Dike surface lineations as magma flow indicators within the sheeted dike complex of the Troodos Ophiolite, Cyprus

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. B3, PAGES , MARCH 10, 1998 Dike surface lineations as magma flow indicators within the sheeted dike complex of the Troodos Ophiolite, Cyprus Robert J. Varga Department of Geology, The College of Wooster, Wooster, Ohio Jeffrey S. Gee, Hubert Staudigel, and Lisa Tauxe Scripps Institution of Oceanography, La Jolla, California Abstract. Mesoscopic flow lineations and anisotropy of magnetic susceptibility (AMS) have been measured for dikes within the Cretaceousage Troodos ophiolite with the goal of comparing the direction of initial magma flow through dike conduits immediately following crack propagation with that of flow of subsequent magma emplaced during later stages of dike growth. Dike margin indicators of flow include cusp axes and elongate vesicles found high in the ophiolite pseudostratigraphy and ridgeandgroove stmcalres termed hot slickenlines found throughout the complex. A unique flow direction is determined where elongate vesicles near dike margins display imbrication with respecto the margin. Significant changes in vesicle elongation directions across dikes likely indicate either changes in magma flow direction after dike propagation or backflow of magma during the waning stages of intrusion. Surface lineations generally lie subparallel to the direction of flow inferred from AMS determinations on cores within 5 cm of dike margins. Surface lineations also lie subparallel to the long axis (el) of the orientation ellipsoid defined by long axes of groundmass plagioclase phenocrysts measured in sections from AMS cores. Correlation of surface lineations with interior indicators of flow (AMS, plagioclase trachytic texture) indicate that the surface features are good proxies for groinscale magma flow directions during dike propagation in Troodos dikes. Orientations of surface flow features in the dikes of the Troodos ophiolite indicate an approximately equal mix of subhorizontal to nearvertical magma flow, contradicting the paradigm of primarily vertical flow of magma beneath continuous axial magma chambers at oceanic spreading centers. Our data are consistent with a model of magma eraplacement both vertically and horizontally away from isolated magma chambers beneath axial volcanoespaced along a ridge crest. 1. Introduction A variety of studies suggest that steeply dipping dikes emplaced beneath subaerial volcanoes preferentially propagate laterally in the upper crust from centralized magma chambers. Surface deformation and seismicity patterns in the Krafla region of Iceland, for example, document lateral propagation of dikes for distances of up to 70 km away from central volcanoes during major eruptive events [e.g., Brandsdottir and Einarsson, 1979; Sigurdsson and Sparks, 1978]. Similar studies of active eruptions on Hawaii suggest that eruptive cycles at Pu'u O'o may be fed from a central magma chamber located 25 km beneath the summit of Kilauea [Dzurizin et al., 1984; Wilson and Head, 1988; Wolfe et al., 1987]. In addition to such direct observations during active eruptions, field studies of exposed vertical dikes also support a significant component of lateral magma migration from central magma source regions [e.g., Baer, 1991' Knight and Walker, 1988; Smith, 1987; Walker, 1987a]. Theoretical studies of dike propagation also support the concept of lateral magma flow [Fiske and Jackson, 1972; Rubin, 1990; Rubin Copyright 1998 by the American Geophysical Union. Paper number 97JB /98/97JB and Pollard, 1987; Ryan, 1987]. Magma, rising vertically within the crust, may begin to flow laterally at a level determined by the point where magma density equals the in situ density of the surrounding country rock [Ryan, 1987] or by a point controlled by the ratedependenthermal structure of the ridge [Hoofi and Detrick, 1993]. In contrast to subaerial volcanoes, there are few convincing data which directly constrain the emplacement directions within the sheeted dike sections of oceanic crust. Sheeted dikes are traditionally thought to intrude vertically to feed surficial volcanics [e.g., Cann, 1970; Kidd, 1977; Kidd and Cann, 1974], presumably reflecting the passive upwelling of magma as a direct consequence of seafloor spreading. This simple model is certainly viable along fast spreading centers, such as the East Pacific Rise, where geophysical evidence exists for a relatively continuous axial magma chamber [Detrick et al., 1987; Toomey et al., 1990], but may be less likely along slow spreading ridges which apparently lack steady state axial magma chambers [Detrick et al., 1990]. The recent acoustical detection of a seafloor spreading event on the Juan de Fuca Ridge suggests that lateral dike propagation may have occurred on this intermediate spreading rate ridge [Dziak eta/., 1995]. The northward (~60 kin) migration of seismicity along the CoAxial segment together with the presence of recent fissuring, hydrothermal venting, and a pristine lava flow at the distal end of the seismic path have widely been

2 5242 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE interpreted as evidence of lateral dike propagation [Baker et al., 1995; Embley et al., 1995], although a magma source related to Axial Seamount cannot be excluded [Fox et al., 1995]. Further evidence of dike emplacement direction in the oceanic crust is desirable, as the mechanisms of magma transport have important implications for crustal generation processes and for the interpretation of geochemical data from the ocean crust. In this paper we report emplacement directions, inferred from mesoscopic structural features and corroborated by analysis of silicate and magnetic fabrics, for sheeted dikes in the Troodos ophiolite. The purpose of this paper is twofold. First, we explore the fidelity of dike surface lineations as indicators of magma flow within dikes. The impetus for this is the distinction that has been drawn in the literature between propagation directions in the vicinity of dike tips and emplacement directions of magma in the open conduits [e.g., Baer and Reches, 1987; Rickwood, 1990]. We present field data on mesoscopicscale surface features of dikes indicative of magma flow from all pseudostratigraphic levels of the sheeted complex and volcanic section. The correlation of these surface data to internal flow within the dikes is made by comparison with the threedimensional distribution of elongate phenocrysts (the "orientation ellipsoid") near dike margins and with anisotropy of magnetic susceptibility results. Second, we consider the magma flow patterns within the sheeted dike section of the Troodos ophiolite, Cyprus, and interpret these patterns as evidence of magma dynamics beneath an oceanic spreading center. Flow lineations derived from mesoscopic features, restored to their paleoridge orientations using field criteria and palcomagnetic results, reveal a range of emplacement directions for dikes in the Troodos ophiolite with no preference for vertical over horizontal flow. 2. Field Setting of the Troodos Sheeted Dike Complex While no ophiolite is yet thought to provide a perfect analog to modern ocean crust [see Gass, 1990], the comparisons of their gross lithospheric/crustal layering [e.g., Moores and Vine, 1971], structure [e.g., Nicolas, 1988; Varga, 1991], and hydrothermal processes [e.g., Constantinou and Grovett, 1973; Schiffman et al., 1990] are sufficiently good to deduce that their accretionary processes are indeed quite similar. The Troodos ophiolite (Figure 1) constitutes a littledisrupted ideal Penrose ophiolite [American Geological Institute, 1972] formed during the Late Cretaceous [Blome and Irwin, 1985; Mukasa and Ludden, 1987] in a probable suprasubduction zone setting [e.g., Miyashiro, 1973; Rautenschlein et al., 1985]. The complex is exposed within an approximately eastwestelongate dome which, as a result of erosion, reveals the ophiolite pseudostratigraphy within approximately concentric bands, with mantle peridotites exposed in the axis of the dome and successively higher units (plutonics, sheeted dikes, and volcanics) exposed toward the periphery of the complex. The extensively exposed sheeted dike complex is the hallmark of the Troodos ophiolite and was central to the correlation of ophiolites with the spreading environment of oceanic crust [Gass, 1968; Moores and Vine, 1971]. Over vast regions of the complex, the dikes strike northnorthwest and indicate spreading orthogonal to that direction during seafloor spreading. In the eastern and southern parts of the complex, the dike strikes swing relatively smoothly to nearly eastwest orientations near the Arakapas fault zone, widely interpreted as a fossil transform fault [Gass, 1968; Gass et al., 1994; Moores and Vine, 1971; Simonian and Gass, 1978]. Several paleomagnetic studies have shown that the sigmoidal pattern Mandres Evrykhou + Akaki Tembria Potamos Kionia Onouphrios Cyn Figure 1. Outline of main Troodos ophiolite massif showing generalized exposure of volcanic section (stippled) and sites where mesoscopic flow structures were measured within dikes (dots); ATD is the Arakapas transform domain. Averaged dike (great circle) and flow line data (dots) for these sites are shown in adjacent stereonets.

3 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE 5243 of dike strike orientations near this fossil transform is the result of post dike emplacement structural rotation and not the result of emplacement of dikes into a crust with curved stress trajectories [Allerton and Vine, 1987, 1990; Bonhommet et al., 1988; Gass, 1968; MacLeod, 1990; MacLeod and Murton, 1995]. In addition to vertical axis rotations, dikes have been variously rotated about subhorizontal, largely approximately northstriking axes related to rotationalplanar and listric, ridgecrest normal faulting. Rotations related to extensional faulting vary up to 90 ø or more JAilerton and Vine, 1987; Hurst et al., 1994; Varga, 1991; Varga and Moores, 1985, 1990]. Paleomagnetic constraints indicate that the ophiolite has undergone approximately 90 ø of counterclockwise, vertical axis rotation since formation and prior to the early Eocene [e.g., Clube et al., 1985]. During this study we measured mesoscopic lineations within and on dike margins throughout the sheeted dike complex and the lower portions of the volcanic section (see Figure 1). Sites at Kamara Potamos, Akaki Canyon, and Onouphrios Canyon expose dikes cutting pillow lavas and We note, however, the caveat discussed by Rickwood [1990] that when viewed over their entire extent, dikes can record complex behavior during emplacement. Individual dikes can propagate in several directions at the same time [Bussell, 1989] and flow streamlines in the open conduit can vary considerably along the dike length and height [e.g., Walker, 1987a]. Flow direction can also change substantially with time, particularly in longlived thick dikes [Philpotts and Asher, 1994]. In this paper we distinguish between flow directions of (1) magma emplaced immediately after crack propagation and as evidenced by surficial flow indicators on dike margins, (2) magma emplaced after propagation and as measured within ~5 cm of dike margins, and (3) magma emplaced during the waning stages of dike growth and represented by material in dike centers. In most dikes we have studied, flow indicators are best developed in situations 1 and 2 where velocity gradients in the magma due to friction are highest. Thus our conclusions apply most strictly to magma emplaced during and shortly after crack propagation. Flow direction results from nearmargin dike material are probably best extrapolated to later emplaced dike material in relatively thin dikes such as those in ophiolites. Although the interpretations reached in this paper are based on a limited sampling of dikes within the sheeted complex of the Troodos ophiolite, we feel that the results can be generalized to the complex as a whole. Our unpublished field and AMS data on over 300 dikes in the region give results similar to those presented here. 3. Mesoscopic Flow Lineations A wide variety of structural features of dikes have been used in previous studies to assess magma flow directions in dikes. Among the most commonly used mesoscopicscale structures observable in the field are stretched and elongate vesicles, surface lineations such as "finger and groove" marks and scour marks, asymmetric folds, and ramp structures [Baer, 1991; Corry, 1988; Johnson and Pollard, 1973; Philpotts and Asher, 1994; Smith, 1987; Walker, 1987a]. Less direct measurements of magmatic flow include morphological features of dikes, including orientations of dike segments, steps, and buds [Currie and Ferguson, 1970; Delaney and Pollard, 1982], although interpretation of such features is controversial [see Baer, 1991]. Petrographic measurement of elongation [e.g., Philpotts and Asher, 1994; Wata, 1992] and crystallographic orientation of phenocrysts and groundmass crystals [Shelley, 1985] measured in oriented thin sections have also proved useful in determining dike flow directions, although in such studies, assumptions have been made about the orientation of the flow plane. Under most circumstances, it is difficult to determine the sense of magma flow along the intrusive direction. In this regard, imbricated elongate and/or platy structures (crystals, vesicles) near dike margins have been of the most use (Figure 2). In ideal cases, the margins of a dike will display opposite senses of imbrication, uniquely fixing the flow direction, although in most instances, imbrication along a single margin is generally sufficient to determine shear sense [Philpotts and Asher, 1994; Wata, 1992]. sheet flows and represent the shallowest (<1 km depth) crustal level sampled during this study. The geology of these sites is wellcharacterized by Schmincke and Bednarz [1990]. Sites at Mendres and Evrykhou represent shallow crustal levels (13 km depth) within the dike complex while Tembria, Kionia, and sites 91281, 91284, 91303, and represent intermediate to deep (24 km depth) crustal levels. Paleomagnetic samples 3.1. Dike Morphology Indicators of Flow were collected from most sites (except the last four) to assess structural rotations and anisotropy of magnetic susceptibility Dikes of the Troodos ophiolite vary in width from a few (AMS) of the dikes. In addition, we analyzed the preferred centimeters to about 5 m, although most are between about 0.5 alignment of plagioclase laths from three representative dikes to 1 m in width. In outcrop at most levels in the sheeted from which we also collected surface lineation and AMS data. complex, dikes have relatively planar and subparallel boundaries with little significant morphological texture. At high levels in the sheeted complex and volcanic section where dikes cut volcanics, however, many dikes have highly curved boundaries (Figure 3a). In dike parlance, such structures are termed "cusps." The axes of cusps are believed to be parallel to the direction of magma transport within dikes [Pollard, 1987] Hot S!ickenlines Grooves and striations occur irregularly on dike surfaces at all levels in the sheeted dike and volcanic sections of the ophiolite (Figures 3b and 3c). These features generally consist of flattopped ridges and corresponding grooves decorating the exterior margins of dikes where in contact with wall rock comprising volcanic material or older dikes. Surface lineations on most dikes are slightly to highly curved, particularly in the vicinity of dike margin irregularities, such as cusps (Figure 3a), indentations (Figure 4c), or bulges. The geometry and curvilinear nature of the surface features discussed here have similarities to the "flow lineations" discussed by Walker [ 1987a] from the Koolau dike complex in Hawaii, to surface lineations described by Roberts and Sanderson [1971] and to the dike margin "scour marks" documented at Spanish Peaks by Smith [1987]. Those on the margins of Troodos dikes, however, differ from some of the above descriptions in that they only occur on dike margins, at the interface with wall rock, and not within the interiors of dike margins. Similar to the conclusions of the above authors, we interpret the grooves and striations described as the result of flow of semiplastic magma during dike injection. This interpretation is strengthened by the parallelism of the lineations with stretched vesicles where both are found together (Figure 2b).

4 5244 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE ^ B. Vesicle : :';.:2:: _:::::::"..."::"::::"'4';:::::';':;:"::'"'": '' ":! ':::i': '": :::::::$:'Y':;:"::":::::::" g::::" " : :.2 ::::::::i:i:i:i?' i' Slickenlines" Relative Velocity N RetatiVe ow Vetoi "' Imbncatlon Hot ¾1 directions C Slickenlites W margin ¾1 directions E margin Inferred Magma Flow Direction Figure 2. Relationship between surface lineations dike margins to internal flow indicators. In both Figures 2a and 2b, arrowshow relative flow velocity which is lowest near dike margins and greatest near dike center. Velocity gradient is, however, greatest near dike margins. (a) Dark lines on east dike margin indicate trend of surface lineations ("hot slickenlines") and indicate subhorizontal flow. Tabular phenocrysts (parallelopipeds) within dike are aligned with long axes subparallel to flow direction but are arranged en echelon (imbricated) with respecto dike margins. Opposite sense of imbrication results from velocity gradients near dike margins and indicates flow direction (to south and down in this case). Dark regions between phenocrysts represent magnetite crystals. The distribution anisotropy of magnetite imposed by the silicate fabric is important in defining the rock AMS fabric [Hargraves et al., 1991]. (b) Similar to Figure 2a but shows alignment and imbrication of ellipsoidal volumesuch as vesicles within magma near margin or AMS representation ellipsoids. Intersections of imbricated vesicles with dike margin are ellipsoids elongate in flow direction. (c) Lower hemisphere, equalarea stereonet shows imbrication of internal flow lineations relation to dike trace and to surface lineations. Note that the sense of flow along flow lineation is uniquely provided by the sense of imbricationear each dike margin. We find "flow lineation," however, too general a term for the surface lineations on Troodos dikes because of the existence of other types of flowaligned features, such as elongate vesicles and prismatic crystals. We also find the term "scour mark" inappropriate for these features for they do not appear to representhe "scouring and erosion of chilled, viscous magma near the dyke contact by more fluid magma flowing in the dyke's interior" as described by Smith [1987, p. 49]. Such a mode of formation apparently results in a zone of lineated and platy material near the dike margin [Rickwood, 1990; Smith, 1987]. The restriction of the striations described here to the outer surface of chilled margins suggests that they formed by late movement of magma at the edge of the dike past asperities and irregularities within the enclosing wall rocks. They are distinct, however, from the tectonic slickenlines sometimes observed on dike margins in the ophiolite which comprise sharply angular grooves in cross section, do not occur in association with ridges, and are often associated with secondary features indicating their tectonic origin, such as Riedel, "PT" and other secondary shear features [Petit, 1987] as well as breccia [Varga, 1991]. In addition, the outer, finegrained to glassy marginal material of dikes follows the grooves and striations and is not cut by them, as would be the case if the features represented secondary abrasion features.

5 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE 5245 Figure 3. (a) Photograph of steep highlevel dike in Akaki Canyon cutting gently dipping sheet flows with welldeveloped, nearhorizontal cusps (arrows). Hot slickenlines on dike margins in (b) Onouphrios Canyon (dike TR53) and (c) Kamara Potamos. Note irregularity of lineations and flattopped, ridge and groove morphology. Number and distribution of paleomagneticore holes on dike margin in Figure 3b are typical for flow determination using anisotropy of magnetic susceptibility (AMS).

6 5246 VARGA ET AL.' DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE Figure 4. (a) View orthogonal to steep dike margin showing surface lineation defined by elongate vesicles indicating subhorizontal flow. (b) View vertically down on dike shown in Figure 4a parallel to dike margin (bar) and nearhorizontal lineations showing imbrication of elongate vesicles (parallel to pencil) indicating flow from right to left (to south in geographic coordinates). Note change in apparent elongation of vesicles toward dike center (bottom of photograph) where the axes of elongated vesicles are nearvertical. (c) View of steeply dipping dike margin showing part of a pillow lava (dark, rounded area to left) embedded within dike margin. Silicafilled vesicles (white streaks parallel to bars) are deflected around pillow margin from horizontal orientation more typical of this dike. All photographs are from Akaki Canyon. Thus we prefer the term hot slickenlines for the flow lineations on the margins of dikes as described here to reflect their formation by the relative motion between wall rock and mobile magma and to distinguish them from their tectonic counterparts Vesicles Dikes bearing vesicles are relatively common within the highest levels of the complex where dikes cut volcanics. Vesicles near dike margins vary from nearly spherical to highly elongate ellipsoids with axial ratios up to 10:2:1 and long axes up to 7.5 cm in length. In some areas, such as Akaki Canyon (Figure 1), vesicles are filled with secondary minerals (amygdules) enhancing their field recognition (Figure 4c). When viewed perpendicular to dike margins (Figure 4a), the elongate vesicles provide a twodimensional lineation which is usually subparallel to the orientation of hot slickenlines where present (Figure 2b). Where exposures permit examination of dike cross sections parallel to the vesicle elongation direction, many dikes show imbrication of vesicles with respect to dike margins; imbrication angles vary up to a maximum of about 25 ø. Several dikes showed imbrication of vesicles along both margins which invariably was opposite in sense. Imbrication of vesicles commonly is lost within several centimeters of dike margins (Figure 4b) where vesicles are either nearspherical or elongated with long axes in a different, commonly subvertical orientation. For example, vesicles in the interior of dike TRO l a (Figure 5a) are much steeper than those at the dike margins.

7 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE 5247 Bo Kamara Akaki Canyon ' ' TROla flow direction.175ø/38 ø Potamos flow direction 176o/_9 ø o vesicles x * vesi½ on man rear west margin 0 vesicles mar east marin n vesicles rear dl center Figure 5. Equalarea stereonets summarizing dike and lineation data for dikes TR01a and TR04 in Akaki Canyon and two dikes in Kamara Potamos. Great circles, average dike orientations; solid star, pole to average sheet flow; open star, pole to average orientation of bedding in Cretaceous sedimentary section overlying ophiolite. however, produce highly elongate vesicles (pipe vesicles) in some marie melts. One theory suggests that pipe vesicles are due to buoyant ascent of gas through the magma [e.g., Walker, 1987b] which, alone, cannot account for the many subhorizontal vesicles noted in Troodos dikes. On the other hand, a model invoking exsolution of gas from the magma at a migrating crystallization front can produce pipe vesicles of any orientation, including horizontal, depending upon the geometry of cooling within the melt [Philpotts and Lewis, 1987]. Extending this latter model to the subvertical dikes of the Troodos ophiolite, we would expect horizontal pipe vesicles to form perpendicular to the dike margins as they cooled inward and not subparallel to their margins as is observed in Troodos dikes. Stretching of vesicles during magma flow seems, then, to be the most reasonable explanation for the geometry of vesicles in the dikes described here. Figure 5a summarizes data from two dikes (TROla, TRO4) exposed in Akaki Canyon. Surface lineations for dikes TRO4 and TROla suggest moderate and shallow flow directions, respectively. For both dikes, imbricated vesicles adjacent to the western dike margins have elongation directions that trend to the west of the dike strike, indicating oblique upward flow to the south in the case of dike TRO4 and nearhorizontal flow to the south in the case of dike TROla. With one exception, imbricated vesicles adjacent to the preserved, eastern margin of dike TROla have elongation directions in the opposite sense to that along the western margin. Vesicles near the centers of both dikes have orientations that differ from those near the dike margins. In the case of dike TR01a, the vesicle lineation is significantly steeper which is the usual observation where flow directions are different in the dike interiors. Several authors [e.g., Philpotts and Asher, 1994; Wata, 1992] have suggested back flow of magma during the late stages of emplacement to explain steeper flow directions near dike centers. An additional cause of more vertical flow in the interiors of some dike margins may be the buoyancy effect of rising vesicles in the magma. In any case, we concur with previous workers that vesicle elongations measured near dike margins probably reflect more accurately magma emplacement directions during dike propagation and growth. We follow previous authors in interpreting elongate vesicles in dikes as indicators of magma flow direction [e.g., Shelley, 1985; Smith, 1987]. Supporting evidence linking 4. Relationship of Mesoscopic Flow Lineation elongate vesicles with magma flow is found by patterns of to Internal Flow Indicators vesicles in the vicinity of flow obstructions. Figure 4c shows a relatively planar dike margin that is deflected around a basalt pillow. Silicafilled vesicles display the subhorizontal flow orientation typical for this particular dike except near the pillow where the vesicles are oriented subparallel to the pillow. These relationships indicate flow divergence around the obstruction (obviously lineations in the vicinity of such irregularities do not represent the overall magma flow A fundamental question in interpretation of dike fabrics in the field is how faithfully mesoscopic features, such as hot slickenlines and vesicles, reflect the internal, grainscale flow regime of magma inward from the dike margin. Assuming that indicators of flow on dike chilled margins (i.e., hot slickenlines, stretched vesicles) record initial magma movement immediately following crack propagation, does direction in a dike and such areas were avoided during the this initial flow vary significantly in direction from that of course of this study). In addition, where elongate vesicles are subsequent magma flowing through the dike conduit? We have imbricated near dike margins, they indicate opposing shear explored this question by comparing dike surface indicators of senses compatible with velocity gradients near dike margins flow with determinations of the internal material flow in the during unidirectionalaminar flow (Figure 2b). Other possible causes of elongate vesicles in dikes might include tectonically dikes using two commonly used flow indicators, AMS and trachytic flow texture orientation. induced plastic deformation and elongation through processes in a static magma during crystallization. A tectonic cause can easily be ruled out by the complete lack of crystalplastic deformation in Troodos ophiolite at levels above the plutonic complex [Varga, 1991]. Purely magmatic processes can, 4.1. Anisotropy of Magnetic Susceptibility (AMS) AMS [Graham, 1954] has been widely used to assess flow direction in a variety of igneous rocks, including ignimbrites [e.g., Ellwood, 1982; Knight et al., 1986; MacDonald and

8 ,,,,,,,,,,,, 5248 VARGA ET AL.' DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE Palmer, 1990], lava flows [e.g., Baer eta/., 1994; Canon et al., 1995, 1997], and dikes [Ernst, 1990; Ernst and Baragar, 1992; Knight and Walker, 1988; Rochett eta/., 1991; Wolff et al.r 1989]. These studies have generally found close correspondence between the principal axis of the susceptibility ellipsoid, /1 associated with the maximum susceptibility eigenvalue x 1 (where'c1>'c2>x3), and flow direction as determined by independent methods. The orientation of the susceptibility ellipsoid reflects the statistical alignment of magnetic grains, either the orientation of individual elongate grains (Khan, 1962)or the distribution anisotropy (Figure 2a) of these grains imposed by their late crystallization in the interstices of a preexisting and aligned silicate fabric (Hargraves et al., 1991). Thus T1 provides a measure of internal magma flow direction independent of that provided by surface lineations which record the flow direction of the magma immediately behind the propagating dike tip. In concert with our study of the mesoscopic structures of Troodos sheeted complex, we have studied AMS in over 300 dikes [Staudigel et al., 1992, also unpublished data, 1997]. Our sampling strategy for such studies in dikes focuses sampling within approximately 5 cm of dike margins to ensure sampling of the high shear strain component of magma flow. Typically, 510 cores were taken on each margin of the dike, with the mean orientation of the principal susceptibility axes and 500 bootstrapped eigenvectors calculated by the Figure 6. Histogram showing correlations between AMS parametric bootstrap method of L. Tauxe et al. (Flow /1 directions and surface lineations for 36 dikes with two directions in dikes from AMS data: The bootstrap way, preserved chilled margins. (a) Plot of difference between rake submitted to Journal of Geophysical Research, 1997). of /1 on east dike margins and /1 on west dike margins. One measure of the fidelity of AMS to record magma migration directions is the consistency of flow directions across dikes. L. Tauxe et al. (submitted manuscript) have (b) Plot of difference between T1 rake and corresponding surface (sfc.) lineation rake; n = 72 on this plot because surface lineation is compared to t I on both dike margins of determined that T1 directions on opposite dike margins are indistinguishable for 95% of dikes for which AMS ellipsoids were defined on both margins. in Figure 6a we show the individual dikes. relationship between AMS flow directions on opposite dike margins for 36 of this subset of dikes that also possess measured flow lineations on their margins. The similarity of flow near both margins is quite good; the average angular difference between rake of T1 on opposite margins is 7.5 ø. The relationship between surface flow lineations and AMS T1,directions is also good, although somewhat more divergence is seen between the two types of flow indicators (Figure 6b). The average difference between surface lineation and 1 directions for dikes with two preserved margins is 12.6 ø, with specimens were cut at successively greater distances from the chilled margin. In all cases, these multiple specimens have statistically indistinguishable susceptibility ellipsoid orientations, indicating that the flow delineated by the /1 directions (and coincident with surface lineations) is representative of at least the outer 58 cm of the dike. Approximately 35 cores taken from two sampling localities (separated by 60 m) on a single dike (TR42/43; Figure 8) illustrate the alongstrike consistency of the flow direction inferred from AMS and the similarity to flow direction defined most of the dikes showing less than about 20 ø of divergence. by elongate vesicles in this dike. Again, /1 and /2 for both Similar results were reported for dikes on Hawaii by Knight and Walker [1988]. Results from three representative dikes illustrate the close correspondence between mesoscopic surface lineations, vesicle elongation and AMS data. Dike TR53 illustrates the case where surface lineations, in this case defined by hot slickenlines, have a consistent orientation over an area of several square meters (Figure 3b). Six cores taken from the east margin of this dike (Figure 7) provide remarkably welldefined magnetic fabric; /1 and /2 lie near the dike plane, while t 3 lies at a high angle to the dike. The value t 1 lies very close to the surface lineation orientation, and although only a single margin was sampled, the tight clustering and margins lie close to the plane of the dike, while /3 axes lie at a high angle to it. In detail, the average maximum principal susceptibility axes from the two margins may be interpreted as imbricated, with /1 for cores drilled on the western margin lying slightly to the west of the dike margin and with /1 for cores drilled on the eastern margin lying to the east of the margin. These data are consistent with nearhorizontal flow to the southwest for this particular dike inferred independently from vesicle elongations. Although the majority of dikes studied from the Troodos ophiolite show close agreement between mesoscopic flow indicators and the flow direction inferred from AMS, the two types of data may diverge by 3050 ø in a small number of imbrication of /1 directions (see Figure 2) is sufficient to delineate a unique flow direction (up to the NW). For many of the cores (drilled perpendicular to the chilled margin), multiple west margin /1 steeper!... st& dev. = 9.6ø... than east margin 1 I i i _ [ i i ast m rgin '1 steeper n10[ i... i ".l than west margin 1 A O East Margin West Margin AMS ¾ 1 Rake mean = 0.6 ø. std. dev. = 16.1 ø,, sfc.!ineation 10 ß steeper than 1 '' ¾1 steeper than sfc.!ineatlon B AMS 1 Rake Surface Lineation Rake cases (Figure 6b). For example, /1 for cores drilled close to the eastern margin of dike TR05 (Figure 9) lie close to the plane of the dike, while /1 for cores drilled close to the western ': i,

9 ß VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE 5249 up ion I flow direction 310o135 ø I i i " scctiou 2 '!..."l'' I I I I!! sect i on 3 TR53 Figure 7. Field, AMS, and plagioclase orientation ellipsoi data for dike TR53. Lower hemisphere great circle traces indicate average orientation of dike margin (solid) and thin sections (dashed). Solid square, triangle, and circle indicate /1, /2, and /3 axes, respectively, of AMS ellipsoid for cores taken on east margin of dike. Dots are the bootstrapped eigenvectors for AMS data: gray circles, long axis of twodimensional orientation ellipse for each thin section; solid star, average orientation of lineation on dike margin; crosses, and orientations of E 1, E 2, and œ3 of orientation ellipsoid for plagioclase fabric. Rose diagramshow long axis orientation data for plagioclase for each of three thin sections. Planes of rose diagrams are parallel to thin sections with arrows showing long axis of twodimensional orientation ellipse; sections viewed looking into lower hemisphere. margin lie to the west of the dike plane indicating nearhorizontal flow to the northnortheast. Although Y2 and /3 for both sides of the dike lie close to and at a high angle, respectively, to the dike, the orientation of the minimum and intermediate susceptibility axes is poorly defined. The substantial error ellipses for /2 and /3 in part reflect the pronounced prolate character of susceptibility ellipsoids for many samples (four samples have 'c2/'c 3 of <1.006). The vesicle lineations recorded for this dike have rakes varying from 46 ø to 90 ø, with a mean of 66 ø, 30 ø steeper than the shallow northward flow inferred from the imbrication of the /1 directions. Because the vesicle!ineation data were obtained over a much larger area (40 m) than the drill cores (5 m), the discrepancy between the two inferred flow directions may, in part, reflect local variations in flow direction related to the somewhat undulose margin of the dike ThreeDimensional Trachytic Texture Another sensitive recorder of fl0w direction in laminarly from oriented paleomagnetic cores. The distribution of particles within each core was measured with respecto thin flowing magmas is texture resulting from statistical alignment of elongate phenocrysts or other linear objects [e.g., Elston and Smith, 1970; Philpotts and Asher, 1994; Schmincke and Swanson, 1967; Varga, 1983; Wata, 1992]. Theoretical and experimental studies of elongate markers embedded within a viscous matrix have shown that during both simple and pure shear, initially randomly oriented particles become statistically aligned [Fernadez, 1987; Harvey and Laxton, 1980; March, 1932; Owens, 1973; Sanderson, 1977; Soto, 1991]. The axes of the particle frequency distribution define ellipses in two dimensions and an ellipsoid, the "orientation ellipsoid," in three dimensions with ellipsoid orientations and axial ratios closely aligned with the finite strain ellipsoid for the matrix. Thus, like AMS, alignment of elongate particles should provide a measure of integrated magma flow direction following dike propagation and freezing of flow features on dike margins. To test further the association of grainscale magma flow with surface features on dikes, we have determined the orientation distribution of aligned plagioclase microphenocrysts for the three dikes, TR05 (Akaki Canyon), TR53, and TR42/43 (both from Onouphrios Canyon) discussed above. To avoid the potentially circular results provided by assuming knowledge of the flow plane, we have determined the orientation and shape of the orientation ellipsoid for the trachytic fabric following the methodology of Harvey and Laxton [1980]. For each dike, three thin sections were cut section coordinates and defined by the tensor function G; where n is number of measurements and x i and Yi are the direction cosines from the orthogonal x and y thin section coordinates, respectively, for the ith measurement. Tensor G

10 ß ß. ß 5250 VARGA ET AL ß DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE up TR42/43 N '2,.:i.',..'i,, o ß ß. up ß '.e;,....t. ß :.:...'. ß :.. ß. '.... e :¾ ß :.?:.' ;.,.. ß '.. ß ß... o... sect i on ß, ß,.. '.:".. ß... ß...,.... ß... up ß.,,,.. 'flow direction 218ø/31 ø Figure 8. Field, AMS, and plagioclase orientation ellipsoidata for dike TR42/43. See caption for Figure 7 for description of symbols except for open square, triangle and circle which are orientations of y], Y2 and Y3, respectively, for AMS data for cores taken on the west margin of dike. flow direction ~025%42 ø section l TR05 " n=302 u section 2 Figure 9. Field, AMS, and plagioclase orientation ellipsoid data for dike TR05. Box with cross indicates location of core 'f,' from which two of the oriented thin sections were cut. See Figures 7 and 8 for further explanation.

11 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE 5251 defines the twodimensional orientation ellipse for the particle distribution and has eigenvectors œ1 and œ2 parallel to the maximum and minimum direction of particle elongation for the distribution. Similarly, eigenvalues of G, E 1 and are the lengths of the maximum and minimum axes, respectively, of the twodimensional orientation ellipse [Harvey and Laxton, 1980]. Similar to the problem of interpreting threedimensional strain from twodimensional strain ellipses [e.g., Ramsay, 1967], we can determine the orientation and shape of the threedimensional orientation ellipsoid from three or more twodimensional ellipses. For each of the three thin sections for the three dikes studied (Figures 7, 8, and 9), the orientation and shape of the twodimensional orientation ellipse was determined. These ellipses were checked for mutual compatibility using the methodology described by De Paor [1990]. For the three examples described here, the ellipse axes were compatible within 5 ø or closer. After final compatibility adjustments, the threedimensional orientation ellipsoid (elongations œ1>e2>e3) was constructed from the sectional data. Equations for this construction can be found in a number of sources, including Ramsay [1967]. We used computer programs (MacStrain 2.4 and 2D>3D Strain) written by K. Kanagawa to build the threedimensional tensors. This step can also be done graphically as described by De Paor [ Ellipticities (R = (I+E1)I(I+E3)) of orientation ellipsoids for plagioclase fabrics are 1.92, 2.27, and 6.3 for samples TR53, TR05, and TR42/43, respectively. Percent anisotropies (%h = 100*(E1E3)/(El+E2+E3))of 22%, 26%, and 53% are several orders of magnitude greater than that shown by the typical AMS fabric in Troodos dikes (%h averages 0.5%). The resulting orientation ellipsoid data correspond closely to the flow directions inferred from both vesicle and hot slickenline lineation data; departures from parallelism are 25 ø or less. For TR53, l is essentially coincident with the uniform surface lineations and cluster of )q directions (Figure 7), indicating that flow determinations by both AMS and hot slickenline orientations reflect grainscale flow close to the dike margins. Likewise, l of the orientation ellipsoid for dike TR42/43 lies sufficiently close to both the AMS )q axes and vesicle lineation direction to suggest that all three methods reflect internal magma flow (Figure 8). The intermediate axes of the AMS and grain orientation ellipsoids do not correspond, however, but lie approximately along the same great circle. For dike TR05, l of the orientation ellipsoid for plagioclase also lies reasonably close to the direction of vesicle elongation (Figure 9). The offset of l to the west of the dike plane may, in part, reflect imbrication of the silicate fabric, as both cores used for the petrographic analysis were taken from the western margin of the dike. However, the orientation ellipsoid for the plagioclase fabric makes a moderate (25 ø) angle to the nearhorizontal mean )q direction. As discussed above, this discrepancy may be related to local variations in the magma flow direction. This interpretation is supported by the much better correspondence (15 ø ) between the plagioclase fabric and AMS for core "f" (which provided two of the three thin sections for the petrographic study of this dike) than for the overall mean 'lq orientation. In addition, the sections from which the orientation ellipsoid was constructed were from cores taken within 1 ½m of the dike margin while the AMS ellipsoid averages cores taken between 1 and 5 ½m of the margin. Because both the surface lineation direction and 1 of the plagioclase orientation ellipsoid diverge from the AMS )'1 direction, it is likely that this discrepancy reflects a real change in magma flow, either spatially or temporally. 5. Structural Corrections of Flow Directional Data Figure 1 shows the orientation of dikes and associated mesoscopic flow lineations as measured in the field for the various sites within the Troodos ophiolite. Each dike and lineation direction are generally the average of between 2 and 10 readings on individual dikes and, in some cases, for groups of subparallel dikes. As is readily apparent, no consistent pattern of horizontal versus vertical flow exists in these dikes in their present orientation. This observation is even more apparent in the display of lineation rakes within dike planes seen in Figure 10a. The average lineation rake is 46 ø, with no skewedness toward either updip or horizontal trends. To consider the true flow direction of magma at the Troodos spreading center, we must restore each dike to its ridge crest position prior to tilting. Dikes of the Troodos ophiolite have suffered a variety of structural rotations since their emplacement at the paleoridge crest. Fortunately, the various structural rotations and their geologic context for the Troodos ophiolite are reasonably wellunderstood as the result of detailed structural and paleomagnetic investigations of the complex [Allerton and Vine, 1987, 1990; Bonhomrnet et al., 1988; Hurst et al., 1994; MacLeod, 1990; Varga, 1991; Varga Updip % Data Updip B % Data 7.25 Horizontal Horizontal Figure 10. Partial rose diagrams showing rakes of 33 field determinations of magma flow within dike planes. Solid arrows show mean rakes of 46 ø and 48 ø for (a) dikes in field orientation and for (b) structurally corrected orientation.

12 5252 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE and Moores, 1985]. Our approach to restoration of dikes to pretilt conditions is to consider reasonable geologic rotational axes based on field evidence and to use, where possible, paleomagnetic data to help constrain magnitudes of rotation and provide a check for vertical axis rotations which are otherwise difficult to verify independently. In general, successful rotations restore dikes to nearvertical orientations [Varga, 1991] and the stable characteristic component of remanent magnetization (ChRM) to the Troodos mean direction (TMD; 2740/36 ø [Clube et al., 1985]) for the Late Cretaceous. For this study, we apply geologically appropriate rotations in reverse order of formation and track the movement of the ChRM direction, where available, with each rotation. We prefer this sequential approach over that of Allerton and Vine [1987] because it uses geologic relations to guide the restoration process rather than solving for a single rotation axis that potentially averages across discrete structural events. Table 1 shows a summary of rotational axes used in our restoration, while Table 2 tracks the orientation of site ChRM directions with each rotation for seven of the sites where palcomagnetic data were collected. The youngest, and there fore first, structural rotation removed is the northward tilt of Cretaceous and younger strata lying above the ophiolite along its northern margin. Our data combined with that from the various Memoirs of the Cyprus Geological Survey suggest northward tilts of about 10 ø (Figure 5) for the eastern group of relatively shallow sites and about 17 ø for the sites to the west. This rotation was not applied to sites deep within the sheeted complex because they lie close to the axis of the structural arch that defines the regional structure of the ophiolite. The second correction applied to most sites was a correction for tilting about horizontal axes related to ridge axis extensional faulting which is ubiquitous in the ophiolite [Varga, 1991]. At Kamara Potamos (Figure 5b) where sheet flows are parallel to overlying sediments, this correction was not applied. However, in most instances where sheet flows are found as screens between dikes, such as in Akaki Canyon (Figure 5a), they are not parallel to overlying sediments and the second correction is warranted. In general, rotation about the average dike strike at a site restores dike ChRM to near the TMD and results in dikes with the steep dips (>78 ø) that characterize the sheeted complex where other geologic evidence exists for original orientations [see Varga, 1991]. At moderate to deep levels in the dike complex where volcanic screens do not exist, the average dike orientation at each location (typically 510 dikes were sampled at each locality) was restored to vertical or to a near vertical orientation. Figure 11 shows ChRM directions for each site before any structural rotations and after each sequential rotation. Considerable scatter in the data are apparent before correction and no site has a direction indistinguishable from the TMD (Figure 11 a). After the stratal tilt correction (Figure 1 lb) and correction for normal fault rotations (Figure l 1 c), sites at Akaki Canyon, Kamara Potamos, Mandres, and Evrykhou are indistinguishable from the TMD at the 95% confidence level. Although having no effect on the rake of magma flow, dikes at several sites appear to have been affected by various degrees of vertical axis rotation. Sites at Onouphrios and Kionia Peak can be restored to a position indistinguishable from the TMD by a vertical axis, counterclockwise rotation (Table II; Figure 11d). Such rotations are justified by the welldocumented evidence for clockwise rotations of dikes near the Arakapas and in the eastern part of the ophiolite [Gass et al., 1994]. Final restoration of the Tembria data (Figure 11d) is more problematic. Dikes at this site are contained within a small region of the dike complex in the Solea graben region of the ophiolite with nearly eastwest strikes as compared to the regional north to west/northwest strike [Hurst et al., 1994; Varga, 1991], although this site is wellremoved from any possible effects of the Arakapas transform domain. We suggest that the structural block containing Tembria was rotated about a vertical axis during obductionrelated strikeslip faulting as documented by Varga [1991]. Although not shown on Figure 11, vertical axis, clockwise rotation of the Tembria data by 52 ø restores this site to a position indistinguishable from the TMD. Figure 12 shows dikes restored to their probable ridge axis orientation. Note that on Figure 12, the Tembria data have been rotated 52 ø clockwise about a vertical axis as discussed above. Comparison with Figure 1 shows that structural corrections have not substantially changed the pattern of flow directions as indicated by mesoscopic lineations on dike surfaces; subhorizontal flow, oblique, and nearvertical flow appear equally represented. Restored lineation rakes within dike surfaces are seen in Figure 8 in comparison to nonrestored lineations. After restoration, the average lineation rake is 48 ø. 6. Interpretations and Conclusions Lineations on the margins of mafic dikes in the Troodos ophiolite are relatively common and potentially provide Table 1. Summary of Rotation Axes Used for Data Correction Site Correction for Correction for Correction for Post Volcanic Tilting Due to TransformRelated Stratal Tilt Normal Faultin[?' Vertical Axis Rotation b' Akaki Canyon 094.3/0.0 (17) 214.1/0.0 (32) none Kamara Potamos 094.3/0.0 (17) none none Evrykhou 090.0/0.0 (10) 139.7/0.0 (6) none Mandres 090.0/0.0 (10) 159.8/0.0 (37) none Tembria 090.0/0.0 (10) 297.6/0.0 (38) none ½' Onouphrios Canyon 094.3/0.0 (17) 12.4/0.0 (30) 000.0/90.0 (46) Kionia Peak none 088.2/0.0 (41) 000.0/90.0 (113) none 165.0/0.0 (35) none none 179.0/0.0 (36) none none 010.0/0.0 (36) none none 025.0/0.0 20) none Data shown are rotation axis trend/plunge (magnitude), all in degrees. a. Rotation restores dike to steep orientations and dike ChRM to statistical agreement with TMD except for sites at Tembria, Onouphrios Canyon, and Kionia Peak. b. Rotations are counterclockwise. c. Restores after 52 ø clockwise rotation (see text for discussion).

13 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE 5253 Table 2. Site Mean ChRM Data Site After Correction After Correction After Correction for Uncorrected for Post Volcanic for Tilting Due to for TransformRelated Measurements a' n Stratal Tilt lqormal Faultinf Vertical Axis Rotation Akaki Canyon 286.1/6.2 (11.0) / /32.5 na b' Kamara Potamos 256.5/23.2 (12.3) /27.3 na na Evrykhou 263.3/21.0 (4.4) / /26.1 na Mandres 272.5/ 12.0 (8.3) / /24.0 na Tembria 192.7/67.3 (14.0) f/ /38.6 na Onouphrios Canyon 321.2/57.2 (6.3) / / /24.0 Kionia Peak 053.3/59.3 (5.9) 11 na 026.3/ /27.2 a. Measurements given as trend/plunge (0t95). b. Correcfon not applied; see text. easily measurable indicators of magma flow. Our data have shown a sufficiently close correspondence between surface indicators of flow and indicators of internal, grainscale flow of dike material to suggesthat such surface indicators are a good proxy for magma flow directions in these dikes. The deviation between internal, but nearmargin indicators of flow and surface lineations is generally small enough in most instances to rely on the more easily measured field criteria. Although surface lineations record magma flow within several centimeters of dike margins, flow directions can change quickly in the interiors of dikes; fortunately, this does not seem to have been a common occurrence in Troodos dikes as surface and AMS flow directions are generally within about 25 ø of parallelism. In addition to changes in flow during ascent of magma along dikes, back flow of magma appears to be a feature of some Troodos and other mafic dikes during their waning stages [e.g., Philpotts and Asher, 1994; Wata, 1992]. It is probable that changes in flow direction are more likely in longlived, thick dikes than in the relatively thin dikes typical of ophiolites and ocean crust. Both surface features and internal indicators (AMS, trachytic textures, vesicle elongation) of flow in dikes from the Troodos ophiolite indicate that magmas rose with no preference for vertical over horizontal flow. Similar results were reported by Rochette et al. [1991] from the S ophiolite in Oman. These results contradict simple models of primarily vertical ascent of magmas along continuous axial magma chambers at ridge crests [Cann, 1970; Kidd, 1977; N N C) ø TMD I i i!! i i i i i KP '' 'T... A. Measured Data B. After Correction for Postvolcanic Stratai Tilt C. After Correction for Tilting due to Normal Faulting D. After Correction for Clockwise Vertical axis Rotation Due to Arakapas Transform Shear Figure 11. Equalarea stereonets showing site mean ChRMs (crosses) and x95 confidence intervals for O, Onouphrios Canyon; K, Kionia Peak; T, Tembria; KP, Kamara Potamos; E, Evrykhou; M, Mandres; A, Akaki Canyon sites; TMD, Troodos mean direction (274ø/36ø). Dotted error ellipses are in upper hemisphere. See text for discussion.

14 5254 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE Mandres ou \ k Akaki I + Kamara \ Tembria Kionia Figure 12. Summary of mesoscopic flow lineation data for various Troodos ophiolite sites following corrections for structural rotations. See Figure 1 for explanation of symbols. The summary of rotations used in these reconstructions is that explained in Tables 1 and 2 except that the Tembria data have been rotated 52 ø clockwise; see text for explanation. Kidd and Cann, 1974]. Rather, the variable flow directions documented here support a model in which magma supply is focused near centralized magma chamber spaced along the ridge crest. In this model, flow is near vertical to steeply oblique near axial volcanoes and more horizontal away from these magmatic centers. As horizontally moving magmas move into areas of the ridge crest near a central volcano, there exists the possibility of juxtaposition of dikes with contrasting flow directions and steepness of flow as well as of dikes with contrasting compositions. Baragar et al. [1987] presented a prescient model to explain mutually cross cutting but compositionally distinct dikes near Kionia Peak (Figure 1). Their model (see Figure 12 of Baragar et al.) predicts many of the flow featuresuggested by our data, particularly the mix of horizontal and more vertical flow in relatively small regions of the sheeted dike complex. The horizontal movement of magma away from central volcanoes helps explain the paradox of slow spreading ridges that have welldeveloped magnetic anomaly lineations yet lack geophysically resolvable magma chambers [Detrick et al., 1990]; much of the crust along such ridges might be constructed by laterally moving magmas. In addition, the possibility of significant horizontal magma transport in dikes suggests caution in interpreting alongaxis variations in lava chemistry [e.g., Langmuir eta/., 1986; Sinton et al., 1991] as necessarily reflecting heterogeneities in the underlying mantle. Ridge parallel geochemical segmentation may, in part, be the result of lateral migration of geochemically distinct magmas. For example, $igurdsson [1987] noted that lateral transport from chemically distinct central volcanoes could effectively explain the enigmatic juxtaposition of alkalic and tholeiitic dikes in Iceland. Flowage differentiation within dikes [e.g., Bhattacharji and Smith, 1964] and magma mixing arising from interaction of laterally moving melts with existing magma bodies [e.g. Garcia eta!., 1989] may introduce further complexity in the interpretation of along axis geochemical variations. We suggest that similar along axis transport at midocean ridges may contribute to the smallscale geochemical heterogeneity that has been documented by detailed nearridge sampling [Perfit eta/., 1994; Reynolds et al., 1992]. Acknowledgments. This work was supported by NSF grant EAR We thank the Cyprus Geological Survey Department and especially its Director, George Constantinou, for their logistical help during our field studies. We also thank Declan De Paor and Kyu Kanagawa for use of various computer routines used to solve for orientation tensors. Critical reviews of this manuscript by Michele Cooke, Michael Knight, and Chiyuen Wang are gratefully acknowledged. References Allerton, S., and F.J. Vine, Spreading structure of the Troodos ophiolite, Cyprus: Some paleomagneticonstraints, Geology, 15, , Allerton, S., and F.J. Vine, Palaeomagnetic and structural studies of the southeastern part of the Troodos complex, in Ophiolites and Oceanic Crustal Analogues, Proceedings of the Symposium '7'roodos 1987'; edited by J.G. Malpas, et al., pp , Geol. Sur. Dept., Nicosia, Cyprus, American Geological Institute, Penrose field conference on ophiolites, Geotimes, 17, 2425, 1972.

15 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE 5255 Baer, G., Mechanisms of dike propagation in layered rocks and in massive, porous sedimentary rocks, J. Geophys. Res., 96, 11,911 11,929, Baer, G., and Z. Reches, Flow patterns of magma in dikes, Makhest Ramon, Israel, Geology, 15, , Baer, G., C. Palmer, A. Heimann, and L. Aharon, Determination of flow directions in Israeli basalts by anisotropy of magnetic susceptibility; preliminary results, Current Res. Geol. Surv. lsr., 9, 2732, Baker, E.T., G.J. Massoth, R.A. Feely, R.W. Embley, R.E. Thomson, and B.J. Burd, Hydrothermal event plumes from the CoAxial seafloor eruption site, Juan de Fuca Ridge, Geophys. Res. Lett., 22, , Mechanisms, edited by A.J. Parker, P.C. Rickwood, and D.H. Tucker, pp , A.A. Balkema, Rotterdam, Netherlands, Ernst, R.E., and W.R.A. Baragar, Magma flow patterns in the Mackenzie giant radial dyke swarm: Evidence from magnetic fabric, Nature, 356, , Fernadez, A., Preferred orientation developed by rigid markers in two dimensional simple shear: A theoretical and experimental study, Tectonophysics, 136, , Fiske, R.S., and E.D. Jackson, Orientation and growth of Hawaiian volcanic rifts: The effects of regional structure and gravitational stress, Piu'los. Trans. R. Soc. Londorg Ser. A, 329, , Fox, C.G., W.E. Radford, R.P. Dziak, T.K. Lau, H. Matsumoto, and A.E. Schreiner, Acoustic detection of a seafloor spreading episode Baragar, W.R.A., M.B. Lambert, N. Baglow, and I.L. Gibson, Sheeted dikes of the Troodos ophiolite, Cyprus, in Mafic Dyke Swarms, edited on the Juan de Fuca Ridge using military hydrophone arrays, by H.C. Halls and W.F. Fahrig, pp , Geol. Assoc. of Can., Geophys. Res. Left., 22, , Vancouver, B.C., Garcia, M.O., R.A. Ho, J.M. Rhodes, and E.W. Wolfe, Petrologic Bhattacharji, S., and C.H. Smith, Flowage differentiation, Science, 145, constraints riftzone processes, Bull. Volcanol., 52, 8196, , Gass, I.D., Ophiolites and oceanic lithosphere, in Ophiolites and Blome, C.D., and W.P. Irwin, Equivalent radiolarian ages from Oceanic Crustal Analogues, Proceedings of the Symposium "I'roodos ophiolitic terranes of Cyprus and Oman, Geology, 13, , ", edited by J.G. Malpas, et al., pp. 110, Geol. Sur. Dept., Bonhommet, N., P. Roperch, and F. Calza, Palaeomagnetic arguments Nicosia, Cyprus, for block rotations along the Arakapas fault (Cyprus), Geology, 16, Gass, I.G., Is the Troodos Complex, Cyprus formed by sea floor , spreading?, Nature, 220, 3942, Brandsdottir, B., and P. Einarsson, Seismic activity associated with the Gass, I.G., C.J. MacLeod, B.J. Murton, A. Panayiotou, K.O. Simonian, September 1977 deflation of Krafla Volcano in northeastern Iceland, and C. Xenophontos, The Geology of the Southern Troodos Transform J. Volcanol. Geotherm. Res., 84, , Fault Zone, Me 9, 218 pp., Geol. Sur. Dept., Nicosia, Cyprus, Bussell, M.A., A simple method for the determination of the dilation Graham, J.W., Magnetic susceptibility anisotropy, an unexploited direction of intrusive sheets, J. Struct. Geol., 11(6), , petrofabric element, Geol. Soc. AM Bull., 65, , Cann, J.R., New model for the structure of ocean crust, Nature, 226, Hargraves, R.B., D. Johnson, and C.Y. Chan, Distribution anisotropy , The cause of AMS in igneous rocks, Geophys. Res. Lett., 18, 2193 Canon, T.E., G.P.L. Walker, and B.E. Herrero, Magnetic fabric and 2196, flow direction in basaltic pahoehoe lava of Xitle Volcano, Mexico, J. Harvey, P.K., and R.R. Laxton, The estimation of finite strain from the Volcanol. Geotherrn Res., 65, , orientation distribution of passively deformed linear markers: Canon, T.E., G.P.L. Walker, and B.E. Herrero, The internal structure of Eigenvalue relationships, Tectonophysics, 70, , lava flows; insights from AMS measurements, II, Hawaiian Hooft, E.E., and R.S. Detrick, The role of density in the accumulation of pahoehoe, toothpaste lava and 'a'a, J. Volcanol. Geotherrr Res., 76, basaltic melts at midocean ridges, Geophys. Res. Left., 20, , 1946, Clube, T.M.M., K.M. Creer, and A.H.F. Robertson, Palcorotation of the Hurst, S.D., E.M. Moores, and R.J. Varga, Structural and geophysical Troodos microplate, Cyprus, Nature, 317, , expression of the Solea graben, Troodos ophiolite, Cyprus, Tectonics, 13(1), , Constantinou, G., and G.J.S. Grovett, Geology, geochemistry and genesis of Cyprus sulphide deposits, Econ. Geol., 68, , Johnson, A.M., and D.D. Pollard, Mechanics of growth of some Con'y, C.H., Laccoliths, mechanisms of emplacement and growth, U.S. laccolithic intrusions in the Henry Mountains, Utah, I, Field observations, Gilbert's model, physical properties and flow of the Geol. Sur. Spec. Pap. 220, 110 pp., magma, Tectonophysics, 18, , Curtie, K.L., and J. Ferguson, The mechansim of intrusion of Khan, M.A., The anisotropy of magnetic susceptibility of some igneous lamprophyre dikes indicated by "offsetting" of dikes, Tectonophysics, and metamorphic rocks, J. Geophys. Res., 67, , , , Kidd, R.K.W., A model for the process of formation of the upper Delaney, T., and D.D. Pollard, Solidification of basaltic magma during oceanic crust, Geophys. J. R. Astron. Soc., 50, , flow in a dike, AM J. Sci., 282, , Kidd, R.K.W., and J.R. Cann, Chilling statistics indicate an ocean floor De Paor, D.G., Determination of the strain ellipsoid from sectional data, spreading origin for the Troodos ophiolite, Cyprus, Earth Planet. Sci. J. Struct. Geol., 12, , Left., 24, , Detrick, R.S., Buhl, E. Vera, J. Mutter, J. Orcutt, J. Madsen, and T. Knight, M.D., and G.P.L. Walker, Magma flow directions in dikes of the Brocher, Multichannel seismic imaging of a crustal magma chamber Koolau Complex, Oahu, determined from magnetic fabric studies, along the East Pacific Rise, Nature, 326, 3541, J. Geophys. Res., 93, , Detrick, R.S., J.C. Mutter, P. Buhl, and I.I. Kim, No evidence from Knight, M.D., G.P.L. Walker, B.B. Ellwood, and J.F. Diehl, Stratigraphy, multichannel reflection data for a crustal magma chamber in the palcomagnetism, and magnetic fabric of the Toba tuffs: Constraints MARK area on the MidAtlantic ridge, Nature, 347, 6164, on the sources and eruptive styles, J. Geophys. Res., 91, 10,355 Dziak, R.P., C.G. Fox, and A.E. Schreiner, The JuneJuly 1993 seismo 10,382, acoustic event at CoAxial segment, Juan de Fuca Ridge: Evidence Langmuir, C.H., J.F. Bender, and R. Batiza, Petrological and tectonic o o for a lateral dike injection, Geophys. Res. Lett., 22, , segmentation of the East Pacific Rise, 5 30'14 30'N, Nature, 322, Dzurizin, D., R.Y. Koyanagi, and T.T. English, Magma supply and storage at Kilauea, Hawaii, J. Volcanol. Geother Res., 21, , Ellwood, B.B., Estimates of flow direction for calcalkaline welded ruffs , MacDonald, W.D., and H.C. Palmer, Flow directions in ashflow tuffs: A comparison of geological and magnetic susceptibility measurements, Tshirege member (upper Bandelier Tuff), Valles and palcomagnetic data reliability from anisotropy of magnetic caldera, New Mexico, USA, Bull. Volcanol., 53, 4559, susceptibility measurements, Earth Planet. Sci. Lett., 59, , MacLeod, C.J., Role of the southern Troodos transform fault in the rotation of the Cyprus microplate: evidence from the Eastern Elston, W.E., and E.I. Smith, Determination of flow direction of rhyolitic Limassol Forest Complex, in Ophiolites and Oceanic Crustal ashflow ruffs from fiuidal structures, Geol. Soc. AM Bull., 81, 3393 Analogues, Proceedings of the Symposium '7'roodos 1987", edited by 3406, J.G. Malpas, et al., pp. 6574, Geol. Sur. Dept., Nicosia, Cyprus, Embley, R.W., W.W.J. Chadwick, I.R. Jonasson, D.A. Butterfield, and E.T. Baker, Initial results of the rapid response to the 1993 CoAxial MacLeod, C.J., and B.J. Murton, On the sense of slip of the southern event: Relationship between hydrothermal and volcanic processes, Troodos transform fault zone, Cyprus, Geology, 23, , Geophys. Res. Lett., 22, , March, A., Mathematische Theorie der Regelung nach der Korngestalt Ernst, R.E., Magma flow directions in two mafic Proterozoic dyke bei affiner Deformation, Z. Kristallogr., 81,285297, swarms of the Canadian Shield as estimated using anisotropy of Miyashiro, A., The Troodos ophiolite was probably formed in an island magnetic susceptibility data, in Mafic Dykes and Emplacement arc, Earth Planet. Sci. Lett., 19, , 1973.

16 5256 VARGA ET AL.: DIKE MAGMA FLOW INDICATORS, TROODOS OPHIOLITE Moores, E.M., and F.J. Vine, Troodos massif, Cyprus and other Sigurdsson, H., Dyke injection in Iceland: A review, in Mafic Dyke ophiolites as ocean crust: Evaluations and implications, Philos. Trans. Swarms, edited by H.C. Halls and W.F. Fahrig, pp. 5564, Geol. R, Soc. Londor Ser. A, 268, , Assoc. of Can., Vancouver, B.C., Mukasa, S., and J.N. Ludden, Uraniumlead isotopic ages of Sigurdsson, H., and S.R.J. Sparks, Lateral magma flow within rifted plagiogranites from the Troodos Ophiolite, Cyprus, and their tectonic Icelandic crust, Nature, 274, , significance, Geology, 15, , Simonian, D.O., and I.G. Gass, Arakapas fault belt, Cyprus: A fossil Nicolas, A., Structure of Ophiolites and Dynamics of Oceanic transform fault, Geol. Soc. AM Bull., 89, , Lithosphere, 367 pp., Kluwer Acad., Norwell, Mass., Sinton, J.M., S.M. Smaglik, J.J. Mahoney, and K.C. Macdonald, Owens, W.H., Strain modification of angular density distributions, Magmatic processes at superfast spreading midocean ridges: Glass Tectonophysics, 16, , compositional variations along the East Pacific Rise 13.ø23 ø, Perfit, M.R., D.J. Fomari, M.C. Smith, J.F. Bender, C.H. Langmuir, and J. Geophys. Res., 96, , R.M. Haymon, Smallscale spatial and temporal variations in mid Smith, R.P., Dyke emplacement at Spanish Peaks, Colorado, in Mafic ocean ridge crest magmatic processes, Geology, 22, , Dyke Swarms, edited by H.C. Halls and W.F. Fahrig, pp. 4754, Geol. Petit, J.P., Criteria for the sense of movement on fault surfaces in brittle Assoc. of Can., Vancouver, B.C., rocks, J. Struct. Geol., 9, , Soto, J.I., Strain analysis method using the maximum frequency of Philpotts, A.R., and P.M. Asher, Magmatic flowdirection indicators in a unimodal deformed orientation distributions: Applications to gneissic giant diabase feeder dike, Connecticut, Geology, 22, , rocks, J. Struct. Geol., 13, , Philpotts, A.R., and C.L. Lewis, Pipe vesiclesan alternate model for Staudigel, H., J. Gee, L. Tauxe, and R.J. Varga, Shallow intrusive their origin, Geology, 15, , directions of sheeted dikes in the Troodos ophiolite: Anisotropy of Pollard, D., Elementary fracture mechanics applied to the structural magnetic susceptibility and structural data, Geology, 20, , interpretation of dykes, in Mafic Dyke Swarms, edited by H.C. Halls and W.F. Fahrig, pp. 524, Geol. Assoc. of Can., Vancouver, B.C., Toomey, D.R.S., G.M. Purdy, S. Solomon, and W. Wilcox, The three dimensional seismic velocity structure of the East Pacific Rise near Ramsay, J.G., FoMing and Fracturing of Rocks, 555 l p., McGrawHill, latitude 93ø0'N, Nature, 347, , New York, Varga, R.J., A statistical study of phenocryst orientation fabrics in a Rautenschlein, M., G. Jenner, J. Hertogen, A.W. Hofmann, J. Kerrich, dacite ignimbrite, J. Volcanol. Geothern Res., 19, 3743, H.U. Schmincke, and W.E.M. White, Isotopic and trace element Varga, R.J., Modes of extension at oceanic spreading centers: evidence composition of volcanic glass from the Akaki Canyon, Cyprus, Earth from the Solea graben, Troodos ophiolite, Cyprus, J. Struct. Geol., Planet. Sci. Lett., 75, , , , Reynolds, J.R., C.H. Langmuir, J.F. Bender, K.A. Kastens, and W.B.F. Varga, R.J., and E.M. Moores, Spreading structure of the Troodos Ryan, Spatial and temporal variability in the geochemistry of basalts ophiolite, Cyprus, Geology, 13, , from the East Pacific Rise, Nature, 359, , Varga, R,J., and E.M. Moores, Intermittent magmatic spreading and Rickwood, P.C., The anatomy of a dyke and the determination of tectonic extension in the Troodos ophiolite: Implications for propagation and magma flow directions, in Mafic Dykes and exploration for black smokertype ore deposits, in Ophiolites and Emplace nent Mechanisms, edited by A. Parker, P. Rickwood, and Oceanic Crustal Analogues, Proceedings of the Symposium '7'roodos D. Tucker, pp. 8197, A.A. Balkema, Rotterdam, Netherlands, ", edited by J.G. Malpas, et al., pp. 5364, Geol. Sur. Dept., Roberts, J.L., and D.J. Sanderson, The intrusive form of some basalt Nicosia, Cyprus, dykes showing flow lineation, Geol. Mag., 108, , Walker, G.P.L., The dike complex of Koolau volcano, Oahu, internal Rochette, P., L. Jenatton, C. Dupuy, F. Boudier, and I. Rueber, Diabase structure of a Hawaiian rift zone, U.S. Geol. Sur. Prof. Pap. 1350, dike emplacement in the Oman ophiolite: a magnetic fabric study , 1987a. with reference to gexx:hemistry, in Ophiolite Genesis and Evolution Walker, G.P.L., Pipe vesicles in Hawaiian basaltic lavas: Their origin of the Oceanic Lithosphere, edited by T. Peters and R.G. Coleman, and potential as paleoslope indicators, Geology, 15, 8487, 1987b. pp. 5582, Kluwer, Norwell, Mass., Wata, Y., Magma flow directions inferred from preferred orientations Rubin, A.M., A comparison of riftzone tectonics in Iceland and of phenocrysts in a composite feeder dike, MiyakeJima, Japan, J. Hawaii, Bull. Volcanol., 52, , Volcanol. Geothern Res., 49, , Rubin, A.M., and D.D. Pollard, Origins of bladelike dikes in volcanic Wilson, L., and J.H.I. Head, Nature of local magma storage zones and rift zones, U.S. Geol. Sur. Prof. Pap., 1350, , geometry of conduit systems below basaltic eruption sites: Pu'u O'o, Ryan, M.P., Neutral buoyancy and the mechanical evolution of Kilauea East Rift, Hawaii, J. Geophys. Res., 93, 14,78514,792, magmatic systems, in Magmatic Processes: Physiochemical Principals, edited by B.O. Mysen, Geochern. Soc. Spec. Pub. 1, pp. Wolfe, E.W., M.O. Garcia, D.B. Jackson, R.Y. Koyangi, C.A. Neal, and , A.T. Olaura, The Puu Oo eruption of Kilauea volcano, episodes 1 Sanderson, D.J., The analysis of finite strain using lines with an initial 20, January 3, 1983 to June 8, 1984, U.S. Geol. Sur. Prof. Pap., 1350, random orientation, Tectonophysics, 43, , , Schiffman, P., L.A. Bettison, and B.M. Smith, Mineralogy and Wolff, J.A., B. Ellwood, and S.D. Sachs, Anisotropy of magnetic geochemistry of epidosites from the Solea graben, Troodos ophiolite, susceptibility in welded tuffs: applications to a weldedtuff dyke in CylmaS, in Ophiolites and Oceanic Crustal Analogues, Proceedings of the Tertiary TransPecos Texas volcanic province, USA, Bull. the Symposium 'Troodos 1987", edited by J.G. Malpas, et ai., pp. 673 Volcanol., 51,299310, , Geol. Sur. Dept., Nicosia, Cyprus, Schmincke, H.U., and U. Bednarz, Pillow, sheet flow and breccia flow volcanoes and volcanotectonic hydrothermal cycles in the Extmsive Series of the northeastern Troodos ophiolite (Cyprus), in Ophiolites J. S. Gee, H. Staudigel, and L. Tauxe, Scripps Institution of and Oceanic Crustal Analogues, Proceedings of the Symposium Oceanography, MC 0215, La Jolla, CA ( jsgee ucsd.edu) '7'roodos 1987", edited by J.G. Malpas, el al., pp , Geol. Sur. R. J. Varga, Department of Geology, The College of Wooster, Wooster, Dept., Nicosia, Cyprus, OH ( rvarga acs.wooster.edu) Schmincke, H.U., and D.A. Swanson, Laminar viscous flowage structures in ashflow tuffs from Gran Canaria, Canary Islands, J. Geol., 75, , Shelley, D., Determining palcoflow directions from groundmass fabrics in the Littleton radial dykes, New Zealand, J. Volcanol. Geother (Received December 13, 1996; revised July 20, 1997; Res., 25, 6979, accepted July 22, 1997.)