Determining sediment transport direction in ancient mudrocks is difficult. In order to determine both process and direction of mud transport, a portion of a well-mapped Cretaceous delta system was studied. Oriented samples from outcrop represent prodelta environments from ca 10 to 120 km offshore. Oriented thin sections of mudstone, cut in three planes, allowed bed microstructure and palaeoflow directions to be determined. Clay mineral platelets are packaged in equant, face-face aggregates 2 to 5 μm in diameter that have a random orientation; these aggregates may have formed through flocculation in fluid mud. Cohesive mud was eroded by storms to make intraclastic aggregates 5 to 20 μm in diameter. Mudstone beds are millimetre-scale, and four microfacies are recognized: Well-sorted siltstone forms millimetre-scale combined-flow ripples overlying scoured surfaces; deposition was from turbulent combined flow. Silt-streaked claystone comprises parallel, sub-millimetre laminae of siliceous silt and clay aggregates sorted by shear in the boundary layer beneath a wave-supported gravity flow of fluid mud. Silty claystone comprises fine siliceous silt grains floating in a matrix of clay and was deposited by vertical settling as fluid mud gelled under minimal current shear. Homogeneous clay-rich mudstone has little silt and may represent late-stage settling of fluid mud, or settling from wave-dissipated fluid mud. It is difficult or impossible to correlate millimetre-scale beds between thin sections from the same sample, spaced only ca 20 mm apart, due to lateral facies change and localized scour and fill. Combined-flow ripples in siltstone show strong preferred migration directly down the regional prodelta slope, estimated at ca 1 : 1000. Ripple migration was effected by drag exerted by an overlying layer of downslope-flowing, wave-supported fluid mud. In the upper part of the studied section, centimetre-scale interbeds of very fine to fine-grained sandstone show wave ripple crests trending shore normal, whereas combined-flow ripples migrated obliquely alongshore and offshore. Storm winds blowing from the north-east drove shore-oblique geostrophic sand transport whereas simultaneously, wave-supported flows of fluid mud travelled downslope under the influence of gravity. Effective wave base for sand, estimated at ca 40 m, intersected the prodelta surface ca 80 km offshore whereas wave base for mud was at ca 70 m and lay ca 120 km offshore. Small-scale bioturbation of mud beds co-occurs with interbedded sandstone but stratigraphically lower, sand-free mudstone has few or no signs of benthic fauna. It is likely that a combination of soupground substrate, frequent storm emplacement of fluid mud, low nutrient availability and possibly reduced bottom-water oxygen content collectively inhibited benthic fauna in the distal prodelta.
The erosion, transport and deposition of mud are currently subject to intense interest. In the marine realm, the classical view, that river-supplied mud is deposited vertically from suspension in relatively deep, quiet water, is rapidly being dispelled (Schieber, 1994a,b; Macquaker & Bohacs, 2007; Schieber et al., 2007; Hovikoski et al., 2008; Bhattacharya & MacEachern, 2009; Ichaso & Dalrymple, 2009; Aplin & Macquaker, 2011; Ghadeer & Macquaker, 2011, 2012). Studies of ancient mudstone successions (e.g. Schieber, 1994a,b, 1998, 1999; Macquaker & Keller, 2005; Ghadeer & Macquaker, 2011) have shown that apparently monotonously laminated or massive mudrocks actually contain a wealth of sedimentary features indicative of deposition from vigorous currents with abundant evidence of syn-depositional erosion and lateral transport of muddy sediment, usually in the form of various types of aggregate grains. Schieber et al. (2007) and Schieber & Southard (2009) have shown that mud, transported as bedload aggregates, can be moulded into ripple bedforms at current velocities comparable to those that transport fine sand. Additional flume experiments show that weakly consolidated mud (ca 85% water) can be eroded to form platy intraclasts, and the eroded surface can have steep and complex relief on a millimetre to centimetre-scale (Schieber et al., 2010).
Bhattacharya & MacEachern (2009) provided a comprehensive review of current thinking on the subject of mud transport in deltaic and shelf settings. These authors emphasized the practical difficulty of determining microstructure, and hence deciphering transport processes in ancient mudrocks; it was concluded that muddy hyperpycnites were probably much more common in the geological record of shelf seas than are appreciated presently. Bhattacharya & MacEachern (2009) examined cores through allomember E of the Upper Cretaceous Dunvegan Formation (Fig. 1), in the Western Canada Foreland Basin. These heterolithic, silty and sandy prodelta deposits contained features typical of hyperpycnites that were inferred to have been deposited from density flows that issued directly from the mouths of flooding rivers, and/or had been formed and sustained by storm waves. Ichaso & Dalrymple (2009) and MacKay & Dalrymple (2011) also illustrated and provided diagnostic criteria for fluid mud deposits in nearshore, tide-dominated and wave-dominated settings. These authors pointed out that, although modern examples of dense fluid mud flows are now reasonably widely documented, descriptions of their ancient counterparts are still very few.
Thin (≪ 1 m thick), gravity-driven downslope flows of fluid mud, in which mud was supported by storm wave-generated turbulence rather than by auto-suspension, were documented from the Eel shelf of California (Ogston et al., 2000; Traykovski et al., 2000), and recognized subsequently on other muddy shelves (e.g. Traykovski et al., 2007). Macquaker et al. (2010) made an important link between the process of ‘wave-enhanced sediment gravity flow’ (WESGF) and the resulting sedimentary deposit. These authors described, from the Eel shelf, event beds that had a distinctive tripartite microfacies succession (rippled, laminated and structureless), that was interpreted to record a change from turbulent to laminar flow as the concentration of suspended mud increased during a WESGF, followed by gelling of fluid mud as wave energy diminished. Similar tripartite beds were identified in Jurassic and Cretaceous shelf mudstone successions and interpreted as evidence for WESGF on these ancient shelves.
The recognition that some ancient shelf mudstone successions are composed largely of clay minerals packaged into various types of aggregate grain (e.g. Schieber et al., 2010; Plint et al., 2012; Cheadle & Jiang, 2013) implies that, in addition to widespread mud deposition, these environments were also subject to pervasive storm wave reworking that lead to wholesale ‘repackaging’ of cohesive clay-size material into silt-sized grains to coarse sand-sized grains, with attendant changes in the hydraulic behaviour of the original clay mineral crystals.
The problem: Sediment transport direction across the Dunvegan prodelta
Because the core examined by Bhattacharya & MacEachern (2009) was not oriented, there remained uncertainty as to the palaeoflow direction with respect to both palaeo-shoreline and palaeo-bathymetry; nevertheless, this study provided inspiration to examine prodelta mudrocks of the Dunvegan Formation at outcrop where palaeoflow data could be measured. Because both the palaeogeography of the delta-front sandstones and the strike and dip of the prodelta area are well-known from regional mapping (Plint, 2000, 2003), it is possible to view site-specific palaeoflow data in a palaeogeographic and palaeobathymetric context, and hence understand palaeoflow patterns in relation to shoreline trend, distance from shore and bathymetric slope.
Previous study of sediment dispersal across the Dunvegan prodelta (Plint et al., 2009) found that wave-formed structures in fine to very fine-grained sandstone were restricted to the upper parts of deltaic clinothems. Sandstone beds were interpreted to have been deposited in a palaeo water depth of no more than ca 40 m; a depth that was interpreted as storm wave base. The lower, more seaward parts of clinothems, which consist of silt and clay, extend to palaeo water depths inferred to have been as much as ca 100 m in some allomembers. However, Plint et al. (2009) acknowledged that they had neither data on sediment transport process nor transport direction from this offshore part of the prodelta succession. The question therefore arose: What process or processes transported silt and clay across that more distal part of the prodelta? The purpose of the present investigation is, therefore, four-fold: (i) to investigate the microstructure of the mudrock and hence determine the possible origin of grain types and the effect of sediment reworking; (ii) to document mudstone microfacies and hence determine possible transport processes; (iii) to obtain palaeocurrent information from both sandstone and mudstone facies; and (iv) to reconstruct the palaeogeographic and palaeobathymetric setting of sand-rich and mud-rich prodelta facies. This study has shown that the Dunvegan prodelta mudstone contains abundant silt-sized aggregates of clay minerals that appear to preserve substantial original porosity despite burial to > 3 km; these grains imply a high degree of wave and current reworking of the prodelta to an inferred water depth of ca 70 m.
Stratigraphic Setting of The Study Site
The Dunvegan delta complex
The Dunvegan Formation represents a large delta complex, of Lower to Middle Cenomanian age, that prograded more than 400 km from north-west to south-east along the axis of the Western Canada Foreland Basin (Bhattacharya & Walker, 1991a; Plint, 2000; Plint et al., 2009; Fig. 1). The Formation undergoes dramatic lateral facies change from delta plain in the north-west through sandy delta-front to muddy prodelta towards the south-east. A genetic stratigraphy was first developed for this Formation by Bhattacharya & Walker (1991a), by using regional marine flooding surfaces, recognizable in wireline logs and core, to define seven allomembers within a subsurface study area in west-central Alberta. This stratigraphic framework was subsequently expanded northward and westward to encompass 10 allomembers; these allomembers were traced to outcrop in the Rocky Mountain Foothills and in the Peace River valley (Plint, 2000). Prodelta mudstones form prominent clinothems that dip to the south-east and downlap onto a highly radioactive phosphatic sandstone that constitutes a condensed section. This phosphatic sandstone was termed the ‘Fish Scales Upper’ (FSU) marker by Bhattacharya & Walker (1991a), and is recognized readily in gamma ray and resistivity logs.
An extensive well log data base (>3000 wells) enabled the regional isopach, and hence the palaeobathymetric pattern for each allomember, to be mapped. In addition, the regional palaeogeography was reconstructed by using gamma ray logs to map the progradational limits of successive delta-front sandstones (Bhattacharya & Walker, 1991b; Plint, 2000, 2003; Hay & Plint, 2009). This comprehensive regional picture makes it possible to place individual outcrop sections exposed in the Rocky Mountain Foothills in their stratigraphic, palaeogeographic and relative palaeobathymetric context (Fig. 2).
Linking outcrop data to subsurface
An outcrop section through allomember G of the Dunvegan Formation, well-exposed on the Smoky River near Grande Cache, Alberta (Fig. 3), was selected for detailed investigation. Because the distinctive phosphatic sandstone of the FSU marker was recognizable at the Smoky River site (Plint, 2000; Fig. 3), the base of allomember G could be identified. The most regressive allomember G delta-front sandstone did not extend as far south as the Smoky River site, and hence the uppermost part of the studied succession consists of centimetre-bedded heterolithic facies of offshore aspect. The upper allomember boundary is marked by a regionally traceable intraclastic transgressive lag. The sandier-upward facies successions in allomembers G and F exposed in outcrop can be matched with similar successions in the closest available wells 7-5-58-6W6 and 7-24-60-7W6, and hence can be related to the regional allostratigraphic framework (Plint, 2000; Fig. 4).
Lateral facies relations in allomember G
A grid of regional correlation lines (Plint, 2000), permits vertical and lateral facies relations to be determined, from which the palaeogeography of successive deltas can be reconstructed. From the 7-5-58-6W6 tie well (Fig. 4), the silty claystone in the lowest part of the Smoky River section (denoted a in Figs 4 and 5) can be traced updip for ca 120 km where it passes laterally into nearshore sandstone (inferred from the gamma ray log character) of the contemporaneous delta-front at point a′ in Fig. 4. This stratigraphic relationship indicates that the mudstone at the base of allomember G at the Smoky River site was deposited as much as 120 km from the contemporaneous shoreline. The top of the sampled succession (Fig. 5, point c), and equivalent level in subsurface (Fig. 4, point c), is relatively sandstone-rich; this most regressive part of the succession is stratigraphically equivalent to the most regressive part of the contemporaneous delta-front sandstone, located between wells 10-13-62-9W6 and 2-20-61-7W6 (point c′ in Figs 4 and 5). Correlation through well log data suggests that the heterolithic muddy sandstone, projected into Fig. 4 at point c, was deposited ca 25 km seaward of the contemporaneous shoreline. Alternatively, correlation of the Smoky River section directly updip (north-west) to the nearest equivalent delta-front sandstone exposed on Sheep Creek and Beaverdam Creek (Plint, 2000; Fig. 2), suggests that point c in the Smoky River section was located only ca 10 to 15 km offshore. The top of an intermediate upward-shoaling succession (point b in Figs 4 and 5) can be traced updip over ca 60 km where it passes into contemporaneous delta-front sandstone (point b′ in Fig. 4). Thus, the stratigraphic succession between points a and c (Figs 4 and 5) can be shown to have been deposited between as much as ca 120 km and as little as ca 10 km offshore; the spatial relations of points a, b and c are shown in map view in Fig. 2.
Sample Preparation Method
The outcrop section on Smoky River provided the opportunity to make in situ palaeoflow measurements in the sandier part of the succession. Sandstone beds were excavated carefully, permitting the trends of wave ripple crests, gutter casts and the migration direction of combined-flow ripples to be measured directly in the field, structural tilt being removed using the carpenter's rule method (Andrews, 1982).
At the study site, intense frost shattering (annual temperature range ca −40 to +30°C) has rendered the rock too fractured to permit block samples to be removed. In consequence, it was necessary to take ‘box cores’; steel electrical junction boxes proved ideal for this purpose. The most heavily frost-shattered surface layer was excavated to a depth of 0·5 to 0·7 m and a smoothed, clean surface was prepared. The sample boxes, 60 mm wide, 100 mm tall and 45 mm deep, were hammered into the surface, oriented with respect to North before extraction, then taped up to stabilize the sample during transport. Eleven samples, located in Fig. 5, were obtained.
In the laboratory, samples were opened, dried at room temperature for six weeks, then warmed to 50°C, after which the open face of the sample box was saturated with liquid Epo-Fix® (Electron Microscopy Sciences, Hatfield, PA, USA) resin and allowed to cure for a week at room temperature in order to stabilize and maintain the integrity of the samples. The steel box was sawn progressively away from the sample using a diamond saw running without lubrication to minimize the effects of clay hydration and subsequent sample distortion that would result from continued contact with water. Each freshly sawn face of the mudstone block was saturated with resin and allowed to cure, yielding a block of resin-impregnated mudstone that was then dry-sawn in the north/south, east/west and north-west/south-east planes, yielding 33 individual slabs. The cut faces of each slab were again saturated with resin and cured in an oven at 50°C for 24 h. Each slab was polished with 400 grade Carborundum grit, and mounted on a 50 × 75 mm glass slide (600 grade grit was found to make the surface too smooth, weakening its adhesion to the glass slide). Thin sections were ground to a thickness (ca 25 μm) where sedimentary lamination could be readily seen; no cover glass was used. Each thin section was scanned on an Epson Perfection 3170 flatbed scanner (Epson America Inc., Long Beach, CA, USA) at 1200 dpi using transmitted light film scan mode. Enlarged colour prints from these scans then provided ‘maps’ of each sample. Micro-stratigraphic sections for each thin section were prepared using a binocular microscope to observe textural and structural features that were then drafted directly on a strip log laid across the paper print. Although all samples had been variously frost-shattered, it was generally possible to recognize and match individual fragments and reconstruct the full stratigraphy in each slide. It was found that coating the uncovered surface of the slide with a thin film of mineral oil significantly improved the resolution of the microstructure.
The least frost-shattered thin section of the three prepared from each of the 11 box cores was chosen as representative of each sample, and provided the basis for the analysis of mudstone microfacies through 41 m of sampled section. Note that the 11 box samples collectively span a vertical distance of only ca 670 mm (after trimming and thin-section preparation), and therefore the statistics given below are for a very small sample of ca 200 individual beds (sedimentation events). Nevertheless, the same suite of microfacies is observed in each sample, regardless of height in the stratigraphic section, and hence the samples are believed to be reasonably representative of the entire succession.
Silt layers containing ripple cross-lamination were digitally traced (during which fractures were restored), in each of the 33 oriented thin sections and, for each sample, the three planes were integrated into a single, summary three-dimensional representation from which it was possible to make the best estimate of true palaeoflow direction. Inevitably, this is an imprecise method, and results were binned into 45° sectors (north, north–east, east, south–east, etc.).
In order to observe microstructure at the micron scale, oriented mudstone slabs ca 4 × 5 × 30 mm were snapped off of the master slabs, mounted on standard glass slides and carbon coated in order to observe grain structure in broken surfaces. Additional oriented samples were mounted in the same way but polished and carbon coated. Samples were examined in both secondary electron and backscattered mode using a Hitachi SU6600 VP-FEG scanning electron microscope (SEM; Hitachi High Technologies America Inc., Dallas, TX, USA) at 15 kV and a working distance of ca 9 mm.
Mudstone lithology and microstructure
The lower part of the studied section, between samples 1 and 6, consists entirely of silt and clay. Above sample 6, millimetre to centimetre-scale beds of very fine to fine-grained sandstone become increasingly common upward (Fig. 5). Samples 7 to 11 were taken between sandstone beds in order to focus on mudstone lithology throughout the studied section. All the imaged samples described below were chosen from the lowest part of the section in which bioturbation, even at microscopic scale, is not evident; this strategy was intended to eliminate, as much as possible, the complicating effect of bioturbation on mud microstructure (O'Brien, 1987).
Examination in transmitted light revealed quartz and chert grains to be angular to sub-angular, and of coarse to very fine silt size; mica flakes and black to brown organic fragments are accessory components (Fig. 6C). Quartz silt grains are organized into well-defined layers, some only one to two grains thick (Fig. 6D). Some silt layers contain cross-laminae of pale yellow-brown particles composed of clay minerals (Fig. 6A). Many of these particles consist of several smaller sub-grains (Fig. 6B). Scattered amongst the amorphous clay particles are more opaque, near-circular structures, typically 5 to 10 μm in diameter that subsequent SEM observation revealed to be weathered pyrite framboids (Fig. 6B and C). Low-power SEM backscatter images of polished vertical sections revealed distinct laminae enriched in medium-grained and fine-grained quartz silt, alternating with laminae in which clay minerals, mica flakes and fine-grained quartz silt are the dominant components (Fig. 7). In places, clay minerals appear to be packaged into fine to medium-grained, silt-sized intraclastic aggregates (terminology of Plint et al., 2012; Fig. 7B).
Examination of both broken and polished sections with SEM shows that clay mineral particles appear to be packaged at three different scales (Figs 8 to 12). From smallest to largest, these are here termed ‘domains’, ‘face-face (FF) aggregates’ and ‘intraclastic aggregates’.
At the smallest scale, near-parallel packages of clay platelets, ca 0·5 to 1 μm in length and of subtly lenticular cross-section, appear to correspond to the ‘domains’ of Bennett et al. (1981, 1991; Fig. 8A). Domains are, in turn, packaged into larger, roughly equant ‘face-face aggregates’ (terminology of O'Brien, 1987; Nishida et al., 2013), typically 2 to 5 μm in diameter, composed of broadly parallel clay domains (Figs 8B, 8C and 10). In broken surfaces (Figs 11 and 12A), the ‘ragged’ margins of clay domains are readily recognized, and it is possible to distinguish individual FF aggregates by tracing microscopic ‘unconformities’ where different groups of domains intersect at different angles (Fig. 12).
At the largest scale, FF aggregates are packaged into ‘intraclastic aggregates’ that are typically 5 to 15 μm, exceptionally 20 μm or more, in diameter and comprise a few to many, tightly packed FF aggregates. Intraclastic aggregates may also contain particles of quartz silt and organic matter (Figs 8 to 10). Intraclastic aggregates commonly have a distinct internal texture and fabric that make them distinguishable from adjacent grains (Figs 7 to 11). It is not uncommon to see a thin layer of clay domains ‘wrapped’ around part of the periphery of an intraclastic aggregate (Fig. 8C). Intraclastic aggregates are distributed through a matrix of FF aggregates, and are readily recognized in thin section (Fig. 6B), in polished sections (Figs 8 to 10) and in broken surfaces (Fig. 11). Broken surfaces show that the boundaries of aggregates are characterized by narrow zones where clay domains are aligned parallel to the margin of the aggregate (Fig. 11). Sub-spherical pyrite framboids, typically 5 to 10 μm in diameter, are distributed throughout the clay matrix. However, EDS spectra from many framboids lack an iron peak, suggesting that pyrite has undergone surface weathering and is largely preserved as crystal moulds (Fig. 11).
Detrital mica flakes exhibit well-developed cleavage planes, but are distinctly ‘blotchy’ in backscatter mode, suggesting some degree of diagenetic alteration (Fig. 9B and C). Detrital mica flakes, although having a broadly bedding-parallel orientation (Fig. 7), are sometimes deformed around quartz silt grains and, rarely, are deformed around intraclastic aggregates as a result of compaction (Fig. 10).
In summary, SEM images show that clay mineral particles lack a strong, bedding-parallel preferred orientation (cf. O'Brien, 1987; O'Brien et al., 1998); instead clay particles have a random orientation and are organized into a hierarchy of aggregates ranging from ca 1 μm platy ‘domains’ through ca 2 to 5 μm approximately equant ‘face-face aggregates’ to ‘intraclastic aggregates’ ca 5 to 20 μm in diameter and consisting of closely packed FF aggregates and fine quartz silt (Fig. 13). The effect of this packaging of clay platelets is to give the sediment a poorly sorted ‘grainy’ texture.
Interpretation of grain microstructure
Possible modification of original microfabric
The fact that the sampled rock has undergone weathering and repeated freeze-thaw cycles, as well as deep burial, raises the possibility that the microstructure might be an artifact of, or modified by, freeze-thaw and/or the growth of diagenetic clays. However, a number of lines of evidence suggest that the primary fabric is largely preserved. The clay domains and FF aggregates are interpreted as primary sedimentary particles rather than diagenetic clays because the clay flakes have a ‘ragged’ appearance (Fig. 12A), and lack both the well-developed crystallinity and microfabric typical of clay cements (e.g. Houseknecht & Pittman, 1992; Carozzi, 1993; Worden & Morad, 2003; Figs 8 and 12). The rocks have been buried to ca 3 km, and probably experienced a temperature of ca 100 to 120°C. Under these conditions some diagenetic kaolinite, and possibly illite may be expected to have formed (Longstaffe et al., 1992). However, none of the samples examined with SEM showed diagenetic clay textures, such as those illustrated by Longstaffe et al. (1992); this probably reflects both a lack of pore space in which diagenetic clays could form, and low permeability which would inhibit introduction of ions necessary for clay mineral growth. Because intraclastic aggregates typically have a distinct internal texture and fabric, lack evidence for internal recrystallization or deformation, and are consistently bounded by aligned clay platelets, they are also interpreted as primary sedimentary grains.
The observation that micro-lamination, visible in transmitted light in all thin sections, shows no evidence of penetrative, ductile deformation, suggests that freeze-thaw has had no effect on the microstructure; the effect of ice appears to be limited to pervasive brittle fracturing on a scale of millimetres to tens of millimetres. Shrinkage during drying of the samples appears to have been concentrated around the margins of intraclasts, and has served to emphasize original grain boundaries. It is therefore concluded that the microfabric observed with SEM is largely representative of the original grain organization, albeit after substantial compaction. Support for this conclusion is provided by O'Brien (1987), who showed that the microfabric of a wide variety of Palaeozoic mudstones had not been modified significantly by burial and compaction: an originally random orientation of flocculated clay aggregates remained random, despite burial compaction. Similarly, flakes that were oriented randomly as a result of bioturbation retained this orientation during burial (O'Brien, 1987). Because macroscopic or microscopic bioturbation is not evident in the Dunvegan samples studied with SEM, it is concluded that the microstructure probably reflects an original depositional fabric of randomly oriented F-F aggregates, rather than biological reorganization (cf. O'Brien, 1987; Bennett et al., 1991; Schieber, 1994b).
As a further check, prodelta mudstone from Dunvegan allomember E was sampled in core from well 15-31-62-26W5 at a depth of 1940 m, where there was no possibility of freezing. Mudstone was resin-impregnated and polished in the same manner as the outcrop samples, and imaged in electron backscatter mode. Clay domains, FF aggregates and intraclasts indistinguishable from those seen in outcrop samples were observed. Electron backscatter imaging of Cretaceous Colorado Group mudstone sampled in core, 1866 m subsurface, prepared with ion-milled surfaces (Cheadle & Jiang, 2013), also revealed clear evidence of clay floccule domains and aggregate grains that showed little evidence of bedding-parallel compaction. It is concluded that the observed microstructure must be an original sedimentary feature, although the preservation of porosity and lack of pronounced compactional flattening of aggregate grains remain to be explained.
Origin of clay domains, face-face aggregates and intraclasts
Clay ‘domains’ comprising ‘two or three’ clay flakes were described by Moon (1972), and figured and defined by Bennett et al. (1991) as: “…a multiplate particle composed of parallel or nearly parallel plates…”; domains can be considered as the fundamental unit particle that have significant structural integrity and comprise the ‘building blocks’ of clay sediment (Bennett et al., 1991). Clay plates are likely to be bound by electrostatic and van der Waals forces, particularly in salt water where repulsive surface charges on clay plates are balanced by mobile ions, especially Ca 2+ and Na+ (Rand & Melton, 1977; Grabowski et al., 2011). Following deposition, initial edge to face particle contacts gradually change to face to face orientation (Bennett et al., 1991).
The densely packed and randomly oriented FF aggregates described above resemble those illustrated by O'Brien (1987) from Permian mudstones deposited under evaporitic conditions where high salinity promoted intense flocculation. Experiments by Nishida et al. (2013) showed that abundant, dense aggregates with face-face contacts were produced in fluid muds suspended in dilute sea water containing between 10 and 30 g l−1 suspended bentonitic clay. In contrast, clay suspensions with only 1 g l−1 produced random particles with edge–face and edge–edge contacts. The experimental FF aggregates were approximately equant and ranged in mean size from 4·6 to 6·5 μm. Similar FF aggregates, but with a mean diameter of 9·4 μm were observed in natural estuarine sediments. Nishida et al. (2013) postulated that the clay particles that formed the FF aggregates were bound by van der Waals forces that became more effective under saline conditions and where clay particles were closely packed, as in a dense fluid mud. However, the details of the mechanism that produced FF aggregates were not understood. It would seem likely that the strength and integrity of FF aggregates would be increased by a coating of extracellular polysaccharides (e.g. Grabowski et al., 2011).
Intraclastic aggregates, composed of tightly packed FF aggregates, and sometimes including quartz silt, are interpreted as the product of storm reworking of shallowly buried mud that had become sufficiently cohesive that the component FF aggregates resisted disaggregation during erosion. The cohesion and adhesion of clay particles increase on a time scale of days to decades as a result of a variety of mechanisms that include compaction, electro-chemical bonding and the production of extracellular polysaccharide coatings by bacteria and diatoms (e.g. Melton & Rand, 1977; Sutherland et al., 1998; Winterwerp & van Kesteren, 2004; Ravisangar et al., 2005; Debnath et al., 2007; Lundkvist et al., 2007; Grabowski et al., 2011). The presence of a thin layer of clay domains apparently ‘wrapping’ around the margin of some intraclastic aggregates (for example, Fig. 8C) suggests that the exterior of the intraclasts was sticky, causing fine clay particles to adhere during transport.
Thus, a variety of physico-chemical and bio-chemical processes operate from the sediment surface to the shallow subsurface environment to convert loosely bound clay particles into a cohesive sediment. Subsequent erosion of that material by currents will yield cohesive grains, rather than dispersed clay flakes (cf. Einsele et al., 1974; Ravisangar et al., 2001; Schieber et al., 2010; Plint et al., 2012; Figs 6 to 11).
In summary, the Dunvegan prodelta mudstone is composed of quartz silt grains and platy clay domains, the latter organized into FF aggregates typically ca 2 to 5 μm in diameter that give the sediment a strongly ‘grainy’ appearance. The recent recognition that FF aggregates form in abundance in fluid muds of various densities (Nishida et al., 2013), may provide evidence that the ubiquitous FF aggregates in the Dunvegan prodelta mudstone may have a similar origin. The FF aggregates have a strongly random orientation, with little evidence that clay domains have been rotated bedding-parallel, despite substantial burial and compaction. This evidence suggests that the individual FF aggregates had considerable internal strength, despite a presumed initially high water content. The reason for this apparent strength is unknown, but bacterially secreted mucous may have been an important factor. Similarly, intraclastic aggregates show little evidence of compaction, suggesting that these grains also retained the inherent strength of their component FF aggregates. The strength of the intraclastic aggregates is supported by the observation that detrital mica flakes are, in places, deformed around these grains (Fig. 10).
The Dunvegan prodelta mudstone can be broadly divided, on the basis of field observation, into two main facies: a more distal ‘laminated mud’ facies, and a more proximal ‘heterolithic’ facies, including sand and mud. Given the present focus on mud transport and deposition, it is useful to take a microfacies approach to the mudstone component, in order to distinguish distinct depositional processes.
Observation of the 33 thin sections spanning the study interval showed that mudstone, in all samples, could be divided into four microfacies, distinguished on the basis of the style of stratification and the relative proportions of silt and clay. The four microfacies are arranged in various combinations to form millimetre-scale beds, each of which appears to represent a distinct sedimentary event. For each of the 11 samples, the mean thickness of beds, the relative proportion of each of the four microfacies, and the thickness of bioturbated layers as a percentage of total observed thickness, was determined and plotted against stratigraphic position (Fig. 14). Vertical changes in these sedimentary features are discussed below. No invertebrate macrofossils were observed in the studied section, but phosphatic fish debris is very abundant in the rippled sandstone forming the condensed facies of the FSU marker (Fig. 5).
Microfacies 1: Rippled or structureless, well-sorted coarse to medium-grained siltstone
Microfacies 1 forms lamina-sets ranging from 0·1 to 2·4 mm thick, with a mean of 0·97 mm (n = 107); the microfacies forms ca 15% of the total sampled thickness. The base of each lamina-set is almost invariably sharp, commonly with millimetre-scale scours that, in rare cases, resemble very small gutter casts (Fig. 15A to G). Silt laminae, particularly at higher stratigraphic levels, may be disrupted by sub-millimetre scale bioturbation. Silt lamina-sets typically have a symmetrical, lenticular form, and pinch-out laterally on the scale of the thin section (i.e. over <40 mm; Fig. 15A and C to F). Commonly, the silt is cross-laminated, and in most cases the cross-laminae dip in one direction (Fig. 15A, B and E). Occasionally however, opposed dip directions are present in a single lamina-set, and off-shooting and draping laminae are present (Fig. 15F). Less commonly, silt beds are structureless and, very rarely, planar laminated. The top of each silt layer almost invariably shows an abrupt, not graded, transition to more clay-rich sediment above. Quartz and chert are the dominant grains, but laminae rich in clay aggregates are usually also present (FF and intraclastic aggregates are not readily distinguishable in thin section and hence are not differentiated here; Fig. 6A). Organic matter, in the form of irregular, flattened black or reddish-brown flakes are rare. Within the sampled section, well-sorted siltstone tends to become more abundant upward (i.e. more proximal to the delta-front), forming <10% of the thickness of the samples near the base and >25% near the top (Fig. 14).
The sharp, scoured base, well-sorted texture, grain-supported fabric and well-developed cross-lamination suggest deposition from relatively energetic, turbulent flows that caused erosion of the underlying substrate prior to deposition. The absence of load casts below silt layers further suggests that deposition took place on a substrate that was cohesive, suggesting removal of an unconsolidated surficial layer prior to the deposition of silt. Rippled siltstone implies deposition from a turbulent flow having a relatively low concentration (probably <2% by volume) of suspended clay particles (cf. Baas & Best, 2002). The prevalence of symmetrical ripple forms, draping and off-shooting laminae are all typical of wave-formed ripples (De Raaf et al., 1977). The preponderance of form-discordant cross-laminae is also typical of wave-formed ripples, and is suggestive of a superimposed unidirectional flow (cf. facies PCS2 of Schieber, 1994b; and facies SM1 of Schieber, 1999); palaeoflow observations are discussed below.
Microfacies 2: Silt-streaked claystone
Microfacies 2 forms lamina-sets that range from 0·3 to 10·6 mm thick, with a mean of 1·81 mm (n = 115); the microfacies forms ca 31% of the total observed thickness. Silt-streaked claystone comprises sets of parallel, sub-millimetre thick, very subtly lenticular laminae of medium to fine-grained quartz silt alternating with clay aggregates (Figs 6D, 16A and 16B). A majority of lamina-sets become more clay-rich upward, although a minority either become siltier upward, or show a symmetrical coarsening–fining succession (Fig. 16F). Rarely, lamina-sets of silt and clay aggregates are inclined and show evidence of lateral accretion (Fig. 16C to E). Flattened, black to reddish-brown organic flakes, typically ca 50 to 200 μm, but occasionally up to ca 1 mm in length, are abundant and usually are concentrated into individual laminae.
Commonly, silt-streaked claystone overlies rippled siltstone of microfacies 1, almost invariably with a sharp but non-erosional contact. Silt-streaked claystone may also sharply and apparently erosively overlie silty claystone (microfacies 3) or claystone (microfacies 4). In a minority of cases, microfacies 2 has a gradational lower contact with silty claystone or claystone. Microfacies 2 forms between 25% and 55% of the samples, but shows little vertical trend in abundance (Fig. 14).
Very fine-scale inter-lamination of silt and clay has been attributed to a variety of processes, including fluctuation in current strength, interaction of waves with fluid mud, ripple migration, hydraulic sorting into laminae of like grain size and ‘shear sorting’ (e.g. Kuenen, 1966; Stow & Bowen, 1980; Kuehl et al., 1988; MacKay & Dalrymple, 2011). Recent work on the effect of suspended clay on flow structure (e.g. Baas & Best, 2002; Baas et al., 2009, 2011; MacKay & Dalrymple, 2011) has shown that, with increasing concentration of suspended clay, flow structure changes gradually from turbulent and vertically mixed, through transitional states, to a quasi-laminar ‘plug’ flow. In the latter, a dense, near-bed layer of fluid mud, having a relatively high concentration of suspended clay (in the order of 10% by weight), is sufficiently cohesive to suppress turbulence: Under these conditions, ripples do not form.
Because clay mineral matter is present in the studied samples in the form of cohesive, silt-sized aggregates, it seems unlikely that the ‘shear sorting’ mechanism of Stow & Bowen (1980) was responsible for the lamination because this mechanism involves the break-up, within the boundary layer of loosely bound flocs. A more plausible explanation of silt-clay lamination was suggested by Baas & Best (2002), in which Kelvin–Helmholtz instabilities develop at the boundary between a basal zone of high shear and an overlying, faster-flowing clay-rich ‘plug’ layer lacking turbulence (i.e. their transitional plug flow). Successive vortices at this interface would result in quasi-periodic pulses of faster and slower flow near the bed, leading to alternating deposition of silt and clay laminae. It has also been suggested that surface wind waves might produce near-bed shear stress fluctuations when superimposed on a unidirectional flow, and hence lead to alternate entrainment and deposition of silt and clay particles (e.g. Kuehl et al., 1988; Ichaso & Dalrymple, 2009; MacKay & Dalrymple, 2011).
The observation of rare, wavy, silt and clay lamina (Fig. 16C and D) suggests that, at times, turbulence was not suppressed fully, and wave action was able to mould the bed into subtle ripple forms (cf. Schieber & Yawar, 2009). In most cases however, parallel, albeit subtly lenticular lamination shows that the bed remained planar, probably due to the suppression of turbulence by the high viscosity of a layer of fluid mud immediately above the bed (Wolanski et al., 1992; Baas & Best, 2002; Traykovski et al., 2007). The organic fragments in this facies do not resemble pelagic marine algal cells (e.g. Macquaker & Gawthorpe, 1993) and are interpreted as terrestrially sourced phytodetrital material.
Microfacies 3: Structureless silty claystone
Microfacies 3 forms layers that range in thickness from 0·2 to 7·4 mm with a mean of 2·01 mm (n = 141); the microfacies forms ca 42% of the total observed thickness. Silty claystone typically lacks stratification, and is characterized by fine-grained quartz silt grains distributed randomly through a clay-rich matrix consisting of a high proportion of intraclastic clay aggregates (Figs 6C, 6D and 16A to E). In a minority of examples however, a very faint sub-millimetre scale inter-stratification of silt and clay is discernible. Bioturbation is only evident where silt is mixed from adjacent layers. Black to reddish-brown organic flakes, 50 to 200 μm in length, typically are dispersed through the silty claystone. Microfacies 3 generally has a sharp to rapidly gradational contact with underlying silt-streaked claystone of microfacies 2 (Fig. 16A and C to E). Structureless silty claystone forms between 15% and 60% of the total thickness of samples, and the proportion of this microfacies tends to diminish upward, although reversals of this trend appear to take place at base and top of the section (Fig. 14).
The lack of stratification and dispersal of fine-grained siliceous silt through a matrix of clay aggregates suggest that this microfacies was deposited from a fluid mud following deceleration of flow to the point where internal cohesion of particles led to gelling of the fluid. The matrix strength of the fluid mud was apparently sufficient to support suspended fine silt grains (cf. Rine & Ginsberg, 1985; Ichaso & Dalrymple, 2009; Baas et al., 2011; MacKay & Dalrymple, 2011; Talling et al., 2012). Even where lithological contrast is provided by adjacent facies, bioturbation is rare, and it is concluded that most of the silty clay microfacies suffered little or no bioturbation. The scarcity of infauna supports the interpretation as a rapidly deposited fluid mud (Bhattacharya & MacEachern, 2009). The tendency for the proportion of silty claystone to diminish up-section suggests that fluid mud was less likely to settle, or be preserved, in shallower water where wave action would have been stronger (Fig. 14).
Microfacies 4: Clay-rich mudstone
Microfacies 4 forms layers that range in thickness from 0·2 to 4·9 mm, with a mean of 1·17 mm (n = 61); the microfacies forms ca 11% of the total observed thickness. Clay-rich mudstone contains few dispersed grains of fine siliceous silt, few or no organic fragments, and has a homogeneous appearance, lacking stratification (Fig. 17A and B). Microscopically, clay-rich mudstone has a blotchy, inhomogeneous appearance and is composed of FF and intraclastic aggregates. Clay-rich mudstone almost invariably gradationally overlies silty claystone of microfacies 3 and, in turn, is sharply overlain by either siltstone or silt-streaked claystone of microfacies 1 or 2. Microfacies 4 diminishes in abundance upward, from ca 20% near the base of the section to ca 2% near the top (Fig. 14).
Clay-rich mudstone resembles silty claystone in its lack of stratification, and may therefore also represent settling of fluid mud. In this case, at least some of the fluid mud was composed of fine silt-size aggregates that probably were dispersed through a matrix of less consolidated clay flocs. Because clay-rich mudstone usually gradationally overlies silty claystone, it is possible that this finest-grained microfacies represents the final settling of a layer of fluid mud from suspension, from which silt grains had already been deposited. A second possibility is that the clay-rich mudstone represents post-storm settling from buoyant plumes of muddy water issuing from river mouths. However, the offshore increase in the abundance of clay-rich mudstone suggests that this possibility is less likely, given that settling of aggregates, coupled with Coriolis veering of plumes would concentrate clay inshore, rather than offshore. A third possibility is that this facies represents settling of clays from a dilute suspension that formed as a result of storm wave action that was sufficiently intense that it disrupted a bottom-hugging fluid mud and dispersed the clays upward into the water column (cf. Fan et al., 2004). The clay-rich mudstone facies is volumetrically the least important in the studied section, and becomes less abundant up-section. This stratigraphic distribution suggests either preferential deposition in deeper water, and/or preferential reworking and remobilization in shallow water.
Bioturbation is scarce to absent in samples 1 to 6 from the laminated siltstone–claystone portion of the section [Bioturbation Index (BI) 0 to 1]. As pointed out by Bhattacharya & MacEachern (2009), sediment with fine-scale lithological heterogeneity, as in the studied samples, would undoubtedly reveal bioturbation, if it were present. With the appearance of sandstone beds above sample 6, bioturbation, albeit at a very small scale, increases rapidly in abundance, affecting up to ca 40% of the thin-section thickness in sample 8 (BI = 1 to 2; Figs 5 and 14). Discrete burrows are not recognizable, but bioturbation can be recognized where fine lamination is abruptly disturbed and homogenized (Fig. 18A, C and F). Alternatively, silt layers may be dispersed downward into underlying silty mud (Fig. 18A to C). In other instances, clay may be dispersed downward into the top of a siltstone layer (Fig. 18C and D). Isolated pods of silt in silty mud may be examples of ‘mantle and swirl’ structures (Lobza & Schieber, 1999; Fig. 18A). In overall terms, however, bioturbation is scarce, of small scale and diffuse character. These characteristics are typical of fluid mud substrates through which organisms swim, rather than burrow (Lobza & Schieber, 1999; MacEachern et al., 2005; Bhattacharya & MacEachern, 2009) and which, in general, provide an inhospitable environment for benthic macrofauna (e.g. Aller & Aller, 1986).
The mudrocks preserve a wealth of micro-stratigraphic information when observed at thin-section scale, as emphasized in previous studies (e.g. Davis & Byers, 1993; Schieber, 1994b, 1998, 1999; Macquaker & Gawthorpe, 1993; Macquaker et al., 2007; Aplin & Macquaker, 2011; Ghadeer & Macquaker, 2011). The vertical successions of sedimentary microfacies in each sample are portrayed as graphic logs (Figs 19 to 22). Individual thin beds are recognized primarily on the basis of sharp facies contacts. The stratigraphic succession in the lower, sand-free part of the section (Fig. 5; samples 1, 2, 4, 5 and 6), is represented for each sample by three graphic logs drafted from each of the three oriented thin sections (Figs 19 to 21). Samples 3 and 7 to 11 are, for brevity, represented by single graphic logs (Figs 20 and 22); nevertheless lateral variability is just as complex as that seen in the other samples.
The most striking feature of all the samples is the extreme lithological heterogeneity, in both the vertical and lateral dimensions. Although stratigraphic sections were measured only ca 20 mm apart in each sample (Figs 19 to 21), it proved remarkably difficult to correlate all but the most robust beds from one thin section to another. The abundance of scoured surfaces and the close vertical juxtaposition of different microfacies show that the depositional environment was very dynamic; most sedimentation events were preceded by an episode of erosion (cf. Schieber, 1994a,b). Equally remarkable is the extreme lateral variability of microfacies, on a scale of millimetres to centimetres. This variability is due to both lateral change in the depositional thickness of individual microfacies and to millimetre-scale erosion on the base of beds.
Recent flume experiments involving erosion of mud substrates (Schieber et al., 2010), showed that mud with as much as 85% water developed steep and complex, ≤10 mm-scale erosional relief comparable in lateral and vertical dimensions (following compaction) to that observed in the Dunvegan prodelta mudstones. Lateral inhomogeneity on a comparable vertical and lateral scale to the Dunvegan samples has been revealed by X-radiographs of cores through shallow water, storm-influenced mudstones from the modern Amazon subaqueous delta (Kuehl et al., 1986) and the Atchafalaya shelf (Allison et al., 2000; Neill & Allison, 2005).
Most siltstone layers are lenticular with a symmetrical external form, and many show unidirectional, form-discordant cross-lamination typical of combined-flow ripples (Fig. 15). Digital tracings of rippled siltstone layers in all thin sections were assembled into ‘pseudo-perspective’ panel diagrams for each sample. By visually integrating the apparent dip of cross-laminae in 54 individual siltstone layers, as seen in each of the three planes, it was possible to determine the most likely true sediment transport direction (Fig. 23A to C).
Results from samples 1 to 4, 5 to 8 and 9 to 11 were pooled in three groups and plotted with respect to stratigraphic position (Fig. 5). This grouping was used to determine whether there was any systematic variation in flow direction with stratigraphic position, and hence with distance from the palaeoshoreline. All three groups show a strong preferred transport mode towards the south-east, with a secondary mode to the north-west (Fig. 5).
Beds of very fine sandstone, typically 10 to 30 mm thick, are common between 22 and 49·5 m in the studied section (Fig. 5). Sandstone beds are sharp-based and ubiquitously wave-rippled; a majority of beds show combined-flow ripples characterized by symmetrically rippled tops and internal, form-discordant, unidirectional cross-lamination. A minority of beds have symmetrically rippled tops but show symmetrical internal cross-lamination. The trend of wave ripple crests shows a strong north-west to south-east mode, whereas combined-flow ripple cross-lamination is directed between south-west and south-east (Fig. 5).
Microfacies Successions, Palaeoflow Patterns and Sediment Transport Processes
In the studied samples, beds can be divided into three types of microfacies succession. Normally graded beds form ca 73% of all beds observed; reverse-graded beds form ca 14%; and reverse-to-normally graded beds form ca 13%. For interpretive purposes, beds are grouped into two types: type A – normally graded (73% of beds), and type B – reverse and reverse-to-normally graded beds (27% of beds). The variability of these two categories is summarized in Fig. 24. The interpretation of the depositional processes responsible for these two bed types is in part dependent on palaeoflow and palaeogeographic data.
Implications of palaeoflow data
The palaeogeographic framework available for the Dunvegan Formation provides a basis for interpreting palaeoflow patterns in terms of sediment transport processes. The wave-influenced silt ripples (microfacies 1), regardless of whether they formed ca 90 to 120 km offshore (samples 1 to 4) or within ca 10 to 50 km of the delta-front (samples 9 to 11), record preferential migration of silt directly down the prodelta slope (Fig. 25A and B). These palaeoflow data strongly suggest that silt and clay aggregates were transported by a combined-flow process that included both oscillatory wave motion, as well as a downslope component of flow. It is possible that the offshore-directed flow could be attributed to storm-driven downwelling, although such flows would be expected to have undergone Coriolis deflection to coast-parallel within tens of kilometres of shore (Swift et al., 1986). Alternatively, various types of gravity-driven flow are known to operate in prodeltaic and shelf settings, including flood-generated hyperpycnal flows and wave-enhanced sediment gravity flows (e.g. Ogston et al., 2000; Mulder et al., 2003; Fan et al., 2004; Bhattacharya & MacEachern, 2009; Macquaker et al., 2010). The details of microfacies successions within beds may help to distinguish these processes.
Type A beds: Wave-enhanced sediment gravity flows of mud
Type A beds encompass several types of normally graded bed, distinguished by subtle variation in the relative proportions of microfacies. The base of each bed is always sharp, sometimes irregular, never shows load casts and is cut in muddier sediment. Typically, beds comprise three divisions; in ascending order, well-sorted siltstone (microfacies 1), sharply overlain by silt-streaked claystone (microfacies 2), overlain by unstratified silty claystone (microfacies 3) and/or clay-rich mudstone (microfacies 4; Fig. 24A to G).
A sharp basal surface overlain by wave-rippled silt implies that wave-induced turbulence was responsible for initial scouring of the sea bed, leading to winnowing, re-suspension of weakly cohesive mud, and probably introduction of new silt. The sharp upward transition from rippled siltstone to silt-streaked claystone suggests a distinct change in flow character, from a turbulent or turbulence-enhanced state to transitional or quasi-laminar plug flow in which turbulence was suppressed to a greater or lesser extent (e.g. Baas et al., 2011). Such changes in flow state are commensurate with deposition from wave-enhanced sediment gravity flows in which the proportion of suspended clay is initially low, but over time increases to the point where flow structure is modified (Ogston et al., 2000; Fan et al., 2004; Friedrichs & Wright, 2004; Macquaker et al., 2010; Baas et al., 2011). During the passage of a storm across a shelf, waves would initially erode the sea bed, winnowing and concentrating silt from underlying silty mud. Wave action would also cause re-suspension of flocculated, river flood-sourced mud in the nearshore region (e.g. Fan et al., 2004), potentially forming a thin, near-bed fluid mud (typically having >10 g l−1 suspended sediment); this fluid mud would be strongly confined below the wave boundary layer (i.e. the lutocline; Traykovski et al., 2007). Fluid mud can become sufficiently dense that it will undergo downslope flow in response to gravity. Macquaker et al. (2010) suggested that the sedimentary record of this process will consist of a distinct ‘triplet’ bed that comprises a lower, sharp-based rippled sand (division ‘A’), a middle parallel laminated division of silt and clay (division ‘B’) and an upper structureless mud (division ‘C’). Rippled sand records initial wave-generated turbulent flow, followed by the formation and subsequent downslope flow of fluid mud, forming division B, followed by cessation of wave action and vertical deposition of structureless mud, division C (Macquaker et al., 2010; Ghadeer & Macquaker, 2011).
In the Dunvegan samples, the preponderance of south-east directed cross-lamination in silt layers indicates that these ripples, although experiencing oscillatory wave effects, also underwent preferential downslope migration; this is attributed most reasonably to the flow of bottom-hugging fluid mud (cf. Ogston et al., 2000; Macquaker et al., 2010; Fig. 25). Wave-rippled siltstone (microfacies 1) records a low suspended sediment concentration (probably <10 g l−1) and turbulent combined flow. The incursion of a dense fluid mud from upslope resulted in a change to transitional plug flow or even quasi-laminar plug flow, and the suppression of both turbulence and of ripple formation (Baas et al., 2011). Subtle velocity fluctuations in the near-bed region below the fluid mud are postulated to have resulted in a planar depositional surface on which was deposited silt-streaked clay (microfacies 2; cf. Wolanski et al., 1992; Teisson et al., 1993; Traykovski et al., 2007). The upward gradation from silt-streaked claystone to structureless silty claystone (microfacies 3) is analogous to the transition from divisions B to C of Macquaker et al. (2010). This facies change is interpreted to record the cessation of both oscillatory and unidirectional flow at the sea bed as the storm waned, wave base-lifted and wave-driven production of fluid mud ceased. Without energy input from waves, the suspended fluid mud collapsed vertically, depositing an unstratified mud, the cohesive properties of which would have inhibited the settling of suspended fine silt grains and clay aggregates (Mehta, 1991; cf. Ichaso & Dalrymple, 2009).
The inter-stratification of centimetre-scale sandy storm beds and millimetre-scale silt-clay storm beds in the upper, more proximal part of the studied section (Fig. 5, samples 7 to 11), shows that muddy, wave-enhanced sediment gravity flows operated across most of the prodelta. Based upon analogy with modern storm beds (Aigner & Reineck, 1982), it is inferred that, for any given palaeogeographic location, thin muddy beds record smaller storms, whereas thicker sandy beds record larger storms. Regardless of palaeogeographic position on the prodelta, however, silt ripples in all samples show a strong south-east directed mode that indicates that wave-enhanced sediment gravity flows experienced negligible Coriolis deflection as they travelled across the prodelta (Fig. 25). This finding is consistent with calculations (Traykovski et al., 2000), that showed, for a mid-latitude setting (such as the Dunvegan Formation), that the force exerted by gravity would greatly exceed the Coriolis effect, and hence wave-enhanced sediment gravity flows would have travelled directly down the bathymetric slope.
Although the normally graded, type A beds in the Dunvegan Formation superficially resemble distal muddy turbidites, they lack a number of key features, including an absence of load structures, a basal layer of planar-laminated silt, climbing and/or fading current ripples, and a gradational boundary between rippled siltstone and silt-streaked claystone; features that Schieber (1999) considered diagnostic of true turbidites. Therefore, type A beds are interpreted as the deposits of wave-enhanced sediment gravity flows rather than classical turbidites. This interpretation is supported by the fact that Dunvegan prodelta mudstone shows a relatively distinct lap-out onto the Fish Scales Upper (FSU) marker, which implies that wave-enhanced gravity flows stopped rather abruptly as they dropped below wave base. Had the flows been able to achieve true auto-suspension, it might be expected that they would have travelled further offshore, and generated a more extended toeset geometry.
The small-scale variability of microfacies successions in type A beds (Fig. 24A to G) is interpreted to record the considerable natural variability of the wind direction and duration of storms, the thermal and salinity structure of shelf waters, and the availability of river-supplied sediment, all of which would have affected the effectiveness of sediment re-suspension, the thickness and density of the resulting fluid mud layers, and the distance that they would have travelled downslope (Allison et al., 2000; Ogston et al., 2000; Fan et al., 2004; Neill & Allison, 2005; Traykovski et al., 2007).
Although downslope flow predominates, a minority of silt ripples indicate upslope bedform migration, particularly for samples 5 to 8 (Fig. 25A). It is possible that these bedforms record landward-directed asymmetrical flow beneath shoaling waves (e.g. De Raaf et al., 1977). If so, they imply waves approaching from the south-east, which is perpendicular to the direction determined for sand beds in the upper part of the section (Fig. 25B); unfortunately, it is not possible to determine crestal trends for silt ripples in thin section but, given the probability of a complex, combined-flow current regime, it is likely that they would have been strongly three-dimensional. The origin of these upslope directed ripples remains enigmatic.
Type B beds: Muddy gravity flows of indeterminate origin
Type B beds commonly lack a sharp basal contact and grade upward from clay-rich mudstone or silty claystone to silt-streaked claystone that may be capped by one or more thin, sometimes lenticular, laminae of well-sorted siltstone (Figs 16F and 24H and I). In some instances, silt-streaked claystone sharply overlies structureless clay-rich mudstone and shows an upward-thickening of siltstone laminae, capped by a thicker, apparently winnowed siltstone lamina (Fig. 24J). Instead of having a sharp top, reverse-graded beds may have a more symmetrical profile, with microfacies 2 grading upward to microfacies 3 and/or microfacies 4 (Fig. 24K).
Inversely graded beds suggest a gradual increase in flow strength, and a normally graded cap suggests a gradually waning flow. These types of bed show little or no evidence of turbulent flow or erosion at the base; instead the transition from microfacies 3 to microfacies 2 suggests gradual onset of flow, such as might be expected beneath a slowly accelerating muddy density current. The upward increase in the frequency and/or thickness of silt laminae in microfacies 2 suggests increasing flow velocity and more effective transport (and shear segregation) of silt grains. The layer of well-sorted silt at the top of some beds may show very subtle lenticularity, suggestive of very small wave ripples, but cross-lamination is not developed (Fig. 24J). Such silt layers might be interpreted as a terminal phase of flow that was sufficiently vigorous to rework and winnow the underlying silt and clay or, alternatively, a completely separate event characterized by reworking, rather than the introduction of new sediment (comparable to the ‘low concentration’ depositional regime of Fan et al., 2004).
Inversely graded, and inverse to normally graded beds of sand, silt and clay have been recognized as characteristic of hyperpycnites (Mulder et al., 2003), deposited from sustained density flows issuing from river mouths. Variations in the discharge of the river are recorded by subtle coarsening and fining patterns, and by minor intra-bed erosion surfaces. Soyinka & Slatt (2008) predicted that muddy hyperpycnites must be just as abundant as their sandy equivalents, but noted that their recognition would be hampered by fine grain size, lack of obvious sedimentary structures and greater susceptibility to surface weathering. Soyinka & Slatt (2008) made very detailed observations of sedimentary lamination and grain-size variability in millimetre to centimetre-bedded siltstones and mudstones in the Upper Cretaceous Lewis Shale of Wyoming. These authors recognized both normally graded and reverse-to-normally graded thin beds of silt and mud that were interpreted as flood-sourced hyperpycnites. Importantly, no evidence of wave ripples was observed in these ‘deep water’ mudstones.
Similarly, guided by the hyperpycnite model, Bhattacharya & MacEachern (2009) interpreted centimetre-scale normally graded sand-mud beds, as well as reverse-graded and reverse-to-normally graded beds of sand and mud in proximal Dunvegan prodelta deposits in terms of flood-generated hyperpycnal flows that may have issued from deltaic distributaries. Unlike the Lewis Shale, however, the sandstone beds illustrated by Bhattacharya & MacEachern (2009) are pervasively wave-rippled, suggesting that wave effects were important in generating and sustaining the flows.
Given the deltaic setting of Dunvegan allomember G, it is possible that type B beds are also the deposits of waxing and waning, river-sourced hyperpycnal flows. However, the prodelta gradient is estimated at ca 0·05° (see 'Discussion' below), which is much lower than the 0.7° necessary to sustain a river-fed hyperpycnal flow (Friedrichs & Scully, 2007). The absence of wave-rippled siltstone layers in type B beds suggests that wave action was not an important influence, at least at the site of deposition, yet the abundance of silt-streaked claystone is indicative of effective boundary-layer shearing, as would be expected beneath a flowing fluid mud. The simplest explanation may be that type B beds record deposition from flows of fluid mud that were generated by waves in shallow water, but subsequently underwent deposition close to effective wave base. At this depth, waves were insufficiently energetic to erode the sea bed and hence neither a scoured surface nor ripples were formed. Wave influence at the bed might have been further diminished because the density of the fluid mud suppressed the propagation of wave energy down through the lutocline, and hence to the bed (Traykovski et al., 2000). Upward-coarsening lamina-sets could be interpreted as the downslope record of the gradual onset of wave-supported hyperpycnal flow as winds strengthened, whereas upward-fining could record the gradual waning of the storm. Type B beds might therefore be interpreted as base-absent type A beds, deposited by relatively weak storms or below effective wave base.
Sand transport processes
The palaeoflow pattern for sand is radically different from that for silt. Wave ripple crests in sandstone beds that were deposited an inferred ca 10 km to ca 80 km offshore, trend shore normal (Fig. 25B). All but one of the observed sandstone beds contained combined-flow ripples that indicate ripple migration obliquely across the prodelta slope directed between south-west and south-east (Fig. 25B). This pattern of sand transport may be explicable in terms of storm winds that blew across the delta front from the north-east. Such winds would be expected for mid-latitude storms tracking west to east across the Interior Seaway, as modelled by Slingerland & Keen (1999). As storm systems began their passage across the Seaway, offshore winds blowing from the south-west, off of the leading edge of the storm, would have had little opportunity to build large waves because of limited fetch. Later, however, winds blowing from the north-east, off of the trailing edge of the storm, would travel unimpeded across hundreds of kilometres of water, building large waves. Storm winds from the north-east would entrain surface water that, under Coriolis deflection, would flow to the right of the wind, causing coastal set-up along the Dunvegan shoreline to the north-west. The resulting downwelling, geostrophic and combined flow would drive bottom sediment transport alongshore to obliquely offshore; i.e. to the south-west and south. This is the pattern observed (Figs 25B and 26).
Need for storms and river floods
There is evidence that both type A and type B beds record wave-enhanced sediment gravity flows, albeit of different intensity. Storm waves clearly are essential to the generation and maintenance of these flows. On the Eel shelf (California), the most vigorous density flows of fluid mud took place during storms that occurred during, or soon after, major river floods that supplied large volumes of flocculated mud to the inner shelf (Ogston et al., 2000). Conversely, on the Po shelf (Italy), storm-generated density flows preceded river floods, showing that wave-enhanced sediment gravity flows could be driven simply by wave re-suspension of previously deposited mud (Fan et al., 2004; Traykovski et al., 2007).
Deposition, consolidation and re-suspension of mud
River-supplied mud must initially have accumulated on the sea bed in the form of weakly consolidated organo-minerallic aggregates bound by extracellular polysaccharides, van der Waals forces and electrochemical attraction between charged particles (Kranck & Milligan, 1980; McCave, 1984). When freshly deposited, these aggregates form a low density ‘fluff’ layer. Gradually, on a time scale thought to be from months to a few years, the fluff layer consolidates as the edge to face fabric of the aggregates breaks down and particles adopt a more closely packed, face to face fabric because repellant forces on grain surfaces are diminished by polysaccharide coatings. Closer packing is also promoted by reduction in the thickness of the charged double layer by Na, and particularly Ca, cations in the pore water, which allows Van der Waals forces to become more effective (e.g. Grabowski et al., 2011). Friedrichs & Scully (2007) suggested that the early consolidation of mud, on a time scale of only weeks, may have been responsible for inhibiting storm wave re-suspension of river-supplied mud on the Po shelf.
Storm waves readily re-suspend the ‘fluff’ layer to form fluid mud (e.g. Winterwerp & van Kesteren, 2004; Grabowski et al., 2011), although the microstructure of clay particles in fluid mud appears to be relatively poorly known. Recently, however, (Nishida et al., 2013) have shown experimentally that relatively dense, face-face aggregates of clay plates form readily in saline fluid muds containing between 10 g l−1 and 30 g l−1 suspended clay. The abundance of face-face aggregates in the Dunvegan mudstones (Figs 8 to 13), suggests that the as-yet poorly understood aggregation process, documented by Nishida et al. (2013), also operated in fluid muds generated across the Dunvegan prodelta. The abundance of intraclastic aggregates in the Dunvegan mudstones attests to widespread erosional reworking of partially consolidated mud, most probably attributable to storm waves scouring the shallower portion of the prodelta (cf. Fan et al., 2004). Wave reworking may have been most effective during ‘dry’ storms that were not accompanied by heavy rain. At such times, no new mud was introduced to the sea floor by river floods and, in consequence, wave-scour of previously deposited mud would have been more effective (cf. Bentley et al., 2006).
Sediment partitioning on the prodelta
The admittedly modest data set presented in this study suggests that sand and mud followed divergent transport pathways across the prodelta (Fig. 26). In sandy heterolithic strata, palaeoflow data show that combined and geostrophic flows moved fine-grained and very fine-grained sand obliquely offshore towards the south-west and south (Fig. 25B). Interpolation between stratigraphy and palaeogeography (Figs 4, 5 and 25), suggests that effective wave base for sand intersected the prodelta surface ca 70 to 80 km offshore (i.e. between samples 6 and 7), at an inferred palaeo water depth of ca 40 m (based on the geometry and decompaction procedure in Plint et al., 2009). The lack of sand in more distal (i.e. stratigraphically lower) locations indicates that sand was not transported further offshore by density flows, presumably because wave-induced turbulence was insufficient to maintain sand in suspension.
In muddy strata, combined-flow silt ripples indicate sediment transport directly down the prodelta slope, effected by gravity-driven fluid mud flows. The divergent transport pathways indicated by palaeoflow data from the Dunvegan Formation support the interpretation of Harazim et al. (2011), who observed divergent palaeoflow directions between interbedded, storm-deposited layers of sandstone and mudstone of Ordovician age. Although these shelf deposits lacked a well-defined palaeogeographic framework, it was inferred that storm waves and geostrophic flows drove sand alongshore to obliquely offshore during storm peak, but as the storm waned, wave-supported gravity flows transported fluid mud offshore (downslope).
The silt ripples in the Dunvegan mudstones imply that effective wave base for mud was, not surprisingly, significantly deeper than the ca 40 m wave base inferred for sand. Using the decompaction factor of 1·4 employed in Plint et al. (2009), the 18 m of sand-free mudstone in the lower part of the Smoky River section (Fig. 5) might represent an additional ca 25 to 30 m of water below ‘sand wave base’. Thus, ‘mud wave base’ for the Dunvegan prodelta might be estimated at ca 70 m water depth. Based on this bathymetric estimate, the height of the allomember G prodelta, from beach to lap-out limit, measured at maximum regression, might have been ca 70 to 80 m, over a horizontal distance of ca 80 km (Fig. 25B). This estimation indicates an average prodelta gradient of ca 1 : 1000 or 0·05°.
With this new appreciation of prodelta mud transport processes, it is reasonable to conclude that the prominent downlap pattern, so distinctive of the Dunvegan Formation (Figs 4 and 25), is primarily an indication of the seaward limit of mud transport by wave-enhanced sediment gravity flows. Beyond this point, fluid mud settled because the water was too deep for wave re-suspension to be effective, and also because of the gradual reduction in gradient towards the toe of the clinoform. The fact that fine to medium-grained silt is present as far seaward as the toe of the prodelta clinoform suggests that, unlike sand and coarse-grained silt, the internal yield strength of wave-supported density flows was sufficient to prevent medium-grained and fine-grained siliceous silt from settling out; alternatively, clays were packaged as aggregates that had similar hydraulic properties to silt grains and hence were not hydraulically sorted (Kuehl et al., 1988; Mehta, 1991; Neill & Allison, 2005).
Controls on benthic fauna
The prodelta setting of the studied succession may have been subject to a variety of stresses that could have included low bottom-water oxygen content, rapid sediment accumulation rate, frequent sedimentation events, emplacement of fluid mud, increased water turbidity, reduced salinity and low organic matter concentration (e.g. MacEachern et al., 2005; Bentley et al., 2006; Wheatcroft et al., 2007; MacEachern & Bann, 2008). The extreme scarcity of recognizable bioturbation in the lower part of the studied succession (samples 1 to 6) suggests that benthic fauna were strongly inhibited in the more distal region of the prodelta (ca 90 to 120 km offshore). Closer to shore, the increase in the intensity of bioturbation coincides with the upward appearance of wave-rippled sandstone interbedded with mudstone (samples 7 to 11), and suggests ameliorating environmental conditions. In the more distal prodelta, the most likely stresses may have been frequent, storm-related deposition of fluid mud, resulting in soupground conditions, and possibly diminished bottom-water oxygen content. The lack of synaeresis cracks in any of the samples suggests that reduced salinity was not an inhibitory factor (e.g. MacEachern & Bann, 2008). There is little evidence that the upward increase in the intensity of bioturbation in samples 7 to 11, representative of environments ca 10 to 90 km offshore, can be attributed to a diminished frequency of sedimentation events, fewer pulses of reduced salinity or a more stable substrate. It is possible that the increased bioturbation reflects lower turbidity, and perhaps greater nutrient availability in the form of benthic cyanobacteria, algae and diatoms that, in this mid-latitude deltaic setting, might have been able to photosynthesize to a water depth of ca 60 m (cf. Wheatcroft et al., 2007). There is also the possibility that a more abundant benthic fauna was favoured by shallower water and more frequent wave agitation that resulted in better oxygenation.
In a study of delta front and inner prodelta deposits of Dunvegan allomembers D and E, that were deposited ca <20 km from the contemporaneous shoreline, Bhattacharya & MacEachern (2009) observed low to very low levels of bioturbation (Bioturbation Index of 0 to 2). In this nearshore setting, a high sedimentation rate from hyperpycnal flows, fluid mud substrates and reduced salinity were all inferred to have inhibited benthic fauna. Further seaward, bioturbation intensity increased as environmental stresses diminished, as also seen on modern river-influenced muddy shelves (e.g. Allison et al., 2000).
Collectively, the observations of Bhattacharya & MacEachern (2009) for allomember E, and those presented here for allomember G, suggest that a synthetic transect across the entire Dunvegan prodelta would show low levels of bioturbation close to shore (ca 0 to 20 km), due to high sedimentation rate, high turbidity, soupground substrate and possibly salinity stress. Further seaward, a lower sedimentation rate, lower turbidity, more normal salinity and possibly better oxygenation and more abundant nutrients afforded by microphytobenthos, promoted higher levels of benthic activity that extended to the seaward limit of sand transport by storm waves, ca 80 km offshore in ca 40 m of water. On the distal prodelta, in muddy sediment, bioturbation decreased to very low levels. In this offshore setting, it seems possible that a combination of soupground substrate, frequent storm emplacement of fluid mud, low nutrient availability and reduced oxygenation collectively inhibited benthic fauna (cf. Macquaker & Gawthorpe, 1993; Schieber, 1994b; Harazim et al., 2011; Fig. 26).
Prodelta sediments of the Cretaceous Dunvegan Formation, sampled in outcrop, were correlated with a subsurface allostratigraphic framework. Correlation with the regional framework showed that the studied succession was deposited between ca 10 km and 120 km from shore. Regional facies and isopach mapping showed that the shoreline trended broadly north-east to south-west, and the prodelta sloped to the south-east.
Prodelta sediments of Dunvegan allomember G form progradational, sandier-upward successions. Within ca 80 km of shore, centimetre-scale very fine to fine sandstone beds are interbedded with mud. Sandstones show abundant combined-flow ripples; wave ripple crests trend north-west to south-east (shore-normal), and internal cross-lamination shows migration towards the south-west to south-east (i.e. obliquely offshore). More than ca 80 km offshore, sand is absent and silt and clay form millimetre-scale beds. Millimetre-scale combined-flow silt ripples occur throughout the mudstone; oriented thin sections reveal a strong preferred migration direction to the south-east, down the prodelta slope.
Clay mineral matter is organized at three scales. Clay domains comprise a few closely bound clay plates ca 1 μm in diameter; domains are grouped into roughly equant ‘face-face’ (FF) aggregates typically 2 to 5 μm in diameter. Face-face aggregates are randomly oriented and show no evidence of compactional rotation parallel to the bedding. Intraclastic aggregates, composed of tightly packed FF aggregates are typically 5 to 20 μm in diameter, show little evidence of compaction, and are interpreted to have been eroded from consolidated mud substrates by storm waves. The lack of flattening of FF and intraclastic aggregates suggests that they had considerable strength, possibly in part attributable to binding by bacterial mucous; this observation requires further investigation and explanation.
Four mud microfacies are recognized. Well-sorted siltstone forms sharp-based, combined-flow ripples and was generated by storm waves with a superimposed density flow. Silt-streaked claystone comprises parallel sub-millimetre laminae of silt and clay aggregates that were sorted by fluctuating bed shear stress in the boundary layer beneath a dense, wave-supported fluid mud. Structureless silty claystone has fine silt grains floating in a matrix of clay aggregates and was deposited from the gelling and vertical settling of fluid mud without wave or current shear. Structureless clay-rich mudstone has little silt and may represent late-stage settling of fluid mud, or possibly settling from a dilute suspension resulting from dissipation of fluid mud by intense wave action.
Facies are organized into two bed types. Type A beds have scoured bases and are normally graded through microfacies 1, 2, 3 and/or 4; they are interpreted to record deposition from wave-enhanced sediment gravity flows. Initially, storm waves scoured and winnowed the sea floor forming ripples under turbulent conditions. Later in the storm, river-supplied mud on the inner prodelta was re-suspended in sufficient concentration that downslope flow of fluid mud was initiated, with the development of plug-flow structure that suppressed turbulence. Type B beds are reverse-graded or reverse-to-normally graded and lack a sharp base; they appear to represent gradually accelerating and decelerating density flows. Flows may have originated as hyperpycnal flows from river mouths, but are more likely to be ‘base-absent’ wave-enhanced sediment gravity flows deposited by weaker storms below effective wave base.
Effective wave base for fine sand is estimated at ca 40 m water depth; this hydraulic boundary intersected the prodelta surface ca 80 km from shore. Effective wave base for mud is estimated at ca 70 m and intersected the sea bed ca 120 km offshore. The lap-out of prodelta mud approximates the point at which the sea floor became too deep for waves to maintain silt and clay aggregates in a fluid mud suspension.
The transport paths of sand and mud across the prodelta diverge markedly. Sand was transported shore-parallel to shore oblique by combined flows driven by storm winds from the north-east. Simultaneously, wave-supported fluid mud flowed directly down the prodelta slope to the south-east under the influence of gravity.
Small-scale bioturbation is present where sandstone beds are inter-stratified with mud, but bioturbation becomes very scarce in the lower part of the section where sand is absent. Benthic fauna probably were inhibited by a combination of frequent fluid mud flow events, soupground conditions, high turbidity, low organic matter concentration and possibly reduced bottom-water oxygen concentration.
I am grateful to the Natural Sciences and Engineering Research Council of Canada which has supported my research on the Dunvegan system for over 25 years. Field sampling was conducted with the assistance of Piotr Angiel, Robin Buckley, Meriem Grifi, Tessa Plint, Joel Shank and Kristyn Smith. I thank Steve Wood for his careful preparation of the thin and polished sections. All SEM imaging was undertaken at the UWO Zircon and Accessory Phase Laboratory, operated by Dr. Desmond Moser, with the technical assistance of Ivan Barker. Informal discussions with Joe Macquaker, Burns Cheadle, Fred Longstaffe, Naohisa Nishida, Piotr Angiel and Joel Shank helped shape and clarify my thinking about prodelta mud transport, but these gentlemen share no blame for any shortcomings in my interpretations. I am very grateful to Joe Macquaker, Jürgen Schieber and Esther Sumner for their constructive suggestions and penetrating questions (not all of which I have been able to answer!), and to Associate Editor Jaco Baas for additional helpful comments. The author declares no conflict of interest with regard to this manuscript.