Chronostratigraphy of the Brazos–Trinity Depositional System, Western Gulf of Mexico: Implications for Deepwater Depositional Models
- File PDFChapter PDF
Carlos Pirmez, Bradford E. Prather, Gianni Mallarino, Walter W. O’hayer, Andre W. Droxler, Charles D. Winker, 2012. "Chronostratigraphy of the Brazos–Trinity Depositional System, Western Gulf of Mexico: Implications for Deepwater Depositional Models", Application of the Principles of Seismic Geomorphology to Continental Slope and Base-of-Slope Systems: Case Studies from SeaFloor and Near-Sea Floor Analogues, Bradford E. Prather, Mark E. Deptuck, David Mohrig, Berend Van Hoorn, Russell B. Wynn
Download citation file:
A series of four intraslope basins linked by submarine channels in the northwestern Gulf of Mexico form part of a source-to-sink depositional system that starts in the headwaters of the Brazos and Trinity Rivers and terminates in a ponded intraslope basin offshore Texas—the Brazos–Trinity depositional system. The system is well imaged with 3D seismic data, and two of the basins have been drilled, with three Integrated Ocean Drilling Program wells and two geotechnical wells. Using an integrated approach, we have combined seismic-litho-bio-tephro-stable-isotope-radioisotope stratigraphic methods, using both new and published results, to generate a millennial-scale-resolution chronostratigraphy for this system. Basins I through IV are infilled with about 62 km3 of sand-rich sediments (∼ 1.6 ×1011 metric tons) transported by sediment gravity flows since the last interglacial (Oxygen Isotope Stage 5e). The bulk of the sediments, about 49 km3, were deposited within a short time period within Oxygen Isotope Stage 2, starting at 24.3 ka at the latest and ending at ∼ 15.3 ka, before meltwater pulse 1A. Sediment accumulated in the slope basins at rates which varied over time between 1.4 and 55 million tons per year. Except for a short time interval when the Brazos River was diverted to the shelf edge at the head of Basin 1, sediment flux to deepwater was on average less than the present-day sediment discharge of the Trinity–Brazos–Sabine Rivers combined. In the period 24-15 ka the sediment sinks comprising the slope basins and shelf-margin delta can be balanced against the fluvial sources if their discharges are somewhat lower than present day, and if the contribution from incised-valley erosion was relatively small. The history of sedimentation on the slope basins is modulated by sea-level changes, but it is strongly influenced by basin topography and by the dynamics of delta development on the shelf. During peak high stands of sea level the slope area receives only pelagic sediments; during low sea-level stands, the sedimentation in each basin results from a complex combination between fluvial input at the head of the first basin, and the rate of subsidence/sedimentation causing basin topography. The ages of sediments in separate basins show that sedimentation occurs at the same time in multiple basins with trapping of sand in updip basins, while mud is preferentially deposited in downdip basins.
The Brazos–Trinity (B–T) depositional system offshore Texas, U.S.A., is composed of a series of four intraslope basins linked by a network of submarine channels extending from the shelf edge to about 100 km onto the continental slope (Fig. 1; Winker, 1996; Satterfield, 1988; Satterfield and Behrens, 1990). Sediment sourced in the watersheds of the Brazos, Trinity, Neches, Sabine, Calcasieu, and western Louisiana Rivers during the late Pleistocene reached the edge of the shelf and the upper reaches of Basin I on the upper continental slope (Suter and Berryhill, 1985; Anderson et al., 1996; Anderson et al., 2004; Simms et al., 2007a). From there sediment gravity flows transported sand and mud as far as 100 km down the depositional system, eventually reaching Basin IV, which represents the terminal end of the system during the latest Pleistocene. Basin V, farther south, remained isolated and received sediment flows derived only from the basin flanks (Fig. 2; Winker, 1996). The land, shelf, and submarine components of the B–T depositional system have been the subject of numerous studies by industry and academia. The system not only represents an excellent laboratory to investigate submarine deposition from source to sink in a relatively small, well imaged area, but it also is an excellent analog for hydrocarbon reservoirs in ancient deepwater (Beaubouef and Friedmann, 2000; Beaubouef et al., 2003; Beaubouef and Abreu, 2006; Mallarino et al., 2006; Prather et al., 1998; Winker and Booth, 2000) and shallow marine sequences (e.g., Anderson et al., 2004; Suter and Berryhill, 1985). The B–T system is particularly well suited to investigate the fate of sediments derived from land as they are carried into deepwater, because: (1) the sediment sources are well studied, and the history of the fluvial-deltaic system is well understood through numerous studies using seismic and borehole data, (2) the points of input are limited, with sediment entering the deepwater system at the shelf edge north of Basin I; (3) the pathways in deepwater are well defined by detailed 2D and 3D seismic coverage, and (4) the main “sink”, Basin IV, is well defined in a single, presently closed intraslope basin (Fig. 2). With the availability of extensive seismic coverage and detailed chronostratigraphic control of the various sequences in each basin as shown in this paper, we can address in detail the main controls on sediment delivery from the shelf edge to the upper slope, and attempt to quantify the sediment budget along an entire source-to-sink depositional system. The interpretation of the lithostratigraphy and seismic stratigraphy is presented in a companion paper by Prather et al. (this volume). In this paper we focus our attention on the age dating and implications of the timing for basin-fill evolution, relationship to sea level and fluvio-deltaic evolution on the shelf, and quantification of the sediment fluxes into deepwater.
The submarine portion of the Brazos–Trinity depositional system has a complex and irregular topography typical of basins underlain by thick salt layers (Figs. 2, 3). The highly mobile substrate leads to the formation of irregularly shaped basins, ridges, diapirs, and fault scarps. Prather (2003) termed these slope systems as above-grade slopes, to indicate the potential for sediment gravity flows traversing these areas to erode (deposit) locally, as the substrate is steepened (flattened) by tectonic deformation, as opposed to graded systems, in which the slope profile tends to equilibrium when sedimentary and tectonic processes balance each other (Pirmez et al., 2000). Studies of the geology of the B–T system have, until recently, focused on the seismic stratigraphy and geometry of the sediment-gravity-flow deposits infilling the various slope basins, since lithologic calibration of the basin fill was previously unavailable. As a result, the history of sediment delivery, the timing of basin fill, and the relationship to sea-level changes were largely inferred. Piston cores in the area were studied by Satterfield and Behrens (1990), but these cores sampled only the uppermost few meters. The lithology of the fill in Basin II was calibrated by two geotechnical wells drilled by Shell in 1990, but the results remained unpublished until now (Prather et al., this volume, page 79–105; and this paper). The lithology of the fill in Basin IV was calibrated more recently by a series of long piston cores (Mallarino et al., 2006) and by three holes cored and logged during IODP Expedition 308 (Expedition 308 Scientists, 2006). Mallarino et al. (2006) analyzed the biostra-tigraphy and stable isotopes of planktonic foraminifer shells in piston cores from the margins of Basin IV. They conclude that sediment delivery to Basin IV is controlled largely by sea level, with rapid influx during periods of sea-level fall.
In this paper, we report on analysis of cores from IODP Expedition 308 in Basin IV and cores from the Shell Rudder wells in Basin II, and provide an integrated chronostratigraphy that augments and revises the study of Mallarino et al. (2006). After obtaining the absolute age for the various depositional units in Basins II and IV, we estimate the sediment flux into the basins over time and compare the sedimentation patterns in the slope basins with basin subsidence patterns, and with other studies reporting independently dated sea-level changes and fluvial– deltaic evolution on the shelf. Our results provide for an evaluation of all the elements of the sediment budget in the Brazos– Trinity depositional system from source to sink during the last glacial cycle.
Morphology of the Brazos–Trinity System
The morphology of the B–T depositional system was first described by Satterfield (1988) and Satterfield and Behrens (1990) based on a few 2D profiles, and later in detail from a detailed grid of high-resolution 2D seismic profiles by Winker (1996). Later studies by Beaubouef and Friedmann (2000) and Badalini et al. (2000) using 2D seismic and seafloor imagery discussed the evolution of the basins over time. Pirmez et al. (2000) showed detailed imagery of Basin I channel networks based on partial 3D coverage and investigated the processes of development of longitudinal channel profile. More recently Beaubouef et al. (2003), Beaubouef and Abreu (2006), and Mallarino et al. (2006) showed multibeam bathymetry and high-resolution 3D seismic that covered most of Basin IV. In this paper and the companion study by Prather et al. (this volume), we have compiled 3D seismic which offers a nearly complete high-resolution coverage (25 ×25 m, ∼ 60 Hz near the surface) of the Brazos–Trinity intraslope basins and surrounding areas, including the shelf (Figs. 2, 3). The 3D seismic coverage together with the high-resolution seismic lines and borehole data offers new perspectives into the sediment dispersal from the shelf to deepwater in this system.
The sea-floor map derived from 3D seismic coverage (Fig. 2) displays a very complex topography in much more detail than the previously available multibeam bathymetry (Holcombe et al., 2002). Basin I is surrounded by steep scarps to the east and west which are associated with normal faults at the edge of salt-cored highs (Fig. 2). The edge of the shelf is marked by a series of faults, including both counterregional and regional normal faults with relay zones in between them (Figs. 2, 3). These large growth faults offset the seafloor, indicating recent fault activity and a likely influence on shelf-margin delta deposition and delivery of sediment to Basin I. The maps of paleo–fluvial valleys on the shelf show a complex network of valleys, with the Brazos, Trinity, Sabine, Neches, and Calcasieu rivers directly linked to the paleo-deltas near the head of Basin I (Fig. 1; Anderson et al., 2004; Simms et al., 2007a). The entry point into the slope basins is a relatively narrow segment of the shelf no more than about 10 km wide. This means that to capture fluvial drainage and sediments into the slope depositional system, advancing delta lobes had to be positioned in a relatively small region, or else sediments would flow to basins either to the west or to the east, away from the Basins I–IV slope corridor.
The Colorado River drainage fed the Colorado shelf-edge delta and did not link directly with the Brazos shelf valley (Anderson et al., 2004). The mineralogy of sediments deposited in the slope basins of the B–T system confirms that the Colorado was a minor contributor, with a possible influence through mixing of sediments on the shelf by longshore currents (Prather et al., this volume). Rivers from the western Louisiana area also built shelf-edge deltas just to the east of the Basin I area, but their drainage shifted east before they could reach Basin I, and they did not appear to have contributed to the B–T system during the last sea-level cycle (Anderson, 2005; Wellner et al., 2004). Sediments transported by the Mississippi River, on the other hand, appear to have contributed to sedimentation in the B–T slope area, particularly during earlier periods (older than OIS 5).
On the slope, Basins I–IV are characterized by a relatively flat floor, in contrast to the basin flanks, which dip several degrees (Fig. 2). Basins I–III are open, in that flows can exit the basin unimpeded by a high sill. Instead, the sill between basins shows a variably incised channel, cutting into the steeper interbasin areas (Prather et al., this volume; Pirmez et al., 2000). A continuous channel exists starting at the exit of Basin I through the eastern part of Basin II, across Basin III, and terminating in the proximal part of Basin IV. An additional channel in the western part of Basin II links it to Basin IV. Basin IV, in contrast, is completely enclosed by salt-cored highs, with no channels exiting the basin. The present seafloor indicates a spill point at about 1250 m below sea level on the northeast side of Basin IV, about 200 m above the basin floor. The basin boundary on the spill point is represented by a “knife-edge” sill, defined by a steep fault scarp down to the Gyre Basin (Fig. 2). Basins II–IV also show complex fault networks along the basin edges as a result of accentuated subsidence along the edges of salt bodies (Fig. 2). The basins downdip of Basin IV, namely Basin V (or Barton Basin) and the Gyre Basin have relatively rough and irregular topography, which results from mass-wasting deposits derived from the basin edges (cf. Winker, 1996). Other basins to the east of Basins I–IV, such as the Gealy and Bowie basins, display a relatively smooth topography but lack the nearly flat basin floor characteristic of infilled basins such as Basins I–IV. Basin A, to the west of Basin I, was partly infilled by sediment gravity flows, and contains locally rough seafloor topography associated with mass wasting from the shelf margin (Fig. 2; Wellner et al., 2004, their Figure 13).
The seismic stratigraphic framework as defined by Prather et al. (this volume) uses key seismic horizons identified on high-resolution 2D seismic lines (e.g., Figs. 4–6), tracked the same horizons on the equivalent 3D seismic lines (e.g., Fig. 3), and then correlated across the entire area of Basins I–IV using the 3D seismic data. These horizons provide the physical stratigraphy used to define depositional series. Details of the data and methods used to tie the wells to the seismic data and to compute the volumes of each stratigraphic unit bounded by the key horizons are described in Appendix A, Sections A1 and A2.
Each key horizon is mappable throughout the area and has distinguishing geometric or seismic facies characteristics. Horizon 10 separates laterally continuous, moderate- to high-amplitude reflectors below, from a laterally continuous low-amplitude to transparent unit above. Horizon 20 is defined on the basis of onlapping reflections above it, observed particularly in Basin IV (Fig. 4) and Basin I (Fig. 6). Horizon 20 can be traced to Basin II, where it shows conformable reflections above and below. Horizon 30 is a laterally continuous reflector defined by onlapping reflections above it in all basins; locally it is truncated by an acoustically transparent unit (Fig. 4). Horizons 40, 50, and 60 are laterally continuous horizons that mark a vertical change in seismic facies. Horizon 40 is an erosional surface locally near the exit of Basin II (Fig. 5). Horizon 61 (Fig. 4) is an erosional surface with onlapping reflections above. It is a very weak reflection and is mappable only in Basin IV. Horizon 70 is well defined in Basin IV, where it forms a baselap surface at the base of a fan wedge sourced from the eastern channel (Fig. 4).
The stratigraphic units defined by these key horizons are described in detail by Prather et al. (this volume), and only a brief summary is presented here. Figures 3–6 illustrate the seismic character, geometry, lithofacies, and well logs obtained in each of the basins along the system (Basins I, II, and IV).
Series 10 comprises subparallel reflections that mostly drape pre-existing topography. Where penetrated by the wells it is composed of mud, probably deposited by muddy surface plumes or by very low-density turbidity currents (or nepheloid layers). A similar pattern of low-amplitude parallel reflections draping topography is observed in Basins I–IV. It is likely that Basin IV, and possibly Basins I and II, did not have significant topographic expression at this time. Mud composition suggests that sediments in these basins were derived mainly from the Mississippi River (Prather et al., this volume; Expedition 308 Scientists, 2006). An ∼ 2-m-thick hemipelagic interval rich in calcareous microfos-sils that represent a decrease of terrigenous sediment input caps this series.
Series 20 comprises convergent reflections in Basin IV, parallel reflections in Basin II, and convergent reflections in Basin I. It contains thin-bedded turbidites and mud in Basin IV and mud in Basin II. According to Prather et al. (this volume) Basin II probably did not have a basin expression during deposition of this unit, whereas Basins I and IV were enclosed ponds. Figure 6 shows high-amplitude onlapping reflections at the base of Series 20 in Basin I. A thick (∼ 8 m) hemipelagic unit rich in calcareous microfossils that represents a marked decrease in terrigenous sediment input caps this series, and was sampled in both Basins II and IV. This hemipelagic unit contains a distinct ash bed.
Series 30 and younger deposits are composed of seismic reflections that onlap the basin margins and contain sand-rich turbidites. The basin relief and morphology at the time of deposition was well developed in all basins at the onset of Series 30, caused by subsidence during the end of Series 20. As a result, ponded aprons formed in Basins I–II and IV during Series 30, including mass-transport deposits sourced from the basin flanks and laterally extensive turbidites.
Series 40–70 comprises perched aprons in Basin II, whereas in Basin IV Series 40 reflections are represented by a low-relief ponded apron and Series 50–70 by high-relief ponded aprons. Series 70 in Basin IV represents a high-relief apron within a ponded-basin setting, which is fed by a channel that traverses Basin III (Prather et al., this volume). In Basin I we did not refine the stratigraphic framework above Series 30 because of difficulty in correlating with other basins and lack of cores and age dating. Series 40–70 contains a very high sand:mud ratio in both basins II and IV with packages of thick-bedded turbidites cored in Basin IV. Series 70 is capped everywhere by an ∼ 5 m hemipelagic layer rich in calcareous microfossils that represents a pronounced decrease of input of terrigenous sediment.
To determine the ages of the sedimentary units in the B–T system we used an integrated stratigraphic approach applying a wide variety of data types and methods. These are presented in detail in Appendix A and are summarized briefly here.
Biostratigraphic zonation of planktonic foraminifers conducted in samples from the Rudder cores follows the methodology of Kennett and Huddleston (1972) (details in Appendix A3). These are the same methods used by Mallarino et al. (2006) to study forams from piston cores collected from Basin IV. Nanno-fossil biostratigraphy of the IODP cores was conducted by Shumnyk, written communication, and Flemings et al. (2006). Oxygen-isotope measurements from Mallarino et al. (2006) and from O’Hayer (2009) in the IODP cores provide some additional constraints on age, but the records are complex due to variable input of fresh water from the Mississippi drainage (Appendix A4). Using light-reflectance measurements on the split-core face obtained in both the Mallarino et al (2006) cores and the IODP cores, we were able to include their measurements and biostrati-graphic observations into our integrated stratigraphy (Appendix A5). Ash geochemistry provides independent correlation evidence between ash layers encountered in different basins and with other ash layers in the Gulf of Mexico (Appendix A6). Radiocarbon ages on planktonic foraminifers, reworked plant debris, and shell material provided key age constraints in the interval younger than 40 ka, which represents most of interval studied (Appendix A7). The initial attempts to use magnetostratigraphy on the IODP cores proved difficult due to the presence of sand and silt layers and extremely variable sedimentation rates (Expedition 308 Scientists, 2006). No further attempt was made in this study to integrate the measurements of magnetic remnant intensity into the stratigraphy. Basins I and III were not cored during either IODP Expedition 308 or Shell geotechnical drilling campaign; as a result, ages in these two basins rely solely on seismic stratigraphic correlation to Basins II and IV.
Absolute ages of biostratigraphic events are based on the time scale of Kohl et al. (2004) (Fig. 7). The key events for the interval of interest are: the last occurrence (LO) of the planktonic foraminifer Globorotalia inflata, the Emiliana huxleyi acme (bloom of a calcareous nannofossil), and the last-appearance datum (LAD) of the planktonic foraminifer Globorotalia menardii flexuosa. Various authors disagree on the exact ages for each of those events. The LO of G. inflata is at 12 ka according to Kohl et al. (2004) and 10.5 ka according to Kennett et al. (1985). The E. huxleyi acme (EHA) is at 90 ka according to Berggren et al. (1985) and at 85 ka according to Kohl (1986), but based on tephra correlation we believe that the EHA may occur later in the Brazos–Trinity area (see below). The G. flexuosa LAD (or Y/X boundary of Ericson and Wollin, 1955) is dated at 68 ka by Berggren et al. (1995), at 85 ka by Kohl (1986), and at 89 ka by Kohl et al. (2004).
A volcanic ash layer about 2 cm thick was cored in Sites U1319, U1320 (Expedition 308 Scientists, 2006), and Rudder 1-Core 7 and was also recovered from piston cores from the margin of Basin IV (Mallarino et al., 2006). Based on seismic correlation we are confident that the ash layer encountered in the Basin IV wells U1319 and U1320 belongs to a single ash-fall event (Fig. 4). On the basis of biostratigraphic correlation Mallarino et al. (2006) identify this ash as tephra Y8, corresponding to the eruption of the Los Chocoyos volcano in Nicaragua, estimated to have occurred 84,000 years ago (Drexler et al., 1980) (Fig. 7). We conducted geochemical analyses of an ash sample from U1319 and from Rudder #1 to address two questions: (1) whether the ash layer cored in Basin II represents the same event as the ash cored in Basin IV, and (2) whether we could obtain independent evidence for correlation with ash stratigraphy in the Gulf of Mexico.
The ash material in the Rudder 1 borehole is indistinguishable from the ash encountered in U1319A (Fig. 8). At Site U1320 the ash layer was not sufficiently distinct to obtain a clean sample for chemical analysis, due to coring deformation. Comparison of our chemical analyses with that of the ashes cored farther south in the Gulf of Mexico and the Caribbean by Drexler et al. (1980) and Rabek et al. (1985) shows that the ash layer cored in Basin II and IV is chemically distinct from the Y8 ash, particularly in its iron content, and in fact it appears to be more closely related to ash W1. However, the position of the ash with respect to the biostrati-graphic zonation (Mallarino et al., 2006 and below) makes a correlation with W1 impossible. It is possible that the ash layer sampled in Basins II and IV is related to some other eruption than the Los Chocoyos (for instance an eruption in the North American continent), but this cannot be independently confirmed, because the methods used here (XRF) are not the exactly the same as used by Rabek et al. (1985) (ICP-MS).
In summary, the ash layer is a correlative event between Basins II and IV based on chemical composition confirming the seismic stratigraphic correlation. According to biostratigraphy (see below) the ash is likely the same as ash Y8 of Drexler et al. (1980), as proposed by Mallarino et al. (2006), but their conclusion is not supported independently by chemical analyses.
Biostratigraphy and Oxygen Isotopes
Results from biostratigraphic and isotope studies are presented separately for each basin.
Samples in Rudder 1–Core 8 show a paucity of warm-water specimens such as G. menardii and P. obliquiloculata and are dominated by cooler-water species such as G. inflata (Fig. 9). This core is interpreted to be within biozone W. Core Rudder 1–7 contains abundant planktonic foraminifers, indicating that this interval consists of hemipelagic mud. The faunal assemblage at its bottom is distinguished by the decreasing frequencies of the G. menardii group, which disappear just below 98.15 mbsf (322 ftbsf). Within the G. menardii group, G. menardii flexuosa disappears around 98.3 mbsf (322.5 ftbsf). This, along with the low frequency of other warm-water forms such as P. obliquiloculata and G. truncatulinoides right coiling, point to a transition from a warmer to a cooler period, placing this lower interval in the uppermost part of subzone X1 and the boundary with the overlying zone Y (Fig. 7). The overlying section, up to 97.8 mbsf (321 ftbsf), is marked by a very distinct faunal change. Nearly all taxon frequencies decrease abruptly, with the exception of P. obliquiloculata and O. universa, which keep moderate frequencies. G. ruber forms a peak of high frequency and, by far, is the dominant taxon, with abundance percentage up to 67% of the entire assemblage. The ash layer observed at 98 mbsf (321.4 ftbsf) corresponds to tephra layer Y8 (Kennett and Huddleston, 1972, Drexler et al., 1980; Rabek et al., 1985). Similarly, a foraminiferal assemblage dominated by G. ruber and an ash layer mark subzone Y8 in the sedimentary fill of Basin IV (Mallarino et al., 2006).
The surface cores in both Rudder 1 and Rudder 2 wells in Basin II contain abundant foraminifers, indicating that they represent hemipelagic deposits. The assemblage of warm-water species is dominant, indicative of biozone Z. The relative proportion of various species suggests that the subzone boundary Z1–Z2 occurs within these surface cores. Because of coring procedures at the time it is unclear whether the seabed interface was captured in these cores, and so the exact depth from the seafloor may not be accurate in these short surface cores; therefore, these cores represent only part of the Holocene drape.
Oxygen isotope measurements from the Rudder cores were conducted to correlate with other cores in the region and with global records. The two samples from core Rudder 1–8 at 117.1 mbsf (384.2 ftbsf) and 116.8 mbsf (383.2 ftbsf) have delta;18O values of -1.95% and -1.16% respectively. In core Rudder 1–7, the lower part of the delta;18O curve shows a shift from relatively lighter, -2.46%, to heavier, -1.43%, values followed by a distinct shift back to more depleted values as light as -3.31%. Such a distinct depleted peak occurs at 97.9 mbsf (321.3 ftbsf). The upper part of the oxygen curve is characterized by a shift towards the heaviest value of -1.05% at 97.3 mbsf (319.2 ftbsf), followed by slightly more depleted values in the uppermost segment of the curve. The results suggest the presence of a melt-water-related event near the Y8–Y7 biozone boundary. The apparent increase in 18O-depleted fresh water into this area of the Gulf of Mexico at this time is accompanied by a distinct faunal change to fresh-water-tolerant G. ruber and other warm-water species. A similar delta;18O light peak is observed in core MD-2642 from Basin IV analyzed by Mallarino et al (2006) (Fig. 10).
In summary, biostratigraphy in Basin II constrains the age of Series 10 as most likely within the W biozone, confirms the presence of the X/Y biozone boundary and ash Y8 in Series 20 in an interval of hemipelagic deposition, and establishes that Basin II sedimentation was dominated by hemipelagic deposits some time before the Z1/Z2 boundary.
A detailed biostratigraphic analysis conducted by Mallarino et al. (2006) in some of the piston cores obtained in the basin margin was carried to the IODP wells using the light-reflectance logs measured in the split-core face (Fig. 10). The correlation of characteristic lightness peaks and troughs enabled the characterization of the Z–Y, Y–X, and other biozone boundaries in U1319 and U1320. This correlation enables the establishment of a system-wide framework that ensures consistency between the dataset studied by Mallarino et al. (2006) and in this and other studies using IODP well results.
A hemipelagic interval sampled in U1319A (between 30 and 32 mbsf, Fig. 10) shows a characteristic peak in light reflectance that is also observed in some of the longer piston cores (e.g., MD03-2641 and MD03-2640, Fig. 10). Mallarino et al. (2006) inferred this interval to represent Oxygen Isotope Stages (OIS) 5c to 5d based on the character of the delta;18O measurements in their cores (see MD03-2640-41-42 in Fig. 10). More recently, O’Hayer (2009) observed that the interval below 31 mbsf in U1319 and the correlative interval below 178 mbsf in U1320 showed a complete absence of specimens from the G. menardii complex, with a faunal assemblage similar to core Rudder 1C-8 (Fig. 9), suggesting that the interval is most likely within the W biozone. The light delta;18O values measured by O’Hayer (2009) combined with the absence of G. menardii indicate that the muddy interval of Series 10 represents the upper part of OIS 6 and transition to OIS 5 (Termination II, or upper part of the W biozone) (Fig. 10; O’Hayer, 2009). An influence of fresh water, possibly from the Mississippi River, also is likely, given the light delta;18O values and mud composition rich in detrital dolomite (Expedition 308 Scientists, 2006). Thus, the fossil-rich hemipelagic mud cored in U1319A at 30–32 mbsf, capping the laterally continuous Series 10 at the base of the infill of Basin IV, must represent sedimentation during OIS 5e or lower part of the X biozone.
Another fossil-rich hemipelagic interval, in the middle of the Basin IV fill, was completely cored in both the IODP wells at the basin margin (U1319) and center (U1320). This interval marks the top of Series 20 and is well defined in the high-resolution 2D seismic lines as a nearly constant-thickness interval that can be correlated across all basins (Figs. 4–6, and Prather et al., this volume). Based on seismic stratigraphic correlation, biostratig-raphy, and presence of the ash layer, this is the same hemipelagic interval encountered at the base of the fill in Basin II and sampled in Core Rudder 1-7. The same interval was correlated into Basin I using the high-resolution 2D and 3D seismic data (Figs. 3, 6; Prather et al., this volume).
Mallarino et al. (2006) note that in Basin IV piston cores the Y– X boundary (G. flexuosa LAD) occurs below the Y8 ash layer (e.g., MD03-2642, Fig. 10). In the IODP308 wells, the E. huxleyi acme (EHA) appears to occur above the same ash layer (Shumnyk, written communication). The Y8 ash dates between 80 and 90 ka, with the exact age assigned representing an estimate based on delta;18O stratigraphy (84 ka; Drexler et al., 1980). The exact age of the EHA also varies from place to place, probably due to local oceanographic conditions influencing the nannofossil bloom. In this case, the high-resolution stratigraphy provided from the hemipelagic interval in U1320 and U1319 provides the best local calibration for these events and indicates that in this area of the northern Gulf of Mexico the nannofossil bloom occurred at ∼ 83 ka, later than in other areas (Fig. 7).
Mallarino et al. (2006) show that the upper 5 m of the Basin IV fill contains a microfossil-rich hemipelagic unit that, based on biostratigraphy and oxygen isotopes, represents biozones Y1 and Z (Fig. 10). The delta;18O measurements show a significant peak suggestive of a meltwater pulse within the Y1 subzone. Mallarino et al. (2006) believe that this peak represents meltwater pulse 1A (∼ 14 ka). The hemipelagic interval obtained in the IODP cores was somewhat thinner (∼ 4 m), probably due to the difficulty in obtaining a complete core top in the soft seabed by the IODP piston core system. However, correlation of the lightness curve shows that the IODP cores recovered the lower part of this hemipelagic unit, including the Z1/Z2 subzone boundary (Fig. 10).
In summary, biostratigraphy in Basin IV reveals that Series 10 is within the W biozone and that the hemipelagic unit at the base of the basin fill must correspond to the last interglacial, OIS 5e. Correlation with the cores studied by Mallarino et al. (2006) places the IODP cores within the same biostratigraphic framework. The hemipelagic interval in the middle of the basin fill corresponds to a pause in sediment input lasting through most of the Y biozone, which is interrupted by a short pulse of sedimentation between biozones Y3 and Y1 as established by Mallarino et al. (2006).
Radiocarbon ages provide a tool complementary to biostra-tigraphy, with the potential to obtain a millennial-time-scale resolution in these rapidly accumulated sediments, particularly in the interval 15–40 ka, during which there are no biostrati-graphic data. A total of 52 ages were obtained from the IODP cores and from the Shell Rudder cores in Basin IV and II, respectively (data and methods discussed in Appendix A7). We sampled a range of materials, from reworked wood, large shell fragments, dispersed organic matter in mud, and planktonic foraminifera (Fig. 11). In addition to AMS 14C ages we obtained 13C/12C measurements in all the samples (Appendix A7). The δ13C of the organic matter dispersed in three mud samples is 23.7%. This suggests a marine influence (Faure, 1986), which we interpret to indicate prodeltaic muds, with organic matter dated between 28.5 and 30.7 ka. The wood samples show a δ13C between -25.3 and -29.7%. Modern wood from humid climate forests show a δ13C of about -30%, with heavier values usually associated with drier climate (Carrie Maziello, personal communication). The type of plant material could not be distinguished in our samples with confidence, but variations in δ13C suggest contribution from both grasses and woody plants.
Radiocarbon ages, particularly those measured in wood, mollusk shells, and organic-rich mud show numerous age reversals with depth. An age–depth curve was produced by starting from an age within hemipelagic sediments and successively taking the next youngest age above by eliminating the age values where a reversal is observed (Fig. 12, Table 1). Radiocarbon ages in planktonic foraminifera obtained from the intervals of hemipe-lagic fossil-rich mud (at the top of the Series 20 and at the top of Series 70) bracket the main pulse of turbidite sedimentation (Series 30–70) into Basins II and IV. While there is some uncertainty in the age–depth curve, particularly because any specific reworked material age is likely to be older than the depositional age, the relatively large number of samples provides a high-resolution curve with minimum estimates of the average sedimentation rates over each interval.
Specific ages for each seismic horizon were obtained by interpolation, by plotting the horizon depth at the well onto the age–depth curve (Fig. 12; Appendix A2). The difference in interpolated age between each well represents the error in the technique, including errors in seismic–well tie, and difference between depositional age and radiocarbon age. Differences of horizon age between the various wells are of the order of 1000 years or less in most cases. In some instances, such as in rapidly deposited sequences of turbidites, the interpolated age is not considered meaningful because these are “event” beds; nevertheless, the ages above and below bracket the series-boundary age (Fig. 12).
The radiocarbon ages reveal an extremely fast pulse of sediment accumulation in both Basin II and Basin IV in the period between ∼ 24.3 ka and ∼ 15.3 ka. Differences in timing of deposition between Basins II and IV, if any, are generally very subtle and difficult to resolve with the available dating tools (see discussion below). In Basin IV turbidite accumulation between Series 30 and 70 occurred within a period of less than 8000 years, with the bulk of the accumulation occurring during an even shorter period between 20.3 and 15.7 ka.
Summary of Chronostratigraphy
The summary of ages and sedimentation patterns in Basins II and IV is presented in Figures 12 and 13, and is discussed below. The age–depth points used to construct Figure 12 are listed in Table 1. The spatial and temporal pattern of deposition between Basins II, III, and IV is illustrated using a Wheeler diagram (Fig. 14), and key characteristics, sand content, and sedimentation rates are summarized in Table 2.
Series 10 was deposited between ∼ 140 ka and 115 ka and is represented by terrigenous-rich hemipelagic mud. The age at the base of the cored section in Basin IV is less than the H. inversa LAD (Expedition 308 Scientists, 2006; Shumnyk, written communication) and within the W biozone. Linear accumulation rates in Series 10 are high, as much as 7 m/ky in the Basin IV area and 9 m/ky in the Basin II area (Table 2). The high accumulation rates and the high detrital carbonate and dolomite in the sediments together with the freshwater signature in foraminiferal delta;18O are consistent with a strong influx from the Mississippi River, and are similar to rates measured in the late Quaternary for the deepwater region near the Mississippi River mouth (Expedition 308 Scientists, 2006), but much higher than those measured near Bryant Canyon by Tripsanas et al. (2007). Series 10 appears to be slightly thinner in the Basin I area (∼ 50 m), compared with the thickness observed in Basin II (∼ 90 m) and Basin IV (∼ 60 m). The variations in thickness of Series 10 have not been studied in detail (Prather et al., this volume) but likely reflect spatial trends in the redistribution and transport processes of the fine-grained material originating from the Mississippi and other coastal rivers. Series 10 is capped by a fossil-rich hemipelagic mud about 2 m thick that was deposited during the lower part of biozone X, when terrigenous sediment to the area was significantly reduced during the last interglacial (OIS 5e).
Series 20 consists of a terrigenous-rich hemipelagic mud in Basin II and of mostly muddy turbidites in Basin IV, where it forms a low-relief ponded apron. The turbidite interval was deposited between ∼ 115 ka and ∼ 89 ka (OIS 5b–5d). The accumulation rates were relatively low during deposition of these muddy turbidites, not exceeding 1 m/ky in Basin IV and even less in Basin II, where there was no relief to trap sediments. An ∼ 8-m-thick sand-rich unit occurs at the base of this unit. The sand was identified on the well logs at U1320 but was not recovered in core (Fig. 4). The ash-bearing microfossil-rich hemipelagic drape capping this interval (Horizon 29) accumulated at a rate of the order of 0.1 m/ky across the area, and was deposited between ∼ 89 ka, close to the Y–X boundary, and 24.3 ka, when the first turbidite in Basin IV was deposited in the basin center at Site U1320.
We obtained a series of samples in Basin IV cores near the transition between fossil-rich hemipelagic sediments and turbidite deposits with the objective of dating the onset of turbidity-current activity at the boundary between Series 20 and 30 (Figs. 11H, J). The age estimate for the onset of turbidite deposition at the Basin IV margin is slightly younger (23.8 ka) than in the basin center (24.3 ka), which would suggest that the initial flows were confined to the basin center (Figs. 11H, J).
The onset of turbidite deposition for Series 30 in Basins I and II cannot be dated, since no continuous core is available across the Series 20/30 boundary. In Basin II Series 20 is relatively thin and laterally continuous, with no evidence of onlapping reflections in it. The age for Horizon 30 is estimated to be practically the same as in Basin IV, based on age extrapolation (Fig. 12A). In Basin I we rely on seismic stratigraphic correlation, since no age control is available (Fig. 6). In Series 20, some minor thickening and high amplitudes near the Basin I center suggests accumulation of sediment-gravity-flow deposits during the same stratigraphic interval as observed in Basin IV, but since there are no ages in Basin I, we cannot determine whether deposition there started earlier or at the same time. Series 20 in Basin I is capped by a laterally continuous unit with character similar to that observed in Basins II and IV.
Series 30 in Basin IV contains a basal muddy slump deposit which is derived from the eastern basin flank. The slump unit sits on top of the thin turbidites that mark the onset of Series 30. An age reversal (36 ka at 124.5 mbsf) on a foram sample near the top of the slump indicates reworking (Fig. 12B). The slump unit is capped by muddy turbidites forming a low-relief ponded apron (Fig. 4). This apron was deposited between 24.3 and 22.8 ka, based on an interpolated age for Horizon 40 at the top of the series (Fig. 12B). The same interval above horizon 30 in Basin II contains a thick interval of sand-rich deposits (as much as 57 m, ∼ 50% sand:mud; Table 2) estimated to have been deposited before 23.4 ka, based on a radiocarbon age near the top of Series 30 (Fig. 12A). Comparing the age of deposition in this interval is of interest because it provides insight into whether deposition in adjacent basins is coeval or sequential (fill and spill). In Basin IV turbidites in Series 30 are younger than 24.3 ka and older than 22.8 ka, whereas in Basin II the ponded turbidites in Series 30 are older than 23.4 ka, and possibly older than 22.3 ka (the radiocarbon age of the sample immediately above at 59 mbsf; Fig. 12b; Table A2). Thus the ponded aprons in Basin IV and II formed within an overlapping age interval, demonstrating that coeval sedimentation occurred. Horizon 40 near the exit of Basin II is a pronounced erosional surface, confirming the bypass of sediment into Basin IV at the end of Series 30, when Basin II was filled to the spill point (Fig. 5; Prather et al., this volume). The maximum linear accumulation rate in Basin II was almost three times that in Basin IV during Series 30 deposition, and the average sand percentage is about five times higher in Basin II compared with Basin IV (Table 2).
During Series 40 sediment continued to accumulate at a high rate in Basin II. A perched apron was deposited there between 23.4 and 19.5 ka. Most radiocarbon ages in the reworked material in this interval in Basin II and in Basin IV occur in the range between 20.4 and 22.3 ka (Fig. 12; Table 2). During deposition of this series, relief in Basin II had diminished as a result of basin infill by the underlying series, and became too small to accommodate all of the sediment entering that basin. As a result, significant amounts of sand bypassed Basin II and entered Basin IV via the western channel (Figs. 2, 14; Prather et al., this volume), forming thick sand beds that extend laterally across nearly the entire basin (Fig. 4). The average sand percentage in Basin IV increases to 35% during Series 40, compared with 10% in the underlying series, whereas the sand percentage remains about the same at 50–55% in Basin II (Table 2).
Series 50 and 60 are combined, since the accumulation rates are so high in this sand-rich interval that it is not possible to ascertain their ages separately. In Basin IV Series 50 and 60 sediments were deposited between 20.3 and 19.2 ka, with as much as 55 m of sandrich turbidites accumulating in 1100 years or less (Figs. 12, 13). The equivalent interval in Basin II appears slightly younger, with Series 50 and younger sediments in that basin containing ages younger than 19.5 ka. However, we were not able to retrieve sufficient organic material in the few core samples available from this interval, so the slight differences in age probably are not significant. During Series 50 and 60 there is a switch in sand content between the basins, with Basin IV increasing to 60% and exceeding Basin II (sand content 45%). Maximum linear accumulation rates in Basin IV reach ∼ 34 m/ky vs. ∼ 20 m/ky for Basin II (Table 2).
Series 70 began at ∼ 19.2 ka in Basin IV, based on age–depth interpolation, and it appears to have accumulated quite rapidly as well, with most reworked material obtained in the turbidites dated between 18.2 and 19.2 ka. The onset of deposition of Series 70 in Basin II is estimated to have occurred at practically the same time (19.4 ka) as in Basin IV (Figs. 12, 14; Table 2).
Series 70 shows a dramatic contrast in sedimentation between Basins II and IV (Prather et al., this volume). A leveed-channel system becomes established in Basins II and III, forming a conduit for sediment to reach Basin IV. Most of the sands bypass the upper basins and form a high-relief apron in Basin IV (Fig. 4). The accumulation rates near the thickest part of this apron (about 100 m based on seismic) reach as much as 29 m/ky, with a very high sand:mud content, as inferred from the sand-rich deposits recovered in the distal portions of the apron at Site U1320 (Fig. 13; Table 2) and in more proximal apron positions by piston cores (Mallarino et al., 2006).
The age of cessation of turbidite deposition in Basin IV is determined to have occurred at 15.7 ka, based on radiocarbon ages on forams deposited in the hemipelagic interval just above the turbidites (Fig. 11A). The observed cessation of turbidite deposition in the Basin IV margin (Fig. 11I) appears to have occurred earlier, at about 17.2 ka, suggesting that turbidity-current discharge (and the thickness of the currents) began to decrease already after 17 ka. The cessation of turbidite deposition in Basin II is slightly later than in Basin IV, with ages in the foram-rich clay of Series 79–80 being as young as 15.3 ka. However, the cessation age in Basin II is less certain because the short surface cores from the Rudder wells in Basin II do not capture the transition from turbidite deposition to hemipelagic deposition. Sedimentation in the shelf-margin delta continued after 14.8 ka (Wellner et al., 2004), after cessation of turbidite influx into Basins IV and II, indicating a rapid landward shift in the depocenter after ∼ 15.7 ka.
Stratigraphy of Shelf Deposits
The deposits of the Trinity, Sabine, Brazos, and other coastal rivers have been the subject of numerous studies over the last several decades, and as a result we now have a good picture of where the main deposits are and when they accumulated (Winker, 1979; Morton and Suter, 1996; Anderson et al., 1996; Anderson et al., 2003; Anderson et al., 2004; Anderson, 2005; Anderson and Rodriguez, 2000; Thomas and Anderson, 1994; Abdullah et al., 2004; Fraticelli and Anderson, 2003; Wellner et al., 2004; Blum and Aslan, 2006). The key results and events described by these studies that are related to the sedimentation in the slope basins are summarized briefly here.
After the end of the last interglacial, the deltas of the Colorado, Brazos, Trinity, Sabine, and rivers from western Louisiana (Red River drainage?) advanced over the shelf as sea level began to fall. The deltas of the larger rivers advanced the farthest, and during OIS 3 the Colorado, Brazos, and western Louisiana deltas (WLD) arrived close to the shelf margin. During OIS 3 the Brazos and the western Louisiana deltas were located, respectively, to the west and to the east of Basin I. The Colorado delta was even farther to the west and did not influence sedimentation in the B–T slope system. As the Brazos and western Louisiana systems reached the shelf margin, major avulsions occurred, with the respective rivers shifting to the east. For the Brazos the avulsion timing is bracketed between ∼ 36 ka (33,720 14C years) and 20 ka (16,970 14C years) according to Fraticelli and Anderson (2003). Wellner et al. (2004) timed the abandonment of the WLD after ∼ 36 ka as well (33,000 14C years, Wellner et al., 2004). The avulsion of the Brazos River resulted in the Trinity–Sabine valley capturing the Brazos drainage, forming the Trinity–Sabine–Brazos valley (Figs. 1, 2; Wellner et al., 2004).
In contrast to the larger rivers to the west and east, the Trinity, Sabine, and other smaller rivers did not form significant deltas as sea level fell (Anderson et al., 2004). Instead, during the sea-level fall the valley incised and widened, and progressively extended across the shelf. Terraces in the valley indicate that the incision occurred in a stepwise fashion, cutting the pre-existing shelf deposits by as much as 40 m (Thomas and Anderson, 1994). The well-defined incision on the inner shelf became shallower and wider towards the middle and outer shelf (Simms et al., 2007). The lower one-half of the incised valley is filled with fluvial deposits, and fluvial deposits are also found in higher-standing terraces (Thomas and Anderson, 1994). The fluvial terraces on the inner shelf were correlated with the Deweyville age (OIS 5c–5a) terraces exposed onshore (Thomas and Anderson, 1994), whereas the age of the lower fluvial deposits infilling the valley is not well constrained. The upper portion of the valley is filled with younger shallow marine deposits (Thomas and Anderson, 1994). The Trinity–Sabine valley over the shelf represents the convergence of several other smaller coastal rivers, including the Trinity, the Sabine, the Neches, and the Calcasieu rivers (Simms et al., 2006; Simms et al., 2007a) (Fig. 1).
Once the Brazos and the Trinity–Sabine valleys joined, sediment input to the area increased and a significant delta reached the shelf margin, the Trinity–Sabine–Brazos (TSB) shelf-margin delta (Fig. 2). Wellner et al. (2004) obtained an age below the base of the TSB delta (near their sequence boundary 2), indicating that the delta is younger than ∼ 31 ka (26,360 14C years). This possibly constrains the age of avulsion to an age younger than the ∼ 36 ka age presented by Fraticelli and Anderson (2003). The TSB Delta built a wedge near the present-day shelf edge, with maximum thickness reaching ∼ 120 m and thinning both landward and seaward. The delta extends for ∼ 60 km along strike and ∼ 20 km across, based on the maps and seismic profiles of Fraticelli and Anderson (2003), Wellner et al. (2004), and Anderson (2005) (Fig. 2). Prodelta deposits extended onto the slope, and laterally, with a mud belt extending towards the west, where it lapped against the OIS-3 Brazos Delta (Fraticelli and Anderson, 2003; Anderson et al., 2003). As sea level rose, the high sediment discharge resulted in continued progradation and aggradation of the TSB delta, until eventually the Brazos River switched back to the east (Wellner et al., 2004; Abdullah et al., 2004). Radiocarbon ages in the TSB delta indicate that about half of the shelf-margin delta accumulated after ∼ 14.8 ka (13,025 14C years; Wellner et al., 2004). As a result of much-diminished discharges in the Trinity–Sabine system without the contribution from the Brazos, the valley remained underfilled as sea level rose further (Anderson et al., 2004)
Major deposition in Basin I started with Series 30 at 24.3 ka or earlier, and is, therefore, largely coeval with sedimentation on the TSB shelf-margin delta. Deltaic sedimentation at the outer shelf continued after 15 ka, when turbidite sedimentation in Basins II and IV had already ceased. In addition to sediment delivery to Basin I, the TSB delta also delivered sediment into Basin A to the west of Basin I and to a minor extent into the Bowie Basin to the east of Basin I (Fig. 2).
Discussion—Implications for Deepwater Depositional Models
A high-resolution chronostratigraphic framework for the Brazos–Trinity depositional system derived by integrating a range of stratigraphic methods and data, including seismic stratigraphy, biostratigraphy, stable isotopes and radioisotopes, tephrostratigraphy, and lithostratigraphy provides a means to correlate events within the various parts of the Brazos–Trinity depositional system, and with the well-established chronology of fluctuations in sea level and climatic for the late Quaternary (e.g., Lambeck and Chappell, 2001; Balsillie and Donoghue, 2004; Clark et al., 2009). This high-resolution stratigraphy allows us to link cause and effect, and forcing with depositional-system response. In this section we discuss the implications of our results; in particular we focus our discussion on three main aspects: (1) depositional models of deepwater slope sedimentation, in particular the evidence for sequential (i.e., fill and spill) vs. coeval basin fill; (2) sediment transfer budget from the fluvial and shallow-marine domain to the deepwater domain across this depositional system; and (3) influence of sea-level changes and delta dynamics on the depositional record on the slope.
Fill and Spill Versus Coeval Sedimentation
Winker (1996) and Satterfield and Behrens (1990) initially, and later Beaubouef and Friedmann (2000), suggested a model in which basins infilled progressively from source to sink, the so-called fill-and-spill model. Later Badalini et al. (2000) suggested that the basins could fill coevally, but with partitioning of sand and mud between basins because of the interaction between a stratified flow—where coarser grains are carried lower in the flow—and basin sill topography, the so-called flow-stripping model. The chronostratigraphy obtained here allows testing which of these hypotheses, fill and spill vs. coeval fill, is at work in this system.
Series 10 deposition of muddy sediments shows variations in thickness across the entire slope area, but overall the deposits drape the pre-existing topography. There are two possible explanations: either deposition is dominantly hemipelagic, originating from mud-rich surface plumes, or sediment was carried near the bottom by low-density gravity flows or nepheloid layers over a smoother basin topography where Basins I–IV were not yet developed. The high rates of sedimentation observed in this interval (Table 2) suggest that sedimentation may have overtaken basin subsidence. There are only rare silt layers and no sand layers in this interval, suggesting deposition from surface “rain” rather than by bottom-hugging sediment gravity flows. We know that at a later time, during Series 20, Basin II was not developed, because turbidites which traversed that basin do not show evidence of onlapping reflections, whereas the coeval deposits in Basins I and IV do. So if Basin II topography did not exist during Series 20, most likely the same was true during Series 10 at Basin II, and, by inference, during Series 10 for Basins I and IV as well.
Subsidence in Basins I and IV must have occurred during the end of Series 10 and continued into Series 20, as recorded by the onlapping fill in those basins, which is dated between ∼ 115 ka and ∼ 89 ka in Basin IV (Figs. 4, 6). Series 20 in Basin II shows no significant accumulation and no deposition of sandy sediments during this time, which resulted from a lack of significant basin topography (accommodation). Turbidity currents largely passed through Basin II without depositing sand or even eroding a significant channel, and leaving only a thin muddy unit. The minor amount of sediment thickening near the Basin I center indicates that some basin relief must have existed then (Fig. 6), although probably not sufficient to completely trap sediment flows that reached Basin IV during this time. Unfortunately there are no core samples to date the fill of Basin I, so while we observe that sediment gravity flows were depositing there during Series 20 we cannot determine whether that basin was filled first before sediment was carried through Basin II to Basin IV. Sedimentation during Series 20 demonstrates a strong control of basin topography on infill history, with a reverse pattern of “spill and then fill” between Basins II and IV. Subsidence history in these salt basins plays a key role in sediment trapping, and cannot be assumed to be static with respect to sedimentation, even over relatively short time intervals, as was generally assumed in the previous fill-and-spill interpretations.
During the late stages of deposition of Series 20, major subsidence resulted in building of significant relief in all basins (Prather et al., this volume). With the onset of Series 30 at about 24.3 ka, muddy turbidites were arriving in Basin IV and onlap-ping the edges of the basin. The exact onset in Basins I and II is not known, but a radiocarbon age in reworked material at the top of Series 30 in Basin II demonstrates age overlap between the ponded apron in Basin IV and the ponded apron in Basin II. The fact that the age of the Basin IV turbidites in Series 30 (younger than 24.3 ka and older than 22.8 ka) overlaps in age with the those in Basin II (older than 23.4 ka, possibly older than 22.3 ka) demonstrates that coeval sedimentation occurred, at least for the basal part of the Series. Until we obtain a direct age of the first turbidite deposit in Basin II we cannot rule out a scenario where there was some infill in Basin II older than 24.3 ka without a coeval unit in Basin IV. However, the ages prove that at least part of the history in Series 20 involved coeval fill. The maximum relief of Basin II at the onset of Series 30 is interpreted to correspond to the maximum thickness of the ponded apron in Basin II (∼ 60 m; Fig. 5, Table 2). This implies that turbidity currents must have exceeded this thickness in order for mud to bypass Basin II (Toniolo et al., 2006)
For Series 40 and younger, the age of sedimentation in each basin is practically coeval, all occurring between ∼ 23 ka and ∼ 15.3 ka. The high sedimentation rates suggest that basin subsidence during this period either was less active or occurred at rates that are well exceeded by the rates of sediment accumulation. Submillennial-scale dating would be required to resolve whether fill and spill occurred during this interval; however, evidence from the seismic and lithostratigraphy shows that partitioning of sands and muds in the various basins during these times was a strong function of basin relief and degree of channelization across the basins (Prather et al., this volume). Once Basin II was infilled at the end of Series 30, bypass of sands into Basin IV increased significantly, while sand accumulation was still relatively high in Basin II. The situation reversed once a continuous channel was established in Basin II during Series 70, at which time mostly fine-grained overbank sediments were retained in Basin II while sand content in Basin IV increased dramatically (Table 2, Fig. 14).
In the case of the Brazos–Trinity depositional system, rates of basin subsidence appear to have been significantly higher than previously thought, and their effect on sediment partitioning and distribution has been very significant. Basin II formed entirely during the latter part of Series 20 deposition (> 60 m subsidence in ∼ 60 ky), and Basin IV must have subsided significantly after deposition of Series 10 (> 180 m in ∼ 100 ky; Prather et al., this volume). These average subsidence rates, of the order of a few meters per thousand years, are lower than the ultra-high linear accumulation rates observed for the sandy turbidite intervals (Table 2), but can generate significant accommodation, particularly during periods of condensed sedimentation (e.g., Series 19, 29, and 79).
Sediment Flux from Source to Sink
Having obtained constraints on the timing of infill of the deepwater system, we can now proceed to compute the volumes of sediment delivered and their temporal changes. After evaluating the sediment flux in deepwater we attempt to estimate a sediment budget for the entire source-to-sink system using information available from the literature for the updip portions of the system.
Sediment Volumes in the Slope Basins
We use the maps illustrated in Prather et al. (this volume) for the various basins along the system and estimate the volume of sediment deposited and the volumetric accumulation rates in each time interval and in each basin (Figs. 15A, B). There is uncertainty about the exact age of Basin I sedimentation. The bulk of the sediment volume lies above Horizon 30 (Fig. 6), but we were unable to confidently trace horizons 40–70 into Basin I, and as a result we cannot determine how the various units are partitioned in that basin. The basin configuration, showing on-lapping reflections for all of Series 30, suggests that most of the subsidence occurred during the upper part of Series 20, similarly to the other basins. This, and the chaotic nature of the ponded apron at the base of Series 30 (Fig. 6; Pirmez et al., 2000), suggests that at least some of Series 30 was completely trapped in Basin I. For computation of volumetric accumulation rates we assume that after 22 ka (Series 40, 50, 60, and 70) all of the sediment entering the system bypassed Basin I through the channels clearly visible towards the top of the basin fill near the exit of the basin (Figs. 2, 6; Pirmez et al., 2000).
A portion of the sediments in Basin I certainly is younger than 22 ka, particularly the overbank deposits in the vicinity of the channels. Only a very small amount of shelf-margin prodelta deposits is included in the outline of Basin I used for volumetric calculations (Fig. 2).
Volumes and Fluxes in Deepwater
The total sediment volume accumulated in Basins I–IV since ∼ 115 ka is about 62 km3, with most of it (∼ 49 km3) being deposited since 24.3 ka (Series 30 and younger). Of the total volume accumulated since 24.3 ka, Basin I contains about 83%, and Basins II and III combined and Basin IV contain 5% and 12% of the total volume, respectively.
The total volume of sediment delivered to Basins I–IV and the total accumulation rate show a dramatic increase after 24 ka, from ∼ 1.5 ×106 metric tons/year during Series 10–20 to 5.5 ×107 tons/year, an increase of 40-fold (Fig. 15). This sediment flux assumes no Series 40–70 deposited in Basin I after the period 24– 22 ka as discussed above and is, therefore, a high-end estimate. Subsequently, the accumulation rate into the slope system oscillated between ∼ 2 and 5 ×106 tons/year. Considering only Basins II–IV, the sediment accumulation rate appears to increase over time, from a low of 1.6 ×106 tons/year during the 24– 22 ka interval, peaking in the interval 18–19.5 ka at 5 ×106 tons/year, and then decreasing slightly after that, before finally dropping to hemipelagic rates (less than about 1 ×105 tons/year) with the abandonment of Basin II after ∼ 15.3 ka.
The present-day river discharges provide a measure against which to compare the sediment accumulated in the terminal portion of the source-to-sink system and help set the stage for evaluating the mechanisms of sediment transfer across the system. The present-day discharge of the three major rivers feeding this depositional system have been significantly affected by dam construction and land-use changes over the last 100 years (Dunn and Raines, 2001; Phillips et al., 2004). Estimates of river discharges by different studies vary quite widely because of the methods used and the particular reference gauge station. The Brazos River, by far the largest of the three main rivers associated with the B–T depositional system, has an average annual sediment discharge as low as 2.2 ×106 tons/year (Dunn and Raines, 2001) to as high as 3.5 ×107 tons/year (Holeman, 1968). Simms et al. (2006) and Meade and Parker (1984) give average annual sediment discharges for the Brazos River of 1.6 ×107 tons/year and 1.1 ×107 tons/year, respectively. The average annual sediment discharge of the Trinity River is ∼ 7 ×104 tons/year (Phillips et al., 2001), but an estimate of the pre-dam values give much higher discharge of ∼ 6.2 ×106 tons/year (Simms et al., 2007). The average annual sediment discharge of the smaller Sabine River ranges from 7.3 ×105 tons/year (Holeman, 1968) to 7.5 ×105 tons/year (Anderson et al., 2004). Given these ranges, the total present-day average sediment discharge of the combined Trinity–Sabine–Brazos Rivers (T–S– B) is estimated at ∼ 2.3 ×107 tons/year, but possibly as high as ∼ 4.2 ×107 tons/year.
Sediment Flux in Deepwater Versus River Discharge
We estimate a sediment flux into the slope basins of 1.4 ×106 tons/year during accumulation of Series 20, six times smaller than the present-day combined discharge of the Trinity and Sabine (T–S) rivers. The T–S delta never reached the shelf margin (Anderson, 2005), and sediment gravity flows that reach the slope basins most likely originated from failure of previously deposited shelf-margin sediments. During accumulation of Series 30, when the T–S–B rivers had joined into one valley, the estimated sediment flux delivered to the slope basins could have been as high as two to four times the present-day combined river discharge. This is a high-case estimate, given that we assume all of Basin I sediment volumes to have accumulated by the end of Series 30, and it ignores the probable initiation of infill of Basin I somewhat earlier than 24 ka. During accumulation of Series 40 and younger, the sediment flux to the slope was five to ten times smaller than the present-day combined sediment discharge of the T–S–B rivers. This disparity would be smaller if we included some of the sedimentation of Basin I into this period of time.
Before reaching any conclusions from this initial comparison, we need to examine in more detail each element of the sediment budget from source to sink.
Key components of the sediment budget in this source-to-sink system during the period 24.3 ka to 15.3 ka include: (1) sediment delivered from the drainage basin through the fluvial system, (2) sediment generated locally by erosion in various parts within the system, (3) sediment accumulated on the shelf and outer-shelf deltas, and (4) sediment deposited in deepwater.
The deepwater sediment sinks are reasonably well constrained by the work in this and the companion paper by Prather et al. (this volume), and is estimated at ∼ 50 km3 for the period between 24.3 and 15.3 ka for Basins I–IV. Possible additional sinks include sediment delivered by the T–S–B system to other slope basins during the same interval of time. From the 3D seismic data we estimate that about 20 km3 of sediment accumulated in Basin A, but the age of these deposits is uncertain. The deposits are acoustically chaotic and have a rough surface topography, suggesting that they formed by mass wasting of the delta front (Fig. 2; Wellner et al., 2004 their Figure 13). The Bowie Basin did not receive a significant amount of sediment from the T–S–B system during this time interval, judging from seismic geometries.
There are some local sources of sediment within the deepwater system, including sediment removed by channel incision across the basin sills and mass wasting of the basin flanks. Sediment eroded from the sill between Basins I and II represent about 5% of the total volume accumulated in Basin II, whereas sediment eroded in the two channels connecting Basins II and IV represent about 2% of the total sediment volume accumulated in Basin IV. This represents less than 1% of the total sediment deposited in Basins I–IV. Including basin-flank slumps (e.g., immediately above Horizon 30 in Basin IV; Fig. 4), locally sourced sediment in deepwater represents less than 5% of the total deepwater sink (< 3 km3).
The volume accumulated by the Trinity–Sabine–Brazos on the shelf and shelf-margin deltas has not been published. Assuming a simple triangular wedge with maximum thickness of 120 m (150 ms) along the entire strike extent of the TSB shelf-margin delta, and an average porosity of 40%, we obtain a sediment volume of about 45 km3. According to Wellner et al. (1994), most of this sediment accumulated after 15 ka. We estimate that at least 20–25 km3, or about half of the total, was deposited after Basins II– IV ceased receiving sediment gravity flows. It is also possible that some of this volume accumulated earlier, perhaps starting as early as 31 ka (Wellner et al., 2004). A maximum value of ∼ 20–25 km3 is estimated for the shelf-margin delta “sink” for the period between 24.3 ka and 15.3 ka.
Using a discharge of 3 ×107 tons/year as a basis, the combined T–S–B Rivers would contribute about 11 km3/ky, or about 100 km3 of sediment in the period 24 to 15 ka. River discharges in the past are difficult to estimate independently. It has been argued that discharges of Texas rivers were higher, based on evidence from fluvial channel morphology on Deweyville-age terraces (discussion in Blum and Aslan, 2006) or from sediment-budget estimates (Anderson et al, 2004). Anderson et al. (2004) indicate that sediment discharges for the Brazos River during past high stands of sea level ranged between 0.3 and 3.2 km3/year, or about 30–300 times higher than present day. It has also been argued, based on climate models, that cooler climates along the Texas coast during glaciations would have resulted in lower overall discharges, by as much as 10–40% less during glacial vs. intergla-cial times (e.g., Blum and Garvin, 2010)
The sediment volumes removed by erosion on the shelf have been estimated by Anderson et al. (1996) to be of the order of 120 km3, primarily a result of the formation of sequence boundary 2 (SB2), formed during OIS 2. Examination of the maps of the incised-valley depth (Thomas and Anderson, 1994; their Figure 7) indicates a volume of about 40 km3 just for the inner shelf portion of the Trinity–Sabine valley (∼ 180 km total length of valleys, 10 km width, 35 m depth, and assuming 40% porosity). Anderson et al. (1996), and Anderson et al. (2004) indicate that the eroded sediment from the incised valley bypassed the shelf and was delivered to deepwater. Wellner et al. (2004) suggest that most of the sediment yield from valley incision accumulated in the TSB shelf-margin delta.
Balancing the Sediment Budget
The sediment-budget items for the period ∼ 24–15 ka can be summarized as follows, with values in km3, positive (+) for sources, and (-) for sinks: river input (+100), yield from incised-valley erosion (+120), shelf-margin-delta accumulation (-25), Basins I–IV (-50), local sources in deepwater (+3), and Basin A (-20), resulting in a net +128 km3 of excess sediment that cannot be accounted for.
We believe that errors in these estimates are significantly smaller than the excess sediment balance. For Basin I and the TSB shelf-margin delta, estimates of sediment volume are probably a maximum, given that sedimentation likely started a bit earlier than 24 ka based on the ages of the TSB delta (as early as 31 ka; Wellner et al., 2004). Basin A sediments also may have been deposited prior to 24 ka, and thus represent a maximum sink. The volume of the delta is approximate, given that iso-pachs have not been mapped in detail, but we tried to make it as large as possible given the available published images—for instance, by assuming that the maximum thickness of 120 m extended laterally over the strike extent of 60 km. Further studies of the shelf margin with detailed mapping of the sequences would certainly help refine the sediment volumes in the middle portion of the source-to-sink system. However, even if we doubled the volume of the shelf-margin delta, an unlikely scenario in our view, we would still have an excess of > 100 km3 of sediment. There are only a few possible explanations for the apparent excess of sediment.
First, the sediment yield from the river could have been smaller than at the present day. This would be in agreement with what has been proposed by Blum and Aslan (2006) and Blum and Garvin (2010). According to the climate modeling used by Blum and Garvin (2010), reduced temperatures during glacial intervals would have reduced the sediment discharge of the T–S–B rivers by as much as 10–40% (10 to 40 km3).
Second, the contribution of sediment exported from valley incision may be an overestimate because assuming that the total volume of the incised valley eroded was exported away may be incorrect (Blum and Aslan, 2006; Strong and Paola, 2008). The lower ∼ 50% of the incised valley is filled with fluvial sediment, in addition to the fluvial deposits preserved in older terraces (Thomas and Anderson, 1994). The results of recent flume studies and theoretical considerations suggest that the bulk of the fluvial infill is coeval with valley incision, with most of the accumulation occurring during lateral river migration and associated valley widening (Strong and Paola, 2008; Blum and Garvin, 2010). As a result, the additional sediment yield from incised-valley formation would be small. If we further assume that overall discharges were reduced during glacial periods, the total sediment discharge, even during periods of rapid valley widening, would likely be less than at the present day. If these hypotheses are correct, as much as 80–100% of the sediment yield from valley incision may not need to be accounted for, reducing the contribution from valley incision to 0–25 km3.
In summary, the revised budget elements, considering the above discussion, in km3 of sediment, are: river input (+60 to +90), yield from incised-valley erosion (+0 to +25), local deepwater sources (+3), shelf-margin-delta accumulation (-25), Basins I–IV (-50), and Basin A (-20), with a net balance ranging from a deficit of 32 km3 to an excess of 23 km3.
These results indicate that: in order to balance the sediment budget for the period 24-15 ka (1) sediment yield from the incised valley likely was much smaller than presently assumed; (2) the sediment discharge from the coastal rivers during glacial time also was smaller than at present; or (3) a combination of these factors.
The period 24–22 ka, near the last glacial maximum, was the only time when the sediment discharge to deepwater may have exceeded the present day T–S–B river discharge, based on our sediment-flux estimates from the slope-basin accumulation (Fig. 15). This time also coincides approximately with the avulsion of the Brazos River to the west, as it abandoned its OIS-3 delta lobe and shifted to the east, joining the Trinity–Sabine valley. While the exact timing of river avulsion is not known, it occurred between 31 ka (base of TSB delta; Wellner et al., 2004) and 24.3 ka (the onset of turbidity currents in Basin IV). Sediment yield from the incised-valley erosion could have been enhanced during this period as the Brazos River adjusted to the new drainage-basin configuration. After 15 ka, the fluvial sediment was captured in the shelf-margin delta (∼ 25 km3), within the incised valleys over the flooding shelf, as fluvial and shallow marine deposits, and along the coastal areas adjacent to the fluvial valley, forming sand banks (Thomas and Anderson, 1994).
Impact of Sea-Level Fluctuations and Deltaic Dynamics on Basin-Fill History
The succession in Basins II and IV can be correlated with independently dated sea-level curves (Fig. 16). In Figure 16A we plot a longer time span illustrating the various sedimentary units against the sea-level curve of Lambeck and Chappell (2001), and in Figure 16B we compare the depositional record with the sea-level curve since the last glacial maximum (Balsillie and Donoghue, 2004; Clark et al., 2009).
During the periods of highest sea level, such as during OIS 5e and OIS 1, there was a general paucity of sediment input into the deepwater basins, as observed globally in ocean basins with wide continental shelves. During OIS 6, the area received muddy sediments that accumulated at relatively high rates of sediment accumulation (up to 7 m/ky). No sand was recovered in Basin II and IV wells during this interval. Sand was probably restricted to the more proximal basins near the Mississippi delta to the northeast during this time, or to the west off the Colorado and Brazos river deltas. During OIS 5d–5c, sea level fell by several tens of meters and sediment gravity flows reached Basin IV while bypassing Basin II, which did not exist as a basin then. Sediment accumulation rates of Series 20 during that period were relatively low (Fig. 15), probably a reflection of the fact that the deltas had not yet reached the shelf edge (Anderson et al., 2004). The source of the sediments that formed these early turbidite deposits in Basin IV are most likely related to failure of shelf-edge deposits formed during previous lowstands of sea level.
A long period of very low sediment input (Series 29; Fig. 16), is observed from OIS 5a through OIS 3 in both Basins II and IV. This occurred despite significant sea-level variations observed during this period. Shelf sedimentation at this time was occurring to the west of Basin I in the Brazos and Colorado deltas and to the east of Basin I in the western Louisiana deltas (Anderson et al., 2004; Anderson, 2005). The smaller Trinity–Sabine river system did not build a significant delta at this time. Fraticelli and Anderson (2003) observe that the slope to the west of Basin I also contains a long period of condensed sedimentation (130–47 ka), reflecting the time it took for the Brazos delta to reach the shelf margin and begin influencing upper-slope sedimentation.
The onset of turbidity-current input into the slope basins appears to correlate well with the onset of OIS 2, and could possibly be somewhat earlier in Basin I. The onset of turbidity-current influx into Basins II and IV occurred about 2000 years before the Last Glacial Maximum (LGM, 22 ka) but several thousand years after sea level dropped to the levels observed in the LGM (Fig. 16A; Clark et al., 2009). During the latter part of OIS 2, sea level gradually rose and at the same time sediment accumulation into Basins II and IV progressively increased until about 18 ka (Fig. 15B). Sediment accumulation in Basins II and IV continued until about 15.3 ka, when it dropped dramatically, about 1000 years or more before the rapid rise in sea level associated with meltwater pulse 1A (Fig. 16B). After ∼ 15 ka sedimentation shifted entirely to the shelf-margin delta, with prodelta deposits reaching Basin I, and progressively shifting landwards as sea level continued to rise (Simms et al., 2007b; Wellner et al., 2004).
Our results revise and augment the previous chronostratigra-phy developed by Mallarino et al. (2006); in particular, we observe that a pause of sediment influx (Horizon 29) occurred over a much longer period and extended through OIS 3. Significant sediment accumulation in Basins II and IV in the new chrono-stratigraphic framework occurred over a very short period of time (∼ 8600 years between 24.3 ka and 15.7 ka). Further, we show that initial influx of sandy turbidity currents into the basin occurred during OIS 5d–5c, earlier than originally inferred by Mallarino et al. (2006).
We observe that when dated with such a high degree of precision, the sediment influx into the deepwater portion of the Brazos–Trinity depositional system was modulated by the large-scale variations in sea level, with sedimentation ceasing during very high sea-level stands and resuming at low sea level. However, in detail it is seen that variations in sediment influx into the deepwater basins cannot be linked directly to fluctuations in sea level. At the time of the relatively large drops in sea level observed during OIS 4 or at the end of OIS 3, or at the time of rapid sea-level rise such as observed during meltwater pulses 1A, there was no coeval change in the sediment flux. Sedimentation also continued to occur in the latter half of OIS 2, with a significant increase in sediment flux despite the relatively rapid rise in sea level at that time (MWP 19k; Fig 16B). A corollary to this is that high-frequency sea-level changes cannot be directly inferred from the sedimentary record by simply examining lithostratigraphic variations, such as changes in sand content. The interpretations need to be taken within a regional context, and take into account fluvial–deltaic sedimentation on the shelf, and basin subsidence.
As noted by Anderson et al. (1996), deltaic sedimentation, including sediment flux and lateral depocenter shifts, is an important control in delivering sediment to deepwater. In the case of Basins II and IV, the evidence suggests that these controls can overwhelm the high-frequency signal observed in the slope-basin fill. Whether sediments enter the deepwater system at the shelf edge depends in part on sea level, because it takes time for the deltas to reach the shelf edge even after sea level drops significantly (Anderson, 2005; Anderson et al., 1996). Once the deltas begin approaching the shelf edge, sediment is delivered into the Brazos–Trinity system only if the deltas occupy a narrow region at the head of Basin I. This is particularly significant at the shelf margin, where delta systems tend to elongate along strike (Anderson et al., 2004). Fault-induced subsidence and river avulsion play a role in determining where a delta lobe will be located. Relatively large sediment discharge, such as for the Brazos River, favors frequent avulsions, and thus less time for the river to “lock” in place and feed an individual string of deepwater basins. Faults near the shelf edge display topography on the seafloor, indicating recent activity, and thus likely influenced the accumulation of the shelf-margin delta and routing of sediment to the slope (Fig. 2). The entry area immediately updip of Basin I, for instance, occurs near the relay zone between two sets of opposing normal faults (E–W and NE–SW), possibly forming a passage where accommodation was diminished. Depending on accommodation for deltaic sediments on the shelf and the location of the delta front, the B–T slope basins may not receive any significant influx of sediment, even at very low sea level.
A detailed chronostratigraphic framework was developed for the Brazos–Trinity depositional system by integrating seismic stratigraphy developed from very high-resolution 2D and exploration 3D seismic data, biostratigraphy using calcareous nannofossils and planktonic forams, tephrostratigraphy, stable-isotope stratigraphy, radiocarbon dating, and lithostratigra-phy. We show that a detailed age model with millennial-scale resolution can be built in an environment with ultra-high accumulation rates.
A total of about 62 km3 of sediment derived from the Trinity– Sabine–Brazos rivers accumulated in Basins I–IV since OIS 5e. Prior to that, during OIS 6, muddy sediments derived from the Mississippi River delta formed a blanket of sediment over the B– T system, which accumulated at relatively high rates of 5–7 m/ky. As sea level fell about 50 m from the peak during OIS 5e, sediment derived from shelf-edge failures reached Basin I and turbidity currents traveled across Basin II to reach Basin IV, forming a muddy apron there. At about 24.3 ka, after a long period of sediment starvation, sediments from the Trinity–Sabine–Brazos shelf-margin delta deposited a sandy apron within a ponded Basin II. Soon after the initial relief in Basin II began to decrease by infilling, muddy sediments were reaching Basin IV. At about 23.8 ka, sandy sediments began to reach Basin IV and formed low-relief sandy aprons there. At about 19.3 ka, Basin II was almost completely filled up and a network of leveed channels carried sediment trough to Basin IV, depositing a sand-rich, high-relief apron there, while leaving mostly muddy sediments on the levees of Basins II and III. At about 15.7 ka, prior to the rapid rise in sea level during Melt-Water Pulse 1A, Basin IV was no longer receiving turbidity currents. Soon after, at ∼ 15.3 ka, Basin II ceased receiving turbidity currents, and after 15 ka, sedimentation was confined to the shelf-margin delta, with prodelta sediments extending onto the upper slope.
The well-dated slope succession shows that basin-fill models cannot assume that basin morphology is static with respect to sedimentation, since rates of basin subsidence, of the order of a few meters per thousand years, are comparable to rates of sedimentation. While still one order of magnitude smaller than the rapid sedimentation pulses observed in these basins, subsidence can lead to the formation of sufficient topography to influence sedimentation patterns over intervals of sedimentation pauses caused by fifth-order sea-level cycles (∼ 10–100 ky) or even delta-lobe avulsion cycles (∼ 5–10 ky). As a result, the basic fill-and-spill model derived from seismic stratigraphic observations in the B–T system is overly simplistic. For the early part of the basin-fill history (Series 20), the oldest fill occurred farther seaward as a result of Basin II not having significant topographic expression. We show that in at least one case (Series 30) coeval sedimentation occurred in Basins II and IV and that sand was preferentially trapped in Basin II while mud accumulated in Basin IV. In the younger series, Basin II relief became more subdued, and sediment accumulation was essentially coeval in both basins. Basin I was probably infilled during and after Series 30, i.e., over the same time interval as the other basins.
The sediment flux into the deepwater basins over the period 24.3 to 15 ka varied between 1.6 ×106 and 5.5 ×107 tons/year, and appeared to have exceeded the present-day discharge over only a short period around 24–22 ka when the Brazos River drainage was captured by the Trinity–Sabine valley. The sediment budget for the period 24–15 ka includes ∼ 45–50 km3 of sediment deposited in Basins I–IV, ∼ 20 km3 in Basin A to the west of Basin I, and ∼ 25 km3 of sediment deposited in the shelf-margin delta. We can demonstrate a balanced sediment budget if the discharge of the Trinity–Sabine–Brazos River system was ∼ 10–40% smaller than today, contributing 60–90 km3; with an additional contribution from fluvial valley incision over the shelf, ranging from ∼ 0 to 25 km3.
Sediment flux into the deepwater basins II and IV at the end of the source-to-sink system appear to increase over time, despite the fact that sea level was rising overall during the same period between the LGM and the end of basin infill at 15.7 ka. This lagged response can be attributed to the time it takes for the deltas to reach the shelf edge, but it also depends on the dynamics of delta-lobe development in delivering the sediment to the head of the deepwater system. Once the delta reaches the head of the deepwater system and advances over the paleo–shelf edge the sediment flux is extremely high, with linear sedimentation rates at the basin center reaching as much as 60 m/ky, far exceeding estimated subsidence rates.
APPENDIX 1. DATA AND METHODS
In this section we describe in detail the data used in this study and the methods of analyses applied to our integrated stratigraphic approach.
A1. Core and Well-Log Data
Three sites were drilled in Basin IV by the Integrated Ocean Drilling Program (Expedition 308 Scientists, 2006). Site U1320 is near the basin center, Site U1321 is on the basin margin, and Site U1319 is on the basin flanking slope (Figs. 2, 4). All sites were logged continuously from the seafloor with logging-while drilling (LWD), which are sensors encased in the drill string just above the drill bit, measuring natural gamma ray, electrical resistivity, bulk density, and porosity. Site U1320 also included wireline logs, with additional measurements such as sonic logs, borehole resistivity images, and seismic velocity survey (check shots). Continuous cores were attempted at Sites U1319 and U1320 with nearly complete recovery at the basin margin site U1319 and moderate to high recovery at Site U1320 (Fig. 4).
Sites Rudder #1 (Ru-1) and Rudder #2 (Ru-2) in Basin II were drilled in the early 1990s as part of a geotechnical and geological proprietary study led by Shell (Fig. 5). During drilling the well was cored at regular intervals, with Rudder 1 being cored in two closely spaced holes (Rudder #1 and #1c) (Fig. 5). Each short core (∼ 1.2 m long) was slabbed, described, and sampled for routine core analyses. Additional analyses were conducted recently and are described as part of this study. After drilling the well, wireline tools were lowered in the hole for measurements of gamma ray, electrical resistivity, sonic velocity, density, and porosity. Depths in the Rudder wells were originally measured in feet whereas depths in the IODP are measured in meters. Here we report the core depths in meters below seafloor (mbsf), adding the original depth in feet below seafloor (ftbsf) in parentheses for the Rudder samples.
A2. Seismic Stratigraphy
Time–depth curves for correlating between well and seismic data were built using sonic logs from the Shell Rudder wells in Basin II, and with sonic logs and check-shot survey in the IODP Basin IV wells (see Expedition 308 Scientists, 2006, for details). At Sites U1319 and U1321 no sonic logs were obtained (only LWD logs without sonic). At Site U1320 a wireline sonic log and a check-shot survey were obtained in the cored Hole U1320A, and LWD logs were obtained in Hole U1320B about 20 m away. The corresponding travel times and depths for the mapped horizons are listed in Table A1.
A synthetic seismogram constructed for Site U1320 (Expedition 308 Scientists, 2006) shows a reasonable correspondence for most reflections, although log quality near borehole washouts prevents a detailed match. Nevertheless, key mapped horizons correspond closely to observed lithologic units when converted to depth using the check-shot survey. A time–depth equation for the Rudder wells (constructed from sonic logs in Ru-1 and Ru-2; see Expedition 308 Scientists, 2006) was used for tying the reflections to the borehole data at Site U1319 and U1321 and at the Ru-1 and Ru-2 wells.
Calculations of sediment volume and masses for the various mapped intervals were as follows. First the surfaces mapped through the seismic volumes using 2D and 3D seismic data were subtracted from one another to compute isochron maps (with contour values in milliseconds of two-way time). Second, an average interval velocity was determined from the time–depth curves at the wells. Third, an average porosity (ϕ) for each interval was determined for the average depth below seafloor of the interval using the trend of porosity vs. depth derived from the IODP and Rudder wells. Finally, deposit gross volume was computed from the isochron maps multiplied by the average interval velocity, and deposit mass was computed by multiplying deposit volume by the average bulk density (ρg(1-ϕ) with ρg being the grain density, 2650 kg/m3). Deposit mass is reported in metric tons (1 metric ton = 1000 kg).
The work on planktonic foraminifers was conducted in two separate studies, one by O’Hayer (2009) for the IODP samples and another by Mallarino and Droxler for the Rudder samples (unpublished internal report, with results described here). In each study the procedure for sample preparation was the same. The 20 to 40 cm3 samples were disaggregated with a Calgon solution and wet-sieved at 63 µm. The sand size fraction was dried and sieved to retain the > 150 µm fraction. In the Rudder cores, if enough specimens were available in a sample, 100 to 300 planktonic foraminifer tests were picked and identified, and their abundance was recorded on a range chart (Fig. 9). Only 18 out 161 samples contained sufficient forams for this detailed biozonation work in the Rudder cores, the other samples being generally silty or sandy turbidite and mass-flow deposits containing few or no forams.
The biostratigraphic analyses are based on the planktonic-foraminifera zonation of Kennett and Huddleston (1972) for the late Quaternary of the western Gulf of Mexico, following the same procedure described in Mallarino et al. (2006). Kennett and Huddleston (1972) refined the five zones (V, W, X, Y, Z) of Ericson and Wollin (1968) that were based on frequency curves of the warm species Globorotalia menardii complex, and changing in coiling direction of Globorotalia truncatulinoides. By using an assemblage of selected taxa, Kennett and Huddleston subdivided the five zones in 18 subzones, thus providing a higher-resolution biostratigraphy. Following Kennett and Huddleston (1972) and Kennett et al. (1985), the following taxa were identified: Globorotalia menardii complex (G. menardii, G. tumida and G. flexuosa), Pulleniatina obliquiloculata, Neogloboquadrina dutertrei, Globorotalia truncatulinoides (left and right coiling), Globigerinoides sacculifer, Globorotalia inflata, Globorotalia crassaformis, Globigerinoides ruber, and Orbulina universa.
In the IODP samples obtained for this study (particularly in the deeper sections), the abundance of planktonic foraminifers was generally insufficient to conduct a detailed biozonation study (< 300 tests), so only the presence or absence of specimens from the Globorotalia menardii complex were noted (O’Hayer, 2009).
Planktonic nannofossils were studied only in the IODP Expedition 308 cores. The methods and results are described in detail in Flemings et al. (2006) and Shumnyk, written communication).
A4. Stable-Isotope Ratios 18O/16O and 13C/12C
In both IODP and Rudder cores, several (between 6 and 30) specimens of Globigerinoides ruber (white variety) were picked from the > 150 mm fraction for stable-isotope analyses. The isotope analyses were performed at the laboratory of the Department of Geology, University of California at Davis, following standard procedures (see O’Hayer, 2009, and Mallarino et al., 2006, for details). In a few samples, G. ruber (pink variety) specimens had to be used to provide sufficient material for the isotope analyses. The foraminiferal 18O/16O and 13C/12C isotopic ratios are expressed as 18O and 13C in per mil versus Pee Dee belemnite, and the average precision is ± 0.05% for oxygen and carbon. Analyses of Mg/Ca ratios in the foraminifer tests also were conducted but are not discussed here (see O’Hayer, 2009). Carbon isotope ratios also were measured on the plant and shell debris used for radiocarbon dating (see below).
Significant light peaks are observed in the delta;18O records in the Gulf of Mexico, including those reported by Mallarino et al. (2006) and O’Hayer (2009), with values of -3.5% to as low as -5.5% being interpreted as indicative of freshening of surface waters (meltwater pulses).
A5. Correlation Using Color Reflectance
A detailed biozonation and oxygen isotope measurements were obtained in some Basin IV piston cores by Mallarino et al. (2006), using the same methodologies employed in this study. Because their piston cores sample the same sequence cored in the upper part of Site U1319, we can correlate measurements obtained by Mallarino et al. (2006) with measurements in IODP cores. We do so by using measurements of color reflectance which were performed in both IODP cores (Expedition 308 Scientists, 2006) and the piston cores studied by Mallarino et al. (2006) (Fig. 9). A volcanic ash layer further constrains the correlation between cores. Correlation between sediment color lightness in marine cores and calcium carbonate content has been well documented by Balsam et al. (1999). The lightness curve was shown by Mallarino et al. (2006) to correlate with the abundance of fora-minifers in the cores from the basin margin.
A6. Volcanic-Ash Stratigraphy
We conducted X-ray fluorescence analysis on a sample from the Rudder #1 well in Basin II and another from IODP well U1319 in Basin IV. Each sample was carefully taken from the core to avoid contamination from adjacent sediments. The samples were split, with one aliquot crushed into a powder for XRF analysis and the other mounted and thin sectioned for examination under a scanning electron microscope. The results are compared with the chemical composition of ashes in the Caribbean and Gulf of Mexico published by Rabek et al. (1985) and by Drexler et al. (1980) (Fig. 8).
A7. Radiocarbon Analyses
Diverse types of materials were obtained for radiocarbon dating (Fig. 11), and each required different treatment before analyses. Pieces of wood, leaves, and fine plant debris were picked from the core and washed in double-distilled water over a wet-sieve to remove mud. The retained coarse fraction containing fine plant debris was examined under a microscope, and about 10 mg of plant material were picked for dating. For larger plant-debris fragments a small piece was cut from the center of the wood branch, root, or seed. A few samples consisted of mollusk shell fragments, which were picked and simply washed with double-distilled water. A few other samples consisted of bulk mud that was rich in organic material dispersed in the mud size fraction.
With radiocarbon samples, particularly wood and plant material, potential contamination by young carbon could affect the ages obtained. The cores were kept refrigerated from the time they were taken on the ship until the time we sampled them at the IODP Gulf Coast repository. The Rudder cores were originally acquired in 1990 and are stored at room temperature in a core storage facility in Schulenburg, Texas. Wood samples were carefully handled with metal tools only, washed in double-distilled water, stored in aluminum foil, and kept frozen once prepared. Examination under the microscope and careful picking of individual plant fragments ensured that we did not include any modern carbon in the samples. Only one sample, containing a very small amount of organic carbon (< 1 mg), was clearly contaminated and was eliminated from the analysis.
At selected intervals we encountered sufficient planktonic foraminifer tests to obtain about 10 mg of CaCO3 for radiocarbon dating. The samples were washed through a 63 mm sieve, and the retained fraction was dried and sieved again to keep the > 150 mm fraction. Tests of G. ruber were picked until obtaining a sufficient number to make about 10 mg of sample (about 300 specimens). In samples where we could not find sufficient white variety tests, we included some pink-variety tests. Some samples consisted of G. sacculifer only. Foraminifer tests were ultrasonically cleaned for a few seconds, washed in distilled water, dried, and stored in a glass vial.
All samples were submitted for radiocarbon analyses at the Keck Carbon Cycle Accelerator Mass Spectrometer facility at the University of California, Irvine. The samples were further treated following standard procedures for measuring 14C/12C and 13C/12C ratios in carbonate, wood or soil (Santos et al., 2007). The reported radiocarbon ages were corrected to calendar ages by first applying a calibration curve based on samples dated by both 14C and U/Th method (Fairbanks, online) and then applying a reservoir age correction of –205 years to the carbonate samples (forams and shell fragments) (Fairbanks, online; Butzin et al., 2005; Cao et al., 2007). The measurements, corrections, and final results are presented in Table A2.
Some foram-rich intervals had sufficient material for duplicate or subsample analyses. The difference in age between G. sacculifer and G. ruber can be associated with different depths where the species live, which should lead to different radiocarbon reservoir ages. Dating subsamples of each species in one sample showed that G. ruber gives an age only 169 calendar years older than G. sacculifer (Table A2).
In several other samples we compare the age of different materials in the same sample interval, such as different sizes of plant debris, wood vs. shell fragments (Table A2, replicates). Because these materials are clearly reworked (transported by turbidity currents to the site of deposition), only the younger ages in the same sample have stratigraphic significance. However, the age range obtained in different wood fragments provides insight into the residence time of organic material in this source-to-sink system. Once dead, a piece of wood can be trapped in fluvial– deltaic deposits, as well as in intermediate basins within the deepwater system, before reaching the sample position. Different pieces of wood within the same depth interval show an age difference of 656 and 1135 years (U1320 at 92.93 mbsf and Rudder #1 at 42.7 mbsf). Within the same (amalgamated) bed, age difference in various pieces of wood span as much as 1918 years (U1320 between 92.5 and 93.1 mbsf). Other replicates show that larger pieces of wood and mollusk shell fragments are older than smaller pieces of wood in the same sample. The age span of wood samples in the same bed provides an indication of the storage time, and of the error with respect to the depositional age of the bed containing the shell or wood fragment. A piece of wood, such as one of those encountered in between 92.5 and 93.1 mbsf at Site U1320 (Figs. 11F, G), may remain in intermediate storage within the system for as much as 2000 years before reaching the sink of the system in Basin IV. Dispersed organic matter in bulk mud samples gave ages in the range of 28–30 ka, and also appear to be quite old, by as much as 5000 years or more with respect to their depositional age.
This research used samples and data from Expedition 308 provided by the Integrated Ocean Drilling Program (IODP). Support for radiocarbon and oxygen isotope sample analyses and for O’Hayer came from a JOI–USSAC grant to Droxler and Pirmez. Julia Gutiérrez-Pastor assisted with IODP core sampling and description at the core repository in College Station. Maria Banks worked on the ash composition and helped prepare Figure 9. We thank Zoltán Sylvester, Carrie Mazzielo, and Brandon Dugan for discussions that helped improve our understanding of various aspects discussed in this paper. Ben Sheets is thanked for the insightful review and suggestions. A special thank you to John Anderson for a detailed review and suggestions that helped improve the manuscript significantly. We thank Shell for permission to publish this material and for releasing the results from the Rudder wells.