sign in sign up
<<
Home , Carbonatite

A merican M i neralo gist, Volume73,pages 1468-1471. 1988

Partial melting of fenitized crustal xenoliths in the Oldoinyo Lengai carbonatitic ., T rnzania: Reply

Rosrnr F. MlnrrN Department of Geological Sciences,McGill University, 3450 University Street,Montreal, Quebec}{3A 247, Canada Vronrc.q.Monoc.qs GeologiskaInstitutionen, Stockholms Universitet, Box 6801,S-113 86 Stockholm,Sweden

Ansrn-q,cr There is perhaps no more exotic an igrreousrock type than natrocarbonatite, erupted from the volcano Oldoinyo Lengai, in northern . Its origin is still the object of debate. Gittins (1988) has proposed that the alkalic carbonatite is a relatively evolved derivative of a mantle-derived -bearing sdvite. We agree that it is geo- chemically an evolved magma, as are the associatedsilicate ; however, the bulk of the evidencesuggests to us that the natrocarbonatitic magma has a crustal component and may well have formed within the crust. In this essay,we review the backgroundinformation pertinent to this proposal, made in our paper on xenoliths of fenitized crustal material brought up in the natrocarbonatite (Morogan and Martin, 1985).

INrnooucrrox "metasomatic ," charucrerizedby high-grade, me- The investigation of Morogan and Martin (1985) was dium-grade, and low-grade fenites, could be measuredin motivated by a natural curiosity concerning the assem- centimetersor meters,as might be expectednear the out- blage of xenolithic materials of crustal origin brought to er, near-vertical contact of an intrusive alkaline complex the Earth's surfacein the highly unusual natrocarbonatite emplaced in cold host-rocks near the Earth's surface,or lava ofOldoinyo Lengai volcano, located in the southern on either side of a crack along which the fenitizing fluids part of the Gregory , in northern Tanzania. The focus are percolating.But the metasomaticinteraction was con- of our investigation was a suite of nine hand specimens sidered to be developed on a broader scaleby Morogan offenitized rocks, and not the host natrocarbonatitelava. and Martin (1985), in part becauseof the anomalous We realized at the outset that as in any investigation of thermal gradient found in this area (cf. the "alkaline a suite of metasomatic rocks in which the unmetasoma- province geotherm"of Joneset al., 1983). tized starting material is unexposedand unsampled,ours "That fenitization might occur on a regional scale . . . was perforce an open:ended survey of the phase assem- is a new idea," saysGittins (1988, p. 1465).We were blagesdeveloped prior to and during entrainment in the attracted to it becauseofthe unusual regional geophysical unusual lava. Clearly, therefore, we did not set out to characteristicsof the upper mantle and lower crust along addressthe evidently controversial question of the ulti- various parts of the . An elliptical area mate origin of the Oldoinyo Lengai natrocarbonatite. In (the Kenya dome) approximately 900 km along its short our Discussion section, however, and specifically in the axis dominates the easternrim of the East African pla- last two paragraphsofthe paper, we did proposea petro- teau; the dome results from a maximum uplift of 1.5 to geneticscheme that could explain, at leastin a qualitative 2 km since mid-Tertiary time (Baker and Wohlenberg, way, the variety ofigneous products in this sector ofthe l97l; Savageand Long, 1985).The GregoryRift, a 70- rift system,as describedby others, and the metasomatic km-wide zone of normal faults, bisectsthe domal feature phenomena that we had just documented. We welcome along its crest.Regional gravity and seismicsurveys show this unexpected opportunity to clarify our views and that the dome reflects the presenceof a broad elliptical evaluatethe backgroundinformation that led to our nov- area of anomalous upper mantle. The mantle below the el proposal that the natrocarbonatite could have origi- Gregory Rift has an anomalously low specificgravity of nated in the suite of metasomatizedcrustal rocks rather 3.2 and is characterizedby attenuatedP- and S-waveve- than deep in the upper mantle. locities. At the normal depth of the Moho, a 30-km-wide bulge of anomalous upper mantle crossesinto the lower Sclr,e oF THE METASOMATTCTRANSFoRMATIoNS crust, and narrows upward (Savageand Long, I 985). This There is obviously nothing recorded in the nine hand lower crust, characterizedby a P-wave velocity of 7.3 specimensof fenitized crustal rocks to justify a firm state- km/s, presumably represents a mixture of shield-type ment one way or the other concerning the scale of the lower-continental crust, as characterizedoff the rift axis zones of metasomatic assemblages.The width of the (6.5 km/s), and alkaline igneousbasic and ultrabasic ma- 0003-004x/88/I l l 2-1468$02.00 1468 MARTIN AND MOROGAN: ORIGIN OF ALKALIC CARBONATITE LAVA-REPLY r469 terial crystallized from injected from the upper into the crust, which is in a state of tension becauseof mantle (Maguire and Long, 1976; Mooney et al., 1983; regionaldoming, or very efficiently,as direct "emanations Savageand Long, 1985).Mooney et al. (1983)considered from below." There remain problems concerning the the development of a lower crust modified in this way to mechanismsof migration of a discrete volatile phase in be a common feature of continental . The proportion the mantle, but the evidence for regional of intruded material probably decreasesupward; the rel- in the mantle beneath continental rifts, in our opinion, is atively high density inferred for the upper halfofthe crust compelling. (3.05 g/cm3)is consistentwith the presenceof igneous ..EMANATIoNS bodies there as well and with a concomitant alteration of N,trunn oF THE FRoM BELowtt the crust. The area is consideredgenerally to be in iso- Recent contributions on the topic of mantle metaso- static equilibrium, and domal uplift is viewed as the re- matism make a clear casefor the introduction of a suite sponseto the presenceof a regional low-density mass in of incompatible elements, including the alkalis, but are the upper mantle. less categorical concerning the respective roles of HrO Heat lossalong the GregoryRift (l l-30 MWkm: Crane and COr, which are likely to be the two dominant species and O'Connell, 1983) is focusedthrough the central part in the relatively oxidized continental lithosphere (Eggler, of the Kenya dome, but geothermal manifestations do 1987). The two speciesare likely to play diferent roles, occur on a regional scale.The heat flow is consideredan HrO effective in complexing most major and trace ele- expressionof focusedmantle degassing(Bailey, 1983)and ments in the solute, and CO' more effectivein mobilizing the unusual intensity of injections of mantle-derived the alkalis and the rare earths.The relevant experimental magmas into the crust. In our opinion, fenitization of dat'aare still sparse.In the opinion of Bailey ( I 980, I 983), crustal rocks could be effectedby both agents,i.e., (l) a the important volatile constituents to escapefrom the fluid medium releasedlocally along an intrusive contact mantle into continental rift systemsare CO2,HrO, F, and as a result of crystallization of an alkaline magma derived cl. from the anomalous mantle below and (2) a fluid phase Joneset at. (1983) have documentedmetasomatic re- releasedfrom a leaky anomalous mantle. Is the shortest actions involving CO, in the lower crust in the Gregory distancebetween two intrusive contactsin this zone mea- Rift: relicts of olivine-normative alkali gabbrosare meta- sured in terms of meters or in kilometers? We believe morphosedinto granulitesthat contain sulfate- and that there is such an intense veining of the crust that a -rich meionitic scapolite.These xenoliths were proposal of a regionally affected,metasomatically modi- collected from the Lashaine tuff cone in northern Tan- fied crust below the Oldoinyo Lengai area is not to be zania.The development of scapoliteis consideredmainly consideredfar-fetched. a result of CO, metasomatism at an unusually high re- What is going on geochemicallyin the anomalous up- gionally developedtemperature (1200 K) at a pressureof per mantle under the Kenya dome obviously is relevant 14 kbar, but HrO is likely also involved, as in the recon- to the question ofthe nature ofthe fluid leaking into the structive transformation of any (Donnay et al., crust above. As in the caseofour knowledgeconcerning 1959).Metasomatism involving K (and HrO?) seemsto the crust, any insight into the state of the upper mantle have occurred later, near the surface,after the incipient is based on meager samples brought to the surface in decompression-relatedmelting recordedin the xenoliths. eruptions of alkali and and on geo- In a companion study, an important 2.0 Ga event of physical modeling. Textural and mineralogical studies of chemical fractionation involving the U-Pb, Sm-Nd, and ultramafic xenoliths from this and other areas along the Rb-Sr systemswas discovered in garnet lherzolite xen- East African Rift have revealed abundant evidence for oliths from Lashaine by Cohen et al. (1984). The pref- regionalmetasomatic enrichment ofthe upper mantle via erential mobilization of U, Pb, Rb, and Sr probably re- a mobile fluid, considereddominantly COr-bearing(Lloyd flects the influence of HrO more than COr. and Bailey,1975; Bailey, 1980, 1982; Cohen et a1.,1984; Morogan and Martin (1985) did not investigatethe na- Dawson, 1987;Lloyd, 1987).There is clearevidence for ture of trapped fluids in the metasomaticsuite and so did regional metasomatism as long as 2 Ga ago in this area, not have direct information on the nature of the meta- as well as an episodejust prior to the generation of the somatizingfluid. In view of the major reconstructivesteps host basalt. The metasomatic reactions have consumed involved in the transformation of the pre-existingplagio- olivine and produced clinopyroxene and new accessory clase to , and of amphibole to alkali-rich cli- phases.They have effectively added alkalis and a variety nopyroxene,and the formation of disordered alkali feld- of incompatible elementsto "normal" and, in some cases, spars, we inferred that the fluid phase was aqueous and depleted mantle. The increase in volume that reflects peralkaline. In view of the presenceof accessorycalcite metasomatismand ensuingpartial melting is responsible in most samples of the fenitized rocks, and the experi- for the regional uplift and the anomalies in heat flow. mental data of Cermignani and Anderson (1983) showing Morogan and Martin (1985) tacitly assumedthat the fluid that the nephelinization of plagioclaseis more likely in phase that has been focused into the anomalous zone concentrated carbonate than chloride solutions, we in- from deeperin the mantle eventually must move on into ferredthalthe metasomatizingfluid was COr-bearing.We the crust, either via the batches of partial melt that rise also showed that in some cases,the metasomatic reac- 1470 MARTIN AND MOROGAN: ORIGIN OF ALKALIC CARBONATITE LAVA-REPLY

tions had completely transformed the pre-existing min- area. Additional problems with contamination by addi- eral assemblagesby the time incipient melting began.As tion of trona werementioned by Donaldsonet al. (1987). in the study of Joneset al. (1983), the melting was attrib- The approachof Williams et al. (1986),based on Ra-Th uted to entrainment in a rising batch of magma. disequilibria, indicates that the carbonatemagma formed and eruptedwithin 7 to l8 1r and that it probably exsolved Evronucn FoR A cRUSTALcoMpoNENT IN THE from nephelinite that was itself Ra-enriched. If the as- NATROCARBONATITE sumption of Williams et al. concerning the geochemical propose Isotopic systemsconstitute the most sensitive tool to coherencebetween Ra and Ba holds, one could (and, evaluatethe question of ultimate derivation of the natro- that enrichment of nephelinite in Ra by extension, 8tSr carbonatite and associatedrocks. The Oldoinyo Lengai and Ba) reflects the incorporation of a crustal com- natrocarbonatite is so unusual that it has been investi- ponent, possibly variably fenitized and thus difficult to process. gatedby a number of sophisticatedtechniques. Bell et al. characterizeas an "end member" in a mixing (1973)found an unusuallylarge spread ofinitial 875r/865r CoNcr-unrNc REMARKS values(between 0.7034 and 0.7096;er. between - 18 and +70) in the various products of igneous activity at Ol- The ultimate origin of alkali carbonatite magma evi- doinyo Lengai. Values at the low end ofthis range refer dently remains one of the controversial topics of igneous (1987) to xenoliths of plutonic aspect (melteigite, ), but petrology. Twyman and Gittins cogently argued other samples of similar aspect have a value closer to in favor of an origin by fractional crystallization of a rel- 0.707. Two values are reported for modern carbonatite atively dry, mildly alkalic olivine sovite magma, which flow rocks: 0.7059 and 0.7061. Finally, and in turn probably separatedas an immiscible melt from nephelinite are typically higher than 0.705 (maximum olivine nephelinite magma "deep within the mantle," at pro- 0.7096), although one sample of nephelinite has a value a confining pressureofapproximately 27 kbar. The of 0.7034.The suiteis anomalousin its scatterof values: posed site of immiscible separationis thus in the anom- compared to carbonatiteselsewhere, which are more de- alous, geochemically"fertilized" upper mantle. The car- pleted in 87Srthan the bulk Earth at the time of formation bonatitic melt must rise 40 km through anomalousupper (e.g.,Andersen, 1987, Fig. 9), the Oldoinyo Lengainatro- mantle to reach the Moho, further through 25 km of carbonatite clearly is anomalous. anomalous, altered lower crust, then through 15 km of The e*ovalue of +0.1 + 0.9 determinedfor the natro- fractured upper crust before making it to the surface. carbonatite (DePaolo and Wasserburg, 1976) is similar During this trajectory, the stivitic magma is said to frac- to that of silicic rocks in many continental massifs. In tionate anhydrous and alkali-poor , which sink Iight of the enrichment in 87Sr,an e*ovalue close to zero efficiently, making the derivative carbonatite magma al- may simply imply that the Nd-Sm system remained vir- kaline and, where the fugacity of water is low, strongly tually unaffectedin the crustal reservoir. The single de- peralkaline and enriched in the halogens.However, Twy- termination is inconsistent with the signature expected man and Gittins (1987)and Gittins (1988)have made no for the geochemicallyenriched upper mantle that under- referenceto the likelihood that such a melt will interact lies the area. An integrated investigation of the isotopic with (and become contaminated by) its surroundings at composition of Sr, Nd, and Pb will be required to assess any stage.In view ofthe reactivity ofa natrocarbonatitic properly the respective roles of crust and upper mantle melt in contact with silicate materials that are not equil- in the volcanic products of Oldoinyo Lengai and the stage ibrated with it, closed-systemfractional crystallization is (prior to melting, during or after crystallization?)at which considereda highly unlikely possibility. We do agreewith interaction with crustal materials occurred. Twyman and Gittins that alkalic carbonatite magmasare Petersonand Marsh (1986)and Donaldsonet al. (1987) highly evolved, and not to be construed as primary mag- found it impossible to explain the interrelationships mas. However, we consider the bulk of the evidence, point, among silicate members of the Oldoinyo Lengai suite us- which is admittedly largely circumstantial at this ing closed-systemcrystal-liquid processes.Peterson and to favor an origin that involves a component of meta- Marsh (1986) interpreted the presenceof combeite somatically modified crust. Like King (1965), we contend (NarCarSi.On)phenocrysts and pseudomorphsof wollas- that the rheomorphism of fenites can account for the more tonite phenocrystsin the (which are unusu- evolved members of the magmatic associationsalong the ally peralkaline and highly evolved: Donaldson et al., East African Rift. 1987) as a product of carbonate addition to the AcrNowr,nocMENTS nephelinitic magma. They proposed that the natrocar- bonatite exsolved from carbonated nephelinite in the We thank R. J. Arculus for his perceptivecomments on an early version of this reply. courseof its crystallization and that troniferous sedimen- tary rocks are the probable crustal contaminant; however, RprnnnNcps crrBt the23oTW232Thvalue (Williams et al., 1986),like the 6'80 (l in a carbonatitecom- values (O'Neil and Hay, 1973), precludes possibility Andersen,T. 987) Mantle and crustalcomponents the plex, and the evolution ofcarbonatite magma: REE and isotopic evi- that the carbonatite is simply a remobilized trona-rich dence from the , southeastNorway. Chemical Geology evaporite deposit, such as are found in lake basins in the (IsotopeGeoscience Section), 65, 147-166. MARTIN AND MOROGAN: ORIGIN OF ALKALIC CARBONATITE LAVA_REPLY 147|

Bailey,D.K. (1980)Volcanism: Earth degassingand replenishedlitho- morphism, , and K-metasomatism of gamet-scapolite- spheremantle. PhilosophicalTransactions ofthe Royal SocietyofLon- kyanite granulite xenoliths from Lashaine, Tanzania Journal of Ge- don, 4291,309-322. ology,9 1, 143-165. - (1982)Mantle metasomatism-Continuingchemical change with- King, B C ( I 965) Petrogenesrsof the alkaline igneousrock suites of the in the Earth Nature,296, 525-530 volcanic and intrusive centresofeastern Uganda.Journal ofPetrology, -(1983) The chemicaland thermalevolution of rifts.Tectonophys- 6, 67-100. ics.94. 585-597. Lloyd, F.E. (1987)Characterization ofmantle metasomaticfluids in spi- Baker,B.H, and Wohlenberg,J. (1971)Structure and evolution ofthe nel lherzolitesand alkali clinopyroxenitesfrom the West Eifel and South Kenyarift valley.Nature, 229,538-542. West Uganda In M.A. Menzies and C.J Hawkesworth, Eds., Mantle Bell, K., Dawson,JB., and Farquhar,RM (1973) Strontium isotope metasomatism,p. 9l-123 Academic Press,London. studiesofalkalic rocks: The active carbonatitevolcano Oldoinyo Len- Lloyd, F.E, and Bailey,D K (1975)Light-element metasomatism of the gai,Tanzania. Geological Society ofAmerica Bulletin,84, 1019-1030 continental mantle: The evidence and the consequences.Physics and Cermignani,C., and Anderson,G M (1983) The plagioclaseexchange Chemistryof the Earth,9,389416 reaction in carbonate solutions and application to nephelinization. Maguire, P.K.H., and Long, R E. (1976)The structureof the westernflank American Journal of Science,283A, 314-327. of the Gregory Rift (Kenya). I. The crust. GeophysicalJournal of the Cohen,R.S., O'Nions, R K, and Dawson,J.B (1984)Isotope geochem- Royal AstronomicalSociety, 44, 661-675 istry of xenoliths from East Africa: Implications for development of Mooney,W D., Andrews,M.C, Ginzburg,A., Peters,D A., and Hamil- mantle reservoirs and their interaction. Eanh and Planetary Science ton. R M (1983)Crustal structure ofthe northernMississippi embay- ktters, 68, 209-220 ment and a comparison with other continental nft zones. Tectono- Crane,K., and O'Connell,S. (1983)The distributionand implicationsof physics,94,327-348. heal flow from the Gregory Rift in Kenya. Tectonophysics,94,253- Morogan, V., and Martin, R.F (1985) Mineralogy and partial melting of 275 fenitized crustal xenoliths in the Oldoinyo Lengai carbonatiticvolcano, Dawson,J.B. (1987)Metasomatized harzburgites in kimberlite and al- T anzania.American Mineralogisr, 70, | | | 4-l 126. kaline magmas: Enriched restites and "flushed" lherzolites. In M.A. O'Neil, J R., and Hay, R L. (1973)"O/'60 ratiosin chertsassociated with Menziesand C.J. Hawkesworth,Eds, Mantle metasomatism,p. 125- the saline lake deposits of East Africa. Earth and Planetary Science 144.Academic Press, London. Letters,19, 251-266 DePaolo,D.J, and Wasserburg,G.J (1976) Inferencesabout magma Peterson,T.D., and Marsh,B D. (1986)Sodium metasomatism and min- sourcesand mantle structurefrom variations of r4rNd/r44Nd.Geophys- eral stabilities in alkaline ultramafic rocks: Implications for the origin ical ResearchLetters. 3. 743-746. ofthe sodiclavas ofOldoinyo L'engai.EOS, 67, 389-390. Donaldson,C H , Dawson,J.B., Kanaris-Sotiriou,R., Batchelor,R A , Savage,J.E.G., and Long,R.E (1985)Lithospheric structure beneath the and Walsh,J.N. (1987)The silicatelavas of Oldoinyo Lengai,Tanza- Kenya Dome GeophysicalJoumal of the Royal Astronomical Society, nia. NeuesJahrbuch fiir MineralogieAbhandlungen, 156,247-279. 82.461477. Donnay,G., Wyart, J., and Sabatier,G. (1959)Structural mechanism of Twyman, J.D., and Gittins, J (1987)Alkalic carbonatilemagmas: Paren- thermal and compositional transformations in Zeitschrift fiir tal or derivative?Geological Society ofLondon SpecialPublication 30, IGistallographie,I 12, l6 l-168. 85-94. Eggler, D H. (1987) Solubility of major and trace elements in mantle Williams.R W , Gill, J.B.,and Bruland,K.W. (1986)Ra-Th disequilibria metasomatic fluids: Experimental constraints. In M.A. Menzies and systematics:Timescale of carbonatite magma formation at Oldoinyo C.J. Hawkesworth,Eds., Mantle metasomatism,p 2141. Academic Lengai volcano, Tanzania Ceochimica et Cosmochimica Acta, 50, Press.London. t249-1259 Gittins, J. (1988)Partial melting offenetized crustal xenoliths in the Ol- doinyo Lengai carbonatitic volcano, Tanzania: Discussion American Mineralogist, 73, 1465-1467. MaNuscnrpr REcETvEDJeNuenv 15, 1988 Jones,A.P, Smith,JV., Dawson,J.B., and Hansen,EC (1983)Meta- M.q.NUscnrPTACcEPTED Jurv l, 1988