Re-examination of textural and structural features of the Skaergaard Intrusion has clarified the process responsible for various forms of layering. Two general types are distinguished, one produced by magmatic flow and another by non-dynamic processes, such as compaction and varied nucleation and growth of crystals. Our concern here is mainly with layering of the first type. Magmatic flow results in contrasting layers and strong foliation accompanied by a linear orientation of elongated minerals. It is associated with slumping and cross-bedding in a zone near the margins of the intrusion. Rocks in the interior of the Layered Series are foliated and commonly layered but have little or no lineation; their layering cannot be due solely to deposition from currents. The principal structural elements produced by dynamic processes in the magmatic state—foliation, lineation, shear zones, folds, and current structures—define parallel belts, mainly around the western and northern margins. Their distribution suggests that the strongest component of flow was for the western side and that its strength decreased toward the east. Shear flow trajectories deduced from lineations and kinematic indicators confirm this asymmetry and point to mass movement toward the southeastern part of the floor.
Despite the intensive investigations focused on mafic and ultramafic intrusions, especially in the last decade or so, no consensus has been reached on the origin of the layering that is such a conspicuous feature of these rocks. Most recent attempts to account for layering have invoked a variety of mechanisms involving nucleation and growth of crystals in gradients of temperature and composition [see Parsons, (1987) and Naslund & McBirney, (1996) for reviews]. Much less attention has been given to dynamic factors related to viscous flow and deformation during consolidation of a zone of crystallization. It is now realized that viscous flow of magmas can produce a variety of features that resemble those in other types of plastically deformed rocks, whereas certain textural features that in the past have been attributed to magmatic flow can be shown to have a non-dynamic origin (Nicolas, 1992). Our purpose here is to clarify the respective roles of dynamic and non-dynamic processes in producing the structural features of layered igneous rocks, such as those of the Skaergaard Intrusion.
The principal geologic and petrologic features of the intrusion have been summarized by Wager & Brown, (1968) and more recently by McBirney, (1996). The structure and compositional variations of the Layered Series have been described in Part I of this series (McBirney, 1989), and crystallization and mechanisms of differentiation were the subject of a more recent study (McBirney, 1995).
Structures and Textures Produced by Magmatic Deformation
Because igneous petrologists may not be familiar with recent studies of the structural aspects of plutonic rocks, it will be helpful to outline some of the principles on which this work has been based. Most of the following summary is condensed from a more complete review by Nicolas, (1992).
Although the term foliation is sometimes used in a general sense for any planar feature, we apply it here only to a plane of mineral flattening that is part of the internal structure of the rock. It may have any origin, including viscous or plastic deformation. It is not synonymous with layering, but includes the ‘igneous lamination’ of Wager & Deer, (1939). Although most foliation is normally attributed to deformation and compaction, some types may result from preferred orientation of crystals growing or recrystallizing in a gradient of temperature, liquid composition, or, more generally, chemical potential. Foliation of this latter kind normally lacks lineation (Jackson, 1961).
Lineation is usually defined by a preferred orientation of mineral grains that are elongated in the plane of foliation, but it can also result from the intersection of two planar features, such as a plane of foliation with a plane of layering. One should therefore make a distinction between mineral lineation and intersection lineation. Although crescumulate and harrisitic textures, commonly attributed to constitutional supercooling, are characterized by a pronounced linear orientation of elongated minerals, the minerals are not aligned in a plane of foliation.
Layering connotes a variation of composition, grain size, or texture that may or may not be associated with a mineral orientation. We do not follow Irvine's, (1987) definition, which recognizes only modal layering, but prefer to include grain-size variations and other textural features that impart a macroscopically visible planar structure, even in rocks of uniform modal compositions.
Although layering may have many forms and origins, it can be divided into two general types, one produced mainly by dynamic processes, such as magmatic flow, and a second resulting from non-dynamic processes, such as differing rates of nucleation and growth of crystals under mechanically static conditions. These are not mutually exclusive, nor are they uniquely related to foliation or layering. Both can produce modal or grain-size variations, and both can have a preferred orientation of crystals.
Layering produced by deformation tends to have sharply defined dark and light layers that in extreme cases may be nearly monomineralic. The spacing is typically on the scale of centimeters or decimeters. Layers of this kind, as well as the lenses and schlieren that are commonly associated with them, have compositions markedly different from that of their host. Some are more mafic, others more felsic, and some have both mafic and felsic rocks in close association. They normally have a marked foliation parallel to the layering or making a small angle with it.
Characteristics of magmatic deformation
Deformation of crystallizing magmas differs in several respects from that of both solid rocks and silicate liquids. Unlike pure liquids, crystal-rich magmas are normally anisotropic. In this sense they have much in common with metamorphic rocks, but they differ from solids in that almost all the strain is taken up by the liquid matrix, the solid particles normally being undeformed. Crystals rarely have dislocation structures, mechanical twinning, strain shadows, or other evidence of plastic strain. This is not to say that minerals are unaffected; in a dense suspension, some may be deformed, but evidence of this strain is usually weak and may later be erased by recrystallization or annealing. Crystals that grow or recrystallize while magma is being deformed may take on new shapes, sizes, or orientations that serve as active markers and record the kinematic conditions under which the magma crystallized.
Because the behavior of a partly crystallized magma is intermediate between that of solid rock and a viscous liquid, the way in which it responds to stress is very sensitive to the proportions of solids and liquid. As the fraction of liquid increases beyond 15 or 20%, bulk viscosity decreases sharply by several orders of magnitude, and beyond a critical melt fraction of ∼40% it approaches that of a liquid. Thus, deformation features produced in magmas with less than the critical melt fraction differ from those in which enough liquid is present to prevent extensive grain-to-grain contact of the suspended solids (Fig. 1). The exact value of this critical melt fraction depends on a variety of factors including the preferred orientation of solid particles suspended in the liquid (Nicolas et al., 1993).
Structures developed during magmatic flow
Development of magmatic foliation and lineation during suspension flow is a complex topic involving a number of factors, such as the flow regime (pure shear, simple shear, etc.) and the concentration and shapes of the solid particles (Nicolas, 1992). We summarize here a few principles that are especially relevant to layered intrusions. Depending on the melt fraction, the different structures produced by deformation may serve as kinematic indicators. With decreasing proportions of liquid, crystals tend to pile up in imbricated structures making a small angle with respect to the shear direction (Den Tex, 1969), and magmatic flow becomes increasingly focused into discrete shear zones where the residual liquid tends to concentrate. The volume between these shear zones has less liquid and behaves more or less rigidly (Fig. 1). In an intermediate range, the result is centimeter- to decimeter-scale ‘S/C structures’ in which the foliation (S) is cut at intervals by narrow bands of shear (C for ‘cisaillement’) making a small angle with the foliation. At even smaller melt fractions, the shear bands grow in size and their spacing increases to a meter scale, with visible offset of planar structures, such as layering (Fig. 2).
Formation of dynamic layering
Compositional layering of the dynamic type can be created in a magmatic medium by one or any combinations of the following mechanisms.
Injections of parallel dikes and sills
This may take place in a thick zone of mush that, despite its weak rheological properties, behaves like a brittle solid under short-term stresses. It has been documented, for example, in the Oman Ophiolite (Benn et al., 1988).
This potentially important mechanism can result from either magmatic or solid-state flow. Two distinct but often related aspects are relevant to layered gabbros: (1) rotation of planar inhomogeneities, such as dikes or fissures, to an orientation close to the deformation plane (Nicolas, 1989); (2) stretching causing elongation of modal heterogeneities into thin streaks. The first of these is seen in the gabbros of the Oman Ophiolite (Benn & Allard, 1989); the second is well documented in granitic intrusions where mafic inclusions have been stretched into parallel lenses, especially near the margins of the intrusion (Williams & Tobisch, 1994).
Flow-induced segregation of solids
Segregation of solids of differing sizes and densities is easily achieved in a flowing medium. For instance, in flows of granular materials of two different sizes, segregation may result from preferential trapping of the small particles in spaces between grains resting on a sloping floor; other factors being equal, the efficacy of trapping probably depends on the magnitude of inertial forces (Jullien et al., 1993). Often referred to as the ‘Bagnold effect’ (Barrière, 1976), this mechanism tends to concentrate crystals in zones of minimum shear. Commonly seen in dikes, segregations produced in this way are induced by the dispersive shear pressure associated with a velocity gradient near the margins of a flowing medium. Solid particles are carried to regions of minimum shear where the velocity gradient is least, typically in the axial part of the dike.
Flow-induced segregation of liquids
Whenever a crystal–liquid assemblage is mechanically disturbed in a stress field, the mobile liquid fraction tends to move toward zones of minimum principal stress to form lenses and schlieren. The ‘filter-press’ mechanism proposed by Bowen, (1928) and more recently by Sleep, (1988) in the context of convection, is based on this principle. Segregations of this kind are common in zones of disturbance, particularly near the margins of intrusions.
Layering may be produced by any of these types of deformation. Mechanical sorting could be effective in dilute suspensions but much less so in zones of partitioned structures where the melt fraction was smaller. New layers that cut across older layers and foliation at oblique angles seem to be the result of segregation of liquids into zones of shear and relative tension. This is seen in schlieren around the northern margins of the Skaergaard Intrusion (Fig. 3). It cannot be the only explanation, however, for some of these schlieren are highly irregular with all orientations with respect to the foliation in surrounding rocks. Moreover, they tend to have extreme compositions, either very felsic or very mafic, that could not have been magmatic liquids. In many places, these bodies appear to be the result of metasomatic replacement of normal gabbro (McBirney & Sonnenthal, 1990; McBirney, 1995).
Structures and Textures Produced by Non-Dynamic Processes
Although a variety of types of non-dynamic layering have been proposed (Naslund & McBirney, 1996), most of these can be grouped in one of two main categories. The first includes layering produced by a number of petrologic processes involving variations in the rates of nucleation and growth of different minerals. These are thought to result from changes of intensive parameters, such as liquid composition, pressure, or oxygen fugacity. A second type is directly related to compaction and expulsion of interstitial liquids from a zone of crystallization. In a sense, it is intermediate between dynamic and non-dynamic processes, for it results from deformation, but it differs from the former in that it involves pure rather than simple shear, and flow is confined mainly to interstitial liquids. The distinguishing marks of these types of layering are only summarized here; a more complete discussion is deferred to Part III where it is dealt with in a more appropriate context (Boudreau & McBirney, 1997).
Layering produced by variations of intensive parameters
Most layering resulting from purely petrologic processes tends to be diffuse, normally with smooth transitions from one unit to the other on a scale of centimeters up to meters. Except in the special case of crescumulate or harrisitic textures, fabrics are isotropic or nearly so, and modal contrasts tend to be weak. Though conspicuous when viewed from a distance, thick layers may be faint and scarcely discernible at close range. Layering of this kind has been ascribed to minor excursions of the liquid composition from its cotectic, possibly as a result of injections of new magma, convective overturn, or some other factor affecting liquidus conditions. Examples in the Skaergaard Intrusion are found chiefly in the Upper Border Series and in the prominent Triple Group near the top of Middle Zone in the Layer Series. The layering of the Marginal Border Series has also been attributed to variations of nucleation and crystal growth, owing, perhaps, to shifting relations between the liquid and cotectic compositions (Wager, 1961). In this case, however, the layers are prominent, perhaps because their slower rate of accumulation caused them to be thinner.
Fine-scale rhythmic layering is another variety of diffuse layering of this kind. It is normally on a scale of a few centimeters and is thought to be due to a process of re-equilibration in which small initial differences of grain size or modal proportions become accentuated and repeated by cyclical solution and crystal growth under conditions similar to those of Ostwald ripening (Boudreau, 1987). In the Skaergaard Intrusion, this type of layering has been found only in a few localized parts of the Layered Series, mainly in the Upper Zone.
Layering produced by compaction
Evidence for compaction layering is both textural and geochemical. The most obvious chemical effect is exclusion of interstitial liquids as seen in lower concentrations of incompatible elements (McBirney, 1995). Petrographic evidence takes the form of bent plagioclase laths, shadowy extinction, and distorted twin planes. Foliation, in itself, is not a reliable criterion, for some of the most strongly laminated gabbros and anorthosites have been shown to have little if any relationship to compaction (Higgins, 1991; McBirney & Hunter, 1995). Lineation is normally weak or totally absent except where local deformation or irregular compaction has caused warping of the layers.
Compaction layering may result in part from mechanical sorting in the manner proposed by Coats, (1936), but most is probably the result of differential solution and reprecipitation (Dick & Sinton, 1979; Boudreau, 1994). In this respect, the mechanism resembles that responsible for fine-scale, rhythmic layers, but it is greatly enhanced by advective transfer of components in liquid expelled during compaction.
Compaction layering is found mainly in the interior region of the Layered Series. It tends to become increasingly sharp upward through the Layered Series and reaches its strongest expression at the horizon of trough structures near the base of Upper Zone B. Above this level, layering is very weak, if not altogether absent. Superimposed on this general pattern is a zone of very prominent layering in the vicinity of blocks that fell from the roof and settled into the middle part of the Layered Series.
Kinematic Analysis of the Skaergaard Intrusion
The distribution of these various types of dynamic and non-dynamic layering can be explained in terms of the processes affecting the magma in different parts of the intrusion. Systematic mapping of structural features can define the extent and trajectories of magmatic flow and thereby lead to a better understanding of internal processes during cooling and crystallization. Recent studies of the layered rocks of the Oman Ophiolite, for example, have revealed that the gabbros have been intensely deformed while in the magmatic state (Nicolas et al., 1988). During this deformation, the gabbro developed a number of macroscopic structures, as well as a marked foliation and lineation. The layering in ophiolitic gabbros can no longer be considered a direct product of crystal accumulation or in situ crystallization. Rather, it is a ‘transposed layering’ in the structural sense, meaning that it results when a heterogeneity in the crystal–liquid state is stretched and rotated into parallel layers or lenses by large amounts of deformation. Structural analyses of these features have provided a means of interpreting the dynamic behavior of the magma.
The Skaergaard Intrusion lends itself well to an analysis of this kind. Thanks to its excellent exposures, convenient size, and wealth of petrologic information, one can identify and interpret a rich variety of internal structures. Many of these were recognized during earlier studies by Wager & Deer, (1939) and Wager & Brown, (1968), but the body as a whole has not been examined systematically in the context of modern structural concepts. In an effort to remedy this deficiency, we have mapped foliations, lineations, and other dynamic structures in a large part of the Layered Series and Marginal Border Series. We have too few measurements in the Upper Border Series to warrant more than general comments on that part of the intrusion.
Magmatic structures in the Skaergaard Intrusion
Well-defined foliation can be found throughout the layered gabbros (Brothers, 1964). This foliation is normally parallel to the layering, but in imbricated zones the two planar elements tend to be oriented at acute angles to one another. It is defined by a subparallel orientation, mainly of plagioclase laths but also of pyroxene and olivine.
Within planes of magmatic foliation, a mineral lineation may be defined by a preferred alignment of elongated grains of plagioclase and, less commonly, pyroxene, olivine, or aggregates of iron–titanium oxides. Conspicuous lineation is confined almost exclusively to layered gabbros of the peripheral parts of the intrusion (Fig. 4). The main exceptions are near blocks that have deformed the host gabbro or in the trough structures where localized compaction caused strong warping of the layers. Elsewhere in the central part of the Layered Series, it is rarely possible to detect lineation, even on flat, glacially polished outcrops that expose tens of square meters of rocks with well-developed foliation. The fact that lineation is so well developed near the margins of the Layered Series but rare in the center suggests that flow was much weaker, if not totally absent, in the central parts of the Layered Series.
As Wager and his co-workers noted (Wager & Deer, 1939; Wager & Brown, 1968), the foliation and layering, which are normally parallel to each other in the center of the Layered Series, dip inward from the borders and, on average, toward the south–central part of the intrusion with an average angle of about 20°. The mineral lineation follows the same general trend, plunging inward from the borders. Although our measurements of lineation in the interior areas are scarce and uncertain (as indicated by thinner symbols in Fig. 4), their consistent trend on the map encourages us to consider them significant.
We have used the distribution of magmatic structures to divide the layered gabbros into two main zones (Fig. 4).
A zone of non-dynamic structures makes up most of the interior of the Layered Series. As we have already noted, apart from the trough structures and disturbed zones around fallen blocks, it is characterized by a lack of lineation and greater proportions of massive rocks with diffuse layering.
A zone of dynamic structures includes what Wager & Brown, (1968) called the ‘cross-bedded belt’ but also a more extensive belt in which the rocks are deformed but lack cross-bedding. It abuts against the Marginal Border Series along the western and northern sides of the intrusion. Deformation features include contrasting dark and light layers offset by listric shear zones, folds, and S/C and imbricate structures. All these kinematic indicators point to a similar shear sense in which the upper unit has slid toward the center of the intrusion (double arrows in Fig. 4). The gabbros have a comparatively strong foliation and lineation.
The contact with the Marginal Border Series on the western side of the intrusion is well defined and normally less than a meter or so wide (Fig. 5). In places, shear bands cut the steeply layered rocks with an inward dip of ∼45°, rapidly rotating to the more moderate dips of the interior parts of the Layered Series. Slices of the Border Series can be found isolated between these shears. The zone of shear flow has an average width of ∼200 m but reaches a maximum of 700 m on the west and disappears around the northeastern and eastern margins.
This zone of dynamic structures can be divided into two parts, an outer subzone of partitioned structures and an inner one dominated by cross-bedding. In the first, the flow-induced surfaces are partitioned between the penetrative foliation and discrete movement zones, such as S/C bands and shear planes. In the second, there is little or no partitioning, despite the presence of complex structures, such as folds and cross-bedding. The deformation evolves from flow that, in the outer subzone, is partitioned on progressively smaller scales with increasing distance from the margin to penetrative suspension flow in the inner zone. This transition shows the declining strength and duration of gravity-driven flow as the slope becomes more gentle toward the interior. It may be the result of a progressive increase of the fluid phase, owing to less mechanical segregation of interstitial liquids.
Evidence of current erosion is found mainly in the inner subzone. The most characteristic structures are cross-bedding, channeling, and disturbed layers, but, as in the outer subzone, both foliation and lineation are well developed. Strongly contrasting dark and light layers tend to have sharply defined boundaries.
Orientation of structural and petrologic units
Structural features caused by magmatic deformation of Lower and Middle Zones show a conspicuous asymmetry between the western and eastern parts of the intrusion. On the west, the Layered Series abuts against the Marginal Border Series with a sharp angular discordance, whereas on the east the two units merge gradually with no clear demarcation. As seen on the map (Fig. 4), however, the partitioned subzone is locally discordant with respect to the petrologic units and remains more or less parallel to the margins of the intrusion. It disappears toward the north where the dip of the contact becomes more gentle. From the western margin inward, flow-related structures become progressively weaker, and the relative waning of lineation indicates that the strength of magmatic flow decreased in that direction. The inner limit of the cross-bedded zone is poorly defined in the northern part of the intrusion. It appears to become parallel to the petrologic units, but the pattern of lineation is everywhere normal to the margins and converges toward the south–central part of the intrusion. Similarly, all shear indicators point to a motion of the upper unit toward the south and center (double arrows in Fig. 4).
As originally postulated by Wager & Deer, (1939; Wager, 1963), magmatic flow is clearly responsible for the sharp layers near the western and northern margins of the Layered Series. The strength of this flow was probably related to the steepness of the walls, but it may have been augmented by compaction and subsidence of the interior parts of the Layered Series. The strong foliation and lineation, shear bands, imbrication, S/C structures, folds, and schlieren were produced by magmatic flow in the crystal mush banked against the steep wall. The cross-bedding and channeling, which extend farther into the interior, were produced by magma descending from the walls and flowing across weakly consolidated crystals on the floor. The limits of the structural units shown in Fig. 4 are related to the intensity of these currents.
Except where currents were generated by masses of material falling from the roof, evidence for magmatic flow is rare beyond the zone adjacent to the western and northern walls. The effects are normally weaker and have no consistent orientation. Although currents were able to erode and winnow weakly consolidated crystals, even in the far interior of the Layered Series (Conrad & Naslund, 1989), little if any of the layering in the interior could have been produced by magmatic flow.
Origin of magmatic flow
The possible relations between the zones of dynamic and non-dynamic layering are illustrated in a schematic east–west section (Fig. 6). The zone of flow-related structures in the western parts of the Layered Series is shown interfingering with the non-dynamic layering in the interior. A division is shown on the map (Fig. 4) between an outer belt of shear structures and an inner one of lamination produced by magmatic flow. The gravity-driven flow seems to have persisted as long as crystallization continued on the steep western wall, but as the height of the wall decreased, the energy of these currents decreased, and the zone of dynamic flow retreated. When the crystallizing fronts on the floor and under the roof finally converged, the steep slopes at the margins could no longer generate strong currents.
The directions of lineation shown on the map in Fig. 4 indicate a flow convergence toward the south–central part of the floor, suggesting that there was a sink in this area. Compaction, which would have been more important where the intrusion was thickest, would tend to perpetuate this depression. It is worth noting, however, that the location of the sink does not correspond precisely to the roots as defined by gravity (Fig. 7); it appears to have had its center 1 or 2 km to the east. The difference may be due to greater amounts of crystals coming from the steep western wall and producing a pattern of asymmetrical deposition displacing the center of accumulation to the east of the root.
Thus we find that, in most respects, the structural evidence is consistent with gravity-driven flow, with the strongest component coming from the steep western side. The initial asymmetry of the body was probably the result of the regional tectonic environment into which the magma was intruded; it does not seem to have developed during the course of crystallization, for the magnetic orientation of the gabbros indicates that the body was tilted toward the south-southeast after the Layered Series had solidified but was still above the Curie temperature (Schwarz et al., 1979).
Layering produced by magmatic flow can be distinguished from that due to non-dynamic processes. The first is normally sharp with strong modal variations that are typically on a scale of centimeters. Layered rocks that have undergone simple shear have well-developed foliation and lineation, but layering associated with pure shear and compaction has a strong mineral lineation only where it has been superimposed by other forms of deformation that have caused distortion of the layering. Layering caused by differing rates of nucleation and crystal growth tends to be more diffuse with more gradational variations on a scale ranging from centimeters to meters, but it may be augmented by later recrystallization.
Dynamic processes were strongest along the western and northern margins of the intrusion. They declined as the zones of crystallization on the floor and under the roof converged and crystallization on the steep western wall finally came to an end. In the interior, however, non-dynamic layering became stronger as the rate of crystallization declined and compaction became more effective. Layering in the Upper Border Series is of the diffuse, non-dynamic type, as would be expected in the absence of gravity-driven deformation.
Kinematic analysis of the magmatic deformation recorded in the Layered Series shows that the pronounced asymmetry of flow directions resulted from the steeper inclination of the western wall. The direction of flow was toward a sink located in the south–central part of the floor.
This work has been supported by grants from the National Science Foundation to A.R.M. and from CNRS-INSU, Contribution 401 from Contract 803824, to A.N.