Sr–Nd–Pb isotope systematics and clinopyroxene-host disequilibrium in ultra-potassic magmas from Toro-Ankole and Virunga, East-African Rift: Implications for magma mixing and source

Lithos 210–211 (2014) 260–277

Contents lists available at ScienceDirect

Lithos

j ourna l homepage: www.e lsev ie r .com/ locate / l i thos

Sr–Nd–Pb isotope systematics and clinopyroxene-host disequilibrium inultra-potassic magmas from Toro-Ankole and Virunga, East-African Rift:Implications for magma mixing and source heterogeneity

N.S. Muravyeva a,⁎, B.V. Belyatsky b, V.G. Senin a, A.V. Ivanov c

a Vernadsky Institute of Geochemistry and Analytical Chemistry RAS, Kosygin st. 19, Moscow, 119991 Russiab VNIIOkeangeologia, Antarctic Geology Department, St. Petersburg, Russiac Institute of the Earth's Crust RAS, Siberian Branch, Lermontova St., 128, Irkutsk, 664033, Russia

⁎ Corresponding author. Tel.: +7 499 9397027; fax: +E-mail address: [email protected] (N.S. Muravyeva).

http://dx.doi.org/10.1016/j.lithos.2014.09.0110024-4937/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t
a r t i c l e i n f o
Article history:Received 11 March 2014Accepted 13 September 2014Available online 4 November 2014

Keywords:Isotope compositionMantle source heterogeneityEast African RiftClinopyroxeneMagma mixing

Nd, Pb and Sr isotope ratios have been determined for kamafugite lava and clinopyroxene phenocrysts fromBunyaruguru (Toro-Ankole) and Virunga volcanic fields of the East African Rift. The whole rock Sr–Nd isotopic sig-natures of kamafugites (87Sr/86Sr: 0.70463–0.70536; 143Nd/144Nd: 0.51249–0.51255) suggest derivation from anEM1-type mantle source. In contrast, Pb isotopic compositions of the same samples (206Pb/204Pb: 19.00–19.57;207Pb/204Pb: 15.69–15.74; 208Pb/204Pb: 39.30–40.26) reveal a similarity to EM2-typemantle. NewNd, Pb and Sr iso-topic data for clinopyroxene (87Sr/86Sr: 0.70473–0.70503; 143Nd/144Nd: 0.51250–0.51254; 206Pb/204Pb: 18.04–18.17; 207Pb/204Pb: 15.58–15.60; 208Pb/204Pb: 38.09–38.23) suggest derivation from an EM1-like source, and indi-cate Sr and Pb isotope disequilibrium between clinopyroxene and corresponding host rock. Moreover,clinopyroxenes exhibiting a greater degree of isotopic disequilibriumwith their host rock aremore sodic in compo-sition. The isotopic disequilibrium is corroborated by the presence of chemical zoning within clinopyroxene, whichsuggests rapidmagmaascent rates preventingmelt homogenization. The Pb isotopic ratios for bothmineral and cor-responding whole rock, together with published data on East African rift-related alkaline centers, define a trendinterpreted to represent amixing line formelts derived from sources such as EM1 and as HIMU. The similar isotopiccompositions for clinopyroxene from the different volcanic rockswithin the East African Rift suggest the existence ofa common, oldermantle source for their parentalmelts. The origin of thesemelts can be attributed to an enrichmentevent ~400–500 Ma, i.e., significantly prior the younger ultrapotassic magmatism. Our preferred interpretation forthe results reported here involves the mixing of melts derived from EM1- and HIMU-like sources, which were rap-idly transported to the Earth's surface. The primarymagmas formed as the result of melting of a heterogeneous (onkilometer scale) mantle source consisting of peridotite and pyroxenite.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Clinopyroxene and olivine formed during the early crystallizationstage of mantle-derived magmas carry important information aboutthe parental melt composition and hence its source. These mineralscan serve as capsules preserving the Sr and Nd isotopic signature ofmagmas fromwhich they have crystallized. They provide important in-sights into the range of mantle heterogeneities sampled by a single lavaflow, and ultimately degree of themantle source heterogeneity sampledby rift volcanoes. The investigation of the isotopic relationships betweenclinopyroxene and their host lavas provides significant insight into thefundamental nature of the mantle source (e.g. Simonetti and Bell,1993, 1994, 1995; Bryce and DePaolo, 2004; Ramos and Reidy, 2005;Jackson et al., 2009).

7 495 938 20 54.

TheWestern Branch of the East African Rift is a regionwith classic oc-currences of ultra-potassic magmatism (Bailey, 1974; Eby et al., 2003;Foley et al., 1987; Rosenthal et al., 2009). The trace element geochemistryof these ultrapotassic rocks indicates derivation from an enriched mantlesource (e.g. Eby et al., 2003). The great diversity in the modal and chem-ical composition of the volcanic rocks within a limited geographic regionis a reflection of upper mantle heterogeneity at the kilometer scale. Forexample, the mantle source of the ultrapotassic rocks is thought to belherzolite (harzburgite?) characterized by numerous veins and inter-layers of phlogopite-bearing pyroxenite (Lloyd et al., 1999).

This paper focuses on an investigation of kamafugites from the Toro-Ankole and Virunga provinces in the northern part of the WesternBranch of the East African Rift (Fig. 1). Sr, Nd and Pb isotopic data areavailable for the kamafugite rocks (Davies and Lloyd, 1989; Rosenthalet al., 2009) but there is a paucity of isotopic compositions for constitu-ent minerals. Isotope data for minerals and xenoliths contained inkamafugite from Katwe-Kikorongo were first reported by Davies and

http://crossmark.crossref.org/dialog/?doi=10.1016/j.lithos.2014.09.011&domain=pdf

http://dx.doi.org/10.1016/j.lithos.2014.09.011

mailto:[email protected]

http://dx.doi.org/10.1016/j.lithos.2014.09.011

Unlabelled image

http://www.sciencedirect.com/science/journal/00244937

www.elsevier.com/locate/lithos

Fig. 1. Simplifiedmap showing theKivu–Edward–Albert–West Nile rift zone of the East African Rift System. Insertmap shows parts of East Africa, with the location of the two rift branches,and the location of the rift zone (after Lærdal and Talbot, 2002). Inset shows sketch-map of Bunyaruguru volcanic field, showing the explosion craters and lava occurrences that are illus-trated at 11.6 × 6 km (after Holmes, 1942). Numbers in the figure indicate places, where studied specimens were sampled.

261N.S. Muravyeva et al. / Lithos 210–211 (2014) 260–277

Lloyd (1989). Subsequent studieswere undertaken on nephelinite lavasfrom Napak and Mount Elgon volcanoes, eastern Uganda-westernKenya (Simonetti and Bell, 1993, 1995). The objectives of the presentwork are to (1) elucidate the mantle source composition on the basisof Nd, Pb and Sr isotope composition of rock-forming minerals, (2) doc-ument the isotopic relationships betweenminerals and their host rocks,and (3) assess the isotopic evolution of the kamafugite magmas. Thisapproach will help enhance understanding of the nature and degree oflithosphere–asthenosphere interaction in areas of continental rifting.

2. Geological setting and sample localities

The ToroAnkole volcanicfield is the northernmost expression of rift-related magmatism in theWestern Branch of the East African Rift. Geo-logical study of this region was first conducted in 1929 byWayland andCombe (Combe, 1930, 1937) and Holmes (e.g. Holmes, 1942, 1950);subsequent researchers include Gerasimovskiy and Polyakov (1974).

Volcanism in theWestern Rift is limited to four main regions in addi-tion to isolated small areas where volcanic intrusions are present; fromnorth to south these are Toro-Ankole, Virunga, South Kivu and Rungwe.In general, the volcanics form spatially isolated rock units that occur inconjunction with the rift grabens. Western Rift volcanism in the VirungaProvince approximately 12 million years ago and is still active today. Inthe Kivu province, basaltic–trachyte volcanism has been active from8 million ago till themid-Pleistocene (Logatchev et al., 1972), and appearsto predate the beginning of volcanic activity within the Rungwe provincein southern Tanzania. Volcanism in the Toro-Ankole province at thenorthern end of the Western Rift appears to be restricted to the past

1 million years (Ebinger, 1989; Nyablade and Brazier, 2002). K-Ar and40Ar/39Ar ages indicate that volcanism is at most 50,000 years old(Boven et al., 1998). The explosive eruptions that occurredwithin this re-gion continued until about 4000 years ago (Logatchev et al., 1972).

The Toro-Ankole province consists of three volcanic fields: Fort Portal,Katwe-Kikorongo and Bunyaruguru. The Bunyaruguru volcanic field –

which forms the focus of this research – is located south of Lake Georgeand east of Lake Edward (Fig. 1). It consists of numerous, closely spacedLate Pleistocene–Quaternary explosion craters, many of which presentlyhave crater lakes. Most of the volcanics are tuffs; lavas are scarce. Themagmatism of the Bunyaruguru field is generally similar to that of neigh-boring Katwe-Kikorongo, but the volcanic rocks are characterized byhigher Mg# [Mg#= Mg/(Mg + Fe2+) in moles, with Fe2+ = 0.85 totalFe] and awider distribution of olivine phenocrysts (Eby et al., 2003). Sam-ples of ugandite and mafurite used in this study were collected from theBunyaruguru volcanic field (Kazimiro, Mafuru and Nyungu craters;Fig. 1). Katunga, an isolated undissected tuff cone with associated lavaflows located east of the Bunyaruguru field, is the southernmost featureof an N–S-trending chain of high-potassic foiditic volcanism (Fig. 1).Katunga (the type locality for katungite, an olivine-melilitite) is locateddirectly atop themetamorphic basement rocks and contains a freshwaterlake in its summit crater. Two lava flows occurred to the NE of vents onthe northern and NE flanks. The sample of katungite reported here isfrom the crater on the left bank of NE section of Katunga Lake. The ageof the cone is uncertain, but it is contemporaneous with Late Pleistoceneto Recent tuff cones in the Bunyaruguru area, and the undissected condi-tion of the tuff cone and associated lava flows implies a young age(Combe, 1937; Holmes, 1950).

image of Fig.�1

Table 1The mineral composition of the studied ultrapotassic rocks from East African Rift and sample description.

SampleNN

Rock name Phenocrysts Megacrysts Groundmass Inclusions in olivine Notes

11497 Mafurite Olivine, clinopyroxene,Ti-rich magnetite, Cr-spinel

Ti-rich magnetite, phlogopite Cr-spinel, melt and fluid, orthopyroxene,Ti-rich magnetite

11494 Mafurite Olivine, clinopyroxene, Cr-spinel,perovskite, Ti-rich magnetite

Perovskite, apatite, phillipsite, nepheline,kalsilite

Cr-spinel, calcite, Ti-rich magnetite,dolomite, barite, phlogopite, meltand fluid

Xenoliths ofPhl-pyroxenite

11503 Mafurite Olivine, clinopyroxene,perovskite

Clinopyroxene, phlogopite(?), Ti-richmagnetite, calcite, dolomite, barite, glass,nepheline, kalsilite

Xenoliths ofpyroxenite

11513 Katungite Olivine, melilite, perovskite,Ti-rich magnetite

Clinopyroxene, kalsilite, apatite, Ti-richmagnetite phillipsite

Cr-spinel, phillipsite, clinopyroxeneTi-rich magnetite, apatite.

11523 Ugandite Olivine, clinopyroxene, leucite,Cr-spinel, phlogopite, Ti-richmagnetite

Phlogopite Perovskite, nepheline, kalsilite, glass Cr-spinel, fluid, melt + calcite Xenoliths ofglimmerite

11525 Ugandite Olivine, clinopyroxene leucite,Ti-rich magnetite

Phlogopite Perovskite, nepheline, kalsilite,phillipsite, glass

Cr-spinel

11529 Ugandite Olivine, clinopyroxene leucite,Cr-spinel

Perovskite Orthopyroxene

11530 Ugandite Olivine, clinopyroxene leucite,Ti-rich magnetite, Cr-spinel

Phlogopite,olivine,

Leucite, nepheline, kalsilite Cr-spinel, melt, Ti-rich magnetite

11642 Ugandite Olivine, clinopyroxene,phlogopite, leucite, ilmeniteTi-rich magnetite

Phlogopite Olivine, apatite, phlogopite, nepheline,kalsilite

11641 Nepheline–leucitite

Clinopyroxene, leucite,Ti-rich magnetite

Perovskite, apatite, ilmenite, nepheline,kalsilite

11640 Melaleucitite Clinopyroxene, leucite,Ti-rich magnetite

Perovskite, apatite, glass

Note: 11497, 11494—mafurites, sampled on the southern slope of Kyambu crater to the west of Mafuri Lake; 11503—mafurite from the lava flow to the west of the crater Nyungu; 11513—

katungite from the left bank to NE from Katunga Lake; 11523— ugandite lava from the southern slope of the crater Kazimiro; 11525— ugandite, “aa-lava” from the top of the hill of Kazimirocrater; 11529— ugandite (picrite) from the lower part of the southern slope of Kazimiro crater; 11530— ugandite from the a relic of the volcano's crater of the Isinga island on the south side oflake Edward; 11642 — ugandite (leucite picrite) lava from the eastern slope of the crater of the Visoke volcano; 11641— nepheline leucitite from the eastern slope the crater of the volcanoVisoke; 11640 — nepheline melaleucitite from the eastern slope of the crater of Visoke.

262 N.S. Muravyeva et al. / Lithos 210–211 (2014) 260–277

The regional scale geodynamic setting is discussed in more detail byEbinger and Furman (2002). Nyiragongo and nearby Nyamuragiramarkthe western edge of the Virunga volcanic field, with Virunga volcanoeserupting extremely fluid, low-silica, high-alkaline lavas. Several faultsystems meet at Nyiragongo: (1) the main N–S fault which was activein the 1977 and 2002 eruptions, (2) a NW–SE trending system linkingNyiragongo to Nyamuragira, (3) a NE–SW trending fault system includ-ing the Rushayo chain of scoria cones (formed in 1948), and possibly(4) a further fault approximately parallel but west of the Rushayo chain.

Several samples of the ugandite and its probable differentiates(melaleucitite and leucitite) from Visoke volcano in the Virunga arealso included in this work. These samples are from eastern slope ofthe volcano. Visoke is a 3711-m-high extinct stratovolcano, which con-tains a 450-m-wide summit crater lake (Fig. 1). The age of Visoke volca-nism is unknown, but whole rock K–Ar ages for the nearby Karisimbivolcano range from 30,000 to 120,000 years (De Mulder, 1985; Rogerset al., 1992, 1998). Visoke together with Nyiragongo and Nyamuragirais the only volcano that has been active historically in the area.

3. Analytical methods

3.1. Electron microprobe analysis

Minerals were analyzed at the Vernadsky Institute of Geochemistryand Analytical Chemistry RAS, Moscow, Russia, using a САМЕСA SX100 electron microprobe equipped with four wavelength-dispersivespectrometers. Analyses were conducted at 15 kV accelerating voltageand a beam current of 30 nA with a focused beam. For analyses ofmicas and feldspars, a defocused beam diameter of 10 μm was used tominimize the local damage of samples andmigration of ions (especiallyK andNa); these elementswere thefirst to be determined for each anal-ysis. Fwasmeasuredwith a beam current of 15nAanddiameter varyingfrom 5 to 25 μm; counting timewas 20 s. Concentrations were calculat-ed from relative peak intensities of samples using theφ(ρz) algorithmof

Pouchou and Pichoir (1984) and analyses of natural and syntheticstandards.

3.2. Bulk rock major and trace element analysis

Major elements abundances for whole rock samples were analyzedby X-ray fluorescence (Roshchina et al., 1982) with a Phillips PW-1600 instrument at the Vernadsky Institute of Geochemistry and Ana-lytical Chemistry RAS, Moscow, Russia. The samples were carefullycleaned with distilled water (but not leached) before crushing in steeland grinding with a jasper pestle and mortar. Analytical accuracy andprecision were determined by regular measurements of internationaland domestic certified reference materials and internal standards. De-tails of the analytical techniques can be found in Roshchina et al. (1982).

Trace elements were determined by ICP-MS at the Baikal AnalyticalCenter, Irkutsk, which is a joint facility of the Institute of the Earth'sCrust SB RAS, the A.P. Vinogradov Institute of Geochemistry SB RAS andthe Limnological Institute SB RAS. Analyses were conducted using a VGPlasmaQuad 2+ quadrupole mass-spectrometer. Details of the analyticaltechniques can be found in Ivanov et al. (2000) and Yasnygina et al.(2003). The accuracy of the measurements for trace elements based onrepeated analyses of BHVO-1 is estimated to be better than 4.5% (1sd)with exception for Ho, Sm (5–7%), Tm, Yb, Pb (8–10%) and V, Cr, Ni, Zn(10–15%).

3.3. Isotope analysis

Isotopic measurements were conducted on both the whole rock andhandpickedmonomineralic fractions (99% pure). About 500mg of sam-ple powder for each of the studied samples was washed with dilute ni-tric acid on a hot plate for 15 minutes. After drying the samples wereweighed, spiked with 149Sm–150Nd and 87Rb–84Sr tracer solutions anddigested in concentrated HF + HNO3 + HClO4 at 120 °C for 2–5 days.Separation of Rb, Sr, Sm, Nd and Pb was conducted following standard

Table 2The major, trace element and isotope composition of Sr, Nd and Pb of the studied ultrapotassic rocks from East African Rift.

Sample 11494 11497 11503 11513 11523 11525 11529 11530 11640 11641 11642

SiO2 38.7 39.92 37.68 36.37 39.88 38.98 41.58 40.44 40.41 43.49 43.49TiO2 2.96 3.35 4.82 4.85 4.86 4.73 3.92 5.02 5.48 5.05 4.69AL2O3 7.11 7.46 7.35 6.13 7.68 6.77 5.68 9.27 10.76 14.67 8.92FeO* 11.69 9.80 13.12 11.53 11.90 12.27 12.24 14.26 14.18 10.12 10.36MnO 0.154 0.22 0.23 0.23 0.14 0.15 0.23 0.20 0.23 0.20 0.19MgO 19.89 17.71 9.39 13.21 16.86 17.35 21.48 9.76 5.72 3.53 10.96CaO 10.09 10.99 14.58 16.45 9.57 11.20 6.52 12.24 13.00 7.90 13.50Na2O 0.58 1.55 1.42 2.02 0.56 0.89 1.24 2.61 3.56 4.46 1.89K2O 5.41 6.25 5.81 4.12 5.61 4.51 4.52 4.17 5.06 8.76 3.88LOI 2.11 1.48 2.93 3.47 2.02 3.13 1.24 0.64 0.50 1.28 0.70P2O5 0.55 0.69 1.18 n.d. 0.22 0.66 0.16 0.36 0.98 n.d. n.d.Total 99.24 99.81 99.40 99.08 99.29 100.64 99.35 99.88 99.88 99.80 99.48Mg# 0.78 0.79 0.60 0.71 0.75 0.75 0.79 0.59 0.46 0.42 0.69V 161 170 184 203 152 136 142 286 427 272 311Cr 605 859 194 372 666 451 1169 443 48 11 728Ni 535 460 75 208 373 427 758 142 52 20 139Zn 92 87 118 119 107 112 99 125 141 124 97Rb 164 157 115 112 167 140 154 112 137 270 111Sr 1535 1602 2754 2537 1300 1714 860 1433 1636 1514 860Y 12 13 17 16 10 12 6.9 14.5 25 25 15Zr 204 210 452 315 266 333 153 379 376 385 224Nb 156 150 276 244 145 184 94 166 195 216 107Ba 1964 2149 2372 1954 2370 1987 1207 1140 1899 1566 1153La 151 160 241 244 118 143 68 133 152 167 84Ce 281 342 343 347 222 271 132 252 279 292 162Pr 30.7 32.6 50.9 52.1 24.0 29.8 14.0 28.3 30.7 32.2 18.0Nd 106 113 184 188 89 110 50 100 112 116 67Sm 13.3 13.8 21.9 21.7 11.7 13.5 6.53 13.7 15.3 15.4 9.51Eu 3.61 3.73 5.98 5.94 3.25 3.84 1.81 3.61 4.18 4.07 2.65Tb 1.07 1.14 1.76 1.69 0.95 1.06 0.55 1.21 1.58 1.51 0.98Gd 8.23 8.79 14.3 13.7 7.51 8.91 4.35 9.21 11.5 11.0 7.17Dy 4.16 4.33 6.85 6.25 3.51 4.30 2.11 4.54 6.71 6.68 4.15Ho 0.53 0.57 0.79 0.70 0.45 0.54 0.31 0.61 1.05 1.01 0.65Er 1.14 1.26 1.71 1.55 1.11 1.04 0.69 1.49 2.50 2.62 1.59Tm 0.18 0.20 0.22 0.21 0.14 0.17 0.11 0.25 0.40 0.34 0.24Yb 1.05 1.08 1.27 1.16 0.77 0.86 0.55 1.15 2.34 2.20 1.39Lu 0.15 0.17 0.20 0.17 0.13 0.13 0.12 0.16 0.31 0.36 0.20Pb 7.60 7.48 10.84 9.09 4.96 6.49 3.83 1.28 9.61 5.58 4.10Th 18.7 20.6 35.9 33.4 14.6 19.0 9.58 16.9 22.6 26.9 12.4U 4.50 4.80 7.78 7.30 3.13 4.26 1.91 4.06 4.98 5.23 2.65143Nd/144Nd 0.51249 0.51249 0.51252 0.51253 0.51250 0.51249 0.51254 0.51255 0.51254 0.51250 0.5125387Sr/86Sr 0.70519 0.70514 0.70463 0.70481 0.70520 0.70515 0.70536 0.70477 0.70570 0.70577 0.70565206Pb/204Pb 19.439 19.431 19.350 19.566 19.283 19.383 19.328 18.998 19.292 19.385 19.432207Pb/204Pb 15.722 15.720 15.703 15.702 15.737 15.719 15.715 15.686 15.669 15.700 15.722208Pb/204Pb 40.264 40.163 39.958 40.186 39.991 40.158 39.999 39.303 40.096 40.431 40.433

LOI— loss on ignition, including H2O and CO2; * — total Fe as FeO; Mg# = Mg/(Mg + Fe2+) in moles, calculated by setting Fe2+ = 0.85 of total Fe.


two stage ion-exchange and extraction chromatography (Richard et al.,1976) with minor modifications according to Pin et al. (2003). Isotopeabundance measurements were performed on a Triton (ThermoFisher)mass spectrometer equipped with nine collectors under static mode atCIR VSEGEI, St. Petersburg. The 143Nd/144Nd ratios are normalizedwithin-run to 146Nd/144Nd = 0.72190 and adjusted to 143Nd/144Nd =0.511860 for the La Jolla standard. Strontium isotope ratios were nor-malizedwithin-run to 88Sr/86Sr= 8.375209. The value of SRM-987 dur-ing this work was 87Sr/86Sr = 0.710248 ± 15 (2σ, 6 measurements).Assigned errors (2σ) for 147Sm/144Nd and 143Nd/144Nd were ± 0.3%and ± 0.000015, respectively, and for 87Rb/86Sr and 87Sr/86Sr ±0.5% and ± 0.000025. The 2σ errors cited in Table 2 for 143Nd/144Nd and 87Sr/86Sr are in-run precisions. The total procedure blankswere 0.01 ng for Sm, 0.02 ng for Nd, 0.01 ng for Rb and 0.1 ng for Sr. Thedata obtained for the international standard BCR-2 during this analyticalwork yielded average values of 143Nd/144Nd= 0.512648±4 (2σ, 6 runs)and 143Nd/144Nd = 0.512106 ± 5 (2σ, 6 measurements) for JNdi-1.

Analysis of the isotopic composition of Pb was made on a TRITONTI (Thermo Fisher) solid-state multicollector mass spectrometer inthe static mode. Chemical separation of Pbwas carried out by columnion-exchange chromatography as described above. Blank levels dur-ing the analytical work did not exceed 0.02 ng for the Pb standard.Measurements of standard NIST-981 yield average values of 206Pb/

204Pb = 16.937 ± 0.011, 207Pb/204Pb = 15.492 ± 0.017, 208Pb/204Pb = 36.722 ± 0.017. The associated absolute uncertainties arequoted at the 2 s level (2σ, abs).

4. Results

4.1. Petrographic sample description

Ultrapotassic lavas in theWestern Rift have a complexnomenclaturethat obscures the petrologic and geochemical similarities among vari-ous rock types. Specifically, kamafugutes – a collective term proposedfor a katungite, mafurite, ugandite, and other less abundant rockstypes (Sahama, 1974) – are characterized by relatively low SiO2, Al2O3

and FeOTot but extremely high CaO and Na2O (Foley et al., 1987).Kamafugites are feldspar-free rocks dominated by kalsilite, nepheline,melilite and sometimes leucite as felsic phases while mafic phases con-sist dominantly of olivine, clinopyroxene and phlogopite.

The specimens investigated here consist of fresh, homogeneouslavas with porphyritic texture containing predominantly phenocrystsof olivine and clinopyroxene. Both mafurites and ugandites contain xe-noliths of glimmerite and pyroxenite (Table 1). Ugandite or olivinemelaleucitite contains phenocrysts of olivine, clinopyroxene, spineland leucite; smaller amounts of nepheline, kalsilite, ilmenite,

Fig. 2. Content of CaO andMgO in the rocks from various volcanicfields ofWestern branchof the East African Rift Zone: Toro-Ankole; field North from Kivu Lake; silica-rich (b8%MgO) volcanits; South Kivu field; Visoke volcano; Muhavura. The two fields, delineatedby dashed lines in the diagram, correspond to trends of a change in the melt compositionthrough the fractionation of olivine and clinopyroxene. Toro-Ankole kamafugites in con-trast to lavas from other volcanic fields of Western branch (Virunga, Kivu) characterizedby only olivine fractionation. Data from present work and other our unpublished dataare shown by black symbols. The grey symbols denote published data from Muhavura:low silica lava — Rogers et al., 1998; Karisimbi: primitive K-basanite (PKB) — Rogerset al., 1992; Kivu lavas: Furman and Graham, 1999; Rungwe mafic lavas — Furman,1995; Toro-Ankole — Rosenthal et al., 2009.


perovskite, apatite and phlogopite are alsopresent in these rocks. Phlog-opite is found both as megacrysts (~1–2 mm) and phenocrysts (~100–200 μm). Megacrysts exhibit marginal zoning and nonequilibrium tex-tures relative to the surrounding groundmass. Some megacrysts ofphlogopite have been significantly altered, resulting in the formationof new reaction minerals. Olivine and clinopyroxene are commonlyzoned, sometimes occurring as megacrysts with irregular, corrodededges. Veins of fresh yellow glass are found in the cracks of these crys-tals. Opaque phases range in composition from Cr-rich spinel to Ti-

Fig. 3. Trace element characteristics of Toro-Ankole kamafugites and Virunga lavas studied, nocomparison (see set in Fig. 3) with representative samples of HIMU, EM-1, EM-2, average normsediment, GLOSS (after Hofmann, 2003). Numbers in the figure correspond to the numbers of

rich magnetite, with rare ilmenite and perovskite. Leucite is present inall rocks as phenocrysts and in the groundmass.

Mafurite lavas consist of phenocrysts of olivine and clinopyroxene,with small amounts of groundmass perovskite, nepheline, kalsilite, apa-tite, clinopyroxene, mica, Ti-richmagnetite, calcite, dolomite, barite andglass (fresh and devitrified). Cr-spinel occurs as inclusions in olivinephenocrysts. Katungite or olivine kalsilitite is distinguished from otherkamafugites by the presence of melilite instead of clinopyroxene inthe phenocryst suite.

Melaleucitite and nepheline leucitite are distinguished from theother volcanic rocks by the absence of olivine. Toro Ankolemelaleucititeis glassywith rare clinopyroxene phenocrysts (grain size up to 2.5mm),Ti-richmagnetite, leucite andperovskitemicrophenocrysts and ground-mass apatite. Nepheline leucitite contains large euhedral leucite pheno-crysts surrounded by finer needle-shaped clinopyroxene and Ti-richmagnetite phenocrysts. Ilmenite occurs in the groundmass and as la-mellae of solid solution breakdown in Ti-rich magnetite; groundmassnepheline is also present. Perovskite occurs as microphenocrysts(~50–70 μm) and inclusions in leucite and clinopyroxene.

4.2. Whole rock compositions

The major and trace element compositions of the samples studiedhere are listed in Table 2. The low SiO2 (b42 wt.%), high MgO (up to21.5 wt.%, withMg# up to 79), low Al2O3 (b10wt.%), and high CaO con-tents (up to 16.45 wt.%) of these volcanics reflect the presence of modalclinopyroxene, kalsilite, leucite, melilite and perovskite. The major ele-ment compositions of the rocks from different volcanic fields of theWestern Branch of the East African Rift define two trends distinguishedby variations in CaO and MgO contents that correspond to olivine andclinopyroxene fractionation (Fig. 2). The Toro-Ankole kamafugiticlavas are characterized by fractional crystallization of olivine, whilelavas from other volcanic fields of the Western Branch (e.g. Virunga,Kivu, Rungwe) record predominantly clinopyroxene fractionation.

There are also consistent differences in the trace element contents ofsamples from the two volcanic provinces (Table 2). Toro-Ankolekamafugites have high Mg# (59–79) and Ni and Cr contents (75–758

rmalized to primitive-mantle (McDonough and Sun, 1995) are shown on spidergramm inal MORB, Average Mauna Loa (Hawaii), average continental crust and average subductingsamples studied (the rock names see in Table 1).

image of Fig.�2

image of Fig.�3

Table3

Oliv

inecompo

sition

(selectedrepresen

tative

analyzes).

Sample11

523a

1152

3a11

523a

1152

3a11

523a

1164

211

642

1164

211

503

1150

311

503

1150

311

503

s114

94a

1149

4a11

494a

1149

4a11

494a

1149

4a11

494a

1153

011

530

1153

011

530

1153

011

530

Grain

12c

2r3c

3r1

23

12c

2r3

41

2c2r

3c3r

4c4r

1c1r

2c2r

3c3r

SiO2

40.29

40.37

39.91

40.3

40.53

40.23

39.46

39.45

39.14

39.2

39.19

39.01

39.39

41.03

41.19

39.88

40.26

40.74

40.53

40.18

39.62

39.63

39.03

38.63

39.31

39.98

FeO

10.57

9.62

10.44

9.64

9.94

13.05

13.62

14.58

14.09

12.58

13.42

13.94

12.7

8.47

8.00

14.65

12.93

9.94

10.11

10.37

12.13

12.55

15.06

16.18

13.9

14.55

MnO

0.13

0.14

0.19

0.12

0.12

0.21

0.19

0.21

0.45

0.23

0.4

0.44

0.29

0.10

0.14

0.59

0.24

0.19

0.15

0.28

0.17

0.20

0.30

0.36

0.28

0.23

MgO

49.19

49.02

48.56

48.71

48.46

46.67

46.37

45.69

44.2

46.09

44.8

44.19

45.77

50.63

50.08

44.38

46.55

48.83

48.05

47.22

48.17

46.36

44.73

43.42

45.73

44.78

CaO

0.23

0.19

0.23

0.16

0.15

0.38

0.40

0.50

1.25

0.31

0.82

1.12

0.45

0.12

0.15

0.87

0.23

0.30

0.32

0.91

0.26

0.28

0.41

0.30

0.29

0.38

NiO

0.25

0.41

0.29

0.48

0.44

0.21

0.19

0.22

0.1

0.18

0.09

0.08

0.11

0.41

0.41

0.14

0.14

0.17

0.17

0.11

0.30

0.25

0.18

0.09

0.24

0.13

Total

100.73

99.78

99.69

99.46

99.69

100.76

100.24

100.64

99.24

98.59

98.74

98.79

98.71

100.67

100.76

99.97

100.51

100.35

100.17

99.33

99.07

100.65

99.27

99.71

98.98

99.75

Si0.98

60.99

40.98

80.99

50.99

90.99

50.986

0.98

60.99

30.99

10.99

40.99

30.99

50.99

41.00

30.99

90.99

90.99

91.00

21.00

00.97

90.99

30.98

70.98

90.98

71.00

2Fe

0.21

60.19

80.21

60.19

90.20

50.27

00.28

40.30

50.29

90.26

60.28

50.29

70.26

80.17

20.16

30.30

70.26

80.20

40.20

90.21

60.25

10.26

30.31

90.34

60.29

20.30

5Mn

0.00

30.00

30.00

40.00

30.00

30.00

40.00

40.00

40.01

00.00

50.00

90.00

90.00

60.00

20.00

30.01

30.00

50.00

40.00

30.00

60.00

40.00

40.00

60.00

80.00

60.00

5Mg

1.79

51.79

81.79

11.79

31.78

01.72

11.72

61.70

21.67

01.73

61.69

41.67

61.72

21.82

81.81

71.65

71.72

11.78

41.77

11.75

21.77

41.73

21.68

61.65

71.71

21.67

2Ca

0.00

60.00

50.00

60.00

40.00

40.01

00.01

10.01

30.03

40.00

80.02

20.03

10.01

20.00

30.00

40.02

30.00

60.00

80.00

80.02

40.00

70.00

80.01

10.00

80.00

80.01

0Ni

0.00

50.00

80.00

60.01

00.00

90.00

40.00

40.00

40.00

20.00

40.00

20.00

20.00

20.00

80.00

80.00

30.00

30.00

30.00

30.00

20.00

60.00

50.00

40.00

20.00

50.00

3To

tal

3.01

33.00

63.01

23.00

43.00

03.00

53.01

43.01

43.00

73.00

93.00

63.00

73.00

53.00

62.99

73.00

13.00

13.00

12.99

83.00

03.02

13.00

63.01

33.01

13.01

22.99

8Fo,%

89.2

89.3

90.1

89.2

90.0

89.7

86.4

85.8

84.8

86.3

84.8

86.7

85.6

91.3

91.4

91.8

84.4

86.5

89.7

89.4

89.0

87.6

86.8

84.1

82.7

85.4

c—

Core,r

—rim

ofgrain;

mineral

form

ulae

werecalculated

forO

=4;

Fo,%

—mol

percen

tof

forsterite

inolivine.

aDatapu

blishe

din

Murav

'eva

andSe

nin(200

9).


and 194–1169 ppm, respectively), which are accompanied by extremeenrichment of some trace elements (notably LREE and Zr; Fig. 3)resulting in high values of Ce/Y (17.4 to 26.4), Zr/Y (16.2 to 28.7), (La/Yb)n (78.2 to 142.7) and (La/Sm)n (6.1 to 7.3), where “n” denoteschondrite-normalized values. Lavas from Kazimiro crater (samples11529, 11525, 11523) in the Bunyaruguru volcanic field display thelargest HREE depletions, with concentrations as low as 1–2 ppmresulting in (Tb/Yb)n values of 3.02–3.15. The low Yb contents suggestthe presence of residual garnet in the mantle source. Melaleucitite(11640) and leucitite (11641) samples from Visoke volcano in Virungaare the least depleted in HREE, with (Tb/Yb)n values of 4.54–6.47. Thesegeochemical features of mafic lavas in the Virunga and Bunyaruguruvolcanic fields likely reflect differences in the composition of their man-tle sources.

4.3. Mineral compositions

4.3.1. OlivineOlivine occurs in all kamafugite samples, where it exhibits both nor-

mal and reverse zoning, but is absent in melaleucitite and leucitite(samples 11641 and 11640). The forsterite content of the olivinepheno-crysts varies from 82.7 to 91.8 mol% (Table 3). Sparse angular clasts oflarger crystals (Fo90) are also present. There are some polycrystallineaggregate nodules with 2–3 mm crystals of olivine in mafurite sample11494 as described by Murav'eva and Senin (2009). Olivines in thisstudy have 0.08–0.48wt.% NiO and 0.07–1.25wt.% CaO; these elementsshow positive (NiO) and inverse (CaO) correlations with forsterite con-tent. Olivine phenocrysts often contain melt, fluid and/or crystalline in-clusions. Among the common crystalline inclusions are Cr-rich spinel,titanomagnetite and clinopyroxene; less common are carbonates, sul-fates, and mica (Murav'eva and Senin, 2009).

4.3.2. SpinelSpinel occurs as phenocrysts and as inclusions in olivine. Cr-rich spi-

nel (with≤58.4 wt.% Cr2O3,≥10.65wt.% Al2O3, and≥2.36 wt.% TiO2) ispresent in ugandite and mafurite samples. In some samples (11530,11503, 11642, 11641, 11640), Ti-rich magnetite is the only spinel phe-nocryst phase present. The composition of the spinels varies withinone sample from Cr-spinel with low Al content to almost pure Ti-richmagnetite (Table 4). The rims of zoned spinel phenocrysts are enrichedin Ti-rich magnetite. The presence of a continuous solid solution seriesof spinel group minerals in ugandites and mafurites indicates continu-ous crystallization of a spinel phase at all stages during the evolutionof their parental melts.

4.3.3. MicaPhlogopite and fluorophlogopite in ugandites occur as megacrysts

(11530, 11525, 11523) and phenocrysts (11523, 11642). In mafuritesthese micas are present only as inclusions in olivine and in the ground-mass. Phlogopite can occur as fine-grained aggregates in high-Mg oliv-ine. The compositions of phlogopite are given in Table 5. These arehigh-magnesian (0.83–0.91 Mg#) compositions with variable Cr, Ti, Aland F contents. Phlogopites in the mafurite (11494) and ugandite(11523) contain up to 7.53 wt.% fluorine.

4.3.4. ClinopyroxeneClinopyroxene occurs in all samples as phenocrysts, megacrysts and/

or groundmass microlites; it is occasionally present as inclusions inother minerals, e.g., in katungite (sample 11513) where it occurs only asinclusions in olivine. Phenocrysts are generally quite primitive (Table 6)with Mg#s up to 89. In most kamafugites, clinopyroxene is characterizedby very low Cr2O3 contents (0.02–0.36 wt.%). Within each sample, thevariation in major element abundances is significant (e.g., sample 11503cpx Mg# ~0.45–0.9), between different grains or within single crystals.Both normal and reverse zonation are present and are clearly visible onbackscattered electron images (Fig. 4). Variations of titanium and sodium

Table 4Spinel composition (selected representative analyzes).

Sample 11523a 11523a 11523a 11523a 11523a 11523a 11523a 11494a 11494a 11494a 11494a 11494aa

Grain 1c 1r 2 3c 3r 4 5 1c 1r 2c 2r 3

SiO2 0.08 0.07 0.79 0.11 0.43 0.04 0.62 0.13 1.29 0.06 0.12 0.12TiO2 5.21 12.66 13.57 6.55 10.75 11.92 9.65 3.42 11.53 4.51 11.58 3.52Al2O3 10.28 0.28 0.07 1.55 0.25 0.24 4.47 4.92 0.08 6.74 0.07 4.84Cr2O3 33.1 11.76 0.78 34.04 17.24 18.67 35.15 51.33 5.49 45.63 6.04 52.03V2O5 0.19 0.44 0.5 0.2 0.33 0.43 0.21 0.08 0.38 0.12 0.41 0.11Fe2O3

b 17.14 33.90 41.51 23.26 30.77 29.53 13.49 7.58 37.78 10.13 40.05 7.55FeOb 26.09 36.97 39.80 23.37 31.12 32.17 26.00 24.40 39.03 25.56 37.08 24.78MnO 0.78 1.03 1.11 3.09 2.53 2.34 0.38 1.11 0.81 1.1 0.92 1.2MgO 7.43 3.39 2.54 7.45 5.05 5.25 10.23 6.79 2.11 7.01 2.34 6.77NiO 0.13 0.09 0.17 0 0 0.12 0.16 0.01 0.93 0.07 0.10 0.04Total 100.42 100.59 100.84 99.62 98.47 100.70 100.36 99.77 99.43 100.93 98.72 100.95

Sample 11642 11642 11642 11642 11530 11530 11530 11530 11530 11530 11530 11530

Grain 1 2 3 4 1 2 3 4 5 6 7 8

SiO2 0.07 0.05 0.08 0.02 0.09 0.07 0.05 0.05 0.05 0.04 2.54 4.16TiO2 21.77 21.99 22.02 23.46 13.37 19.76 16.88 17.76 22.17 17.03 16.21 15.08Al2O3 1.88 1.85 1.94 1.89 0.04 0.07 0.05 1.50 0.44 0.03 1.11 2.90Cr2O3 0.27 0.34 0.31 0.07 0.37 0.37 0.54 8.44 1.37 0.41 0.19 4.20V2O5 0.55 0.57 0.51 0.69 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.Fe2O3b 27.35 27.25 26.58 24.94 42.90 30.22 38.09 25.32 24.75 36.42 30.72 25.65FeOb 41.09 40.66 41.37 41.78 40.92 46.75 40.72 41.89 47.41 43.59 45.80 36.85MnO 1.07 1.04 1.03 0.97 0.81 0.97 0.97 1.32 0.80 0.77 0.96 0.45MgO 6.23 6.70 6.22 6.89 1.08 0.96 3.44 2.93 2.12 1.63 1.73 8.55NiO 0.06 0.00 0.00 0.11 0.09 0.07 0.06 0.04 0.08 0.04 0.06 0.09CaO n.d. n.d. n.d. n.d. 0.14 0.09 0.18 0.07 0.08 0.15 0.13 2.23Total 100.33 100.46 100.07 100.81 99.81 99.33 100.99 99.32 99.28 100.11 99.44 100.15

C — core, r — rim.a Data published in Murav'eva and Senin (2009).b Calculated by stoichiometry.


against Mg# (Fig. 5) show two distinct chemical trends. Trend A is char-acterized by high Na2O contents that decrease from 2.6 to 0.2 wt.% withlow near-constant TiO2 (1 ± 0.6 wt.%); this pattern occurs in cores of

Table 5Mica composition (selected representative analyzes).

Sample 11642 11642 11642 11523 11523

Grain 1 2 3 1 2

SiO2 35.86 36.02 34.72 36.96 35.42TiO2 7.72 7.80 9.42 6.66 7.13Al2O3 15.58 15.37 15.53 11.15 10.49Cr2O3 0.31 0.23 0.15 0.03 0.04FeOa 9.03 9.22 9.15 5.84 6.28MnO 0.08 0.04 0.06 0.08 0.00MgO 17.29 17.80 16.83 22.57 21.86CaO 0.03 0.01 0.06 0.02 0.03BaO 0.69 0.65 1.49 2.42 2.06Na2O 0.55 0.54 0.54 0.55 0.66K2O 9.13 9.53 9.05 9.50 9.47F 0.47 0.43 0.42 5.87 5.35Total 99.99 100.97 100.77 100.38 97.99

apfuSi 2.594 2.586 2.513 2.700 2.664Ti 0.420 0.421 0.513 0.366 0.403Al 1.329 1.301 1.325 0.960 0.930Cr 0.018 0.013 0.009 0.002 0.002Fe 0.546 0.554 0.554 0.357 0.395Mn 0.005 0.002 0.004 0.005 0.000Mg 1.864 1.905 1.815 2.457 2.450Ca 0.002 0.001 0.005 0.002 0.002Ba 0.020 0.029 0.067 0.069 0.061Na 0.077 0.075 0.076 0.078 0.096K 0.843 0.873 0.836 0.885 0.909F 0.108 0.098 0.096 1.396 1.273OH 1.892 1.902 1.904 0.604 0.727Mg# 0.77 0.77 0.77 0.87 0.86

Mineral formulae were calculated for O = 11; Mg# = Mg/(Mg + Fe) in moles.a All Fe as FeO.b Rim of grain.

zoned phenocrysts, crystalline inclusions and pyroxene from xenoliths(Lloyd, 1981). Trend B, which is observed in some phenocrysts (e.g., insample 11641), microlites and the outer rims of most phenocrysts, is

11525 11530 11494 11494 11494 11494

b 1 2 3 4

35.24 37.17 37.86 38.01 42.28 41.535.86 6.19 5.19 5.13 2.08 2.67

15.09 14.08 15.12 14.89 8.63 8.730.36 1.51 0.99 0.95 0.05 0.06

18.65 7.54 7.01 6.88 4.55 4.620.26 0.05 0.00 0.00 0.00 0.008.64 18.84 19.36 19.08 23.88 23.720.31 0.02 0.05 0.04 0.00 0.060.29 0.54 0.37 0.55 1.20 1.392.73 0.23 0.12 0.16 0.96 0.859.10 9.69 10.26 10.08 10.49 10.330.12 0.81 0.82 0.44 7.53 7.58

100.75 100.25 100.56 99.96 99.04 98.90

2.687 2.676 2.697 2.732 2.859 2.8180.336 0.335 0.278 0.277 0.106 0.1361.357 1.195 1.270 1.262 0.688 0.6980.022 0.086 0.056 0.054 0.003 0.0031.189 0.454 0.418 0.414 0.257 0.2620.017 0.003 0.000 0.000 0.000 0.0000.982 2.021 2.056 2.044 2.406 2.3980.025 0.002 0.004 0.003 0.000 0.0040.014 0.024 0.016 0.025 0.051 0.0590.404 0.032 0.017 0.022 0.126 0.1120.885 0.890 0.933 0.924 0.905 0.8940.027 0.185 0.188 0.101 1.723 1.7351.973 1.815 1.812 1.899 0.277 0.2650.45 0.82 0.83 0.83 0.90 0.90


characterized by varying TiO2 content (from 0.3 to 5.8 wt.%) at almostconstant Na2O content (0.5–0.3 wt.%). Both trends occur within sample11503 (Fig. 5) where the dashed lines connect the composition of coreand rim for each clinopyroxene grain.

4.4. Sr–Nd–Pb isotope systematics

4.4.1. Whole rocksStrontium and neodymium isotopic ratios measured for the Toro-

Ankole kamafugites range from 87Sr/86Sr: 0.704629 to 0.705356 and143Nd/144Nd: 0.512488 to 0.512550 (Table 1, Fig. 6) (Muravyeva andBelyatsky, 2009), consistent with previous studies (Davies and Lloyd,1989; Rosenthal et al., 2009) and within the range of EM1-type oceanicisland basalts (Hofmann, 2003; Stracke et al., 2005). In contrast, the leadisotopic compositions for the same samples (206Pb/204Pb: 18.998–19.566, 207Pb/204Pb: 15.686–15.737, 208Pb/204Pb: 39.303–40.264 —

Table 1) indicate similarities to ocean island volcanics derived from amore radiogenic EM2 source (Hofmann, 2003).

4.4.2. MineralsSr, Nd and Pb isotopic compositions have been determined for

clinopyroxene and mica phenocrysts from six samples: four kamafugitesfrom Toro Ankole and one sample each of ugandite and leucitite fromVisoke volcano. The Sr and Nd isotopic compositions for theclinopyroxenes (Table 7) demonstrate varyingdegrees of isotopic equilib-rium between mineral and corresponding whole rock isotopic composi-tions. Clinopyroxene data indicate consistent differences in the degreeof mineral-rock disequilibrium between Nd and Sr: in cases where the143Nd/144Nd values are close to equilibrium, the majority ofclinopyroxenes are not in 87Sr/86Sr equilibriumwith the whole rock. Fur-ther, phenocrysts may be either depleted or, rarely, enriched relative tothe corresponding host rock (Fig. 7).

Clinopyroxene Pb isotope compositions also differ from those oftheir host rocks. Measured values of 206Pb/204Pb, 207Pb/204Pb and208Pb/204Pb in crystals are appreciably less radiogenic (Table 7) thanthose of their host rocks, with the exception of Visoke leucitite 11641,in which clinopyroxene is in Pb isotope equilibrium with its host rock.The Pb isotopic ratios (Fig. 8) form a sub-parallel trend above theNorth Hemisphere Reference Line (NHRL, Hart, 1984).

5. Discussion

5.1. Chemical and isotopic disequilibrium between clinopyroxene and hostrocks

A sign of the chemical equilibrium of crystals with melt is their ho-mogeneity, whereas zoning of phenocrysts indicates a clear absence oflong-term equilibrium. We focus on Mg# as the indicator for zoning,as this parameter changes predictably during normal melt evolutionasmanifest in decreasingMg# from core to rim. Reverse zoningmay re-cord mixing of melts, capture of xenocrysts, or a change in the redoxcondition and sudden onset of titanomagnetite crystallization. Herewe consider these processes by exploring the relationship of Na2O andTiO2 with Mg# (Fig. 5).

The cores of the individual pyroxene grains and clinopyroxene inclu-sions in phenocrysts are characterized by elevated concentrations ofiron (up to 17 wt.% FeO) and sodium (up to 2.5 wt.%. Na2O), while rimcontents range from 4 to 6 wt.% FeO and ~0.4–0.7 wt.% Na2O. Suchclinopyroxenes have been described previously for alkaline basaltsfrom Uganda (Lloyd, 1981) as well as from other provinces,e.g., Central French massif (Pilet et al., 2002, 2004), West Eifel (Dudaand Schminke, 1985), Italy (Barton et al., 1982) and Western Yunnan,China (Xu et al., 2003). Crystallization of pyroxene cores and inclusionscould have taken place frommeltswith significant compositional differ-ences from their corresponding whole rock. The formation of such“green-core” clinopyroxenes is associated with high-pressure

crystallization of alkaline basalt magma (Duda and Schmincke, 1985)and/ormantlemetasomatism (Pilet et al., 2002). Clinopyroxenes of sim-ilar composition (poor in Al2O3, rich in Na2O) were obtained in meltingexperiments conducted at a pressure of 30 kbar and 1230 °C usingUganda xenoliths (Lloyd et al., 1985).

The natural clinopyroxenes define two evolution trends (Fig. 5)suggesting different petrogenetic histories. Similar trends wereoutlined for green-core clinopyroxenes from ultrapotassic lavasWestern Yunnan, China (Xu et al., 2003). Clinopyroxenes that dis-play reverse zonation define a trend that may be attributed to crys-tallization under conditions of decreasing pressure andtemperature: the distribution coefficient for Na betweenclinopyroxene and liquid increases with increasing pressure(Blundy et al., 1995; Putirka et al., 2003), while the opposite is ob-served for Ti (Walter, 1998; Prytulak and Elliott, 2007; Pertermannand Hirschmann, 2003; Blundy et al., 1995; Pertermann andHirschmann, 2003; Walter, 1998; Prytulak and Elliott, 2007). Amore quantitative evaluation of pressure was carried out usingsingle-pyroxene geothermobarometry (Mercier, 1980). For green-core clinopyroxenes (sample 11503) results were obtained only forthe outer zones, since the method is not suitable for alkaline-richcompositions. The calculated depths of crystallization are extremelyvariable and range between 10 and 40 kbar, suggesting continuousclinopyroxene crystallization that occurred between ~120 and~30 km.

The degree of isotopic disequilibrium between clinopyroxene andhost lavas is much larger for 87Sr/86Sr and 206Pb/204Pb, 207Pb/204Pb and208Pb/204Pb than for 143Nd/144Nd (Fig. 7). The Sr isotopic disequilibriumis coherent with major element clinopyroxene composition (Fig. 7).Clinopyroxenes enriched in radiogenic strontium relative to their hostrocks are from samples in which reverse zonation predominates (espe-cially 11503, Figs. 5 and 7), suggesting that clinopyroxenes which crys-tallized at high pressure recordmore radiogenic Sr signatures than theircorresponding host rocks, whereas clinopyroxenes formed at lowerpressures are more depleted than their host rock.

The Pb isotope compositions of the kamafugite clinopyroxenes arefor the most part not in equilibriumwith their host rock; the exceptionis nepheline leucitite sample 11641 where isotope equilibrium is ob-served. Interestingly, the Pb isotope compositions of the clinopyroxenefrom kamafugites plot closer to the EM1 field than their correspondingwhole rocks (Fig. 8). This observation might be explained by mixing ofthe clinopyroxene-bearing melts with EM1 signatures with a meltfrom a more enriched (HIMU-like?) source. The Pb isotopic similarityof the clinopyroxene crystals to the EM1mantle endmember is also ap-parent in plots of 206Pb/204Pb versus 87Sr/86Sr and 206Pb/204Pb versus143Nd/144Nd (Fig. 9). This relationship suggests that the clinopyroxenecrystals (and a fraction of their host lavas) were entrained by newmagma that came from greater depth, i.e., these younger magmaswere most likely derived from sublithospheric depths and likelyplume-related. This assumption is supported by our pyroxenegeobarometry calculations. Alternatively, it is possible that the isotopicdisequilibrium documented here results from disintegration of litho-spheric xenoliths; Pb isotopic compositions of the majority of studiedclinopyroxenes approach to the lower limit values obtained for Toro-Ankole pyroxenite xenoliths (see Fig. 10.7 in Davies and Lloyd, 1989).At present we are unable to distinguish between these two possible in-terpretations but favor the incorporation of sublithosphericmelts in theabsence of isotopic data that would require disaggregation of pyroxe-nite xenoliths.

The inferred sublithospheric kamafugite melts that mixed with theclinopyroxene-bearing liquids appear to be considerably more magne-sian, as indicated by the rim compositions of the zoned phenocrysts; re-versely zoned clinopyroxene has rimMg# higher than those of the hostrock. For example, clinopyroxene from sample 11503 has a rim withmaximum Mg# = 0.88. Using the equilibrium value for Fe–Mgpartitioning of Putirka et al. (2003), the calculated melts in equilibrium

Table 6Clinopyroxene composition (selected representative analyzes).

Sample N 11523 11523 11523 11497a 11497a 11497a 11503 11503 11503 11503 11503 11503 11530 11530

Grain 1c 1r 2 1c 1r 2 1c 1r 2c 2r 3c 3r 1 2c

SiO2 51.13 51.33 50.83 53.40 54.60 53.45 50.44 51.78 52.02 51.63 50.84 52.37 48.14 51.51TiO2 1.41 1.38 2.27 1.45 1.44 1.06 0.77 1.79 0.55 1.74 0.69 1.60 3.19 1.57Al2O3 1.99 1.95 2.15 2.29 0.54 0.43 2.25 1.97 0.89 1.89 1.18 1.19 4.05 1.82Cr2O3 0.01 0.10 0.18 0.07 0.05 0.07 0.00 0.20 0.00 0.02 0.00 0.01 0.11 0.46FeOb 5.53 5.65 4.42 3.75 3.68 3.87 15.39 4.35 11.63 4.48 16.70 4.11 6.48 4.86MnO 0.10 0.09 0.09 0.08 0.12 0.07 0.38 0.08 0.26 0.00 0.25 0.09 0.09 0.13MgO 15.64 15.44 16.17 16.34 16.32 16.18 8.61 16.05 11.48 15.55 8.03 16.32 14.50 15.30CaO 23.68 23.18 24.23 21.73 22.06 23.41 20.23 23.77 21.50 24.05 20.13 24.40 23.82 22.84Na2O 0.72 0.61 0.51 0.64 0.49 0.45 2.45 0.48 1.72 0.34 2.13 0.30 0.38 0.33K2O 0.04 0.09 0.02 0.09 0.05 0.03 0.01 0.01 0.00 0.01 0.02 0.01 0.01 n.d.Total 100.25 99.82 100.87 99.84 99.35 99.02 100.53 100.48 100.05 99.71 99.97 100.40 100.77 98.82Si 1.893 1.906 1.866 1.946 1.997 1.976 1.939 1.900 1.973 1.909 1.973 1.921 1.790 1.921Ti 0.039 0.039 0.063 0.040 0.040 0.029 0.022 0.049 0.016 0.048 0.020 0.044 0.089 0.044Al 0.087 0.085 0.093 0.098 0.023 0.019 0.102 0.085 0.040 0.082 0.054 0.051 0.178 0.080Cr 0.000 0.003 0.005 0.002 0.001 0.002 0.000 0.006 0.000 0.001 0.000 0.000 0.003 0.014Fe 0.171 0.175 0.136 0.114 0.113 0.120 0.495 0.133 0.369 0.139 0.542 0.126 0.201 0.152Mn 0.003 0.003 0.003 0.002 0.004 0.002 0.012 0.002 0.008 0.000 0.008 0.003 0.003 0.004Mg 0.863 0.854 0.885 0.888 0.890 0.892 0.493 0.878 0.649 0.857 0.464 0.892 0.803 0.850Ca 0.940 0.922 0.953 0.849 0.865 0.927 0.833 0.934 0.874 0.953 0.837 0.959 0.949 0.913Na 0.052 0.044 0.036 0.045 0.035 0.032 0.183 0.034 0.126 0.024 0.160 0.021 0.027 0.024K 0.002 0.004 0.001 0.004 0.002 0.001 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000Total 4.051 4.036 4.041 3.989 3.969 4.001 4.080 4.023 4.055 4.014 4.060 4.020 4.044 4.000Mg# 0.83 0.83 0.87 0.89 0.89 0.88 0.50 0.87 0.64 0.86 0.46 0.88 0.80 0.85Wo 0.48 0.47 0.48 0.46 0.46 0.48 0.46 0.48 0.46 0.49 0.45 0.49 0.49 0.48En 0.44 0.44 0.45 0.48 0.48 0.46 0.27 0.45 0.34 0.44 0.25 0.45 0.41 0.44Fs 0.09 0.09 0.07 0.06 0.06 0.06 0.27 0.07 0.20 0.07 0.29 0.06 0.10 0.08

Mg# = Mg/(Mg + Fe⁎) in moles; c — core. r — rim of grain; i — clinopyroxene inclusion in clinopyroxene; mineral formulae were calculated for O = 6.a Data from Gurenko and Kononkova (1991).b All Fe as FeO.


with this clinopyroxene rim are characterized by Mg # ~0.69, substan-tially higher than the observed whole rock value (Mg#= 0.6).

The difference in degree of equilibrium between Pb, Sr and Ndisotope systems in clinopyroxene and whole rock can be attributedto the difference in the diffusion rates of these elements and the res-idence time of melts in the magma chamber (Jackson et al., 2009). Inmost geochemical environments the diffusion rate for Nd is muchlower than that for Sr or Pb (e.g. Albarede, 2003). The clinopyroxeneSr and Pb isotope ratios will thus experience diffusion re-equilibration with the host melt on the time scales that are typicalfor the “life” of a magma chamber, without the input of new batchesof isotopically distinct melts. The observed Sr and Pb isotope disequi-librium between clinopyroxene and corresponding whole rock thusindicates the rapid change in composition of the surrounding melt,and requires that the time interval between mixing of isotopicallydistinct magmas and subsequent ascent and eruption may be veryshort. A realistic explanation of local radiogenic isotope inhomoge-neity is the highmobility of Rb and U in comparison to REE in volatilerich fluids/melts released from or by the ascending plume. Such pro-cesses result in the formation of veined mantle that has been sug-gested to be the source for the deep-seated alkaline ultrabasicmagmas (Foley, 2008).

5.2. Clinopyroxene — mica relationship

Comparison of the isotopic characteristics of the clinopyroxeneand mica within the Visoke ugandite shows that the neodymium isoto-pic composition of pyroxene (143Nd/144Nd = 0.512535) and mica(143Nd/144Nd = 0.512527) phenocrysts is in equilibrium with the hostrocks (143Nd/144Nd= 0.512530), but the mica strontium isotopic com-position (87Sr/86Sr = 0.705703) differs appreciably from that of theclinopyroxene 87Sr/86Sr= 0.705033 (Table 7, Fig. 11). In this case,we as-sume that strontium isotopic composition of themica phenocrysts reflectthat of the host melt composition (whole rock, 87Sr/86Sr = 0.70565),

whereas the isotopic composition of the clinopyroxene reflects contribu-tions from another mantle source.

5.3. Single isotope source for different clinopyroxene

In contrast to the hydrousminerals, clinopyroxene phenocrysts in theeffusive rocks are resistant to processes of secondary alteration, and thusretain primary isotopic signatures. It is very likely that the clinopyroxenespreserve the isotopic signatures of the melts from which they wereformed. Sr isotopic data of clinopyroxene and mica (Fig. 10) documentthe degree of isotopic equilibrium of minerals with host rocks. The ar-rangement of clinopyroxene points allows us to draw another dottedline, which passes through the Sr isotopic data for all studiedclinopyroxenes. This observation that all clinopyroxenes are groupedalong the same line may be interpreted as evidence for a common (rela-tively depleted) source to which contributions from an enriched compo-nent are added in different proportions. Taken together, the mantlesource of the studied clinopyroxenes is a result of the interaction ofthese components. Extrapolation of the correlation based on the cpxdata suggests that the common mantle source contributing to all of thelavas has 87Sr/88Sr b0.7046. We presume that this source compo-nent, located beneath both volcanic areas and presumably close tothe boundary of the lithosphere and asthenosphere, was generatedduring an ancient event, long before the ultrapotassic magmatismof the Toro-Ankole and Virunga provinces. Lloyd et al. (1999) arguedfor a laterally variable clinopyroxenite layer in Uganda'sWestern Riftdeep mantle from which fragments occur as xenoliths in alkalinelavas. Such an alkali clinopyroxenite layer may be equated with thestockwork of metasomatized mantle that was assumed to exist be-neath continental rift zones where alkaline magmatism dominated(Lloyd and Bailey, 1975; Bailey, 1982). Edgar et al. (1980) inferredthe presence of a wehrlite or lherzolite source for K-enrichedugandite at depth ~90 km. This depth corresponds to the layer ofmagma generation that is defined by the mechanical and thermal

Table 6Clinopyroxene composition (selected representative analyzes).

11530 s11530 11530 11530 11642 11642 11642 11642 11642 11642 11641 11641 11641 11641 11641 11641

2r 2i 3 3i 1 2 3 4 5c 5r 1c 1r 2c 2r 3c 3r

47.75 50.35 51.41 51.82 51.53 47.03 44.53 51.35 50.20 50.53 46.42 44.68 48.26 45.98 48.52 46.613.35 0.63 1.81 0.44 1.37 3.20 4.67 1.27 1.49 1.52 4.13 5.46 3.31 4.60 3.12 3.303.77 3.65 1.88 2.49 2.38 4.52 6.29 2.19 2.84 2.66 4.95 6.38 3.74 5.45 3.24 2.210.24 0.04 0.42 0.07 0.71 0.04 0.00 1.00 0.17 0.30 0.00 0.00 0.01 0.00 0.02 0.006.42 13.72 5.17 13.04 3.92 7.09 7.38 3.71 4.30 4.42 7.02 7.74 7.04 7.01 6.58 10.900.07 0.42 0.06 0.33 0.02 0.14 0.15 0.04 0.03 0.09 0.18 0.14 0.11 0.11 0.15 0.23

13.47 9.15 15.08 12.44 17.01 13.76 12.72 17.28 16.59 16.55 12.83 11.68 13.22 12.68 13.60 11.2022.69 18.84 23.01 18.66 23.23 22.88 22.94 23.24 23.35 23.07 22.90 22.62 23.57 23.50 23.94 22.490.41 2.46 0.36 1.05 0.62 1.08 0.97 0.62 0.61 0.57 0.53 0.70 0.61 0.63 0.49 0.98n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 0.00 0.03 0.03 0.01 0.01 0.01

98.17 99.26 99.20 100.34 100.79 99.74 99.65 100.70 99.58 99.71 98.96 99.43 99.90 99.97 99.67 97.931.817 1.934 1.913 1.949 1.882 1.774 1.691 1.878 1.861 1.870 1.762 1.698 1.813 1.733 1.825 1.8240.096 0.018 0.051 0.012 0.038 0.091 0.133 0.035 0.042 0.042 0.118 0.156 0.094 0.130 0.088 0.0970.169 0.165 0.082 0.110 0.102 0.201 0.282 0.094 0.124 0.116 0.222 0.286 0.166 0.242 0.144 0.1020.007 0.001 0.012 0.002 0.020 0.001 0.000 0.029 0.005 0.009 0.000 0.000 0.000 0.000 0.001 0.0000.204 0.441 0.161 0.410 0.120 0.224 0.234 0.113 0.133 0.137 0.223 0.246 0.221 0.221 0.207 0.3570.002 0.014 0.002 0.011 0.001 0.004 0.005 0.001 0.001 0.003 0.006 0.005 0.004 0.004 0.005 0.0080.764 0.524 0.836 0.697 0.926 0.774 0.720 0.942 0.917 0.913 0.726 0.662 0.740 0.712 0.762 0.6530.925 0.775 0.918 0.752 0.909 0.925 0.934 0.911 0.928 0.915 0.932 0.921 0.949 0.949 0.965 0.9430.030 0.183 0.026 0.077 0.044 0.079 0.071 0.044 0.044 0.041 0.039 0.052 0.044 0.046 0.036 0.0740.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.001 0.001 0.000 0.000 0.0004.014 4.056 4.002 4.021 4.041 4.073 4.070 4.047 4.054 4.046 4.029 4.029 4.033 4.039 4.033 4.0650.79 0.54 0.84 0.63 0.89 0.78 0.75 0.89 0.87 0.87 0.77 0.73 0.77 0.76 0.79 0.650.49 0.45 0.48 0.40 0.47 0.48 0.49 0.46 0.47 0.47 0.50 0.50 0.50 0.50 0.50 0.480.40 0.30 0.44 0.37 0.47 0.40 0.38 0.48 0.46 0.46 0.39 0.36 0.39 0.38 0.39 0.330.11 0.25 0.08 0.22 0.06 0.12 0.12 0.06 0.07 0.07 0.12 0.13 0.12 0.12 0.11 0.18


boundary between lithosphere and asthenosphere (Rogers et al.,1998).

5.4. Pb isotope data

The Pb–Pb isotopic ratios for the lavas and whole rocks (Table 1)are similar to those of ocean island volcanics (modern OIB) derivedfrom an EM2 source (Zindler and Hart, 1986; Hofmann, 2003) andapproach those measured in primitive low-silica lavas from Virunga(Karisimbi and Muhavura — Rogers et al, 1992, 1998; Nyiragongoand Nyamuragira — Vollmer and Norry, 1983a, 1983b; Chakrabartiet al., 2009) and Katwe-Kikorongo (Davies and Lloyd, 1989). Rela-tively high lead isotope ratios for kamafugites agree well with theirTh/U values (4.16–5.01) which are high compared to the primitivemantle value of 3.9 (McDonough and Sun, 1995), but close to ~5.0mea-sured in Nyiragongo nephelinites (Vollmer and Norry, 1983a; b). Traceelement ratios involving Pb (Pb/Nd = 0.01–0.08, Ce/Pb = 34.51–45.85) indicate crustal contamination of kamafugite lavas in this studyis unlikely (average for MORB and OIB ~0.04, e.g., Hofmann al., 1986;Hofmann, 1988, 2003). Hofmann et al. (1986) demonstrated thatmany oceanic basalts (both MORB and OIB) have low Pb contents andhigh Ce/Pb values (27± 7), which contrast with the high Pb concentra-tions and low Ce/Pb ratios in rocks and melts derived from the conti-nental crust. The values obtained here for Toro-Ankole and Virungakamafugites (35–46) extend the trend of the Virunga PKB and K-basanites (Rogers et al, 1992, 1998) to higher Ce concentrations andfall within the range of mafic lavas derived from HIMU mantle sources(27–55; Chauvel et al., 1992). The exception is sample 11530 where alow Pb abundance may be due to its dissolution in lake water (Brenanet al., 1995; Ayers, 1998).

The usefulness of the parameter 208Pb*/206Pb* can be related to thehistory of Th/U fractionation. 208Pb*/206Pb* is a measure of the cumula-tive radiogenic Pb in a sample corrected for the presence of terrestrialprimordial Pb. For continental volcanics such as potassic basanitesfrom Karisimbi volcano in the East African Rift there is a positive corre-lation between 87Sr/86Sr ratios and 208Pb*/206Pb* (Rogers et al., 1992).The data we obtained for the Toro-Ankole and Virunga whole rocks fitwell this trend (Fig. 12a in Rogers et al., 1992) that is manifest on a

global scale (Nicolaysen et al., 2007). The observed correlation of 87Sr/86Sr with 208Pb*/206Pb*may be explained by a scenario inwhich the up-permost mantle has experienced melt enrichment/vein metasomatism(Nicolaysen et al., 2007).

Based on the Sr–Nd isotope whole rock data, the mantle source giv-ing rise to the kamafugite primary melt was of EM1-type composition.Pb isotope ratios are closer to an EM2 type model mantle source, butPb isotope signatures for clinopyroxenes are close to EM1 type source.The highly restricted range of Pb isotope ratios for the clinopyroxenesuggests that all of them have crystallized from a series of melts withsimilar isotopic composition. The clinopyroxenes have considerablylower lead isotopic ratios than those from Napak and Mount Elgon,which resemble modern OIB and even MORB. However, trend lines(Fig. 8) connecting clinopyroxenes and host rock compositions for oursamples are similar to those defined for the Napak and MountElgon nephelinite lavas, which are consistent with mixing betweenmelts derived from EM1-like and HIMU-like sources (Hofmann,2003; Stracke et al., 2005). A comparison of our data with Pb isotoperatio ranges of oceanic basalts, young carbonatites and kimberlites(Bell, 1998; Chakrabarti et al., 2009) shows that while whole rockPb isotope data are close to the fields of most African carbonatitesand volcanites, clinopyroxene data are close to fields for OIB (EMI)and Group II kimberlites, especially with regard to radiogenic Pb ra-tios (Fig. 11).

Our new Pb isotope data on studied clinopyroxenes and enclosinglavas correspond to prior results for Katwe-Kikorongo xenoliths andlavas (Davies and Lloyd, 1989). These data define a 207Pb/204Pb–206Pb/204Pb isotope array equivalent to an age of 1850 ± 160 My, althoughthe young Nd isotope model ages (b1000 My) of the xenoliths argueagainst the direct age significance of the Pb isotope array (Davies andLloyd, 1989). Zoned clinopyroxene macrocrysts within the Katwe-Kikorongo host magmas have Pb–Sr–Nd isotope ratios comparable tothose of the xenoliths, demonstrating that they are xenocrysts andthus must be derived from depths similar to or shallower than themagma source region.While thepresence of xenocrystswithin the sam-ples does affect thewhole rock isotopic composition to somedegree, thedata allow observation of general patterns of themagmatismhistory forthe area.

Fig. 4. The two types of core-to-rim zonation in clinopyroxene: a, b — normal and c, d — reverse zonation. (a) The phenocryst in leucite nephelinite (sample 11641) Visoke volcano, Vi-runga; (b) the same grain image in Fe-Kα emission; (c) large clinopyroxene inmafurite (11503)with clearly expressed reverse zonality; (d) the same grain image inMg-Kα emission. Thevarious nature of zoning indicates a difference in the primary melt composition evolution.


It is likely that some of clinopyroxenes studied in our work arexenocrysts. Isotopic analyses of pyroxenes studied here used mineralseparates from whole rocks; thus the corresponding isotope ratios rep-resent average weighted values. Obviously, this approach does not per-mit documentation of the isotopic heterogeneity within single grains asthe result of metasomatism as described for clinopyroxenes from man-tle xenoliths (e.g., Schmidberger et al., 2003). On the other hand, ourwork allows one to observe more general patterns of the magmatismhistory for the area. Davies and Lloyd (1989) suggested that the subcon-tinental lithosphere beneath Uganda records a regional enrichmentevent that extends for at least 200 km along the rift valley. They inter-pret the Pb–Pb array as indicating mixing of at least two components“A” and “B” with different values initial U/Pb values (μ1 and μ2). Daviesand Lloyd (1989) conclude that the most plausible explanation for thePb–Pb mixing arrays is that components A and В were derived fromthe convecting upper mantle, as small degree partial melts/fluids (rela-tively poor in H2O and rich in CO2, F and Cl) that became trapped in thesubcontinental lithosphere where the unradiogenic Nd isotope ratiosevolved.

If one takes the clinopyroxene isotope 206Pb/204Pb and 207Pb/204Pbdata at face value (i.e., without any correction of lead isotope compositionon possible growth due to in-situ uraniumdecay), and considers their po-sitions on the Stacey-Kramers (1975) lead-isotope growth curve, the ageof the source region for potential parentalmelts for clinopyroxenes can beestimated as about 300–400 Ma (Fig. 9). This interpretation is consistent

with the Pb–Pb isotope data obtained on xenolith minerals of Katwe-Kikorongo. Model calculations based on a depleted mantle compositiongive an age of 900 Ma if the Sm/Nd ratio of the melt is used, or 1100 Maif a higher Sm/Nd ratio for the source is assumed (Davies and Lloyd,1989) and these values also overlap the ~1 Ga model age for the sourceof the primitive K-basanites from Karisimbi (Rogers et al., 1992). TheNyiragongo source region, by contrast, has a younger age of enrichment,estimated to be ~500 Ma by Vollmer and Norry (1983a, 1983b) on thebasis of a Pb–Pb isochron age of 484 ± 28 Ma.

In our preferred model, the young (50 Ka) kamafugite magmas arederived from a sublithospheric mantle source (a deep-seated plume).During ascent to the surface they interacted with older and isotopicallymore depleted clinopyroxene-bearing veins. The initial melts for theclinopyroxenes were generated as the result of migration of volatile-rich melts/fluids into shallow mantle levels, where they reacted withsurrounding depleted peridotite. The resulting melts solidified in theformof veins or layers,whichwere stored in the stable uppermostman-tle for ~300–400 Ma and then remelted by heating from ascending as-thenospheric plume, resulting in the entrainment of clinopyroxenes.

5.5. Partial melting and fractional crystallization (effects on whole rockcompositions)

As noted above, the Toro Ankole volcanic rocks, in contrast to otherlavas from the southern part of the Western Branch (Virunga, Kivu,

image of Fig.�4

Fig. 5. The plots showing the relationship TiO2 vs. Mg# (a) and Na2O vs. Mg# (b) of investigated clinopyroxenes. In figure (a) and (b) two trends are clearly defined, that probably cor-respond to changes in temperature and pressure of crystallization; (c) and (d) — an example of the core (fill diamond) to rim (empty square) changing of clinopyroxene compositionin the mafurite (11503). Note that the zoning of clinopyroxene in this sample reflects changes in the composition of pyroxenes in all the studied rocks. Symbols— numbers in the figurecorrespond to the numbers of samples (the rock's name see in Table 1), filled stars — xenoliths and open stars — phenocrysts of clinopyroxene (from Lloyd, 1981).


Rungwe volcanic fields), are characterized by only minor effects of meltdifferentiation. Olivine is principal mineral involved in the melt differ-entiation of these rocks. As olivine fractionation cannot exert an

Fig. 6. Nd–Sr isotopic composition of ultrapotassic rocks of the East African Rift and theirphenocrysts of clinopyroxene and mica of the present study. Also shown are the domainsofMORB, OIB,mantle reservoirs, other Virunga volcanics (Vollmer and Norry, 1983a; Rog-ers et al., 1992, 1998), Kivu (Furman and Graham, 1999), Turkana (Furman et al., 2004)and Kenya rift (Rogers et al., 2000) volcanics of EARS located around the Tanzanian craton.Data for Toro-Ankole lavas of Davies and Lloyd (1989) andRosenthal et al. (2009) coincidewith our data and are shown in Fig. 6 as field delineated by the dashed line. Filled circle—Nyiragongo 2002 lava; open circle—Nyiragongo older lava (fromChakrabarti et al., 2009).

influence on the isotopic composition of a magma, we suggest thatwhole rock compositions closely reflect the composition of the primi-tive mantle magmas.

Weused the algorithmPRIMELT2.XLS (Herzberg andAsimow, 2008)to evaluate ourmodel ofmelting of garnet peridotite followed by the ol-ivine fractionation.We note that fertile peridotite was used in this algo-rithm and, therefore, the petrologic models represent a first orderapproximation. The composition of the melts source will be discussedin more detail in a subsequent article (Muravyeva et al., in prep.). Thealgorithm uses all major element concentrations in addition to Ni andCr contents to define the possible chemical nature of parental melts de-rived from garnet or spinel peridotite. The geochemical features of ourmafic samples (specifically, low Yb; Fig. 3) point to the presence of re-sidual garnet in their mantle source. Modeling results show that paren-tal melts for most kamafugites investigated could be formed duringmelting of garnet peridotite followed by the removal (for differentiates)or addition (for accumulates) of olivine.

The twomost Ca-rich samples (katungite and Bunyarugurumafurite11503) cannot be modeled as melts from peridotite source and there-fore additional componentsmust be included.We assume the existenceof pyroxenite (or eclogite) that has mixed with peridotite in some pro-portion. The presence of pyroxenite and/or eclogite is geologically rea-sonable, i.e., Lloyd et al. (1999) argued for a laterally variableclinopyroxenite layer beneath the Western Rift. More recent experi-ments have shown that partial melts of pyroxenite lithologies (or peri-dotite that has beenmetasomatized by pyroxenite partialmelts)mayberequired for highly SiO2 undersaturated melts (Dasgupta et al., 2007).Partial melts of carbonated peridotite with small amounts of CO2 (1

image of Fig.�5

image of Fig.�6

Table7

TheSr,N

d,Pb

isotop

iccompo

sition

ofmineralsfrom

ultra-po

tassium

volcan

icrocksof

theEa

stAfrican

Rift.

Sample

SmNd

147Sm

/144Nd

143Nd/

144Nd±

2σRb

Sr87Rb

/86Sr

87Sr/8

6Sr

±2σ

206Pb

/204Pb

±2σ

207Pb

/204Pb

±2σ

208Pb

/204Pb

±2σ

1149

7cp

x36

.12

121

0.18

096

0.51

2506

±3

19.15

421

0.13

146

0.70

4944

±12

18.157

4±

815

.595

5±

738

.190

1±

2411

503cp

x57

.25

198

0.17

504

0.51

2511

±4

17.21

324

0.15

352

0.70

4725

±10

18.170

5±

515

.584

0±

638

.230

8±

1911

523cp

x36

.39

145

0.15

173

0.51

2508

±4

16.54

366

0.13

066

0.70

4824

±8

18.115

1±

715

.597

1±

738

.212

2±

2411

641cp

x27

.91

116

0.14

491

0.51

2504

±6

24.44

426

0.16

614

0.70

4913

±7

19.425

7±

615

.729

3±

640

.574

1±

2311

530cp

x48

.59

161

0.18

194

0.51

2525

±6

22.51

523

0.12

453

0.70

4705

±6

18.037

5±

915

.588

3±

938

.089

9±

3011

642cp

x51

.76

136

0.23

061

0.51

2535

±4

18.68

382

0.14

134

0.70

5033

±8

18.048

1±

715

.577

2±

838

.109

7±

2311

642mica

2.48

24.35

0.06

150

0.51

2527

±12

132

96.62

1.28

371

0.70

5703

±18

n.d.

n.d.

n.d.

Note:

Allelem

entconten

tsarede

notedin

ppm;a

llun

certaintiesarequ

oted

at2σ

leve

land

referred

tothelast

decimal

digits;u

ncertainties

ofisotop

e14

7 Sm/1

44Ndan

d87Rb

/86 Srratios

determ

inationare0.3an

d0.4%

(2σ).


and 2.5 wt.%) at 3 GPamay yield liquids potentially parental for such al-kaline lavas. Dasgupta noted that eclogite partial melts may contributeto the petrogenesis of alkalic magmas, includingmetasomatism leadingto a TiO2- and FeO*-enriched carbonated peridotite. The most Ca-richsamples compositions (11503 and 11513) lie close to trend in binary di-agram “CaO–SiO2” obtained in experiments (Fig. 9 and 11 in Dasguptaet al., 2007).

5.6. Mantle metasomatism and the source heterogeneity

The topic of mantle source heterogeneity and its relation to alkalinesilicatemagmas in the East African Rift has been discussed by many au-thors (e.g. Lloyd and Baily., 1975; Lloyd et al., 1999; Eby et al., 2003;Furman, 1995; Furman and Graham, 1999; Bell and Tilton, 2001;Ivanov et al., 1998; Rogers et al., 1992, 1998). The core of the debateconcerns the contributions of different proportions of asthenosphericand lithosphericmantle sources to the origin of the primitivemelts. Fur-ther questions focus on the style and age of the metasomatism and itsrelationships with magmatism.

There is substantial evidence for mantle metasomatism in the min-erals from Toro-Ankole kamafugites studied in our work. There are nu-merous melt, fluid and crystalline inclusions in olivine andclinopyroxene phenocrysts, some of which contain glass inclusionsthat resemble results of secondary melting of mantle xenoliths. Thepresence of phlogopite and carbonates as megacrysts and inclusionsin olivine clearly indicates modification of the source compositionprior to kamafugiemelt formation, i.e., an episode ofmantlemetasoma-tism. The Cr2O3 contents (0.02–0.36 wt.%) in clinopyroxenes of Toro-Ankole kamafugites are much lower than for primary diopsides fromthemantle xenoliths and similar to secondarymetasomatic xenolith di-opsides (Dawson and Smith, 1988; Dawson, 2002), further supportingthe idea that some metasomatic event had taken place in their origin.

Anomalously high abundances of LREE (La/Ybnup to 143), Nb (up to276 ppm), Zr (up to 452 ppm), and Sr (up to 2754 ppm) also point tointensive mantle metasomatism in source of Toro-Ankole kamafugites.The enrichment agents may be fluids (CO2, H2O, F) and small degreemantle melts of either silicate or carbonatite composition (e.g.Thibault et al., 1992; Furman, 1995; Furman and Graham, 1999; Lloyd.et al., 1999; Eby et al.., 2003; Rosenthal et al., 2009). The mantle sourceof Toro-Ankole lavas has likely experienced several stages and styles ofmantlemetasomatism. Davies and Lloyd (1989) argued for silicatemeltmetasomatism, while more recent authors point to the effects of bothsilicate and carbonate low degree melts (e.g. Thibault et al., 1992; Ebyet al., 2003; Rosenthal et al., 2009). Rosenthal et al. (2009) used Os–Sr–Nd–Hf isotopic data to suggest a complex model in which thekamafugite source contains two metasomes — one that is carbonatiteand the other resembling MARID inclusions in kimberlites. They inter-pret the range of 87Sr/86Sr values (0.704599–0.705402) as related to acarbonate-rich metasome, which also imparts a Nd and Hf signaturesimilar to convecting upper mantle (i.e. asthenosphere). Thus, the pos-sible sequence of source enrichment (metasomatism) beneath theToro-Ankole field that is required to produce the variation in Nd, Srand Hf isotopes is potassic alkaline silicate metasomatism in the firststage, which is then followed by carbonate-rich metasomatism.

Lateral mantle heterogeneity that records different styles of mantleenrichment can be observed in the Bunyaruguru volcanic field. Themafurites from two neighboring volcanic craters Kyambu (sample11494) and Nyungu (sample 11503), located at a distance of ~1 kmapart (Fig. 1), differ in their major element chemistry and strontium iso-topic composition (87Sr/86Sr— 0.70519 and 0.70463, respectively). Thesebulk rocks also have different values of key geochemical indicators: Ba/Rb— 11.96 and 20.614, Ba/Nb— 12.60 and 8.60, Rb/Sr— 0.11 and 0.04, La/Ta— 9.01 and 13.39, Ni/Co— 9.42 and 1.81, and Zr/Y— 17.49 and 26.09 forsamples 11494 and 11503 respectively. There are also differences in themineral composition of phenocrysts and xenoliths in these lavas: sample11503 is dominated by clinopyroxene phenocrysts whereas sample

Fig. 7. (a) Variants of Sr–Nd isotopic relationships clinopyroxeneswith host rocks, expressedas the difference between isotopic data for mineral and whole rock. The isotopic differenceΔ143Nd/144Nd and Δ87Sr/86Sr between whole rock (Wr) and clinopyroxene (Cpx) isexpressed in parts permillion (ppm). The size of symbols in Fig. 8a exceeds themeasurementerrors of the relevant isotope ratios. The symbols denote: black diamond — present workstudied ultrapotassic rocks Toro-Ankole and Virunga. The open symbols denote the pub-lished data: triangle — kamafugites Katwe-Kikorongo (Davies and Lloyd, 1989); square —

nephelinite lavas from Napak volcano (Simonetti and Bell, 1993); circle — nephelinite lavasfrom the Mount Elgon volcano (Simonetti and Bell, 1995); tie-lines connect the values forthe core and the outer zone of the phenocrysts; (b) Δ143Nd/144Nd and Δ87Sr/86Sr versusNa2O content in clinopyroxene (average composition for each host rock).

Fig. 8. Pb isotope composition of host rock (open diamond) and clinopyroxene (filled di-amond) studied, shown on plots 208Pb/204Pb vs. 206Pb/204Pb (a) and 207Pb/204Pb vs. 206Pb/204Pb (b) together with other data for African rift: kamafugites (open triangle) withclinopyroxene phenocrysts (filled triangle) from Katwe-Kikorongo (Davies and Lloyd,1989), open circle — nephelinite lavas and filled circle — diopside phenocrysts fromcarbonatite–nephelinite centers Mount Elgon (Simonetti and Bell, 1995) and Napak(shaded area enclosed by dashed line) (Simonetti and Bell, 1993). HIMU and EMI mantlecomponents are from Hart (1988). S/K — Stacey–Kramers growth curve (Stacey andKramers, 1975). Open squares along S/K curve are at 200 Ma intervals. NHRL— NorthernHemisphere Reference Line (Hart, 1984).


11494 (Murav'eva and Senin, 2009) contains phlogopite in olivinemacrocrysts. Carbonates are found in ugandite (sample 11523) andmafurites (samples 11494 and 11503), but they differ in type and

quantity: sample 11494 contains inclusions in olivine microxenolithswhile sample 11503 has discrete crystals in the groundmass. In uganditesample 11523 rare carbonateswere found only in the crystallizedmelt in-clusions in olivine phenocrysts (Fig. 4a in Murav'eva and Senin, 2009),which estimated bulk compositions are close to those for carbonatites.The petrographic and geochemical evidence indicates the prevalence ofcarbonate-rich metasomes in the source region of the Nyungu craterlavas, but of both phlogopite- and carbonate-rich metasomes in theKyambu crater lavas source. As noted above, the major element modelof garnet peridotite melting has no solution for sample 11503. On the ac-cumulatedweight of the above arguments,we propose that a phlogopite-carbonate-bearing peridotite source melted to form Kyambu lava 11494and a predominantly pyroxenitic source modified by carbonate-containing fluid/melt produced Nyungu lava 11503. The latter assump-tion is in good agreement with the work of Lloyd et al. (1999) andRosenthal et al. (2009).

On a larger scale the lateral heterogeneity is observed via correlationsbetween the strontium isotopic ratios and Rb/Sr ratios (Fig. 12). The sam-ples define three groups differing in the degree of radiogenic 87Sr enrich-ment: the clinopyroxenesdefine a localizedfieldwhichoverlapswith twowhole rock values, the most Sr-depleted and CaO-enriched samples:

image of Fig.�7

image of Fig.�8

Fig. 9. 143Nd/144Nd vs. 206Pb/204Pb (a) and 87Sr/86Sr vs. 206Pb/204Pb (b), HIMU, EM1andEMII mantle components are from Hart (1988). Symbols: fill diamond — studied in presentwork clinopyroxenes from Toro-Ankole and Virunda lavas; open diamond — whole rockToro-Ankole and Virunga lavas; dots— diopside phenocrysts fromMount Elgon nephelin-ites; circles — whole rock Mount Elgon nephelinites (Simonetti and Bell, 1995); filled tri-angle — clinopyroxenes Katwe-Kikorongo; open triangle — host rock Katwe-Kikorongo(Davies and Lloyd, 1989).

Fig. 10. Sr isotopic disequilibrium: 87Sr/86Sr ratio in minerals and host rocks for the sam-ples studied. The solid line corresponds to the isotopic equilibrium betweenminerals andhost rocks. The dashed line passes through the Sr isotopic compositions of the studiedclinopyroxenes. The left end of the line indicates the composition of the proposed com-mon source for melts from which all clinopyroxenes have been formed. This plot showsthat mica is in isotopic equilibrium with host rock, while the most of clinopyroxene areout of equilibrium with the host rock. Symbols: open square — clinopyroxene, filledsquare — mica. Numbers in the figure correspond to the numbers of samples (the rock'sname see in Table 1).

Fig. 11. 208Pb/206Pb versus 207Pb/206Pb plot of the present study samples for Toro-Ankoleand Virunga (whole rocks — open square and triangle, clinopyroxene — black diamond)and data from Chakrabarti et al. (2009). The mantle reservoirs EM I, EM II, DMM andHIMU are also plotted. The published data are shown for Nyiragongo (grey and black cir-cles) and Nyamuragira (black squere) lavas from Virunga and the fields of Group I andGroup II kimberlites (Chakrabarti et al., 2009).


mafurite 11503 and katungite 11513. One may suggest, therefore, thatrocks with the least radiogenic Sr isotope compositions were producedfrom a pyroxenite source that was metasomatized later by carbonatitemelt-fluid. The high strontium content in the sample and the presenceof large carbonates in the groundmass support this model. In addition,oxygen fugacity estimates for the Toro-Ankole kamafugite are typicalof those for metasomatized mantle: fO2 excess buffer NNO (ΔQFM is0.3–3) over a temperature range of 1230–750 °C (Muravyeva andSenin, 2008; Murav'eva and Senin, 2009).

The depth of the mantle heterogeneity for East Africa (chemical andisotopic) has been discussed previously (e.g. Furman, 1995; Furman andGraham, 1999; Rogers et al., 1992, 1998). The depths from which allmagmas in the Western Rift have been derived have been discussedpreviously (Furman, 1995; Rosenthal et al., 2009). The Virunga K-basanites were inferred to have been derived from similar depths tothe Rungwe basanites and the Kivu basalts, whereas the potassic lavasfrom Katwe-Kikorongo, and the nephelinites from Rungwe and Nyira-gongo show a greater involvement of garnet in their source regions.

These relative depth constraints can be further refined by consider-ing the minor phases involved in the generation of the differentmagmas. For example, the development of Nyiragongo nephelinites

Fig. 12. The relationships between strontium isotopic composition (87Sr/86Sr) and rubidi-um–strontium ratios (87Rb/87Sr) in studied samples of the East African rift. Symbols: opendiamond — Toro-Ankole kamafugites, filled diamond — clinopyroxene phenocrysts fromToro Ankole kamafugites, open circle — foidites of Visoke volcano, filled circle —

clinopyroxenes from foidites of Visoke volcano (Virunga province).

image of Fig.�9

image of Fig.�10

image of Fig.�11

image of Fig.�12

Fig. 13. Sr–Nd isotope ratios of studied kamafugite clinopyroxene (filled diamond) andwhole rocks (open diamond) in comparison with data on clinopyroxenes from Mt. Elgonand Napak nephelinite and carbonatite lavas. EACL from Bell and Blenkinsop (1987). Sym-bols: dots — diopside phenocrysts from Mount Elgon nephelinites; circles — whole rockMount Elgon nephelinites. Tie-lines join diopside phenocryst with corresponding wholerock analysis. Isotopic data fromNapak whole rock nephelinites (open square) and diopsidephenocrysts (filled square) shown in area enclosed by dashed line. Open star — Napakcarbonatite.


requires the presence of carbonate in the magma source region (Rogerset al., 1998). Carbonate is only stable within themantle at depths great-er than ~100 km (Rogers et al., 1998; Olafsson and Eggler, 1983), consis-tent with the requirement for residual garnet in the magma sourceregion. Toro-Ankole kamafugites were derived from the deepest partsof the mantle in the Western Branch of the Rift. Supporting evidencefor this statement includes the presence of trace diamonds in a xenolithfrom Toro-Ankole lavas (P. Nixon,1973 - cited by Rosental et al., 2009),and the presence of Fo90–92 olivine phenocrysts with dolomite inclu-sions (Murav'eva and Senin., 2009). The classic scenario of melt mixingfrom various depths is confirmed by our data, which indicates isotopicdisequilibrium between the clinopyroxenes and their correspondinghost rock, as well as by the chemical zonation of clinopyroxene andlarge differences in Pb isotope compositions.

5.7. Mixing of mantle melts

Petrographic features of clinopyroxene and spinel phenocrysts sug-gest that kamafugite lavas formed by mixing at least two parentalmelts. This scenario does not preclude the entrapment of disintegratedmantle xenoliths, the proof of which we found in the morphology ofsome phenocrysts. Such phenocrysts contain veins of glass, indicatingsecondary melting. However, the complex nature of the zoning of indi-vidual clinopyroxene grains clearly favors capture of clinopyroxenes bya later melt (Fig. 4c and d), which is followed by typical melt evolutioninvolving chemical changes resulting from crystallization during de-creasing temperature and pressure conditions.

Olivine with up to Fo92 and high-Cr# spinel inclusions and pheno-crysts indicate a refractory source for some kamafugite melts (e.g. sam-ple 11494). At the same time the cores of zoned clinopyroxenes inmafurite have much lower Mg# (~0.55), and some clinopyroxenemacrocrysts (xenocrysts?) and inclusions in olivine from this lavahave Mg# ~76. The existence of minerals with such different character-istics in a single lava may indicate mixing of kamafugite magma withsome melts from shallower levels.

The isotopic results from this study echo the findings of prior studiesat Toro Ankole (Davies and Lloyd, 1989), but are distinct relative tothose from Napak and Mt. Elgon. Note that all of the isotope data forthe clinopyroxenes investigated here plot at the high-Sr/low-Nd endof the East African Carbonatite Line (EACL) (Fig. 13). This Nd–Sr isotopicarray is definedmainly by the isotope characteristics of young (b30Ma)

carbonatite complexes from Kenya, Tanzania and Uganda (Bell andBlenkinsop, 1987), and represents a mixing trend of melts generatedfrom EM1 and HIMU mantle components, which are also identified forOIB magmatism (Bell and Blenkinsop, 1987; Bell and Tilton, 2001).Using Nd, Pb and Sr isotopic constraints, Bell (1998) demonstrated theneed for open-system behavior that involvesmixing either amongmul-tiple mantle melts or sources, or of mantle-derived melts with lowercontinental crust. The location of our data near the radiogenic end ofthe EACL trend suggests a close genetic relationship betweenkamafugite and carbonatite melts in the Toro-Ankole that was identi-fied previously (e.g. Bell, 1998; Bell and Tilton, 2001; Bailey et al.,2005; Eby et al., 2009; Murav'eva and Senin, 2009).

There are currently different views on the location of these enrichedsources for African Rift: subcontinental lithosphere, asthenosphere orheterogeneous mantle plume. When investigating African carbonatitelavas, Kalt et al. (1997) suggested that these sources were located in aheterogeneous lithospheric mantle and were produced by enrichmentand depletion processes at different times and degrees. Their argumentagainst a simplemixingmodelwas the lack of straight lines in Sr–Pb andNd–Pb isotope space for carbonatites. In contrast we see a linear corre-lation in both whole rock and clinopyroxene Sr–Nd isotopic data(Fig. 9).

More recently, Bell and Simonetti (2010) argued that carbonatitesare not derived from a lithospheric source but rather from the astheno-sphere or a heterogeneous mantle plume. Their constraints imposed byboth radiogenic and stable isotopic data from carbonatites world-wideare consistent with a sub-lithospheric source for the parental melts, as-sociated with either asthenospheric upwellings or more deep-seated,plume-related activity. This latter model (a heterogeneous plume) isdifficult to determine uniquely on basis of our results; our preferredmodel is that the kamafugites and related lavas document mixing ofmelts from two mantle sources, probably asthenospheric (plume) andlithospheric mantle. Taking into account the large difference of Pb iso-tope ratios of clinopyroxene and their host rocks, it seems reasonableto assume contributions from the convecting upper mantle (Daviesand Lloyd, 1989) or a heterogeneous plume. In any case, the mantlesource for the clinopyroxenes-bearing melts is most probably locatedon the same level or shallower than the source for parental kamafugitemelts.

6. Conclusions

The results obtained in our study show significant isotope disequi-librium between clinopyroxene and the host alkaline primitive melts,as recorded previously in East Africa by Simonetti and Bell (1993,1995) andDavies and Lloyd (1989). The nature and degree of Sr isotopicdisequilibrium are correlated with clinopyroxene compositional zon-ing; isotopic and elemental composition disequilibria indicate a high-pressure trend (primarily in high Na2O) and correspond to isotope-enrichment relative to the host rock clinopyroxene, whereas a low-pressure evolution trend is correlated to isotopically depletedclinopyroxene compositions.

We infer the existence of a common source for the parental meltsfrom which all of the clinopyroxenes crystallized. The isotopic variationsof Sr and Pb in clinopyroxenes are confined to a limited range, regardlessof occurrence over large area. Such a sourcemay be earliermagma cham-bers or veins within the lower lithosphere, generated during an ancientmetasomatic or magmatic event long before the modern ultrapotassicmagmatism.Melts in these locations, where clinopyroxene crystallizationtook place, have Pb–Sr–Nd isotopic signatures similar to those of EM1mantle. Later kamafugite magmas, which differ in the isotopic composi-tion of strontium and lead, formed during ascent as mixing betweenmelts entrained the previously formed clinopyroxenes. The rate ofmixingwas faster thanneeded to establish the isotopic and chemical equilibrium.

We interpret the ultrapotassic rocks studied here as recording themixing of melts derived from two different sources (lithosphere and

image of Fig.�13


asthenosphere), which took place during rapid ascent of the magmas.The origin of the primary melts themselves is associated with meltingof a heterogeneous (kilometer-scale)mantle source in which peridotiteand pyroxenite were present in varying proportions. The variable na-ture of the Rb–Sr, Sm–Nd and U–Th–Pb isotope systems is likely indica-tive of the complex processes involved in the genesis of the kamafugitemagmas. These processes include enrichment (metasomatic) events bysilicatemelts and H2O–F–CO2 fluids, convecting asthenosphere, ascend-ing plume and magma mixing.

Acknowledgements

Samples studied in this paper were collected by V.I. Gerasimovskyand A.I. Polyakov during the Upper Mantle International GeophysicalProject in 1968–1969. We thank I.A. Roshchina (Vernadsky Institute ofGeochemistry, Russia) for help with the XRF — analyses our samples.We thank Roger Mitchell and Tanya Furman for spelling English in ourarticle. We are also very grateful to Nelson Eby (Editor in Chief), TanyaFurman and Antonio Simonetti for careful reading of the article andvery helpful constructive scientific and editorial comments. We arealso very thankful to reviewers Felicity Lloyd and Wendy Nelson forcareful reading of the article and detailed comments.

References

Albarede, F., 2003. Geochemistry: An Introduction. Cambridge University Press, UK(248 pp.).

Ayers, J., 1998. Trace element modeling of aqueous fluid–peridotite interaction in themantle wedge of subduction zones. Contribution to Mineralogy and Petrology 132,390–404.

Bailey, D.K., 1974. Continental rifting and alkaline magmatism. In: Sorensen, H. (Ed.), Thealkaline rock. John. Wiley and Sons, London – New York – Sidney – Toronto,pp. 148–159.

Bailey, D.K., 1982. Mantle metasomatism— continuing chemical change within the Earth.Nature 296, 525–530.

Bailey, K., Lloyd, F., Kearns, S., Stoppa, F., Eby, N.,Woolley, A., 2005. Melilitite at Fort Portal,Uganda: another dimension to the carbonate volcanism. Lithos 85, 15–25.

Barton, M., Varekamp, J.C., Bergen, M.J., 1982. Complex zoning of clinopyroxenes in thelavas of Vulsini, Latium, Italy: evidence for magma mixing. Journal of VolcanologyGeothermical Researches 14, 361–388.

Bell, K., 1998. Radiogenic Isotope Constraints on Relationships between Carbonatites andAssociated Silicate Rocks—a Brief Review. Journal of Petrology 39, 1987–1996.

Bell, K., Blenkinsop, J., 1987. Nd and Sr isotopic compositions of East African carbonatites:implications for mantle heterogeneity. Geology 15, 99–102.

Bell, K., Simonetti, A., 2010. Source of parental melts to carbonatites—critical isotopic con-straints. Mineralogy and Petrolology 98, 77–89.

Bell, K., Tilton, G.R., 2001. Nd, Pb and Sr isotopic compositions of east African carbonatites:evidence for mantle mixing and plume inhomogeneity. Journal of Petrology 42,1927–1945.

Blundy, J.D., Falloon, T.J., Wood, B.J., Dalton, J.A., 1995. Sodium partitioning betweenclinopyroxene and silicate melts. Journal of Geophysical Researches 100,15501–15515.

Boven, A., Pasteels, P., Punzalan, L.E., Yamba, T.K., Musisi, J.H., 1998. Quaternaryperpotassic magmatism in Uganda (Toro-Ankole Volcanic Province): age assessmentand significance for magmatic evolution along the East African Rift. Journal of AfricanEarth Science 26, 463–476.

Brenan, J.M., Shaw, I.H.F., Ryerson, F.J., Phinney, D.L., 1995. Mineral-aqueous fluidpartitioning of trace elements at 900 °C and 2.0 GPa: constraints on the trace elementchemistry of mantle and deep crustal fluids. Geochimica el Cosmochimica Acta 59,3331–3350.

Bryce, J.G., DePaolo, D.J., 2004. Pb isotopic heterogeneity in basaltic phenocrysts.Geochimica et Cosmochimica Acta 68, 4453–4468.

Chakrabarti, R., Basu, A.R., Santo, A.P., Tedesco, D., Vaselli, O., 2009. Isotopic and geochem-ical evidence for a heterogeneousmantle plume origin of theVirunga volcanics,West-ern rift, East African Rift system. Chemical Geology 259, 273–289.

Chauvel, C., Hofmann, A.W., Vidal, P., 1992. HIMU-EM: the French Polynesian connection.Earth and Planetary Science Letters 110, 99–119.

Combe, A.D., 1930. Recent volcanic area of Bunyaruguru. Annual Reports Geological SurztUganda 1929, pp. 16–18.

Combe, A.D., 1937. The Katunga volcano, southwest Uganda. Geological Magazine 74,190–200.

Dasgupta, R., Hirschmann, M.M., Smith, N.D., 2007. Partial melting experiments on peri-dotite + CO2 at 3 GPa and genesis of alkalic ocean island basalts. Journal of Petrology48, 2093–2124.

Davies, G.R., Lloyd, F.E., 1989. Pb–Sr–Nd isotope and trace element data bearing on theorigin of the potassic subcontinental lithosphere beneath south-west. Uganda:Proceeding of the 4th International Kimberlite Conference. 2, pp. 784–794.

Dawson, J.B., 2002. Metasomatism and partial melting in upper-mantle peridotite xenolithsfrom the Lashaine volcano, Northern Tanzania. Journal of Petrology 43, 1749–1777.

Dawson, J.B., Smith, J.V., 1988. Metasomatised and veined upper-mantle xenoliths from PelloHill, Tanzania: evidence for anomalously-light mantle beneath the of the East AfricanRift Valley Tanzanian sector. Contributions Mineralogy and Petrology 100, 510–527.

De Mulder, M., 1985. The Karisimbi Volcano, (Virunga). Musee Royale de l'AfriqueCentrale, Tervuren, Belgique Annales, Serie in Octavo Sciences Geologique. p. 90.

Duda, A., Schmincke, H.-U., 1985. Polybaric differentiation of alkali basaltic magmas:evidence from green-core clinopyroxenes (Eifel, FRG). Contribution to Mineralogyand Petrology 91, 340–353.

Ebinger, C.J., 1989. Tectonic development of the western branch of the East African riftsystem. Geological Society of America Bulletin 102, 885–903.

Ebinger, C., Furman, T., 2002. Geogynamical setting of the Virunga volcanic province. ActaVulcanologica 14 (15), 9–16.

Eby, G.N., Lloyd, F.E., Woolley, A.R., Stoppa, F., Weaver, S.D., 2003. Geochemistry and man-tle source(s) for carbonatitic and potassic lavas, Western Branch of the East-AfricanRift System, SW Uganda. Geolines 15, 23–27.

Eby, G.N., Lloyd, F.E., Woolley, A.R., 2009. Geochemistry and petrogenesis of the FortPortal, Uganda, extrusive carbonatite. Lithos 113, 785–800.

Edgar, A.D., Condliffe, E., Barnett, R.L., Shirran, R.J., 1980. An experimental study of anolivine ugandite magma and mechanisms for the formation of its K-enriched deriva-tives. Journal of Petrology 21, 475–497.

Foley, S.F., 2008. Rejuvenation and erosion of the cratonic lithosphere. Nature Geoscience1, 503–510.

Foley, S.F., Venturelli, G., Green, D.H., Toscani, L., 1987. The ultrapotassic rocks: character-istics, classification, and constraints for petrogenetic models. Earth Science Review24, 81–134.

Furman, T., 1995. Melting of metasomatized subcontinental lithosphere: undersaturatedmafic lavas from Rungwe, Tanzania. Contribution to Mineralogy and Petrology 122,97–115.

Furman, T., Graham, D., 1999. Erosion of lithospheric mantle beneath the East African Riftsystem: geochemical evidence from the Kivu volcanic province. Lithos 48, 237–262.

Furman, T., Bryce, J.G., Karson, J., Iotti, A., 2004. East African Rift System (EARS) plumestructure: insights from Quaternary mafic lavas of Turkana, Kenya. Journal of Petrol-ogy 45, 1069–1088.

Gerasimovskiy, V.I., Polyakov, A.I., 1974. Geochemistry of volcanic rocks of East African riftzones. In: Beloussov, V.V., Gerasimovskiy, V.I., Rykunov, L.N. (Eds.), East-African RiftSystem. Nauka Publishing House, pp. 5–195.

Gurenko, A.A., Kononkova, N.N., 1991. Zoning ofminerals as an indicator of changes in theredox conditions of crystallization (for example of high-Kmagmas of the East AfricanRift): Reports of the Academy of Sciences of the USSR. 318, 181–186 (In Russian).

Hart, S.R., 1984. A large scale isotope anomaly in the Southern Hemisphere mantle.Nature 309, 753–757.

Hart, S.R., 1988. Heterogeneous mantle domains: signatures, genesis and mixing chronol-ogies. Earth and Planetary Science Letters 90, 273–296.

Herzberg, C., Asimow, P.D., 2008. Petrology of some oceanic island basalts: PRIMELT2.XLSsoftware for primary magma calculation. Geochemistry, Geophysics, Geosystems 9,1–25.

Hofmann, A.W., 1988. Chemical differentiation of the Earth: the relationship between man-tle, continental crust, and oceanic crust. Earth and Planetary Science Letters 90,297–314.

Hofmann, A.W., 2003. Sampling mantle heterogeneity through oceanic basalts: isotopesand trace elements. Treatise on Geochemistry 2, 61–101.

Hofmann, A.W., Jochum, K.-P., Seufert, M., White, W.M., 1986. Nb and Pb in oceanic basalts:new constraints on mantle evolution. Earth and Planetary Science Letters 79, 33–45.

Holmes, A., 1942. A suite of volcanic rocks from south-west Uganda containing kalsilite (apolymorph of KAlSi04). Mineralogical Magazine XXVI, 197–217.

Holmes, A., 1950. Petrogenesis of katungite and its associated. American Mineralogist 35,772–792.

Ivanov, A.V., Rasskazov, S.V., Boven, A., André, L., Maslovskaya, M.N., Temu, E.B., 1998. LateCenozoic alkaline-ultramafic and alkaline basanitic magmatism of the Rungwe prov-ince, Tanzania. Petrology 6, 208–229.

Ivanov, A.V., Rasskazov, S.V., Chebykin, E.P., Markova, M.E., Saranina, E.V., 2000. Y/Ho ra-tios in the Late Cenozoic basalts from the eastern Tuva, Russia: an ICP-MS study withenhanced data quality; Geostandards Newsletters. Journal of Geostandards andGeoanalyses 24, 197–204.

Jackson, M.G., Hart, S.R., Shimizu, N., Blusztajn, J.S., 2009. The 87Sr/86Sr and 143Nd/144Nddisequilibrium between Polynesian hot spot lavas and the clinopyroxene they host:evidence complementing isotopic disequilibrium in melt inclusions. Geochemistry,Geophysics, Geosystems 10, N3.

Kalt, A., Henger, E., Satir, M., 1997. Nd, Pb and Sr isotopic evidence for diverse lithosphericmantle sources of East African Rift carbonatites. Tectonophysics 278, 31–45.

Lærdal, T., Talbot, M.R., 2002. Basin neotectonics of Lakes Edward and George, East AfricanRift. Palaeogeography, Palaeoclimatology, Palaeoecology 187, 213–232.

Lloyd, F.E., 1981. Upper-mantle metasomatism beneath a continental rift: clinopyroxenesin alkaline mafic lavas and nodules from South-West Uganda. MineralogicalMagazine 44, 315–323.

Lloyd, F.E., Bailey, D.K., 1975. Light element metasomatism of the continental man-tle: the evidence and the consequences. Physics and Chemistry of the Earth 9,389–416.

Lloyd, F.E., Arima, M., Edgar, A.D., 1985. Partial melting of a phlogopite-clinopyroxenitenodule from south-west Uganda: an experimental study bearing on the origin ofhighly potassic continental rift volcanics. Contribution to Mineralogy and Petrology91, 321–329.

Lloyd, F.E., Woolley, A.R., Stoppa, F., Eby, G.N., 1999. Rift Valley magmatism — is thereevidence for laterally variable alkali clinopyroxenite mantle? In: Ulrych, J., Cajz, V.,

http://refhub.elsevier.com/S0024-4937(14)00345-4/rf0005
















































































































































Adamovic, J. (Eds.), Magmatism and Rift Basin EvolutionGeolines 9. IGCP Liblice,Czech Republic 1998, pp. 76–83.

Logatchev, N.A., Beloussov, V.V., Milankovsky, E.E., 1972. East African rift development.Tectonophysics 15, 71–81.

McDonough, W.F., Sun, S.-S., 1995. The composition of the Earth. Chemical Geology 120,223–253.

Mercier, J.-C.C., 1980. Single-pyroxene thermobarometry. Tectonophysics 70, 1–37.Murav'eva, N.S., Senin, V.G., 2009. Carbonate–silicate equilibria in the high-magnesia

ultrapotassic volcanics of the Toro-Ankole province (Eastern African Rift zone). Geo-chemistry International 47, 882–900.

Muravyeva, N.S., Belyatsky, B.V., 2009. Petrology and geochemistry Toro Ankolekamafugite magmas: isotopic constraints. Geophysical Research Abstracts 11(EGU2009–12651–1, EGU General, Assembly 2009).

Muravyeva, N.S., Senin, V.G., 2008. Kamafugites of the Eastern-African rift zone: redoxstates of crystallization. Electronic Scientific Information Journal “Bulletin of theDepartment of Earth Sciences RUS” ( 1, http://www.scgis.ru/rissian/cp1251/h_dgggms/1-2008/magm-23e.pdf).

Nicolaysen, K.P., Frey, F.A., Mahoney, J.J., M. Johnson, K.T., 2007. Influence of theAmsterdam/St. Paul hot spot along the Southeast Indian Ridge between 77° and88° E: correlations of Sr, Nd, Pb, and He isotopic variations with ridge segmentation.Geochemistry, Geophysics, Geosystems 8, N9.

Nixon, P., 1973. Kimberlitic volcanoes in East Africa. Oversea Geol. Mineral. Resources.Inst. Geol. Sci. 41, 119–138.

Nyablade, A.A., Brazier, R.A., 2002. Precambrian lithospheric controls on the developmentof the East African rift system. Geology 30, 755–758.

Olafsson, M., Eggler, D.H., 1983. Phase relations of amphibole, amphibole–carbonate andphlogopite–carbonate peridotite: petrologic constraints on the asthenosphere.Earth and Planetary Science Letters 64, 305–315.

Pertermann, M., Hirschmann, M.M., 2003. Anhydrous partial melting experimentson MORB-like eclogite: phase relations, phase compositions and mineral-meltpartitioning of major elements at 2–3 GPa. Journal of Petrology 44, 2173–2201.

Pilet, S., Hernandez, J., Villemant, B., 2002. Evidence for high silicic melt circulation andmetasomatic events in the mantle beneath alkaline provinces: the Na–Fe–augiticgreen-core pyroxenes in the tertiary alkali basalts of the Cantal massif (FrenchMassifCentral). Mineralogy and Petrology 76, 39–62.

Pilet, S., Hernandez, J., Bussy, F., Sylvester, P.J., 2004. Short-term metasomatic control ofNb/Th ratios in the mantle sources of intraplate basalts. Geology 32, 113–116.

Pin, C., Joannon, S., Bosq, Ch., Le Fèvre, B., Gauthier, P.J., 2003. Precise determination of Rb,Sr, Ba, and Pb in geological materials by isotope dilution and ICP-quadrupole massspectrometry following separation of the analytes. Journal of Analytical Atomic Spec-trometry 18, 135–141.

Pouchou, J.-L., Pichoir, F., 1984. A new model for quantitative X-ray microanalysis. Part I:application to the analysis of the homogeneous samples. Recherche Aerospatiale 3,13–38.

Prytulak, J., Elliott, T., 2007. TiO2 enrichment in ocean island basalts. Earth and PlanetaryScience Letters 263, 388–403.

Putirka, K., Mikaelian, H., Ryerson, F., Shaw, H., 2003. New clinopyroxene-liquidthermobarometers for mafic, evolved and volatile-bearing lava compositions, withapplications to lavas from Tibet and the Snake River Plain. American Mineralogist88, 1542–1554.

Ramos, F.C., Reidy, M.R., 2005. Distinguishing melting of heterogeneous mantle sourcesfrom crustal contamination: insights from Sr isotopes at the phenocryst scale, PisgahCrater, California. Journal of Petrology 46, 999–1012.

Richard, P., Shimizu, N., Allègre, C.J., 1976. 143Nd/146Nd, a natural tracer: an application tooceanic basalts. Earth and Planetary Science Letters 31, 269–278.

Rogers, N.W., De Mulder, M., Hawkesworth, C.J., 1992. An enriched mantle source for po-tassic basanites: evidence from Karisimbi volcano, Virunga volcanic province.Rwanda Contribution to Mineralogy and Petrology 111, 543–556.

Rogers, N.W., James, D., Kelley, S.P., De Mulder, M., 1998. The generation of potassic lavasfrom the Eastern Virunga Province, Rwanda. Journal of Petrology 39, 1223–1247.

Rogers, N., Macdonald, R., Fitton, J.G., George, R., Smith, M., Barreiro, B., 2000. Twomantleplumes beneath the East African rift system: Sr, Nd and Pb isotope evidence fromKenya Rift basalts. Earth and Planetary Science Letters 176, 387–400.

Rosenthal, A., Foley, S.F., Pearson, D.G., Nowell, G.M., Tappe, S., 2009. Petrogenesis ofstrongly alkaline primitive volcanic rocks at the propagating tip of the WesternBranch of the East African Rift. Earth and Planetary Science Letters 284, 236–248.

Roshchina, I.A., Shevaleevsky, I.D., Korovkina, N.A., Maiorov, A.P., 1982. X-ray fluorescenceanalysis of rock samples of variable composition. Journal of Analytical ChemistryXXXVII, 1611–1618 (In Russian).

Sahama, T.G., 1974. Potassium-rich alkaline rocks. In: Sorensen, H. (Ed.), The AlkalineRocks. Wiley, London, pp. 96–109.

Schmidberger, S.S., Simonetti, A., Francis, D., 2003. Small-scale Sr isotope investigation ofclinopyroxenes from peridotite xenoliths by laser ablation MC–ICP–MS—implicationsfor mantle metasomatism. Chemical Geology 199, 317–329.

Simonetti, A., Bell, K., 1993. Isotopic disequilibrium in clinopyroxene from nephelinitelavas from Napak volcano, eastern Uganda. Geology 21, 243–246.

Simonetti, A., Bell, K., 1994. Nd, Pb and Sr isotopic data from the Napak carbonatite-nephelinite centre, eastern Uganda: an example of open-system crystal fractionation.Contribution to Mineralogy and Petrology 115, 356–366.

Simonetti, A., Bell, K., 1995. Nd, Pb and Sr isotopic data from the Mount Elgon volcano,eastern Uganda-western Kenya: implications for the origin and evolution of nephe-linite lavas. Lithos 36, 141–153.

Stacey, J.S., Kramers, J.D., 1975. Approximation of terrestrial lead isotope evolution by atwo-stage model. Earth Planetary Science Letters 26, 207–223.

Stracke, A., Hofmann, A.W., Hart, S.R., 2005. FOZO, HIMU, and the rest of the mantle Zoo.Geochemistry, Geophysics, Geosystems 6 (N 5), 1–20.

Thibault, Y., Edgar, A.D., Lloyd, F.E., 1992. Experimental investigation of melts from a car-bonated phlogopite lherzolite: implications for metasomatism in the continental lith-ospheric mantle. American Mineralogist 77, 784–794.

Vollmer, R., Norry, M.J., 1983a. Unusual isotopic variations in Nyiragongo nephelinites.Nature 301, 141–143.

Vollmer, R., Norry, M.J., 1983b. Possible origin of K-rich volcanic rocks from Virunga,EastAfrica, by metasomatism of continental crustal material: Pb, Nd and Sr isotopicevidence. Earth and Planetary Science Letters 64, 374–386.

Walter, M.J., 1998. Melting of garnet peridotite and the origin of komatiite and depletedlithosphere. Journal of Petrology 39, 29–60.

Xu, Y., Huang, X., Menzies, M.A., Wang, G.R., 2003. Highly magnesian olivines and green-core clinopyroxenes in ultrapotassic lavas from western Yunnan, China: evidence fora complex hybrid origin. European Journal of Mineralogy 15, 965–975.

Yasnygina, T.A., Rasskazov, S.V., Markova, M.E., Ivanov, A.V., Demonterova, E.I., 2003. De-termination of trace elements in volcanic rocks of basic and intermediate composi-tion by ICP-MS method using microwave oven acid decomposition technique. In:Burenkov, E.K., Kremenetskii, A.A. (Eds.), Applied Geochemistry. Issue 4. AnalyticalStudies 48, pp. 48–56 (In Russian).

Zindler, A., Hart, S., 1986. Chemical geodynamics. Annual Review Earth Planetary Science14, 493–571.














http://www.scgis.ru/rissian/cp1251/h_dgggms/1-2008/magm-23e.pdf

http://www.scgis.ru/rissian/cp1251/h_dgggms/1-2008/magm-23e.pdf





























































































Sr–Nd–Pb isotope systematics and clinopyroxene-host disequilibrium in ultra-potassic magmas from Toro-Ankole and Virunga, East-African Rift: Implications for magma mixing and source

Documents