to 12 km thick. In contrast with western Hoggar and central Iforas, where Late Proterozoic pan-African volcanic and volcaniclastic assemblages display geochemical affinities of an active conti- nental margin (Chikhaoui et al. 1978) and are floored by pre- C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. CABY ET AL. 1137 pan-African basement, the study area is composed entirely of volcanic, volcaniclastic, and plutonic rocks. These include mid-ocean ridge basalt (M0RB)-like basalts, arc tholeiites, and calc-alkaline series (Caby et al. 1986; C. Dupuy and J. Dostal, unpublished results, 1985). To the east, the studied domain is truncated by a major, vertical megashear (at 0°55'E), a left-lateral, synmeta- morphic, strike-slip fault, which was reactivated during late- and post-pan-African events. Pan-African overprinting (of greenschist facies grade to the north) progressively reaches amphibolite facies conditions south of 19 "25'N. Late-kinematic, low-pressure-type associa- tions of the granulite facies are developed in the high- temperature Aguelhoc province (Fig. 1) where no retrogressive alteration took place. A conventional Rb - Sr age on one biotite separate at 588 f 10 Ma (Bertrand and Davison 1981) may indicate the approximate cooling date of this province. To put time constraints on the precollisional evolution of the Tilemsi magmatic arc (TMA) and to draw some inferences about its geodynamic setting, U-Pb, Sm-Nd, and Rb-Sr systematics of the major elements of three representative pre- pan-African massifs (Fig. 1) were investigated. Geologic framework and petrographic notes on the Tilemsi magmatic arc Detailed mapping has shown that all igneous complexes of the TMA are intrusive into volcanic-volcaniclastic assem- blages of the Tilemsi Group. Also, detailed mapping has docu- mented a gradual transition from greenschist and greenstone of the Tilemsi Group in the north, southward to pre-pan-African grey gneisses, which are spatially associated with gabbro- noritic lopoliths (Fig. 1). Tilemsi Group Bimodal volcanics overlying slices of a marble-dolomite sequence represent the oldest regionally extensive unit of the Tilemsi Group. Metabasalts, up to 800 m thick, are commonly pillowed, brecciated, and interfingered with dacitic breccias. Metabasalts are arc tholeiites according to major-element, trace-element, and rare-earth-element (REE) geochemistry (Caby et al. 1986; C. Dupuy and J. Dostal, unpublished results, 1985). Volcanic greywackes make up a regularly layered unit more than 3 km thick, with layers and lenses of plutonovolcanic-pebble conglomerate. These overlie the volcanics and are, in turn, cut by dyke swarms. Locally pre- served sedimentary features in undeformed hornfels (slumps, convolute laminations, incomplete Bouma's sequences, etc.) suggest a turbiditic origin. The greywackes contain crystal and lithic fragments of basalt, andesite, and dacite. Chemical com- positions range from dacitic to andesitic; there is also a Na20 enrichment (Caby et al. 1986). In the vicinity of gabbro-noritic intrusions, such as exposed at Techalrk (Figs. 1, 2), metagreywackes are progressively affected by a pre-pan-African, high-temperature metarnor- phism that is responsible for their transformation into banded gneissic hornfels cut by several generations of leucodiorite, tonalite, and trondjhemite veins. The hornfels units were fur- ther transformed into grey gneiss, which includes foliated dio- rite and tonalite sheets. The mineral assemblage of the grey gneiss (partly retrogressed during pan-African greenschist metamorphism), which includes calcic plagioclase + brown amphibole + ilmenite + CPX OPX, indicates low-pressure (?) granulite facies conditions. Chemical compositions range from quartz-dioritic to granodioritic and trondjhemitic (Caby et al. 1986; C. Dupuy and J. Dostal, unpublished results, 1985). Leucosomes are leucodioritic to trondjhemitic (sample IC 1569). The gneissic units were involved in a high- temperature polyphase folding, associated with the intrusion of gabbro -norite. Low-grade mineral associations (chlorite, epi- dote, actinote) are mostly developed as veinlets in the less duc- tile lithologies and indicate variable extent of retrogression of both grey gneiss and plutonic rocks. Volcanosedimentary formations that outcrop to the southwest and north of Tessalit are petrographically similar to those of the Tilemsi Group. However, they are overlain by a metapelitic unit including a polygenic mixitite with abundant matrix, interpreted as a glaciomarine deposit. This unit is cor- related with the Tafeliant formation (outcropping 150 km to the south-southwest) where it unconformably overlies both the 2000 Ma old basement and a quartz diorite pluton dated at 693 +! Ma (Caby and Andreopoulos-Renaud 1985). Intrusive rocks Several generations of dykes and plutons were emplaced in the Tilemsi Group. Dykes of dolerite, microdiorite, and dacite are ubiquitous within the greywackes. Dacites also form north-northeast- trending screens of dykes, several kilometres wide and some tens of kilometres long. Some siliceous dyke swarms cut across earlier mafic dykes (dolerite, microgabbro). These dyke complexes may be tentatively related to extrusion of basalt and dacite stratigraphically above the turbiditic greywacke, and later removed by erosion. Plutons form more than 70% of outcrops in the arc. The syn- to late-kinematic suite (of pan-African age and not dealt with here, Fig. 1) intruded around 620 Ma in the central Iforas (Caby and Andreopoulos-Renaud 1989). They are spatially unrelated and partly younger than the dyke suite described above. The lack of any subsequent, retrogressive metamorphic imprint allows these plutons to be distinguished from the pre-pan-African suite. There are four main pre-pan-African plutonic associations in the Tilemsi arc: (1) Laccoliths of gabbro, in many places layered, with microgabbroic margins and associated sills. These are slightly differentiated clinopyroxene gabbros with minor troctolites and dunites. Although they resemble ophiolitic gabbros, they clearly intrude the oldest carbonate formation. They are char- acterized by low incompatible element abundances, flat or slightly depleted light-rare-element (LREE), and an increase of V and Ti with differentiation index. These criteria are in agreement with a tholeiitic character, and suggest an affinity with MORB (Caby et al. 1986; C. Dupuy and J. Dostal, unpublished results, 1985). (2) Gabbro-noritic lopoliths, such as the >200 km2 body at Techalr6, spatially associated with grey gneisses. These are gneissic quartz norites with OPX, CPX, brown amphibole, ilmenite, and calcic plagioclase. Magmatic layering is partly affected by high-temperature folding of a style similar to that observed in the grey gneisses. The layering is crosscut by both plagioclase-rich and pyroxenite segregations, displaying a complex, anastomosing geometry. Crosscutting veins of noritic gabbro, and later plagiogranite and quartz diorite pegmatoids, suggest complex and multiple magma injections. (3) Metaquartz-dioritic sheets, up to several hundreds metres in thickness, intrude the grey gneiss and mobilized metagrey- wacke. A magmatic layering is defined in outcrop by globular, C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. 1138 CAN. J . EARTH SCI. VOL. 26,1989 C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. CABY ET AL. 1139 bluish, quartz phenocrysts and by mafic layers richer in OPX and CPX. Sample IC 1581 was selected (for its unusually high zircon content) from a 200 m thick sheet related to the southern part of Techald noritic gabbro (see Appendix for petrography of analy sed samples). (4) Southwest of Tessalit, granitoids occur as elongate batho- liths intruding volcanosedimentary formations. They have narrow contact metamorphic aureoles. K-feldspar and biotite are scarce, while plagioclase and amphibole are abundant. This suggests a calc-alkaline affinity for this granodiorite family. Sample IC 329 was collected from a small intrusion (with a strong postmagmatic greenschist facies imprint) out- cropping along the Tilemsi - Gao road (Fig. 1). Pan-African structures and tectonic setting of the Tilemsi magmatic arc Pan-African deformation and metamorphism affected the volcanosedimentary formations and the four plutonic associa- tions of the Tilemsi arc. This major tectonothermal event is considered as a shield-scale, 620-600 Ma old orogenic event related to continent-continent collision (Caby 1987). In the northern part of the study area, pan-African deforma- tion is characterized by the occurrence of upright folds with associated steep, axial-plane, slaty cleavage generated in a pre- dominantly flattening regime. This is demonstrated by moderate stretching and flattening of pebbles and basalt amyg- dules, parallel to the slaty cleavage. This faint deformation, suggestive of a moderate (less than 50 %) east - west shorten- ing, progressively increases toward the steep, north - south- trending, sinuous shear zones. These are delineated by thick zones of ultramylonite (> 100 m wide) of greenschist grade, with nearly horizontal stretching and mineral lineations. Shear criteria (sense of bending of the regional foliation on both sides, shear bands, and rotated porphyroclasts) indicate dextral strike-slip movement. To the south, inclined and recumbent structures progres- sively develop, and are associated with amphibolite facies metamorphism. Fold geometry was partly controlled by the synkinematic massifs. Refolding of the primary metamorphic fabric also characterizes the high-temperature Aguelhoc gneisses that outcrop south of a large synkinematic tonalite- trondjhemite massif. This high-temperature province is charac- terized by upper amphibolite facies and granulite facies mineral assemblages and by synmetamorphic norite and trocto- lite bodies. The lack of any retrogression allows separation of these pan-African metamorphic assemblages from those in the grey gneiss (north of 20°N). Structural investigations in an area 50 km west of the Tilemsi Mesozoic trough have revealed west-directed overthrust sheets involving both passive conti- nental margin sediments and an oceanic assemblage (Caby 1987). The latter includes metabasalt with blue amphibole, diabase, slices of Mg-metagabbro, and serpentinized mantle peridotite. These nappes suggest that the Tilemsi magmatic arc was also thrust westward onto the west African craton (Fig. 3). The syn- and post-kinematic suite emplaced in the Tilemsi domain includes diorite, trondjhemite, and layered complexes (gabbro - troctolite - norite - anorthosite) but no granites. Therefore, we hypothesize that the Tilemsi magmatic arc did not form above older sialic crust. This hypothesis has been tested by Sm-Nd and Rb-Sr data, and the results are pre- sented below. U - Pb Results Samples for U-Pb zircon analyses were collected from the leucosome portion of the grey gneisses of the Tilemsi Group (IC 1569), the pre-pan-African metaquartz diorite sheets (IC 1581), and the calc-alkaline granites (IC 329). Zircon populations were split into size and magnetic frac- tions using conventional methods. The grains were carefully hand picked, but were not further selected (e.g., by air abra- sion). The analyses were performed at Unit6 propre no 361, Centre national de la recherche scientifique, Montpellier. Dissolution and separation of Pb and U followed the proce- dures described by Krogh (1973), with total lead blanks rang- ing from 60 to 560 pg. The mass spectrometric analyses were carried out on a Cameca TSN 206 S mass spectrometer equipped with an electron multiplier. Analytical data are reported in Table 1 and plotted on a Con- cordia diagram (Fig. 4). Zircon data from the metadiorite (IC 1581) are nearly concordant and plot along a line intersecting the Concordia curve at 726+: Ma and 103 f 270 Ma. Three size fractions from the leucosome (IC 1569) were analysed. These define a line whose intercepts with the Concordia curve give ages at 710'9 Ma and 23 f 100 Ma. The linear array of the granodiorite points yields an upper intercept at 635 f 5 Ma, and a lower intercept at 28 f 48 Ma. In each case, the zircons show minor discordance. There is no evidence for inheritance, and it is considered that the upper intercepts may confidently be interpreted as reflecting magmatic crystalliza- tion ages. In contrast, the lower intercepts are very imprecise, and do not differ from zero, within error. Although they indi- cate minor, recent lead loss, they have no simple geological interpretation. Nd - Sr results To complement the geochronological results, a preliminary Sr and Nd isotopic study was done at Unit6 associ6e no 10, Centre national de la recherche scientifique, Clermont- Ferrand. Two samples from each formation studied by zircon dating were analysed, following the procedure described by Pin and Poidevin (1987). The results are reported as E units (DePaolo and Wasserburg 1976). It is apparent that the results for samples from a same geological unit agree closely FIG. 1. Geological sketch map of the Tilemsi magmatic arc. ( 1 ) Molasse and associated volcanics; (2) undifferentiated syn- to late-tectonic batholith; (3) synmetamorphic intrusions: (a) tonalite and (b) troctolite-norite; (4) Aguelhoc gneisses with high-temperature low-pressure pan-African metamorphism. Proterozoic magmatic suite of the Tilemsi arc: (5) granodiorite; (6) quartz diorite; (7) gabbro-norite lopoliths; (8) ~ilemsi Group: (a) m&abasalts, (b) acid me&volcanics, and (c) mafic dyke complexes; (9) idem: (a) volcanic greywackes and (b) grey gneisses; (10) main shear zone. Unpattemed areas: undifferentiated phanerozoic. INSET: ( 1 ) West African craton stable since 2000 Ma (a) and its possible extension beneath the Late Upper Proterozoic (b); (c) granulite nuclei with in the pan-African belt. The pan-African belt: (2) Upper Proterozoic parautochthons and allochthons of the frontal part of the belt; (3) undifferentiated rocks of (a) Late Proterozoic age and (b) the Tilemsi magmatic arc; (4) monocyclic and polycyclic gneisses with mesocatazonal, pan-African metamorphism; (5) eastern Hoggar domain stabilized 730 Ma ago (a) and its eastward extension (b); (6) pan-African suture. Ti, Timetrine; G, Gourrna. C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. CAN. J . EARTH SCI. VOL. 26,1989 C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. CABY ET AL. I - " a 'E 2 : (':: 4 E C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. 1142 CAN. J . EARTH SCI. VOL. 26,1989 TABLE 1. U -Pb analytical results Atomic ratio Apparent age rubPb Conc. (pglg) - 9 Atomic percent 204Pb 206pb* 207pb* 207pb* 206pb 207pb 207pb Sample fraction Weight - - - --- (am) (mg) U Pb measured 206Pb* 207Pb' 2WPb* 238U 2 3 5 ~ 206pb* 238U 235U 206pb Metaquartz diorite Leucosome (IC 1569) 63- 80, M l o 8.4 92.4 11.89 840 74.53 4.67 20.78 0.111 35 0.9671 0.06299 681 687 708 80-100, M l o 16.6 84.8 10.61 1439 72.81 4.59 22.60 0.10575 0.9185 0.06299 648 662 708 100-125, M l o 4.8 82.2 11.13 1196 73.30 4.62 22.08 0.115 17 1.0012 0.06305 703 704 710 Granodiorite (IC 329) 43- 74, NM 0" (A) 4.0 105.0 11.41 765 80.66 4.91 14.43 0.101 90 0.8549 0.06084 626 627 634 63- 80, M 5" (B) 5.7 145.3 13.50 888 80.78 4.91 14.31 0.08723 0.7310 0.06078 539 557 631 80-100, 5" (B) 3.1 132.2 12.85 566 80.64 4.90 14.47 0.091 10 0.7631 0.06075 562 576 630 100-160, 5" (B) 1.0 112.6 11.74 692 80.28 4.89 14.83 0.09720 0.8161 0.06089 598 606 625 NOTES: NM, nonmagnetic; M, magnetic; *, radiogenic. Isotope composition (Z04~b:206~b:mPb:208Pb) used for common lead correction: metadiorite and mobilizate, 1: 17.89: 15.50:37.55; granodiorite A, 1: 18.03: 15.62:37.33; granodiorite B, 1: 17.53: 15.42:37.25. Chemistry blanks: metadiorite and plagiogranite leucosome, 60 pg Pb; granodiorite, 560 pg Pb. Decay constants: XZ3'U = 1.551 25 X a-l, XZ3'U = 9.8485 X lo-'' a-'. TlLEMSl MAGMATIC ARC METAQUARTZ DlORlTE 726 r : ~ a CALCIC LEUCOSOME 710 +EM= P R E M E T A M O R P H I C GRANODIORITE FIG. 4. Concordia diagram: Metadiorite (IC 1581); Ir plagiogranite leucosome (IC 1569); W granodiorite (IC 329). (Table 2). This suggests that the formations studied are Srisotoperatios (0.7024-0.7025). Fromthis, weinfera time- roughly homogeneous. integrated, highly depleted mantle source. Liegeois (1987) The 730 Ma old metadiorites (IC 431 and IC 1582) have obtained higher positive eNd (+ 8.3, +9.5) on two metanoritic high initial eNd (+6.3, +6.6), and quite nonradiogenic initial gabbros from the Techalr6 complex. Despite their elevated C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. CABY ET AL. TABLE 2. Isotopic Sr-Nd results 87Rb 87Srb 1 4 7 s ~ ~ Sample 143Ndd - - No. Rba Sra "Sr Smc NdC 144Nd '44Nd "Rb and Sr determined by X-ray fluorescence spectrometry; precision 1 % for Sr, 5 % for Rb. bNormalized to 86Sr/88Sr = 0.1194; precision 20.1 % (20)-NBS 987 = 0.71023 f 3. "Srn and Nd determined by isotope dilution mass spectrometry; precision of 147Sm/'"Nd 0.2%. dNormalized ratio 146Nd/1"Nd = 0.7219; uncertainties quoted at the 95% confidence level, based on within-mn statistics. The La Jolla isotopic standard gave a 0.51 1 847 f 6 (N = 19) value during the period of these analyses. e87Sr/86Sr after age correction using the U -Pb data (t). feNd, = relative deviation, parts in 1@, with respect to a chondritic reservoir of identical age (DePaolo and Wasserburg 1976). gModel age (in Ma) calculated with respect to a chondritic reservoir T,,,, (DePaolo and Wasserburg 1976) and with respect to a depleted-mantle reservoir T,, (DePaolo 1981). values, our eNDi are significantly lower than those predicted by evolutionary models of MORB source - 8.5, Zindler 1982; Jacobsen and Wasserburg 1980), and do not support derivation from oceanic crust. However, these esti- mates are consistent with an intraoceanic island-arc setting (DePaolo 1981 ; White and Patchett 1984), in agreement with geological evidence presented in the previous sections. Similar results were reported for roughly coeval ophiolites of inferred back-arc origin in Saudi Arabia (Claesson et al. 1984). The greywacke units, in which the dioritic plutons were emplaced, have positive EN^^^^, albeit significantly lower than the diorites (+4.4, +5.8). This difference is reflected also by slightly more radiogenic initial Sr (0.7027 -0.7028). Never- theless, the data indicate the limited contribution of old, sialic components, and therefore substantiate a predominantly juve- nile origin of the greywacke units. In marked contrast to other units, the 635 Ma old granodiorites have strongly negative initial eNd (-6.1, -6.4), coupled with more radiogenic (87Sr/86Sr)i (0.7046). The eNd, values show that the grano- diorites were extracted from a time-integrated, LREE-enriched source with continental affinities. A significant contribution of ancient, sialic crust is suggested by Nd model ages (TCHUR = 1250 Ma, TDM = 1700 Ma) well in excess of the emplace- ment age. The moderate enrichment in radiogenic Sr could be inter- preted in two different ways. In the first model, the negative eNdi and comparatively low 87Sr/86Sr could be interpreted as reflecting a source reservoir characterized by a long-term LREE enrichment, and a low, time-integrated Rb/Sr ratio. A felsic, high-grade basement domain depleted in Rb during an old granulite facies metamorphic event would fit these require- ments. Alternatively, the apparent decoupling of Sr and Nd isotopes might result from mixing process between a mantle component characterized by a high Sr/Nd ratio (e.g., a calc-alkaline basalt), and an old crustal end member with a much lower Sr/Nd. If so, the mass input (DePaolo and Wasseburg 1976) would be dominated by the crustal component for Nd, and by the mantle component for Sr. Thus, the resulting blends would exhibit crustal Nd isotopic signatures, but would be interme- diate in terms of Sr isotopes. Geologically, the first hypothesis would suggest that old granulitic terrains occur beneath the area where granodiorites were emplaced. Such a high-grade domain outcrops in the neighbouring central Iforas unit. However, its (87Sr/86Sr)700 values fall near 0.7095 (Liegeois 1987). If a ca. 2 Ga age is assumed for this hypothetical basement, then the younger model ages require the involvement of a more juvenile com- ponent. The second model assumes pervasive contamination of a mafic melt (with high Sr/Nd) by low SrINd, isotopically ancient material. Granitic melt, generated by anatexis of metasedimentary rocks would be an appropriate contaminant. In this case, and ensialic setting is not a necessary requirement. In any case, a sialic component is indicated for the granodio- rites. This highlights the contrasting settings, that is, sialic crust, either cratonic or sedimentary, versus ensimatic arc, of crustal domains situated on both sides of the 0°55E trans- current fault zone. Summary and geodynamic implications The Tilemsi arc differs from other domains of the trans- Saharan pan-African belt by the abundance of pre-, syn-, and post-tectonic, mafic plutons and the lack of granites. It also differs from other volcanic-volcaniclastic units of the belt (i.e., the Tafeliant Group exposed 150 krn to the south- southwest, the SCrie Verte of Hoggar (Caby 1987)) that are floored either by pre-pan-African basement or by 1 1.8 Ga old basinal sequences of orthoquartzites and Al-metapelites (Caby and Andreopoulos-Renaud 1983). The Tilemsi arc is delimited to the west by an east-dipping oceanic unit of metabasalt and mantle peridotite that lies in sharp contact with nappes of passive continental margin metasediments deposited on the eastern edge of the west African craton. The latter include very high-pressure metamorphic assemblages typical of the subduc- tion -0bduction geodynamic settings (Caby 1987). U-Pb radiometric and Sr-Nd geochemical data allow an assessment of the geodynamic evolution of the Tilemsi zone, and its consequences for the pan-African suturing in north- ern Mali. A pretectonic metadiorite, emplaced within the metagrey- wackes of the Tilemsi Group and belonging to the first plutonic suite, yields an age of 726:: Ma. Values of eNdi of about +6 are consistent with geological evidence indicating an origin in an intraoceanic island arc. C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. 1144 CAN. I. EARTH SCI. VOL. 26,1989 FIG. 5. Evolutionaly plate tectonics diagram. TA, Tilemsi arc; WAC, west African craton; ACM, active continental margin; EC, eastern continent. Metagreywackes of volcanic derivation have EN^,^,, of +4.4 and + 5.8. These estimates are not very sensitive to age correc- tion (e.g., +4.9 and +6.0 at 800 Ma). These high values pre- clude a major contribution of ancient terrigeneous, sialic crust and further substantiate an ensimatic origin for the volcanic arc. With increasing metamorphism, the metagreywackes grade into grey gneisses, of the hornblende granulite facies. A predominantly thermal origin of the metamorphic overprint may be inferred from its close association with gabbro-noritic and dioritic intrusions. U -Pb dating of a plagiogranite leuco- some from the grey gneiss yields a 710:: Ma age for the high- temperature metamorphic episode and, by inference, for the intrusion of gabbro-norite plutons. The ensimatic Tilemsi magmatic suite formed between 730 and 710 Ma, although a slightly older age may be inferred for deposition of the earlier volcanics and greywackes. Interestingly, the major arcs in Saudi Arabia were built during the same epoch, 732 -725 Ma ago (Stacey et al. 1984). There is, however, significant diachronism between the pre- 730 Ma Tilemsi arc and volcaniclastic suites deposited on the active margin of the eastern continental plate, which are younger than a 690:; Ma old metaquartz diorite (Caby and Andreopoulos-Renaud 1985). The pretectonic granodiorite plutons, outcropping at the eastern margin of the arc, are younger (635 f 5 Ma) and are of continental origin, as documented by their continent-like eNdi of about -6. This marked contrast emphasizes the first-order role of the strike-slip 0°55'E fault zone, which separates a western, ensimatic block from an eastern, predominantly sialic domain. In the Iforas central batholith (Liegeois et al. 1987), syn- to late-tectonic plutons were emplaced at 6202; Ma (Caby and Andreopoulos-Renaud 1989). Considering these ages, the pan- African tectonic episode can be precisely bracketed as having occurred between 635 f 5 and 620:; Ma. From these data, the following tentative plate tectonic evolu- tionary scheme is inferred (Fig. 5): Formation of an intra- oceanic volcanic arc (Tilemsi) took place prior to 730 Ma, the age of early plutons emplaced in the volcanic -volcaniclastic pile. At 710 Ma, a high-temperature, synkinematic metamor- phism related to intrusion of gabbro-noritic plutons occurred. After 695 Ma, volcaniclastic rocks derived from penecontem- poraneous andesitic volcanoes accumulated on the active con- tinental margin of the eastern continental plate. At 635 Ma, calc-alkaline, pretectonic granitoids of predominantly crustal derivation were emplaced. This magmatism might have occurred as a result of the eastward subduction of oceanic crust, in a back-arc position relative to the Tilemsi arc. Alter- natively, these granodiorites might be related to partial melting of the continental crust during assembly of the arc along the C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. CABY ET AL. 1145 edge of the eastern continent. The final suturing of continental land masses, that is, the pan-African event s.st., occurred between 620 and 600 Ma, through oblique collision, as evi- denced by major syn- and late-orogenic strike-slip shear zones. This was followed by uplift of the belt, emplacement of late- to post-kinematic granites and molasse deposition. Acknowledgments This work was supported by Centre national de la recherche scientifique (CNRS), Action spkcifique Afrique. We are indebted to R. Black, C. Dupuy, and J. P. Liegeois for fruitful discussions. We greatly appreciated critical reviews by R. S. Hildebrand and by an anonymous reviewer, which much improved the manuscript. Thanks also to J. Clemens for his help for the language improvement. BAYER, R., and LESQUER, A. 1978. Les anomalies gravimktriques de la bordure orientale du craton ouest-africain : gkomktrie d'une suture pan-africaine. Bulletin de la Sociktk gkologique de France, 6: 863-876. BERTRAND, J. M. L., and DAVISON, I. 1981. Pan-African granitoid emplacement in the Adrar des Iforas mobile belt (Mali): a Rb/Sr isotope study. Precambrian Research, 14: 333 - 361. CABY, R. 1978. Palkogkodynamique d'une marge passive et d'une marge active au Pn5cambrien supkrieur : leur collision dans la chaine pan-africaine du Mali. Bulletin de la Sociktk gkologique de France, 20: 857-861. 1987. The pan-African belt of west Africa from the Sahara desert to the gulf of Benin. In Anatomy of mountain ranges. Edited by J. P. Schaer and J. Rodgers. Princeton University Press, Prince- ton, NJ, pp. 129-170. CABY, R., and ANDREOPOULOS-RENAUD, U. 1983. Age ?I 1800 Ma du magmatisme sub-alcalin associk aux mktaskdiments mono- cycliques dans la chaine pan-africaine du Sahara central. Journal of African Earth Sciences, 1(3/4): 193 - 197. 1985. ~ t u d e petrostructurale et gkochronologique d'une metadiorite quartzique de la chaine pan-africaine de 1'Adrar des Iforas, Mali. Bulletin de la Sociktk gkologique de France, 8: 899 - 903. 1989. Age U-Pb h 620 Ma d'un pluton synorogenique de 1'Adrar des Iforas : cons6quences pour la datation de la phase majeure de l'orogknkse pan-africaine au Mali. Comptes Rendus de 1'AcadCmie des Sciences, Series 2, 308 : 307 -314. CABY, R., DOSTAL, J., and DUPUY, C. 1977. Upper Proterozoic volcanic graywackes from western Hoggar (Algeria). Geology and geochemistry. Precambrian Research, 5: 283-297. CABY, R., BERTRAND, J. M. L., and BLACK, R. 1981. Pan-African ocean closure and continental collision in the Hoggar-Iforas seg- ments, central Sahara. In Precambrian plate tectonics. Edited by A. Kroner. Elsevier, Amsterdam, The Netherlands, pp. 407-434. CABY, R., LIEGEOIS, J. P., DUPUY, C., and ANDREOPOULOS- RENAUD, U. 1986. The Tilemsi magmatic arc and the Pan-African suture zone in northern Mali. International Field Conference on Proterozoic Geology and Geochemistry, central Colorado, U. S. A. Abstract, p. 88. CHIKHAOUI, M., DUPUY, C., and DOSTAL, J. 1978. Geochemistry of late Proterozoic volcanic rocks from Tassendjanet area (N.W. Hoggar, Algeria). Contributions to Mineralogy and Petrology, 66: 157- 164. CLAESSON, S., PALLISTER, J. S., and TATSUMOTO, M. 1984. Samarium-neodymium data on two Late Proterozoic ophiolites of Saudi Arabia and implications for crustal and mantle evolution. Contributions to Mineralogy and Petrology, 85: 244-252. DEPAOLO, D. J. 1981. Neodymium isotopes in the Colorado Front Range, and crust-mantle evolution in the Proterozoic. Nature (London), 291: 193-196. DEPAOLO, D. J., and WASSERBURG, G. J. 1976. Nd isotopic varia- tions and petrogenetic models. Geophysical Research Letters, 3: 249 -252. FABRE, J. 1982. Carte gwlogique et gravimktrique de 1'Adrar des Iforas (Mali) au 11500.000e. Direction nationale de la gkologie et des mines, Bamako, Mali, and Centre gkologique et gkophysique, Montpellier, France. FABRE, J., BA, H., BLACK, R., CABY, R., LEBLANC, M., and LESQUER, A. 1982. La chaine pan-africaine, son avant-pays et la zone de suture au Mali. Notice explicative de la carte gkologique et gravimktrique de 1'Adrar des Iforas au 1/500.000e. Direction nationale de la gkologie et des mines, Bamako, Mali. JACOBSEN, S. B., and WASSERBURG, G. J. 1980. Sm-Nd isotopic evolution of chondrites. Earth and Planetary Science Letters, 50: 138-155. KROGH, T. E. 1973. A low contamination method for decomposition of zircon and the extraction of U and Pb for isotopic age determina- tions. Geochimica et Cosmochimica Acta, 37: 485 -494. LIEGEOIS, J. P. 1987. Le batholite composite de l'Adrar des Iforas (Mali). Gwchimie et gkochronologie d'une succession mag- matique du calco-alcalin ?I l'alcalin dans le cadre de l'orogenkse pan-africaine. Thesis, Universitk libre de Bruxelles, Brussels, Belgium. LIEGEOIS, J. P., BERTRAND, J. M., and BLACK, R. 1987. The sub- duction and collision-related pan-African batholith of the Adrar des Iforas (Mali). In African geology. Edited by J. Kinnaird and P. Bowden. John Wiley, London, United Kingdom, pp. 185 -212. LUDWIG, K. R. 1980. Calculation of uncertainties of U/Pb isotope data. Earth and Planetary Science Letters, 46: 212-220. LY, S., ALBOUY, Y., CHAUVIN, M., FOY, R., LACHAUD, J. C., and LESQUER, A. 1980. Apport de la gravimktrie 5 la compn5hension de la chaine pan-africaine dans 1'Adrar des Iforas. Cahiers O.R.S.T.O.M. Geophysique, 11: 37 -58. PIN, C., and POIDEVIN, J. L. 1987. U -Pb zircon evidence for a pan- African granulite facies metamorphism in the Central African Republic. A new interpretation of the high-grade series of the northern border of the Congo craton. Precambrian Research, 36: 303-312. STACEY, J. S., STOESER, D. B., GREENWOOD, W. R., and FISCHER, L. B. 1984. U - Pb zircon geochronology and geologic evolution of the Halaban - A1 Amar region of the eastern Arabian shield, Kingdom of Saudi Arabia. Journal of the Geological Society, London, 141: 1043 - 1055. WHITE, W. M., and PATCHETT, P. J. 1984. Hf -Nd -Sr and incom- patible element abundance in island arcs: implications for magma origins and crust -mantle evolution. Earth and Planetary Science Letters, 67: 167-185. ZINDLER, A. 1982. Nd and Sr isotopic studies of komatiites. In Kornatiites, Edited by N. T. Arndt and E. G. Nisbet. George Allen & Unwin, London, United Kingdom, pp. 399-420. Appendix. Petrography of analysed samples Sample IC 1581 is a mesocratic metaquartz diorite with OPX, CPX, brown amphibole, biotite, plagioclase (An 50), and bluish quartz. A cataclastic foliation is defined by ribbon quartz and appears coeval with retrogression of the igneous minerals under greenschist facies conditions (actinolite and blue-green amphibole, chlorite, epidote, calcite, white mica). Zircon is intergranular or included in amphibole and plagio- clase, as are other accessory minerals such as ilmenite, rutile, and apatite. Zircon crystals are subhedral, display short and simple prisms with rounded edges, and are translucent. Sample IC 1559 is a calcic leucosome with bluish quartz, minor chloritized amphibole, biotite, ilmenite, sphene, and zircon. Pan-African metamorphism is responsible for blastesis of epidote and white mica and the cloudy appearance of plagio- clase. Zircons are euhedral, and exhibit simple, slightly C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y. 1146 CAN. J. EARTH SCI. VOL. 26,1989 acicular prisms with a pale, yellow-pink colour. gioclase is zoned and partly replaced by epidote, calcite, and Sample IC 329 is a pink cataclastic granodiorite cut by white mica. Globular quartz is strained and partly recrystal- schistose metabasic dykes containing actinolite. Green home- lized. Zircons are euhedral and yellow-orange in colour. They blende is partly replaced by actinolite. This amphibole and display pitted surfaces and contain numerous inclusions, but green chloritized biotite are the main Fe-Mg minerals. Pla- lack any feature suggestive of inherited cores. C an . J . E ar th S ci . D ow nl oa de d fr om w w w .n rc re se ar ch pr es s. co m b y U N IV E R SI T Y O F M IC H IG A N o n 11 /1 1/ 14 Fo r pe rs on al u se o nl y.