Dissolved rare earth elements and hydrography in the Sulu Sea

Dissolved rare earth elements and hydrography in the Sulu Sea

Geochimica et Cosmochimica Acta, Vol. 63, No. 15, pp. 2171–2181, 1999 Copyright © 1999 Elsevier Science Ltd Printed in the USA. All rights reserved 00...

437KB Sizes 1 Downloads 27 Views

Geochimica et Cosmochimica Acta, Vol. 63, No. 15, pp. 2171–2181, 1999 Copyright © 1999 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/99 $20.00 ⫹ .00


PII S0016-7037(99)00142-8

Dissolved rare earth elements and hydrography in the Sulu Sea YOSHIYUKI NOZAKI,* DIA-SOTTO ALIBO, HIROSHI AMAKAWA, TOSHITAKA GAMO, and HIROSHI HASUMOTO The Ocean Research Institute, University of Tokyo, Nakano-ku, Tokyo 164-8639, Japan (Received December 23, 1998; accepted in revised form March 24, 1999)

Abstract—Because most Southeast Asian basins are flushed rapidly by waters from the western Pacific, the effects of respiration and silica dissolution within those basins are hardly discernible based on general hydrographic and geochemical observations (Broecker et al., 1987). However, a different situation is expected for the Sulu Sea because its deep water is isolated from other deep sea basins by shallow sills of ⬍400 m depth. Hence, any heterogeneity of elemental distributions within the basin must be ascribed to vertical physical and biogeochemical processes. We have obtained detailed vertical profiles of dissolved rare earth elements (REEs) together with conductivity-temperature-depth (CTD) and hydrographic measurements in the Sulu Sea during the 1996 –97 Hakuho-Maru cruise. Vertical profiles clearly indicate that the REE(III)s are enriched in the deep water due to regeneration and are involved in vertical biogeochemical cycling. The light REE(III)s (La, Pr, and Nd) showed a subsurface maximum at 500 m and a minimum ⬃1500 m in depth. None of the hydrographic properties or nutrients shows such features; therefore, the REE(III)s and their elemental ratios may be useful as tracers of water masses. The relationship of REE(III)s versus dissolved Si suggests that the REE(III)s are regenerated in a delayed fashion relative to Si, and it is likely that their dissolution largely occurs at the bottom interface. The mean residence time of the deep water is estimated to be 300 ⫾ 150 yr. Consequently, the calculated benthic flux of REE(III)s becomes roughly comparable with the river and atmospheric fluxes. Because of oxidation to Ce(IV), the Ce profile shows a decrease from ⬃6 pmol/kg at the surface to a constant value at 3.9 pmol/kg below 1500 m, unlike the other REEs. The REEs in the Sulu Sea are enriched in the middle REEs and Ce relative to those of the North Pacific Deep Water and must derive from local sources around the Sulu Sea. Copyright © 1999 Elsevier Science Ltd surements indicated that salinity and temperature inside the basin exhibit nonuniform and complex vertical features that may be related to the renewal mechanism of the deep water. This paper describes the results of dissolved rare earth element (REE) analyses that show vertical profiles influenced by physical and biogeochemical processes active within the basin. In particular, we show that dissolution of particles at the sedimentwater interface is an important source for dissolved REEs whereas physical processes within the basin are of primary importance in governing the redistribution of dissolved REEs within the overlying water column of the Sulu Sea.


Among the Southeast Asian basins, the Sulu Sea exhibits unique hydrography such that potential temperature is higher than 9.8°C throughout the water column of ⬃5000 m (Wyrtki, 1961). Although the basin is located along the important pathways from the western Pacific to the Indian Ocean (Fig. 1), the shallow sill depths (⬍400 m) of the Mindoro, Sibutu, and other straits prevent deep water of the Sulu Sea from communicating with that of other deep sea basins, such as the South China, Philippine, and Celebes Seas. Because of this, the Sulu Sea provides an excellent opportunity to study purely vertical processes of physical mixing and biogeochemical cycling that govern the elemental distributions in the water column. However, hydrographic and geochemical investigations of the Sulu Sea have been few to date. Broecker et al. (1986) have summarized the hydrography, 14C, and 226Ra measurements made in the Southeast Asian Basins during two cruises (the 1976 –77 INDOPAC and Vema 33 expeditions) and confirmed the earlier description given by Wyrtki (1961) that the deep waters in those basins are rapidly fed and flushed with deep water from the western North Pacific. However, very little attention was paid at that time to the hydrography, geochemistry, and flushing rate of deep water in the Sulu Sea. During the 1996 –97 R. V. Hakuho-Maru cruise, we had occupied a single station in the southeastern basin of the Sulu Sea to collect various hydrographic and geochemical data. High-resolution conductivity-temperature-depth (CTD) mea-

2. MATERIALS AND METHODS During the Piscis Austrinus Expedition with R. V. Hakuko-Maru, waters were collected at the Sulu Sea station (PA-1 in Fig. 1; 8°50⬘N, 121°48⬘E; water depth, 4950 m) on December 25, 1996. The South China Sea station (PA-11 in Fig. 1; 15°25⬘N, 115°20⬘E) also was occupied during the same cruise on February 10 –11, 1997. For sampling, we used 12-l lever-action type Niskin bottles mounted on a 36-position Sea-Bird’s 911plus CTD-rosette, hung from a titanium armored cable. The Niskin bottles were pre-cleaned successively with distilled HCl and deionized water. Immediately after sampling, seawaters were filtered, onboard, through a 0.04 ␮m hollow-fiber membrane (Millipore HF-400) in a built-in clean room of the vessel. The use of the filter is based on the smaller pore size and rapid filtration and is almost free from contamination (Alibo and Nozaki, 1999). Filtered samples were transferred into pre-cleaned 5-l plastic cubic containers and then acidified to pH ⬍ 1.5 with ultrapure HCl (TAMAPURE-AA-100). The container was wrapped in a plastic bag, placed in a large plastic box, and transferred to the Ocean Research Institute, University of Tokyo for REE analysis. The analytical method for RREs by inductively coupled plasma (ICP) mass spectrometry is identical to that described in Zhang and Nozaki (1996) and Alibo and Nozaki (1999). The method can determine yttrium and all lanthanides including monoisotopic Pr, Tb, Ho, and Tm

* Author to whom correspondence should be addressed ([email protected]). 2171


Y. Nozaki et al.

Fig. 1. Map showing the Southeast Asian basins (A) and bathymetry of the Sulu Sea enlarged in (B). The station locations of PA-1 in the Sulu Sea and PA-11 in the South China Sea where water samples also are obtained for measurements of dissolved REEs are indicated by the solid circle and cross, respectively. The open circle and triagle show the stations T97-2 (Alibo and Nozaki, 1999) and TPS 24, 271-1 (Piepgras and Jacobsen, 1992), respectively, in the western North Pacific where REE data are reported previously.

Rare earth elements in the Sulu Sea


Fig. 2. Vertical profiles of potential temperature and salinity (a). The temperature profiles indicated by the squares are enlarged in b and c.

with an accuracy and precision comparable to those of the isotope dilution method by using thermal ionization mass spectrometry (e.g., Piepgras and Jacobsen, 1992). 3. RESULTS

3.1. Hydrography The waters in the Sulu Sea are considered to flow from the South China Sea through the Mindoro Strait, which has a maximum sill depth of ⬃400 m. Consequently, there may be some complementary outflow to the Celebes Sea through the Sibutu Strait. Figure 2 shows the salinity and potential temperature profiles for the Sulu Sea on expanded scales focused on the deep water. The salinity profile indicates a sharp maximum of 34.523 psu around 190 m. Below this depth, it decreases sharply to a broad minimum of 34.440 psu between 700 m and 1100 m and then increases gradually toward the bottom up to 34.462 psu. Because the subsurface salinity minimum in the Sulu Sea occurs well below the sill depth, we must seek its cause. Figure 3a compares the Sulu Sea (PA-1) and the South China Sea (PA-11) stations in the ␪-S diagram. The inset area B is expanded below (Fig. 3b). In the South China Sea, a salinity maximum at 135 m and a salinity minimum ⬃520 m can be seen because of influences of the Kuroshio Current and the North Pacific Intermediate Water, respectively (Wyrtki, 1961). Nitani (1972) noted that the 84-S characteristics at ⬃3500 m in the South China Sea is simlar to that at ⬃2000 m in the east of Luzon Strait, suggesting that the deep water is flushed from the western North Pacific. The ␪-S curves for PA-1 and PA-11

crosses at ⬃350 m roughly corresponding to the sill depth. In the Sulu Sea, waters less saline than 34.440 psu at the subsurface salinity minimum are found only in depths shallower than 150 m, and the temperature of those waters is higher than 17.5°C (Fig. 2). Because the temperature of waters at the salinity minimum is lower than 10.2°C, it seems unlikely that the salinity minimum is formed by overturning of the surface water due to seasonal cooling: such a large change in water temperature has never been observed in the area. Thus, the low salinity waters observed at 700 to 1100 m depth must result from lateral transport of the South China Sea water into the basin, across the sill, and subsequent subduction. In addition, the waters with salinity higher than that of the bottom value (34.462 psu) only can be found between 170 and 310 m depth in the Sulu Sea (Fig. 2). Therefore, entrainment of the high-salinity water is necessary during the formation of bottom water, if its mechanism is operating at present. (There is no evaporites on the seabed that can raise the salinity of deep water). Potential temperature monotonically decreases from 27.63°C at the surface to 9.883 at ⬃3050 m and then appears to slightly increase with depth to 9.885 m at the bottom. Consequently, potential density rapidly increases from ␴␪ ⫽ 21.48 at the surface to 26.50 at the top of the salinity minimum (⬃750 m), and then the density gradient gradually decreases with depth. From the bottom of the salinity minimum at ⬃1100 m, density only increases slightly with depth but is almost linear from 26.525 to a constant value of 26.552 in bottom waters below ⬃3200 m. The bottom density corresponds to that of water at a depth of several hundred meters in open oceans


Y. Nozaki et al.

Fig. 3. The ␪-S diagram in the Sulu Sea and in the South China Sea (a). The area indicated by the square B is enlarged below.

like the South China and Philippine Seas. So, the entire basin is similar to much shallower open-ocean waters. The ␪-S diagram (Fig. 3b) shows nonlinear and complex features, suggesting that vertical flushing of the deep and bottom waters probably takes place in a complex manner. Although very little information is available on the renewal mechanism, subduction of high-density near-surface waters must reach to the bottom to refresh and ventilate the water column. It is likely that the renewal occurs sporadically along the side slopes, being followed by cascade-type vertical mixing. Hydrographic data of the Piscis Austrinus Expedition are available elsewhere (Gamo, 1997). The vertical profiles of dissolved oxygen, nutrients, pH, and alkalinity in the Sulu Basin are shown in Figure 4. Dissolved oxygen shows a sharp minimum of 1.55 mL/L at 300 m near the salinity maximum and a broad subsurface maximum between 500 and 700 m, just below sill depth. Dissolved oxygen then monotonically decreases with depth down to 1.2 mL/L at the bottom. All nutrients are depleted in the surface water. Nitrate, phosphate and pH values change abruptly from the surface down to 300 m, whereas below that depth their variations are relatively small. Dissolved Si concentrations increase from zero at the surface to 50 ␮M at 300 m and then gradually increase with depth to 97 ␮M near the bottom. Because the Sulu Sea basin is completely isolated from outside below 400 m, dissolved Si in the deep waters must have been regenerated largely within the basin. Table 2 compares hydrographic characteristics together with some rare earth element concentrations in deep waters around 4000 m for different oceanic basins. High potential temperature and low dissolved oxygen are pronounced in the Sulu Sea. Nutrients are lower in the Sulu Sea than the other basins, whereas pH and alkalinity are similar to each other. Dissolved

Fig. 4. The vertical profiles of dissolved oxygen, silicate, phosphate, nitrate, pH, and alkalinity.

Rare earth elements in the Sulu Sea


Table 1. Dissolved rare earth elements (in pmol/kg) in the Sulu Sea and average concentrations in the North Pacific Deep Water (NPDW). Depth (m)
















0 50 101 199 498 794 992 1240 1486 1733 1979 2226 2472 2717 2962 3207 3452 3697 3942 4185 4429 4672 4854 4899 4944

100 109 101 115 132 140 137 140 144 144 149 154 162 165 171 172 181 183 182 186 162 184 185 185 184

8.6 9.2 10.1 15.3 21.0 19.4 19.4 18.9 18.3 18.3 19.0 19.7 21.6 22.6 23.8 24.4 26.5 26.4 27.1 26.9 22.3 26.9 27.1 26.9 26.9

5.90 6.82 5.00 4.51 5.95 3.85 4.44 4.41 4.06 3.97 3.77 3.46 3.91 3.77 4.11 3.85 4.12 3.53 4.32 3.88 3.95 3.86 4.61 3.96 4.03

1.77 2.07 1.88 2.72 3.42 3.32 3.20 3.07 2.94 2.99 3.12 3.30 3.63 3.89 4.17 4.36 4.75 4.85 4.96 4.94 4.05 4.93 4.98 4.99 4.96

9.2 10.6 9.5 12.4 15.0 15.6 14.7 15.0 14.4 14.5 15.1 16.0 17.7 19.1 20.3 21.4 22.6 23.0 23.7 24.4 19.6 24.4 23.8 23.9 23.6

2.24 2.64 2.24 2.77 3.02 3.38 3.16 3.17 3.16 3.28 3.36 3.55 3.83 4.21 4.62 4.69 5.07 5.18 5.18 5.25 4.16 5.20 5.27 5.36 5.09

0.71 0.82 0.68 0.84 0.86 0.93 0.87 0.92 0.89 0.96 1.00 1.02 1.14 1.28 1.36 1.39 1.49 1.58 1.58 1.56 1.30 1.51 1.53 1.49 1.51

3.09 3.96 3.36 4.00 4.38 4.56 4.53 4.60 4.73 4.80 5.04 5.34 5.66 6.22 6.31 6.47 7.25 7.21 7.36 7.14 5.83 7.09 7.03 7.21 7.01

0.64 0.70 0.58 0.65 0.70 0.75 0.72 0.75 0.77 0.75 0.80 0.84 0.90 0.97 1.01 1.04 1.12 1.15 1.15 1.20 1.01 1.14 1.18 1.19 1.13

4.30 4.95 4.24 4.69 5.03 5.39 5.09 5.25 5.56 5.62 5.83 6.13 6.53 7.02 7.15 7.40 7.99 7.95 7.97 8.21 6.89 8.18 8.12 8.19 7.89

1.04 1.21 1.06 1.18 1.29 1.40 1.36 1.41 1.44 1.44 1.51 1.56 1.66 1.79 1.82 1.90 1.98 2.02 2.02 2.06 1.74 2.02 2.07 2.03 1.99

3.29 3.85 3.34 3.66 4.24 4.38 4.24 4.46 4.53 4.59 4.97 4.98 5.36 5.76 5.90 6.05 6.28 6.45 6.35 6.43 5.53 6.45 6.47 6.35 6.30

0.44 0.53 0.46 0.54 0.61 0.64 0.58 0.65 0.66 0.67 0.72 0.73 0.79 0.84 0.85 0.86 0.91 0.95 0.95 0.92 0.79 0.93 0.93 0.93 0.93

2.70 3.24 3.02 3.33 3.73 4.15 3.97 4.20 4.31 4.41 4.68 4.74 5.12 5.48 5.42 5.68 5.85 6.05 6.03 6.05 5.10 6.07 5.98 5.88 5.87

0.41 0.50 0.47 0.51 0.62 0.70 0.66 0.69 0.71 0.73 0.77 0.79 0.84 0.89 0.89 0.93 0.99 0.98 1.00 1.00 0.84 0.98 1.02 1.02 0.97

















* 2500 ⫾ 100 m (After Alibo and Nozaki, 1999).

SiO2 is more depleted relative to nitrate and phosphate in the Sulu Sea than in the other basins. 3.2. REEs The results of RRE analyses are given in Table 1. The REEs do not resemble any of the nutrients in their vertical profiles, although some correlation does exist between the heavy REE(III)s and dissolved Si or alkalinity. Shale-normalization generally has been used to discuss fractionation of the REEs (e.g., Ce anomalies and heavy REE enrichment that are characteristic of seawater; e.g., Elderfield, 1988; Alibo and Nozaki, 1999). Here, we use the average values of Post-Archean Australian Sedimentary rocks (PAAS; Taylor and McLennan, 1985) as representative of upper continental crust. Alternatively, for detailed arguments concerning fractionation between different water masses, we can normalize our data relative to the REE values of some reference water mass. Here, we use the average values obtained by Alibo and Nozaki (1999) for three deepwaters at 2400, 2500, and 2576 m depths in the western North Pacific (station T97-2 in Fig. 1; referred to as NPDW hereafter) because the Southeast Asian basins are believed to be fed from the deep western North Pacific (Broecker et al., 1986). Those REE data were obtained by the same method as described here, including 0.04 ␮m filtration, subsequent acidification treatment, and determination by ICP mass spectrometry. The data are summerized in the bottom column of Table 1. The REE composition of our NPDW shows a good agreement, when compared with the available data measured by isotope dilution method (Piepgras and Jacobsen, 1992) for unfiltered

waters at the station of 24°N, 150°E in the western North Pacific. Some REEs in the deep waters are compared in Table 2. Like dissolved SiO2, Y, and La are lower in the Sulu Sea than in the other basins. This trend varies across the lanthanide series and weakens at the middle REEs, e.g., Gd. Cerium, which may be in tetra-valency state, ranges from 4 to 6 pmol/kg for the deep waters. 4. DISCUSSION

4.1. Vertical Profiles of REE(III)s Figure 5 compares the vertical profiles of the REEs. The light REE(III)s (La, Pr, and Nd) show a pronounced subsurface maximum at 498 m and a minimum around 1500 m, which are not clearly seen in any of the profiles of dissolved oxygen, nutrients, pH, and alkalinity. This trend weakens as the atomic number increases and is not visible for the heavy REEs and yttrium, consistent with a systematic fractionation across the lanthanide series. Thus, the REE(III)s and their elemental ratios (e.g., Lu/Pr and Er/Nd; Fig. 6) have large signal-to-noise ratios in the water column and may be useful as tracers of water masses. For example, in the deep water between 1100 m and 3200 m, salinity and potential density vary almost linearly, whereas the light REE(III)s show a minimum around 1500 m and elemental ratios such as Lu/Pr and Er/Nd show a maximum at ⬃2000 m (Fig. 6). Note that there are some disturbances in the smooth curve of potential temperature at those depths (Fig.


Y. Nozaki et al. Table 2. Comparison of hydrographic properties and some REEs of the deep water at 4000 m depth in different oceanic basins.

␪ (°C) Salinity Dissolved O2 (ml/l) SiO2 (␮M) PO4 (␮M) NO3 (␮M) pH Alkalinity (meq/l) Y (pmol/kg) La (pmol/kg) Ce (pmol/kg) Gd (pmol/kg) Er (pmol/kg) a b

Sulu Sea (PA-1; 9°N, 122°E)

South China Sea (PA-11; 15°N, 115°E)

NW Pacific (Geosecs 224; 34°N, 142°E)

9.88 34.46 1.32 97 2.21 28.5 7.77 2.41 185 27 4.1 7.2 6.4

2.09 34.60 2.63 154 2.63 36.6 7.68 2.46 285 40 5.9 8.1 8.8

1.17 34.68 3.34 151 2.47 37.0 7.87 2.43 300a 55b 5.0b 9.5b 9.7b

After Nozaki et al. (1997) for unfiltered water at 29°N, 143°E. After Piepgras and Jacobsen (1992) for unfiltered water at 24°N, 150°E.

2b and c). Such a break may be an indication of meeting and mixing of two water masses. Below 3940 m, the REE(III)s reach their highest concentrations and become uniform with depth. Because such high REE(III) concentrations cannot be found at any other depth in the water column, these must have resulted from regeneration

within the deep water and/or at the bottom interface much like dissolved Si and alkalinity or represent in-flow; but it is unlikely for a water mass at ⬍1000 m to have such high REE concentrations. The nonuniformity of REE(III) concentrations when compared to the phosphate and nitrate profiles suggests that vertical mixing is limited relative to the biogechemical

Fig. 5. Vertical profiles of dissolved REEs (pmol/kg) in the Sulu Sea.

Rare earth elements in the Sulu Sea


Fig. 6. Vertical profiles of elemental ratios of Lu/Pr, Er/Nd, and Y/Ho.

regeneration cycle of the REE(III)s within the basin. This conclusion differs from observations in the other Southeast Asian (South China, Celebes, Mollucca, and Banda) basins (Broecker et al., 1986), which indicated that those basins are rapidly flushed, such that no evidence for respiration, silica dissolution, 14C decay, or 226Ra production can be seen there. From that work, Broecker et al. (1986) suggested that the flushing times for those basins did not exceed 100 yr. The light REE(III) maximum (498 m) underlies the depth of the salinity maximum (⬃190 m), which is probably formed by lateral flow from the South China Sea. It is not clear how this light REE(III) maximum is formed, although it is likely affected by the surface current system as the internal soliton waves propagate well over 500 m in the Sulu Sea (Apel et al., 1985). Note that Ce, although its vertical profile is different from the other REE, also shows a pronounced subsurface maximum at 498 m (Table 1 and Fig. 5). Such a high Ce concentration normally only can be seen in shallower depths in the open seas. The light REE(III) minimum ⬃1500 m is also deeper than the salinity minimum layer of 700 m to 1100 m (Fig. 2). Such a light REE(III) minimum can be caused by preferential scavenging over the middle and heavy REEs because the particle reactivities of the light and middle REEs are higher than those of the heavy REEs (Sholkovitz et al., 1994; Alibo and Nozaki, 1999). However, there is no evidence for particle scavenging in the NPDW-normalized REE patterns, as will be shown later. Rather, we believe that the mid-waters light REE minimum is formed as a result of the physical mixing of different water masses. Maintaining such stratification in steady-state would require sources of the low salinity and low REE(III) waters to those particular depths, but where and how those waters are supplied from and formed are not yet understood. Because no lateral transport is expected from other basins, their formation must occur within the Sulu Sea, perhaps

in association with the dynamics of a surface current system that varies seasonally: a strong southward flow in summer and a counterclockwise current in winter. To better understand the renewal mechanism, further investigations obviously are needed through time-series oceanographic observations as well as both theoretical and numerical modeling. Figure 7 shows a correlation diagram for some REE(III)s and dissolved Si. The curves become concave upward as dissolved Si approaches its bottom value, suggesting that the REE(III)s are regenerated more slowly than Si. A similar diagram (not shown) is obtained when one plots REEs versus alkalinity, such that the same conclusion can be drawn with respect to carbon-

Fig. 7. Correlation between REE(III)s (Pr, Sm, and Yb) vs. dissolved Si.


Y. Nozaki et al.

Fig. 9. Diagram showing correlation of Ce anomaly vs. dissolved oxygen content. Fig. 8. Vertical profile of Ce anomaly.

ate dissolution, although the REE contents in calcareous tests are relatively low. Thus, it is likely that the REE(III) distributions in the Sulu Sea are significantly influenced by regeneration across the sediment–water interface. 4.2. Ce and Y Profiles The Ce profile is uniquely different from those of the other REEs because of the oxidation of Ce(III) to insoluble Ce(IV). The Ce concentrations in the surface waters are high at 6 to 7 pmol/kg and somewhat variable with a distinct subsurface maximum at 498 m. There appears to be a secondary subsurface maximum at ⬃992 to 1240 m where a broad oxygen maximum exists (Fig. 4). Below 1500 m, the Ce concentration becomes constant at 3.94 ⫾ 0.28 pmol/kg, and there is no evidence that Ce oxidation continues in the deep sea. The Ce anomaly is calculated as Ce/Ce* ⫽ 2[Ce]/([La] ⫹ [Pr]), where [ ] indicates PAAS-normalized value. It is always ⬍1—the so called “negative Ce anomaly”—indicating a depletion of Ce in the seawater relative to the REE(III)s (Fig. 8). Although the Ce anomaly continually decreases from the surface toward the bottom, it may not be due to continued Ce oxidation at depths but rather due to preferential addition of La and Pr to the water by regeneration. This is because the Ce concentration does not show any decrease with depth (or time) and remains constant in the deep water, suggesting that removal of Ce from the waters by scavenging may be balanced by particle dissolution. It is interesting to note here that there exists a strongly positive correlation between Ce anomaly and dissolved oxygen (Fig. 9). This apparent correlation is understandable because the more oxygen is consumed for decomposition of particles, the more REE(III)s are generated to the water, resulting in a Ce/Ce* ratio decrease. The vertical profile of nonlanthanide Y resembles those of

the heavy REEs, in particular that of Ho (Nozaki et al., 1997). However, the Y/Ho molar ratios in seawater (Fig. 6c) are significantly higher than the value of 55 expected from crustal abundances, suggesting that fractionation between the two must take place in the marine environment. The Y/Ho ratio below 3200 m is ⬃90, which is lower by ⬃10 than the ratios in shallower depths, indicating that the REEs regenerated on the bottom must have a lower and more crust-like Y/Ho ratio than in ambient water. 4.3. NPDW-Normalized REE Patterns The NPDW normalized REE patterns are shown in Figure 10. The Ce values, although not plotted in the figure, were distinctly positive compared to that expected from the La and Pr values, indicating that Ce in the Sulu Sea waters are enriched relative to NPDW. The Y values also are not shown but closely follow the trend of heavy REEs, particularly Ho. Below 3200 m, the NPDW normalized patterns are almost identical and show marked middle REE enrichment with a maximum at Eu (Fig. 10c). This pattern is also maintained in the mid-depths (Fig. 10b), although the middle REE enrichment is reduced as depth decreases. A few remarks need to be made concerning Figure 10a for the shallower depth samples. The surface samples (0, 50, and 101 m) clearly show a depression at Gd that is characteristic of South China Sea waters (Fig. 11). Thus, it is clear that the surface water of the Sulu Sea is originated from the South China Sea, consistent with the hydrographic description of Wyrtki (1961). In contrast, the subsurface waters between 199 m and 1240 m, where a salinity maximum and a salinity minimum both exist, are relatively enriched in the light REEs as compared to those in other depths and do not show the REE patterns characteristic of the South China Sea. Therefore, there is a good chance that the waters at intermediate depth in the Sulu Sea originate from elsewhere. The best candidate is the Celebes Sea through the Sibutu Strait with a sill depth of

Rare earth elements in the Sulu Sea


Fig. 11. NPDW normalized REE patterns of the South China Sea waters from the PA-11station (Fig. 1; 25°25⬘N, 115°20⬘E). Note the distinct Gd depression for all the samples. After Alibo and Nozaki (submitted).

Fig. 10. NPDW normalized REE patterns of the Sulu Sea waters. The Ce and Y values are not plotted to avoid complexity.

340 m, but we have no data for REEs there to further this assessment. The high REE concentrations near the bottom of the water column (Fig. 5) and their middle REE enrichment (Fig. 10) clearly indicate that REEs in the deep Sulu Sea also must have derived from local sources with REE patterns quite different

from that of NPDW. The REE enrichment in the bottom water suggests that dissolution of particles at the sediment-water interface is a predominant source in the Sulu Sea. There is some evidence that terrestrial sources might show such a middle REE enrichment. Greaves et al. (1994) have conducted dissolution experiments on the Saharan dust in seawater that indicate a release of REEs with middle REE enrichment. Lerche and Nozaki (1997) also have shown that mild acid leaching of settling particles collected by sediment trap in the Japan Trench has resulted in a shale-normalized REE pattern with middle REE enrichment being measured in the leachate. Thus, there is a possibility that the Southeast Asian aeolian dust is an ultimate source for REEs characteristic of middle REE enrichment in the Sulu Sea. However, aeolian dust should also affect to the South China Sea, resulting in the similar REE pattern to that of the Sulu Sea. Obviously, this is not the case because the dissolved REE pattern in South China Sea waters, showing a marked Gd depression and a less pronounced Eu maximum (Fig. 11), is different from that in the deep Sulu Sea (Fig. 10). Alternatively, the supply of REEs from Southeast Asian rivers can have a significant influence on the REE pattern of local seawaters, although the river REE flux to the ocean may be significantly modified through estuarine and coastal reactions (Elderfield et al., 1990). In fact, we have measured dissolved REE concentrations in the Chao Praya river flowing into Gulf of Thailand that indicated a middle REE enriched pattern with a maximum at Eu relative to NPDW (Fig. 12; Nozaki, et al., submitted). Nevertheless, the Chao Praya river should influence more strongly the South China Sea than the Sulu Sea and, therefore, it seems difficult to ascribe the dissolved REE pattern in the Sulu Sea to large rivers from Asian continent. Thus, more local river drainage, from the Philippine-Indonesian Archipelago and other nearby volcanic islands, and dissolution of its suspended load are most likely to be an important source for REEs in the Sulu Sea. This is consistent with measurements of Nd isotopes made on surface samples collected during the same cruise; more radiogenic value (⑀Nd ⫽ ⫺1.4) due to volcanic sources for PA-1 in the Sulu Sea vs. less


Y. Nozaki et al.

Fig. 12. NPDW (a) and PAAS (b) normalized REE patterns of the Chao Praya River waters (Nozaki et al., submitted). Salinity for the samples is indicated in the parentheses.

radiogenic values (⑀Nd ⫽ ⫺6.4 to ⫺9.5) due to old continental sources in the South China Sea (Amakawa et al., submitted). Recently, Sholkovitz et al. (in press) have shown that island weathering is a significant source for dissolved REE in the equatorial Pacific. 4.4. Renewal of Deep Water and REE Regeneration Although we do not know how the deep waters of the Sulu Sea are ventilated, it is possible to put some constraints on the time scale. Broecker et al. (1986) obtained two radiocarbon data for 3250 m and 4775 m which were the same at ⌬14C ⫽ ⫺122‰. This corresponds to a radiocarbon age (or ventilation age) of 660 yr relative to the pre-bomb surface ⌬14C value of ⫺50‰. This should be regarded as an upper limit for flushing time if the deep water is renewed by entertainment of subsurface waters that elapsed for some time since the last atmospheric contact. Broecker et al. (1986) also reported a ⌬14C value of ⫺106‰ for 420 m in the South China Sea. Assuming that the deep water in the Sulu Sea originates from South China Sea water at a depth of ⬃420 m, then the flushing time estimated from the difference of their radiocarbon ages becomes ⬃150 yr. Dissolved oxygen in the deep waters is consumed by decomposition of organic matter. The deep water in the Sulu Sea

contains 55 ␮M of dissolved O2 and may be compared with the values of 101 ␮M at 420 m in the South China Sea or 93 ␮M at the subsurface oxygen maximum at 500 m in the Sulu Sea. Nozaki et al. (1998), on the basis of the GEOSECS Pacific 14C and dissolved O2 data, have estimated an oxygen consumption rate to be 0.13 ␮My⫺1 for Pacific deep waters. If this rate is used here, the differences in the dissolved O2 concentrations correspond to 380 to 440 yr. If the oxygen consumption in the Sulu Sea is enhanced due to high temperature, the mean residence time of the deep water needs to be reduced. Considering these, the Sulu Sea deep waters may be renewed in roughly 300 ⫾ 150 yr during which time significant amounts of REE(III)s and Si are regenerated into the deep water. This is much longer than the flushing times of other Southeast Asian basins (⬍100 yr) as argued by Broecker et al. (1987). From the above, a “back of the envelope” type calculation is now possible for the balance of REEs in the Sulu Sea. If we take Nd as an example of the REE(III)s and 300 yr as a mean residence time, then assuming an excess concentration of ⬃5 pmol/kg for the bottom water with an average thickness of ⬃1000 m, we can estimate that the benthic flux of Nd is ⬃1.7 pmol/cm2/yr. This is somewhat lower than the estimate of 12 ⫻ 106 mol/yr (or 3.2 pmol/cm2/yr) for the Nd flux from the world’s rivers by Goldstein and Jacobsen (1988), which was not corrected for estuarine removal. However, the value can be compared with the global average Nd fluxes of 4.6 ⫻ 106 mol/yr and 4.4 ⫻ 106 mol/yr (both, 1.2 pmol/cm2/yr) from rivers and the atmosphere, respectively, as estimated by Greaves et al. (1994). Thus, a large portion of the REE(III)s that reach the seafloor in association with settling particles must return to the water column and become involved in the vertical biogeochemical cycle, much like Si. This is consistent with the recent estimate of mean oceanic residence times of the REE(III)s between 800 yr for Pr and 5000 yr for Y (Alibo and Nozaki, 1999). 4.5. Implication to Global Problems The land-locked Sulu Sea basin can be treated as a natural laboratory in which we can study vertical cycling, and it has clearly been demonstrated from our data that a unique REE pattern can be evolved in the deep sea via dissolution of particles. In the open oceans, such local dissolved REE signatures also can be developed in certain areas and recorded in water masses. This provide a basis on how dissolved REEs and their pattern (or isotopic signature) are influenced by geochemical processes and ocean circulation and may be used as tracers of water mass in the open oceans; an aspect that has been discussed and disputed by many workers (e.g., Piepgras and Jacobsen, 1992; Zhang and Nozaki, 1996). For example, Zhang and Nozaki (1996) have shown that the heavy REE ratios such as Ho/Dy and Ho/Er in the South Pacific Intermediate Water are very different from those of the North Pacific, suggesting lateral transport of the REE signature from a local source, somewhere around the Southern Ocean, to the north. Such information cannot be obtained only from nutrients like dissolved SiO2 or other hydrographic properties such as temperature and salinity. The conventional technique to precisely determine dissolved REEs is now available using ICP mass spectrometry, and therefore, it will be possible to further argu-

Rare earth elements in the Sulu Sea

ments and resolve some issues in the field of chemical oceanography. 5. CONCLUSIONS

Based on the hydrographic observation and measurements of REE profiles in the Sulu Sea described above, we conclude the following: Because of the shallow sill depths, the deep and bottom waters of the Sulu Sea are ventilated at a slower rate, (⬃300 yr as opposed to ⬍100 yr) than other Southeast Asian basins. Therefore, changes in the seawater constituents by geochemical processes can be clearly seen within the basin. REE(III)s are released to the bottom water due to particle dissolution, most likely at the sediment-water interface rather than within the water column, and then redistributed by physical mixing. Thus, the REE(III)s and their elemental ratios are useful as tracers of water masses. REEs in the Sulu Sea, when normalized to those of NPDW, show a characteristic pattern enriched in the middle REEs with a maximum at Eu, suggesting that they are largely derived from local Southeast Asian sources. Although a negative Ce anomaly can be seen throughout the water column, dissolved Ce in the deep Sulu Sea is enriched relative to NPDW. The development of negative Ce anomaly is probably due to preferential addition of La and Pr over Ce during dissolution of particles in the deep water. Judging from the ␪-S diagram, the renewal of the deep and bottom waters occurs in a complex manner. The time-series observation of chemical tracers should help us understand the dynamic mechanism of flushing of the Sulu Sea. Also, theoretical and numerical modeling works should also be elaborated. Acknowledgments—We thank two anonymous reviewers for providing useful comments on the manuscript. We are also grateful to Captain Y. Jinno, officers and crew of R. V. Hakuho-Maru for their collaboration in the sampling, and the scientific party of the Piscis Austrinus Expedition for stimulation and cooperation. We are especially thankful to the Philippine Government for permission of investigation in the Sulu Sea. This work is supported partially by the Ministry of Education, Sports, Science and Culture, Japan through Grant-in-Aid No. 7404051 to the University of Tokyo (Y. Nozaki, principal investigator). REFERENCES Alibo D. S. and Nozaki Y. (1999) Rare earth elements in seawater: Particle association, shale-normalization and Ce oxidation. Geochim. Cosmochim. Acta 63, 263–272. Alibo D. S. and Nozaki Y. (submitted) Dissolved rare earth elements in the South China Sea: Geochemical characterization of the water masses. J. Geophys. Res. Amakawa H., Alibo D. S., and Nozaki Y. (submitted) Nd isotopic composition and REE pattern in the surface waters of the eastern Indian Ocean and its adjacent seas. Geochim. Cosmochim. Acta.


Apel J. R., Holbrook J. R., Liu A. K., and Tsai J. J. (1985) The Sulu Sea internal soliton experiment. J. Phys. Oceanogr. 15, 1625–1651. Broecker W. S., Patzert W. C., Toggweiler J. R., and Stuiver M. (1986) Hydrography, chemistry, and radioisotopes in the Southeast Asian Basins. J. Geophys. Res. 91, 14345–14354. Elderfield H. (1988) The ocean chemistry of the rare earth elements. Phil. Trans. Roy. Soc. Lond. A 325, 105–126. Elderfield H., Upstill–Goddard R., and Sholkovitz E. R. (1990) The rare earth elements in rivers, estuaries, and coastal seas and their significance to the composition of ocean waters. Geochim. Cosmochim. Acta 54, 971–991. Gamo T. (ed.) (1997) Preliminary report of the Hakkuho-Maru Cruise KH-96 –5 (the Piscis Austrinus Expedition). The Ocean Research Institute, University of Tokyo. Goldstein S. J. and Jacobsen S. B. (1988) Rare earth elements in river waters. Earth Planet. Sci. Lett. 89, 35– 47. Greaves M. J., Statham P. J., and Elderfield H. (1994) Rare earth element mobilization from marine atmospheric dust into seawater. Mar. Chem. 46, 255–260. Lerche D. and Nozaki Y. (1998) Rare earth elements of sinking particulate matter in the Japan Trench. Earth Planet. Sci. Lett. 159, 71– 86. Nitani H. (1972) Beginning of the Kuroshio. In Kuroshio—Its Physical Aspects (eds. H. Stommel and K. Yoshida), pp. 129 –164. University of Tokyo Press. Nozaki Y., Zhang J., and Amakawa H. (1997) The fractionation between Y and Ho in the marine environment. Earth Planet. Sci. Lett. 148, 329 –340. Nozaki Y., Yamada M., Nakanishi T., Nagaya Y., Nakamura K., Shitashima K., and Tsubota H. (1998) The distribution of radionuclides and some trace metals in the water columns of the Japan and Bonin trenches. Oceanologica Acta 21, 469 – 484. Nozaki Y., Larche D., Alibo D. S., and Snidvongs A. (submitted) The estuarine geochemistry of rare earths and indium in Chao Phraya River, Thailand and three Japanese rivers. Geochim. Cosmochim. Acta. Piepgras D. J. and Jacobsen S. B. (1992) The behavior of rare earth elements in seawater: Precise determination of variations in the North Pacific water column. Geochim. Cosmochim. Acta 56, 1851– 1862. Sholkovitz E. R., Landing W. M., and Lewis B. L. (1994) Ocean particle chemistry: The fractionation of rare earth elements between suspended particles and seawater. Geochim. Cosmochim. Acta 58, 1567–1580. Sholkovitz E. R., Elderfield H., Szymczak R., and Casey K. (in press) Island weathering: River sources of rare earth elements to the western Pacific Ocean. Mar. Chem. Taylor S. R. and McLennan S. M. (1985) The Continental Crust: Its Composition and Evolution. An Examination of the Geochemical Record Preserved in Sedimentary Rocks. Blackwell. Wyrtki K. (1961) Scientific Results of Marine Investigations of the South China Sea and the Gulf of Thailand 1959 –1961, Vol. 2. NAGA Report, University of California, Scripps Institute Oceanography. Zhang J. and Nozaki Y. (1996) Rare earth elements and yttrium in seawater: ICP-MS determinations in the East Caroline, Coral Sea, and South Fiji basins of the western Pacific. Geochim. Cosmochim. Acta 60, 4631– 4644.