Journal of Petrology Advance Access originally published online on July 22, 2004
Journal of Petrology 2004 45(9):1747-1776; doi:10.1093/petrology/egh032
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology 45(9) © Oxford University Press 2004; all rights reserved
Rates and Timescales of Fractional Crystallization from 238U–230Th–226Ra Disequilibria in Trachyte Lavas from Longonot Volcano, Kenya
1 DEPARTMENT OF EARTH SCIENCES, THE OPEN UNIVERSITY, MILTON KEYNES, MK7 6AA, UK
2 DEPARTMENT OF GEOLOGY, UNIVERSITY OF PLYMOUTH, DRAKE CIRCUS, PLYMOUTH, PL4 8AA, UK
3 DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF BRISTOL, QUEENS ROAD, BRISTOL BS8 1RJ, UK
RECEIVED MARCH 8, 2002; ACCEPTED MARCH 30, 2004
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONA suite of peralkaline trachytes from Longonot volcano, Kenya, which erupted during the last 6000 years, has been analysed for major and trace elements, Pb and Nd isotopes, and U–Th–Ra disequilibria. The lavas are divided into three stratigraphic groups of trachytes (Lt2a, Lt2b and Lt3), and hybrid lavas, designated LMx1 and LMx2, which, respectively, pre-date and post-date the Lt2 lavas. Major and trace elements are consistent, with up to 37% within-group fractional crystallization of predominantly alkali feldspar. The parental magma for the different trachyte groups had a more mafic composition—probably hawaiitic. Nd and Pb isotopes show minimal variation, both within and between magma groups, and indicate that up to 10% comendite magma from the neighbouring Olkaria volcanic field may have intermixed with the Longonot magma. (230Th/238U) disequilibria indicate that limited U/Th fractionation occurred during the past 10 kyr, whereas (226Ra/230Th) disequilibria reflect the effect of alkali feldspar fractionation >8 kyr ago in the Lt2a lavas, between 3 and 7 kyr ago in the Lt2b lavas and in the past 3 kyr for the Lt3 lavas. (226Ra/230Th) disequilibria in the Lt2b lavas are interpreted using a model that combines the equations of radioactive decay and in-growth with Rayleigh crystallization to give fractionation rates of about 0·2 x 10–4/year for the evolution of hawaiite to trachyte, but more rapid rates of up to 3 x 10–4/year for fractionation within the trachyte sequence. (226Ra/230Th) from two whole-rock–alkali feldspar pairs are interpreted to show the crystals formed at 5800 years BP (Lt2b) and 2800 years BP (Lt3), implying that phenocryst formation continued almost up to the time of eruption. The results strongly indicate that fractionated magmas can be stored for periods on the order of 1000–2500 years prior to eruption, whereas other magmas were erupted as fractionation was proceeding.
KEY WORDS: trachyte; magma chambers; u-series; Kenya
|
INTRODUCTION |
|---|
A major challenge in igneous petrology is the determination of the rates at which magmas evolve, the factors that contribute to controlling those rates and how they relate to the thermal budgets of volcanic systems. Apart from academic interest in understanding the evolution of magmatic systems, the dynamics of the deeper regions of volcanoes are critical to the development of models for accurate forecasting of volcanic hazards (Sparks, 2003
) and complement information from observations of eruption frequency and volume made from surface and geophysical observation. One of the most promising methods for determining rates of magma evolution involves the use of short-lived isotopes in the decay chains of 238U and 235U, particularly the use of disequilibrium between 238U, 230Th and 226Ra. These systems allow the investigation of element fractionation on timescales commensurate with the half-lives of the isotopes concerned, namely 1599 years for 226Ra and 75 690 years for 230Th, and have the potential to define timescales of crystallization and fractionation processes within sub-volcanic magma chambers prior to eruption.
There have been numerous studies of U–Th disequilibrium in volcanic rocks, both on whole-rock samples and mineral separates, and these have been extensively reviewed by Hawkesworth et al. (2000) and, more recently, by Condomines et al. (2003). For basaltic magmas, the effects of crustal-level processes on U-series disequilibria are, in most cases, relatively insignificant compared with those inherited from melting and melt transport within the magma source region [see Lundstrom (2003) and Bourdon & Sims (2003) for comprehensive reviews on mid-ocean ridge basalt (MORB) and ocean island basalt (OIB), but also Vigier et al. (1999) and Cooper et al. (2001)]. By contrast, silicic magmatic systems are often dominated by magma chamber processes, and the timescales from U-series isotopes and the Rb–Sr system are interpreted in terms of crustal residence times and rates of magma fractionation. (e.g. Widom et al., 1992; Bourdon et al., 1994; Reid, et al., 1997; Bohrson & Reid, 1998; Reid, 1998; Heath et al., 1998 Heumann & Davies, 2002; Heumann et al., 2002).
Many of these investigations involve the analysis of mineral separates, which may date specific crystallization events and reflect storage of crystals prior to eruption. In many cases, such separates are mixtures of crystals with contrasting histories and they provide no direct indication of the storage location, e.g. whether crystals and melt were preserved in a slow-cooling, long-lived magma body (Halliday et al., 1989) or in a largely solid cumulate pile, remobilized by more recent events (Heath et al., 1998). Thus, the rates derived from these studies are not simply related to the rate at which a magma body evolves. In addition, age estimates derived from different isotope systems can give different timescales on the same samples, and these, in turn, differ from timescales from other physical methods (e.g. crystal size distribution and intra-mineral diffusion). These discrepancies probably reflect the times at which different aspects of the final texture of the rock were generated and relate to multi-stage processes, involving crystals produced during earlier magmatic episodes and reactivated immediately prior to more recent phases of activity (e.g. Turner et al., 2003).
Rates of fractionation are more difficult to determine, as they relate to the rate of change of melt composition. One example where a rate of fractionation has been successfully determined is at Ardoukoba, a basaltic volcano in Djibouti, where Vigier et al. (1999) related changes in (226Ra/230Th) disequilibrium in whole-rock samples to fractional crystallization, defined by both major and trace elements. From a quantitative model that integrated crystal fractionation with radioactive decay, they deduced a fractionation rate of (3·3–3·6) x 10–4/year, depending on whether the system was closed or open. Here, we present data on whole-rock major and trace element, radiogenic isotope and U–Th–Ra disequilibria from a series of mafic to trachytic lavas from Longonot volcano in the Kenya rift. The lavas are related by fractional crystallization and the analyses reveal subtle variations in composition that are related to the separate and rapid evolution of distinct magma batches and their storage for significant periods of time prior to eruption.
|
LONGONOT VOLCANO: GEOLOGICAL BACKGROUND |
|---|
Longonot volcano is located at 0°55'S, 36°26'E, within the central graben of the Kenya rift (Fig. 1), and approximately 100 km west of Nairobi, south of Lake Naivasha. It is one of 12 Quaternary volcanoes that are aligned along the inner trough of the rift valley, collectively known as the ‘Quaternary volcanic province of central Kenya’ (Baker, 1987
) and, together, they represent the most recent activity in the world's most extensive and voluminous alkaline igneous province (Mechie, 1994). The Kenya rift is located along the boundary of the Tanzanian craton and the Mozambique mobile belt, formed during the Pan-African orogenic event in the late Proterozoic (Smith & Mosely, 1993), and Longonot lies close to the craton boundary, over a basement zone, remobilized during the Pan-African event.
|
The evolved volcanic rocks of the Quaternary volcanic province of Kenya have been reviewed by Macdonald (1987)
. They are overwhelmingly peralkaline in character and individual volcanoes are dominated by trachytes, although at least two centres (Eburru and Olkaria) erupt mainly comendites and another (Suswa) phonolites. Seven of the 12 volcanoes have undergone caldera collapse and, whereas those in the north are dominated by lavas with minor associated pyroclastic material and up to 40% basalt, the southern volcanoes (Menegai, Suswa and Longonot) show evidence of violent eruptions, with thick pyroclastic deposits and minimal basalt (Macdonald, 1987). Longonot itself lies between the rhyolite dome field of Olkaria, 20 km to the north-west (Macdonald, 1987; Macdonald et al., 1987; Black et al., 1997) and the phonolite-dominated Suswa, 
30 km to the south (Skilling, 1988; Scott & Skilling, 1999).
The volcanic history of Longonot has been the subject of a series of detailed studies (Scott, 1980; Clarke et al., 1990; Scott & Skilling, 1999) and it is subdivided into three distinct stages, as illustrated by the stratigraphic column in Fig. 1. The earliest ‘Olongonot’ stage (0·4 Ma to c. 21000 years BP) encompassed the growth of a large composite trachyte cone (Lt1), that culminated in the formation of a large caldera, 7·5 km in diameter, at c. 21000 years BP. This event marked the onset of the ‘caldera pyroclastic’ stage, during which early ignimbrites and surge deposits (Lp1–4) were followed by a series of airfall pumices (Lp5–6). The oldest of these has been dated at 9150 years ± 150 years BP by 14C (Clarke et al., 1990). The estimated total volume of the caldera pyroclastics is 20 km3 (Scott, 1980) and a 14C age of sediments overlying Lp6 indicates that it had ended by 5650 ± 120 years BP (Clarke et al., 1990). Several of these airfall pumice beds are compositionally zoned, with the lower sections being enriched in incompatible trace elements relative to the top (Scott, unpublished data).
An abrupt shift to predominantly effusive eruptions marked the onset of the final ‘lava sequence’ stage. The earliest lavas are heterogeneously mixed trachyte–hawaiite flows (Lmx1), followed by a protracted sequence of trachytes (Lt2) that accumulated to form the Longonot cone. Growth of the cone terminated with the eruption of an ash horizon (Lp8), 14C dated at 3280 ± 120 years BP (Clarke et al., 1990), closely followed by collapse of the summit, leaving a 2 km diameter pit crater. Subsequent eruptions of trachyte–hawaiite mixed lavas in the pit crater (Lmx2) were accompanied by flank eruptions of trachyte (Lt3), although the ages of individual eruptions are poorly constrained. The nature of the vegetation cover on some of the Lt3 flank eruptions suggests that they may be substantially younger than 3280 years and possibly <1000 years. The total volume of the lava sequence is estimated to be 18–20 km3, most of which comprises the main sequence of trachyte eruptions (Scott, 1980). This study is concerned with the compositional and isotopic differences within the lava sequence stage (Lt2 and Lt3), where detailed stratigraphy of flows and 14C ages permit these variations to be evaluated within a well-constrained temporal framework.
|
SAMPLES AND ANALYTICAL TECHNIQUES |
|---|
Thirty-eight samples were selected, to span the entire stratigraphic sequence of the lava sequence. Major and trace elements (Zr, Nb, Y, Rb, Cr, Ni, Cu, Zn and Ga) were analysed by X-ray fluorescence spectrography (XRF) at the Open University and the remaining trace elements by inductively coupled plasma mass spectrometry (ICP–MS) at the University of Plymouth. Routine precision for these techniques is better than ±5% (2
) and often better than ±2% (2), even at the low abundances of Ba and Sr that are typical of these samples. The abundances of Th and U determined by ICP–MS are in good agreement (within ±5%) with those subsequently determined by isotope dilution. Fluorine was determined by ion-selective electrode at Imperial College.
Nd and Pb were separated following standard procedures, and their isotope ratios were determined using Finnigan MAT 262 and 261 mass spectrometers, respectively. Pb was analysed in temperature-controlled runs and the ratios were corrected for 1
mass fractionation using the NBS981 standard (206Pb/204Pb: 16·937; 207Pb/204Pb: 15·493; 208Pb/204Pb: 36·705). During the period of this study, the 2 errors were 206Pb/204Pb ± 0·02, 207Pb/204Pb ± 0·02 and 208Pb/204Pb ± 0·04 (n = 45). Neodymium isotope ratios were normalized to Johnson and Matthey Nd = 0·511778, with an average laboratory uncertainty of ±2 x 10–5 (2). Total procedure blanks for neodymium and lead were <0·5 ng and <1 ng, respectively. Sr isotopes were not determined because the very low Sr abundances made them unusually susceptible to secondary processes of contamination or alteration.
U, Th and Ra isotopes were determined using a Finnigan MAT 262 mass spectrometer with an RPQ II deceleration lens that ensured an abundance sensitivity of <10 p.p.b. (van Calsteren & Schweiters, 1995). Samples were spiked with a combined 229Th–236U tracer (see Turner et al., 1997). Total procedural blanks are negligible (<100 pg for U and <50 pg for Th). The accuracy of the U and Th isotope ratios was determined by multiple analyses of the rock standard A-THO, which yielded values of (238U/232Th) = 0·919, (230Th/232Th) = 1·018, U = 2·22 p.p.m. and Th = 7·34 p.p.m., in close agreement with results from other laboratories (Williams & Gill, 1992). The precision of the U and Th isotope ratios was monitored through routine analysis of two solutions. U456 is a dilute solution of SRM U112a (natural U), and it yielded an average value of 234U/236U = 0·0975 ± 0·0003, a 2 deviation of ±0·62%. Repeated analysis of the Open University Th‘U’ standard (van Calsteren & Schweiters, 1995) yielded a 2 error of ±0·7% for (230Th/232Th) (n = 15) over the 6 month period of this study.
New protocols were developed for the measurement of 226Ra by thermal ionization mass spectrometry (TIMS), based upon previously published methods. Samples were spiked with a 228Ra tracer, prepared from NIST 3159 Th solution, to give an approximate 228Ra/226Ra = 1 in solutions with 50 fg of sample 226Ra. The decay-corrected 228Ra/226Ra of the spike was determined to an analytical precision of better than ±0·5% (2). Pre-concentration of Ra and Ba was achieved using a double pass-through cation exchange resin, in a procedure modified from that of Cohen & O'Nions (1991). Ra and Ba were then separated using ElChrom Sr-spec resinTM, in a procedure modified from Chabaux et al. (1994). The combined effect of these procedures was to achieve a yield of >80% and Ba/Ra of <102, compared with a natural Ba/Ra ratio of >108. Samples were loaded onto degassed Re filaments, using a Ta–HF–H3PO4 activator solution, and analysed dynamically in a Finnigan MAT 262 mass spectrometer. This procedure yielded beams of 200–2000 c.p.s. of 226Ra and 228Ra, which were stable over the duration of the analysis (1/2 h), to give an analytical precision of better than ±1% (2). Interferences on the masses were monitored at mass 227, which, in all cases, was negligible relative to the size of the Ra beam (<<1%).
There is, at present, no agreed standard for determining the accuracy of 226Ra analyses. Repeat analyses of a Mt Lassen sample yielded an average value of 1065 ± 9 fg/g (n = 6), in close agreement with the published values of 1063 ± 10 and 1068 ± 11 fg/g by Volpe et al. (1991). Precision was assessed using the Mt Lassen sample and an in-house standard derived from a Longonot sample (ThITS, n = 6) and these yielded 2 standard deviations of 1·5 and 1%, respectively. Repeat analyses of samples for U–Th–Ra and Ra alone are also within 1% (2) error (see also Turner et al., 2000). Total procedural blanks were below the detection limit of the mass spectrometer (currently 0·1 fg/g). The abundance of Ba in the sanidine mineral separates and in sample L33 was determined by isotope dilution using a 135Ba tracer, and with a precision of <1% (2). The decay constants used in the calculation of activity ratios were those compiled by Bourdon et al. (2003): 
238U = 1·55125 x 10–10, 230Th = 9·1952 x 10–6 and 226Ra = 4·335 x 10–4. Activities are reported as decays per year per gram.
Sanidine was separated by hand from a fine (250 µm) crush of the whole rock. Despite the low phenocryst abundance (<5%), the contrast in colour and lustre of the mineral and glass allowed the preparation of clean mineral separates. Each dissolution comprised 100 sanidine mineral shards. Prior to dissolution, each fragment of sanidine was individually examined for residual glass, before ultrasound washing in TD H2O to remove any dust particles and examined a second time to ensure >99% purity.
|
MINERALOGY AND PETROGRAPHY |
|---|
Trachytes from both the Lt2 and Lt3 units at Longonot are typical of this rock type throughout the Kenya rift, being either glassy or microcrystalline with a trachytic texture and, typically, <5% phenocrysts, with sanidine as the dominant phase. The sanidines are Na-rich and, within the Lt2 lavas, fall into two distinct compositional groups (Fig. 2). They vary from Ab62, Or36 and An2 in the earliest lavas, to Ab66, Or33 and An
1 in the most evolved Lt2 lavas. Other phenocrysts include rare Ca-pyroxene, fluor-amphibole and fayalitic olivine. Neither quartz nor feldspathoids are present and accessory phases are restricted to magnetite. The Lt3 trachytes are compositionally and mineralogically similar to the Lt2 lavas, but they also contain a large number of mafic xenoliths, the abundance of which decreases with time such that the youngest flow is free of mafic enclaves. The sample studied in detail contains alkali feldspar phenocrysts of similar composition to the group labelled Lt2a in Fig. 2.
|
The mixed lavas (LMx1 and LMx2) are highly porphyritic (
50%), with a dominantly glassy matrix. Two feldspars coexist—plagioclase with An80 and sanidine—similar in composition to those in the trachytes but with somewhat greater An contents (up to 6% An). Mafic minerals include both forsteritic and fayalitic olivines and aegirine–augite. Small anhedral magnetite crystals are ubiquitous and have characteristically high TiO2 contents (21–23 wt %). The contrasting compositions of the coexisting minerals, particularly the coexistence of calcic plagioclase and alkali feldspar, reveal the hybrid origins of the mixed lavas which result from the mingling of trachytic and basaltic or hawaiitic magmas. Syenitic xenoliths found in the Lp5 unit are dominated by exsolved alkali feldspar, with iron-rich olivine, pyroxene and amphibole and magnetite as interstitial phases. In addition, small (<10 µm) interstitial grains of a P, U and REE-rich accessory mineral, possibly apatite, were also identified from electron back-scattered images.
|
MAJOR AND TRACE ELEMENTS |
|---|
The major and trace element, and Nd and Pb isotopic data for the Longonot lava sequence are presented in stratigraphic order in Table 1, with the oldest sample in the first column. The lavas are predominantly peralkaline trachytes, with 0–5% normative quartz and molecular (Na2O + K2O) > Al2O3. On the total alkali–silica plot (Fig. 3), the Lt2 and Lt3 trachytes form a tight cluster, similar in composition to the trachytes from the earlier caldera pyroclastic stage and trachytes from Suswa (Skilling, 1988
; Scott & Skilling, 1999). The mixed lavas (Lmx1 and Lmx2) and one of the Lt3 lavas extend from the trachyte field to more mafic compositions.
|
|
Although silica shows limited variation in the trachytes and the compositional ranges of alkali feldspar are small, incompatible elements (e.g. Th, Nb and Zr) and Fe2O3 show marked and systematic variations in concentration. Figure 4 shows the stratigraphic variation of Fe2O3 and Th in the lavas, divided according to their stratigraphic units. Within the Lt2 unit, both Th and Fe2O3 show a gradual increase with time, Th doubling in concentration from 12 to 24 p.p.m. and Fe2O3 increasing from 7·8 to 10 wt %. Th is strongly correlated with other incompatible elements, such as Nb and Zr (Fig. 5)—a feature typical of evolved lavas from other peralkaline centres in the Kenya rift (Black et al., 1997
). Moreover, the Th/Zr, Th/Nb and Nb/Zr ratios (Fig. 5) are within the ranges of Kenya rift mafic lavas (Macdonald et al., 2001). The Lt3 lavas conform to the trace-element trends defined by the Lt2 trachytes, with Th contents ranging from 14·7 to 22 p.p.m., and inter-element ratios for Zr, Nb and Th indistinguishable from the older Lt2 lavas.
|
|
In the mixed lavas (Lmx1 and Lmx2), Fe2O3 is displaced to higher values than in the stratigraphically nearest trachytes, whereas Th concentrations are markedly lower. The contrasting behaviour of Fe2O3 in the mixed lavas and trachytes is clearly revealed on a plot of Fe2O3 against Zr (Fig. 6). In the trachytes, Fe2O3 and Zr show a good positive correlation, whereas the mixed lavas trend towards more mafic compositions (high Fe, low Zr). In the Lt3 lavas, Fe2O3 is similarly elevated and Zr depressed relative to the youngest Lt2 trachyte, consistent with the petrographic presence of enclaves of more mafic material.
|
Within the Lt2 lavas, there is a distinct hiatus between two samples (24Lt2 and 23Lt2) when Th and Fe2O3 concentrations jump from 19 to 22 p.p.m. and from 8·7 to 9·4 wt %, respectively (Fig. 4). This compositional shift is used to subdivide the Lt2 lavas into two separate groups, designated Lt2a and Lt2b. There is also an inflection in the Zr–Fe2O3 trend at this point (Fig. 6), reflecting an increased incompatibility of Fe2O3 in the Lt2b sequence. Where analyses are available, this subdivision is consistent with mineral compositions, the Lt2a group containing slightly more orthoclase-rich feldspars than the Lt2b lavas (Fig. 2). Within the Lt2b sequence, three samples (7Lt2, 21Lt2 and 28Lt2) have distinctly lower Fe2O3 and Th contents and are compositionally closer to the Lt2a lavas. These three samples also plot along the trachyte trend in Fig. 6 (Zr vs Fe2O3) and are not displaced towards more mafic compositions.
Selected REE analyses are illustrated in Fig. 7. Abundances generally increase with other incompatible elements, although correlations between individual REE and Zr are not as tight as those between Th, Nb and Zr. The trachytes and the LMx2 lavas are also characterized by pronounced negative Eu anomalies, consistent with the removal of alkali feldspar, which has a much larger D value for Sr and Eu than plagioclase (Blundy & Wood, 2003). By contrast, the positive Eu anomalies in the LMx1 lavas indicate alkali feldspar accumulation.
|
In marked contrast to other trace elements, Ba and Sr decrease rapidly as Zr, Th and Nb increase within the trachytes. This behaviour is consistent with their compatibility with alkali feldspar, D values for both these elements being consistently >1 (Long, 1978
; Guo & Green, 1989; Icenhower & London, 1996). Pb, however, shows a positive correlation with Zr and Th, suggesting that it is not as compatible with sanidine as might be expected from theoretical considerations (Blundy & Wood, 2003). A more detailed analysis of element compatibility follows in the discussion. Whereas all of the lavas in the Lt2 and Lt3 units are trachytic, there are subtle differences between and within the stratigraphic groups. The correlation between total iron, Th and Zr reveals that changes in incompatible trace elements are coupled with changes in the major elements. For example, pronounced linear trends are apparent in the variation of K2O and Al2O3 with Zr within the trachytes, both elements decreasing as Zr and Th increase (Fig. 8), consistent with removal of alkali feldspar similar in composition to the phenocrysts. In addition, other major elements not accommodated by sanidine also decrease with increasing Zr and Th, although in a less pronounced fashion. For example, MgO decreases from 0·6 to 0·2 wt %, CaO from 1·50 to 1·06 wt %, TiO2 from 0·67 to 0·57 wt % and P2O5 from 0·114 to 0·038 wt % as Zr increases from 576 to 1128 p.p.m. Of the remaining major and minor elements, only MnO behaves incompatibly, whereas Na2O increases slightly with Zr and Th in the Lt2a group but shows no particular trend in the Lt2b or Lt3 lavas.
|
Together, these variations imply the removal of a ferromagnesian and a phosphatic phase, possibly apatite, in addition to sanidine during fractionation. Although the P2O5 concentration of the trachytes is low for apatite saturation, based on their SiO2 contents, (Green & Watson, 1982
), the progressive decrease in P2O5 as Zr increases is strong evidence that a phosphate-bearing phase is being removed, consistent with the presence of small grains rich in P, U and the REE in syenite xenoliths. |
RADIOGENIC ISOTOPES |
|---|
The Longonot trachytes have a restricted range in Pb and Nd isotope ratios, with 206Pb/204Pb = 19·62–19·69, 207Pb/204Pb = 15·66–15·73 and 208Pb/204Pb = 39·36–39·50 (Fig. 9), and 143Nd/144Nd = 0·51265–0·51258. Pb isotopes plot above the Northern Hemisphere Reference Line (NHRL) of Hart (1984)
and are within the range of values typical of basaltic rocks from the craton and craton margin settings of the Kenya rift (Rogers et al., 2000). Similarly, 143Nd/144Nd ratios of the trachytes are comparable with those of basalts from the same geological settings. However, there is no systematic variation between Pb and Nd isotopes in the Longonot data set.
|
143Nd/144Nd with 207Pb/204Pb in Longonot, Suswa and Olkaria. Kenya basalt data from Rogers et al. (2000)In detail, the Lt2a and Lt2b lavas define a steep linear trend on the 207Pb/204Pb vs 206Pb/204Pb plot. This trend is too steep to have any age significance but might reflect inadequacies in the conventional mass fractionation correction of Pb-rich samples in comparison with repeated analyses of NBS981. The gradient of the trends in both Pb isotope diagrams is similar to that defined by the uncorrected standards and may represent the effects of residual mass fractionation. However, two samples with similar Pb contents have well duplicated but analytically different isotope ratios that define the extremes of the Longonot array (207Pb/204Pb—20Lt2 15·73, 15·75, 15·74; 15Lt2 15·65, 15·66), implying that if some of the analytical spread is the result of mass fractionation, then it is not simply related to Pb contents. Notwithstanding such possible uncertainty on the within-suite variations, the Nd and Pb isotope fields for Longonot are distinct from those for the adjoining evolved volcanic centres. In particular, the Olkaria comendites have higher 207Pb/204Pb and low 143Nd/144Nd, indicative of a crustal origin (Davies & Macdonald, 1987
), and they also define steep trends in both conventional Pb isotope diagrams, broadly collinear with the Longonot arrays. |
238U–230Th–226Ra DISEQUILIBRIA |
|---|
U, Th and Ra contents and isotope ratios of the Longonot lavas are listed in Table 2 and illustrated in conventional equiline plots in Fig. 10. U and Th contents in the trachyte whole-rock samples range from 1·1 to 5·6 p.p.m. and 5·4 to 23·8 p.p.m., respectively. By contrast, (238U/232Th) activity ratios lie within the restricted range of 0·60–0·71 and (230Th/232Th) between 0·77 and 0·81. All of the trachytes have excess 230Th, with (230Th/238U) = 1·11–1·28, and define a sub-horizontal array on a conventional U–Th isochron diagram (Fig. 10a). There is no systematic variation of (238U/232Th) or (230Th/232Th) with stratigraphic position or Th concentration. These results contrast with those from the adjoining Olkaria comendite field, in which most samples have excess 238U (Black et al., 1987). However, subsequent mass spectrometric analyses of comendites from two of the centres within the Olkaria complex show either secular equilibrium or slight 230Th excess (Heumann & Davies, 2002
). The range of (230Th/232Th) ratios from this later study (0·76–0·78) is similar to that shown by the Longonot lavas, but (238U/232Th) ratios are higher.
|
|
In contrast to the uniform (230Th/232Th) ratios, there is considerable variation in the present-day (226Ra/230Th) ratios. The lavas are either close to secular equilibrium (±2%) or show 230Th excess over 226Ra, although one sample has an 11% 226Ra excess. All have very high (230Th/Ba) ratios, 1–2 orders of magnitude higher than those determined for basalts from Afar (Vigier et al., 1999
). In detail, the different magma groups show contrasting behaviour on the Ra equiline diagram (Fig. 10b and c). Three of the older Lt2a lavas generally plot close to the equiline, whereas the Lt2b trachytes have excess 230Th and define a clear linear array, with a slope corresponding to an ‘age’ of 4270 (+1390/–860) years. The Lt3 lavas also plot in 230Th excess but at lower (230Th)/Ba ratios than the Lt2b lavas.
Two sanidine separates were also analysed for Ra–Th disequilibrium, one (28Lt2) from the Lt2b group and the second (L33) from the Lt3 trachytes. Results are given in Table 2. The two-point regressions between the separates and whole rocks yield ages of 6700 (+700/–540) years from the older Lt2b trachyte (28Lt2), and 2630 (+130/–120) years from the younger Lt3 trachytes (L33). Both ages are slightly older than their maximum stratigraphic ages. However, given the clear control exerted by partition coefficient on the (226Ra)/Ba ratio of feldspars and coexisting liquid (Cooper et al., 2001), these ages should be regarded as purely indicative and the data are subject to a more detailed interpretation below.
|
TRACHYTE EVOLUTION |
|---|
The generation and evolution of evolved peralkaline magmas pose a number of problems, not least of which is the large variation of incompatible element abundances but minimal change in major element composition. One popular mechanism for the generation of these characteristics involves enrichment in the upper layers of a magma chamber by fluid movement, enhanced by high concentrations of halogen elements, particularly fluorine, which are inferred to concentrate selected trace elements in the upper, more fractionated layers of the magma chamber (e.g. Bailey & Macdonald, 1975
; Macdonald, 1987; Macdonald et al., 1987). Despite having been degassed on eruption, Longonot trachytes are rich in fluorine (Table 1); some of the lavas contain in excess of 2000 p.p.m. F and the effects of such high levels of halogens may be significant. Alternatively, as trachytes are close to the composition of the alkali feldspar minimum in Petrogeny's Residua System, such variations could be a consequence of fractional crystallization, with the major elements remaining relatively buffered at a pseudo-eutectic composition (e.g. Baker, 1987). The problem is whether the amount of fractionation implied by the incompatible elements is consistent with the limited major element variations.
Major elements
When the major element compositions of the Longonot trachytes are recalculated into normative components and plotted on Petrogeny's Residua System, they fall close to the Ab–Or divide, with small but significant normative silica, and close to the thermal minimum in the Ab–Or binary system (Fig. 11). Thus, with respect to SiO2, Al2O3 and the alkali elements, they are close to a eutectic composition. Given that the dominant phenocryst phase in the trachytes is alkali feldspar, then, because most of the trachytes are slightly quartz normative, fractional crystallization should drive the liquid composition towards the granite minimum. However, the bulk composition of the trachytes in this system is so close to that of the alkali feldspar and the albite–orthoclase minimum point that such an effect is marginal, even after substantial fractionation. Thus, if a parental trachyte has 1% normative quartz and loses 50% of a feldspar with Or33–37 (Fig. 11), then the final bulk composition will contain only 2% normative quartz and still fall within the compositional field of trachytes. Such a small compositional shift is apparent from the data, the more trace element-enriched Lt2b lavas plotting further away from the Ab–Or join than the Lt2a group, but, within each group, there is no overall trend.
|
The amount of fractional crystallization and the possible phases involved were assessed using a least-squares method. Mineral analyses from the least-evolved lava in each group, namely 26Lt2 for the Lt2a lavas and 23Lt2 for the Lt2b group, were used in separate calculations and the results are listed in Table 3. In both cases, the fractionating assemblage is dominated by alkali feldspar (88 and 86% in groups a and b, respectively), with minor amounts of clinopyroxene, fayalite, magnetite and apatite, as predicted from the binary element variation diagrams. The sums of squares in the Lt2a group are <1, but are >1 in the Lt2b lavas, the poorest agreement in both cases, and particularly for the Lt2b lavas, being for Na2O. The total amount of fractionation within each group is 38% for the Lt2a lavas and 7% for the Lt2b group. These calculations are a clear demonstration of how significant crystal fractionation can occur within the trachyte compositional field.
|
Least-squares models that attempt to relate the most primitive Lt2a lavas to the most evolved Lt2b lavas have much higher residuals and sums of squares, implying that the two magma groups may not be simply related to one another, but that they were derived from distinct parent magmas. This latter conclusion supports the stratigraphic division into the two magma groups and is consistent with the two groups of alkali feldspars corresponding to the Lt2a and b lavas.
Trace elements
One of the most striking features of evolved peralkaline magmatism is the excellent correlations shown by the incompatible trace elements, Nb, Zr and Th, and such correlations are a prominent feature of the Longonot lavas. These almost perfect correlations imply that all three elements are behaving with virtually identical and probably perfect incompatibility, and that they are not controlled by accessory phases, particularly zircon.
The stability of zircon at magmatic temperatures is a function of melt composition, and Watson & Harrison (1984) have shown that it is related to a parameter, M = molar (Na + K + 2Ca)/(Al x Si), similar to the peralkalinity index (molar (Na2O + K2O)/Al2O3), and that as M increases, so does Zr solubility. The least-evolved Longonot trachyte has a value of M of 1·37, and would therefore become saturated in zircon when temperatures fall below 900–910°C (Watson & Harrison, 1984). The liquidus temperatures of peralkaline trachytes are not well determined, but, considering they are poor in water and the temperature of the anhydrous alkali feldspar minimum is 1063°C, it is unlikely that the least-evolved Longonot trachyte had a liquidus temperature as low as 910°C and, therefore, was not saturated in zircon. Like the peralkalinity index, the value of M increases with feldspar fractionation and so Zr solubility also increases. Thus, as concluded by Watson (1979), if a trachyte is initially Zr-undersaturated, it will never become saturated as a result of feldspar fractionation alone. Given that fractionation in the Longonot lavas is dominated by feldspar, zircon will not become stable at any stage in the evolution of the trachytes and Zr may therefore be regarded as a truly incompatible element. Furthermore, given the lack of fractionation of the Zr/Nb (4·85 ± 0·05) and Zr/Th (47 ± 1) ratios throughout the trachyte lava sequence, Nb and Th are also behaving as truly incompatible elements during trachyte fractionation. Hence, all three can be used as quantitative indices of fractional crystallization. For example, the difference between the Zr, Nb and Th contents of the least- and most-evolved Lt2a lavas (26Lt2a and 17 Lt2a) imply 36%, 35% and 35% fractional crystallization, respectively, agreeing remarkably well with the estimate of 38% from the least-squares calculations. Similarly, for the Lt2b lavas, the three estimates are all 7%, again agreeing closely with the least-squares calculation.
Other trace elements, such as Rb, U, Y and the REE, show poorer correlations with Zr and Th over comparable relative concentration ranges. U, Y and the REE are affected by apatite fractionation and, given that P2O5 behaves as a compatible element during fractionation and apatite is part of the fractionating assemblage in the Lt2a calculations, this scatter may relate to the presence of apatite or another phosphate in the fractionating assemblage. The poorer correlation shown by Rb cannot be explained by apatite fractionation but may be related to the formation of amphibole or other minor phases.
By contrast with most trace elements, Ba and Sr show marked decreases as Zr increases, reflecting their compatibility in a fractionating assemblage dominated by alkali feldspar. This variation is best illustrated on a logarithmic plot (Fig. 12a), on which trends generated by fractional crystallization plot as straight lines. Each of the three lava groups have broadly comparable Ba contents, ranging from 5 to 200 p.p.m., but plot at progressively higher Th contents for a given Ba content.
|
The variation of Ba with Th can be modelled using D values calculated after Blundy & Wood (2003)
from the composition of the alkali feldspars and their calculated modal abundance in the fractionating assemblage (Table 3). The calculated fractionation vectors are superimposed on the data in Fig. 12. They show good agreement with both the data and the results from the least-squares calculations, suggesting 35% and 10% fractionation within the Lt2a and b groups, respectively. However, it is equally apparent that the Lt2a and Lt2b lavas do not lie on a common fractionation trend. Extrapolation of the Lt2a trend in Fig. 12a intersects the field defined by basaltic lavas from the Kenya rift at approximately 750 p.p.m. Ba and 10·5 p.p.m. Th; fractional crystallization of alkali feldspar from such a starting composition can easily reproduce the Ba, Th and Th/Ba ratios of the Lt2a samples after 14–44% fractionation (0·86 > F > 0·56). Similarly, the Lt2b lavas can be derived from a mafic parent with 800 p.p.m. Ba and 12·5 p.p.m. Th after 40–48% fractionation (0·6 > F > 0·52) (Fig. 12b). The trends in Fig. 12a are therefore consistent with the conclusions from the least-squares calculations that the Lt2a and Lt2b lavas cannot be directly related to one another through fractional crystallization. All appear to have been derived from more mafic parents with higher Ba but lower Th contents. Given that fractionation is largely controlled by alkali feldspar, then other trace elements, notably Eu and Pb, should also behave compatibly, decreasing as Th and Zr increase. A plot of Eu/Eu* against Th (Fig. 12c) shows that Eu/Eu* decreases with fractionation, but the rate of decrease is small, consistent with an Eu D value of 0·7, significantly lower than the values >2, expected from the similar ionic radii of Sr2+ and Eu2+. The relatively small DEu probably results from a high fO2, typical of evolved peralkaline melts in general, and may also explain the relative incompatibilities of Fe as Fe3+ and Pb as Pb4+.
In contrast to the Lt2a and b lavas, the Lt3 lavas do show petrographic evidence for magma mixing, the earlier flows in the group containing mafic enclaves resembling the mixed lavas of the LMx1 and LMx2 units. The four Lt3 analyses do not lie on a linear trend in Fig. 12a, but a convex-upwards curve. In addition, two lavas from the Lt2b group also plot at high Ba concentrations, but with similar Th contents to the main Lt2b group. These data are more consistent with mixing trends and lie along a calculated mixing line between the LMx2 group and the Lt2b lavas. They suggest that, in addition to crystal fractionation, there is some degree of magma mixing between stratigraphically separated magma lineages and that variations within the Lt3 lava group are dominated by magma mixing between pre-existing magma types, namely the Lt2b and LMx2 magmas. However, those lavas that do not show petrographic evidence for magma mingling conform to models of closed-system fractionation, which is the dominant mechanism whereby the trachytes have evolved.
Magma mixing is also implied by the Pb isotopes, which lie on a steep trend on conventional Pb isotope diagrams, broadly collinear with the Olkaria comendites (Fig. 9). The total range of Pb isotope ratios within Longonot can be modelled by addition of up to 10% of a typical Olkaria comendite, because the latter has very high abundances of Pb relative to the Longonot trachytes. The positive correlation between Pb content and Zr in the Longonot lavas suggests that Pb is behaving as an incompatible element during magmatic evolution, but there is no clear correlation between 207Pb/204Pb or 208Pb/204Pb and Pb concentration. Thus, any mixing that may have occurred between these two magmas—in effect, contamination of the trachyte by a crustal melt—must have taken place prior to the onset of alkali-feldspar fractionation.
In conclusion, both mixing and fractional crystallization can be recognized as controlling processes in the evolution of the Longonot trachytes. The preferred model for the Lt2a and b lavas is that they are not related to each other but represent the products of crystal fractionation of more mafic, probably hawaiitic, parents. By contrast, the Lt3 lavas reflect mixing between the Lt2b magma and more mafic material, possibly represented by the LMx2 lavas. Trace-element variations suggest that the mafic parent had the trace-element contents of an evolved basalt or hawaiite and that up to 50% fractionation is required to generate the most evolved trachyte from such a mafic parent.
|
TIMESCALES OF TRACHYTE FRACTIONATION |
|---|
238U–230Th disequilibrium
The whole-rock analyses of all of the samples from the lava sequence show an excess of 230Th over 238U, with minimal variation in (230Th/232Th), whereas (238U/232Th) varies from 0·6 to 0·72. The flat array on the equiline diagram implies that U/Th fractionation occurred very recently and within the uncertainty of the zero age derived from the data, i.e. within the last 10 kyr. This uncertainty reflects the inability of the U–Th system to record ages of <10 ka, except in those rare cases where there has been substantial U–Th fractionation (e.g. zircon, Charlier & Zellmer, 2000
). Within Longonot, U–Th fractionation may be related to the removal or accumulation of minor amounts (<<1%) of apatite during magma evolution but there is no clear relationship between U/Th and Zr or P2O5 content. Perhaps the most significant observation is that there is no difference between the measured (230Th/232Th) ratios of any of the magma groups (Lt2a, b, Lt3 or LMx), implying that the differences both within and between magma groups were generated on a similar (<10 kyr) timescale. This is particularly significant for the LMx samples, which were generated by mixing trachytic magma with a more mafic, probably hawaiitic, magma, and suggests that the mafic component being added in also had a similar (230Th/232Th) ratio to the bulk of the trachytes. It is also consistent with the trachytes' forming by protracted fractional crystallization of a more mafic (hawaiitic) magma.
226Ra–230Th disequilibrium
It is clear from Fig. 10 that there are significant differences in (230Th/226Ra) disequilibrium between the different magma groups. Of the four Lt2a samples analysed, two plot within error of the equiline, with (226Ra/230Th) ratios of 1 ± 0·01, implying that (230Th/226Ra) fractionation must have occurred >8000 years ago (i.e. >5 half-lives of 226Ra). Given that the stratigraphic ages of these lavas are between 5650 and 3280 years, this indicates storage of fractionated trachyte magma beneath Longonot for >2500 years prior to eruption. A third sample has a (226Ra/230Th) ratio of 1·11 and so is out of secular equilibrium, which implies that (226Ra/230Th) fractionation occurred within the last 8000 years, although the nature of that fractionation is difficult to establish. The fourth sample in this group and all of the Lt2b samples have an excess of 230Th over 226Ra of >2·7%, i.e. greater than the analytical uncertainty of 1–2%, and so they record fractionation ages of <8000 years.
The gradient of a linear regression through the Lt2b samples is equivalent to an age of about 4200 years, which is similar to the stratigraphic age. Thus, a first-order conclusion is that (226Ra/230Th) fractionation occurred at about the same time as eruption. However, this ‘age’ interpretation implies that all samples fractionated from a common parent with a (226Ra)/Ba ratio equivalent to the intercept of the regression with the equiline, i.e. 105. It has already been shown that the Lt2a, b and Lt3 groups originated from mafic parents that had much lower Th/Ba, and, hence, (226Ra)/Ba ratios. Furthermore, in a system fractionating alkali feldspar, in which both Ba and Ra are compatible elements, the (226Ra)/Ba ratio will also change with fractionation according to the partitioning of both elements into alkali feldspar. Thus, the interpretation of the linear array in Fig. 10 is far from straightforward.
The relationship between XOr and DBa from Blundy & Wood (2003) allows the calculation of the DBa/DRa ratio. Although this relationship is strictly only appropriate to specific conditions (0·2 GPa and 650–750°C) that are different from those for fractionation in Longonot, the D value for Ba calculated in this way for the Lt2a lavas accurately models the variation of Ba within the sequence, suggesting that the approximation is valid. Hence, the Blundy & Wood equation for relating the DBa/DRa ratio to XOr is adopted here without further modification but with those uncertainties in mind. The nature of the relationship is such that DRa is always <DBa and both increase with XOr, and, because Longonot sanidines are relatively homogeneous, Ra and Ba partition coefficients vary little throughout the fractionating sequence. For example, in the case of sanidines from the Lt2b lavas, in which XOr = 0·35, the bulk D value of Ba during Lt2b fractionation is 8·85 and that for Ra is 2·61, whereas, for XOr = 0·33, a composition appropriate for the Lt2a lavas, DBa = 8·44 and DRa = 2·40. Such small differences are considered insignificant in the following analysis of the data.
The effects of these values on the instantaneous position of a fractionating magma on the Ra equiline plot are shown in Fig. 13. The curves assume a parent mafic magma with (226Ra/230Th) = 1 and 2, similar to values from the mixed lava, 6LMx1, and basalts from Ardoukoba (Vigier et al., 1999), respectively. The latter are the only other (226Ra/230Th) data from the African Rift and indicate the likely maximum disequilibrium of basaltic lavas when they enter a crustal magma chamber. Moreover, a review of available (226Ra/230Th) data from within-plate basalts (mainly OIB) worldwide (Bourdon & Sims, 2003) shows that initial (226Ra/230Th) ratios are seldom >2. Given that the Longonot lavas are significantly more evolved than those from Ardoukoba, implying that the Longonot parental magma would have been resident in the deep crust for a significant period relative to the half-life of 226Ra, the Ardoukoba data provide a maximum constraint on the initial (226Ra/230Th) ratio of any likely parental magma feeding the Longonot system. Initial (226Ra)/Ba ratios of 1000 and 2000 are assumed and an initial (230Th)/Ba ratio of 1000 in both cases. These values are similar to those from the Ardoukoba basalts and 6LMx1.
|
The calculated instantaneous fractionation path of the evolving magma is plotted in Fig. 13. As fractionation and the (230Th)/Ba ratio increase, so does the (226Ra)/Ba ratio, but at a slower rate because Ra is more compatible in sanidine than Th. Consequently, fractionated magmas have 230Th excesses relative to 226Ra. Using this model fractionation curve as a baseline, it is then possible to calculate a model age for each sample from the difference between the measured (226Ra)/Ba and the modelled value at the measured (230Th)/Ba value, i.e. the vertical displacement of a measured sample above the modelled curve. Model ages are based on the simple relationship
 | (1) |
is the decay constant of 226Ra. The model ages are tabulated in Table 4, and are generally similar to or greater than the minimum eruption age of 3280 years. It is not possible to calculate model ages for those samples with (226Ra/230Th) 
1 and this is the case for three samples from the Lt2a and one Lt2b lava. However, the observation that one sample has a significant 226Ra excess suggests that (226Ra/230Th) fractionation must have occurred within the past 8000 years (i.e. within 5 half-lives). By contrast, those samples with close to unity have not fractionated (226Ra/230Th) within the past 8000 years.
|
The remaining Lt2b samples all have model ages between 3·6 and 7·5 ka, all but one being within the bracketed stratigraphic age bounds of 3280–5650 years. If the parental magma had (226Ra/230Th) = 2, but an otherwise similar composition, then the fractionation curve is shifted upwards (Fig. 13) and model fractionation ages are reduced by 400–600 years (Table 4). Such differences do not alter the main conclusion that fractionation ages are similar to or slightly greater than stratigraphic ages.
This method allows model ages to be calculated for the Lt3 lavas and the one Lt2a lava that lies below the equiline, and these ages for the two starting conditions are also listed in Table 4. For the Lt2a lava (25ALt2), the fractionation age is about 7000 years—significantly older than the maximum eruption age of 5250 years. However, the two Lt3 lavas have fractionation ages of 2500 and 3400 years, compared with their maximum stratigraphic age of 3280 years.
In the simplest case, model ages relate to the time when the bulk-rock composition was established and, because the composition is related to fractional crystallization, they are directly related to the time that (226Ra/230Th) fractionation ended. Furthermore, because they are calculated relative to a model of instantaneous crystal fractionation, they represent maximum ages for fractionation. Those samples that lie on or close to the equiline must have fractionated alkali feldspar >8000 years ago and, as their maximum stratigraphic age is 5650 years, secular equilibrium implies storage of fractionated magma beneath Longonot for at least 2500 years prior to eruption. By contrast, the majority of those lavas that show substantial (226Ra/230Th) disequilibrium have model ages that are bracketed by the stratigraphic ages and imply that fractional crystallization was ongoing, either immediately prior to or synchronous with eruption.
|
THE EFFECTS OF RADIOACTIVE DECAY AND IN-GROWTH OF 226Ra DURING FRACTIONAL CRYSTALLIZATION |
|---|
The above examples show how Ra/Ba and Ra/Th ratios can be affected by simple crystal fractionation. However, 226Ra variations also result from radiogenic in-growth or radioactive decay, depending upon the (226Ra/230Th) ratio of the system in question, and that process continues while fractional crystallization is occurring. Clearly, if fractional crystallization is rapid relative to the half-life of 226Ra, then the effect of in-growth will be minimal. By contrast, if fractionation occurs over a period of some hundreds or thousands of years, then decay and in-growth need to be taken into consideration in a detailed analysis of 226Ra–230Th disequilibria. The situation is complicated because, at any given stage of fractionation, the evolving melt has a unique (226Ra/230Th) ratio, determined by the amount of fractionation and the D values of Ra, Ba and Th. Vigier et al. (1999)
modelled combined radioactive decay and in-growth for systems in which the absolute rate of fractionation varied exponentially with time. In the following analysis, we investigate the situation in which the amount of fractionation is a linear function of time (i.e. the absolute rate remains constant), and then explore the implications for the fractionation rates and residence times of magmas beneath Longonot. This model is considered more appropriate for evolved systems, because they often have eutectic or pseudo-eutectic compositions, and the crystallization rate will be governed by the rate at which heat is conducted away from the system rather than by the rate at which the temperature of the magma falls. However, the deviations between the two models are not great, except when the amount of crystallization becomes large (i.e. >30%), as is the case for Longonot. The derivation of the governing equations is outlined in the Appendix and the following section is based on the numerical solutions of equations (A7), (A8) and (A9a). Before these equations can be applied to the Longonot data, assumptions have to be made concerning the precise age of eruption. All of the lavas from the Lt2b series were erupted between 3300 and 5250 years BP. The Lt2b lavas are high in the Lt2 stratigraphic sequence (Fig. 4) and so (226Ra/230Th) ratios are corrected back to 3300 years BP. (Note that the effect of correcting to older ages is to reduce the initial (226Ra/230Th) ratios, further increasing the calculated fractionation rates and reducing the timescales of magma differentiation. Hence, the timescales determined from the following analysis should be regarded as maxima.) In the following section, we investigate fractionation from an original hawaiitic magma with initial (226Ra/230Th) ratios of 1 and 2 (as in Fig. 13) and fractionation within the Lt2b trachyte sequence, using the calculated (226Ra/230Th) ratio of the least-evolved member of that stratigraphic unit as a parent.
In Fig. 14a, the parental hawaiite magma has a Th content of 12·5 p.p.m. (determined from the modelled trend in Fig. 12b), the Th partition coefficient is assumed to be zero and the bulk partition coefficient for Ra was calculated after Blundy & Wood (2003) and the modal composition of the cumulate phases (Table 3). Curves for different values of a, the rate of fractionation, that encompass the range of (226Ra/230Th) in the Lt2b lavas are superimposed on the data. Also shown is a curve for instantaneous Rayleigh fractionation and a curve based on the closed-system equation from Vigier et al. (1999) for a value of f of 0·0002/year. The general coincidence between the curve from the Vigier et al. equation and our method at low degrees of fractionation (i.e. low Th content) for the same value of a demonstrates the consistency between the two approaches, whereas the deviation at higher Th concentrations is a reflection of the exponential decline in absolute fractionation rate in the case of the Vigier et al. model. In Fig. 14b, similar curves are plotted for the same values of fractionation rate but based on a parental hawaiite with an initial (226Ra/230Th) of 2.
|
The majority of the samples lie between curves corresponding to fractionation rates of (0·5–2) x 10–4/year, whereas the least-evolved sample (lowest Th content) lies closer to the curve corresponding to a rate of 2 x 10–5/year, and one sample lies close to the instantaneous fractionation curve. The same is true for Fig. 14b, although the sample that plotted on the instantaneous fractionation curve in Fig. 14a now plots below that curve. Thus, in this system, the (226Ra/230Th) ratio of lavas on eruption appears to be relatively insensitive to the (226Ra/230Th) ratio of the mafic parent magma after extensive (>20%) fractionation.
The total amount of fractionation required to generate the most evolved trachyte from the hawaiite parent is 0·47 (47%) (Fig. 12b) and so, for a rate of between 0·00005 and 0·0002/year, this implies a duration of between 10 000 and 2000 years. However, the least-evolved trachyte, 21Lt2b, lies close to the curve corresponding to a fractionation rate of 0·00002/year. The amount of fractionation required to generate this lava from the hawaiite parent is 0·31 (31%), implying a time period of 15000 years for differentiation. All of these timescales are within the absolute timescales of the evolution of the volcano, as defined by 14C dates.
We have also used the model to explore the rates and timescales of fractionation within the Lt2b trachyte series and the results of these calculations are shown in Fig. 14c. The least-evolved lava in this group is 21Lt2, with 18·2 p.p.m. Th. This is one of the three lavas with relatively low Zr, Th and Fe2O3, identified in Fig. 4 as having strong affinity with the Lt2a sequence and, hence, possibly of a different magmatic lineage from the remaining Lt2b lavas. Notwithstanding this difference, appropriate curves originating from 21Lt2 for varying fractionation rates are shown in Fig. 14c. Three samples plot within error of the curve for a fractionation rate of 0·0002/year, whereas two are within error of the instantaneous fractionation curve and two lie below it. These differences almost certainly relate to uncertainties in exact eruption ages and the probability that 21Lt2 is not an appropriate parental composition for all of the lavas. It should be noted that all (226Ra/230Th) ratios have been corrected to 3300 years BP and, if the lavas are older than this, then their initial (226Ra/230Th) ratios would be even lower.
Trace-element abundances indicate that a maximum of 24% (0·24) fractionation is required to generate the most evolved trachyte (15Lt2 with 23·8 p.p.m. Th) from 21Lt2 (18·2 p.p.m. Th). At a rate of 0·0002/year, this corresponds to a timescale of 1200 years and, given that other samples lie below this curve, this timescale should be regarded as a maximum. If, for example, the 7% fractionation implied by major and trace elements within the majority of Lt2b lavas, as discussed previously, occurred at a rate of 0·0002/year, then that would take only 350 years.
These fractionation timescales within the Lt2b lavas are significantly shorter than those for the evolution of trachytes from their hawaiitic parent, in part because the amounts of fractionation are lower, but also because fractionation occurred more rapidly. The 350–1200-year timescale is well within the 2400-year duration of the Lt2 lava sequence, and suggests that differentiation of the Lt2b lavas was occurring as the Lt2 lava sequences were being erupted.
|
AGE OF SANIDINE CRYSTALS |
|---|
The above fractionation rates and timescales relate to whole-rock analyses and the evolution of the magma chamber, but age information can also be derived from crystal–whole-rock pairs. Ages derived from phenocryst assemblages relate to the development of the texture of the rock and may be regarded as ages of crystallization, or phenocryst formation. Conventionally, ages have been calculated from the Ba-normalized equiline diagram, but it is now clear that the assumptions inherent in this approach when applied to plagioclase–groundmass pairs result in erroneous ages because of differences in the partitioning behaviour of Ra and Ba (Cooper et al., 2001
). In the example studied in detail by Cooper et al. (2001), the calculated equilibrium age for plagioclase and groundmass is 500 years, which is substantially less than 3000 years, derived from the Ba-normalized 226Ra–230Th isochron diagram. For alkali feldspar, in which Ra and Ba are both compatible, the effect might be expected to be even more significant and, in the following section, we adopt the method developed by Cooper et al. (2001) and apply it to alkali feldspar, using the relationship between XOr, DBa and DRa from Blundy & Wood (2003).
The curves in Fig. 15 track the in-growth of 226Ra in the whole rock and its decay in alkali feldspar back through time, assuming that the initial (230Th/226Ra) of the alkali feldspar is zero. The curve labelled Meltaf tracks the (226Ra)/Ba ratio of a melt in equilibrium with the alkali feldspar, calculated using the DRa/DBa ratio determined from the Blundy & Wood (2003) equations. The age of the crystals is determined where the two curves cross, and for the whole rock–feldspar pair in 28Lt2 this gives an age of 5200 years, which is 1500 years less than the age determined from a ‘two-point isochron’. A similar set of curves is shown for the sample L33 and, in this sample, the Meltaf and whole-rock curves cross at 2800 years, which is 800 years less than the simple whole-rock–feldspar pair and within the stratigraphic limit of <3280 years.
|
Assessing the uncertainties in these ages is difficult because the equations describing the partitioning behaviour of Ra and Ba are not yet fully developed and the whole rock is assumed to be representative of the liquid from which the feldspar crystallized. In the case of 28Lt2, this is not unreasonable, as the measured partition coefficient for Ba (alkali feldspar/WR) is 11·3 and the predicted value is 10·5. By contrast, L33 has a measured Ba D value of 17·8 against a predicted value of 10·8, implying a lack of equilibrium between the bulk feldspar separate and the whole rock. Moreover, as Ra is compatible in alkali feldspar, a significant amount of the whole-rock Ba and Ra will be accommodated in the phenocrysts, the effect being more marked for Ba because of its higher D value. The consequence of this is that both the (226Ra)/Ba and (230Th)/Ba ratios of the groundmass are significantly greater than that of the whole rock, and this modified composition is probably closer to that of the melt with which the alkali feldspar was originally in equilibrium.
Phenocryst contents of Longonot lavas are low and, for the samples under discussion, are 5% (28Lt2) and 3% (L33). However, the effects of removing even these small fractions of crystals on the whole-rock (226Ra)/Ba ratios are marked, as revealed by the initial (226Ra)/Ba ratios of the curves marked groundmass (5% af) and (3% af) in Fig. 15a and b. However, because of the exponential nature of radioactive decay and in-growth, the effect on the intersection of the crystal age is less dramatic, resulting in an increase from 5200 to 6000 years for 28Lt2 and 2800 to 3200 years for L33. These measures of phenocryst contents are regarded as maxima, and so these latter ages are considered to be maximum limits on the ages of the crystals in these two samples.
The ages determined from modelling the melt in equilibrium with the alkali feldspars are younger than those derived from the ‘two-point isochron’ but the differences between the ages derived from the two approaches are not as great as those revealed by Cooper et al. (2001) for plagioclase–groundmass pairs. This is largely because alkali feldspar crystallization generates very large (226Ra)/Ba ratios in the whole rock and coexisting groundmass relative to the crystals themselves. Moreover, the modelled increase in the (226Ra)/Ba ratio of the liquid in equilibrium with the alkali is small compared with the measured (226Ra)/Ba ratio of the whole rock and very small compared with the (226Ra)/Ba ratio calculated for the groundmass. Hence, the difference in the ages from the two approaches is comparatively small.
Notwithstanding this comparison, the alkali feldspars from 28Lt2 suggest a crystallization age of between 5200 and 6000 years, compared with a model fractionation age (Table 4) of 6700–7500 years. The similarity of the phenocryst age to the maximum stratigraphic age and the slightly older model age for fractionation may relate to the presence of a spread of crystal ages, ranging from possibly as old as 7500 years up to the age of eruption. Moreover, 28Lt2 has incompatible element characteristics comparable with the Lt2a lavas, which also have model fractionation ages of >8000 years. Thus, 28Lt2 may represent a batch of magma inherited from an earlier stage in the evolution of the Longonot system, in which fractionation ceased possibly 7000 years ago but further crystal growth occurred in the intervening 2000–2500 years prior to eruption.
The stratigraphic age constraints for L33 are less precise, but the phenocryst age implies formation during the emplacement of the Lt3 lava flows. However, if the flows are younger than 1000 years, as inferred from field characteristics, then the results would imply, once again, a contribution from crystals that may be up to 2000 years older than eruption.
|
SUMMARY AND CONCLUSIONS |
|---|
The evolution of trachytic magmas at Longonot is dominated by fractional crystallization, with a secondary contribution from magma mixing. Despite the high silica contents of the Longonot trachytes, fractionation rates [(0·5–2) x 10–4/year] are comparable with those from the basaltic system at Ardoukoba [(3·3–3·6) x 10–4/year], and the rates may be >2 x 10–4/year within the trachytes themselves. Crystallization rates are dominantly controlled by the rate of heat loss from the system, whereas differentiation also depends on the rate at which crystals are removed from the magma. The latter is controlled largely by melt viscosity, which is generally higher in more silica-rich compositions. However, the shield morphology of Longonot and the nature of the trachyte lava flows (sheets rather than domes) both testify to the low viscosity of the Longonot magmas, which may, in turn, be related to their high fluorine contents (Table 1). The effect of fluorine in any magma is to break down the network-forming bonds between Si and O, raising the oxygen fugacity and reducing melt viscosity. Thus, the high halogen contents of peralkaline magmas may have an effect on magma evolution, but one that enhances crystal removal by reducing viscosity, rather than one of mobilizing incompatible elements through the formation of halogen complexes (see Bailey & Macdonald, 1975
; Macdonald, 1987; Macdonald et al., 1987). This conclusion is in accordance with that of Heumann & Davies (2002) concerning the evolution of the Olkaria comendites. The possibly slower rates of fractionation inferred for the differentiation of hawaiite to trachyte may reflect a two-stage process in the evolution of the Longonot lavas. Initially, hawaiitic magma differentiates in a large magma reservoir at mid-crustal levels to produce a parental trachyte magma, which is then introduced into the magma reservoir in the immediate sub-volcanic environment, in which the parental trachyte further differentiates into the range of compositions erupted as lavas at the surface.
In more detail, the differences in (226Ra/230Th) disequilibrium between the Lt2a and b lavas, and the inferred storage periods of the two lava groups, suggest that the Lt2a magma was stored in the sub-volcanic system for at least 2500 years before eruption. Crystal ages from 28Lt2, which is compositionally similar to the Lt2a lavas, also suggest storage for possibly 1000 years prior to eruption. However, the implication is that the Lt2a magmas acquired their range of compositions at an early stage and did not undergo further differentiation before they were erupted. Although it is not immediately apparent how the range of compositions could be maintained in an active system for such a period of time, especially given the rapidity with which the younger Lt2b lavas evolved, any alternative model has to explain both the major- and trace-element systematics of the Lt2a group, combined with their state of (226Ra/230Th) secular equilibrium.
The conclusion that the Longonot lava system has evolved within the past 15 000 years is similar to results from other trachyte-dominated centres, such as the Laacher See, and the Azores (Widom et al., 1992; Bourdon et al., 1994), where timescales were based solely on U–Th disequilibria. In the case of Longonot, because the mixed lavas have identical (230Th/232Th) ratios to the trachytes, this time period includes fractionation from a parental hawaiite magma. The (226Ra/230Th) disequilibria in the Lt2b lavas are also consistent with evolution from a hawaiite parent over this timescale. Moreover, as hawaiites are more likely to have been derived from a basaltic parent, this further implies that the Longonot trachytes are ultimately derived from a mafic magma via extensive fractional crystallization, rather than as partial melts of gabbroic underplate. Such an origin is consistent with the similar trace-element characteristics of the trachytes and basalts in Kenya (Fig. 5) and would also explain their excess of 230Th over 238U, as recent basalts from the Kenya rift have similar or greater 230Th excesses (Black et al., 1998; N. W. Rogers, unpublished data). The 230Th excess in the trachytes also contrasts with the secular equilibrium or 238U excess in the Olkaria comendites, which are more probable candidates for crustal melts (Black et al., 1998; Heumann & Davies, 2002).
The fractionation rates and magma storage times implied by the Longonot disequilibria contrast with the evolution of the neighbouring volcanic complex of Olkaria, where both U-series and Rb–Sr isotope analyses suggest rapid crystallization followed by magma storage for periods of up to 22 kyr prior to eruption (Heumann & Davies, 2002). Furthermore, the Longonot storage times are shorter by 1–2 orders of magnitude than the periods of up to 150 kyr inferred for the much larger Long Valley volcanic system (Reid et al., 1997; Heumann et al., 2002).
Comparison with the fractionation rates derived from other centres is difficult because they were obtained by alternative methods. For example, rates are frequently determined by dividing the total volume of an individual volcanic centre by the uncertainty in the crystallization age. Isochrons derived from phenocrysts and glasses are analogous to the sanidine–whole-rock pairs in this study and give a mean age to the development of the preserved texture of the rock. As shown above, such ages are not easily related to fractionation rates as defined here. Notwithstanding this difference in approach, the maximum fractionation rate from Olkaria is given as 2·5 x 10–3 km3/year, which reduces to 5 x 10–4/year when the 5 km3 volume of the system is taken into consideration. This rate is significantly more rapid than that derived for Longonot. The rate for the Long Valley system (7·5 x 10–4 km3/year, Davies & Halliday, 1998) translates to 1·5 x 10–5/year to 1 x 10–6/year when the total volume of the system is taken into account. These rates are somewhat slower than those from Longonot and Ardoukoba, and, given the differences in composition, volume and eruptive styles between these examples, they should be treated with caution.
The link between ages derived from separated crystals and magmatic storage periods is still unclear and any conclusions drawn from such age information on the rates of magmatic evolution remain conjectural. Examples such as Longonot and Ardoukoba, where variations in U-series disequilibrium in the whole rock can be related to magma differentiation, are always going to be the exception but the insight that they provide will be critical to the resolution of these important debates. The similarity between the rates of fractionation for Longonot and Ardoukoba suggest that the rates of evolution in relatively small volcanic systems are possibly related to cooling rate and melt viscosity; the challenge now is to focus on larger-volume volcanic centres, where a similar approach can be applied.
|
APPENDIX: DERIVATION OF THE GOVERNING EQUATIONS FOR COMBINED RADIOACTIVE DECAY AND IN-GROWTH WITH FRACTIONAL CRYSTALLIZATION |
|---|
The rate of change of the radioactive parent, 230Th, and its radioactive daughter, 226Ra, are described by differential equations that account for the changing number of these atoms in the differentiating and ageing liquid, as follows. Consider the general case of a liquid of mass, M, containing a number, N, of atoms of a particular trace element (or isotope) of atomic mass, W. The concentration by mass, C, is therefore
 | (A1) |
M, incorporates N atoms of the trace element, so that the concentration in the crystals is
 | (A2) |
 | (A3) |
 | (A4) |
 | (A5) |
p, the number of parent atoms, P, in a liquid evolving by simultaneous fractional crystallization and radioactive decay changes according to
 | (A6) |
 | (A7) |
The simplest model for the crystallization rate is to assume a constant rate, a; hence
 | (A8) |
Equations (A6) and (A8) are readily solved:
 | (A9a) |
 | (A9b) |
The results can be compared with those for instantaneous fractional crystallization,
 | (A10) |
, where the mass of liquid decreases exponentially:
 | (A11) |
 | (A12) |
 | (A13a) |
d – p 
d (Vigier et al., 1999),
 | (A13b) |
|
ACKNOWLEDGEMENTS |
|---|
Much of the analytical work was carried out by P.J.E. while in receipt of an NERC Ph.D. studentship at the Open University. We would like to thank Louise Thomas and Nathalie Vigier for their comments on, and thank Mary Reid and Phil Leat for perceptive reviews of, an earlier version of this paper. Constructive reviews from Kari Cooper and Arndt Heumann on the subsequent version resulted in further significant improvements. U-series research at the Open University is in part funded by NERC.
|
FOOTNOTES |
|---|
* Corresponding author. Telephone: 01908 652013. Fax: 01908 655151. E-mail: n.w.rogers{at}open.ac.uk
Present address: LGC, Queens Road, Teddington, Middlesex, TW11 0LY, UK. 
|
REFERENCES |
|---|
Bailey, D. K. & Macdonald, R. (1975). Fluorine and chlorine in peralkaline liquids and the need for magma generation in an open system. Mineralogical Magazine 40, 405–414.[Web of Science]
Baker, B. H. (1987). Outline of the petrology of the Kenya rift alkaline province. In: Fitton, J. G. & Upton, B. G. J. (eds) Alkaline Igneous Rocks. Geological Society, London, Special Publications 30, 293–311.
Black, S., Macdonald, R. & Kelly, M. R. (1997). Crustal origin for peralkaline rhyolites from Kenya: evidence from U-series disequilibria and Th isotopes. Journal of Petrology 38, 277–297.[CrossRef][Web of Science]
Black, S., Macdonald, R., Barreiro, B., Dunkley, P. & Smith, M. (1998). Open system alkaline magmatism in northern Kenya: evidence from U-series disequilibria and radiogenic isotopes. Contributions to Mineralogy and Petrology 131, 365–378.
Blundy, J. & Wood, B. (2003). Mineral–melt partitioning of uranium, thorium and their daughters. Reviews in Mineralogy and Geochemistry 52, 59–123.
Bohrson, W. A. & Reid, M. R. (1998). Genesis of evolved ocean island magmas by deep- and shallow-level basement recycling, Socorro Island, Mexico: constraints from Th and other isotope signatures. Journal of Petrology 39, 995–1008.[CrossRef][Web of Science]
Bourdon, B. & Sims, K. W. W. (2003). U-series constraints on intra-plate basaltic magmatism. Reviews in Mineralogy and Geochemistry 52, 215–254.
Bourdon, B., Zindler, A. & Worner, G. (1994). Evolution of the Laacher See magma chamber: evidence from SIMS and TIMS measurements of U–Th disequilibria in minerals and glasses. Earth and Planetary Science Letters 126, 75–90.[CrossRef][Web of Science]
Bourdon, B., Turner, S., Henderson, G. M. & Lundstrom, C. C. (2003). Introduction to U-series geochemistry. Reviews in Mineralogy and Geochemistry 52, 1–21.
Chabaux, F., Othman, D. & Birck, J. L. (1994). A new Ra–Ba chromatographic separation and its applications to Ra mass spectrometric measurement in volcanic rocks. Chemical Geology 114, 191–197.[CrossRef][Web of Science]
Charlier, B. & Zellmer, G. (2000). Some remarks on U–Th mineral ages from igneous rocks with prolonged crystallisation histories. Earth and Planetary Science Letters 183, 457–469.[CrossRef][Web of Science]
Clarke, M. C. G., Woodhall, D., Allen, D. & Darling, G. (1990). Geological, Volcanological and Hydrogeological Controls on the Occurrence of Geothermal Activity in the Area Surrounding Lake Naivasha, Kenya. British Geological Survey & Kenyan Ministry of Energy. Derry & Sons Ltd.
Cohen, A. S. & O'Nions, R. K. (1991). Precise determination of femtogram quantities of radium by thermal ionisation mass spectrometry. Analytical Chemistry 63, 2705–2708.[CrossRef]
Condomines, M., Gauthier, P.-J. & Sigmarsson, O. (2003). Timescales of magma chamber processes and dating of young volcanic rocks. Reviews in Mineralogy and Geochemistry 52, 125–174.
Cooper, K. M., Reid, M. R., Murrell, M. T. & Clague, D. A. (2001). Crystal and magma residence at Kilauea Volcano, Hawaii: 230Th–226Ra dating of the 1955 east rift eruption. Earth and Planetary Science Letters 184, 703–718.[CrossRef][Web of Science]
Davies, G. R. & Halliday, A. N. (1998). Development of the Long Valley rhyolitic magma system: strontium and neodymium isotope evidence from glasses and individual phenocrysts. Geochimica et Cosmochimica Acta 62, 3561–3574.[CrossRef][Web of Science]
Davies, G. R. & Macdonald, R. (1987). Crustal influences in the petrogenesis of the Naivasha basalt–comendite complex: combined trace element and Sr–Nd–Pb isotope constraints. Journal of Petrology 28, 1009–1031.
Green, T. H. & Watson, E. B. (1982). Crystallization of apatite in natural magmas under high pressure, hydrous conditions, with particular reference to ‘orogenic’ rock series. Contributions to Mineralogy and Petrology 79, 96–105.[CrossRef][Web of Science]
Guo, J. & Green T. H. (1989). Barium partitioning between alkali feldspar and silicate melts at high temperature and pressure. Contributions to Mineralogy and Petrology 102, 328–335.[CrossRef][Web of Science]
Halliday, A. N., Mahood, G., Holden, P., Metz, J. M., Dempster, T. J. & Davidson, J. P. (1989). Evidence for long residence times of rhyolite magma in the Long Valley magma system: the isotopic record in the precaldera rhyolite lavas of Glass Mountain. Earth and Planetary Science Letters 94, 274–290.[CrossRef][Web of Science]
Hart, S. R. (1984). A large scale isotope anomaly in the Southern Hemisphere mantle. Nature 309, 753–757.[CrossRef]
Hawkesworth, C. J., Blake, S., Evans, P., Hughes, R., Macdonald, R., Thomas, L. E., Turner, S. P. & Zellmer, G. (2000). Time scales of crystal fractionation in magma chambers: integrating physical, isotopic and geochemical perspectives. Journal of Petrology 41, 991–1006.
Heath, E., Turner, S. P., Macdonald, R., Hawkesworth, C. J. & van Calsteren, P. (1998). Long magma residence times at an island arc volcano (Soufrière, St Vincent) in the Lesser Antilles: evidence from 238U–230Th isochron dating. Earth and Planetary Science Letters 160, 49–63.[CrossRef][Web of Science]
Heumann, A. & Davies, G. R. (2002). U–Th disequilibrium and Rb–Sr age constraints on the magmatic evolution of peralkaline rhyolites from Kenya. Journal of Petrology 43, 557–577.
Heumann, A., Davies, G. R. & Elliott, T. (2002). Crystallisation history of rhyolites at Long Valley, California, inferred from combined U-series and Rb–Sr isotope systematics. Geochimica et Cosmochimica Acta 66, 1821–1837.[CrossRef][Web of Science]
Icenhower, J. & London, D. (1996). Experimental partitioning of Rb, Cs, Sr and Ba between alkali feldspar and peraluminous melt. American Mineralogist 81, 719–734.[Abstract]
Long, P. E. (1978). Experimental determination of partition coefficients for Rb, Sr and Ba between alkali feldspar and silicate liquid. Geochimica et Cosmochimica Acta 42, 833–846.[CrossRef][Web of Science]
Lundstrom, C. C. (2003). Uranium-series disequilibria in mid-ocean ridge basalts: observations and models of basalt genesis. Reviews in Mineralogy and Geochemistry 52, 175–214.
Macdonald, R. (1987). Quaternary peralkaline silicic rocks and caldera volcanoes of Kenya. In: Fitton, J. G. & Upton, B. G. J. (eds) Alkaline Igneous Rocks. Geological Society, London, Special Publications 30, 293–311.
Macdonald, R., Davies, G. R., Bliss, C. M., Leat, P. T., Bailey, D. K. & Smith, R. L. (1987). Geochemistry of high-silica peralkaline rhyolites, Naivasha, Kenya rift valley. Journal of Petrology 28, 979–1008.
Macdonald, R., Rogers, N. W., Fitton, J. G., Black, S. & Smith, M. (2001). Plume–lithosphere interactions in the generation of the basalts of the Kenya rift, East Africa. Journal of Petrology 42, 877–900.
Mechie, M. (1994). Crustal structure beneath the Kenya rift from axial profile data. Tectonophysics 236, 179–199.[CrossRef][Web of Science]
Reid, M. R., Coath, C. D., Harrison, T. M. & McKeegan, K. D. (1997). Prolonged residence times for the youngest rhyolites associated with Long Valley Caldera 230Th–238U ion microprobe dating of young zircons. Earth and Planetary Science Letters 150, 27–39.[CrossRef][Web of Science]
Rogers, N. W., Macdonald, R., Fitton, J. G., George, R. M. M., Smith, M. & Barreiro, B. A. (2000). Two mantle plumes beneath the East African rift system: Sr, Nd and Pb isotope evidence from Kenya rift basalts. Earth and Planetary Science Letters 176, 387–400.[CrossRef][Web of Science]
Scott, S. C. (1980). The geology of Longonot volcano, Central Kenya: a question of volumes. Philosophical Transactions of the Royal Society of London 296, 437–465.
Scott, S. C. & Skilling, I. P. (1999). The role of tephrachronology in recognition of synchronous caldera collapse at Quaternary volcanoes in the South Kenya rift. Geological Society, London, Special Publications 161, 47–67.
Skilling, I. P. (1988). The geological evolution of Suswa volcano, Kenya. Ph.D. thesis, University of Lancaster.
Smith, M. & Mosely, P. (1993). Crustal heterogeneity and basement influence on the development of the Kenya rift, East Africa. Tectonics 12, 591–606.[Web of Science]
Sparks, R. S. J. (2003). Forecasting volcanic eruptions. Earth and Planetary Science Letters 210, 1–15.[CrossRef][Web of Science]
Turner, S., Hawkesworth, C. J., Rogers, N. & King, P. (1997). U–Th isotope disequilibria and ocean island basalt generation in the Azores. Chemical Geology 139, 145–164.[CrossRef][Web of Science]
Turner, S., Bourdon, B., Hawkesworth, C. & Evans, P. (2000). 226Ra–230Th evidence for multiple dehydration events, rapid melt ascent and the time scales of differentiation beneath the Tonga–Kermadec island arc. Earth and Planetary Science Letters 179, 581–593.[CrossRef][Web of Science]
Turner, S., George, R., Jerram, D. A., Carpenter, N. & Hawkesworth, C. (2003). Case studies of plagioclase growth and residence times in island arc lavas from Tonga and the Lesser Antilles, and a model to reconcile discordant age information. Earth and Planetary Science Letters 214, 279–294.[CrossRef][Web of Science]
van Calsteren, P. & Schweiters, J. B. (1995). Performance of a thermal ionization mass spectrometer with a deceleration lens system and post deceleration detector selection. International Journal of Mass Spectrometry and Ion Processes 146, 119–129.[CrossRef]
Vigier, N., Bourdon, B. Joron, J. L. & Allègre, C. J. (1999). U-decay series and trace element systematics in the 1978 eruption of Ardoukoba, Asal rift: timescales of magma crystallisation. Earth and Planetary Science Letters 174, 81–97.[CrossRef][Web of Science]
Volpe, A. M., Olivares, J. A. & Murrell, M. T. (1991). Determination of radium isotope ratios and abundances in geologic samples by thermal ionisation mass spectrometry. Analytical Chemistry 63, 913–916.[Medline]
Watson, E. B. (1979). Apatite saturation in basic to intermediate magmas. Geophysicsal Research Letters 6, 937–940.
Watson, E. B. & Harrison, T. M. (1984). Accessory minerals and the geochemical evolution of crustal magmatic systems: a summary and prospectus of experimental approaches. Physics of the Earth and Planetary Interiors 35, 19–30.[CrossRef][Web of Science]
Widom, E., Schminke, H. U. & Gill, J. B. (1992). Process and timescales in the evolution of a chemically zoned trachyte: Fogo A, São Miguel, Azores. Contributions to Mineralogy and Petrology 111, 311–328.[CrossRef][Web of Science]
Williams, R. W. & Gill, J. B. (1992). Th isotope and U series disequilibria in some alkali basalts. Geophysical Research Letters 19, 139–142.[Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Macdonald and B. Baginski The central Kenya peralkaline province: a unique assemblage of magmatic systems Mineralogical Magazine, June 15, 2009; 73(1): 1 - 16. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Macdonald, H. E. Belkin, J. G. Fitton, N. W. Rogers, K. Nejbert, A. G. Tindle, and A. S. Marshall The Roles of Fractional Crystallization, Magma Mixing, Crystal Mush Remobilization and Volatile-Melt Interactions in the Genesis of a Young Basalt-Peralkaline Rhyolite Suite, the Greater Olkaria Volcanic Complex, Kenya Rift Valley J. Petrology, August 1, 2008; 49(8): 1515 - 1547. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. M. Cooper and M. R. Reid Uranium-series Crystal Ages Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 479 - 544. [Full Text] [PDF]
|
|
|
|
 |
|
 S. A. Miller, D. S. Burnett, P. D. Asimow, D. L. Phinney, and I. D. Hutcheon Experimental study of radium partitioning between anorthite and melt at 1 atm American Mineralogist, August 1, 2007; 92(8-9): 1535 - 1538. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. LOWENSTERN, B. L. A. CHARLIER, M. A. CLYNNE, and J. L. WOODEN Extreme U-Th Disequilibrium in Rift-Related Basalts, Rhyolites and Granophyric Granite and the Timescale of Rhyolite Generation, Intrusion and Crystallization at Alid Volcanic Center, Eritrea J. Petrology, November 1, 2006; 47(11): 2105 - 2122. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||