The Recent foraminifera and facies of the Bass Canyon: a temperate submarine canyon in Gippsland, Australia

This study describes the foraminifera and facies of a large submarine canyon: the Bass Canyon, in the Gippsland Basin off the coast of southeastern Australia. The study incorporates facies analyses and interpretations of three types of foraminiferal distributional data: forms alive at time of collection, recently dead forms and relict forms. Four principle biofacies types occur: (1) middle shelf to shelf-break carbonate sand; (2) oxic upper to middle bathyal carbonate sand and gravel, with abundant bryozoans; (3) reduced oxic middle bathyal carbonate sand and gravel and (4) lower bathyal oxic muddy sand to Globigerina Ooze. Correspondence Analysis of the 61 parameters (percentage abundance of foraminifera and % carbonate) in 36 samples yielded a clear depth-related pattern, although other related parameters such as dissolved oxygen and substrate also exert control on the foraminiferal assemblages. Relict foraminifera are restricted to shelfal depths, shallower than 145 m. This pattern is similar to other shelf regions in Australia, where shelf areas were exposed during the Last Glacial Maximum, reworking shelf facies shallower than 150 m. The distribution of living foraminifera is similar to the distribution of the total assemblage, suggesting that the region has not been significantly mixed by wave, slump or bioturbation processes. The majority of the modern Bass Canyon foraminiferal assemblages are cosmopolitan species, with few (semi-)endemic taxa that are mostly restricted to the shelf. These modern deeper-living forms are more conservative since they evolved in relatively lower stress eutrophic environments than their shallower oligotrophic dwelling contemporaries. The foraminiferal and facies analogues of this study on the Bass Canyon may be used as a modern palaeoenvironmental analyses of the Gippsland and Otway Neogene sedimentary deep-sea successions. This will lead ultimately to a better understanding of the evolution of the basins in southeastern Australia, in an area influenced by the Southern Ocean during the Cenozoic.


INTRODUCTION
The Bass Canyon is an east-south-east trending submarine canyon, located in the Gippsland Basin off the coast of southeast Australia (Fig. 1). It is at the confluence of several smaller canyons from the continental shelf and is approximately 10 km wide and 80 km long and extends in an approximate straight line. At 4000 m water depth, the Bass Canyon exits the continental slope and passes onto the abyssal plain, terminating as the Bass Fan.
Summaries of modern marine sediment distribution patterns have been completed in the region by Jones & Davies (1983) and Smith et al. (2001), although there have been no previous detailed studies on the foraminiferal biofacies of the Bass Canyon. The modern foraminiferal distributions are better known in other parts of Australia and New Zealand, and have concentrated on the temperate shelf to marginal marine regions, and have been used as analogues for palaeoenvironmental interpretation of the neritic, Cenozoic, cool water carbonate successions of the Otway and Gippsland Basin (Parr, 1945;Collins, 1953Collins, , 1973Albani, 1968;Johnson & Albani, 1973;Apthorpe, 1980;Cann & Gostin, 1985;Cann et al., 1988;Gallagher & Holdgate, 1996Li et al., 1996aLi et al., , b, 1999Holdgate & Gallagher, 1997;Hayward et al., 1999;Gallagher et al., 1999Gallagher et al., , 2001. It is intended that the foraminiferal and facies analogues of this study on the Bass Canyon will greatly improve palaeo-Three types of distributional data were collected for the foraminiferal assemblage in each sample: forms interpreted to have been alive at time of collection (=live forms) with Rose Bengal solution; forms with no evident protoplasm at time of collection nor any visible diagenetic alteration (=recently dead forms) and relict forms .
For descriptive purposes the fauna and facies of the Bass Canyon will be described in terms of the four dominant faunal types, shelfal, upper to middle bathyal (oxic), bathyal (reduced oxic) and middle to lower bathyal (oxic). The reduced oxic samples are those that occur within the Oxygen Minimum Zone (see Fig. 2) and contain characteristic foraminifera of reduced   Gupta & Machain-Castillo, 1993).
The shelfal samples include: five samples from the northeast section of the area of study from depths between 88 m to 242 m (transect C) at the top of the Bass Canyon; two from the south-east section of the area from depths of 132 m to 300 m (transect D) at the top of a smaller feeder canyon of the Bass Canyon; and four samples from the northern section of the area of study immediately above the steeply dipping canyon wall (transect E and adjacent samples) (see Fig. 1).
The upper-middle bathyal (oxic) samples include: three samples from transect C; one from transect D; and five from transect E (along with adjacent samples), from depths between 366 and 1137 m.
The nine bathyal (reduced oxic) samples studied come from depths between 995 and 2980 m. This zone is classified as 'reduced oxic' due to the high percentage of Rhabdammina spp. and Bulimina marginata. This zone also corresponds with the oxygen minimum zone of the data from Levitus (1982) (Fig. 2).
The seven middle to lower bathyal (oxic) samples come from depths between 1025 and 3931 m, and it is suggested increased dissolved oxygen conditions exist compared to the 'reduced oxic' samples described earlier due to a decrease in abundance of the reduced oxygen indicators, such as Rhabdammina spp. and Bulimina spp. The Levitus (1982) data also show an increase in dissolved oxygen in the region at depths greater than 2000 m (Fig. 2).
In additional to the samples collected aboard the R/V Franklin, samples collected by David Taylor in 1968 (the 'SPB' samples), and samples taken aboard the Kimbla in 1973 (the 'K7/73' samples) have been used to supplement the results.
Percentages of CaCO 3 were calculated for each sample by dissolving approximately 0.3 g of the crushed sample using 50% HCl and measuring the amount of CO 2 gas released (Carver, 1971). Facies analysis was completed by analyses of the dried unprocessed sample, which was split and then viewed under a reflected light microscope.

REGIONAL OCEANOGRAPHIC SETTING
The Gippsland Basin covers 56,000 km 2 (Smith, 1982), of which two-thirds is offshore, and lies between 38(S and 41(S off the coast of southeast Australia (see Fig. 1). The curved morphology of its tapering shelf break is principally determined by the occurrence of a large submarine canyon that reaches a maximum depth of 4000 m (the Bass Canyon). Most of the basin lies in Bass Strait, a shallow sea between mainland Australia and Tasmania in the west, and the Tasman Sea in the east. The slope of the Gippsland shelf ranges from 0.06( in the inner to middle shelf (0-100 m) to 0.3( in the outer shelf (100-200 m). Slopes up to 6( occur at the shelf break at the head of the Bass Canyon (deeper than 200 m), it decreases to 0.6( approaching the bottom of the Bass Canyon in depths of w3500 m. The Gippsland region (Fig. 3) lies 500 km north of the Subtropical Convergence (STC) ocean front, which occurs around 45(S, and marks the boundary between the subantarctic and temperate oceanic regions. The STC is identified by the summer 15(C isotherm and winter 10(C isotherm outcropping on the surface (Martinez, 1994). The contact between the temperate Tasman Current and the subtropical East Australian Current, lies at about 30(S, and is known as the Subtropical Divergence (Martinez, 1994).
The unstable, predominantly wind-driven East Australian Current (EAC) is the western boundary current of the South West Pacific (Hamon, 1965;Tomczak & Godfrey, 1994). Between 27(S and 32(S the current flows southward along the shelf break. South of 32(S however, the current forms numerous anticyclonic eddies (Hamon, 1965;Davies, 1979).
During winter a dense cold saline water mass (35.7‰) leaves the Bass Strait and travels northward along the shelf and sinks at the head of the Bass Canyon. Part of this water mass also travels east across the shelf and sinks into the Tasman Sea just beyond the position of the shelf break between 300 and 400 m (Godfrey et al., 1980;Tomczak, 1985). This is the "Bass Strait Cascade", which is a density current that travels at speeds of up to 0.5 m/sec below 200 m. It is, therefore, an important local control on the facies at the shelf break at the head of the Bass Canyon.
Mean sea surface temperatures for offshore Gippsland vary from 12(C to 16(C in winter and 18(C to 20(C in summer (Levitus, 1982), thus it is a temperate or cool-water region (sensu James, 1997). The 15(C thermocline is at a depth of around 100 m and the 10(C thermocline is at 400 m, where seasonal changes in temperature occur deeper than 200 m depth (Knauss, 1997).
Mean annual salinities across the Gippsland shelf are stenohaline and range from 35.2‰ at 200 m depth to 35.5‰ at 50 m (Levitus, 1982).
Data from Levitus (1982), as shown in Figure 2, records an oxygen minimum zone of oxic condition (3.65 ml/l) near the seabed between 1300 and 1800 m water depth on the west side of the 38.5(S latitude. Dissolved oxygen increases below this depth in the deeper water of the east.

REGIONAL SEDIMENTOLOGICAL SETTING
Southeastern Australia is a typical cool-water carbonate depositional realm. Figure 4 shows a contour map of percentage of Fig. 3. The oceanographic setting of the Gippsland area and shelfal areas of southern Australia, adapted from Martinez (1994) and James et al. (1999). calcium carbonate of the shelf to bathyal realm in southeastern Australia. This map uses data obtained from this study, along with GA (Geoscience Australia) data derived from the Southern Surveyor (Harris et al., 2000) and data from Jones & Davies (1983).
The Gippsland Lakes and the inner shelf yields between 0 and 40% calcium carbonate, where values are increasing with depth ( Fig. 4). This is due to the abundance of siliciclastic sediments in the shallow facies and the relative lack of calcium carbonate producing organisms. This carbonate low stretches along the coast caused by the longshore drift distribution of siliciclastic sediments. From the middle to outer shelf, the calcium carbonate percentage increases with depth, due to dilution of clastic sediments with an increasingly diverse bioclastic assemblage. This is the cool-water carbonate 'factory', the region where particles of all grain sizes are 'born', either by crystallising as skeletons or precipitating directly out of the seawater (James & Kendall, 1992). The relatively low calcium carbonate values in the middle shelf, southeast of Lakes Entrance may be caused by the presence of remnant siliciclastics derived from palaeo-rivers, which meandered across the Pliocene to Pleistocene coastal plain during glacial period(s).
There are two clear calcium carbonate-rich areas due south of the Gippsland Lakes and north of the Bass Canyon (Fig. 4). These sediment-starved regions are densely populated by bryozoans with both living and fossil forms comprising most of the bioclasts in the carbonate 'factory'.
A carbonate high to the north of the Bass Canyon occurs near the outer shelf to upper slope transition. This bryozoan-rich facies is similar to that described by Wass et al. (1970), who observed that living bryozoa are most abundant at depths between 90 and 220 m in southern Australia. High carbonate values are also recorded from deeper, bathyal Globigerina ooze sediments (Fig. 4).
The carbonate low on the northern side of the Bass Canyon may be due to erosion of the steep canyon wall, possibly exposing and reworking parts of the underlying, carbonate poor, lower Tertiary Latrobe Group.
Carbonate values decrease in the deeper part of the canyon due to the abundance of clay and since the deeper samples are taken from below the carbonate compensation depth (CCD). The CCD is estimated to be around 3600 m based on the results of two stations: north-east of Sydney and south-east of Tasmania (Tilbrook, pers. comm.).
The percentage gravel, sand and mud size (Folk and Dunham classification) contour maps (Fig. 5) are derived from GA data from the Southern Surveyor cruise (Harris et al., 2000) and data published in Jones and Davies (1983). These percentages are based on bulk samples; they refer to grainsize percentages and do not distinguish between siliciclastic and bioclastic grains. Figure 5a shows a high percentage of gravel in the western section of the area of study. This is associated with the carbonate high shown in Figure 4 due to the presence of bryozoan-rich sediments. Most of the shelf and slope to the northeast of the Bass Canyon yields low (<5%) gravel percentages where this region is carbonate-rich (Fig. 4). Here the bioclasts in the sediments are sand sized or smaller, perhaps due to the Fig. 4. The percentage of carbonate in the Gippsland Basin, using data from , GA data from the Southern Surveyor (SS01/00) and data published in Jones & Davies (1983).  5. The percentage of a) gravel, b) sand and c) mud size sediments in the Gippsland Basin, using GA data from the Southern Surveyor (SS01/00) and data published in Jones & Davies (1983). instability of the relatively steep shelf/slope environment that did not allow bryozoans and other bioclast-producers to thrive. There is a lobe of gravel-rich sediments on the inner shelf towards the centre of the area of study, this is associated with low inner-shelfal carbonate percentages (Fig. 4) and abundant quartz grains, and may be a related to the reworking and transportation of terrestrial material. The lobe of gravel-rich sediments extending from the middle shelf to the Bass Canyon head may be associated with downslope transport of bioclasts.
The contour map of sand-sized siliciclastics and bioclasts (Fig. 5b) shows an inverse relationship to % gravel, where the highest percentages of sand-size sediments occur in the inner to middle shelf, replaced by mud in deeper facies. The high % sand values along the inner shelf are probably related to longshore drift and entrainment of the Ninety Mile Beach sediments. The high percentage sand values, which extends south from the coast along the 148(30#E latitude, matches the carbonate low in this region (Fig. 4) and is possibly related to remnant river sediments.
The contour map of % mud in southeastern Australia ( Fig.  5c) shows an increase in abundance with increased depth. The most abundant mud-sized sediments (>40%), occur at depths greater than 2000 m, where Globigerina ooze predominates, yielding 60 to 70% carbonate (Fig. 4). There is a lobe of calcareous mud stretching outward from Lakes Entrance. This corresponds with high shallow water percentage gravel values, where poorly sorted bioclasts in a muddy matrix accumulate, reflecting the overall low energy nature of the shelf.
The deeper shelf samples on transect C (Fig. 6) are dominated by rotaliids, with common C. mediocris, Hoeglundina elegans, Lenticulina spp., Parrellina imperatrix and U. bassensis. Miliolids comprise 6 to 15% of the benthic (live + dead) assemblages and textulariids make up between 6 and 9% of these assemblages. Up to 11% of the benthic (live + dead) assemblages were found to be alive at time of collection. Few relict forms occur.
The standardized diversity of the benthic (live + dead) assemblage in these samples is between 9 and 10. Most forms were interpreted to have been recently dead at time of collection, except for 8% live forms recorded at 132 m.
The shelfal transect E (Fig. 8) and adjacent samples are dominated by rotaliids, with abundant H. elegans, Lenticulina spp., P. imperatrix and C. mediocris. Miliolids account for up to 25% and textulariids up to 21% of the benthic (live + dead) assemblages. The Margalef diversity index for this section is between 10 and 13. Between 7 and 9% of the benthic (live + dead) fauna at 119 m (K7/73/19) and 223 m (K7/73/20) are interpreted to have been alive at the time of collection. Minor relict forms occur at 223 m.

Oxic upper to middle slope biofacies
Benthic foraminifera comprise up to 32% of the total foraminiferal assemblages in the upper to middle bathyal oxic zone, decreasing with depth. Parrellina imperatrix, Rosalina spp. and Uvigerina spp. are common in the shallower samples, while A. centroplax, C. mediocris, H. elegans, Lenticulina spp. and G. subglobosa occur between 366 and 1137 m (Figs 6,7 & 8). Foraminifera interpreted to have been alive at the time of collection were rare deeper than 600 m. No relict forms occurred in the oxic upper to middle slope biofacies. Faunal diversity was found to decrease with depth into the bathyal realm.

Oxygenated middle to lower slope
Benthic foraminifera increase in abundance from 2 to 17% with increasing depth in this zone. Typical rotaliids include Cibicidoides wuellerstorfi, Sphaeroidina bulloides, B. marginata and Gyroidinoides spp. Textulariids are abundant in the lower bathyal samples (3114, 3340 and 3931 m), dominated by Rhabdammina spp., Rhizammina spp., Reophax spp., Marsipella spp. and Hyperammina spp. The miliolids Pyrgo lucernula and Pyrgo serrata also occur in the deeper samples. Benthic diversity in this interval is low (Figs 6,7 & 8). Refer to Plates 1 and 2 for images of the most common species of benthic foraminifera mentioned above.

Shelfal biofacies
The neritic realm on the margin of the Bass Canyon is plankton poor in its shallowest parts (8%), with yields of up to 60% in the deeper outer shelf. Globorotalia inflata dominates the planktonic fauna with common Globigerina bulloides (Figs 6,7 & 8).

Oxic upper to middle slope biofacies
Planktonic foraminifera are abundant, increasing from 68 to 96% with depth. Globorotalia inflata dominates the planktonic fauna with lesser amounts of G. bulloides, Globorotalia hirsuta, Fig. 6. Transect C. Bivariate graphs show abundance and distribution of various taxa as percentage of the biococenosis, the live + dead assemblages. Diversity as shown is the Margalef diversity (see Appendix 1). The total abundance of planktonic foraminifera are calculated as percentages of the benthic (live + dead) and planktonic assemblages, with the relative percentages of Globigerina bulloides, Globorotalia inflata and other planktonics (which include all species of planktonic foraminifera other than G. bulloides and G. inflata) shown as percentages of the planktonic assemblages. The Live:Dead:Relict abundances are shown as part of the thanatocoenosis, the entire benthic assemblage (see Fig. 1 for location of samples shown in seabed profile).

Fig. 7.
Transect D. Bivariate graphs show abundance and distribution of various taxa as percentage of the biococenosis, the live + dead assemblages. Diversity as shown is the Margalef diversity (see Appendix 1). The total abundance of planktonic foraminifera are calculated as percentages of the benthic (live + dead) and planktonic assemblages, with the relative percentages of Globigerina bulloides, Globorotalia inflata and other planktonics (which include all species of planktonic foraminifera other than G. bulloides and G. inflata) shown as percentages of the planktonic assemblages. The Live:Dead:Relict abundances are shown as part of the thanatocoenosis, the entire benthic assemblage (see Fig. 1 for location of samples shown in seabed profile).

Slope biofacies in the reduced oxygen zone
A similar planktonic foraminiferal assemblage to the middle to upper bathyal oxygenated zone occurs in this interval.

Oxygenated middle to lower slope
Planktonic foraminifera are abundant in this interval, comprising 83 to 98% of the entire assemblage. Globorotalia inflata is dominant except at 3340 m (MG53) where the assemblage is dominated by G. bulloides. Globorotalia hirsuta, G. truncatulinoides, G. ruber, Orbulina universa, Neogloboquadrina dutertrei, Fig. 8. Transect E. Bivariate graphs show abundance and distribution of various taxa as percentage of the biococenosis, the live + dead assemblages. Diversity as shown is the Margalef diversity (see Appendix 1). The total abundance of planktonic foraminifera are calculated as percentages of the benthic (live + dead) and planktonic assemblages, with the relative percentages of Globigerina bulloides, Globorotalia inflata and other planktonics (which include all species of planktonic foraminifera other than G. bulloides and G. inflata) shown as percentages of the planktonic assemblages. The Live:Dead:Relict abundances are shown as part of the thanatocoenosis, the entire benthic assemblage (see Fig. 1 for location of samples shown in seabed profile).

CORRESPONDENCE ANALYSIS
Correspondence analysis (CA) is an eigenvector method which allows the projection of a large multivariate cloud of points (samples, parameters or both) into a very reduced space (defined by the factor axes) while conserving the major part of the structured, meaningful information (Hennebert & Lees, 1991). The result is a two-dimensional graph where similar samples or parameters (or both) are nearest each other and those which are most dissimilar furthest apart. Correspondence Analysis calculates the variation (as a percentage) represented by each axes, and assigns a Relay Index (RI) to each sample or component, these values are rescaled such that the range RI's lies between 0 and 100 (Pickard & Emery, 1990).
The method is useful in detecting environmentally important trends in foraminiferal assemblage data. If a uni-dimensional variation trend such as a depth relay exists in the data set, projection of the results of CA onto a factorial plane (defined by the first and second CA axes) should produce an Arch effect or Guttman effect (Pickard & Emery, 1990). Correspondence Analysis has been applied successfully to modern and ancient carbonate facies analysis by Melguen (1974) and Hennebert & Lees (1991), Carboniferous limestones by Pickard & Emery (1990) and Carboniferous foraminiferal studies by Gallagher (1997). The computer packages utilized to carry out the analysis on this data, SedUTIL (used to store the data) and AFCal (a correspondence analysis package) were created by Lees (1989). Figure 9 shows the correspondence analysis results for the 36 samples studied from the Bass Canyon, and Figure 10 shows a CA plot of the parameters included in this analysis. The percentage abundance of 61 parameters in each of the 36 samples were included in the analysis: 44 were the most common benthic foraminiferal taxa, 11 were the most common planktonic foraminiferal taxa, other parameters include the total percentages of the four supergroups, 'undifferentiated foraminifera' and the % CaCO 3 (note, only 40 of the most useful parameters are annotated for clarity).
Before beginning analysis on the CA data, the proportion of inertia of each of the axes must be considered. The calculated inertia of the three axes is 38.0%, 26.4% and 9.7% respectively for axes one, two and three, totalling 74.1%. If values greater than 30% are obtained for the first axis, or if the total statistical variation accounted for the three axes is over 50%, then the results of the analysis are useful and a statistically significant arched plot is obtained (Gallagher, 1997).
The scatter of points on the CA plot is clearly related to depth, with the shelfal samples plotted on the left side of the graph and the lower slope cluster on the right. The three shallower samples (MG22, MG23 and MG24) contain a significantly different assemblage to the other middle to outer shelf samples included in this analyses, causing the statistical separation of this cluster. Although depth seems to be the dominant control on the distribution of faunas and parameter points on the correspondence analyses plot, it is not the only controlling factor, with other closely related parameters, such as dissolved oxygen and substrate, exerting control, as shown below.
Samples from 1137 m (MG57, transect C) and 1181 m (MG 63, transect D) contain similar foraminiferal assemblages yielding Relay Indices of 92.0 and 93.8 respectively, since both samples are molluscan-foraminiferal clay to fine sand. Samples from 366 m (SPB5, transect E) and 577 m (MG64, transect D), despite being derived from different depths, contain similar assemblages (with RI's of 81.2 and 85.6) where both samples comprise silt containing molluscs, foraminifera and peloids. In addition, the samples from 134 m (MG, 23, transect C) and 132 m (MG66, transect D), yield different assemblages, with RI's of 0.4 and 60.0 respectively, have contrasting facies: MG23 is a fine to medium sand containing molluscs, bryozoa, foraminifera and quartz fragments and MG66 is a silt containing molluscs, bryozoa, foraminifera and sponge spicules.
There is less variability in the assemblages of the deeper samples (>1000 m) compared to the shallower samples (<1000 m). The deeper samples' RI values range from 91.9 to 100.0, whereas the shallower samples' RIs values range from 0.0 to 93.3. This is partly due to decreased diversity and increased plankton percentage with depth (Figs 6, 7 & 8). Planktonic foraminifera account for at least 80% of the total assemblage in all samples deeper than 1000 m, and only a total of 11 species of planktonic foraminifera occur in this region.
Globocassidulina subglobosa and Cassidulina laevigata occur as outliers due to their dominance in MG65 (300 m) where they account for 40% of the entire assemblage, suggesting ideal conditions for these species.

DISCUSSION
The distribution of the foraminiferal taxa in the Bass Canyon will be discussed in this section. Figure 11 summarizes the distribution of the most common benthic foraminiferal species incorporating quantitative data from transects and correspondence analyses results. In addition, the occurrence of foraminifera interpreted to have been living at time of collection, as well as relict specimens are shown.
The distribution and abundance of Cibicides, the most common benthic genus in southeastern Australia , is similar to the percentage carbonate, which itself is related to abundant bryozoans. This suggests that Cibicides prefers bioclast-rich sediments on which to attach themselves. Foraminiferal diversity also displays a similar pattern to the percentage of carbonate, with the highest diversity occurring in the 'carbonate factory'.
The epifaunal Cibicides refulgens occurs in samples shallower than 145 m on transect C. This taxon is substrate controlled, most commonly found on bioclastic sands, where it attaches itself to hard substrates . The distribution of C. refulgens is inversely related to Uvigerina on this transect. Uvigerina is restricted to the neritic and outer neritic zones in the Bass Canyon. The percentage of CaCO 3 is related to the nature of the substrate, with higher values reflecting the presence of bioclast-rich carbonates and coarser grained bioclastic substrates.
Lenticulina spp. occurs from the Gippsland shelf to the bathyal zone (210 m to 603 m). Parrellina spp. is present from 88 m to 471 m at the head of the Bass Canyon. The most common species of this genus is Parrellina imperatrix. The distribution and abundance of Astrononion spp. in southeastern Australia is similar to the percentage mud distribution. This is the typical substrate type for this infaunal genus (Murray, 1991). Elphidium spp. is common in the middle shelf samples, including minor relict forms. This taxon is rare deeper than 150 m water depth. Discorbis spp. occurs only on the shelf.
The most common benthic rotaliids in the deeper samples are Sphaeroidina bulloides, Gyroidinoides spp., Cibicidoides wuellerstorfi and Bulimina spp. Bulimina spp., like Rhabdammina spp., often inhabits reduced oxygen bottom water conditions (Sen Gupta & Machain-Castillo, 1993). Bernhard (1986) studied benthic rotaliid assemblages from Jurassic to Holocene anoxic deposits and concluded that anaerobic taxa possess predominantly flattened, unornamented tests. The test walls are often highly perforate and thin such as in the taxon Bulimina spp. In the Bass Canyon, Bulimina spp. is most common from 1996 m to 2352 m, in the reduced-oxygen zone.
Miliolids decrease in abundance with depth, with abundant relict and live fauna in the shallow samples. Quinqueloculina spp., the dominant miliolid, is most common in the shallower samples on transect C and transect E. Pyrgo lucernula and Pyrgo serrata are deep-water miliolids, occurring below 3692 m in the Bass Canyon.
Textulariids are geographically the most widespread group of benthic foraminifera, being well represented in marginal marine, hypersaline, hyposaline and bathyal regions (Zheng & Fu, 1990). In the Gippsland Basin, Textularia saggitula, Gaudryina convexa and G. quadrangularis occur as deep as 1137 m, although they are most common at depths less than 300 m.
Rhabdammina spp. is present in minor quantities in the shallower samples. This is the dominant benthic taxon in the reduced-oxygen zone samples over 1000 m depth, often co-occurring with Rhizammina spp. The use of these taxa as reduced-oxygen indicators was reported by Morlotti & Kuhnt (1992). This taxon is also known to live in areas where strong bottom-water currents occur, often the result of continental margin upwelling, which may cause reduced-oxygen bottom water (Linke & Lutze, 1993).
It is clear from the distribution of inferred living forms that reworking by wave action or other processes (such as bioturbation, turbidity currents or scarring) of the Bass Canyon assemblages has not occurred, since the total faunal assemblage data (i.e., total living + recently dead + relict forms) is similar to the distribution of the live forms in the shallower biofacies (Fig. 11). Production rates were also estimated between species using the method described by Murray (1991) and used by Smith et al. (2001). There were no differential production rates greater than 10 or 12 times between species, suggesting that postmortem influences have not affected the biocoenosis.
Live foraminifera are however absent below the reducedoxygen zone. This may be due to the difficulty in distinguishing 'live' Rhabdammina spp. due to their agglutinated tests, constructed from planktonic foraminifera or it may be due to instability below the OMZ. Jorissen's 1988 study on the vertical distribution of living benthic foraminifera in submarine canyons off New Jersey, concludes that the relative lack of live faunas within submarine canyons is a result of the inherent instability of the environment, with small-scale mass wasting and/or scouring unfavourable for a stable ecosystem. Jorissen (1988) also states that straight canyon environments are more susceptible to ecosystem instability than meandering canyons.
As summarised in Fig. 11, the distribution of relict foraminifera in the Bass Canyon samples is confined mainly to shallow shelf samples (shallower than 145 m water depth). With the exception of Gaudryina convexa and Gaudryina quadrangularis, most recorded relict foraminiferal taxa in the Gippsland Basin are commonly found living at depths of less than 242 m. In Fig. 11. An idealized summary of the distribution of benthic foraminifera in the Bass Canyon based on the results of the quantitative foraminiferal analysis and the correspondence analysis. Note: the bars represent the relative abundance of particular taxa from the benthic (live + dead) assemblages. The occurrence of relict forms are noted, however they are not used to calculate abundances. Smith et al. (2001), it is suggested that only samples from less than 150 m water depth will contain in situ relict forms, as it is inferred that these relict forms may be a product of the last glacial maximum (w19,000 years ago; Bard et al., 1990), during which time the sea level was between 120 and 150 m below present day (Chappell & Shackleton, 1986). Figure 12 summarizes the biostratigraphic ranges of 30 Bass Canyon foraminiferal taxa (listed in Fig. 11), based on stratigraphical studies of the Cenozoic successions of southeast Australia (northern Tasmania, the Otway Basin and the Gippsland Basin). In addition, cosmopolitan and (semi)endemic foraminifera are distinguished. The term (semi)endemic here refers to Bass Canyon foraminiferal taxa that are either confined to southern Australia, New Zealand, southwest Pacific or South America (sensu Li et al., 1996c).

CENOZOIC PALAEOENVIRONMENTAL SIGNIFICANCE OF THE BASS CANYON FORAMINIFERAL FAUNAS
The majority of the modern Bass Canyon fauna listed are cosmopolitan species, and show similar depth distributions to those listed in Murray (1991). These taxa are therefore useful for global palaeoenvironmental studies. The few (semi-)endemic taxa, suitable only for regional palaeo-environmental studies, are mostly restricted to the shelf. This is not surprising, as it is suggested in Smith et al. (2001) that modern deeper-living forms are more conservative since they evolved in relatively lowerstress eutrophic environments, than their shallower oligotrophic environment dwelling contemporaries.

CONCLUSIONS + The neritic realm of the Bass Canyon (88 m to 300 m)
contains between 50 to 80% CaCO 3 , with the maximum occurring on the western side of the canyon, due to the dominance of bryozoa in the carbonate 'factory'. Gravel-size sediments account for between 0 to 50% of the sediments, increasing with depth, with the maximum occurring on the western side of the canyon in the carbonate 'factory'.  Fig. 11. The ranges for the taxa are for the Gippsland, Otway and Tasmanian Cenozoic strata, based on data from Parr (1945), Li et al. (1996c), Li & McGowran (1997), Gallagher et al. (1999), Gallagher & Holdgate (1996 and Holdgate & Gallagher (1997). The biogeographic data were obtained from Li et al. (1996c) and Murray (1991).
Sand-size sediments account for between 40 to 90% of the sediments, with mud-size sediments increasing with depth from less than 5% at 80 m to 30% at w200 m water depth. Hastigerina siphonifera and Globorotalia crassula. + The CCD is the region is estimated to occur at around 3600 m, this is reflected by a decrease in percentage carbonate after this depth. + Correspondence Analysis of the 61 parameters (percentage abundance of foraminifera and carbonate) in 36 samples yielded a clear depth-related pattern, although it is suggested that other related parameters such as dissolved oxygen and substrate also control the distribution of foraminifera. + Relict foraminifera are restricted to shelfal depths, shallower than 145 m. This pattern is similar to other shelf regions in Australia, where shelf areas were exposed during the Last Glacial Maximum, reworking shelf facies shallower than 150 m. + The distribution of living foraminifera is similar to the distribution of the total assemblage, suggesting that the surficial sediment is this region has not been significantly mixed by wave, slump or bioturbation processes. + The majority of the modern Bass Canyon fauna listed are cosmopolitan species, with few (semi-)endemic taxa mostly restricted to the shelf, providing further evidence that modern deeper-living forms are more conservative since they evolved in relatively lower-stress eutrophic environments, than their shallower oligotrophic environment dwelling contemporaries.

ACKNOWLEDGEMENTS
This study was supported by a grant from the Australian Research Council. The authors thank David Taylor for providing the additional 'SPB' and 'Kimbla' samples, and the crew of the R/V Franklin for their help with collecting the Gippsland Basin samples during September to October, 1998. The authors thank two reviewers for their valuable assistance. This coefficient more accurately relates each sample to one another as some samples have a different number of specimens picked than others (from Brenchley & Harper, 1998).