Pliocene – Pleistocene history of the Gippsland Basin outer shelf and canyon heads, southeast Australia

Abstract

The Gippsland Basin on Australia's southeastern continental margin is host to a number of large shelf-breaching canyons that form part of the Bass Canyon system. Analysis of high-resolution bathymetry data and biostratigraphically controlled shallow seismic data across the shelf and upper slope and associated canyon heads shows that widespread erosion at the canyon heads began in the earliest Pleistocene (CN13a, 1.95 – 1.72 Ma). Deep V-shaped channels were eroded into pre-existing U-shaped channels. Erosion was not necessarily a response to Pleistocene lowstands but followed a period where (carbonate) sedimentation rates on the shelf were greatly elevated. Both carbonate production and erosion were likely to be amplified, if not caused, by intensification of the Bass Cascade density current during Late Pliocene eustatic highstands. During the Middle Pleistocene (CN14a, 0.95 – 0.47 Ma) outer shelf canyon-head channels were rapidly infilled. Canyon-head channels migrated laterally across the shelf, predominantly in a northeast direction, influenced by the prevailing northeast moving Bass Cascade. Upper slope canyon heads have remained largely non-depositional exposing Pliocene strata at the seafloor. A minor episode of deformation occurred during the Middle Pliocene (CN12, 3.70 – 1.95 Ma) that produced low-amplitude open folds in Upper Pliocene and older sediments, beneath the northern inner shelf. Beneath the central inner shelf and middle shelf, similar open folds developed during the Late Pliocene to at least the Middle Pleistocene (CN13 – CN14a, 1.95 – 0.47 Ma).

Key words:

• Bass Cascade
• density currents
• cool-water carbonates
• Gippsland Basin
• sedimentation rate
• seismic stratigraphy
• submarine canyons
• swath-mapping

Introduction

Modern (largely unfilled) submarine canyons are well documented from a variety of continental-margin settings around the world, where they act as significant modifiers of slope and shelf-break environments (Andrews et al. 1970; Twichell & Roberts 1982; Hagen et al. 1996; Lewis & Barnes 1999; Kenyon et al. 2002). Modern submarine canyons are common on the continental slope across southern Australia (von der Borch 1968; Exon et al. 2005). The most prominent of these are the Albany Canyons (Exon et al. 2005), the Murray Canyons (Sprigg 1947; Hill & De Deckker 2004; Hill et al. 2005), the Glenelg Canyons (Leach & Wallace 2001) and the Gippsland Basin canyons (Conolly 1968). The Bass Canyon of the Gippsland Basin is arguably the largest and most spectacular of these modern canyon systems. Five shelf-breaching tributary canyons coalesce on the lower slope at 3000 m depth, where they are captured by the huge deep-water Bass Canyon, which drains southeast for 80 km to the Tasman Abyssal Plain below 4000 m depth.

However, the age and evolutionary history of the Bass Canyon are largely unknown. Ancient (infilled) canyons are considerably less well documented, and as such, surprisingly little is known about their evolution and significance on palaeo-continental margins (Leach & Wallace 2001; Bertoni & Cartwright 2005). Seismic profiles of the offshore Gippsland Basin reveal complex canyon-fill structures from the Early Oligocene (Conolly 1968; Maung & Cadman 1992; Holdgate et al. 2000; Wallace et al. 2002), and show evidence that the geometry and size of the modern-day Bass Canyon system are influenced at least partially by precursor Pliocene canyon systems. The Glenelg Canyons also have a geological record which extends back into the Oligocene (Leach & Wallace 2001), while Hill et al. (1998, 2005) and Exon et al. (2005) have suggested histories for the Bass, Murray and Albany Canyons that may extend back to Cretaceous continental breakup.

Submarine canyons probably result from a combination of factors (Shepard 1981) including subaerial exposure, structural predisposition, bottom currents, biological activity, sediment loading and long-time persistence. However, studies from a wide range of slope environments suggest that the prevailing canyon-forming process is sediment gravity flows (Conolly & von der Borch 1967; Keller & Shepard 1978; Howard 1992; Inman 1994; Pratson et al. 1994; Kudrass et al. 1998; Khripounoff et al. 2003).

The extensive buried canyon structures, identified from seismic exploration (Maung & Cadman 1992; Holdgate et al. 2000) in the Gippsland Basin's temperate carbonate Seaspray Group, have been problematic for seismic interpretation of underlying hydrocarbon targets. Canyon-fill sediments are often associated with anomalously high sonic-velocity zones, attributed to carbonate-rich sediments and burial diagenesis (Wallace et al. 2002). However, the architecture and distribution of these diagenetically susceptible facies are poorly understood. This study reports on aspects of Late Neogene sedimentology, stratigraphy and sediment architecture preserved in the outer shelf, upper slope and canyon-head environments of the Gippsland Basin, and forms part of a broader study that aims to better understand the sedimentary processes controlling canyon facies architecture.

Geological setting

The Gippsland Basin is the most eastern of Australia's southern continental-margin basins that formed during the breakup of eastern Gondwana in the Mesozoic and Cenozoic ( Figure 1). It occupies an area of 56 000 km², of which two-thirds is offshore (Smith 1982). It is bound to the east by the Tasman Sea; to the west and southwest by uplifted Lower Cretaceous rocks of the Balook Block and Bassian Rise; and to the north by the Eastern Highlands. Fluvial and lacustrine clastic sediments of the Strzelecki Group were deposited during Early Cretaceous rifting and thermal subsidence. Fluvial, lacustrine and marginal-marine clastic sediments of the Latrobe Group were subsequently deposited during the Late Cretaceous to Eocene (Veevers 1986; Rahmanian et al. 1990; Veevers et al. 1991; Willcox et al. 1992; Moore & Wong 2001) and are the primary hydrocarbon-bearing rocks of the Gippsland Basin. Continued thermal subsidence produced a marine transgression over most of the basin during the Oligocene depositing up to 2500 m of fully marine carbonates known as the Seaspray Group. The predominant bathymetric feature of the modern-day Gippsland Basin is the 80 km long southeast-trending, Bass Canyon, located at 3000 – 4000 m water depth. Draining into the Bass Canyon are five shelf-breaching tributary canyons, the heads of which are the focus of this study (Figure 2).

Figure 1 Gippsland Basin locality and major geostrophic currents effecting the southern Australian margin (modified from Smith and Gallagher 2003, who adapted Martinez 1994 and James et al. 1999).

Figure 2 High-resolution swath-mapped seafloor bathymetry image of the offshore Gippsland Basin and Bass Canyon. Locality of R/V Franklin cruise FR11/98 shallow (sparker) seismic survey, piston-core sites, and major Esso/BHP oil and gas production platforms. Bathymetry data provided by Geoscience Australia (Exon et al. 1999; Harris et al. 2000). The study area extends southeast from Ninety Mile Beach to Bass Canyon. Most good-quality shallow seismic lines are figured.

Oceanographic setting

The offshore Gippsland Basin lies between 38 and 39.5°S, and has mean annual surface temperatures ranging from 12 to 20°C. The 15°C and 10°C isotherms occur at 100 m and 400 m, respectively (Levitas 1982). The wind climate is dominated by westerlies, producing a moderate to high-energy wave-dominated environment. Forcing from the west of Bass Strait produces an eastward-moving winter current that moves bottom sediment eastward across Bass Strait (Li et al. 2005) ( Figure 1). This current, called the Bass Cascade (Godfrey et al. 1980), is generated by relatively cold, dense and saline water from Bass Strait sinking below the warmer, less-saline water of the Tasman Sea, which occurs a few kilometres landward of the 200 m isobath.

Gallagher et al. (2003) suggested from planktonic proxy data that cool conditions occurred during the initial stages of the Early Pliocene. Subsequently, conditions warmed and stabilised, while a weaker (compared to today) Sub-tropical Convergence advanced northward causing upwelling at the outer shelf and upper slopes of the basin. During the Middle and Late Pliocene, foraminiferal assemblages fluctuated between cool and warm planktonic taxa, corresponding to fluctuations in global δ¹⁸O levels and the expansion of polar ice sheets. During the Pleistocene, intensification of Northern Hemisphere glaciations periodically exposed the Gippsland shelf, allowing fluvial channels to meander across the shelf platform to the present-day shelf-break and canyon heads (Holdgate et al. 2003).

Previous work

The Bass Canyon and its largest tributary, the Everard Canyon, were first described by Conolly (1968) using soundings from the Royal Australian Navy Hydrographic Office. Facies distribution maps of the surficial sediments of the Gippsland and surrounding shelves were produced by Davies and Marshall (1973), Jones et al. (1975), Davies (1979), Jones and Davies (1983) and Smith et al. (2001). Limited modern samples were taken from the Bass Canyon by Schneider (1985) and Colwell et al. (1987). Upper Cretaceous sediments exposed in the canyon walls were dredged by HMAS Cook (Marshall 1988), and a comprehensive piston core and surface sediment grab sampling scheme was undertaken by the RV Franklin cruise in 1998 (Exon et al. 2002; Holdgate et al. 2003; Smith & Gallagher 2003).

The Oligocene to Holocene sediments of the onshore Gippsland Basin have been well studied (Singleton 1941; Crespin 1943; Partridge 1971; Hocking 1976; Thompson & Walker 1982; Gallagher & Holdgate 1996; Holdgate & Gallagher 1997). Consequently, the stratigraphic relationships onshore are well understood. The offshore equivalents are less well understood as the majority of data acquired from the offshore region have been focused on the oil- and gas-bearing Upper Cretaceous to Eocene Latrobe Group. Interest in the stratigraphy of the overlying Oligocene to Holocene Seaspray Group has been relatively recent (Bernecker et al. 1997; Feary & Loutit 1998; Holdgate et al. 2000). More recent work in the Gippsland Basin has focused on palaeo-environmental data (Gallagher et al. 2001, 2003), sonic-velocity modelling (Wallace et al. 2002) and the Plio-Pleistocene eustasy, tectonics and sedimentology of the shelf (Holdgate et al. 2003).

Holdgate et al. (2003) assigned shelf sediments from eight Esso/BHP foundation bores to a range of ages from Early Pliocene nanno-zone CN10d – 11 through to the Late Pleistocene to Holocene nanno-zone CN15. Additionally, Holdgate et al. (2003) used magnetic imagery of the Gippsland Basin ( Figure 3) to identify a complex network of high-magnetic meandering fluvial channels beneath the shallow subsurface sediments of the shelf that are not expressed on the modern seafloor. The orientation of these channels (Figure 3) shows they drained predominantly south-southeast and are likely to have followed pre-existing depressions such as compacted buried canyons.

Figure 3 (a) Airborne magnetic imagery of the Gippsland Basin showing buried magnetic fluvial channels and magnetic barrier features identified by Holdgate et al. (2003). Image processed by the Geological Society of Victoria. Light pixels equate to high magnetic regions. (b) Simplified interpretation of airborne magnetic image, location of buried Plio-Pleistocene canyon heads, present-day canyon features and isobaths, and major petroleum platforms and pipelines. Numbers correspond with ‘seismic smudges’ identified on FR11/98 shallow seismic survey and match those labelled in Figures 5 and 6.

Methodology

A total of 621 km of shallow seismic (sparker) data was acquired across the Gippsland shelf, upper slope and canyon-head regions ( Figure 2) during the RV Franklin cruise FR11/98 using a single-channel receiver that logged at a 0.5 s firing rate (Keene 1998). Significant surfaces were traced around the survey ensuring closure and were then time/depth converted and structurally projected to age-constrained Esso/BHP foundation bores (Figure 4). Velocities for time/depth conversion were estimated using:

simplified from equation 2 of Wallace et al. (2002) where v is velocity (m/s) and d is depth (m).

Figure 4 Correlation of Esso/BHP foundation bore logs from the main offshore Gippsland Basin oil and gas fields amended from Holdgate et al. (2003). Strata are tied to the FR11/98 shallow seismic survey ( Figure 2) and have been divided into six units of approximately equal seismic thickness (time), defined by seismic facies and significant surfaces. Units are biostratigraphically age-constrained (McMinn 1992; Shafik 2000; Mays 2001; Holdgate et al. 2003).

Most of the FR11/98 data achieve a decametre-scale resolution of reasonable quality to 300 ms two-way travel time (TWT). However, in all lines, a strong sea-bed multiple is generated at approximately twice the time to the seafloor, below which underlying seismic reflections are often difficult to interpret. In addition, all lines have a strong coupling effect at the sediment/surface interface that swamps near-surface sediment reflections in the first 5 – 10 ms TWT, or approximately 4 – 8 m depth below the seafloor.

Age constraints are provided by multiple biostratigraphic studies of punch-core samples from eight Esso/BHP foundation bores (Barracouta 1, Flounder 2, Halibut, Kingfish 1, Mackerel, Tuna, Marlin and Snapper; Figure 2). These include biostratigraphic studies of dinoflagellates by McMinn (1992), calcareous nannofossils by Shafik (2000) and foraminifers by Mays (2001), Smith et al. (2001) and Holdgate et al. (2003). Ages for biostratigraphic units are based on the nannofossil zonation scheme of Okada and Bukry (1980).

Airbourne magnetic imagery of the Gippsland Basin was reformatted by the Geological Survey of Victoria to highlight near-surface magnetic anomalies on the shelf ( Figure 3). The swath bathymetry data come from two cruises, the Sojourn 7 swath-mapping cruise of RV Melville (Exon et al. 1999) and the Southern Surveyor cruise 001/00 (Harris et al., 2000): the data were provided to us as an amalgamated single set by Geoscience Australia.

Pliocene to Holocene tectonics and sedimentology

Our analysis of the FR11/98 survey has divided the stratigraphy into six units (A – F in Figures 4 –  8) of approximately equal seismic thickness (∼50 – 100 ms). Units are separated by significant erosional boundaries across the outer shelf and upper slope, or stratigraphic relationships such as onlapping or downlapping seismic reflections. These units have been correlated to eight age-constrained Esso/BHP foundation bores, which provide the chronostratigraphic framework for discussing the geological history of the shelf to upper slope and canyon-head environments (Figure 4). The stratigraphic units are: (i) Lower to Middle Pliocene (CN11, 4.80 – 3.70 Ma); (ii) Middle to Upper Pliocene (CN12, 3.70 – 1.95 Ma); (iii) uppermost Pliocene to lowermost Pleistocene (CN13a, 1.95 – 1.72 Ma); (iv) Lower Pleistocene (CN13b, 1.72 – 0.95 Ma); (v) Middle Pleistocene (CN14a, 0.95 – 0.47 Ma); and (vi) Upper Pleistocene and Holocene (CN14b and CN15, 0.47 – 0 Ma).

Figure 5 Interpreted and uninterpreted FR11/98 sparker line L22, which runs roughly along-strike (shore-parallel) across the inner and middle shelves. Nearby Esso/BHP foundation bores are structurally projected onto line, and Pleistocene magnetic fluvial channels located. See Figure 2 for traverse location.

Figure 6 Interpreted and uninterpreted FR11/98 sparker lines (a) L21 and (b) L14, which run downdip from shoreface to upper slope. Nearby Esso/BHP foundation bores are structurally projected onto lines, and Pleistocene magnetic fluvial channels located. See Figure 2 for traverse location.

Figure 7 Interpreted and uninterpreted FR11/98 sparker lines (a) L15, middle shelf along-strike transect and (b) L13, L2 and L20, shelf-break and canyon-heads transect. Nearby Esso/BHP foundation bores are structurally projected on to line L15. See Figure 2 for traverse location.

Figure 8 Interpreted and uninterpreted FR11/98 sparker lines (a) L19 and (b) the northern portion of L07, both downdip projections that profile the shelf-break. See Figure 2 for traverse location.

The offshore Gippsland Basin seafloor can be divided into shelf, slope and canyons based on water depth and seafloor morphology. The present-day shelf is 40 km wide in the north of the basin and 120 km wide in the southwest ( Figure 2). The shelf dips seaward from the coastline at 0.08° to a break in slope that occurs at an average depth of 130 m (110 – 150 m range). The shelf is here subdivided, based on water depth, into inner shelf (0 – 50 m), middle shelf (50 – 100 m) and outer shelf (100 – 130 m). Subdivisions of the slope are defined by changes in gradient, which include an upper slope (130 – 600 m, average gradient 1.1°), middle slope (600 – 1750 m, average gradient 4.5°) and lower slope (1750 – 4000 m, average gradient 2.5°). Submarine canyons are defined here as regions of the seafloor that have a channelled morphology with a minimum trough to shoulder height of 150 m and an axial length >10 km.

Quaternary shelf sediments have aggraded to a maximum of 120 – 150 m ( Figure 6) at an averaged compacted deposition rate of 62 – 77 mm/ky, which is comparable to rates estimated by Holdgate and Gallagher (1997) and Bernecker et al. (1997) for the Tertiary Gippsland shelf. Yet, the shelf-break has prograded some 10 km since the latest Pliocene (>5 m/ky).

Inner shelf (0 – 50 m) to middle shelf (50 – 100 m)

Shallow seismic reflections (<500 ms TWT) from the inner and middle shelves (L14, L21, L22 in Figure 2) have been correlated (Holdgate et al. 2003) to Lower Pliocene to Holocene carbonate sediments drilled in foundation bores Barracouta 1, Snapper, Marlin, Tuna and Halibut (Figures 2, 4). The oldest sampled sediment consists of Lower to Middle Pliocene outer shelf and upper slope silty marl, which coarsens upwards to middle shelf quartz – carbonate sand (Gallagher et al. 2003). Upper Pliocene to lowermost Pleistocene sediments consist of muddy fine quartz – carbonate sand and calcarenitic limestone of the upper slope to middle shelf. Lower to Middle Pleistocene sediments are middle shelf calcarenitic limestone, and inner shelf to marginal marine quartz – carbonate sand, which is often micaceous and carbonaceous.

Pliocene to Holocene sediments are conformable over the middle shelf, producing subparallel seismic reflections, with a general strike that follows the modern coastline. Sediments dip at 0.09° towards the centre of the basin. However, beneath much of the inner shelf and shoreface, Lower Pliocene reflections are subtly folded and unconformably overlain by younger sediments, revealing surprisingly recent tectonic activity. Additionally, many shelf sediments are incised by lowstand Pleistocene fluvial channels (Holdgate et al. 2003) and record a history of sea-level falls.

Folding beneath the shelf is observed at the northern end of line L22 ( Figure 5) and at the western end of lines L14 and L21 (Figure 6). Beneath the northern inner shelf, a growth anticline has developed in Lower Pliocene sediments (Figure 5). The structure on seismic line L22 has a maximum dip on the steeper southern limb of 0.9° in lower Middle Pliocene sediments. Uppermost Middle Pliocene seismic reflections unconformably truncate this fold, creating a discordant angle of 0.8°, and restrict the timing of folding to the Middle Pliocene (CN12, 3.70 – 1.95 Ma). Subhorizontal uppermost Pliocene to Middle Pleistocene sediments conformably overlie the fold axis and downlap toward the centre of the basin. Subsequent transgression during the Late Pleistocene and/or Holocene has deposited a veneer disconformably over Upper Pliocene to lowermost Pleistocene sediments. It is possible that limited growth of this structure continued until recently and caused the downlap of Upper Pliocene to Middle Pleistocene sediments. However, it is more likely that the structure ceased growing in the late Middle Pliocene and that the downlap of sediments was caused by lower average sea-levels, as Northern Hemisphere glaciations intensified (Prell et al. 1986; Haq et al. 1988).

On seismic lines L14 and L21 beneath the central inner shelf, folding consists predominantly of very open structures in Lower Pliocene to Lower Pleistocene sediments ( Figure 6a, b). On average, these sediments tilt basinward and are erosionally truncated above by either Upper Pleistocene fluvial channel sediments (Figure 6a), or subhorizontal Upper Pleistocene inner shelf sediments (Figure 6b). Fanning of Upper Pliocene to Middle Pleistocene sediments off anticlines/monoclines (Figure 6a) and thinning of sediments towards the east indicate that folding has occurred gradually from the Late Pliocene to at least the Middle Pleistocene. One structure, referred to as the Tarwhine structure by Dickinson et al. (2001) (Figure 6b), deforms subparallel Lower Pliocene seismic reflections, indicating that deformation was post-Early Pliocene (<3.70 Ma). This age is much more recent than inferred by Dickinson et al. (2001), who suggested that deformation occurred in the Miocene. To the east of the Tarwhine Anticline (below PC07, Figure 6b), a syncline is visible in seismic reflections down to the end of the data at 350 ms TWT. However, this is likely to be an artefact of data acquisition, as it coincides with changes in boat speed and direction (Figure 2).

Across all but some localised inner shelf regions, subhorizontal Upper Pleistocene to Holocene seismic reflections are produced by a conformable veneer of coarse-grained quartz – carbonate sand that onlaps the entire shelf. An exception is where up to 40 m of chaotic reflections, corresponding with Pleistocene fluvial channelling, are preserved in local inner shelf regions ( Figure 3). Conformable marine sediments thicken to the east, reaching their maximum thickness over the shelf-break and upper slope.

Outer shelf (100 – 130 m) to upper slope (130 – 600 m)

Beneath most of the outer shelf, sediment reflections are conformable and subhorizontal, thickening slightly as they approach and prograde over the upper slope ( Figure 6). Sampled sediments consist of Middle to Upper Pliocene calcarenite and muddy, fine, quartz – carbonate sand of the outer shelf (Gallagher et al. 2003); Lower to Middle Pleistocene calcarenite of the outer to middle shelf, and coarse-grained, shelly, quartz – carbonate sand of the middle to inner shelf; and Upper Pleistocene to Holocene coarse-grained, shelly, quartz – carbonate sand (Holdgate et al. 2003).

In regions below the outer shelf and upper slope, seismic reflections adjacent to canyon heads become increasingly complicated by channels and cut-and-fill structures. The most landward extension of canyon-head channelling is observed on the along-strike shallow seismic line L15 ( Figure 7a). Here, a large channel filled with Pliocene to Holocene marine sediments is observed and appears to trend northwest – southeast from Halibut and downdip to the central shelf-break near PC25 (Figure 2) between seismic lines L15 and L20. Bathymetric data suggest that a broad shallow depression in the modern seafloor directly overlies this channel. During the Middle Pliocene, this channel was at least 150 m deep and 30 km across. Lower Pliocene seismic reflections in this channel extend well below the maximum depth of sparker data at 450 ms TWT (>300 m below seafloor), suggesting that the channel, in its original form, was a much older and larger feature.

Three sediment packages can be identified in the channel on line L15. The first succession consists of Lower to Upper Pliocene calcarenite, characterised by subparallel seismic reflections that mantle the palaeochannel. The second succession consists of lowermost to Upper Pleistocene calcarenite, quartz – carbonate sand, and micaceous and carbonaceous carbonate – quartz sand lenses (Holdgate et al. 2003), and is characterised by extensive erosion. During this phase of erosion, sub-channels developed within the main palaeochannel. Seismic reflections in these sub-channels are asymmetrical, and the sub-channels migrated laterally. The third succession consists of uppermost Pleistocene to Holocene coarse-grained quartz – carbonate sand that has subparallel seismic reflections ( Figure 7a).

Lower to Upper Pliocene seismic reflections define a single broad U-shaped and asymmetrical channel with subparallel bedding. The northern wall of the Upper Pliocene channel is significantly higher and steeper than the southern wall, suggesting that sediments mantle a pre-existing structure or morphology ( Figure 7a). A comparison is made here to a meandering fluvial channel, where the higher and steeper bank is typically found on the outer bend of the channel and is prone to erosion, while the less-steep bank is generally the inside bend and associated with point-bar deposits. The same analogy has been used to explain high-velocity downlapping seismic reflections in Miocene Seaspray Group sediments (Wallace et al. 2002), which were also attributed to laterally migrating canyons. This would imply that the pre-Pliocene channel migrated laterally across the outer shelf and upper slope in a northward direction, eroding the northern bank while depositing sediment on the southern bank.

The oldest observed erosional event in L15 ( Figure 7a) occurs in uppermost Pliocene sediment and erodes Middle to Upper Pliocene sediments on the northern wall of the main channel. On the southern wall, sediments conformably mantle the palaeochannel, but onlap unconformably over Middle to Upper Pliocene sediments on the northern wall, confirming a northward lateral migration direction as inferred from underlying morphology.

The onset of the Pleistocene coincides with major erosional events in the main Pliocene channel on the outer shelf, cutting three significant sub-channels into Upper Pliocene sediments ( Figure 7a). The biggest sub-channel approaches 180 m in depth (200 ms TWT) and is 15 km across. Seismic reflections show that deposition within the sub-channels is strongly asymmetrical, causing channels to migrate laterally. During most of this phase, sediments predominantly were built out from the southern sub-channel walls and downlapped unconformably onto the sub-channel floors (1 and 3 on Figure 7a). This caused sub-channels to migrate laterally across the shelf in a northeast direction. Immediately following the initial erosional event in the Early Pleistocene and again in the late Middle Pleistocene (2 and 4 on Figure 7a) migration direction reversed. During these times, sediments predominantly built out from the northern sub-channel walls, causing channels to migrate to the southwest. By the Late Pleistocene, the three sub-channels were filled, leaving only a broad depression in the seafloor above the main palaeochannel.

Canyon heads

Sparker data across the canyon heads are generally of poorer quality than across the shelf, as the deeper water and greater slope attenuate the acoustic signal. However, many regions have been successfully imaged, revealing significant sedimentary phases during the late Neogene. In particular, onset of widespread erosion across the canyon heads occurred from the Plio-Pleistocene boundary to the present day ( Figure 7b).

Seismic data across the canyon heads show that Lower to Middle Pliocene strata are subparallel and mantle older broad U-shaped channel morphologies. Towards the end of the Pliocene, sediments progressively thickened over highs between channels and thinned in troughs, gradually steepening channel walls and narrowing troughs, while U-shaped morphologies are maintained ( Figure 7b). A rapid transformation from net sedimentation to net erosion at the Plio-Pleistocene boundary cut deep V-shaped incisions within the pre-existing U-shaped channels, and exposed Middle Pliocene strata at the seafloor. Seismic data reveal that subsequent sedimentation has been extremely localised and limited to the upper reaches of canyon heads, while deeper regions appear to have remained erosional to the present day. Two seismic facies are recognised in the limited Pleistocene strata: (i) sediment packages with coherent reflections that prograde and generally downlap from the shelf-break and across channel floors; and (ii) complex packages that consist of discordant cut-and-fill or slump-related deposits, which are commonly observed in channel troughs (Figures 7a, 8a). The absence of Pleistocene strata is particularly apparent where the seafloor topography steepens to gradients >20°. In such regions, Pliocene seismic reflections are truncated at the seafloor, implying that channels are actively eroding at present (Figure 7).

Sparker line L7 traverses the southern slope canyon heads along the modern shelf-break ( Figure 2) where a close spatial relationship between pre-existing (Quaternary?) buried channel structures and modern channels is observed. Although the data are of poorer quality, at most localities modern canyon-head channels overlie buried channels, suggesting that channels have remained active and open in some form since at least the Pliocene, or have developed within pre-existing channels.

Buried magnetic fluvial channels

Holdgate et al. (2003) showed that the origins of these magnetic palaeochannels are the modern Tambo and Nicholson Rivers, Ironstone Creek, and Snowy River systems, and they are interpreted as lowstand fluvial features. Although the source of their high magnetic signature is unknown, Holdgate et al. (2003) suggested that ferruginous cementation within channels may cause such magnetism. Additionally, they identified a number of ‘seismic smudges’ on FR11/98 sparker lines (L21, L22) which were interpreted as fluvial channels. The present study successfully correlates 19 of these seismic features identified on the FR11/98 survey lines to magnetic fluvial channels observed in the aeromagnetic image ( Figures 3, 5, 6).

Buried magnetic fluvial channels occur across the inner and middle shelves out to the 80 m isobath, where they appear as U-shaped features ( Figure 9) that interrupt the otherwise planar Plio-Pleistocene shelf strata in the upper 50 ms TWT (∼45 m below sea bed) (Figures 3, 5, 6). Channels range in width from 280 to 1400 m and have troughs from 8 to 27 m deep. Channels trend southeast following the approximate axis of the basin across the middle and outer shelves, where they amalgamate near the head of the Anemone Canyon and the northwest corner of the present-day shelf-break (Figure 3). Seismic stratigraphic correlations suggest that the oldest fluvial channels developed in the middle of the Early Pleistocene (Figures 3, 5, channel 11) but that the majority of preserved channels formed during the Middle and Late Pleistocene, which is consistent with the interpretations of Holdgate et al. (2003). The amalgamation of the fluvial channels at canyon heads suggests that direct fluvial – canyon connections occurred during at least the larger Pleistocene regressions.

Figure 9 Example of a ‘seismic smudge’ as identified by Holdgate et al. (2003). This particular seismic smudge correlates with magnetic fluvial channel 9 ( Figure 3) on FR11/98 line L21 and is Late Pleistocene in age.

Pliocene sedimentation rates

Sedimentation rates for the Early to Middle Pliocene are estimated from the Barracouta foundation bore where a shallowing-upward sequence of outer shelf marl to middle shelf quartzose carbonate sand is recorded (Gallagher et al. 2003; Holdgate et al. 2003). Minimum and maximum sedimentation rates of 100 and 200 mm/ky are calculated ( Figures 4, 10). During the Middle to Late Pliocene, sedimentation rates ranged from negative values on the innermost shelf, to 45 mm/ky over much of the middle shelf, and reached 75 mm/ky on canyon-free regions of the outer shelf to upper slope (Figures 4 –  7, 10). However, the highest rates are recorded from time/depth converted seismic data in upper slope canyon-head environments on Line L20 (Figure 7b), reaching 180 mm/ky (Figure 10).

Figure 10 Maximum (compacted) sedimentation rates and seismic characteristics for the inner shelf to upper slope and canyon-heads regions of the Gippsland Basin, Pliocene to Holocene geology, as expressed in the FR11/98 seismic survey. A – F refer to stratigraphy in Figures 4 –  8.

Latest Pliocene to earliest Pleistocene sedimentation rates

During the latest Pliocene and earliest Pleistocene, sedimentation rates more than doubled across the shelf, upper slope and canyon heads. Rates of 85 mm/ky on the inner shelf, increasing to 400 mm/ky on the outer shelf and upper slope, were recorded from foundation bores ( Figure 4) and time/depth-converted seismic data (Figures 5 –  9). Maximum depositional rates occurred at outer shelf uppermost canyon-head regions near the Mackerel foundation bore and line L15 (Figure 7a) where locally, they exceeded 650 mm/ky (Figure 10).

Pleistocene and Holocene sedimentation rates

Sedimentation across much of the shelf during the Early Pleistocene was episodic because large eustatic fluctuations frequently exposed the shelf (Holdgate et al. 2003). Consequently, averaged sedimentation rates were much lower than those at the Plio-Pleistocene transition period, with rates ranging from negative values across parts of the inner shelf, to 85 mm/ky on the outer shelf ( Figures 4 –  7). Sedimentation rates in the canyon heads dropped significantly, resuming rates comparable to those recorded for the Middle Pliocene. Maximum sedimentation rates of 65 mm/ky were obtained in channels of the outer shelf (Figure 7a), and 195 mm/ky in isolated channels in the lower canyon-head regions (Figure 7b, L24).

In the Middle Pleistocene, shelf sedimentation rates reached their lowest recorded levels. However, in the uppermost canyon-head channels on the central outer shelf ( Figure 7a), sedimentation rates were high, depositing coarse-grained micaceous and carbonaceous quartz-rich carbonate sand of marginal marine and inner shelf environment (Holdgate et al. 2003). Here, maximum rates approached 400 mm/ky (Figure 7a). In deeper regions of the canyon heads, deposition was localised with net erosion occurring in most areas to the present day.

During the Late Pleistocene and Holocene, marine transgressions allowed sedimentation to resume over much of the shelf, particularly on the inner shelf where a Holocene highstand sediment wedge has accumulated at 77 mm/ky. Sedimentation rates elsewhere on the shelf and upper slope have been the lowest since the Middle Pliocene ( Figures 4 –  8, 10). Deposition in canyon heads was extremely localised, with the majority of canyon-head regions undergoing net erosion.

Discussion

Shepard (1981) speculated that modern submarine canyons were likely to be the product of a number of processes and, of more importance, he suggested that the long-time persistence of such processes was of great importance in canyon formation. Major canyon structures in the carbonate Seaspray Group first develop during the Oligocene (Conolly 1968) and Middle Miocene (Maung & Cadman 1992; Bernecker et al. 1997; Feary & Loutit 1998; Holdgate et al. 2000). Yet, it has been suggested that the Bass Canyon (sensu stricto) may have structural origins that relate to the breakup of Gondwana, where it may have acted as a conduit for clastic sediments in the Late Cretaceous (Hill et al. 1998). Recent detailed 3D seismic studies of Neogene buried canyons (Leach & Wallace 2001; Bertoni & Cartwright 2005) have showed that even comparatively small, often slope-confined canyons, may span considerable lengths of time in the geological record, responding to, and preserving, long-term oceanographic conditions. The morphology of much of the present-day Bass Canyon system has been shown here to date from the latest Pliocene, although these modern canyons are themselves nestled in pre-existing Pliocene or older canyons.

Timing and cause of the present-day canyon-head morphologies

The RV Franklin FR11/98 shallow seismic survey across the outer shelf, upper slope and canyon heads shows that widespread deposition occurred throughout the Pliocene. The onset of deep V-shaped erosion in canyon heads occurred in the Early Pleistocene, at approximately the same time that Northern Hemisphere glaciations forced eustatic regressions. This would suggest that erosion of the canyon heads was related to the regressive and lowstand systems tracts, which is in agreement with the traditional sequence-stratigraphic model (Posamentier et al. 1992). However, leading up to this event, from as early as the Middle Pliocene, seismic data show sediments progressively thickening over ridges and thinning on channel floors towards the end of the Pliocene, causing canyon-head channels to narrow and walls to steepen ( Figures 7b, 10). Additionally, sedimentation rates across the entire shelf and upper slope increased two to threefold during the latest Pliocene and earliest Pleistocene (Figure 10). It is likely that the combined effect resulted in slope instability, which may have led to widespread mass-wasting and erosion of the canyon heads. If so, then the Pleistocene regressions may just have amplified an inevitable process of erosion driven by high sedimentation rates on a relatively steep slope.

The cause of increased sedimentation is unclear. Dickinson et al. (2001) argued that there has been significant Late Neogene deformation in all the southeastern Australian basins, with up to 1 km of exhumation in places, and this study identifies the effects of minor compressional tectonics on Pleistocene sediments. Increased erosion of exhumed landscapes could provide a source of increased sedimentation. Several siliciclastic facies are present in Pleistocene shelf deposits (Holdgate et al. 2003), although foraminiferal studies (Mays 2001; Smith et al. 2001) show that these sediments were deposited in inner shelf to euryhaline environments, and therefore do not represent an influx of terrigenous material to the basin. Instead, they represent lowstand depositional environments of laterally migrating sediments, which are analogous to the present-day Gippsland inner shelf to coastal lake environments that trap the majority of terrigenous sediment (Li et al. 2005).

The correlation of lowstand magnetic fluvial channels to biostratigraphically age-constrained seismic packages ( Figures 3, 6) suggests that potential connections of fluvial systems to canyon heads post-date the initial phase of erosion of the canyon heads, which occurred in the latest Pliocene to earliest Pleistocene (Figure 7). However, fluvial systems clearly did reach the canyon heads during Middle and Late Pleistocene lowstands (Figure 3). Nevertheless, direct fluvial – canyon connections must have significantly altered sedimentary processes at the canyon heads and upper slope. Despite this, there is no evidence of increased terrigenous sediments to the deep-water environment during Late Pleistocene glacial periods (J. K. Mitchell, G. R. Holdgate, M. W. Wallace and S. J. Gallagher in press).

Therefore, additional sedimentation is most likely to be derived from high carbonate production, predominantly of shelf molluscs and of outer shelf to upper slope bryozoan communities, and accordingly is likely to reflect favourable highstand oceanic conditions. Planktonic foraminiferal proxy data from the Gippsland Basin (Gallagher et al. 2003) show that cool conditions prevailed during the earliest Pliocene, followed by relatively stable warmer marine conditions. Then, during the Middle and Late Pliocene, the abundances of cool and warm water taxa fluctuated, as did the global δ¹⁸O record. Increases in the abundances of cool-water taxa are interpreted as upwelling events occurring along the outer shelf and upper slope of the Gippsland Basin, caused by northward migration of the Subtropical Convergence ( Figure 1) during the Late Pliocene glaciations (Gallagher et al. 2003). The mixing of water masses, such as by upwelling events, or from the formation of eddies such as those associated with the East Australian Current (Marchesiello & Middleton 2000), may cause nutrient enrichment of coastal waters, and is therefore likely to promote biogenic productivity and increased carbonate production. Late Pliocene cooling is also likely to have intensified local currents such as the Bass Cascade (Figure 1), providing filter-feeding organisms, such as molluscs and bryozoans, with an increased food supply.

Intensification of density currents, such as the Bass Cascade, may have been responsible for causing differential sedimentation across the canyon heads and for steepening channel walls. Such gravity-driven currents would follow pre-existing channels, prevent sedimentation within channel troughs, but allow sedimentation to continue on highs between channels. Additionally, density currents flowing off the shelf could trigger erosive sediment gravity flows and mass-wasting at the shelf-break. Similar high-salinity density currents outflow seasonally from the Spencer Gulf, where they cascade over the shelf-break (Lennon et al. 1987) and are likely to enter the Murray Canyons.

The lateral migration of shelf marine channels and canyon heads may also be related to prevailing oceanic conditions. In the Otway Basin, Leach and Wallace (2001) observed that Miocene canyons migrated laterally to the west, while Pliocene canyons migrated to the east. They suggested that this migration was likely to be caused by a prevailing oceanographic current, which switched direction at around the Miocene/Pliocene boundary. The alternation of northeast and southwest lateral migration of Early and Middle Pleistocene sub-channels across the Gippsland shelf and upper slope is also likely to be caused by prevailing oceanographic currents. The predominant northeast lateral migration, observed in Lower and Middle Pleistocene channel-fill deposits ( Figure 7a), might be caused by a prevailing northeast-moving current, such as the Bass Cascade. Southwest lateral migration might be caused by the East Australian Current extending further south during periods of warmer oceanic conditions (Marchesiello & Middleton 2000). A northerly (compared to today) advanced Subtropical Convergence is likely to restrict the southern limits of the East Australian Current and favour a prevailing Bass Cascade current in the Gippsland Basin. Therefore, northeast lateral channel migration in the Gippsland Basin may relate to cooler oceanic conditions and a northerly advanced Subtropical Convergence.

Finally, the timing of structural activity in the Gippsland Basin may have enhanced Pleistocene erosion. Many of the low-amplitude fold structures identified beneath the inner shelf ( Figure 6) developed during the Late Pliocene through Pleistocene. It is likely that continental margin slopes in relatively inactive tectonic settings, such as the Gippsland Basin, may reach steeper gradients compared to active margins. Thus, a minor tectonic movement (earthquake) following a quiescent period may potentially induce considerable slope failure. The timing of such movements, particularly following a period of higher than normal sedimentation, may well have induced erosion at the canyon heads without a eustatic fall.

High sedimentation rates as a viable cause of canyoning in the Tertiary

Although the cause of the major Middle Miocene canyoning (Maung & Cadman 1992; Holdgate et al. 2000) is not known, it appears that elevated sedimentation rates occurred subsequent to and during erosion in both the onshore (Holdgate and Gallagher 1997) and offshore record (Bernecker et al. 1997). As such, elevated sedimentation rates may well have contributed to Middle Miocene canyon erosion.

Conclusions

This study develops a Late Neogene geological history of the shelf, upper slope and canyon-head environments of the Gippsland Basin, using shallow (500 ms TWT) FR11/98 seismic stratigraphy tied to eight biostratigraphically age-constrained Esso/BHP foundation bores (from Holdgate et al. 2003). Interpretation of shelf seismic stratigraphy reveals that minor tectonic folding occurred on the northern inner shelf during the Middle Pliocene (CN12, 3.70 – 1.95 Ma), and on the central inner/middle shelf from the Late Pliocene to at least the Middle Pleistocene (CN13 – CN14a, 1.95 – 0.47 Ma). Widespread erosion across the outer shelf and upper slope canyon heads occurred during the earliest Pleistocene (CN13a, 1.95 – 1.72 Ma) cutting deep V-shaped incisions into pre-existing U-shaped canyon heads. We argue that Quaternary erosion was not necessarily caused by lowstand exposure of the shelf-break but was induced primarily by mass-failure of oversteepened sediments, caused by: (i) the progressive steepening of canyon channels through differential sedimentation during the Middle to Late Pliocene; and (ii) a period of greatly elevated (carbonate) sedimentation rates, which occurred during the latest Pliocene to earliest Pleistocene, under highstand eustatic conditions.

Both contributing factors are likely to have been promoted by an intensification of the local Bass Cascade density current in shallowing waters towards the end of the Pliocene. The density current can only operate while the Bass Strait is inundated during highstand sea-levels and would be strongest when the water is shallow. Subsequent rapid infilling of the outer shelf canyon-head channels occurred during the Middle Pleistocene (CN14a, 0.95 – 0.47 Ma). The architecture of channel-fill sediments shows strong lateral migration, predominantly in a northeast direction, but reversed periodically to the southwest. Northeast lateral channel migration is likely to develop when the Bass Cascade current prevails over the southwest-moving East Australian Current, during cool oceanic conditions and with a northerly advanced Subtropical Convergence. Quaternary upper slope canyon-head environments have remained largely non-depositional and preserve deep V-shaped morphologies that often expose Pliocene strata (beneath a Holocene veneer) on the modern ocean floor.

Acknowledgements

We wish to thank Geoscience Australia for providing the bathymetry data for the Gippsland Basin and southeast Australia, and the Department of Industry, Technology and Environment (Geological Survey of Victoria) for permission to use the airborne magnetic image of the Gippsland Basin. The University of Sydney's Ocean Drilling Program is thanked for providing and running the seismic-acquisition equipment (particular thanks to Jock Keene and Dave Mitchell). The RV Franklin cruise was partially funded by ARC Research Grant No. A39803024. Reviewers Neville Exon and Barry Bradshaw are thanked for their helpful comments and suggestions.