Backward stacking of submarine channel–fan successions controlled by strike-slip faulting: The Izumi Group (Cretaceous), southwest Japan

Strike-slip faults and related basins are commonly observed at plate boundaries or active plate margins in a setting of convergent or oblique relative plate motion (Fitch, 1972; Jarrard, 1986; Lundberg and Reed, 1991). Such sedimentary basins generally develop in areas of crustal extension related to the geometry and kinematic history of the fault configuration (e.g., Mann, 2007; Nilsen and Sylvester, 1995). Sedimentary fill within many strike-slip basins is classically considered as a repetitive, basinwide, upward-coarsening succession that develops in response to tectonically induced basin migration and deepening, as reported for the Hornelen Basin, Norway, and the Ridge Basin, California (Nilsen and Sylvester, 1995). The Hornelen Basin is a Devonian pull-apart basin made up of axial deposits subdivided into 200 coarsening-upward sequences (Steel, 1976; Steel et al., 1977). The depositional facies shows a shallowing-upward succession, changing from lacustrine mudstone to conglomeratic braided-stream deposits. The repetitive sequences have been interpreted to result from discontinuous strike-slip movement (Steel et al., 1977). Similarly, the Ridge Basin is a fault-bend basin that was initially filled by marine mudstone, followed by shallower facies of alluvial conglomerates and sandstone (Crowell and Link, 1982; Crowell, 2003).

In contrast to these studies, recent reports have described fining-upward successions in strike-slip basins, reflecting progressive deepening of the basin (Kim et al., 2003; Lee and Chough, 1999). These reports ascribe the origin of the sequences to a temporal increase in tectonic subsidence due to dip-slip displacement rather than strike-slip offset along the bounding faults. Similar fining-up successions have been observed in extensional basins with dip-slip displacements, such as half-graben basins (e.g., Postma and Drinia, 1993; Zecchin, 2005).

The lithostratigraphic variations observed in fault-bounded basins have generally been explained in terms of tectonics as an external (allogenic) force (Blair and Bilodeau, 1988; Gordon and Heller, 1993; Heller et al., 1988), although other processes of glacioeustatic (e.g., Heckel, 1986) or climatic oscillations (e.g., Cecil, 1990; Zhang et al., 2001) could also control lithology. Stratigraphic variations in foreland basins have been modeled numerically based on interactions among tectonic uplift or subsidence, sediment flux, and gravel content of the basinfill sediment (e.g., Clevis et al., 2004; Flemings and Jordan, 1989; Marr et al., 2000; Paola et al., 1992; Sinclair et al., 1991). Given that strike-slip basins are also fault-related, it is possible that the basin-filling processes are syntectonic, affected by external factors such as fault activity and sediment supply; however, few studies have quantitatively investigated the interactions between tectonics and sediment dynamics in strike-slip basins. The causes of variations and cyclicity in lithology within strike-slip basins therefore remain poorly understood.

The present study focuses on the Izumi Group, southwest Japan, which is considered to be a strike-slip basin related to rapid oblique subduction (~24 cm yr⁻¹; Engebretson et al., 1985) of the Izanagi plate during the Late Cretaceous (Maruyama et al., 1997; Taira et al., 1983). The axial deposits of the Izumi Group are considered to represent repeated stacks of submarine turbidites and mudstones, with no evidence of shallowing-upward successions within the basin. The objectives of this paper are to (1) describe and interpret the sedimentological characteristics of cyclic lithostratigraphy within the strike-slip basin, and (2) discuss the cause of the stratigraphic cyclicity by using numerical simulations performed with the aim of reproducing the observed repeated coarsening- and fining-upward successions together with basin formation via strike-slip faulting. This multi-disciplinary approach was attempted in order to bridge the gap between qualitative sedimentary facies analysis and quantitative tectonic modeling. The results enable a better understanding of the spectrum of different stratigraphic architecture possible in strike-slip basins for which development is controlled by bounding faults.

The Izumi Group is dominated by Upper Cretaceous subaqueous deposits within a 300-km-long, 10–20-km-wide zone along the Median Tectonic Line extending from Shikoku Island to Kinki district (Figs. 1 and 2). The sediments of the Izumi Group show the following characteristics of deposition in a strike-slip basin (Ichikawa et al., 1981; Miyata, 1990; Tanaka, 1989): (1) migration of depocenters away from the direction of strike-slip motion; (2) deep basin in relation to its width; (3) axial infilling by the dominant basin sedimentation; (4) paleocurrent directions in the main facies largely parallel to the migration direction of the depocenters (i.e., opposite to the younging direction of strata); (5) synclines concentrated in the central part of the basin; and (6) an absence of high-grade metamorphism and sparse igneous activity.

The deposits of the Izumi Group are divided into marginal and main (axial) facies (Ichikawa et al., 1979). The former consists of basal conglomerates and thick mudstones, and the latter consists of turbiditic sandstone alternating with mudstone. The depositional environments are considered to have been a line-sourced slope or fan delta for the marginal facies and a point-sourced submarine fan for the main facies (Nishiura et al., 1993; Tanaka, 1989, 1993; Tanaka and Maejima, 1995). Radiolarian assemblages (Okamura et al., 1984), based on the abundant occurrence of spongodiscid-form spumellars, indicate that the paleobathymetry of the main facies was “intermediate depth,” as defined by Empson-Morin (1984).

The depositional age of the Izumi Group ranges from early to middle Campanian (in the western and central parts of Shikoku Island) to late Campanian (eastern Shikoku) and Maastrichtian (Awaji Island and Kinki district), as determined from macrofossils (Bando and Hashimoto, 1984; Morozumi, 1986; Suyari, 1973), radiolarian zones (Hollis and Kimura, 2001; Yamasaki, 1987), and magnetic polarity data (Kodama, 1990, 2003). These ages indicate a rate of depocenter migration of ~1.4–1.9 cm yr⁻¹ (Fig. 2).

Clasts in conglomerate beds are mainly felsic igneous rocks (rhyolite, dacite, and granite), and sandstones in the main facies are mainly quartzo-feldspathic arenites (Nishimura, 1976; Yokoyama and Goto, 2000). Studies of diagenetic zeolite and mudstone porosity reveal that the maximum burial depth of the Izumi Group shallows to the east, decreasing from 5 to 6 km in the western part of Shikoku Island to 3 km depth in the Kinki district (Nishimura et al., 1980; Nishimura, 1984).

The Izumi Group in the study area (Niihama City) occurs in a narrow zone between granites of the Ryoke belt and high-pressure (P), low-temperature (T) metamorphic rocks of the Sambagawa belt (Fig. 3). The bounding fault to the Sambagawa belt is the Median Tectonic Line, which records dextral displacement during the Quaternary (Kaneko, 1966; Mizuno et al., 1993; Okada, 1980) and sinistral displacement during the Cretaceous (Ichikawa, 1980). The strata in this area generally strike ENE-WSW and dip 30°–60° to the south. In the eastern part of Niihama City, strata are gently folded into east-plunging synclines; strata on the northern limbs strike NE-SW and dip to the south, and strata on the southern limbs strike NW-SE and dip to the north. Several ENE-WSW–striking faults are oriented subparallel to the Median Tectonic Line. Small NW-SE–striking faults are present as secondary faults. Strata are locally tightly folded into left-stepping synclines and anticlines oriented parallel or slightly oblique to the ENE-WSW–striking faults.

In the eastern part of the study area, the Ryoke granite is unconformably overlain by thin packages (<50 m) of basal conglomerate and sandstone, followed by thick packages (~350 m) of thin-bedded mudstone (Fig. 4). These lithofacies correspond to the northern marginal facies of eastern Shikoku Island (Matsuura et al., 2002; Suyari, 1973). The main facies is represented in geological maps by five lithological facies (Figs. 3 and 4): (1) a conglomerate-dominated facies (100–300 m thick) distributed in western and eastern areas across a horizontal distance of ~10 km; (2) a sandstone-dominated facies, consisting of thick- to very thick–bedded sandstones, widely distributed throughout the study area; (3) a sandstone-mudstone facies; (4) a mudstone-dominated facies in southern areas, near the Median Tectonic Line; and (5) felsic tuff beds (1–30 m thick) locally intercalated with sediments within the main facies.

Lithofacies association (LA) I consists of stacks of graded and nongraded clast-supported conglomerates (facies Gc1 and Gc2) along with thick- to very thick–bedded, pebbly (facies Sp), structureless (facies Sm), and graded (facies Sg) sandstones (Table 1; Fig. 5). Subordinate lithofacies include matrix-supported conglomerates (facies Gm), weakly to strongly deformed slump deposits (facies SL1 and SL2), and thin-bedded mudstone and sandstone couplets (facies Zs). Small-scale (10–20 m), fining-upward successions are repeated from facies Gc1 or Gc2 to facies Sp, Sm, or Sg (Fig. 5).

Clast-supported conglomerate beds (facies Gc1 and Gc2) are generally thick- to very thick–bedded (0.5–3 m) and consist of granule- to boulder-sized clasts within a coarse sand matrix (Figs. 6A and 6B). Matrix-supported conglomerate beds (facies Gm) are thick- to very thick–bedded (1–5 m) and consist of granule- to boulder-sized clasts within a muddy sandstone or mudstone matrix (up to 10% of clasts by volume) (Fig. 6C). Clasts are subrounded to well rounded in facies Gc1 and Gc2, and angular to well rounded in facies Gm. Clasts are predominantly felsic volcanics (rhyolite and dacite) and granite, with lesser mudstone and sandstone. Channelization, basal erosion, and amalgamation are commonly observed in facies Gc1 and Gc2. Very thin mudstone beds are intercalated with clast-supported conglomerate beds or are absent altogether. Rare fragments of marine fossils are found in both facies Gm and Gc.

The distribution of LA I corresponds to the conglomerate-dominated facies in the geological map (Fig. 3), and it is divided into western and eastern parts. In addition, small elongate or lenticular bodies are intercalated with the sandstone-dominated facies. LA I is stratigraphically underlain by mudstone-dominated successions (LAs IV and V) and overlain by sandstone-dominated successions (LA II).

This coarse-grained lithofacies association is considered to represent deposition in a subaqueous environment, given the lack of evidence for emergence (i.e., absence of coals, parasols, plant roots, and mud cracks), the presence of a deep-water trace-fossil association, and the presence of fragments of marine fossils. Graded (Gc1) and nongraded (Gc2) conglomerates were possibly deposited from gravelly debris flows or high-concentration turbidity currents (Lowe, 1982; Mulder and Alexander, 2001; Pickering et al., 1986). Structureless parts (Gc2) of the association represent highly concentrated traction carpet material (Walker, 1977), whereas graded parts (Gc1) were deposited from dispersive pressure. Matrix-supported conglomerates (Gm) probably represent gravelly debris flows or cohesive flows (Mulder and Alexander, 2001; Pickering et al., 1986). Channelized or erosional bases of conglomerate and sandstone beds indicate the occurrence of feeder channels. Clast-supported conglomerates might also be interpreted as residual facies in channels (Mutti and Normark, 1991). Lithological variation in the form of small-scale upward fining is interpreted to represent progressive channel abandonment (Walker, 1977).

A high slope gradient is suggested by the occurrence of debris flows and high-concentration turbidity currents. Possible depositional environments for this association include a channel fill (e.g., Bruhn and Walker, 1997), inner fan (Mutti and Ricci Lucchi, 1972), or upper fan (Normark, 1970, 1978; Walker, 1978).

The gradual transition from conglomerate-dominated LA I to sandstone-dominated LA II suggests a lack of fan progradation. One interpretation of the observed upward fining is long-term changes in the source area that possibly produced smaller and finer-grained flows (Walker, 1977). A similar interpretation was made by Bruhn and Walker (1997) for canyonfilling deposits on the Atlantic margin of Brazil, who ascribed an upward-narrowing, thinning, and fining stack of channel complexes to a gradual decrease in tectonic activity during deposition, which influenced turbidite sedimentation by reducing the long-range sediment supply. An alternative interpretation of the fining-upward sequence is backfilling of channels by channel-mouth lobes (Chen and Hiscott, 1999; Normark et al., 1998).

LA II is characterized by structureless (facies Sm), pebbly (facies Sp), and graded (facies Sg) sandstones (Table 1; Fig. 5), with subordinate thin-bedded mudstone and sandstone couplets (facies Zs), graded conglomerate (facies Gc1), and slump deposits (facies SL1). Sandstone beds show a generally tabular geometry and good lateral continuity (Fig. 6D), although shallow channels and lateral amalgamation of sandstone beds are sometimes observed. Structureless sandstone beds (facies Sm) are thick- to very thick–bedded (0.3–3 m), coarse- to very coarse–grained, and show no internal sedimentary structures. Graded sandstones (facies Sg) are thin- to thick-bedded and medium- to coarse-grained. The bases of beds locally contain granule- to pebble-sized clasts (Fig. 6E). Thick- to very thick–bedded pebbly sandstone beds (facies Sp) contain scattered granule- to pebble-sized clasts and are occasionally weakly laminated. Thick sandstone beds of facies Sm and Sg are occasionally accompanied by parallel or wavy laminations in the uppermost parts (Fig. 6F), comparable with Bouma sequence Ta–b. Sole marks of flute, groove, and gutter casts are common at the base of thick sandstone beds, and amalgamation is frequently observed in thick beds. Although body fossils are rare in this association, trace fossils (mainly Nereites ichnofacies) are sometimes observed in thin-bedded mudstone beds.

The ratio of sandstone to mudstone ranges from 10:0 to 7:3, and the sandstone beds become thinner with increasing proportion of mudstone. No systematic lithological cycles are recognized in this association. The distribution of the association corresponds to the sandstone-dominated facies on the geological map (Fig. 3). LA II grades upward into the sandstone-mudstone facies of LA III.

Pebbly sandstone (facies Sp) beds probably formed from traction carpets of gravelly high-concentration turbidity currents (Stow and Johansson, 2000). Structureless sandstone (facies Sm) beds are considered to represent rapid deposition from suspension without the influence of traction (Hiscott and Middleton, 1979) or sustained (prolonged, quasi-steady) flows of high-concentration turbidity currents (Kneller and Branney, 1995). Normally graded sandstone beds associated with parallel laminations, convolute laminations, and ripple cross-laminations are interpreted to represent deposits from waning turbidity currents (Lowe, 1982; Middleton and Hampton, 1973).

A sheet-like geometry and laterally persistent sandstone beds are interpreted to indicate deposition from unconfined turbidity currents upon a submarine fan or basin floor (e.g., Mutti and Normark, 1991). The lack of recognizable channels provides additional evidence for an extensive depositional environment rather than an erosional setting. The depositional environment could be interpreted as lobes or frontal splays (Lien et al., 2003; Posamentier and Kolla, 2003), suprafan lobes (Normark, 1978; Reading and Richards, 1994; Walker, 1978), or depositional lobes (Mutti and Ricci Lucchi, 1972). The dominance of coarse-grained sandstone beds suggests a sand-rich submarine-fan system (Reading and Richards, 1994; Shanmugam and Moiola, 1991). The occurrence of amalgamated sandstone beds and lack of interbedded mud-stone suggest a depositional environment in the proximal part of lobes or frontal splays. A gradual shift to finer and thinner facies (LA III) could be explained by a reduction in sediment supply or the progressive backward-stepping of lobes (Chen and Hiscott, 1999).

LA III is characterized by thin- to medium-bedded sandstones showing grading (facies Sg) and parallel or cross-lamination (facies Sc), along with thin-bedded mudstone and sandstone couplets (facies Zs) (Table 1; Fig. 5). Conglomerate and slump deposits are rare, but several thick-bedded structureless sandstone beds (facies Sm) are recognized. Graded and laminated sandstones are medium- to fine-grained and medium- to thick-bedded (~50 cm). Beds are generally laterally continuous. Compared with LA II, channelization, amalgamation, and basal erosion are rarely observed; the grain sizes of sandstones are finer (fine- to medium-grained); and beds are thinner (thin- to medium-bedded). Trace fossils of Nereites ichnofacies are common in mudstone beds, as in LA II. The ratio of sandstone to mudstone ranges from 7:3 to 3:7. The distribution of this facies corresponds to the sandstone-mudstone facies on the geological map (Fig. 3).

Graded and laminated sandstones are comparable with Ta–d of the Bouma sequence and are therefore considered to represent deposition from low-density turbidity currents (Middleton and Hampton, 1973; Mulder and Alexander, 2001). A reduction in bed thickness, grain size, and ratio of sandstone to mudstone indicates a more distal setting of LA II. Parallel-bedded sandstone and mudstone couplets possibly indicate slow hemipelagic deposition, interrupted periodically by turbidity currents. The depositional environment of this association is regarded as the distal part of lobes or frontal splays (Lien et al., 2003; Posamentier and Kolla, 2003), a lower fan (Normark, 1978; Walker, 1978), or an outer fan (Mutti and Ricci Lucchi, 1972).

LA IV is characterized mainly by thin-bedded mudstone and sandstone couplets (facies Zs), irregularly intercalated with slump deposits (facies SL1 and SL2) (Fig. 5). Facies Zs consists of intercalated thin-bedded mudstone and sandstone, and the ratio of sandstone to mudstone varies from 1:9 to 3:7. Sandstones vary from fine- to coarse-grained and show tabular and channel-like geometries. Some coarse-grained sandstone beds contain pebble- to cobble-sized mud clasts; muddy intraclasts are comparable with LAs I–III. Bioturbation is common in the mudstone beds. Slump deposits of facies SL1 contain folded and contorted strata within a mudstone matrix, as also observed in LAs I and II. No exotic blocks or clasts are found in facies SL1. Slump deposits of facies SL2 are polymictic and characterized by the ductile-deformation–related mixing of coarse-grained sandstone and mudstone. Felsic volcanic and granite clasts occur locally in facies SL2. Syn-depositional low-angle unconformities are also observed, commonly associated with slump deposits. This association is distributed along the coastline in the northern marginal facies.

The monotonous stacking of tabular, laterally continuous mudstone and thin-bedded sandstone beds suggests pelagic sedimentation and expansive turbulent flows on a slope or basin floor. Thin tabular sandstone beds suggest deposition from low-concentration turbidity currents upon an unconfined slope or basin floor. In contrast, the occurrence of coarse-grained, channelized sandstone beds with mud intraclasts indicates high-concentration turbidity currents related to slumps or debris flows (Johansson and Stow, 1995). Muddy slump deposits indicate the presence of unconsolidated mud or semiconsolidated mudstone on the slope. The slump deposits are mainly related to gravity-induced sediment sliding and slumping, caused by a range of mechanisms including seismic activity, excess pore-water pressure in response to active sediment loading, and oversteepening of submarine slopes (e.g., Hampton et al., 1996). Local intercalations of slumps and coarser-grained sediments within finer-grained sediments indicate a slope-apron facies (Hill, 1984; Richards et al., 1998).

LA V is characterized by thick units of thin-bedded mudstone (facies Zs) and interbedded thin-bedded sandstone (facies Sg, Sc, and Sw) (Table 1; Figs. 5 and 6G). The ratio of sandstone to mudstone ranges from 1:9 to 3:7. No thick turbidites or slump deposits are recognized. Structureless mudstone (facies Zm) is thick-bedded, and silt-sized to sandy silt-sized sediments are intensely bioturbated. Marine fossils of Inoceramus sp. and echinoidea are found in this facies. Wavy-laminated sandstone beds (facies Sw) are thin-bedded, medium- to fine-grained, ungraded, and intercalated with thick units of mudstone beds (Fig. 6H). Sand dikes and flame structures occur in thin-bedded sandstones within mudstone-dominated facies.

Parallel-laminated sandstone and mudstone beds are intercalated with hemipelagic muds and low-density turbidity currents. Thick-bedded, homogeneous mudstones possibly reflect prolonged suspension settling of low-concentration turbidity currents, or postdepositional bioturbation. This association is therefore interpreted as a basin-plain deposit. The absence of thick turbidites and slump deposits indicates a depositional environment located distal from a submarine fan or the basin floor.

Paleocurrent data were obtained from nearly 50 points in the study area. All measured data were collected relative to structural dip. Sole marks of flute and groove casts are well developed at the base of coarse-grained sandstones in LAs I–III. Clast imbrication is poorly developed in conglomerate beds due to the high sphericity of clasts. The average paleocurrent direction determined from sole marks is 241° for flute casts and 259° for groove casts, representing NE-ENE to SW-WSW directions (Fig. 7). Clast imbrication indicates a paleocurrent direction of 249°, consistent with the sole marks. The paleocurrent directions are largely subparallel to the strike of bedding, although some N-S directions are recorded in the northeastern part of the study area, where slope-apron facies are distributed.

Slump folds provide clues to paleoslope direction, although some caution is required in making such interpretations (e.g., Lajoie, 1972; Smith, 2000; Strachan and Alsop, 2006; Woodcock, 1979). The limited extent of outcrops means that we adopted the simplest measurement method proposed for slump folds: the mean axis method (Jones, 1939). The results indicate the paleoslope dipped to the SSW (Fig. 8), although there is some variability. The paleoslope direction deduced from slump folds is oblique to that deduced from sole marks and imbrication.

In constructing a depositional model of the Izumi Group, analogue experiments were previously performed in the form of a fault-bend basin along a releasing bend of a northward-dipping, left-lateral (i.e., top to the west) strike-slip fault in a sand box (Iwamoto and Miyata, 1994; Miyata and Iwamoto, 1994). The authors alternately generated displacement along the fault in the sand-filled box and filled the created basin with sand, thereby revealing stepwise eastward migration of depocenters. The experiments also indicated the occurrence of an extensional strike-slip duplex and the generation of secondary NE-SW–striking normal faults, as observed in the field (Ichikawa et al., 1981; Miyata, 1989, 1990). Yamakita and Ito (1999) conducted numerical simulations to evaluate the deformation of the sedimentary body in a fault-bend basin generated by a low-angle strike-slip fault with a releasing bend (i.e., an extensional oblique ramp). The authors reproduced eastward-plunging asymmetric synclinal structures resulting from drag effects along the fault.

Neither of the aforementioned studies considered the depositional environment or the lithological evolution of coarsening- or fining-upward sequences. In addition, they assumed that the basins were always filled by sediment as the fault moved: the accumulation rate was exactly balanced by the subsidence rate. The progressive infilling of a basin is expected to result in shallowing- and coarsening-upward sedimentation; however, the sediments in the axis of the Izumi Group do not shallow upward, suggesting the basin was underfilled throughout its sedimentation history. An appropriate numerical simulation, therefore, should be conducted while taking into account the depositional environment and lithostratigraphic variations.

The diffusion equation is commonly used to describe the transport of sediment down a slope, where the rate of transport is proportional to the local gradient (Culling, 1960; Flemings and Jordan, 1989; Kenyon and Turcotte, 1985). For a given topographic profile, we can use this relationship to specify sedimentation and thereby calculate the change in accumulation at every point along the x axis. Assuming that erosion of the basement and sediments is negligible in the basin, the condition ∂z/∂x = 0 is used at points where ∂²z/∂x² > 0.

To minimize topographic change of the basement during fault displacement, the initial slope profile is assumed to be the same as that of the fault plane (Fig. 9D). This model geometrically yields ~3.2 km² of accommodation space for every 1 km of fault displacement. The total displacement is set to 80 km, corresponding to 256 km² of accommodation space. This implies the equivalent volume of mountain uplifted at the source area. In the simulations, we assume that half of the volume of the uplifted mountains is eroded and supplied to the basin.

The diffusion equation formulated is independent of grain size; thus, the downstream slope decrease forced by deposition does not directly result in downstream fining (cf. Paola, 2000). Therefore, there is no conservation of mass for sediments of each grain size, although the magnitude of the slope is inferred to reflect the energy of deposition. Sediment of a single grain size would be deposited on progressively diminishing slopes that correspond to diminishing energy conditions. The lithology in this study is assumed to depend on the depositional slope (e.g., Flemings and Jordan, 1989).

Equation 1 was solved numerically using an explicit method. Values used in this model are listed in Table 2 and shown in Figure 10. Calculations ceased when the sediment supply Q_s achieved the values shown in Figure 10 at each time step. Five cases were tested, each with varying fault slip (dx) and sediment efflux (Q_s): (A) constant dx and Q_s, (B) constant dx and varying Q_s, (C) varying dx and constant Q_s, (D) intermittent dx and constant Q_s, and (E) varying dx and Q_s.

Constant rates of basement retreat and sediment supply in case A mean linear increases in accommodation space (A) and sediment deposition (S). No periodic variability represented by this case resulted in continuous migration of the depocenter and noncyclic lithostratigraphy (Fig. 10A).

In case B, even though the rate of fault slip was constant, sediment flux into the basin was variable. A relative increase/decrease of sediment input to fault-slip rate (rate of creation of accommodation space) generated a cyclic stratigraphy composed of fining-upward and coarsening-upward successions (Fig. 10B). In this case, a large increase of sediment flux in short duration caused the successions to rapidly coarsen and then gradually fine upward.

Case C is the opposite to that of case B: the slip rate is variable (cf. Chéry and Vernant, 2006), and sediment supply is constant (Fig. 10C). Variation in the slip rate also resulted in temporal changes in the ratio of accommodation generation, yielding cycles of upward-fining and upward-coarsening successions similar to case B.

Case D represents cycles of very rapid and long quiescent phases of fault movement (Fig. 10D). This case resulted in repeated episodes of progradational sedimentation that produced coarsening-upward successions, comparable with typical sedimentary successions found in strike-slip basins such as the Hornelen Basin, Norway (e.g., Steel et al., 1977). Large slip rate caused migration of the depocenter and resulted in cyclic stratigraphy.

In case E, the sediment-supply rate corresponds to the fault-slip rate at a fixed ratio (Fig. 10E). This case represents a basin-fill model with instantaneous response of sediment supply to tectonics. No cyclicity was shown in the simulated lithology due to increase of the accommodation space in proportion to sediment input.

These cases demonstrated that the depocenter gradually shifted toward the sediment source point and that the depositional environment became increasingly distal during the period in which subsidence exceeded sediment accumulation. The period of sediment accumulation in excess of subsidence led to progradation, during which coarse and more proximal sediments were deposited farther from the source.

The occurrence of conglomerate- and sandstone-dominated successions (LAs I–III) indicates a likely depositional environment of a coarse-grained, gravel/sand–rich submarine fan (channel-lobe system). The unidirectional paleocurrent directions recorded throughout these successions suggests an elongate basin and point-sourced submarine fans from a channelized sediment-transfer area (LA I) to a distal depositional zone (LAs II–III) (Fig. 11A). The basin width is estimated to exceed 3–5 km, although the southern margin of the basin is not exposed in the study area. The submarine channel–fan contains a set of coarse-grained channel-fill successions (LA I); thick-bedded, tabular sandstone beds in the proximal part of lobes or frontal splays (LA II); and fine-grained sandstone- mudstone couplets in the distal part of the lobes or frontal splays (LA III). LA I is considered to be a transfer zone of fine-grained material in channel flows, as deduced from the occurrence of channelized erosion, clast-rich conglomerates, and relatively rare mudstone beds. The transition from the transfer to depositional zones was possibly related to changes in the flow condition of turbidity currents at a “transition point” (Posamentier and Kolla, 2003). This change from confined, high-gradient channels to unconfined, gently sloping lobes (depositional) probably resulted in reduced flow velocity and deposition of the sediment load from sand-rich turbidity currents, leading to the stacking of thick-bedded sandstone facies (LA II).

The basin-marginal facies are slope-apron facies (LA IV) rather than an alluvial fan or coarse-grained fan delta (Fig. 11A), in contrast to other strike-slip basins (Crowell, 2003; Steel et al., 1977). This observation indicates that the basin and boundary faults might have subsided completely beneath the sea. The large number of beds in the slope-apron facies affected by gravity-induced deformation suggests seismicity-related sediment instability on the slope, indicating intense syndepositional fault activity.

The geological map and lithological column (Figs. 3 and 4) show two sets of submarine channel–fan units, with the upper unit stacked on the lower unit. The offset between the two units (~10 km) indicates the distance of backward migration of depocenters associated with displacement along the bounding faults. Although many authors have claimed that submarine-fan lobes generally preserve coarsening-upward sequences due to progradation of the fan (Mutti and Ricci Lucchi, 1972; Shanmugam and Moiola, 1988; Walker, 1978), the lithology of each of the fan units in the present study shows upward fining of grain size and thinning of beds rather than systematic coarsening or thickening (Fig. 4). One explanation for the lack of thickening-upward cycles is that the lobes were aggradational rather than progradational (e.g., Hiscott and Ghibaudo, 1981; Ricci Lucchi and Valmori, 1980), or that backward filling of the fan lobes occurred (Chen and Hiscott, 1999); however, this interpretation does not provide a full explanation of the displacement and rejuvenation of the fans. One possible interpretation is a gradual retreat or subsidence of the basement along a strike-slip fault, resulting in a progressive receding of the source area and deposition of increasingly distal facies at a given site, as shown in Figures 11B and 11C. Given that the submarine-fan units would have had point sources, we can ignore the fact that they had multiple source points from basin-side areas, which could also have caused lateral offset of source areas and stacking of fans along the bounding faults (e.g., Crowell and Link, 1982; Reading, 1980; Steel, 1988).

Many processes of cyclical behavior, such as global and local sea-level changes, climatic oscillations, and temporal and spatial variations of tectonic forces, affect stratigraphical architecture. It is, therefore, difficult to constrain a unique control on the stratigraphic cyclicity or to extract the dominant process quantitatively at the present stage. The stratigraphy in the study region is closely associated with migration of the depocenters, although global eustatic or climatic oscillations could be independent to the fault activity. We therefore propose a model where the basin-filling processes were primarily controlled by activity of the strike-slip fault before discussion of some causes of cyclic stratigraphy.

The development of the stratigraphic architecture is basically controlled by the ratio of accommodation space (A) generated by tectonic subsidence/retreat to sediment accumulation (S), which depends on sediment-supply rate. If basin subsidence/retreat exceeds sediment infill (i.e., high A/S), basement retreat would lead to back-filling of the submarine fan, potentially leading to upward-fining and upward-thinning successions; thus, the depositional environment would change to a relatively distal facies. In the case that accommodation space and sediment accumulation are balanced (cases A and E), any stratigraphic trend would not be developed. For low values of A/S, the submarine fan would aggrade or prograde; sediments would become coarser and thicker, and shallower successions would be deposited in the basin. Therefore, upward-coarsening and upward-thickening deposits could be explained by a decrease in tectonic activity or increase in sediment supply. The stratigraphy in the study area shows repetition of rapidly coarsening-upward and gently fining-upward successions with migration of the depocenter. This scenario is represented by cases B and C in Figure 10, with fluctuating A/S values. In both cases, tectonics represent the primarily control on progressive migration of the depocenter.

Case B (Fig. 10B) represents periodic changes in the sediment-supply rate to the basin with constant fault-slip rate. The sediment-supply rate into the deep-sea basin is partly controlled by the denudation rate in the source area; indeed, sediment yield shows a correlation with elevation or uplift rate (Ahnert, 1970; Montgomery and Brandon, 2002; Pinet and Souriau, 1988). Elevation or uplift rate may vary as a function of fault-slip rate, which is controlled by tectonics, and thus sediment yield is ultimately dominated by tectonics (e.g., Heller and Paola, 1992). Paleoclimatic changes, including precipitation, temperature, and biogenic activity, so-called Milankovitch mechanisms, can also influence the erosion rate with frequency of 10⁵ yr (e.g., Zhang et al., 2001). Such climatically driven changes in sediment supply are likely to be independent of tectonism (Koltermann and Gorelick, 1992). The fourth-order (10⁵ yr) glacioeustatic variation is another potentially dominant factor in controlling deposition on submarine fans, despite very limited evidence for glaciation during the Cretaceous. Submarine fans commonly develop in response to sea-level lowering or forced regression (Posamentier et al., 1991; Stow et al., 1985). In the case that rivers are directly connected to the shelf edge or upper slope, frequent deep-water turbidity currents are likely to enter the basin. Whereas coarser sands tend to be deposited behind the shoreline or on the shelf during periods of high sea level, deposition upon the submarine fan by high-concentration turbidity currents would be reduced.

Lithological cyclicity might also result from fluctuations in subsidence rates arising from episodic changes in fault-slip rates (case C in Fig. 10C). Discontinuous strike-slip movement has been deduced from cyclothems in some pull-apart basins (Blair and Bilodeau, 1988; Steel et al., 1977). Especially in the case of a nonmarine basin such as Hornelen Basin, eustatic control on cyclicity should not be considered. Periods of tectonic quiescence (i.e., low A/S) might be expected to result in a relatively small increase of accommodation space to sediment input, thereby enabling the progradation or aggradation of sedimentary successions in the basin. The sense of fault slip (i.e., vertical, transpressional, or transtensional) would affect accommodation generation and sediment supply, and it could be another control on the sedimentation pattern and basin-fill architecture of marginal depositional systems (Kim et al., 2003; Lee and Chough, 1999).

The average rate of depocenter migration along the Median Tectonic Line can be estimated to be 1.4–1.9 cm yr⁻¹, indicating that a unit of the stratigraphic cycle with 10 km offset requires a period of 5–7 × 10⁵ yr. It is possible that fault-slip rates have temporal variations on geological time scale (10⁴–10⁶ years) longer than earthquake elastic cycles (Bennett et al., 2004; Chéry and Vernant, 2006; Friedrich et al., 2003; Kirby et al., 2006). Another example of tectonic deformation with long time scales is coupled lowering and raising of basement in subduction-related active margins (Flint et al., 1991). On the western coast of South America, oblique subduction of multiple aseismic ridges may result in cycles of forearc uplift/subsidence that produce relative sea-level changes at a frequency equal to glacioeustatic processes (~10⁵ yr frequencies). A cyclic ridge subduction (15–35 km wavelength) has also been reported from Japanese forearc (Kodaira et al., 2003), under which the Philippine Sea plate subducts at the rate of 2–4 cm yr⁻¹ (Seno et al., 1993); this implies a frequency of 4–18 × 10⁵ yr. Geology during the Late Cretaceous age in the northwest Pacific was characterized by rapid oblique subduction with a mid-ocean ridge between the Izanagi-Kula and Pacific plates (Maruyama et al., 1997; Uyeda and Miyashiro, 1974). Episodic subduction of ridges or seamounts might induce temporal variability in the fault-slip rate.

We here propose a model where temporal variations of fault-slip rate control the stratigraphic cyclicity. At times when the bounding fault moves fast, the basement rapidly retreats and subsides (Fig. 12A). The basin subsidence makes relative sea level rise, therefore increasing accommodation space. In the initial phase of a large fault slip, rapid retreat/subsidence induces a deficiency of sediment throughout the basin (cf. Pitman and Andrews, 1985) because a certain response time is needed for fault-induced uplift in the source area and hence a rejuvenation of the supply of detritus by erosion—considered to be the same order of 10⁴–10⁶ yr as for large rivers (Castelltort and Van Den Driessche, 2003; Tucker and Slingerland, 1996). Small amounts of fine-grained sediments would therefore accumulate in the basin. Thereafter, sediment input gradually increases in response to enhancement of denudation in the source region (Fig. 12B). Larger amounts of coarse-grained detritus supplied to the basin would prograde/aggrade the coastal delta or submarine fan, as indicated by relative sea-level fall and development of coarsening- and thickening-upward successions. After local relief is subdued by progressive erosion, sediment supply into the basin gradually decreases. In addition, continuous receding of the source area and isostatic subsidence in the basin cause relative sea-level rise and development of fining- and thinning-upward successions (Fig. 12C).

Geological survey and numerical simulation have revealed the depositional environment and the lithological cyclicity related to strike-slip faulting in the Izumi Group (Upper Cretaceous), southwest Japan. Sedimentary facies analysis discriminated five lithofacies associations (LAs): conglomerate-dominated successions representing submarine channel-fill deposits (LA I); coarse-grained, thick-bedded sandstone-dominated sediments indicating proximal part of lobes or frontal splays (LA II); interbedded sandstone and mudstone successions in the distal part of LA II (LA III); slump deposits in mudstone-sandstone sequences of a slope-apron facies (LA IV); and mudstone-dominated facies of the basin floor (LA V). Sole marks and clast imbrication indicate unidirectional paleocurrents from ENE to WSW. The orientations of slump folds indicate that the paleoslope dipped to the SSW for the marginal slope. LAs I–III, therefore, would constitute point-sourced submarine channel–fan successions in the axial facies, and LAs IV–V would represent marginal facies of the slope apron and basin floor.

Two units of submarine channel–fan successions are stacked with 10 km eastward (backward) offset; these consist of a rapidly upward-coarsening and upward-thickening succession in the lower part, and a gradually upward-fining and upward-thinning succession in the upper part. The stratigraphic cyclicity is closely associated with migration of depocenters; it would take 5–7 × 10⁵ yr. Numerical simulations of pull-apart basins produced by strike-slip faults and filled by sediments have indicated that episodic changes of fault-slip rate or sediment-supply rate could explain cyclic stratigraphy. We proposed that temporal variations of the fault-slip rate dominated the cyclic stratigraphy in this case. In the initial phase of active fault movement, rapid retreat/subsidence of the basement caused migration of the depocenter, resulting in accumulation of fine-grained sediments due to deficiency of sediments. Thereafter, progressive erosion on the source land might have increased sediment flux into the basin and prograded/aggraded submarine fans, indicating development of coarsening-upward successions. In the quiescence of the fault, gradual reduction of sediment input by erosion and incremental basement retreat/subsidence initiated fining-upward successions.

We are grateful to Jun Tanaka for insightful and rigorous comments on an early version of the manuscript. We also thank Kazunari Nawa for isostatic response calculations, Kazuhiro Miyazaki for the finite-difference model, and Toshiyuki Kurihara for radiolarian fossils. Excellent thin sections were made by Akira Owada, Takumi Sato, and Kazuyuki Fukuda. We acknowledge anonymous reviewers and Editor Jon D. Pelletier, whose constructive comments significantly helped us to improve the manuscript. This research was part of the “Geological Mapping Project” supported by the Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology (AIST).