The neodymium isotopic (εNd) signature is a widely used proxy in determining changes to ocean currents through Earth’s history. The application of εNd as a proxy of water-mass circulation is based on an assumed quasi-conservative behavior of neodymium isotopes in the ocean. However, our understanding of the factors that govern the oceanic budget of neodymium, including the mechanisms controlling dominant sources and sinks, is not yet complete. Here, I present pore water and water-column neodymium concentration profiles from the western edge of the Tasman Sea to examine the influence of a benthic flux on the northward-flowing Circumpolar Deep Water (CDW). I find that the flux from this region’s calcareous sediments is similar in magnitude to the benthic flux observed in the North Pacific Ocean. In addition, water-column data show a significant increase in the neodymium concentration of the CDW as it moves through the Tasman Sea. Together, these findings indicate that regions with calcareous sediments can account for an important component of the benthic source of neodymium to the oceans.


The neodymium isotopic (εNd) signature of fossil fish teeth, foraminifera, and ferromanganese coatings within marine sediments is widely used for paleoclimatic reconstructions of ocean circulation (e.g., Frank, 2002). The assumptions underpinning these reconstructions are that (1) the εNd of each water mass is distinct and quasi-conservative, and (2) the water-mass signature is imprinted on and can be recovered from fossil fish teeth, foraminifera, or sedimentary coatings. Where these assumptions are met, the εNd proxy provides unique constraints on the movement and origin of water masses during critical periods of Earth history (e.g., glacial-interglacial cycles, opening of the Tasmanian gateway; Burton and Vance, 2000; Scher et al., 2015). However, observations of rapid changes in εNd of bottom water (e.g., Lacan and Jeandel, 2005; Lambelet et al., 2016), of mobilization of Nd during early sediment diagenesis (Abbott et al., 2015a, 2015b, 2016), and from models suggesting that up to 95% of oceanic neodymium may come from previously unaccounted for deep sources (Jones et al., 2008; Arsouze et al., 2009; Rempfer et al., 2011) have challenged the application of εNd as a strictly conservative water-mass tracer (e.g., Abbott et al., 2015b; Jang et al., 2017). These observations have led to the proposal of a benthic flux model in which a flux from the sediments via the pore waters is attributed for giving Nd isotopes their quasi-conservative behavior in a “bottom-up” view of rare earth element (REE) cycling in the ocean (Haley et al., 2017).

A substantial benthic source of Nd to the global ocean challenges the ability of the authigenic phase(s) to record seawater conditions (e.g., Abbott et al., 2016; Du et al., 2016; Jang et al., 2017), particularly if the authigenic phases are forming near the sediment-water interface (Kechiched et al., 2018). If authigenic records are influenced by the pore water, a new framework that can account for a benthic source from diverse sediment compositions and depositional environments is needed for the use of Nd isotopes as a paleocirculation tracer (Du et al., 2016; Haley et al., 2017). Because biogenic carbonate contains relatively little Nd compared to the Nd associated with Fe-Mn coatings (e.g., Parekh et al., 1977; Elderfield et al., 1981; Shaw and Wasserburg, 1985; Toyoda et al., 1990; Chavagnac et al., 2005; Zhang et al., 2017), regions dominated by calcareous sediments have been considered an unlikely contributor of a benthic source of Nd to the ocean (Abbott et al., 2015a). Here, I present pore water profiles from three sites in the Tasman Sea that suggest that regions with dominantly calcareous sediments can be a significant source of Nd to the global ocean through a sedimentary benthic flux.


Pore water and sediment samples were collected at three sites off New South Wales (Australia), in the Tasman Sea, from R/V Investigator in September 2016 during the Australian Marine National Facility IN2016_v04 cruise (Fig. 1; see Table DR1 in the GSA Data Repository1 for sample locations). Pore water and sediment sample sites were located on the continental shelf at a water depth of 150 m (site PH150), on the continental slope at a water depth of 1550 m (site JB1500), and at the base of a submarine canyon at a water depth of 2660 m (site CB2600; see the Data Repository). Additionally, water-column samples were collected at two offshore sites (CTD23 and CTD54) in water depths >4000 m (Fig. 1; Table DR1) using Standard PVC Niskin bottles. Water-column samples were immediately filtered through 0.45 μm Sartorius Stedim single-use filters using an HCl-cleaned 25 mL syringe, and refrigerated. Pore waters were collected from sediments from a KC 6 × Ø100 mm model 70 multi-corer. Core samples were sectioned into HCl-cleaned 85 mL centrifuge tubes in a clean, inert environment (HEPA-filtered N2) then centrifuged at 4500 rpm for 20 min. Each centrifuge tube held approximately a 1 cm slice of sediment from the core. Pore waters were then filtered through 0.45 μm Sartorius Stedim filters using an HCl-cleaned 25 mL syringe in an inert environment (HEPA-filtered N2). A more detailed description of pore water collection is available in Abbott et al. (2015a). All water samples were acidified to pH ≤ 2 using quartz-distilled concentrated HCl. Water samples were analyzed for Nd using an ESI SeaFAST II (automated preconcentration system for undiluted seawater) in line with the Thermo VG ExCell quadropole inductively coupled plasma–mass spectrometer (ICP-MS) at the W.M. Keck Collaboratory for Plasma Spectrometry at Oregon State University, Oregon, USA (after Yang and Haley, 2016). A large-volume seawater sample (NBP95R10) from 1152 m depth in the Bransfield Strait in the Southern Ocean (62°46′S, 59°24′W) was used as an in-house consistency standard (mean 23.4 pM Nd, 1σ = 3 pM, n = 17) and 2% nitric acid as a blank (see the Data Repository; mean 1.4 pM Nd, n = 20).


Pore water Nd concentrations ranged between 28 pM and 350 pM, as much as an order of magnitude higher than water-column Nd concentrations (Figs. 2 and 3). The largest range in pore water Nd concentrations was observed at site JB1500 (from 40 pM at 17 cm sediment depth to 350 pM at 1.2 cm sediment depth), and the smallest range was observed at site CB2600 (from 40 pM to 110 pM; Fig. 2). Pore water Nd concentrations are not correlated to dissolved iron, dissolved manganese, or dissolved phosphorous at any of the sites (see the Data Repository). While pore water Nd concentrations are consistently greater than water-column Nd concentrations, the enrichment is not as high as previously observed (Abbott et al., 2015a; Deng et al., 2017), consistent with calcareous sediments typically being associated with relatively low REE concentrations (Parekh et al., 1977; Elderfield et al., 1981; Shaw and Wasserburg, 1985; Toyoda et al., 1990; Zhang et al., 2017). Due to these relatively low concentrations, calcareous sediments have been dismissed as a likely contributor to the oceanic Nd budget (Abbott et al., 2015a). However, the gradient from the pore water to the overlying water, which exists because the highest observed pore water concentrations are <2 cm below the sediment-water interface, indicates that calcareous sediments may still be a significant source of Nd to the ocean, as I will demonstrate below.

To demonstrate the magnitude of this potential source, I calculate the flux at sites PH150 and JB1500 due to diffusion, following the methods of Haley and Klinkhammer (2003):
where Ds is the diffusion coefficient corrected for temperature and tortuosity (Abbott et al., 2015a), and ∆C/∆z is the concentration gradient between the pore water concentration maximum and the bottom water. This calculation results in a flux of 12.5 pmol cm–2 yr–1 at site PH150 and 25.1 pmol cm–2 yr–1 at site JB1500. These calculations are valid only for sites dominated by diffusion; therefore, calculations were not attempted at site CB2600 because the site’s location at a canyon bottom means that advective processes such as rapid sediment transport likely render these calculations inappropriate for estimating sediment-water exchange rates (see the Data Repository; e.g., Elrod et al., 2004; Schmidt and De Deckker, 2015). The sampling resolution and the shallow subsurface neodymium concentration maximum at PH150 and JB1500 only allowed for linear flux calculations, likely resulting in an underestimate of the flux (Abbott et al., 2015a). However, the resulting fluxes are similar to those observed in the North Pacific (sample sites HH1200, 22 pmol cm–2 yr–1; HH3000, 26 pmol cm–2 yr–1; Abbott et al., 2015a) when calculated using the same linear equation. These calculations suggest that regions consisting dominantly of calcareous sediments can be important contributors to the global benthic flux of Nd to the ocean. As calculated by Abbott et al. (2015a), global benthic flux estimates for Nd increase by 30%–50% if calcareous sediments have a similar-magnitude benthic flux to the sediments in the North Pacific. Specifically, Abbott and coauthors showed that if only the abyssal plains and hills are considered (an area of 151.5 × 106 km2), the benthic flux increases by 30% from 37 × 106 to 48 × 106 mol Nd yr–1 with the addition of regions characterized by calcareous sediments (Abbott et al., 2015a). Similarly, if we consider a wider range of depths (abyssal plains, hills, continental rise, oceanic rise, oceanic ridge, continental slope, and continental shelf) for an area of 344.8 × 106 km2, the benthic flux increases by 50% from 73 × 106 to 110 × 106 mol Nd yr–1 with the addition of regions characterized by calcareous sediments (Abbott et al., 2015a).

A pore water flux driven by a sedimentary source has potential implications on the use of authigenic phases for past water-mass circulation studies. The shallow subsurface maximums observed at sites JB1500 and at PH150 indicate that the mechanism releasing REEs during early diagenesis from these sediments can occur in close proximity to the sediment-water interface, and therefore could exert an influence on phases forming at or near the time of deposition. For instance, the fluorapatite of fish teeth after deposition contains three to six orders of magnitude more Nd than that in living fish teeth (e.g., Toyoda and Tokonami, 1990; Martin and Haley, 2000; Huck et al., 2016). This post-depositional enrichment is likely driven by pore water Nd concentrations, meaning that the authigenic εNd records recovered from fish teeth may be influenced by the sedimentary processes that drive the benthic flux. Previous work has shown that the εNd of the pore water may be determined by a fraction of the bulk sediment being highly reactive, and therefore potentially subject to the influence of sedimentation rate or shifts in sediment chemistry (e.g., Freslon et al., 2014; Rousseau et al., 2015; Abbott et al., 2016). A signature driven by a small reactive fraction of the bulk sediment could be sensitive to even small shifts in sediment provenance (e.g., Rousseau et al., 2015; Du et al., 2016). The potential importance of a small fraction of the sediment implies that large-scale extrapolations of a benthic flux may not be viable based solely on sediment classifications, and rather may require a more detailed analysis of sediment composition. For instance, if iron oxides or iron-rich clays are the phase acting as the primary source of Nd due to reductive dissolution, then we may expect a greater benthic flux of Nd in regions with a higher flux of organic matter to the sediment because increased respiration would deplete pore water oxygen and facilitate a reducing environment. If, instead, a calcareous phase is driving the flux, then we may expect pH and related factors to have a greater role in predicting a benthic flux. Further complicating the identification of such a mechanism is the lower abundance of REEs relative to iron. The result of this difference in abundance is that the REEs could be sensitive to a change in iron concentration that is small enough to be below the analytical precision for iron (e.g., Abbott et al., 2015a). The identification of this sedimentary phase or phases interacting with the pore water and the period over which this interaction takes place would also have implications on the signature of the authigenic component and its ability to evolve during diagenesis.

Water-column Nd concentrations range between 0 pM and 50 pM, with only the bottom water at site CTD54 (CTD—conductivity, temperature, depth) exceeding 40 pM (Fig. 3). Generally, CTD54 Nd concentrations below 1500 m depth are at least double those at corresponding depths from site CTD23. The concentration of Nd in the deepest water sampled by the Niskin bottle rosette at both CTD23 and CTD54 is ∼10 pM higher than that of the second-deepest sample. This abrupt concentration change near the bottom indicates an upward flux rather than the smoother gradient associated with Nd desorption from sinking particulates (Crocket et al., 2018; Hathorne et al., 2015). The near-doubling of Nd concentrations from CTD23 to CTD54 (approximate 10° latitude difference) as the Circumpolar Deep Water (CDW) loops from south to north along Australia’s east coast (Figs. 1 and 3) is in contrast to Nd transects reported to show conservative behavior of the REEs within a water mass with a significant lateral transport component (e.g., Basak et al., 2015; Lambelet et al., 2016; Zheng et al., 2016; Behrens et al., 2018). Temperature and salinity data indicate that the same water masses are present at both CTD locations, making an additional source necessary to explain the observed non-conservative behavior of Nd (Fig. 4). The need for an additional source is consistent with the benthic flux suggested by the gradient between the pore water and bottom-water Nd concentrations and is further supported by the corresponding seawater REE patterns (see the Data Repository). Similar observations of large additions of Nd to the ocean from deep sources have been made in the Angola Basin off the southwest coast of Africa (Zheng et al., 2016), the Rockall Trough in the North Atlantic (Crocket et al., 2018), and the eastern North Pacific (Abbott et al., 2015a). These widespread observations indicate that the benthic flux model previously proposed (Abbott et al., 2015a, 2015b; Du et al., 2016; Haley et al., 2017) may be widely applicable throughout the global ocean.

In summary, this study demonstrates non-conservative behavior of Nd in water masses transported laterally through the Tasman Sea, with the concentrations of neodymium in CDW nearly doubling after moving along Australia’s east coast. The diffusive flux of Nd from the sediments calculated in this region at sites PH150 and JB1500 illustrates a flux commensurate with that observed in the North Pacific, suggesting that regions with dominantly calcareous sediments could add 50% to existing estimates of a global benthic flux. These new data provide additional evidence that there is a large pore water (sedimentary) source of REEs to the oceans (Elderfield and Sholkovitz, 1987; Sholkovitz et al., 1989; Haley and Klinkhammer, 2003; Arsouze et al., 2009; Abbott et al., 2015a, 2015b; Zheng et al., 2016). Moreover, these findings expand the area over which the benthic flux model may be applicable, and support the need for further constraints on the implications of a benthic flux on the interpretation of Nd isotopes as a paleocirculation proxy.


Macquarie University (Sydney, Australia) supported this study. I thank the Australian Marine National Facility, the captain and crew of R/V Investigator, Chief Scientist Martina Doblin, and the science party of cruise IN2016_V04 for their support during the expedition. I appreciate many other individuals for their assistance: Rebecca Darcy, Brian Haley, Jesse Muratli, Damian Gore, Beth Rutilia, Martin Ostrowski, Peter Weiland, Bruce Schaefer, Morgan Perrone, Stephen Tibben, and Kirianne Goosen for help during the field and/or laboratory portions of this research; Alan Mix and Stefan Löhr for helpful discussions; Geomatics Research LLC (Seattle, Washington, USA) for software development to aid in data processing; Chris Russo for ICP-MS facility support at Oregon State University’s W.M. Keck Collaboratory (Corvallis, Oregon, USA); and Russell Field for X-ray diffraction support at Macquarie University, Australia. Two anonymous reviewers provided thoughtful and constructive comments that improved the text.

1GSA Data Repository item 2019124, water-column and pore water neodymium concentrations from the Tasman Sea; water-column REE patterns; pore water iron, manganese, and phosphorous concentrations; and site CB2600 grain size and site map, is available online at http://www.geosociety.org/datarepository/2019/, or on request from editing@geosociety.org.