Magmatic cycles and formation of the upper oceanic crust at spreading centers: Geochemical study of a continuous extrusive section in the Oman ophiolite



[1] The data and analysis presented in this paper provide an assessment of lava morphologies and the geochemistry of lavas from the Oman ophiolite. In order to provide detailed constraints on the construction of the upper oceanic crust, a continuous volcanic transect (300 m-thick) was sampled at high-frequency in the Semail ophiolite along Wadi Shaffan. The Wadi Shaffan section is composed mainly of pillow lavas interbedded with massive flows and occasional hyaloclastites. The sampling performed along Wadi Shaffan implies temporal variations in the activity of the ridge. The section is characterized by chemical compositions consistent with those of V1-Geotimes volcanism. The Wadi Shaffan transect was built through two main petrological and geochemical sequences of volcanic activity. Trace element ratios (e.g. Zr/Nb and La/Yb) allow us to distinguish two main sequences with two different parental magmas. Differences in the degree of partial melting are required to explain these trace element ratio variations. Beyond these differences in parent melt composition, variations in trace element abundances (TiO2, Zr, REE) involve differentiation processes prior to emplacement. In the lower sequence, less differentiated lavas are in the upper part of the cycle. Magma mixing is proposed to explain this reversed geochemical evolution through time. In the upper sequence, geochemical analysis suggests a different magma chamber process. This sequence consists of multiple events of magma emplacement. Variations in trace element abundance suggest four magmatic cycles. Each magmatic cycle is characterized by primitive lavas evolving to more differentiated lavas with time. The upper sequence lavas appear to be in equilibrium with clinopyroxene and lower sills from the MTZ (Mantle-Crust Transition Zone) and with lower gabbros. We propose a model in which the upper sequence lavas were directly derived from the MTZ and lower gabbro sills and then transported to the surface without interaction with higher crustal levels.

1. Introduction: Magmatic Cycles in Oceanic Basement From Petrophysical Properties and Geochemistry

[2] The crustal accretion and volcanic stratigraphy at mid-ocean ridges (MOR) is of primary importance in understanding magma transport and the processes of igneous crustal accretion. In the past decade, the understanding of plumbing in magmatic systems at spreading ridges has been improved significantly by geophysical observations of present-day ridges and by geological studies of ophiolites. Lava emplacement studies in oceans have often been guided by geophysical surveys indicating the presence of tectono-magmatic cycles [Kappel and Ryan, 1986; Gente, 1987]. Such analyses have been conducted, for instance, along the Galapagos Rift [Lonsdale, 1977] and the East Pacific Rise [Ballard et al., 1981; Francheteau and Ballard, 1983], and from deep oceanic drilling [Ayadi et al., 1998]. Several authors have underlined the strong link between lava morphology and spreading rate [Macdonald, 1983; Bonatti and Harrisson, 1988; Perfit and Chadwick, 1998; Brewer et al., 1999]. The proportion of flows increases and the proportion of pillows decreases with increasing spreading rate. However, many aspects of the temporal variations in melt supply process and magma plumbing systems beneath mid-ocean ridges are not well understood. The best way to access this information is to study continuous time series. Continuous sections of volcanic rocks in oceanic basement can be found in tectonic windows by high-resolution dredging, or by drill hole studies.

[3] The drill hole approach allows a temporal study of variations in accretion processes. In the Ocean Drilling Program (ODP, previously DSDP: Deep Sea Drilling Project), deep holes have been drilled in oceanic crust formed at different spreading rates. Slow spreading crust was drilled in the Kane area of the Mid-Atlantic Ridge [Hyndman and Salisbury, 1984, Bartetzko et al., 2001] and intermediate spreading crust was drilled south of the Costa Rica Ridge [Pezard et al., 1992; Ayadi et al., 1998; Brewer et al., 1999]. Multidisciplinary studies have been necessary at each of these two reference sites to investigate accretion processes. Continuous logging data have demonstrated changes in magmatic processes at the ridge axis [Hyndman and Salisbury, 1983; Pezard et al., 1992; Bartetzko et al., 2001]. In the examples considered (Figure 1), the recorded physical properties have been interpreted in terms of volcanic and magmatic processes. To illustrate this, data from two deep holes are presented in the following paragraph, and used as references to investigate connections between downhole physical properties and geochemistry.

Figure 1.

Location and electrical resistivity logging measurements (solid lines) and TiO2 (open circles) analyses from 500 meter sections in DSDP/ODP Hole 395A and DSDP/ODP Hole 504B. Depth is in meters below seafloor (mbsf). Dashed lines are estimated best-fit trends to data for illustrative purposes. (a) Electrical resistivity (SFL: Shallow Spherically Focused) logging data in Hole 395A. The resistivity trends of Hyndman and Salisbury [1983] and Bartetzko et al. [2001] are highlighted with blue segments and magmatic cycles are noted from I to X. The two black stars indicate the presence of aquifers. TiO2 contents are shown with open circles. (b) Electrical resistivity (LLD: Deep Laterolog) logging data in Hole 504B. Trends in TiO2 contents are underlined with light blue shading. Only the upper 600 m are reported, the total depth reaches 2211 mbsf, in the sheeted dike complex. MF for Massive Flow. Magmatic cycles proposed by Pezard et al. [1992] are noted VS1 to VS3. (c) Magnetic susceptibility measured on discrete samples from Wadi Shaffan. The elevation is increasing up-section from the sheeted dykes to lava flows. A simplified lithological column on the right for Wadi Shaffan is used for reference, and shows the 42 different units defined by Einaudi et al. [2000]. TiO2 contents are shown with red circles (unit 2–12), green triangles (unit 13–18) and blue diamonds (unit 19–42).

[4] Along a slow spreading ridge (Mid-Atlantic Ridge, 2.5 cm/yr at full rate, DSDP/ODP Hole 395A), the basaltic section is mainly composed of pillows with numerous breccias [Hyndman and Salisbury, 1984]. In Hole 395A, variations in downhole measurements have been used and interpreted in terms of volcanic cycles [Hyndman and Salisbury, 1983; Bartetzko et al., 2001]. The electrical resistivity measurements reveal a succession of cycles (Figure 1). Each cycle starts with high resistivity values, which gradually evolve toward lower values. Each cycle was postulated to start with the eruption of flows or resistive pillows (corresponding to high resistivity values), followed by more conductive pillows. The end of each cycle corresponds with the occurrence of breccias. However, these electrical cycles could not be correlated with geochemical variations.

[5] At an intermediate spreading rate (Costa Rica Ridge, 7 cm/yr at full rate, DSDP-ODP Hole 504B), the basaltic section has been described as composed of pillows and flows [Adamson, 1985]. Two volcanic sequences have been postulated to explain the architecture of the volcanic section penetrated in Hole 504B [Pezard et al., 1992]. Both magmatic sequences start with the emplacement of a massive unit followed by pillows. The electrical resistivity measurements allow the discrimination of massive units from pillows (Figure 1). The electrical resistivity signal does not record gradual changes like those observed in ODP Hole 395A. In contrast, TiO2 changes gradually, either decreasing or increasing with decreasing depth. Unfortunately, due to low core recovery (less than 20% on average), the exact location of each sample is often unknown, and as a consequence, high-resolution comparisons between logging data and geochemistry are impossible.

[6] At fast spreading ridges, the volcanic architecture of the crust is still poorly known, because of the lack of drillholes. As a consequence very little is known about the temporal variations in the geochemistry of MOR Basalts (MORB) erupted at a fast spreading ridge although along-axis geochemical variations are well documented [Estoubleau et al., 1983; Reynolds et al., 1992; Smith and Perfit, 1994; Niu et al., 1996; Fornari et al., 1998; Regelous et al., 1999; Carbotte et al., 2000]. The approach developed here is to investigate a vertical section through a structure which is interpreted to have been emplaced at the axis of a fast spreading ridge: the Oman Ophiolite [Nicolas et al., 2000, and references therein].

[7] As opposed to ocean crustal drilling that often yields discontinuous core, well-exposed ophiolites, like the Oman Ophiolite, provide a unique opportunity to study continuous sections of oceanic basalts. An integrated study combining geochemistry and physical properties has been conducted on rocks from the Oman Ophiolite along the Wadi Shaffan transect (Figures 2 and 3). In order to compare this section to those drilled in the ocean and where downhole logs were recorded, a complete set of physical property measurements has been performed on minicores sampled along Wadi Shaffan [Einaudi et al., 2000; Pezard et al., 2000]. In these previous studies we measured physical properties on Wadi Shaffan samples to evaluate whether physical properties can be used as a proxy to identify individual magmatic units, and to integrate these variations in terms of temporal changes in volcanic activity. All the measured petrophysical parameters record variations to various extents [Einaudi et al., 2000; Pezard et al., 2000], but the most sensitive parameter appears to be magnetic susceptibility which records wide variations along our Wadi Shaffan section (Figure 1c).

Figure 2.

Location of the Wadi Shaffan transect (a) The Sultanate of Oman (dark gray area) is at the eastern tip of the Arabian plate. (b) The ophiolite (dark gray area) defines the northern coast of the Sultanate of Oman. The Wadi Shaffan transect is in the Sarami massif, in the central part of the ophiolite. (c) Detailed location of the Wadi Shaffan outcrop (black circle) from Yanqul 1/100000 geological map [Villey et al., 1984].

Figure 3.

Stratigraphic section of Wadi Shaffan. The Wadi Shaffan transect is presented in terms of relative elevation, starting from a fault zone located at the base. A total of 42 lithological units were identified and measured. Stars represent sample locations. All samples have been analyzed in terms of physical properties [Einaudi et al., 2000; Pezard et al., 2000]. Black stars: geochemical analysis from Einaudi et al. [2000]; gray stars, new analysis; white stars, no geochemical analysis.

[8] The high frequency sampling performed along Wadi Shaffan provides a temporal relationship between contiguous samples, which reflects temporal variations in the activity of the ridge. The mineralogical and geochemical variations of the Wadi Shaffan lavas are investigated and integrated in terms of emplacement and magma chamber processes.

2. Setting: Wadi Shaffan Field Characteristics

2.1. Oman Ophiolite

[9] The Semail Ophiolite is a 600 km long, 100 to 150 km wide and 5 to 10 km thick nappe along the Gulf of Oman (Figure 2). The ophiolite has been divided into blocks, most of them composed of a complete ophiolitic section, from the mantle to volcanics. The Semail volcanic pile was produced between 97.9 and 93.5 Ma [Tilton et al., 1981; Hacker and Gnoss, 1997] and obduction started during the Late Cretaceous [Alleman and Peters, 1972; Lippard et al., 1986]. Several studies have suggested that the Oman Ophiolite was formed along an intermediate to fast spreading ridge [Pearce et al., 1981; Tilton et al., 1981; Nicolas, 1989; MacLeod and Rothery, 1992; Nicolas et al., 1994].

2.2. Wadi Shaffan Section

[10] Our study focuses on a traverse through the extrusive sequence from the Sarami massif (Figure 2). Excellent and continuous exposure of basalts along the Shaffan river (Wadi Shaffan), as well as the lack of tectonic deformation, allowed the study of spatial, and hence temporal variations of extruded lavas. The sampled transect represents a nearly 300 m-thick section (Figure 3). The basal Wadi Shaffan flows are geometrically consistent in orientation with the underlying sheeted dikes, i.e., lava flows are perpendicular to the sheeted dike complex. However, our sampled section probably does not correspond to the entire extrusive sequence accreted at the axis of the spreading ridge, as the contact between extrusives and sheeted dikes complex is marked by faults. In order to obtain a section as continuous as possible, our sampling begins immediately above this faulted interval and proceeds upward.

[11] Lava samples were collected on a flow-by-flow basis along a continuous section exposed along Wadi Shaffan. The sampling strategy was based on the characterization of each lithological unit, that is, each eruptive event, and the freshest possible samples were chosen. At least two samples per unit were collected, leading to a total of more than 100 samples along the 900 m outcrop, resulting in a sample section with a structural thickness of 280 m (Figure 3). On the basis of abrupt changes in lithology or size of the pillows or structural surface occurrence (Figure 4), we distinguished 42 lithological units along the Wadi Shaffan section. The lateral continuity is poor, so most of the time, units can be followed for only a few tens of meters.

Figure 4.

Field photographs: (b), (c) lower sequence, lithologic units are underlined in Figure 4b. Note the late dike that cuts the whole lower section and illustrates the structural continuity of the units. The interval shows in this picture is in black in the summarized lithologic column (Figure 4a). (e), (f) represent part of the upper sequence. Lithologic units are underlined in Figure 4f. The interval shown in this picture is in black in the summary lithologic column (Figure 4d).

[12] Pillow lavas are the most common lithology in the investigated section (76% of the total thickness). They occur as rounded or elongate tubes (30 cm to 1 m diameter). The largest elongated pillows were found near the base of the section (Unit 2; Figures 3, 4b, and 4c). Interpillow spaces are filled in most units with hyaloclastites consisting of small basalt fragments, rare altered glass and red jasper in a clay-rich mineral matrix, especially in the lower 85 m of the sequence. Above this depth, interpillow material consists of chlorite, red jasper or empty space. Radial patterns of fracturing are common in the cuspate pillows, especially in void upper part of the section. Along the entire section, we found no sediment, or umber horizons.

[13] Massive flows (22% of the total thickness) are the second most common lithology encountered. The thickness of massive flows varies from one to five meters, and some of them have columnar jointing (Units 13 and 24). In the field, the main difficulty is to distinguish massive flows from intrusive sills. Most of the massive units are interpreted as flows. The principal criterion used was the presence of breccias at the bottom part of the massive unit. This thin brecciated interval appears to be linked to massive flow emplacement. For one unit (Unit 22), smooth and draped surfaces (similar to pahoehoe flows) are present at the top of the massive flow and indicate the effusive origin of this unit. In the topmost part of the section (Units 40 and 41), some massive units may correspond to intrusive sills.

[14] The section is composed of 2% hyaloclastites and dykes. Hyaloclastites occur in several thin intervals. Interestingly, these layers occur at the base of massive units and may be linked to their emplacement.

[15] At the topmost part of the section and slightly above the vertical section, a volcanic vent was found (Figure 5). This is a 20 m-high and 200 m-wide semicircular volcano. Only half of the volcano is well preserved, the easternmost part of this volcanic edifice having been eroded. The internal flank is characterized to the west by a radial, fan-wise distribution columnar jointing, dipping inward to the core of the dome. In the central part, a small hill (about 2 m high) consists of yellowish coarse grained, highly crystalline rock, which may correspond to the vent. Pillow lavas and tubes are recognized in the outer flanks, and suggest that this dome had been a volcanic point source. The volcano is interpreted as the very top of the Wadi Shaffan section, representative of the last volcanic events in this region.

Figure 5.

Pictures showing the volcanic point source (ISOV, Isolated Volcano). (a), (b) This structure appears as a 20 m-high and 200 m wide semicircular volcano. The internal flank is characterized to by a radial, fan-wise distribution of columns, dipping inward to the core of the dome. This structure is underlined in Figure 5c.

[16] The extrusives dip between 45 degrees at the base of the section to about 10 degrees toward the top (Figures 3 and 4e, and 4f). On the basis of field mapping, it is difficult to determine whether these variations are a primary characteristic or related to the tectonic emplacement of the ophiolite. However, this change in tilting is within an undisturbed zone (Units 28–29). Such variations in tilting within extrusive section are expected in oceanic crust [Pezard et al., 1992; Schouten and Denham, 2000; Karson et al., 2002], where they are interpreted as the vertical subsidence of the volcanic units, to accommodate the near axial thickening of the basaltic lavas.

3. Petrography

[17] The identification of igneous texture and magmatic mineral phases in the samples is generally possible, in spite of varying degrees of hydrothermal alteration. As previously described [Alabaster and Pearce, 1985; Lippard et al., 1986; Pflumio, 1991; Regba et al., 1991], the Oman lavas have suffered pervasive multistage alteration. The most significant feature of our transect is the large petrographical and mineralogical heterogeneity identified in thin sections (Figure 6). Within a meter (between Units 12 and 13), significant changes in texture, primary and secondary mineralogy are observed. This abrupt change in petrography is used in the following discussion to distinguish between the lower sequence (Unit 1 to 12) and the upper one (Unit 13 to 42). The textural and primary mineralogical changes between the lower and upper sequence cannot be related to a sampling bias because these petrological and mineralogical variations are observed for both pillows and massive units. Therefore they are not simply related to cooling, but most probably to a change in the magma chamber.

Figure 6.

Primary and secondary mineralogical variations summarized according to lithology and relative elevation along the Wadi Shaffan transect, with selected thin section micrographs. Dashed lines indicate the occurrence of a given mineral phase in lesser amounts than solid lines. Both primary and secondary minerals show a significant change in abundance and nature at 85 m (Unit 13). Clinopyroxene phenocryst microprobe analyses are reported in the Wollastonite-Enstatite-Ferrosilite diagram. Representative clinopyroxene analyses are reported in Table 1. A simplified lithological column of the Wadi Shaffan section is shown.

3.1. Lower Sequence: From the Base of the Sequence to Unit 12

[18] The lower sequence lavas are generally aphyric or sparsely phyric. The groundmass is mainly microlitic to cryptocrystalline with abundant plumose quenching and spherulitic features. Overall, the groundmass is red-brown in thin section due to tiny inclusions of oxy-hydroxides. The primary mineral phases in the groundmass are plagioclase and minor clinopyroxene. These minerals present the radiating and acicular shape typical of rapidly cooled oceanic lavas. Although phenocrysts are rare, some plagioclase and unaltered clinopyroxene occur locally as clusters. Pervasive alteration formed iron oxi-hydroxides, quartz, and chlorite in the groundmass (Figure 6). Plagioclase microlites are totally replaced by albite. Vesicles and small veins are filled with quartz, chlorite, celadonite and more rarely zeolites.

3.2. Upper Sequence: From Unit 13 to the Top of the Sequence

[19] The upper sequence lavas are phyric and rarely microlitic with an increase in size of the plagioclase microlites compared to the lower sequence basalts. A mineralogical assemblage dominated by plagioclase, clinopyroxene and lesser amounts of olivine characterizes the upper sequence lavas (Figure 6). Clinopyroxene phenocrysts are abundant, and are the only primary igneous phase to be preserved. Phenocrysts are mainly low Na-low Ti augites (avg. Wo43En47, Figure 6). Their compositions (Table 1) plot along the Geotimes - Lasail trend defined by Alabaster et al. [1982], in both Geotimes and Lasail fields. Euhedral olivine phenocrysts are observed in some units (Units 13–17, 21–22, 30). These olivines are pseudomorphed by secondary phases. The occurrence of olivine phenocrysts is indicative of a more primitive composition for the upper section lavas compared to the lower sequence one. Titanomagnetite occurs as interstitial microcrysts, and in a few cases as subhedral phenocrysts. Secondary mineralogy also changes abruptly at the top of Unit 12, where calcite appears replacing olivine and filling vesicles. Plagioclase crystals are replaced by zeolites and albite. Vesicles are filled with zeolite, chlorite, celadonite and locally quartz in the uppermost part of the section.

Table 1. Representative Clinopyroxene Analyses
SampleSH M 22SH M 26SH M 39SH M 44SH M 64SH M 67SH M 77B

[20] The degree and nature of alteration is variable and directly related to the location of the sample within the transect (Figure 6). As in DSDP/ODP Hole 504B [Honnorez et al., 1983; Alt et al., 1996], and in the Troodos Ophiolite [Bednarz and Schmincke, 1994; Gillis and Robinson, 1990], an abrupt change in alteration style occurs in the Wadi Shaffan section. Such a transition has been proposed to result from the presence of a permeability barrier [Honnorez et al., 1983; Pezard, 1990]. The massive flow of Unit 13 might constitute this barrier, separating high temperature alteration in the lower sequence from lower temperature alteration in the upper sequence. Along the Wadi Shaffan section, a complex low-grade metamorphic history involving several alteration phases related to successive magmatic episodes and off-axis hydrothermal circulation is present, similar to that described for the entire Oman Ophiolite [Pflumio, 1991; Regba et al., 1991].

[21] The variations in primary mineralogy between the upper and lower sequence (e.g. olivine or not) suggest variations in the degree of differentiation and magmatic processes over time. Similarly, abundance of phenocrysts in the upper sequence could indicate a relatively long residence time in magma chamber which could account for other differences in magmatic processes between the lower and upper sequence. To investigate these differences, whole rock geochemical analysis were performed on samples from the entire section.

4. Geochemistry: Results

4.1. Analytical Techniques

[22] In this study, two geochemical analysis data sets are used (Table 2). The first data set includes 60 analyses. For these samples, major and trace elements were determined by X-Ray Fluorescence (XRF) at the University of Leicester, following the method of Harvey [1989]. Rare Earth Elements (REE) were determined for a subset of 21 samples, by inductively coupled plasma-spectrometry (ICP), using a JY-Ultima-2 spectrometer at the University of Leicester [Harvey et al., 1996]. This data set is complimented by 23 analyses from Einaudi et al. [2000].

Table 2. Geochemical Analysis of the Wadi Shaffan Lavas
SampleSH M 84SH M 85SH M 01SH M 02SH A 03SH A 07SH A 05SH M 05SH M 04SH M 07SH M 09SH M 10SH A 12SH M 11SH M 12SH A 14SH A 16SH M 14SH M 15SH A 19SH M 17SH M 16SH A 22SH M 19SH A 23SH M 20SH M 22SH M 25SH M 26SH M 23SH M 24SH A 26SH M 27SH M 28SH M 29SH M 30SH M 31SH A 29SH M 33SH M 34SH M 36SH M 37SH A 31SH M 39SH M 38SH A 32SH M 42SH A 33SH M 44SH M 45SH M 49SH A 35SH M 50SH M 51SH M 53SH M 52SH M 86SH M 55SH M 56SH M 57SH M 58SH A 37SH A 38SH M 60SH M 61SH A 39SH A 40SH M 64SH M 65SH M 66SH A 43SH M 67SH A 44SH M 70SH M 71SH M 72SH M 73SH M 74SH M 75SH M 76SH M 78SH M 77B
   *      * * *    * *  * *  * *    *  *  * *    * *      *      *  *   *    *  ** 
  1. a

    Major elements are in oxide-weight percentage and trace element concentrations in ppm. Mg# is calculated as the atomic ratio of Mg2+/(Mg2+ + Fe2+), assuming Fe2O3/FeO = 0.15. Samples with * are from Einaudi et al. [2000]. These two data sets are distinguished in the following multielement diagrams: samples from Einaudi et al. [2000] are called (ICP-MS) and new analyses are called (XRF).

Hf  4.18      3.38 3.28 3.55    3.09 3.10  2.98 0.86  0.91 1.03    1.78  2.16  1.84 3.49    1.54 2.26      3.27      1.43  1.72   1.75    3.21  3.385.24 
Ta  0.22      0.16 0.16 0.18    0.16 0.16  0.16 0.07  0.08 0.09    0.15  0.16  0.15 0.32    0.12 0.22      0.25      0.11  0.15   0.14    0.27  0.270.43 
Pr  2.99 3.84  3.09 2.462.262.38 2.48  1.97 2.21 2.25 2.262.25 0.77  0.84 0.80 0.87  1.471.96 1.782.09 1.552.652.90 1.19  1.25 1.99 2.21 1.95 1.712.56 2.13 2.692.23 1.30 1.931.39 1.64 1.43 4.55  2.66  2.623.683.26
Sm  5.51 7.67  6.20 4.773.774.45 4.95  3.91 4.34 4.38 4.094.29 1.54  1.64 1.42 1.76  2.503.11 2.593.30 2.663.954.33 2.31  2.32 3.16 3.70 3.42 3.063.70 3.59 4.194.27 1.95 3.262.60 2.69 2.45 7.25  3.94  3.965.696.05
Eu  1.85 2.11  2.50 1.591.511.64 1.70  1.73 1.48 1.51 1.691.63 0.55  0.59 0.61 0.73  0.900.99 1.061.10 0.871.481.62 1.11  0.74 1.16 1.49 1.33 1.281.31 1.50 1.581.18 0.80 1.090.82 1.08 0.86 2.11  1.37  1.402.031.70
Gd  7.17 8.91  10.25 6.005.595.68 5.91  6.36 5.21 5.33 6.115.72 1.75  1.92 1.79 2.56  2.993.80 3.613.64 3.205.305.42 3.83  2.52 3.77 5.49 4.70 4.334.73 5.49 5.694.78 2.44 4.063.04 3.96 2.96 8.61  4.81  4.947.316.71
Tb  1.22      1.03 1.05 1.09    0.97 1.00  0.96 0.29  0.31 0.32    0.53  0.61  0.52 0.95    0.45 0.65      0.85      0.44  0.52   0.54    0.84  0.931.26 
Dy  8.78 10.40  12.04 7.606.737.08 7.74  7.66 6.59 6.88 7.186.90 2.04  2.20 2.28 3.51  3.743.86 4.213.57 3.556.056.55 4.86  3.16 4.64 6.30 5.82 5.225.65 6.27 6.295.64 3.20 3.583.65 4.52 3.52 8.86  5.67  5.628.307.83
Ho  1.92      1.64 1.57 1.74    1.47 1.58  1.50 0.50  0.52 0.51    0.88  0.94  0.84 1.48    0.74 1.06      1.28      0.77  0.78   0.79    1.22  1.211.93 
Er  5.62 7.12  7.78 4.684.404.45 4.80  4.95 3.94 4.26 4.634.03 1.30  1.41 1.43 2.26  2.182.16 2.542.25 2.203.753.87 3.13  1.88 2.71 4.07 3.68 3.213.55 4.00 3.903.81 2.03 2.152.10 2.85 2.08 5.67  3.40  3.264.925.03
Tm  0.86      0.74 0.74 0.76    0.65 0.69  0.68 0.21  0.24 0.22    0.33  0.41  0.37 0.59    0.32 0.47      0.61      0.32  0.34   0.36    0.58  0.600.90 
Yb  5.81 6.32  8.02 4.824.194.74 5.08  5.01 4.25 4.35 4.564.21 1.24  1.47 1.46 1.99  2.372.17 2.762.23 2.523.334.20 2.85  1.97 2.96 3.98 3.27 3.253.69 3.97 3.663.37 1.98 2.232.11 2.52 2.18 5.29  3.64  3.485.314.37
Lu  0.99 0.92  1.19 0.780.610.79 0.81  0.74 0.74 0.77 0.700.72 0.22  0.26 0.26 0.28  0.380.35 0.430.37 0.370.490.62 0.40  0.29 0.45 0.60 0.51 0.490.58 0.60 0.530.43 0.33 0.360.33 0.39 0.33 0.84  0.55  0.530.770.67

4.2. Major and Transition Elements

[23] Loss on ignition (LOI) values range from 0.4 to 3.4% in the lower sequence and from 1.3 to 11% in the upper sequence, though for 53% of the samples (Table 2), LOI is less than 3%. These values reflect the pervasive alteration observed in thin sections. Alteration leads to strong modifications CaO, Na2O and K2O in the primary lava composition, and to a lesser extent, in MgO and SiO2 (Table 2). Nevertheless, despite some scatter in the SiO2 concentration due to alteration, two distinctive differentiation trends appear on Figure 7, which shows the variation in TiO2 with SiO2. The first trend has an inverse correlation between SiO2 and TiO2 and corresponds to the lower sequence lavas. The second displays an increase in TiO2 content with increasing SiO2 and corresponds to the upper sequence lavas. A striking feature of the Wadi Shaffan lavas is also that their SiO2 contents display a wide range of values (from 46 to 68% anhydrous), with the highest SiO2 content occurring in the lower sequence. Such high values in SiO2 are unusual in MORB (see Petrological Database of the Ocean Floor at The associated low TiO2 content indicates a highly differentiated composition for the lower sequence lavas, confirmed by the very low Cr and Ni content (ranging respectively from 2 to 36 ppm and from 0.8 to 17 ppm). Compared to the lower sequence, the upper sequence lavas are less differentiated. They exhibit a wider range in SiO2 content (from 46 to 63 wt.%), and highly variable compatible elements such as Cr and Ni (3 to 397 ppm and 1.5 to 128 ppm, respectively).

Figure 7.

Variation in SiO2 versus TiO2 (wt.%) for the Wadi Shaffan lavas. All results are expressed in dry oxide weight. Analyses are reported in Table 2. Sample symbols are color coded to indicate relative elevation and analysis method (open symbols XRF analysis and closed symbols from Einaudi et al. [2000]).

4.3. Trace Elements

[24] As illustrated on Figure 8, all the lower sequence basalts plot within a restricted range of values (La normalized to Chondrite [Sun and McDonough, 1989] Lan: 16–24) and display flat to slightly Light REE (LREE) depleted patterns, with La/Yb ranging from 1.35 to 0.80. A pronounced decrease in REE contents marks the beginning of the upper sequence. The first lava of the upper sequence (Unit 13) has significant trace element depletion (Unit 13, Lan: 8) compared to the last lower sequence unit (Unit 12, Lan: 19). From Unit 14 to the top of the sequence, lavas exhibit highly variable REE contents, with Lan ranging from 8 to 40. Despite these variations in concentration, all the upper sequence lavas display flat or slightly LREE-enriched patterns (La/Yb: 1.03–1.96). For all the analyzed samples (lower and upper), there is little or no Eu anomaly, except for the most differentiated lavas in each sequence. Overall, the general REE patterns are not very different from typical MORBs with La/Yb ranging from 0.66 to 0.83 (N-MORB average value 0.82 [Sun and McDonough, 1989]).

Figure 8.

Chondrite normalized REE diagrams illustrating differences in the relative abundance of REE between the lower and upper sequence. Patterns are normalized after Sun and McDonough [1989]. Elements are listed from right to left in order of increasing incompatibility during mantle melting. The complete analyses are presented in Table 2. A simplified lithological column of the Wadi Shaffan section is shown, with sample locations. Symbols are the same as in Figures 3 and 6.

[25] Over the whole section, there are no significant variations of High Field Strength Elements (HFSE) content relative to REE and other incompatible elements, except for Nb and Ta, which are slightly depleted relative to Th and La. HFSE concentrations are correlated with REE concentrations along the Wadi Shaffan section. The Zr/Nb ratio is distinctly different in the lower sequence compared to the upper one (Figure 9). Other elements such as Sr, Ba, Rb and more generally Large Ion-Lithophile Elements are probably modified by secondary alteration, and will not be used for geochemical interpretations.

Figure 9.

Zr/Nb ratio versus La/Yb ratio, which clearly distinguish the upper sequence from the lower sequence. Complete analyses are presented in Table 2. Sample symbols are the same as in Figure 6.

4.4. Chemical Stratigraphy

[26] High frequency sampling performed along Wadi Shaffan implies a temporal relationship between contiguous samples. Variations in geochemistry (content - ratios) can be investigated in terms of temporal evolution.

4.4.1. Lower Sequence

[27] The lower sequence consists of the most differentiated lava described in Wadi Shaffan, characterized by high SiO2 contents together with low and constant Cr and Ni contents. It extends from the base of the outcrop (contact between the sheeted dyke complex and the extrusives) to the top of Unit 12. Elements such as TiO2, Zr, and REE exhibit variable concentrations from the base to the top of this sequence (Figure 10). The TiO2 contents increase gradually from the base of the sequence to the top, whereas Yb and Zr decrease in abundance over the same interval. This trend in the lower sequence indicates an evolution from differentiated lavas toward less differentiated lavas with elevation (and hence time). Two units (Units 4 and 5) are out of the main evolution trend and have distinctive geochemical characteristics with very low TiO2 content and the highest Zr and Yb contents. No significant differences in composition are observed between massive and pillowed units in this lower sequence.

Figure 10.

Spatial geochemical variations along the Wadi Shaffan transect for some representative trace element (TiO2, Cr, Ni, Zr, Yb) contents and ratios (La/Yb and Zr/Nb). Analyses are reported in Table 1. A simplified lithological column of the Wadi Shaffan section is presented on the right, see Figure 3 for lithologic symbols. Sample symbols are the same as in Figure 6.

4.4.2. Upper Sequence

[28] The upper sequence lavas have distinct parental magmas compared to the lower sequence. Incompatible element ratios such as Zr/Nb are constant among all upper sequence lavas (Figures 8 and 9), and significantly lower than in the lower sequence lavas. The La/Yb ratio for the upper sequence lavas is generally higher than the lower sequence. On the basis of these trace element ratios, Units 13 to 17 are unequivocally parts of the upper sequence. Lavas from the upper sequence are characterized by large variations in trace element concentrations (Cr, Ni, TiO2, Zr, and REE) resulting in a saw-toothed pattern along the section (Figure 10). The less evolved lavas are located at the base of the upper sequence (Units 13 to 17). The large variations within the upper sequence may be indicative of modifications by crystal fractionation, the less evolved end-members being emplaced at the base of the upper sequence (Units 13 to 17).

[29] In summary, petrographic and geochemical analysis supports the existence of two distinct magmatic sequences: the upper and lower sequence, and show that (1) the lower and upper sequence lavas were formed from distinctly different parental magmas, (2) differentiation processes operated on these magmas before their eruption. These two processes are discussed in the next paragraphs.

5. Discussion

5.1. Wadi Shaffan Lava Composition in the Global Oman Context

[30] As described in numerous ophiolite suites (e.g., Shervais [2001] or Bednarz and Schmincke [1994] for the Troodos Ophiolite), several magmatic episodes are often encountered, illustrating the history from accretion to emplacement. For the purpose of this study, it is important to determine whether the Wadi Shaffan lavas formed through magmatic processes at the ridge axis. The Semail extrusive sequence has been divided into several volcanic units on the basis of field relations and geochemistry: Geotimes-Lasail-Alley-Salahi [Alabaster et al., 1982; Lippard, 1986] also called V1-V2-V3 [Beurrier, 1987; Ernewein et al., 1988]. The Geotimes/V1 unit displays trace element compositions similar to N-MORB with the exception of a slight Nb-Ta negative anomaly [e.g., Pearce et al., 1981; Lippard et al., 1986] and is related to the bulk of the sheeted dike complex [e.g., Lippard et al., 1986; Rochette et al., 1990]. It is interpreted as resulting from accretion magmatism along a spreading ridge, either in an open-oceanic environment [e.g., Nicolas et al., 2000 and references therein], or in a marginal basin (high SiO2 content and HFSE negative anomaly, e.g., Lippard et al. [1986]). The overlying “Lasail-Alley/V2” and “Salahi/V3” volcanics may be linked to the post accretion history of the Semail Ophiolite and obduction [Pearce et al., 1981; Alabaster et al., 1982; Lippard et al., 1986; Ernewein et al., 1988].

[31] A Ti versus Zr plot has been used to discriminate the V1-Geotimes from the late volcanic sequences [Alabaster et al., 1982; Lippard et al., 1986]. Because oxides were fractionated earlier in the V2-Lasail/Alley lavas, the Zr and Ti content in the Geotimes lavas is generally higher than in the V2-Lasail/Alley. On Figure 11a, the lower sequence lavas plot in the Geotimes field. The upper sequence lavas are mainly in the same field, but some basalts are in the V2-Lasail/Alley field, with low Zr and Ti content, in particular at the base of the upper sequence (Units 13 to 17). However, Wadi Shaffan lavas do not show the significant LREE depletion and low Zr/Hf ratios shown by V2-Lasail/Alley lavas (Figure 11b). All Wadi Shaffan lavas are characterized by La/Yb ratios ranging from 0.80 to 1.93 and Zr/Hf ratios in the range of MORB (Zr/Hf = 34–44; N-MORB: 36.1 [Sun and McDonough, 1989]) and plot in the V1-Geotimes field.

Figure 11.

(a) TiO2 versus Zr for the Wadi Shaffan basalts. V1 and V2 trends are from Alabaster et al. [1982]. (b) Plot of Zr/Hf ratio against La/Yb ratio for the Wadi Shaffan basalts. V1 and V2 analysis are from Alabaster et al. [1982], M. Godard et al., A geochemical study (trace elements and Nd, Sr, Pb isotopes) of a tholeiitic volcanic section in the Salahi block: Implications for the origin and evolution of the Oman ophiolite, manuscript submitted to Chemical Geology, 2002, and Godard et al. [2003].

[32] Thus the Wadi Shaffan lavas can be linked unequivocally to the V1-Geotimes volcanic episode and to the Oman paleo-ridge accretion. Decoupling in trace element ratios observed between the lower and upper sequence lava suggests variations in the magma source during the same tectonic event.

5.2. Source Characteristics

[33] Incompatible trace elements such as REE, Nb, Zr are considered to be insensitive to alteration and they are not expected to fractionate greatly during crystallization processes. Therefore their variations can be used to assess the composition of the parental magmas of the Wadi Shaffan section. The La/Yb versus Zr/Nb diagram (Figure 9) suggests that two different parent magmas formed the lower and upper sequences. The differences in Zr/Nb and La/Yb suggest a higher degree of partial melting for the lower sequence lavas compared to the upper sequence lavas.

[34] Such large variations in trace element ratios between the lower and upper sequence may imply two different mantle processes in the origin of the Wadi Shaffan section. The Wadi Shaffan lavas may record two accretion events; the first involving a depleted mantle (corresponding to the lower sequence lava) and the second stage involving a less depleted mantle (upper sequence). Alternatively, the Wadi Shaffan section may have sampled different regions of mantle during a single mantle upwelling event. According to models proposed for present-day spreading ridges [e.g., Bottinga and Allègre, 1973; Plank and Langmuir, 1992; Dunn et al., 2000], the distribution of melts and the extent of partial melting is highly heterogeneous below the accretion zone. The top of the melting column corresponds to the highest degree of melting, while regions that have undergone lower extents of melting are below and slightly off axis. The most depleted, lower sequence lavas may correspond to the first extracted melts. The less melted mantle may have been sampled later, maybe off-axis, and would correspond to the upper sequence lavas.

5.3. Temporal Trends and Fractionation-Magma Chamber Processes: From Gabbros to Basalts

[35] The formation and structure of magma storage systems and the processes in them are fundamental to our understanding of magma differentiation, and to explain the variability of the Wadi Shaffan lavas. Lavas from the Wadi Shaffan transect have significant differences in trace element content between the lower and the upper sequence. Such large variations in trace element abundance along the cross section can be interpreted in terms of magma chamber functioning, differentiation and fractional crystallization processes. Variations in geochemistry reported along Wadi Shaffan section can also be related to gabbros, representing the magma chamber(s).

5.3.1. Lower Sequence and Upper Gabbros (Cycle I)

[36] The geochemical evolution of lower sequence lavas versus elevation shows a reverse fractionation trend, with most samples being Zr and REE depleted, that is, the least differentiated ones, at the top of the sequence (Figure 10, Cycle I). Two models can explain this reverse evolution. The first is a general model of a zoned magma chamber with the most differentiated lavas on the top. In this model, the first emplaced lavas will be the most differentiated and the following ones will be less evolved. However, this model does not explain the absence of phenocrysts. The second process is magma mixing of newly injected magma with highly differentiated melt in a magma chamber. Such a process has been proposed to explain gradual reverse geochemical evolution through time in ODP Holes 504B (Figure 1) and 896A [Brewer et al., 1996]. The homogeneity in trace element ratios as well as the geochemical variations (Figure 12) recorded in the lower sequence leads us to interpret the lower sequence as a single magmatic cycle (Figures 13 and 14, Cycle I).

Figure 12.

TiO2 versus SiO2 diagram. The Wadi Shaffan lavas are compared to gabbros from the Wadi Abyad section [MacLeod and Yaouancq, 2000]. Sample symbols are the same as in Figure 6.

Figure 13.

(a) REE patterns of the Wadi Shaffan upper sequence lavas Cycles II, III, IV and V are distinguished (see text for explanation). In this diagram is also reported liquid in equilibrium with clinopyroxene from the MTZ (sample 90OA61, from Kelemen et al. [1997]). Patterns are normalized to C1 chondrite from Sun and McDonough [1989]. (b) Liquid in equilibrium with clinopyroxene from gabbroic sills from Kelemen et al. [1997], gray shaded area is the range of the Wadi Shaffan lavas. Patterns are normalized to C1 chondrite from Sun and McDonough [1989]. (c) (Ce/Yb)n versus Cen diagram. Our Wadi Shaffan analyses are reported in the diagrams. Three in situ crystallization models [Langmuir, 1989] are integrated in this diagram. Distribution coefficients between minerals and melt used in these models were compiled by Bedini and Bodinier [1999] for clinopyroxene and olivine and by Niu et al. [1996] for plagioclase. Crystallization model considered corresponds to in situ crystallization from a parental magma corresponding to the mean of the units 13–17 composition. Calculated liquids differ only by proportion of crystallizing mineral phases: olivine + clinopyroxene + plagioclase; respectively Model 1, 20–30–50; Model 2, 25–25–50; Model 3, 15–40–55.

Figure 14.

Proposed interpretation of the geochemical variation along Wadi Shaffan. Two main processes are involved; difference in partial melting (Zr/Nb ratio) and fractional crystallization, illustrated with variation the TiO2 content. On this diagram are also reported magnetic susceptibility measurements on the same samples. Excellent correlations are observed. Trends of inferred magmatic cycles are underlined with light yellow shading. A simplified lithological column of the Wadi Shaffan section is presented on the right. Symbols are the same as in Figure 3.

[37] Lower sequence lavas are characterized by an inverse correlation between SiO2, Zr, REE contents and TiO2 (Figures 7, 10 and 11). These compositions points to a considerable degree of pre-eruption fractionation, marked by oxide fractionation. Corresponding cumulates should be found in the gabbroic section of the ophiolite. Numerous studies of gabbroic rocks have been conducted in Oman [e.g., Pallister and Hopson, 1981; Lippard et al., 1986; Kelemen et al., 1997; MacLeod and Yaouancq, 2000]. Evolved and oxide bearing gabbros have been found only in the upper part of the gabbroic section below the dyke root zone. These “varitextured gabbros” are characterized by wide geochemical variations compared with the underlying layered and foliated lower gabbros (Figure 12). Reverse zoning in plagioclase has been described by MacLeod and Yaouancq [2000] and interpreted as the interaction between incoming primitive melt in the existing differentiated liquid. MacLeod and Yaouancq [2000] proposed that varitextured gabbros represent the remnants of a fossil magma lens that could be an analogue to the Axial Magma Chamber (AMC) described at fast spreading ridges [e.g., Detrick et al., 1987; Sinton and Detrick, 1992; Natland and Dick, 1996].

5.3.2. Upper Sequence and Lower Gabbros (Cycles II to V)

[38] In the upper sequence, variations in trace element abundance (REE, Zr) are organized into four magmatic cycles. These cycles have been denoted from II to V (Figures 13 and 14). Each cycle (except Cycle II) starts with primitive lavas (low Zr, REE) evolving to more differentiated (high Zr REE) lavas until the beginning of a new magmatic phase. In contrast with the lower sequence variations, TiO2 variations in the upper section are correlated with Zr and REE contents and to SiO2 suggesting a lesser degree of differentiation before the emplacement of these melts. Nevertheless, the saw-toothed pattern in TiO2 content in Figure 14 suggests that each cycle represents the fractionation of a discrete magma batch of magma in a closed system until replenishment or initiation of a new magma chamber. Cycle II differs from the following ones by its homogeneity in trace element content which is interpreted as being the result of fast emplacement of a large volume of primitive magmas. These most primitive lavas (Cycle II, Units 13 to 17) are in equilibrium with clinopyroxene analyzed in sills emplaced the Mantle-Crust Transition Zone -MTZ- [Kelemen et al., 1997; Korenaga and Kelemen, 1997].

[39] The MTZ represents the deepest location where crystallization took place [Kelemen et al., 1997]. The large range of incompatible element abundances in the upper sequence lavas may result from varying extents of fractional crystallization from a compositionally similar parent (Figure 13). To quantify this hypothesis, simple crystallization models were made (Figure 13), assuming that initial liquids had the mean composition of the most primitive lavas analyzed in Wadi Shaffan (Cycle II, Units 13 to 17). The simplest model involves closed-system crystallization. On Figure 13, different models are proposed to explain variations observed in the upper sequence lavas. Each cycle can be explained by crystallization of various proportions of olivine, clinopyroxene and plagioclase. These crystallizing phases are consistent with phenocrysts in lavas from our sample section (i.e., the observation of olivine and pyroxene in lavas at the base of each cycle), and with the mineral assemblage in the Oman lower gabbros.

[40] Geochemical differences between the upper and lower sequences can be linked to different magma chamber processes at different crustal levels. The high frequency sampling performed along Wadi Shaffan implies a temporal relationship between these two processes. They can reflect temporal variations in the activity of the ridge, and can be integrated in terms of the timing of crustal processes. The Wadi Shaffan lava geochemistry leads us to the conclusion that the section was build through two main volcanic episodes.

[41] 1. Lower sequence lavas (Units 2–12) are characterized by a relatively high degree of partial melting of the mantle source (based on La/Yb and Zr/Nb ratios, Figure 9) and high degrees of differentiation (e.g., Figure 7). These lavas correspond to the first emplaced sequence and are genetically linked to the top of the upper gabbros section and more precisely to gabbro/dyke interface. Geochemical evolution of these lavas can be interpreted as representing a shallow melt lens functioning as an open system.

[42] 2. Upper sequence lavas (Units 13–42) formed by a lesser degree of partial melting of the mantle source; these are less differentiated than lower sequence lavas. Liquids that formed the upper sequence lavas are in equilibrium with minerals from the gabbroic lenses in the MTZ and with lower, layered gabbros. Geochemical variations within each cycle can be generated by fractional crystallization from a common primitive magma in equilibrium with MTZ gabbros. We propose a model in which the upper sequence lavas were directly transported from the MTZ and lower gabbro sills to the surface without interacting with higher crustal levels. When considering variation in tilting from the base to the top of the Wadi Shaffan lavas (see Figure 3), the last lavas may have been emplaced slightly off axis.

[43] The absence of a hiatus (no sediments or hydrothermal deposits) between the upper and lower sequences, suggests a relatively fast transition between the two processes, from a single shallow melt lens one or more magma chambers in the MTZ and/or in the lower gabbros. Our data provide temporal constraints on accretion models proposed for the Oman Ophiolite [Boudier et al., 1996; Kelemen et al., 1997; Korenaga and Kelemen, 1997; MacLeod and Yaouancq, 2000].

5.4. Oman and Oceanic Crust

5.4.1. Lithology and Magmatic Cycles

[44] Along the Wadi Shaffan section, several magmatic cycles are observed, with thicknesses ranging from 25 m (volcanic Unit II) to almost 85 m (volcanic units I and V). Most magmatic cycles can be linked to morphological variations: for instance, cycles II, III and IV start with a 5 to 8 m thick massive unit followed by pillowed units. In contrast, Cycle I is composed of pillow units cross cut by massive units but these morphological variations are not associated with any geochemical or petrophysical changes. In oceanic crust [Crane and Ballard, 1981; Kappel and Ryan, 1986; Perfit et al., 1994; Staudigel et al., 1996; Ayadi et al., 1998] or other ophiolites [Furnes et al., 2001], morphological variations have been linked to the occurrence of cyclic stratigraphic units. Cyclic units in a typical volcanic sequence may consist, from base to top, of sheet flows, pillow lavas and sometimes hyaloclastites [Van Andel and Ballard, 1979; Staudigel et al., 1996; Furnes et al., 2001]. Sheet flows are interpreted as being produced during rapid extension at high eruption rates, followed by pillow lavas at slower extension and lower eruption rates. Along Wadi Shaffan, this scheme is observed only at the top of the section. The strong link between lithological and geochemical changes across the boundaries of these cyclic units indicates that strong links exist between magma chamber processes, magma eruption rates, and the mode and nature of volcanic build up of the seafloor. In the lower sequence, massive units are more abundant but they are not associated with geochemical variations. The lower sequence represents one magmatic cycle with steady geochemical variation. The shift in composition observed for units 4 and 5, might be interpreted as an isolated event, in which volcanic products from a different lens were interlayered in the main cycle. Such interlayered volcanic products have been described in Hole 504B and 896A [Kempton et al., 1985; Brewer et al., 1996]. Consequently, care should be taken when using only morphological criteria. The lower sequence clearly illustrates the ambiguous relations between lithology and geochemical composition.

5.4.2. Off-Axis Lavas?

[45] Another comparison point between volcanics at active mid-ocean ridges and upper Wadi Shaffan cycles (II to V), is the interpretation of the upper sequence lavas as off-axis eruptions. Our data suggest that the fine scale geochemical variations reflect temporal changes in the lava delivered to the axis, and in magma chamber processes within the crust.

[46] Interestingly, the Troodos Ophiolite extrusives display the same lithological and chemical stratigraphic evolution as the Wadi Shaffan section, with a highly differentiated lower sequence (high Ti and Si lavas), and less differentiated lavas at the top [e.g., Schouten and Denham, 2000; Schouten and Kelemen, 2002]. As at Wadi Shaffan, the lower sequence can be geochemically linked to the upper gabbros while the upper sequence is in equilibrium with the lower cumulate section in the Troodos crust [Schouten and Kelemen, 2002]. In Troodos, data on chilled margins in the sheeted dyke complex, which include both low-Ti and high-Ti lavas, suggest that the two sequences are contemporaneous [Baragar et al., 1990; Staudigel et al., 1999]. Schouten and Kelemen [2002] emphasize the effects of chemical variations on melt physical properties, the highly differentiated lower sequence (high SiO2) being more viscous than the top low SiO2, low TiO2 lavas [Shaw, 1972]. These authors show that these viscosity variations may have resulted in different depocenters for the two lava types. The high SiO2, high TiO2 viscous lavas could not flow very far while low viscosity, low TiO2 lavas flowed out of the axial rift onto the ridge flanks. Together with seafloor spreading, this could produce the observed stratigraphy in which the low TiO2 lavas overlie the high SiO2 more viscous lavas.

[47] In Wadi Shaffan, the boundary between the two sequences is marked by the presence of a massive flow unit formed by the most primitive lava. Across this boundary, significant changes in mineralogy and geochemistry are observed. Without constraints on the age of each individual unit, the rate of geochemical changes is unknown. However, a continuum of magmatic activity can be inferred from the absence of volcanic breccias and deep-water sediments within the Wadi Shaffan section. Therefore a scenario similar to that proposed by Schouten and Kelemen [2002] for Troodos lava emplacement may also apply to the Wadi Shaffan section. However, in contrast with Troodos, the transition between the two lava sequence in Wadi Shaffan and therefore the two different magma chamber processes (from an open to a closed system) is sharp and associated with a change in the degree of melting in the magma source. The transport of distinctly different parental magmas to feed lower and upper sequence lavas suggest that the conduits of melt transport in Wadi Shaffan may have been spatially separate. We propose that the conduits for lower lavas were on-axis and the conduits for upper lavas may have been slightly off-axis.

[48] Abrupt changes in geochemical composition often occur across discontinuities in the mid-ocean ridge spreading system such as transform faults and overlapping spreading centers. Within a single segment, geochemical variations are also found [Batiza and Niu, 1992; Regelous et al., 1999; Reynolds and Langmuir, 2000]. High-resolution sampling from dredging and diving have provided data on the temporal variation of lava composition along a fast spreading ridge, revealing heterogeneity in the chemistry and morphology of erupted lavas [Reynolds et al., 1992; Regelous et al., 1999]. Several authors have proposed that the chemistry of the lavas erupted at a fast spreading ridge axis may change cyclically [Thomson et al., 1989; Batiza and Niu, 1992; Reynolds et al., 1992; Regelous et al., 1999], but such magmatic cycles have not yet been demonstrated, possibly due to the lack of drillholes into such structures. These studies have also shown that off-axis lava compositions cover a wider range of composition than in on-axis MORBs from the same region [Niu and Batiza, 1991; Spiegelman and Reynolds, 1999; Reynolds and Langmuir, 2000]. Off-axis lavas may develop very close to the spreading center [Reynolds and Langmuir, 2000], and play a volumetrically important role in igneous accretion of oceanic crust. In this context, our data for the upper sequence lavas is consistent with data on off-axis lavas from fast spreading ridges.

5.4.3. Melt Lens Composition

[49] Magma chambers have an important influence on the chemistry of the lavas that are erupted at the ridge axis [e.g., Natland, 1980; Sinton and Detrick, 1992]. The formation and structure of magma storage systems, and the processes in them are fundamental to our understanding of magma differentiation, crustal evolution and the eruptive cycles of the oceanic crust. An Axial Magma Chamber (AMC) located at the base of the sheeted dike complex has been clearly observed at fast spreading mid-oceanic ridges in seismological studies [e.g., Detrick et al., 1987; Sinton and Detrick, 1992]. However, the composition of melt and crystals in these lenses is still problematic. Ferrogabbros and oxide rich gabbros have also been recovered at slow spreading ridges [Dick et al., 1991] and at fast spreading ridges such as in Hess Deep [e.g., Natland and Dick, 1996]. Interpretation of geochemical variations in the lower Wadi Shaffan sequence involves mixing within the AMC, and convective mixing that will tend to homogenize melt compositions before eruptions sample the mixed melt. Such mixing would produce the gradual reverse fractionation trend recorded in the lower sequence lavas.

6. Conclusion

[50] In this paper the geochemistry of a single volcanic section of the Oman Ophiolite has been documented. The high-resolution investigations of the volcanic lithologies and geochemical compositions of the Wadi Shaffan section establish a detailed volcanic stratigraphy and contribute to a better understanding of the volcanic construction of upper oceanic crust at the paleo-Oman ridge. The extrusive sequence of the Wadi Shaffan section was built in two main events.

[51] The lower sequence shows reverse fractionation trends, with less differentiated lava in the upper part of the cycle. Magma mixing is proposed to explain the reverse geochemical evolution through time. Reverse evolution is interpreted as recording mixing between highly differentiated resident magmas and a less differentiated magma entering a shallow melt lens. Mixing will tend to homogenize melt compositions of the periodically erupted lavas. Highly differentiated lavas showing oxide fractionation can be linked to uppermost gabbros and were probably erupted directly along the Oman paleo-axis.

[52] In the upper sequence, geochemical analyses suggest different magma chamber processes. This sequence was formed in four magmatic cycles, as interpreted from variation in trace element concentration. The boundaries between cycles are marked by sharp shifts in lava composition. Each magmatic cycle begins with primitive lavas, and 3 of the 4 cycles start with a massive unit, suggesting link between morphological and geochemical changes. Wadi Shaffan lavas originated from two distinct crustal levels at different periods of the ridge activity. The upper sequence lavas are in equilibrium with lower gabbros from the MTZ. Within each cycle, variations in trace element contents are explained by crystal fractionation in a closed system. Lavas were directly transported from the MTZ to the surface without interacting with shallower crustal levels. The upper section may have been emplaced slightly off-axis.


[53] In the course of this research we were supported in part by CNRS INSU as part of the IT program. We were generously guided in our field work in the Oman Ophiolite by Adolphe Nicolas and Michel Beurrier, with the logistical support of Dr. Hilal Al Azri, Director of the Geological Survey of Oman. We have benefited greatly from many discussions with the mid-ocean ridge and ophiolite research community. We thank Hans Schouten and Ken Sims for helpful reviews, associate editor Peter Kelemen who substantially improved this manuscript and William White for editorial handling.