JOURNAL OF PETROLOGY VOLUME S7 NUMBER • PAGES 961-991 1996 ANDREAS SPATH1, ANTON P. LE ROEX1* AND ROBERT A. DUNCAN1 'DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF CAPE TOWN, RONDEBOSCH 7700, SOUTH AFRICA "COLLEGE OF OCEANIC AND ATMOSPHERIC SCIENCES, OREGON STATE UNIVERSITY, CORVALLIS, OR 973J1, USA The Geochemistry of Lavas from the Gomores Archipelago, Western Indian D o w Ocean: Petrogenesis and Mantle Source nlo a d e d Region Characteristics fro m h ttp s ://a c a d New mineral and bulk-rock analyses, as well as Nd, Sr and Pb incompatible element abundances of the lavas are interpreted to em isotope compositions are presented for lavas from Grande reflect the effects of a recent mantle enrichment event. At depths ic .o Comore, Moheli and Mayotte, thru of the four main islands of well within the garnet stability field this mantle enrichment is up the Comores Archipelago in the western Indian Ocean, and interpreted to have taken the form of modal metasomatism with .co m these data an used to evaluate the petrogenesis, evolution and the introduction of amphibole (giving rise to the source of LGT /p mantle source region characteristics of Comorean lavas. The magmas), whereas cryptic metasomatism took place at shal- etro typically silica-undersaturated, alkaline lavas from all three lower levels (giving rise to the source of KT lavas). The Nd, lo g islands can be grouped into two distinct types: La Grille-type Sr and Pb isotope signature of the majority of Comorean lavas y /a (LGT) lavas, which display strong relative depletions in K, (both LGT and KT) is proposed to be the result of pre- rtic and Karthala-type (KT) lavas, which do not. With the dominant contributions from a somewhat heterogeneous source le -a exception of the lavas erupted by La Grille volcano on Grande representative of the ambient sub-Comorean mantle, comprising b s Comore, which exhibit the petrographic and geochemical char-a mixture between a HIMU component and a component on the tra c acteristics expected of primary mantle-derived magmas, all depleted portion of the mantle array (possibly the source of t/3 Comorean lavas analysed have experienced compositional modi- Indian Ocean MORB), with only limited contributions from 7/4 fications after they segregated from their source regions. Much ofan EM I plume component. The lavas erupted by Karthala /9 6 1 this variation can be explained quantitatively by fractional volcano (the youngest Comorean lavas), however, have sig- /1 5 crystallization processes dominated by the fractionation ofoliv- nificantly different isotopic compositions from all other Comor- 0 8 ine and clinopyroxene. Semi-quantitative modelling shows that ean lavas (lower 'i3Nd/'"Nd and higher ^Sr/^Sr), 4 8 2 the consistent and fundamental difference in composition suggesting increased contributions from the EM I component. b y between K-depleted LGT lavas and normal KT lavas can be g u attributed to partial melting processes, provided amphibole is aKEY WORDS: basalt petngtntsis; Comores; mantle geochemistry; es residual mantle phase after extraction of LGT magmas at low ocean island basalts t o n degrees of melting. Low absolute abundances of the heavy rare 0 4 earth elements in LGT magmas are interpreted to reflect partial A p melting within the garnet stability field In contrast, KT mag- INTRODUCTION ril 2 mas, which do not show relative K depletions, are considered to Grande Comore, Moheli, Anjouan and Mayotte, the 01 be the products of somewhat larger degrees of partial melting of 9 four main islands of the Comores Archipelago, are an amphibolefree source at comparatively shallower depths. located between ll°30'S, 43°30'E and 13°S, 45°15'E, Whereas the Nd and Sr isotopic compositions of Comorean and form the emergent peaks of a west-northwest- lavas (which show a significant range: r>'Sr/S6Sr = 0-70319- trending volcanic lineament that straddles the 0 • 70393; mNd/M4Nd = 0 -51263-0 -51288) bear evidence for northern entrance to the Mozambique Channel in a time-averaged depletion in incompatible elements, the high the western Indian Ocean (Fig. 1). Detailed petro- •Correjponding author. Telephone: 27-2-6502921. Fax: 27-21-6503781 e-mail: [email protected]. © Oxford University Preu 1996 JOURNAL OF PETROLOGY VOLUME 37 NUMBER • AUGUST 1996 THE COMORES ARCHIPELAGO n. 43*E ri 44°E 46'E GRANDE COMORE (0.13-0 Ml) -12°8 ANJOUAN MOHELI (2.8 - 0.62 Ma) MAYOTTE D o H3'3 5Q^ (5.28-1.49 w 1| n lo a d e d 40°E 60°E 70°E fro m h ttp Nairobi \ s ://a c Mombasa"1: ^Seychelles ad e m ic .o Dares u p Salaam /Aldabra .c o ./ group m 10°S / /pe Farquhar group tro Central Indian Ridge lo g y /a Mascarene Basin rtic le -a b s Tromelin tra c t/3 7 /4 y /9 20°S 6 1 Bassa da India /1 Mauritius 5 0 Europa Reunion Basin Reunion 848 2 b y g u e s t o Fig. 1. Map of the western Indian Ocean, Madagaicar and the toutheait coajt of Africa. The iniet ihowi a doic-up df the Comorci n 0 Archipelago (the range of K—Ar iga of lavas from three of the iilandj ii indicated; EmcricV & Duncan, 1983). 4 A p ril 2 0 logical studies have been conducted on lavas from 1925), and, more recently, the age progression has 1 9 Grande Comore (Flower, 1971, 1973a; Strong, been confirmed by K-Ar dating (Hajash & Arm- 1972a), Moheli (Strong, 19726) and Anjouan strong, 1972; Emerick & Duncan, 1982, 1983). (Flower, 1971, 1972, 1973a, 19736; Thompson & Emerick & Duncan (1982) concluded that the geo- Flower, 1971; Ludden, 1977). The near-linear west- metry and age progression of the Comores Archi- northwesterly trend of the island chain and the pelago are consistent with a hotspot origin. They apparent westward decrease in age from the oldest proposed that the volcanic features which represent and most severely eroded island of Mayotte to the the surface expression of the so-called Comores younger, volcanically active Grande Comore, have mantle plume can be traced in a curvilinear trend long been recognized as significant (e.g. Krenkel, along the length of the Comores Archipelago itself, 962 SPATH tt aL LAVAS FROM THE COMORES ARCHIPELAGO through the northern tip of Madagascar and several the precision and accuracy for all major and minor atolls north of Madagascar to the late Tertiary elements are typically estimated to be below 5%. basaltic dykes which dissect the Seychelles Plateau. Analytical errors and detection limits for the trace Based on chronological and geochemical evidence, elements are similar to those given by le Roex & Devey & Stephens (1991) have, however, suggested Dick (1981), Duncan et al. (1984) and le Roex that the Seychelles dykes are part of the Deccan (1985). Mineral compositions were determined using magmatic event. The Comores hotspot hypothesis a Cameca CAMEBAX electron microprobe (accel- was questioned by Nougier et al. (1986), who argued erating voltage 15 kV; beam current 40 nA) with that no obvious migration of volcanic activity can be on-line data reduction according to the method of demonstrated on the archipelago and suggested that Bence & Albee (1968), using factors from Albee & the Comores were produced by more or less syn- Ray (1970). Rare earth element (REE) abundances chronous volcanic activity along deep lithospheric were determined by high-pressure ion chromato- D o fractures in northern Madagascar and in the graphy (HPIC) following the method of le Roex & wn Mozambique Channel. Yet another model for Watkins (1990), for which accuracy and precision loa d Comorean volcanism proposes that the Comores are typically below 5%. The Sr, Nd and Pb isotopic e d Ridge represents a very slow, and consequently compositions of selected Comorean lavas were fro partly emergent, spreading ridge (Upton, 1982). No determined, using a VG Sector seven-collector m h consensus exists in the literature concerning the thermal ionization mass spectrometer on unleached ttp nature of the crust underlying the Comores Archi- samples. The technique followed for the chemical s://a pelago. The tectonic model for the southeastward separation of Sr and the mass spectrometric deter- c a migration of Madagascar from a pre-rift setting mination of 87Sr/86Sr was similar to that of Hart & de m adjacent to East Africa, starting in the Jurassic (e.g. Brooks (1977). Procedures similar to those of ic Norton & Sclater, 1979; Rabinowitz et al., 1983), Zindler (1980) and Zindler et al. (1979) were used .o u predicts that the archipelago is underlain by oceanic for the separation of Nd and the determination of p.c o crust, whereas some workers have argued for the the Nd isotopic compositions, respectively. The Pb m presence of continental crustal material beneath the isotopic compositions were determined following a /pe Comores (Flower & Strong, 1969). Nougier et al. technique similar to that of Manhes et al. (1978). tro (1986) suggested that the Comores are located near 143Nd/ **Nd ratios were normalized to a value of log y the continental margin in a transitional zone 0*51182 for La Jolla. Because consistently accurate /a between oceanic and continental crust. and precise results were obtained for the 87Sr/86Sr rtic le ratio of the standard NBS SRM 987, no normal- -a In this paper, we present new bulk-rock and b mineral analyses, as well as radiogenic isotope data ization was required for the 87Sr/86Sr ratios of the stra samples. Pb isotope ratios were corrected for frac- c for a selection of lavas from the islands of Grande tionation using the ratios obtained in repeated runs t/37 ACnojmouoaren, wMeorhee lia vaainladb lMe afyoort tes tu(dnyo). saTmhproleusg hf roamn oanf dt hPe bN aBreS aSll RsiMgn if9ic8a1n tslyta nbdelaorwd. 1B nlagn kans df othr eNredfo, reSr /4/961 investigation of their petrography, mineralogy and negligible. /15 bulk-rock chemistry, we attempt to shed some light 0 8 on the role of crystal fractionation processes in the 48 2 petrogenesis of these rocks. Once an understanding b y of these processes in Comorean magmas has been CLASSIFICATION, gu developed, partial melting processes, as well as the e s mantle source region characteristics of the lavas and PETROGRAPHY AND t o n the spatial and temporal evolution of Comorean MINERALOGY OF THE LAVAS 0 4 volcanism are evaluated. A summary of the salient petrographic and miner- Ap alogical features of the Comorean lavas analysed in ril 2 this study is presented in Table 1. The classification 01 ANALYTICAL TECHNIQUES into La Grille-type (LGT) and Karthala-typc (KT) 9 All analytical work was carried out in the lavas which is referred to in Table 1 is based on Department of Geological Sciences, University of geochemical criteria and will be discussed in sub- Cape Town. Bulk-rock samples were analysed for sequent sections of this paper. The compositional major, minor and selected trace element abundances ranges of the phenocryst minerals in Comorean lavas by X-ray fluorescence (XRF) spectrometry, using are presented in Table 1 and selected electron the Norrish fusion technique and pressed powder microprobe analyses arc shown in Tables 2—5. The briquettes (Duncan et al., 1984). International and complete dataset of mineral analyses may be 'in-house' standards were used for calibration and obtained from the first author on request. 963 D o w n lo a d Table 1: Summary ofthepetrographic characteristics ofComorean lavas ed fro m Volcanic centre Rock type" Lava typet Texture Phenocrysts Groundmass O http s or phase ://a Modal proportions Compositions cca d e m Grinds Comas ic Karthala alkali basalt KT slightly to highly vesicular, fine-grainod to .ou op hawaiite sparsely to coarsely cpx7% cryptocrystalllna and partly .c o porphyritic glassy m m/pe La Grille basanlte LGT vesicular, porphyrttic to ol10% very fine-grained to Otrolog us microporphyrttlc cpx6% gcrlayspstoycrystalllne and partly oogy/article s -a MoheJi b s (•uatigraphtc alkali basalt KT vesicular, coarsely 0111% F0«2-B1 fine-grained tra c relationships porphyritic to cpx11% t/3 7 not mtcroporphyrttic /4 /9 established) 6 1 basanite LGT slightly vesicular. 0113% fine-grained to /15 0 porphyritic to cpx9% cryptocrystalllne and partly 8 4 microporphyrrtlc glassy 82 b y g nephelinlte LGT slightly vesicular. oil 4% F077-M very fine-grained to ue s porphyritic cpx3% cryptocrystalllne t o n 0 4 Mtyotte A p RPheajusveenescent ablaksaalin bitaesalt vmeiscircouploarrp.hyritlc to copl6x%4% F073-S6 foicncea- stioo nvaelrlyy fpinaret-lgyr galiansesd, ril 20 1 trachybasalt seriate-textured and amph ~2% 9 porphyritic K-fsp~1% Tlmgt <1% plagrare phonotephrite KT seriate-textured. amph18% subtree hytlc mugearite porphyritlc cpx10% plag4% TTmgt1-2% D K-fsprare ow n lo a benmoreite KT moderately porphyritic amph 4% fine-grained, subtrachytic to de d trachyte ftp 3% trachytic fro phonoIHe cpx <1% m h Tlmgt<1% ttp s ://a basanhe LGT moderately vesicular, ol9% fine-grained c a basalt porphyritic to seriate- cpx 8% de m textured ic .o u p shoshontte LGT porphyritic amph 6% fine-grained, subtrachytic .c o cpx 2% Im /p fsp~1% etro lo g Post-erostonal tephrite LGT moderately porphyritlc ~ 12% of cpx ± neph ± |—WO4^En20rS2S fine- to very fine-grained y/a Phase nephellnite gtt±Fe-Ti oxide rtic le -a Shield-building alkali basalt KT vesicular, porphyritic ol8% F074 fine-to very fine-grained bs Phase cpx 3% tra oct/3 7 basantta LGT slightly altered, ol5% fine- to very fine-grained, /4 /9 nephelinite porphyrttic cpx 15% partly glassy 61 /1 5 •After Le Bas of al. (1986). tKT, Karthala-type; LGT, La Grille-type (see later sections for a definition of these terms). tSiO2 = 29-46; TIO2 = 15-40; AJ2O3 = 2-10; FeO = 19-41; 08 MnO = 0-46; MgO = 0-86; CaO = 32-33. 48 2 b y g u e Os Ot o n 0 4 A p ril 2 0 1 9 JOURNAL OF PETROLOGY VOLUME 37 NUMBER 4 AUGUST 1996 Table 2: Selected clinopyroxene analyses from Comorean lavas Grande Comore Moheli Karl hall LaGrille Alkali bata It Basanite Nephallnfte GC-1 RH-7 RH-11 RH-15 RH-16 RH-33 RH-31 RH-31 SIOj 52-14 54-68 44-64 49-77 49-88 4703 44-39 48-60 T1O2 0 59 n.d. 3-13 1 -61 0-82 2-28 5-48 2-20 D o AljOj 2-29 0-27 9-60 4-46 3-89 6-26 7-80 4-89 w n Cr2O3 101 0-48 n.d. 0-42 0-09 n.d. n.d. n.d. loa d FeO 3-93 4-71 7-38 5-71 11-69 8-37 7-90 6-60 ed MnO n.d. n.d. n.d. n.d. 0-31 0-20 n.d n.d. fro m MgO 16-40 15-78 11-60 14-46 11-73 12-28 11-89 13-77 h ttp CaO 22-72 2308 23-29 22-72 20-52 23-10 23-96 24-19 s NbO 0-26 1-10 0-35 0-42 1-27 0-67 0-65 0-39 ://ac a d e m Total 99-36 100-10 99-99 99-57 100-19 100-18 10007 100-63 ic .o u p .c Wo 0-47 0-47 0-51 0-48 0-45 0-50 0-51 0-50 o m En 0-47 0-45 0-36 0-43 0-35 0-36 0-36 0-40 /p e Fs 006 008 0-13 009 0-20 0-14 0-13 0-10 tro lo g y Mayotta /artic le Shield-building Post-erotional Rejuvenescent -a b s tra KT LGT KT LGT ct/3 7 /4 MA-30 MA-60 MA-55 MA-8 MA-14a MA-59 MA-7 MA-33 /9 6 1 /1 5 SIO2 50-94 49-23 45-36 48-88 49-79 47-34 4807 47-71 08 4 TlOa 105 1-43 3-20 1-68 1-38 1-89 1-90 2-20 8 2 AI20j 2-76 4-20 7-30 5-71 4-88 6-14 5-78 5-45 by g Cr20, 0-52 n.d. n.d. nj. 0-30 n.d. n.d. 0-14 ue s FeO 6-05 8-17 12-71 8 65 7-09 10-31 8-45 6-36 t o n MnO 0-15 0-24 0-32 0-13 0-13 0-30 0-16 n.d. 0 4 MgO 15-61 13-35 8-72 12-86 14-46 10-63 12-60 13-61 A p CaO 21-87 22-36 2201 21-51 21-49 22-59 21 85 23-63 ril 2 0 Na2O 0-29 066 1-11 0-77 0-54 0-95 0-77 0-47 19 Total 99-23 99-64 100-28 100-10 100-07 100-15 99-58 99-56 Wo 0-45 0-47 0-60 0-47 0-45 0-50 0-47 0-60 En 0-45 0-39 0-26 0-39 0-43 0-32 0-38 0-40 Ft 0-10 0-14 0-24 0-14 0-12 0 18 0-16 0-10 n.d., not detected; KT, Karthala-type; LGT, La Grille-type. 966 SEATH et al. LAVAS FROM THE COMORES ARCHIPELAGO Table 3: Selected olivine analyses from Comorean lavas Grande Comoro Moheli Karthala La Grille Alkali basalt Batanrte Nephellnite GC-1 GC-1 RH-7 RH-9 RH-15 RH-16 RH-33 RH-36 RH-31 RH-31 SiO2 39-15 40-22 40-12 37-97 39-65 3605 39-77 39-72 38-44 40-03 AV>3 n.d. nd. 0-08 nd. n.d. n.d. 0-06 006 n.d. 0-08 Cr,O3 n.d. n.d. 0-09 nj. n.d. nd. n.d. n.d. n.d. n.d D FeO 17-84 13-81 11-31 23-49 17-37 33-06 12-08 10-73 20-22 12-95 o w MnO 0-23 0-25 0-13 0-33 0-23 0-61 0-16 0-13 0-49 0-15 n lo MgO 4207 44-82 47-39 38-19 42-77 30-43 47-24 48-30 40-17 46-72 a d CaO 0-44 0-38 0-21 0-15 0-41 0-34 0-24 0-21 0-23 0-25 ed NIO 0-26 022 0-35 nd. 020 n.d. 0-32 0-31 0-10 0-31 fro m h Total 99-99 99-70 99-67 100-12 100-64 100-49 99-87 99-47 100-43 100-50 ttp s ://a Fo 0-80 0-85 0-88 0-74 081 0-62 0-87 0 89 0-77 0 86 c a d e May otto m ic .o u Shield-building Rejuvenescent p .c o m KT LGT KT LGT /p e tro MA-29 MA-30 MA-60 MA-60 MA-1 MA-66 MA-7 MA-12 MA-33 lo g y /a SiO2 38-15 39-95 39-59 40-29 37-54 38-96 40-15 38-47 3901 40-33 rtic AljOj n.d. nxl. n.d. n.d. nA. nd. n.d. n.d. n.d n.d le Cr2O3 n.d. n.d. n.d. nji. n.d. nd. n.d n.d n.d n.d -ab s FaO 23-18 12-67 18-18 12-20 24-33 19-27 13-56 20-41 18-23 12-61 tra MnO 0-31 0-20 0-20 0-20 0-33 0-27 0-19 0-26 0-22 n.d ct/3 MgO 37-98 45-97 42-91 45-85 36-99 41 06 45-06 4003 4200 46-39 7 /4 CaO 0-31 0-28 0-25 0-31 0-20 0-27 0-23 0-16 0-26 0-20 /9 6 NiO 0-21 0 28 0-27 0-18 0-17 0-19 0-29 0-19 0-24 0-32 1 /1 5 0 Total 100-14 99-35 99-40 9903 99-55 10002 99-48 99-51 99-94 99-75 84 8 2 b Fo 0-74 0-86 082 0-87 0-73 0-79 0-85 0-77 0-80 0-87 y g u e n.d., not detected; KT, Karthala-type; LGT, La Grille-type. s t o n Grande Comore Moheli 04 A p The fresh, sparsely to coarsely olivine- and clino- The age relationships between the different volcanic ril 2 pyroxene-phyric lavas of Grande Comore are con- units on Moheli appear to be poorly understood 01 9 veniently divided into two groups according to (Strong, 19726; Emerick & Duncan, 1982; Nougier whether they were erupted from the active Karthala et al., 1986) and until the chronostratigraphy of the volcano in the south of the island, or from the older island is resolved by more detailed field and geo- La Grille volcano in the north (Strong, 1972a). chronological investigations, it appears imprudent to According to the geochemical classification scheme use age constraints as criteria for the classification of of Le Bas et al. (1986), Karthala lavas are alkali lavas from this island. The fresh to slightly altered basalts, whereas La Grille lavas are basanites. Small Moheli lavas examined in this study, which range in lherzolite xcnoliths occur in some of the La Grille texture from sparsely microporphyritic to coarsely lavas, as previously reported by Strong (1972a). porphyritic, containing phenocrysts of olivine and 967 JOURNAL OF PETROLOGY VOLUME 37 NUMBER 4 AUGUST 1996 Table 4: Selected feldspar analyses from Mayotte lavas; Rejuvenescent Phase Karthala-type LJ Grille-typa MA-3 MA-11 MA-14 MA-20 MA-65 MA-66 MA-17 MA-17 MA-34 MA-34 SK)2 53-77 61-64 62-38 66-81 6400 59-02 58-03 61-61 66-52 63-80 AW), 29-47 24-77 2397 19-87 23-14 2604 26-44 24-10 20-58 21-82 FeO 0-36 0-18 0-26 0-33 0-25 0-33 0-32 0-30 n.d. 0-28 MgO n.d. n.d nji. n.d. tuL n.d. 0-06 nJ. n.d. n.d CoO 12-76 6-83 4-93 0-23 3 62 7-94 8-98 5-89 1-79 3-62 D o w Ni20 3-86 6-67 7-17 5-17 6-80 6-44 6-89 7-09 7-61 7-75 n lo K2O 0-27 1-00 1-34 7-78 1-86 0-70 0-49 1-12 4-18 2-44 ad e d Total 100-47 99-99 100-05 100-19 99-67 99-47 100-21 100-00 100-57 99-60 fro m h An 0-64 0-31 0-25 001 0-20 0-43 0-44 0-29 0-09 0-17 ttps Ab 0-34 0 63 087 0-50 0-68 0-53 0 63 0-64 0-67 0-69 ://a c Or 0-02 0-06 0-08 0-49 0-12 0-04 0-03 0-07 0-24 0-14 ad e m n.d., not detected. ic .o u p .c Table 5: Selected amphibole analyses from Mayotte lavas; Rejuvenescent Phase om /p e tro Karthala-type La Grille-type lo g y /a MA-3 MA-6 MA-6 MA-18 MA-35 MA-36 MA-17 MA-17 MA-34 MA-34 rtic le SIOj 39-84 40-07 40-35 38-64 39-26 39-69 40-07 39-76 39-40 39-44 -ab s TiO2 605 4-66 6 81 5-13 6-26 6-97 5-81 6-08 5-40 6-56 tra c AljO, 13-71 13O7 13-54 13-10 13-72 13-71 13-29 13-21 13-62 13-84 t/3 7 FeO 11-66 16-39 11-18 17-45 11-38 11-63 11-35 14-35 12-60 12-83 /4 MnO 0-14 0-28 n.d. 0-33 n.d. 0-16 0-14 0-28 0-14 0-14 /96 1 MgO 12-49 10-08 12-87 8-94 12-65 12-40 1301 11-12 12-31 1202 /1 5 CaO 12-47 11-68 12-09 11-81 12-64 12-68 12-25 11-75 1208 11-91 08 4 Na2O 263 2 64 2-65 2-79 2-40 2-44 2-50 2-72 2-48 2-50 82 K20 0-81 1-13 103 1-33 1-21 1 28 103 MO 1-20 1-23 by g u e Total 99-79 100-00 99-42 99-41 99-50 99-95 99-44 99-37 99-14 99-47 st o mg-no. 0-66 0-52 0-67 0 48 0-66 0-65 0-67 058 0-64 0-62 n 0 4 A n.d., not detected. p ril 2 0 zoned clinopyroxenc, are classified as alkali basalts, fissure-erupted Post-erosional Phase (~3-75 to 2-5 1 9 basanites and nephelinites (Le Bas et al., 1986). Ma) and a Rejuvenescent Phase (~2-5 to 149 Ma). The lavas from Mayotte span a significantly greater compositional range than those from Grande Mayotte Comorc and Moheli, and may be classified into Mayotte, the oldest island in the Comores Archi- basalts, basanites, tephrites, nephelinites, trachy- pelago, was constructed during three main phases of basalts, basaltic trachyandesites, benmoreites, volcanic activity (Emerick & Duncan, 1982, 1983; phonotephrites, phonolites and trachytes (Le Bas et Nougier et al., 1986): a Shield-building Phase (pos- al., 1986). Olivinc and clinopyroxene are the sibly from as early as 15 Ma to ~3-75 Ma), a dominant phenocryst phases in the mafic to inter- 968 SPATH tt aL LAVAS FROM THE COMORES ARCHIPELAGO mediate lavas, but phenocrysts of amphibole, are richer in Ni and Cr than Karthala lavas, but feldspar, titanomagnetite and in one sample there is no significant difference in the Sc content of nepheline and garnet become increasingly important the lavas from the two volcanoes (Fig. 3). La Grille in the more differentiated rocks. lavas are furthermore enriched in Nb, Th, Sr and Ba relative to Karthala lavas and have significantly steeper REE patterns (La/Yb= 15-9-26) than BULK-ROCK CHEMISTRY s Karthala lavas (La/Yb = 10-6-10-9), with the B The bulk-rock major, minor and trace element com- former cross-cutting the latter (Fig. 4). positions of selected lavas from Grande Comore, Moheli and Mayotte are listed in Table 6, and the Moheli REE compositions of a subset of samples are pre- sented in Table 7. The majority of the lavas analysed Well-defined linear trends are uncommon in Moheli D o major and minor element variation diagrams, but w were found to have loss-on-ignition (LOI) values n with decreasing m^-number, a general increase in lo below 3 wt % and are considered to range from fresh a to slightly altered in composition—an interpretation A12O3 and FeO is observed (Fig. 2). Considerable ded consistent with petrographic observations. The com- ranges in the concentrations of the compatible trace fro elements are observed amongst Moheli lavas, with m pfilrestt ea suetth oofr bounl kr-erqouceks ta.nalyses are available from the well-defined trends of decreasing Ni, Cr and Sc with http decreasing m^-number (Figs 3 and 9). In contrast, s Major, minor and trace elements broad positive correlations exist between incompa- ://ac a tible trace element abundances (Fig. 3). Moheli d The compositional ranges of Grande Comore and e nephelinites tend to be more strongly enriched in the m Moheli lavas are compared with individual Mayotte ic light and middle REE, but they have chondrite-nor- .o data in Fig. 2, which shows several major and minor u malized heavy REE abundances similar to or lower p elements plotted against m^-number. Trace element .c than those of the basanites, which in turn show o variations in Comorean lavas are illustrated in Fig. m greater overall REE enrichment than the alkali /p 3. For clarity and to facilitate comparison between e itnhde ivtihdrueea l isdlaantda s, pcooimntpso, siatiorena ls hfoiewldns , froarth eGr rathnadne bdaristael-tns o(rFmiga.l iz4e).d ThRe EneEph pealtitneirtenss ha(vLea /sYtebeBp =e r2 c8h-5o2n)- trology Comore and Moheli in Figs 2 and 3. Chondrite-nor- hthaavne tshtee ebpaesra npitaetst e(rLnsa /YthbaBn = t1h6e-8 a-l3k9a)l,i wbhaiscahl tisn ( Ltuarn/ /artic malized REE abundances are shown in Fig. 4, and le Fig. 5 depicts selected primitive mantle-normalized Yb,,= 12-4-22). The nephelinite REE patterns cross- -ab cut those of the basanites and alkali basalts in the s incompatible element plots (spidergrams). middle to heavy REE range. All of the alkali basalts tra c from Moheli (K/Nb = 175-229) were found to be t/3 Grande Comore 7 KT lavas, whereas all of the basanites and nepheli- /4 /9 The most notable difference between the lavas from nites (K/Nb = 66-150) display the distinct K 6 1 the two volcanoes on Grande Comore lies in their depletion which characterizes LGT lavas (Fig. 5). /1 5 K2O and MgO contents and m^-numbers: La Grille 08 4 lavas have significantly higher MgO concentrations 8 2 and m£-numbers (0-69—0-71) than Karthala lavas Mayotte b y (m£-numbers 0-52—0-54; Table 6), and significantly Mayotte lavas show a considerable range in mg- gu e lower K contents relative to other highly incom- number (0-15-0-64) and relatively well-defined s patible elements when normalized to primitive linear trends arc observed on a number of the vari- t on mantle (Fig. 5). The relative K depletion of La ation diagrams in Fig. 2. Good negative correlations 04 rGarnigllien gl afvroams 7i0s tmo a1n1i6fe, satse dc ominp alroedw w Kith/ Nab r avnagleu eosf, wexhisetr ebaest wpeoesnit imve^ -ncourmrebleart iaonnds SairOe2 o, bAsJe2rOve3d a nbde tKw2eOen, April 2 205-220 for Karthala lavas. This distinction between m£-number and CaO, FeO and to a lesser extent 01 9 lavas that show a relative K depletion and those that TiOz- The P2O5 content of Mayotte lavas generally do not is not restricted to the island of Grande decreases as the m^-number decreases below 0-5, Comore, but is also evident in the lavas of Moheli which is interpreted to reflect the effect of apatite and Mayotte, and it therefore provides a first-order fractionation. Compatible trace element abundances classification criterion for Comorean lavas as a in Mayotte lavas show well-defined trends of whole: lavas with a relative depletion in K will be decreasing concentration with decreasing m^-number referred to as La Grille-type (LGT) lavas, whereas (Figs 3 and 9). The lavas also show considerable lavas that do not show a distinct K depletion are variability in incompatible trace element abund- termed Karthala-type (KT) lavas. La Grille lavas ances, and a number of relatively well-defined trends 969 Table 6: Representative bulk-rock analyses and selected trace element ratios of lavas from Grande Comore, Moheli and Mayotte D o w Grande Comore Moheli n lo a Karthala La Grille Alkali basalts (KT) Basanhos (LGT) NepheUnites (LGT) de d GC-1 GC-3 GC-5 RH-1b RH-7 RH-9 RH-11 RH-15 RH-16 RH-17 RH-28 RH-12a RH-21 RH-26 RH-29 RH-36 RH-39 RH-14 RH-18 RH-31 t from AB AB Ha AB Bat Bas Bas O h cttp s SIO2 46-47 46-60 46-80 46-91 42-81 42-38 41-48 44-97 47-26 45-06 43-46 39-05 42-29 40-81 41-22 42-17 47-38 39-50 38-76 37-83 ://a TIOj 2-67 2-69 2-65 2-49 1-92 2-34 2-17 2-38 2-50 2-60 2-88 2-64 2-47 4-20 3-53 2-32 3-33 3-25 2-91 2-90 >c a AljO, 14-38 14-96 14-63 14-29 12-96 12-45 1208 10-34 13-70 10-92 1300 11-55 11-22 11-33 14-16 11-67 12-10 11-15 10-64 10-29 de 1-58 1-59 1-69 1-58 1-38 1-66 1-54 1-69 1-64 1-63 1-74 1-81 1-61 1-69 1-88 1-62 1-64 1-70 1-84 1-82 Om FeO 10-51 10-58 10-59 10-55 9-21 10-40 10-25 11-24 10-94 10-88 11-63 1209 10-71 11-24 12-54 10-82 10-92 11-30 12-28 12-15 ic.o MnO 0-19 0-18 0-18 0-19 0-18 0-19 0-20 0-18 0-19 0-18 0-20 0-27 0-20 0-19 0-24 0-21 0-19 0-22 0-23 0-22 Sup MgO 6-48 6-72 6-81 6-88 12-44 13-39 12-78 12-59 7-80 11-63 8-41 9-72 11-69 1008 507 11-31 9-84 11-32 10-75 10-92 .c o CaO 10-81 10-25 10-64 11-09 10-12 11-43 12-20 10-21 9-40 11-70 9-65 11-80 12-63 1207 988 1204 10-78 12-66 13-01 14-00 0m No^O 3-39 3-47 3-78 3-47 3-49 2-43 4-34 306 3-08 2-68 2-59 407 2-69 4-74 6-74 4-36 4-76 4-41 4-83 3-49 f0/pe K20 1-34 1-38 1-28 1-25 0-84 0-60 1-29 0-88 1-38 1-00 1-66 107 0-61 0-66 1-14 0-76 0-68 1-25 1-80 1-46 0tro P*OS 0-48 0-50 0-47 0-42 0-44 0-55 0-78 0-43 0-59 0-43 0-75 109 0-89 0-72 1-11 0-72 0 63 0-88 1-27 1-24 >log< LOI 1-64 0-84 0-12 0-82 3-42 1-40 0-89 107 0-97 1-54 2-36 3-61 2-56 2-41 2-88 208 2-21 2-23 1-66 2-48 y/a THo2Ota"l 1000--5447 909--1985 1000--6282 1000--2281 1000--9176 909--2303 1000--1132 909--7715 909--4816 1000--4750 909--8181 909--9762 1000--6163 1001--9174 1000--9365 1000--3496 1001-0573 1000--6555 1000--2120 909--3100 0rticle -a §b Zr 223 226 224 208 223 202 226 174 228 200 278 359 229 192 430 211 214 237 359 299 s Nb 52 56 52 47 87 71 92 34 50 38 66 124 68 65 112 79 64 88 127 103 totrac Y 29 30 29 27 24 24 29 24 27 24 29 42 29 26 42 30 27 32 36 32 t/3 Rb 31 32 31 29 30 8-1 46 17 32 23 40 45 15-5 117 72 20 128 20 48 50 7/4 Ba 363 405 374 324 709 634 623 369 416 306 550 1003 639 800 844 693 679 806 802 657 5/95 Sr 642 603 557 495 659 628 764 446 634 487 731 1404 909 710 1225 754 761 792 1527 1367 s61 Th 6-2 5-2 5-6 49 92 8-6 9-9 2-3 3-5 5-3 6-3 150 9-4 7-1 11-0 9-4 9-1 9-8 14-7 11-8 w/15 Co 52 50 53 53 64 72 68 76 65 69 61 59 63 68 46 67 67 71 65 66 08 Cr 143 142 145 166 535 613 603 644 263 612 268 221 601 454 20 606 387 392 356 403 48 2 Nl 117 122 115 121 327 373 338 403 184 375 219 201 343 254 34 338 269 301 260 267 b V 260 249 253 248 208 246 238 240 181 240 207 208 242 290 219 252 249 257 224 216 >y g Zn 237 110 107 111 89 92 93 107 146 108 127 158 107 96 150 102 104 109 156 141 0ue SCcu 22760 2615 2944 2887 6262 2676 7288 2980 2908 8208 6178-8 5167-9 2747 2727 5183-8 2889 2711 8213 2627 7119-7 cHst on La 41 46 — 39 — — — — — 31 — — — — — 61 53 — — — 0 4 Ce 81 90 — 79 — — — — — 64 — — — — — 110 99 — — — $ A ZNrd/Nb 414-3 4240 —4-3 394-4 —2-6 —2-9 —2-5 —5-1 —4-6 326-3 —4-2 —2-9 —3-4 —30 —3-8 492-7 453-3 —2-7 —2-8 —2-9 pril 2 0 Zr/Y 7-7 7-5 7-7 7-7 9-3 8-4 7-8 7-3 8-4 8-3 9-6 8-6 7-9 7-4 10-2 70 7-9 7-4 100 9-3 1 9 Y/Nb 0-56 0-54 0-58 0-57 0-28 0-34 0-32 0-71 0-54 0-63 0-44 0-34 0-43 0-40 0-38 0-38 0-42 0-36 0-28 0-31 Ba/Nb 70 7-2 72 6-9 8-2 7-5 6-8 10-9 83 8-1 8-3 8-1 9-4 12-3 7-5 8-8 10-6 9-2 6-3 6-4 mg-no. 0-52 0-53 0-63 0-54 0-71 0-70 0-69 0-67 0-56 066 0-56 0-59 0-66 0-62 0-42 0-65 0-62 0-64 0-61 0-62