We used cores and logs from Integrated Ocean Drilling Program (IODP) Expedition 313 to generate biostratigraphic, lithofacies, biofacies, and geochemical data that constrain the ages and paleoenvironments of Pleistocene sequences. We integrate sequence stratigraphy on cores with new seismic stratigraphic data to interpret the Pleistocene history of the Hudson shelf valley and paleoenvironmental and sea-level changes on the inner to middle continental shelf. Improved age control compared to previous studies is provided by integrated calcareous nannofossil biostratigraphy, Sr isotopic stratigraphy, and amino acid racemization. We recognize four upper Pleistocene–Holocene sequences: sequence uP1 is correlated with Marine Isotope Chrons (MIC; “chron” is the correct stratigraphic term for a time unit, not “stage”) 7 or 5e, sequence uP2 with MIC 5c, sequence uP3 with MIC 5a, and sequence uP4 with the latest Pleistocene to Holocene (MIC 1–2). However, within our age resolution it is possible that sequences uP2 and uP3 correlate with MIC 4–3c and 3a, respectively, as suggested by previous studies. Lower Pleistocene sequences lP1 and lP2 likely correlate with peak interglacials (e.g., MIC 31 and MIC 45 or 47, respectively). Thus, we suggest that preservation of sequences occurs only during peak eustatic events (e.g., MIC 45 or 47, MIC 31, and MIC 5), unless they are preserved in eroded valleys. The architecture of the Pleistocene deposits at Sites M27 and M29 is one of thin remnants of highstand and transgressive systems tracts, with lowstand deposits only preserved in the thalwegs of incised valleys. Incised valleys at the bases of sequences uP3 (IODP Site M27) and uP2 (IODP Site M29) document more southward courses of the paleo–Hudson valley, compared to the more southeastward course of the MIC 1–2 paleo–Hudson valley. The patchy distribution of Pleistocene sequences beneath the New Jersey inner-middle continental shelf is due to low accommodation during an interval of large eustatic changes; this predicts that sequences in such settings will be discontinuous, patchy, and difficult to correlate, consistent with previous studies in Virginia and North Carolina.


Sea-level change is one of the primary controls on passive continental margin sedimentation (e.g., Miller et al., 2005), and unconformities formed during relative sea-level lowerings bracket the stratigraphic record into units called sequences (e.g., Mitchum et al., 1977). Large (>100 m) sea-level changes typified the Pleistocene and profoundly affected sedimentation during that time (e.g., Mallinson et al., 2005, 2010; Mountain et al., 2007). During the Brunhes Chron (past 780 k.y.), the growth and decay of ice sheets was paced by the quasi-100 k.y. eccentricity cycle along with subdominant 41 k.y. tilt and 23 and 19 k.y. precessional periods (Hays et al., 1976). The largest of the sea-level changes occurred during terminations of 100 k.y. glacial intervals, with ∼120 m rises (Fairbanks, 1989). Smaller ice volume and sea-level changes (∼20–60 m) were controlled predominantly by the 41 and 23 and 19 k.y. cycles, with the 41 k.y. cycle dominating sea-level changes prior to the Brunhes Chron (e.g., Lisiecki and Raymo, 2005). Though the timing and amplitude of Pleistocene eustatic changes are generally known, the response of continental margin sedimentation to these changes has been difficult to evaluate due to sampling and dating issues.

Several studies have evaluated the Pleistocene sequence stratigraphy beneath the inner to middle continental shelf of offshore New Jersey using seismic reflection profiles and Vibracores (Ashley et al., 1991; Carey et al., 1998, 2005; Knebel et al., 1979; Sheridan et al., 2000; Wright et al., 2009). Uniboom and minisparker data (∼1 m vertical resolution) were used to map a paleo–Hudson valley on the middle continental shelf off New Jersey (Fig. 1) that cut into a seismic reflector named Horizon R (Knebel et al., 1979). Carey et al. (2005) interpreted the sequence stratigraphy of Pleistocene strata imaged on Uniboom data on the inner to middle continental shelf, identifying sequence I to sequence IV from oldest to youngest. Sequence IV has been dated in Vibracores as Holocene (MIC 1; Knebel et al., 1979; Ashley et al., 1991). Sequence III was identified based on reflector truncations. Radiocarbon dates of 28–36 ka on Vibracore samples on lines 1–6 (Fig. 1; Knebel et al., 1979) suggest that the upper part of sequence III, the transgressive systems tract (TST), correlates with MIC 3a (Carey et al., 2005). The lower part of sequence III was interpreted as the lowstand systems tract (LST) and was correlated to MIC 3b (ca. 40–50 ka) and as equivalent to the infill of the incised paleo–Hudson valley of Knebel et al. (1979). Carey et al. (2005) suggested that their sequence boundary III is the same as Horizon R (Knebel et al., 1979). Carey et al. (2005) recognized an older ancestral Hudson valley in their sequence II and dated this as correlative to MIC 3c–4.

These cited studies provide a glimpse of Pleistocene sequences on the New Jersey inner to middle continental shelf. Other studies have examined the late Pleistocene of the outer continental shelf and slope and provided sequences tied to a chronostratigraphic framework (e.g., Goff et al., 2005), although the links to the inner-middle shelf region are unclear. However, several major challenges remain in evaluating Pleistocene sequences across this margin: (1) Pleistocene sequences are generally thin and discontinuous (e.g., Carey et al., 2005); (2) Pleistocene age control is poorly constrained in the region, despite recent attempts at improving resolution (Wright et al., 2009); and (3) core samples of Pleistocene sediments have been sparse in the inner to middle shelf region and limited to relatively short Vibracores or spot-cored AMCOR (Atlantic Margin Coring Project) samples (Hathaway et al., 1979). In 2009, Integrated Ocean Drilling Program (IODP) Expedition 313 drilled Sites M27, M28, and M29 on the continental shelf ∼45 km off New Jersey (Fig. 1), using the L/B (liftboat) Kayd, and recovered several Pleistocene sequences (Mountain et al., 2010).

In this paper, we develop a Pleistocene sequence stratigraphic framework for the inner to middle continental shelf of New Jersey based on cores and logs obtained by Expedition 313 and new seismic stratigraphic data (Figs. 2–5). Expedition 313 cores provide critical material to generate biostratigraphic, lithofacies, biofacies, and geochemical data to constrain the ages and paleoenvironments of Pleistocene sequences. We use this integrated framework to interpret the Pleistocene history of the Hudson shelf valley and the paleoenvironmental and sea-level changes on the inner to middle continental shelf.



Site M27 (33.5 m water depth; Fig. 1) recovered 32 m of continuously cored Pleistocene section, whereas the more seaward Site M29 (35.9 m water depth; Fig. 1) recovered ∼50 m of discontinuously cored Pleistocene sediments. No Pleistocene cores were obtained at Site M28 due to hole stability concerns (Mountain et al., 2010). All sites were continuously logged through the drill pipe. The sites are located near seismic reflection profiles (Knebel et al., 1979; Carey et al., 2005) along with being located on recent R/V Oceanus cruise 270 multichannel seismic profiles (Monteverde et al., 2008; Fig. 2) and unpublished Chirp (compressed high-intensity radar pulse) sonar profiles from R/V Endeavor 370 (Figs. 3–5); detailed tracks charts are given in Supplemental Figures 11 and 22.

Lithology, Logs, and Benthic Foraminifera

Lithology and downhole gamma logs were used to identify sequence boundaries in cores and to make paleoenvironmental interpretations (Fig. 6). Gamma ray measurements made on unsplit core sections confirm the registration of the downhole logs to the core within at least 0.5 m (Mountain et al., 2010). Gamma log data record lithologic variations primarily of quartz sands versus muds at Site M27 (Fig. 7), with low gamma readings in sands and high gamma readings in muds. Glauconite sand and other components that complicate gamma log interpretations (e.g., Lanci et al., 2002) occur in trace abundances or are absent in the Pleistocene succession.

We reevaluated lithofacies and placement of sequence boundaries done by the onshore science party (Mountain et al., 2010) based on cumulative grain size data (Fig. 7), core descriptions, downhole logs (Figs. 2 and 7), and core photographs (Fig. 8). Mountain et al. (2010) differentiated clay, silt, and various sand fractions visually and on smear slides. We quantitatively measured weight percent fine sand, medium-coarse sand, and mud on washed samples and semiquantitatively estimated the percent glauconite, shells, and mica through visual counts (Fig. 7). Washed samples were examined for benthic foraminifera, which were generally rare and thus were not counted. Nevertheless, sufficient benthic foraminifera were present to evaluate water depth variations (Fig. 6) by using biofacies and water-depth markers of Miller et al. (1997a) and Katz et al. (2003); these in turn were based on modern distribution of taxa (e.g., Bandy and Arnal, 1957; Smith, 1964; Walton, 1964; Poag, 1981; Sen Gupta and Kilborne, 1976; Murray, 1991). Lithofacies variations (Fig. 6) are interpreted based on previous studies of shallow-marine sediments using a wave-dominated shoreline model (summarized in Methods discussion in Mountain et al., 2010), recognizing upper shoreface, lower shoreface, shoreface-offshore transition, and offshore environments (Fig. 6). The integrated lithofacies and biofacies model (Fig. 6) provides a means of reconstructing water depth changes through time. (1) Samples with few to no foraminifera, common echinoderm plates and spines, mollusk shell debris, and medium-coarse sand are interpreted as upper shoreface (∼0–5 m paleodepth). (2) Samples dominated by Elphidium spp. and fine sand are distal upper shoreface (∼5–10 m paleodepth). (3) Samples dominated by Elphidium spp. and Pseudononion spp. and sandy muds are interpreted as shoreface-offshore transition (∼10–20 m paleodepth). (4) Samples dominated by Elphidium spp. and Pseudononion spp. and muds are interpreted as offshore (∼30–40 m paleodepth). Evidence for paleowater depths deeper than ∼30–40 m is lacking.

Age Control

Age control was obtained from calcareous nannofossils (D.K. Kulhanek, inMountain et al., 2010; this study; Table 1), strontium isotopic age estimates (Mountain et al., 2010; this study; Table 2), and amino acid racemization (this study; Table 3; Figs. 9–11). A radiocarbon date obtained on Astarte shell fragments from sample 7 of Site M29 (13.4 mcd, meters composite depth) was measured at the National Ocean Sciences Accelerator Mass Spectrometry Facility (NOSAMS), Woods Hole. The date obtained of older than 52 ka is considered beyond the reliable age range for radiocarbon, but places a minimum age constraint (Fig. 11).

Calcareous nannofossil slides were prepared using standard smear slide techniques. Data for Site M29 were collected during the IODP Expedition 313 onshore science party using an Olympus BX51 light microscope. Abundance estimates were made at 1000× utilizing the following scale: B—barren, R—rare (1 specimen/≥51 fields of view), F—few (1 specimen/11–50 fields of view), Fr—frequent (1 specimen/2–10 fields of view), and C—common (1–10 specimens/field of view). Semiquantitative data were collected for 13 samples from Site M27 using a Zeiss Axiophot microscope at 1250× magnification. For each slide, nannofossil specimens were counted in random fields of view until reaching 300 specimens or ∼800 fields of view had been observed. Several additional slides were scanned, with presence/absence data collected due to the sparse nature of the nannofossil assemblages in these samples. All nannofossil data are reported in Table 1.

Sr isotopic age estimates were obtained from mollusk shells and shell fragments of different species. Approximately 4–6 mg carbonate was cleaned in an ultrasonic bath and HCl, and then dissolved in 1.5 N HCl. Sr was separated using standard ion exchange techniques (Hart and Brooks, 1974). The samples were analyzed on an Isoprobe T Multicollector thermal ionization mass spectrometer (Table 2). Internal precision on the Isoprobe for the data set averaged 0.000007 and the external precision is ∼±0.000008 (based on replicate analyses of standards). We measured NBS 987 at 0.710241 normalized to 86Sr/88Sr of 0.1194. Ages were assigned using the look-up tables of McArthur et al. (2001). Sr isotopic age estimates for the past 2.5 m.y. have errors of ∼±300 k.y. based on the rate of change of Sr isotopes for this interval (Farrell et al., 1995). We computed the error based on the look-up tables of Howarth and McArthur (1997) and McArthur et al. (2001) using their LOWESS (locally weighted scatterplot smoothing) estimates of error of the regression of Sr versus ages and an external precision of ±0.000008 (Supplemental Figure 33), confirming that errors are ±0.2 to ±0.4 m.y. for the past 1.8 m.y. We use the astronomical time scale summarized in Gradstein et al. (2004) and correlation to the deep-sea benthic foraminiferal δ18O stack of Lisiecki and Raymo (2005) for chronology (Figs. 9–11).

Eight shell fragments were analyzed for amino acid racemization (AAR) using gas chromatographic methods described in Goodfriend (1991), Wehmiller and Miller (2000), and Wehmiller et al. (2010). Individual amino acid measurements were determined on shells of Mercenaria, Spisula, Ensis, and Astarte. These are common genera found on the modern shelf and in Pleistocene deposits of the adjacent U.S. Atlantic Coastal Plain. Ratios of Dextrorotary/Laevorotary (D/L) values for seven amino acids are plotted (Fig. 12) and tabulated (Table 3). Individual amino acid identifications are listed in Table 3. For purposes of both relative and numerical age estimation, we compare the AAR results from Expedition 313 shells with those for the same taxa from southern New Jersey, Virginia, and North Carolina (Wehmiller et al., 1988; Groot et al., 1995; Sugarman et al., 2007; Wehmiller et al., 2010). Because AAR age estimates are affected by temperature, comparisons of results over wide latitude ranges require assumptions about differences in temperature histories in the region of study (see Results discussion).

Seismic Data

High-resolution (∼5 m vertical resolution) generator-injector air gun multichannel seismic reflection profiles were collected on R/V Oceanus in 1995 as a site survey for Ocean Drilling Program (ODP) Leg 174A coring on the New Jersey outer shelf and slope (Mountain et al., 2001, 2007). Oc270 multichannel seismic (MCS) line 529 (Fig. 2) was extended across the shelf to identify seismic sequences subsequently drilled on IODP Expedition 313 at Sites M27, M28, and M29.

As part of the site surveys for drilling at Sites M27 and M29, Chirp sonar profiles collected from an array towed 5–15 m above the seafloor were recorded on R/V Endeavor cruise 370 in 2002 (McHugh et al., 2010; Figs. 2–5). Chirp sonar profiles directly sampled Sites M28 and M29 but deflected to the north of Site M27. Vertical resolution is ∼0.5 m in these data. With the high resolution of the Chirp sonar data, details such as the near-surface filled channel can be imaged (e.g., Fig. 3). Such small-scale features, both vertical and laterally, were not seen in Uniboom data (Knebel et al., 1979; Carey et al., 2005) or in the air gun data (Figs. 3–5).


Seismic Stratigraphy of Oceanus MCS and Endeavor Chirp Sonar Data

We recognize four Pleistocene sequences, based on seismic profiles, that we term from older to younger uP1, uP2, uP3, and uP4. Three of these (uP1–uP3) are recognized in the cores and dated as middle-late Pleistocene (see following).

Sequence uP2 is imaged near Site M29, but has been eroded between Sites M27 and M28 (Fig. 4). The basal uP2 sequence boundary clearly truncates uP1 on Oc270 MCS line 529 (Fig. 2). Near drill Site M27, the En370 Chirp sonar data image a seismic sequence boundary identified as the base of sequence uP3 (Fig. 4). The Chirp sonar data show erosion associated with the sequence boundary at the base of uP3 as an irregular, notched erosional surface (Fig. 4). A reflector within uP3 is identified as a maximum flooding surface that can be traced into Site M27 (Fig. 4). The Chirp sonar data reveal an erosional surface within sequence uP3 (Fig. 4), a surface interpreted as a tidal ravinement channel. The seismic transparency of the unit above this unnamed surface suggests uniform sand. Such sand-filled channels are common near the baymouth of a larger incised river valley (Ashley and Sheridan, 1994). Carey et al. (2005) also identified similar tidal-cut channels within the complex fill of the larger incised paleo–Hudson valley of sequence uP3 on line 2–14 (Fig. 1) using lower resolution Uniboom data.

At Site M29, the Chirp sonar data trackline crosses the drill location (Fig. 5) where sequences uP2, uP3, and uP4 are well imaged. There is seismic evidence of erosional truncation associated with the bases of uP4 and uP3 (Fig. 5). Unfortunately, the deeper parts of the incised valley in sequence uP2 and most of the basal sequence boundary at Site M29 are obscured by the source-to-surface reflection and the water-bottom reflection multiples (Fig. 5). The exception is the westernmost portion shown in Figure 5 that images the basal sequence boundary and part of sequence uP1.

The seismic data image an incised valley at the base of uP2 that was cored at Site M29 (Fig. 2). The incised valley at Site M29 has cut out sequence uP1; as a result, sequence uP2 is on sediments older than sequence uP1. Sequence uP2 infills the incised valley at M29 (Fig. 2) and is eroded and truncated by the base of sequence uP3. An incised valley also occurs just northeast of Site M27, where the basal sequence boundary of uP3 cuts into uP1 (Fig. 2). Sequence uP4 occurs as erosional remnants in the area of Sites M27 and M29 (Figs. 2–5), with a distinct valley channel imaged on the Chirp sonar data (Fig. 4).

Sequence Stratigraphy Based on Cores and Logs

We identify sequence-bounding unconformities in the core (Figs. 7–11) and use these to place our lithologic, downhole log, core log, and benthic foraminiferal data sets into a sequence stratigraphic framework. The onshore science party (Mountain et al., 2010) identified possible Pleistocene sequence boundaries at Site M27 at 10.3 mcd (at the base of an upward-coarsening shoreface succession), 26.38 mcd (at the base of a gravel; Fig. 8A), and 31.9 mcd (at the base of a gravel layer) based on lithologic successions. Here we confirm two of these sequence boundaries and identify two additional Pleistocene sequence boundaries at Site M27: 13.57 mcd (at the base of a gravel layer; Fig. 8) and 22.5 mcd (at the base of a clay unit, although it is possible that the sequence boundary is at 21.42 mcd based on biostratigraphy; Figs. 7 and 8B). We note that sequence boundaries typically occur at gravel lags (Figs. 7 and 8). The surface identified by Mountain et al. (2010) as a possible sequence boundary at 10.3 mcd is interpreted as a maximum flooding surface (MFS) separating a thin TST from a highstand systems tract (HST) (Fig. 7). Sequence boundaries at 13.57 (Fig. 8A) and 22.8/21.42 (Fig. 8B) mcd are the bases of middle-upper Pleistocene sequences (i.e., nannofossil Zone NN20 and younger) and correspond to seismic sequence boundaries uP3 and uP1, respectively. A younger, uppermost Pleistocene to Holocene seismic sequence uP4 is detected but not cored at Site M29; this sequence was previously dated as Holocene in Vibracores using 14C ages at locations on lines 1–6 (Fig. 1; Knebel et al., 1979). The sequence boundaries at 26.38 mcd (Fig. 8A) and 31.9 mcd are not imaged on the seismic profiles and are lower Pleistocene; they are named lP1 and lP2 (from older to younger).

Site M27

We describe the Pleistocene sequences at Site M27 upsection from the base of the Pleistocene.

Lower Pleistocene sequence lP1. A lower Pleistocene sequence is found from 31.9 to 26.38 mcd (Fig. 7). The base of the sequence is a scoured, cut-and-filled surface at 31.9 mcd that separates a pebbly medium sand above from silty clay. The clay below is inferred to be upper Miocene based on seismic correlations (Mountain et al., 2010). The lower part of sequence lP1 at Site M27 consists of medium sand with two thin gravel beds (30.8–30.7 and 31.3–31.2 mcd). Gamma log values increase upsection to 28.2–28.1 mcd, indicating fining upward to laminated, lignitic clay at this level. Above this, the sequence generally coarsens upward to coarse to very coarse sand (Fig. 7). We interpret this sequence as containing a TST (31.9–28 mcd), an MFS at the clay (28.2–28.1 mcd), and a thin HST (28–26.38 mcd). The dominant environment is upper shoreface with the laminated clay marking shoreface-offshore transition environments.

Lower Pleistocene sequence lP2. A gravel bed at 26.38–25.6 mcd marks the base of a lower Pleistocene sequence (26.38–22.8 mcd) overlain by medium-coarse sand (25.6–22.8 mcd) (Fig. 7). The sand fines upward to fine sand with shell fragments in the upper part of the sequence (Fig. 7). The sand contains Elphidium and Pseudononion, indicating deposition in distal upper shoreface to shoreface-offshore transition environments (Fig. 7). We favor placing the upper sequence boundary at a sharp lithologic contact at 22.8 mcd that is overlain by a slightly shelly, heavily bioturbated clay (Fig. 8B). However, possible clayballs (core 12–1, 9–15 cm; Fig. 8B) at 21.42 mcd could mark the upper sequence boundary, and this interpretation is suggested by nannofossils indicating that Zone NN19 continues above 22.88 mcd, but Zone NN21 (younger than 290 ka) appears above 21.4 mcd at the possible clayballs. It is possible that the interval from 21.24 to 22.8 mcd is a separate sequence, or the nannofossils from 21.24 to 22.8 mcd are reworked. We interpret the sequence from 26.38 to 22.8/21.42 mcd as a deepening-upward section (from shoreface to shoreface-offshore transition), with only the TST preserved (Fig. 7).

Upper Pleistocene sequence uP1. A clay-rich interval (22.8/21.42 mcd to 17.64 mcd) marks the base of sequence uP1 (22.8–13.57 mcd) at Site M27 (Fig. 7). The clay contains shells and rare silt laminae and was deposited in offshore environments. We tentatively place an MFS at ∼21 mcd based upon peak gamma ray values suggesting the finest grain size. Above this, the section generally coarsens upsection and biofacies change from Elphidium and Pseudononion to Elphidium. We interpret this upsection shallowing from offshore to shoreface-offshore transition to distal upper shoreface (Fig. 7) as an HST. A sandy gravel bed (13.57–13.4 mcd) marks the base of the overlying sequence.

Upper Pleistocene sequence uP3. Slightly shelly, bedded, medium and coarse sand (13.4–11.54 mcd) overlies a gravel bed (13.57–13.4 mcd; Fig. 8A) marking the base of sequence uP3 (Fig. 7), although it is possible that the gravel is caved and the sequence boundary is in the small (0.17 m) coring gap. The section fines upsection from fine to medium sand to muddy fine sand at 10.3 mcd; we interpret the section from 13.4 to 10.3 mcd as a TST capped by an MFS. Above 10.3 mcd, the section coarsens upward to coarse sands to 6 mcd (Fig. 7), interpreted as the HST. Seismic stratigraphic interpretations suggest that the upper part of sequence uP3 at Site M27 is a valley-fill deposit (Fig. 4).

Upper Pleistocene sequence uP4. The uppermost sequence imaged in Chirp sonar data (Figs. 3 and 5) is eroded at Site M27. Though imaged seismically at Site M29, no cores were recovered from sequence uP4 and its presence is inferred.

Sites M29 and M28

The Pleistocene section at Site M29 was discontinuously cored, but spot cores and the downhole gamma log (Fig. 11) allow tentative sedimentary facies interpretations of several units.

  1. High gamma values in core 14 (43.8–43.5 mcd) may be associated with muds of the shoreface-offshore transition, although we lack sufficient core samples to verify this interpretation (Fig. 11).

  2. We place a sequence boundary at ∼43.5 mcd based on a gamma log decrease (Fig. 11) and seismic correlations (Fig. 5).

  3. Cores 13–8 (∼40.6–15.5 mcd) had limited recovery containing only coarse sands; they are very likely sands throughout, as indicated by uniform, low gamma log values and low-amplitude seismic reflections (Fig. 5). We interpret this as a thick valley-fill deposit of sequence uP2 based on seismic interpretations (Fig. 5). Marine shell fragments in the upper part of the valley-fill sands in cores 9 and 8 (19.2–15.5 mcd) attest to the high-energy marine environment of this unit.

  4. Gamma log values increase from a coring gap between cores 8 and 7 (∼15.5 mcd; Fig. 8C) to core 6, section 1 (10 mcd), indicating a deepening-upward TST. We tentatively place a sequence boundary in the coring gap at ∼15.5 mcd associated with the gamma log increase (Fig. 11), based on seismic correlations (Fig. 5).

  5. Sample 313-29A-6R-1, 63 cm (10.58 mcd), contains benthic foraminifera (common Elphidium and rare nodosarids) typical of shoreface-offshore transition to offshore environments; this level is associated with peak gamma log values and is interpreted as an MFS (Fig. 11).

  6. The water depths in the section above 313-M29-6R-1 are interpreted to represent environments that shallow upward from shoreface-offshore transition and offshore at the base to shoreface sands on top. Sample 313-M29-5R-1, 28 cm, had rare Elphidium and sample 313-M29-5R-1, 17 cm, was barren. The section is interpreted as a regressive HST (Fig. 11).

Similar to Site M27, the Pleistocene sediments at Site M29 are largely remnants of TST and HST facies, but the lower part of the valley fill of cores 13–7 (40.6–13 mcd) may be an LST at Site M29.

Age Control

Nannofossil Biostratigraphy

Calcareous nannofossils are generally common in the marine sediments of the New Jersey shallow shelf. The presence of Emiliania huxleyi, the first occurrence (FO) of which is at 290 ka, marks the base of Zone NN21 in Sites M27 and M29, and Pseudoemiliania lacunosa, the last occurrence (LO) of which marks the top of Zone NN19 in Site M27, places the Pleistocene sequences in a broad biostratigraphic framework. In addition, changes in species abundance and the occurrence of secondary taxa allow for further refinement of the biostratigraphy.

At Site M27, samples from 0.20 mcd (sample 313-M27-1H-1, 20 cm) and 8.27 mcd (sample 313-M27-4H-1, 78 cm) contain common E. huxleyi (15%–20% of the assemblage) and abundant Gephyrocapsa muellerae (57%–62%). In general, E. huxleyi has dominated the global nannoplankton assemblage for the past ∼73 k.y. (e.g., Flores et al., 2010), although Jordan et al. (1996) showed that the onset of this acme is diachronous and younger toward higher latitudes and in coastal upwelling areas. The E. huxleyi acme is identified by a crossover in the abundance of G. muellerae and E. huxleyi (e.g., Thierstein et al., 1977; Jordan et al., 1996; Hine and Weaver, 1998). The abundances of these taxa in the upper part of Site M27 indicate that this interval is older than the onset of the E. huxleyi acme; however, assigning an age to that event for the New Jersey margin is difficult. The temperate setting suggests these samples are older than 73 ka (Thierstein et al., 1977), but younger than the first common occurrence (FCO) of G. muellerae within MIC 6 (Flores et al., 2003). Therefore, we suggest that cores 1–4 were deposited during MIC 5, although we cannot completely rule out a correlation to MIC 3 if the E. huxleyi acme event is slightly younger (∼45–60 ka; shown as dashed line in Fig. 10) at this site due to the coastal setting (Jordan et al., 1996).

Three samples from 16.05 to 21.25 mcd (313-M27-8H-1, 79 cm, 313-M27-10H-CC, and 303-M27-11H-CC) are dominated (68%–86% of the assemblage) by small Gephyrocapsa spp. (primarily Gephyrocapsa aperta), but also contain few G. muellerae and questionable specimens of E. huxleyi (both <5% of the assemblage). E. huxleyi can be difficult to identify in the light microscope, especially when specimens are rare and small. The (likely) presence of E. huxleyi indicates that the samples are assignable to Zone NN21 (<290 ka). In addition, the significantly reduced abundance of G. muellerae suggests the samples are older than the FCO of this taxon in MIC 6 (ca. 165 ka; Hine and Weaver, 1998; Flores et al., 2003) and that they were deposited during MIC 7. Alternatively, high-resolution Pleistocene nannofossil data often show variation in nannofossil assemblages over short time scales, resulting in intervals of reduced species abundance even within an acme zone. These patterns of species abundance can vary substantially even between sites located in the same region. For example, Liu (2009) presented a high-resolution record for IODP Site 1304 (northwest Atlantic) in which the abundance of G. muellerae is generally high (>15%) back through to MIC 6, with a few samples that contain lower abundances, particularly within MIC 3 and 5. This is in contrast to a nearby record from IODP Site 1308 (northeast Atlantic), in which G. muellerae is not recognized at all in the assemblage in sediments older than ca. 100 ka (Chiyonobu et al., 2010). Flores et al. (2003) also noted a reduction in the abundance of G. muellerae within the early part of MIC 5 following its FCO in MIC 6. Thus, it is possible that the samples from Site M27 could represent one of these intervals of reduced abundance following the FCO of the taxon within MIC 6, allowing correlation to MIC 5 (probably MIC 5e).

Nannofossils are generally rare in samples from lower Pleistocene sequences lP1 and lP2 between ∼21.4 mcd and 31.9 mcd. The presence of P. lacunosa in samples 313-M27-12H-1, 40 cm (21.67 mcd), and 313-M27-12H-1, 43–44 cm (21.70 mcd), indicates assignment to Zone NN19 (1.93- ∼0.425 Ma) if not reworked. A questionable specimen is also present in sample 313-M27-15H-1, 104 cm (27.42 mcd). The absence of key taxa such as Reticulofenestra asanoi (1.14–0.91 Ma) and large Gephyrocapsa spp. (1.56–1.26 Ma) may be due to the sparse nature of the assemblages.

Only two samples from Site M29 contain calcareous nannofossils: 313-M29-5R-1, 34.5 cm (7.245 mcd), and 313-M29-7R-1, 25 cm (13.25 mcd). The presence of E. huxleyi in these samples indicates Zone NN21 (younger than 290 ka; Table 1).

Amino Acid Racemization

The Mercenaria samples from Site M27, core 13 (sample 313-M27-13H-2, 53.0–54.0 cm, 24.7 mcd) and Site M29, core 14 (sample 313-M29-14R-2, 12.0–13.0 cm, 43.9 mcd) are from the same stratigraphic unit and have similar D/L values (Fig. 9A) that are much greater than any values observed for this genus from upper Pleistocene units in the New Jersey, Virginia, and North Carolina region. These high D/L values from the M27 and M29 Mercenaria may represent slightly different ages, but combined they suggest an early Pleistocene (ca. 1.0–1.5 Ma) age based on calibrated racemization kinetics for this genus (Wehmiller et al., 2012). This early Pleistocene age assignment agrees with the Zone NN19 assignment (Table 1) and with the Sr isotope age estimates.

Because different molluscan genera have different racemization rates, and because no single genus was found in all the cores, it is necessary to make some assumptions about similarities or differences in racemization rates among the analyzed samples. In our study, we assume that Spisula, Astarte, and Ensis have similar racemization rates, based on limited evidence from other studies in the mid-Atlantic region (Toscano et al., 1989; York, 1990). Spisula and Astarte samples from Sites M27 and M29 have virtually identical D/L values (Fig. 8) and are interpreted to represent an age older than 52 ka, based on a radiocarbon analysis on Astarte from sample 313-M29-7R-1, 39–41 cm. This interpretation is reinforced by similar D/L values for a radiocarbon-dead (older than 44.6 ka) Spisula from Parramore Island, Virginia. Consequently, based on the AAR results, we infer that the Spisula and Astarte samples from M27 and M29 are older than ca. 55 ka but not older than MIC 5 (ca. 130 ka; Figs. 10 and 11). This age estimate is consistent with assignment to nannofossil Zone NN21 prior to the FCO of E. huxleyi (older than 73 ka).

The D/L values for the Ensis samples from Site M29 (sample 313–M29–9R-1, 0–4 cm) are significantly greater than those for the Astarte and Spisula samples described here (Table 3). Assuming these taxa have comparable racemization rates, this observation suggests that the Ensis samples are older. An estimate of this age difference can be made using the power-law leucine kinetic model (developed by Wehmiller et al., 2012) and mean D/L leucine values of 0.19 (Astarte-Spisula) and 0.285 (Ensis). If the 0.19 D/L values are assigned an age of 80 ka, then the Ensis samples are estimated to be ca. 300 ka. Proportional age differences arise depending on the calibration age used for this calculation. Because the Ensis samples are in a unit (uP2) with younger shells, these observations suggest that this unit contains a mixture of shell ages. Reworking of Ensis from an older unit is the simplest explanation, although other explanations for the different D/L values observed in the uP2 unit (shell alteration or contamination) must be considered. However, a unit with an estimated age of ca. 300–400 ka has been identified using AAR data from the nearby Cape May Zoo core hole (Sugarman et al., 2007), so a potential source unit for reworked shells is known in the region.

Strontium Isotopes

Strontium isotope age estimates were obtained from six samples in Site M27 from cores 1, 6, 9, and 15, and one sample from Site M29 in the lower part of core 7 (Table 2). The samples from core 15 (313-M27–15H-1, 60 cm, 26.98 mcd; two from 313-M27–15H-1, 126 cm, 27.64 mcd), yield Sr isotope age estimates of 1.3, 1.1, and 1.3 Ma, respectively, that are in general agreement with the early Pleistocene age determination from AAR dating and nannofossils. However, the Sr isotopic age estimates from younger sequences in Site M27 (313–M27-1H-1, 144 cm, 1.44 mcd, 1300 ka; 313–M27-6H-1, 145 cm, 12.99 mcd, 400 ka), are much older than the ages determined by AAR and nannofossils. A date of 400 ka (313–M27-9H-3, 0 cm, 19.54 mcd) is consistent within the errors. Also, sample 313–M29-7R-2, 4 cm, 14.54 mcd) in Site M29 has a Sr isotopic age estimate of 200 ka that is quite a bit older than the AAR dates, but within nannofossil Zone NN21.


Integration of Core Sequences, Seismic Sequences, and Ages

Though the seismic stratigraphy and sequence stratigraphic definition of Pleistocene sequences in the Expedition 313 region is clear, age control is still limited. Lower Pleistocene sequences lP1 and lP2 are dated as older than Brunhes (780 ka) by Sr isotopes and AAR (Fig. 9). Nannofossils assign both sequences to Zone NN19 (1.93- ∼0.425 Ma), but are otherwise equivocal on age control; the absence of key early Pleistocene taxa may be due to the sparse nature of the assemblages. Sequence lP1 is dated as 1.3 ± 0.3 and 1.1 ± 0.2 Ma (1σ) using Sr isotopes (errors derived from Supplemental Fig. 3 [see footnote 3]), and the age of the sequence is constrained by superposition to be older than the ca. 1–1.5 Ma sequence above it. We suggest that it may correlate with MIC 45 or 47 because these are intervals of peak global sea level inferred from the oxygen isotopic record (Fig. 9). Sequence lP2 is dated as early Pleistocene (ca. 1.0–1.5 Ma) based on AAR. We tentatively correlate this sequence with MIC 31 because this was an interval of peak global sea level of the early Pleistocene inferred from the oxygen isotopic record (ca. 1 Ma; Fig. 9).

Sequence uP1 is correlative to either MIC 7 or MIC 5e (Fig. 10). The presence of E. huxleyi (Zone NN21; younger than 290 ka) and the low abundance of G. muellerae suggest that the samples are older than ca. 165 ka (Hine and Weaver, 1998; Flores et al., 2003; Fig. 10). This would suggest correlation to MIC 7 (Fig. 10); however, there is enough variation in nannofossil assemblages following the FCO of G. muellerae to allow correlation to MIC 5 (e.g., decreased abundance of G. muellerae in early MIC 5; Flores et al., 2003). Based on Uniboom data in Carey et al. (2005), sequence uP1 appears to be the equivalent of their sequence I. Carey et al. (2005) suggested that sequence I correlates with MIC 5 on the basis of: (1) AAR results that indicate an age older than 50 ka for this sequence at Barnegat Inlet, ∼40 km landward of Expedition 313 sites; and (2) it is above the FO datum of E. huxleyi (younger than 290 ka) at AMCOR 6010 on the outer continental shelf, ∼75 km seaward.

Sequence uP2 was deposited in an incised valley at Site M29 that Carey et al. (2005) identified as cut by their sequence II boundary. Thus, sequence II of Carey et al. (2005) correlates with our sequence uP2, although they correlated sequence II to MIC 3c. We prefer a correlation of the fill to MIC 5c based on AAR estimates on the younger sequence uP3. At Site M27, AAR dates suggest an age older than 55 ka (Fig. 10), whereas nannofossils suggest this sequence predated the E. huxleyi FCO datum level (73 ka; Fig. 10), and thus we favor correlation of uP3 to MIC 5a. The sequence uP2 must be older and our preferred interpretation is that the uP2 sequence boundary was cut during the MIC 5e-5d eustatic lowering. Carey et al. (2005) based their age interpretation of their sequence II on AAR results and ambiguous radiocarbon dates of 44 to older than 49 ka at AMCOR 6020, ∼20 km south of Site M27. There, the AAR results are consistent with correlation to either MIC 3 or MIC 5. Still, we cannot rule out a correlation to MIC 3 if the E. huxleyi acme event is slightly younger (ca. 45–60 ka; Jordan et al., 1996).

Sequence uP3 is dated as older than the FCO of E. huxleyi (older than 73 ka) at Site M27, AAR ages that correlate it to MIC 5 at Site 27, and a radiocarbon date older than 52 ka at Site M29 (Fig. 11). We favor correlation to MIC 5a. Based on Uniboom data in Carey et al. (2005), sequence uP3 appears to be the equivalent of their sequence III that they interpreted as correlative with MIC 3a; they based this on radiocarbon dates of 29–43 ka reported by Duncan et al. (2000) and 28–36 ka of Knebel et al. (1979). This could be explained by uncertain seismic correlations of this unit to the Expedition 313 area; however, as with the underlying sequence, we cannot rule out a correlation to MIC 3 if the FCO of E. huxleyi is slightly younger, but favor the correlation to MIC 5a based on an age older than 73 ka for this event.

Expedition 313 did not core sequence uP4 (Fig. 8). It is seismically imaged in the Chirp sonar data, where it is preferentially preserved in shelf valleys (Fig. 3). Elsewhere in this area, the equivalent sequence IV of Carey et al. (2005) consists of TST sediments deposited during the sea-level rise following the Last Glacial Maximum (LGM; MIC 2). Sequence uP4 has been sampled and radiocarbon dated at several nearshore locations (Ashley et al., 1991; Carey et al., 2005) as a latest Pleistocene to Holocene sequence (IV of Carey et al., 2005).

Systems Tracts and Incised Valleys

The Pleistocene sequences are interpreted as remnants of TSTs and HSTs, with LSTs only possibly preserved as shelf valley fill. Sequence lP1 appears to contain thin portions of both the TST and HST at Site M27 (Fig. 7). Sequence lP2 appears to represent a thin remnant of a TST at Site M27 (Fig. 7). Neither was sampled at Site M29 (Fig. 11). Sequence uP1 may contain a very thin remnant of a TST at Site M27, but is primarily a relatively thick (∼8 m) HST (Fig. 7); it was cut out at Site M29 (Fig. 11). Sequence uP2 is cut out at Site M27 (Fig. 2). At Site M29 it is entirely represented by the incised valley fill and was apparently deposited either during the MIC 5d-5c transition (our preferred interpretation) or the MIC 4-3c transition (Carey et al., 2005). Sequence uP3 at Site M27 contains a TST, an HST, and a valley fill deposited during either MIC 5a (our preferred interpretation) or MIC 3a (Carey et al., 2005). Sequence uP3 appears to consist of a thin TST and thicker (∼9 m) HST at Site M29 (Fig. 11).

The Holocene TST is poorly represented in the inner-middle shelf region of Expedition 313. Holocene TSTs are well represented closer to the modern shoreline, where seismic profiles and Vibracoring data document the facies (e.g., Ashley et al., 1991; Sheridan et al., 1974). In addition, upper Pleistocene–Holocene sequence uP4 is well represented a few kilometers seaward on the mid-shelf sediment wedge (fig. 4 of Carey et al., 2005). Strata in equivalent water depths have been eroded in the immediate area of Expedition 313, presumably by submarine currents.

Previous studies reveal evidence of Pleistocene incised valleys that, because of their orientation and geomorphology, were interpreted to be the ancestral Hudson River drainage and ancestral Hudson shelf valley (Fig. 1). Two distinct incised valleys were detected in the preexisting seismic data: the upper was termed the sequence III (our uP3) incised valley, whereas the underlying was named the sequence II (our uP2) incised valley (Fig. 1; Carey et al., 2005). Knebel et al. (1979) first identified the incised valley associated with the uP3 sequence and called it the paleo–Hudson River Valley (Carey et al., 2005). Based on mapping using the previous seismic reflection data, and the high-resolution air gun data on line Oc270 529, it appears that Site M27 penetrated the Pleistocene near the eastern flank of the uP3 paleo–Hudson valley and Site M29 penetrated the Pleistocene in the center of the sequence uP2 paleo–Hudson incised valley (Fig. 1; Carey et al., 2005). The shelf valley fill at Site M27 appears to be an upper shoreface deposit that coarsens upsection, and thus is inferred to be regressive (Fig. 3). It is not possible to unequivocally place these valley-fill deposits into a sequence stratigraphic framework because they could be interpreted as forming during the regressive lowstand (and thus represent LST) or during the regressive HST. We favor the latter, considering the upper shoreface nature of the fill at Site M27. The sediments in the valley fill at Site M29 are very poorly sampled and the environment uncertain; however, interpretation of the gamma log suggests that much of the section is a regressive LST, whereas the top of the section fines upward and is interpreted as a TST. In both incised valleys, the indicated sea levels and paleoenvironments, taken together, document the late Pleistocene geologic history of the New Jersey continental shelf. Previous seismic mapping of these sequence II–uP2 and III–uP3 valleys (Knebel et al., 1979; Carey et al., 2005) show that they are the ancient equivalents of the modern Hudson shelf valley (Fig. 1). It is interesting that the more southern trends of the uP3 and uP2 Hudson valley (Fig. 1) seem to relate to the fact that, at those times, the Hudson River drainage was south of the Laurentide ice sheet edge. However, in the modern MIC 1 and MIC 2 Hudson shelf valley, the more eastern to southeastern trend of the valley (Fig. 1) reflects its origins when the ice sheet edge reached New Jersey and Long Island, and there was a more east-west–trending forebulge influence on post-LGM (MIC 2) drainage (Carey et al., 2005).

Sea-Level History

Sequences recovered at Sites M27 and M29 have implications for global sea-level changes and their impact on margin sedimentation. Sequence uP1 was deposited in the deepest water depth (∼30–40 m) at Site 27 (33.5 m present water depth; Fig. 7), consistent with correlation to peak global sea levels during MIC 5e (∼8 m above present; Kopp et al., 2009). The base of the shelf valley associated with the basal sequence boundary of uP2 at Site 29 is ∼45 mcd (Fig. 5). Given the water depth at this site (35.9 m), this suggests ∼80 m of relative sea-level fall that we correlate with the global fall from MIC 5e to MIC 5d. However, it is possible that fluvial incision has cut the valley below sea level, and thus the ∼80 m fall must be considered an upper limit. Based on data from Papua New Guinea (Chappell et al., 1996; Cutler et al., 2003) and the Red Sea proxy record (Siddall et al., 2003), global sea level dropped ∼60 m from MIC 5e to MIC 5d (Fig. 2 in Wright et al., 2009). This is consistent with our estimates, considering that we have not accounted for the effects of compaction, loading, and nonthermal subsidence and uplift (including glacial isostatic adjustment, GIA, effects; e.g., Peltier, 1998). However, other studies on the middle Atlantic coast and Florida suggest that the sea-level fall from MIC 5e to 5d was only ∼25 m (Toscano and York, 1992; Toscano and Lundberg, 1999).

Global sea-level records make a prediction about our correlation of sequences uP2 and uP3 with MIC 5c and MIC 5a, respectively, versus MIC 3c and MIC 3a, respectively (Carey et al., 2005). Though there is considerable uncertainty, based on dating of coral in Papua New Guinea and the scaled Red Sea record, global sea level during MIC 5a and 5c appears to have been ∼15–30 m below present; in contrast, peak sea level during MIC 3c was ∼30–60 m and MIC 3a was 40–60 m below present (Chappell et al., 1996; Cutler et al., 2003; Wright et al., 2009). Other syntheses suggest a global peak of 60 m below present during MIC 3 (Siddall et al., 2008). Scaling of the δ18O record to sea level is particularly uncertain in this time interval due to the nature and timing of cooling that occurred following MIC 5e (Wright et al., 2009). We observe that relative sea level at Site M27 was ∼35 m below present during deposition of sequences uP2 and uP3 (∼5–10 m paleodepth, at 5–14 mcd, in 33.5 m modern water depth). This would seem to be consistent with correlation to MIC 5c and 5a; however, on the middle Atlantic coast and Florida, MIC 5a and 5c have been found at +3 to -12 m (Szabo, 1985; Toscano and York, 1992; Toscano and Lundberg, 1999), although the entire middle Atlantic record (including ours) might be strongly affected by GIA effects (Potter and Lambeck, 2004). Future work is needed to backstrip these sites, removing the effects of thermal subsidence (which was minimal during the Pleistocene), compaction, and loading (e.g., Kominz et al., 2008), and to model the GIA effects.


The Pleistocene New Jersey inner-middle continental shelf is analogous to that of the Virginia and North Carolina shelves and adjacent Albemarle Sound. Foyle and Oertel (1997) noted similar features in the Pleistocene of offshore Virginia where they identified six thin sequences consisting of thin fragments of TSTs and HSTs. Mallinson et al. (2005) used high-resolution seismic profiles to identify and map 18 Pleistocene–Holocene seismic units (sequences and parasequences) in Albemarle Sound that were dated with AAR. Like New Jersey, these Pleistocene sequences were complexly cut and filled, in this case by the paleo–Roanoke River. Mallinson et al. (2010) identified six sequences in the Cape Hatteras, North Carolina area and provided age constraints using Sr isotopes and AAR. These studies all point to a similar conclusion, i.e., Pleistocene sequences on this margin are patchy, thin, and difficult to correlate with precision, and it appears that only a handful of the peak Pleistocene sea levels are preserved in any one location.

Pleistocene sequences on the New Jersey inner to middle continental shelf contrast markedly with Miocene and older sequences. Pliocene sequences are absent in the Expedition 313 area and the outer continental shelf (Mountain et al., 2007) and are restricted to upland gravels onshore (Stanford et al., 2002), although they are present and marine onshore in Virginia and Maryland; these differences must be attributed to tectonic, most likely due to GIA effects (Raymo et al., 2011). Miocene sequences tend to be relatively thick onshore (tens of meters to 100 m; Miller et al., 1997b) and are very thick offshore (often 100+ m; Mountain et al., 2010). In contrast, Pleistocene sequences at Sites M27 and M29 are thinner (3–10 m), except in incised valleys where they are nearly 30 m thick (Fig. 11). Onshore, Miocene sequences preserve thin TSTs with relatively thick HSTs (Miller et al., 1997a), whereas offshore sequences preserve very thin TSTs and thick LSTs and HSTs (Monteverde et al., 2008). Pleistocene sequences at Sites M27 and M29 are all highly dissected, preserving only portions of TSTs and HSTs, with LSTs preserved only in incised valleys. We suspect that preservation of sequences occurs only during peak eustatic events (e.g., MIC 45–47, MIC 31, and MIC 5) unless in eroded valleys; the absence of sequences during other peak eustatic events (e.g., MIC 11; Raymo and Mitrovica, 2012) suggests recurring subaerial exposure and erosion that removed large portions of the stratigraphic record. As a result, Pleistocene sequences preserve a patchwork of eustatic peaks. We attribute differences of Pleistocene versus Miocene sequences to the higher amplitude eustatic variations of the Pleistocene versus the Miocene.


Drilling by Expedition 313 provided a means of evaluating Pleistocene sequences. Our preferred chronology interprets sequences preserved during peak eustatic events (e.g., MIC 45–47, MIC 31, MIC 7–5e, MIC 5c, and MIC 5a), although age uncertainties remain and it is possible that uP2 and uP3 correlate with MIC 3, as suggested previously. Our studies show that thin TSTs and HSTs are preserved in a patchy distribution on the inner-middle shelf, although the fill of the paleo–Hudson shelf valley preserves lowstand deposits. Pleistocene sequences differ markedly from Miocene sequences in this region in that they are thinner and less complete, a pattern that we attribute to higher amplitude eustatic changes in the Pleistocene.

We thank the drillers and scientists of Expedition 313 for their enthusiastic collaboration and the Bremen core repository for hosting our studies. Seismic data were collected on cruise Oc270 (G. Mountain, J. Austin, co-chiefs) and En270 (G. Mountain and N. Christie-Blick, co-chiefs), and seismic data were provided by G. Mountain. Funding was supplied by COL/USSP (Consortium for Ocean Leadership/U.S. Science Support Program), samples were provided by the Integrated Ocean Drilling Program and the International Continental Scientific Drilling Program, and the radiocarbon date was provided by the National Ocean Sciences Accelerator Mass Spectrometry Facility (Woods Hole). We thank two anonymous reviewers, the associate editor, and the editor for helpful comments.

1Supplemental Figure 1. Detailed track map of line Oc270 529 (common depth point, CDP) with the En 370 Chirp (Compressed high-intensity radar pulse) sonar data line (time of day [TOD]) closest to Site M27. Also located are near-surface filled channels detected by the Chirp sonar data. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00795.S1 or the full-text article on www.gsapubs.org to view Supplemental Figure 1.
2Supplemental Figure 2. Detailed track map of line Oc270 529 and En370 Chirp sonar data, with Site M29. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00795.S2 or the full-text article on www.gsapubs.org to view Supplemental Figure 2.
3Supplemental Figure 3. Sr isotope resolution for the Pleistocene computed based on the look-up tables of Howarth and McArthur (1997) and McArthur et al. (2001) using their LOWESS (locally weighted scatterplot smoothing) estimates (95% confidence interval) of error of the regression of Sr versus ages, Sr versus age interpolated to 0.05 m.y. increments, and an external precision of ±0.000008 (1σ). Error is ± value shown (e.g., error for a 1 Ma sample is ∼0.2 m.y., for a 1.5 Ma sample is ∼±0.25 m.y.). If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00795.S3 or the full-text article on www.gsapubs.org to view Supplemental Figure 3.