Trace elements in ships' ballast water as tracers of mid-ocean exchange

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

w w w. e l s e v i e r. c o m / l o c a t e / s c i t o t e n v

Trace elements in ships' ballast water as tracers of mid-ocean exchange Kathleen R. Murphy a,b,⁎, M. Paul Field c , T. David Waite b , Gregory M. Ruiz a a

Smithsonian Environmental Research Center, PO Box 28, Edgewater MD 21037, USA The University of New South Wales, School of Civil and Environmental Engineering, NSW 2052, Australia c Institute of Marine and Coastal Sciences, Rutgers University, New Brunswick, NJ 08901, USA b

AR TIC LE I N FO

ABS TR ACT

Article history:

Recent regulation mandates that ships conduct mid-ocean ballast water exchange (BWE)

Received 28 March 2007

prior to discharging foreign ballast in U.S. territorial waters. We investigated the utility of

Received in revised form

dissolved concentration measurements for 6 elements (Ba, P, Mn, U, V and Mo) in the ballast

23 November 2007

tanks of ships operating in the North Pacific and Atlantic oceans as tracers of mid-ocean

Accepted 1 December 2007

BWE. Relatively conservative elements Mo, U and V provided little additional information

Available online 30 January 2008

beyond that obtained from salinity, whereas nonconservative Ba, P and Mn offered greater resolution. The utility of Ba, P and Mn was further examined in the context of three criteria:

Keywords:

(1) stability, or whether tracers maintain stable concentrations in ballast tanks over time; (2)

Trace elements

fidelity, or the degree to which tracer concentrations in ballast tanks faithfully reflect

Tracers

concentrations at their ocean source; and (3) predictability, or the degree to which ballast

Ballast water exchange

tanks have a predictable and restricted range of tracer concentrations following BWE. We

Seawater

found that in water held in ballast tanks over time, average stability increased for Mn b P b Ba,

Biological invasions

as reflected by decreasing coefficients of variation (30% N 21% N 3%) and fidelity increased in the same direction. While Ba and P usually increased discrimination at high salinities, Mn was typically the most sensitive indicator of BWE and the presence of residual port water in partially exchanged tanks. Ba, P and Mn in tanks exchanged in the Atlantic exhibited different concentration ranges compared to tanks exchanged in the Pacific, suggesting that if trace elements are to be used to verify BWE, criteria for discriminating between exchanged and unexchanged ballast tanks may need to be basin-specific. © 2007 Elsevier B.V. All rights reserved.

1.

Introduction

Ships transporting ballast water between geographically isolated ports contribute to the spread of many aquatic species beyond their natural range limits. Such biological invasions threaten the biodiversity of coastal environments, resulting in the homogenization of biota and sometimes causing severe economic, social or ecological impacts through the introduction of harmful species or ecosystem-modifiers (Carlton, 1985; Ruiz et al., 1997; Ruiz and Carlton, 2003; Wonham and Carlton,

2005). To reduce the inter-coastal transfer of invasive species, ships can implement mid-ocean ballast water exchange (BWE). During BWE, ships replace coastal ballast water, usually loaded during cargo operations in port and containing potential invasive species, with oceanic ballast water. This treatment replaces coastal organisms with oceanic species that are poorly adapted for survival in coastal environments, thereby reducing the likelihood of establishment (Verling et al., 2005). BWE is being promoted at national (USCG, 2004) and international (IMO, 2004) levels and is expected to remain

⁎ Corresponding author. Smithsonian Environmental Research Center, PO Box 28, Edgewater MD 21037, USA. Tel.: +61 2 9385 5778; fax: +61 2 9385 6139. E-mail address: [email protected] (K.R. Murphy). 0048-9697/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2007.12.011

12

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

the dominant ballast treatment method until better technological solutions are widely available (Hunt et al., 2005). Since 2004, the United States Coast Guard (USCG) has required BWE for all ships prior to discharging foreign ballast water in U.S. territorial waters (USCG, 2004), yet the requirement cannot be adequately enforced due to the lack of reliable methods for determining compliance. One approach to verifying BWE is to compare concentrations of naturallyoccurring chemical tracers in ballast tanks with their known distributions in the open ocean. Salinity is a primary tracer because salinities in the open oceans are high (typically 32–37) whereas salinities in coastal environments are often lowered by rainfall and river runoff. Although low salinity is an unambiguous indicator of coastal sources, many ports have high salinities overlapping with ranges in the open oceans for all or part of an annual cycle (Murphy, 2007). Allowing for the (legally tolerated) presence of up to 5% residual freshwater in a ballast tank exchanged in the open ocean (IMO, 2004; USCG, 2004), a salinity of ~30 appears to be an appropriate lower threshold for considering ballast water to be of uncertain BWE status. Thus, secondary tracers are required to distinguish coastal from ocean sources when salinities in ballast tanks exceed such a threshold value. A preliminary study of predominantly high-salinity ballast water indicated that chromophoric dissolved organic matter (CDOM), radium isotopes and certain trace elements (including Ba, P and Mn) aided discrimination between coastal and oceanic sources (Murphy et al., 2004a). A more extensive study of CDOM in ballast water from multiple sources in Asia, North America and Europe found that a single fluorescence intensity threshold separated most high-salinity coastal and oceanic ballast water (Murphy et al., 2006). The utility of sensitive, short lived isotopes of radium (223Ra and 224Ra) in this application is constrained by logistical challenges surrounding the collection and processing of samples (Murphy et al., 2004a). A rigorous evaluation of trace elements as ballast water indicators is yet to be performed. Initial results suggested that Ba, Mn and P were more sensitive tracers of ballast water source than salinity, but it was unclear whether the results were robust or could be extended into new ocean regions; furthermore, questions were raised regarding the reliability and stability of trace element measurements in ballast tanks (Murphy et al., 2004a). The distribution of trace elements in seawater is geographically variable with respect to open ocean and coastal environments. U, V and Mo occur in rivers and coastal regions typically at lower concentrations than in the open ocean, where higher concentrations are maintained by long oceanic residence times (5 × 104–7.6 × 106 years) relative to the mixing time of the oceans (~103 years) (Collier, 1984; Palmer and Edmond, 1993; Sohrin et al., 1998). Because Mo, U and V have low particle reactivity and nearly always vary conservatively with salinity, they offer limited sensitivity when salinity differentials are small, and concentrations in high-salinity coastal environments are similar to ranges in the open oceans (U: ~3.3 µg/L; V: ~1.6 µg/L; Mo: ~9–11 µg/L) (Sohrin et al., 1987; Delanghe et al., 2002, this study). In contrast, Ba, P and Mn tend to occur near the coast at higher concentrations than in the surface open ocean due to the predominance of terrestrial sources over oceanic sinks, as

reflected by their short residence times of 10,000, 69,000 and 60 years, respectively (Chan et al., 1977; Martin and Knauer, 1980; Broecker and Peng, 1982). A combination of terrestrial and anthropogenic sources, water column recycling, sediment diagenesis, aeolian inputs, sorption to particles, biological uptake, co-precipitation, biological decay and redox cycling contribute to potentially large oceanic gradients relative to conservative elements (Dehairs et al., 1980; Broecker and Peng, 1982; Sunda et al., 1983; Sunda and Huntsman, 1990; BenitezNelson, 2000; Karl and Björkman, 2002). As a first approximation, typical coastal:oceanic concentration ratios are on the order of N2:1 for dissolved Ba (Hanor and Chan, 1977; Shaw et al., 1998), N10:1 for total dissolved P (Froelich et al., 1982; Baturin, 2003), and N20:1 for dissolved Mn (Bruland and Franks, 1983; Shiller, 1997; Wells et al., 2000) with greater ratios typically correlating with greater salinity differentials. Elevated levels of Ba, P and Mn in ships' ballast tanks are therefore potential indicators of coastal (unexchanged) ballast water. Despite the potential complexity of reactions that could affect trace element speciation and behavior in ballast tanks, previous studies have documented occurrences of conservative behavior of dissolved Ba (Coffey et al., 1997; Taylor et al., 2003), Mn (Moore et al., 1979; Muller et al., 1994; Hatje, 2003) and P (Shammon and Hartnoll, 2002) in natural environments, including in estuaries that experience large physico-chemical gradients. Furthermore, deviations from conservative mixing for these elements are often attributed to sediment interactions (Hanor and Chan, 1977; Laslett and Balls, 1995), particularly in low-oxygen or low-salinity conditions (e.g. Coffey et al., 1997). Ballast water typically experiences relatively stable salinity, oxygen and pH conditions (Gollasch et al., 2000; Wonham et al., 2001; Drake et al., 2002, Ruiz unpublished data), thus Ba, P and Mn could plausibly act as stable and sensitive tracers of midocean ballast water exchange. In this study we examine whether Ba, P and Mn are sensitive and reliable tracers of BWE, or conversely, whether chemical or biological processes in ballast tanks preclude their widespread use as tracers of ballast water origin. The utility of Ba, P and Mn is examined with respect to three criteria: (1) stability, or whether tracers maintain stable concentrations in ballast tanks over time; (2) fidelity, or the degree to which tracer concentrations in ballast tanks faithfully reflect concentrations at their ocean source; and (3) predictability, or the degree to which BWE results in a predictable and restricted range of tracer concentrations in ballast tanks.

2.

Experimental methods

2.1.

Sampling

Ballast water samples were collected during eight cruises in 2001–2004 on commercial ships operating in the north Pacific and Atlantic oceans (Fig. 1 and Table 1). Tanks were ballasted initially at the port of departure or at a previous port of call and exchanged in the open ocean. Implementation of BWE required between 2 and 15 h, depending upon tank volume (~260–11,500 m3) and method of exchange (flow-through (FT) takes approximately three times longer than empty-refill (ER)),

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

13

Fig. 1 – Sources of ballast water (loading and exchange locations) in the Pacific Ocean (upper plot) and Atlantic Ocean (lower plot). Symbols signify cruises: LF ( ), AS ( ), K1 ( ), K2 ( ), SF ( ), LA ( ), BN ( ), Fos ( ).

as the ship tracked a linear distance of ~30–480 nautical miles. A total of 42 ballast tanks were sampled that contained ballast water from various geographic locations. Approximately half of the ballast tanks were inspected while empty prior to the beginning of experiments and none were seen to contain appreciable quantities of residual sediment. Nontoxic Rhodamine WT dye (Bright-Dyes) was added to ballast tanks at the beginning of two cruises (SF, LA) to enable BWE efficacy to be independently quantified. Overall, 239 ballast water samples and 150 seawater (shipside) samples were collected. For ballast tanks, sampling depths were 1–3 m with additional samples from 10–12 m in deep tanks (cruises SF, LA and BN). Seawater (shipside) samples were collected via the ship's engine cooling system, which constantly circulates ambient seawater from a depth of approximately 4–7 m through steel pipes at flow rates of ~200 m3/h. Sampling was via a short length of tubing attached to the pipe as near as possible to the seawater intake and upstream of the engine machinery. Equipment potentially contacting samples was acid washed in 1 N HCl before each cruise and flushed with N10 water volumes prior to collecting samples. Ballast water samples were collected via acid-cleaned plastic pumps (Wilden: Pro-Flo P.025) and tubing (Cole-Parmer: Chemfluor 367). Equipment and methodologies are detailed in earlier publications (Murphy et al., 2003; Murphy et al., 2004a). Salinity, temperature and oxygen measurements were obtained using either (a) a dissolved oxygen and conductivity meter (YSI-85) at discrete depths (cruises LA, SF, K1, K2 and BN); or (b) a CTD (Hydrolab MS4) to obtain profiles through the accessible tank water column (cruises Fos, LF and AS). Instruments were precalibrated with distilled water and NIST-traceable seawater conductivity standard (YSI 3169, 50,000 µScm− 1). A failure in the CTD motherboard 5 days into cruise LF resulted in falling salinity readings; these data were back corrected with an estimated accuracy of ±1 salinity unit.

Filtration on cruises SF, LA, Fos was via individual 0.22 µm syringe filters and on cruises LF, AS, K1, K2, BN via 0.45 µm high-capacity inline polypropylene capsule filters (Osmonics Inc., Memtrex™). Different inline filters, replaced several times during each voyage, were used for ballast versus seawater samples and refrigerated when not in use. The cutoff for ‘dissolved’ concentrations is operationally defined and the larger filter size used in this study (0.45 µm) would not have excluded colloidal material (Nakatsuka et al., 2007). Potential sampling artifacts were investigated on cruise LF, during which ballast water samples were collected by both pump + inline filter (used for multiple samples) and syringe sampler + syringe filter (used once only). No significant differences between replicate samples collected by the two methods were found despite small sample variances (Murphy et al., 2004a), indicating that both procedures were equally reliable and that the elements measured were present mostly in dissolved form or as species of size b0.22 µm. Filtered samples were frozen after collection and shipped to the laboratory at the end of the cruise where they were acidified to pH b2 by addition of 2 mL L− 1 12 M HNO3 (Optima grade, Fisher Scientific) upon thawing. This protocol of sample preservation has been tested and proved reliable when at sea acidification was logistically difficult or impractical.

2.2.

Trace element analysis

The concentrations of Ba, Mn, Mo, P, U and V were analyzed on an Element-1 high resolution inductively coupled plasma mass spectrometer (ThermoFinnigan, Bremen, Germany) at Rutgers Inorganic Analytical Laboratory, Institute of Marine and Coastal Sciences, Rutgers the State University of New Jersey. Samples were diluted 10 fold with 10% V/V ultra-pure HNO3 then analyzed in low and medium resolution using published techniques (Field et al., 1999). Replicate analysis of NASS-5 (North Atlantic Surface Seawater, certified reference

14

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Table 1 – Experimental design indicating numbers of (i) samples in the assessment of stability; (ii) pair-wise comparisons in the assessment of fidelity, and (iii) different ballast water masses (source/tank combinations) in the assessment of predictability Coastal ballast water source

Cruise dates

Ballast tanks

# BWE loc.

BWE types

# and type

Samp. depth (m)

Tank vol. (×102 m3)

1,12 1,11 1–3 1–3 1–3 1–3 1–3 1 1

113.8–114.9 67.5 18.8–48.1 19.1 2.8–4.5 2.8–4.5 2.8–4.5 2.6–2.9 2.6–2.9

4 3 3 1 2 2 6 2 1

Salinity (Avg.) before/after BWE

(i)

(ii)

(iii)

■▼▲ ■▲ ■▲ ■▲ ■▼ ■▼ ■▼ ■▲ ■▲

21.1/32.0 33.3/32.8 33.2/33.0 33.4/32.6 27.2/32.8 28.2/32.9 26.9/32.8 32.0/35.4 36.3/36.0

– – 30 8 10 5 14 – –

– – 6 2 1 1 1 3 3

7 4 12 4 3 3 11 6 7

North Pacific Ocean SF San Francisco (USA) LA Los Angeles (USA) LF Haramachi (Japan) Haramachi/unknown AS Yokkaichi (Japan) Inchon (Korea) Kunsan (Korea) K1 Oakland (USA) K2 Honolulu (USA)

Jun 19–26, 2004 Jul 1–6, 2004

3w 2w 8w 2w 2w 2w 6w 3d 3d

North Atlantic Ocean Fos Fos Sur Mer (Mediterr. Sea)

Jun 13–25, 2001

8w

1–3

6.2–18.0

7

■▼◪

37.6/37

62

16

24

BN

Sep 10–23, 2004

4w 2w

1,12 1

87.6–116.7

2 1

■▲

32.8/36.5

29

12

12

Rotterdam (Netherlands)

Nov 5–13, 2000 Dec 9–12, 2000 Jun 5–16, 2003 Aug 5–17, 2003

Ballast tank types were wing (w) or double-bottom (d). BWE types indicate that tanks sampled were exchanged by empty-refill (ER) method (▼), exchanged by flow-through (FT) method (▲), partially (1/3) FT-exchanged ( ), partially (2/3) FT-exchanged (◪) or unexchanged (■).

material) indicate good precision and accuracy (10%) for all elements of interest (Table 2). Note that our Mn determinations in NASS samples exceeded certified reference values by ~ 9% across all runs. Sporadic analytical errors affected P and V determinations in 2003 (CV N35%); samples that appeared anomalous (N = 13 from cruises LF and AS) were rerun in 2004. A number of samples from cruises K and BN also appeared to be anomalous and 64 were rerun. Data for anomalous samples were replaced upon reanalysis if significantly different or averaged where not different. Thereafter, a small number of individual data points that differed by N2 standard deviations from the means of the remaining (true) replicate samples were excluded only where there were N ≥ 3 replicates from the same time and place; this affected b1% of the dataset for any element. Erroneous data were difficult to diagnose in tanks with small sample N; in

these situations we took the approach of retaining apparent outliers due to the lack of sufficient evidence to justify removing them.

3.

Results

3.1.

Tracer sensitivity

Fig. 2 illustrates the relative sensitivity of trace elements as ballast water indicators for each ballast tank during this study, calculated from the ratio of (percent) change in mean tracer concentration to the (percent) change in mean salinity due to BWE. Where tracers exhibited essentially stable or increasing concentrations over time (i.e. Mn on cruises LF and Fos, together with all other tracers on all other cruises; Section 3.2),

Table 2 – Summary of trace element determination in NASS samples and detection limits (DL) during the analytical period Element

Resolution

Reference Value

2σ

YR 2003 Type

(µgL− 1) 138

Ba Mn 98 Mo 31 P 238 U 51 V 55

Low Medium Low Medium Low Medium

5.1 0.919 9.6 17.7 2.6 1.2

0.057 1.0

p v c p i i

YR 2004

Mean

CV

DL

Mean

CV

DL

(µgL− 1)

(%)

(µgL− 1)

(µgL− 1)

(%)

(µgL− 1)

4.98 0.97 9.64 21.84 2.74 1.63

1.77 8.71 6.38 36.08 8.3 39.65

0.184 0.18 0.32 4.2 0.065 0.06

5.18 1.03 9.46 17.94 2.76 1.16

2.52 8.29 3.11 3.48 8.81 3.78

0.14 0.08 0.04 2.8 0.037 0.014

Reported NASS reference data (determined value and standard deviation, σ) are either certified (c), informative (i) or proposed (p) by Field et al. (2007).

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

mean initial tracer concentrations were calculated from all samples collected prior to BWE. Where concentrations decreased over time (i.e. Mn on cruises BN and AS), mean initial tracer concentrations were calculated from samples collected immediately prior to BWE, thus testing the “worstcase” scenario for tracer sensitivity. Final tracer concentrations represent means of all samples collected from the same tank following BWE. Sensitivity ratios determined according to this method are approximate due to measurement error, which can substantially influence ratios particularly when the effect of BWE on salinity or tracer concentrations is small. The method is most useful for illustrating consistent differences in the relative performance of the six tracers. Physico-chemical conditions in the ballast tanks are summarized in Table 3. In general, the conservative elements Mo, U and V exhibited comparable sensitivity to salinity, as is indicated by a sensitivity ratio close to 1 (Fig. 2). Of the three, V usually exhibited the greatest relative change following BWE. On cruise LF, all trace elements including Mo, U and V performed better than salinity. This result is attributed to the CTD malfunction (Section 2.1) which reduced the precision and accuracy of salinity measurements on cruise LF. If the relative change in salinity had in fact been similar to that of the conservative elements, then the relative sensitivity of all tracers on this cruise should be reduced by a factor of ~10. In the majority of cases especially at higher salinities, Ba, P and Mn were more sensitive than salinity and the conservative elements. Ba exhibited typically 3–10 times greater relative change than salinity, except on cruise AS (salinity ~27 before BWE), where it performed similar to or worse than conservative tracers. P offered no discrimination additional to salinity

15

on cruises AS and SF, but was more sensitive than salinity on all cruises where initial salinity N30. Mn was most sensitive overall, exhibiting 10–30 times greater relative change than salinity, Mo and U on cruise Fos and 10–20 times greater relative change than Mo, U, V and Ba on cruise LF. On cruise AS, Mn exhibited more than 5 times greater relative change than salinity and the conservative elements, despite decreasing in concentration by 3–10 fold prior to BWE (Section 3.2). Since our goal is to study tracers that can offer additional information to salinity, the remainder of this paper will focus upon assessing the suitability of Ba, Mn and P for tracing ballast water sources.

3.2.

Stability

Tracer stability was examined for all unexchanged ballast tanks that were sampled on at least three occasions each separated by a day or longer, and for which replicate measurements existed for multiple sampling occasions. Overall, 18 ballast tanks from four cruises (Fos, LF, AS, BN) are included in the analysis of stability (Figs. 3 and 4). Error bars are for true (independent) replicate samples and thus indicate reproducibility for the entire sampling and analysis process. Ba concentrations were stable ± 0.25 µgL− 1 over the period of sampling, with the exception of cruise BN (Fig. 3A). Trends for Mn were variable (Fig. 3B); Mn concentrations decreased significantly over time on cruise AS (ANOVA by tank: F10,15 = 162.3, p b 0.0001) and cruise BN (ANOVA by tank: F6,28 = 13.8, p b 0.0001) but were quite stable on cruises Fos, and LF, varying by no more than ± 2.5 µgL− 1 in each tank. Phosphorus exhibited highly fluctuating concentrations in

Fig. 2 – Percent change in trace element concentrations relative to percent change in salinity following BWE of ballast tanks during this study.

16

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Table 3 – Average salinity (Salav), temperature (Tav) and dissolved oxygen (DOav) and minimum dissolved oxygen (DOmin) in ballast tanks during this study Cruise SF SF SF SF LA LA LA Fos Fos Fos Fos LF LF AS AS K1 K1 K2 K2 BN BN

BWE Tav Salav DOav DOmin Sampling treatment (°C) (mg/L) (mg/L) depths (m) ■ ▼ ◪ ■ ◪ ■ ▼ ◪ ■ ▼ ■ ▲ ■ ▼ ■ ▼ ■ ▲

14.6 12.9 12.5 12.5 14.1 14.1 12.8 20.7 24.4 21.3 21.6 12.0 9.2 25.6 14.8 19.3 21.9 25.1 19.7 20.4 25.3

22.2 32.6 31.1 31.1 33.6 33.8 33.8 37.6 36.6 36.9 36.7 33.3 32.9 27.2 32.8 31.9 35.4 36.3 36.0 32.8 36.2

8.5 8.0 8.4 8.4 7.6 8.9 7.9 6.8 6.4 6.8 6.7 7.5 8.0 5.5 8.6 5.6 5.9 5.5 6.2 6.6 6.8

7.5 7.5 7.8 7.8 6.8 8.6 7.0 6.4 6.1 6.7 6.7 7.1 7.2 3.5 7.5 5.0 5.8 5.3 6.1 5.9 6.3

1,12,20 1,12,20 1,12,20 1,12,20 1,15,20 1,15,20 1,15,20 0–3.5 0–3.5 0–3.5 0–3.5 0–3.5 0–3.5 0–3.6 0–3.6 1 1 1 1 1 1

BWE treatments indicate that tanks sampled were exchanged by empty-refill (ER) method (▼), exchanged by flow-through (FT) method (▲), partially (1/3) FT-exchanged ( ), partially (2/3) FTexchanged (◪) or unexchanged (■).

ballast tanks held over time (Fig. 4); on cruise AS, P concentrations in individual tanks differed by as much as 20 µgL− 1 on consecutive days. On cruise Fos, there was a general trend of increasing P from ~6 to ~11 µgL− 1, and on cruise LF, P concentrations fluctuated by ± 1 µgL− 1 in some tanks and ± 6 µgL− 1 in others. The relative standard deviation (i.e. coefficient of variation) of elemental concentrations in true replicate samples gives an indication of the relative importance of sampling/analytical error, with higher values of Rcv indicating lower overall measurement precision. In Table 4, Rcv is defined as the CV for replicate samples (N = 2–4) collected in the same tank at the same time, averaged across individual cruise datasets. Rcv for the four cruises in increasing order was AS ~ LF ~ Fos ≪ BN. The low Rcv for cruise BN reflects intermittent problems with an analytical run in 2004, affecting ~20% of ballast water samples from this cruise. For cruises except Cruise BN, Rcv in increasing order was Ba ~ Mo ~ U b V ≪ Mn b P. On Cruise BN, inter-replicate variability was up to 3× higher than on other cruises, with Rcv in increasing order Mo b U b Ba ~ V ≪ Mn ~ P. Total measurement variability in individual ballast tanks over a cruise summarizes information on element stability + precision. Tcv in Table 4 is defined as the CV for all samples (N = 2–16) collected in an individual tank, averaged across tanks for individual cruise datasets. Low values of Tcv indicate that an elements overall stability in ballast tanks on a cruise was high, and that the sampling and analysis process resulted in precise estimates of concentration. Tcv is increased by changes in concentrations in ballast tanks over time, sample contam-

ination, poor analytical reproducibility, or any combination of these factors. Indicating decreasing stability, Tcv for the four cruises increased for LF ~ Fos b AS b BN. On cruises except cruise BN, Tcv for the elements increased for Ba ~ Mo ~ U b V ≪ P ~ Mn. Values of Tcv on cruise BN were 3–10× higher than on the other three cruises, increasing for Mo b U b V b Ba ≪ P ≪ Mn.

3.3.

Fidelity

Tracer fidelity was assessed for all tanks in which BWE coincided geographically with the collection of one or more shipside samples, by comparing the mean concentrations of tracers in ballast tanks (N = 2–24, residence time b80 h) with shipside sample means (N = 1–8). Tracers with high fidelity should exhibit close to a perfect 1:1 correlation between the two measurements (Fig. 5). Ba exhibited greatest fidelity overall, with most measurements from fully exchanged ballast tanks (i.e. triangular symbols) lying close to the line of perfect fit (Fig. 5A). Means for partially exchanged ballast tanks on Cruise Fos lie to the left of the 1:1 correspondence line, indicating higher concentrations in the ballast tanks than in the ocean; but are located progressively closer to the line after successive exchanges. The greatest deviations were exhibited by three tanks during initial ballasting on Cruise BN for which Ba concentrations in the tanks were lower than measurements in the port. Since very few samples from the port were available for comparison (N = 3), low sampling/analytical precision and/or inaccurate characterization of the water entering these ballast tanks could account for

Fig. 3 – Stability of dissolved Ba and Mn in unexchanged ballast tanks sampled repeatedly over time (mean ± SE). ), BN ( ), LF ( ) Symbols signify cruise: AS ( ). and Fos (

17

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Concentrations of P in ballast tanks corresponded well to concentrations in the ocean during BWE, however, the greater number of points lying slightly to the left of the 1:1 correspondence line suggests a bias toward higher concentrations in ballast tanks than in the ocean where BWE took place (Fig. 5C). With few exceptions (initial ballasting on cruise BN and ER exchange on cruise K1) deviations from the 1:1 line were in the same direction as for Ba, but greater in magnitude, reflecting typically greater differences in end-member concentrations. Considering only samples from fully exchanged ballast tanks after removing the six data points associated with the initial ballasting of tanks on cruise BN, the equation of the straight line for the remaining data points has a slope of 0.97 ± 0.05 and x-intercept of 3.59 ± 1.01 µgL− 1 (R2 = 0.90), confirming the bias toward slightly elevated concentrations in the ballast tanks relative to the locations where ballast water was drawn.

3.4.

Fig. 4 – Stability of dissolved P in unexchanged ballast tanks sampled repeatedly over time (mean ± SE). A) cruises AS ) and BN ( ); B) cruise LF; and C) cruise Fos. In ( B) and C), symbols indicate individual ballast tanks.

this deviation; furthermore, no significant deviations are associated with the mid-ocean exchange of the same tanks (blue triangles). Considering only samples from fully exchanged ballast tanks after removing the six data points associated with the initial ballasting of tanks on cruise BN, the equation of the straight line for the remaining data points (mean± standard error) has a slope of 0.88 ± 0.06 and x-intercept of 1.24 ± 0.4 µgL− 1 (R2 = 0.85). This indicates that overall, concentrations of Ba in fully exchanged ballast tanks were slightly greater than concentrations in the ocean where the ballasting occurred, particularly when oceanic concentrations were lowest. Manganese fidelity was comparatively poor (Fig. 5B), with a high degree of scatter in measurements around the 1:1 correspondence line. For most tanks, the deviations were in the same direction as those seen for P and Ba (e.g. tanks on cruise Fos during initial ballasting, tanks on Cruise BN during FT exchange), whereas for others, deviations in the opposite direction were observed (e.g. ER exchanged tanks on cruises K1 and K2). Considering all samples from fully exchanged ballast tanks again after removing the six data points associated with the initial ballasting of tanks on cruise BN, the equation of the line of best fit (R2 = 0.49) has a slope of 1.01± 0.2 and x-intercept of 0.23 ± 0.33 µgL− 1. This indicates generally poor correspondence between the two measurements, although with no particular bias toward higher concentrations in either ballast tanks or the ocean.

Predictability

Tracer predictability was assessed by examining the effect of BWE on the range of tracer concentrations measured in ballast tanks. The multivariate distribution of Ba, P and Mn in exchanged and unexchanged ballast tanks during this study is illustrated in Fig. 6. Overall, smaller concentration ranges were measured in exchanged ballast tanks (triangles) than in unexchanged tanks (diamonds). Also in exchanged ballast tanks, a smaller range of concentrations was observed in cruises that took place in the Atlantic (light blue symbols) compared to the Pacific Ocean (black and yellow symbols). With few exceptions, Ba concentrations in unexchanged tanks containing coastal ballast water exceeded 9 µgL− 1 (Fig. 6). Exceptions were the tanks ballasted in Hawaii on cruise K2 (~5.9 µgL− 1) and two of ten tanks ballasted in Japan on cruise LF (~5.5 µgL− 1). In tanks exchanged in the Pacific and Atlantic open oceans, mean Ba was ~4.6–8 and ~ 5.3–7 µgL− 1, respectively. Partially exchanged tanks contained Ba at levels intermediate to coastal and oceanic concentrations. Whereas it appears possible to discriminate exchanged from unexchanged ballast tanks on both Atlantic cruises on the basis of Ba alone (i.e. Ba b~ 7 ppb indicates fully exchanged ballast tanks), Ba concentrations in nine tanks containing ballast

− × 100%) for Table 4 – Mean coefficients of variation (CV: σ/ x trace elements in ballast tanks sampled over time AS

Ba Mn Mo P U V Nav

LF

BN

Fos

TCV

RCV

TCV

RCV

TCV

RCV

TCV

RCV

2.4 48.7 4.1 11.4 3.5 6.0 4.8

2.1 3.6 3.5 7.0 3.0 5.2 2.0

3.3 14.7 4.2 26.3 5.7 19.0 6.8

2.9 6.1 3.2 17.3 3.2 6.9 2.2

19.7 115.9 12.0 31.9 14.8 16.8 9.3

13.8 29.1 8.9 25.8 10.9 13.8 3.6

3.4 20.0 2.6 28.3 4.6 9.2 15.5

2.7 16.9 2.3 23.3 4.0 7.8 4.0

TCV is the mean CV for all samples from each tank, averaged across tanks on each cruise. RCV is the mean CV for replicate samples collected at the same location and time. The average number of samples in each case is given by Nav.

18

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

In exchanged ballast tanks, Mn concentrations were typically ~0.2–2 µgL− 1 (Atlantic cruises) and b0.1–1 µgL− 1 (Pacific cruises). Fig. 6A shows the joint distribution of Ba and P in exchanged and unexchanged ballast tanks. Ba and P are correlated in tanks exchanged in the Pacific (r = 0.68, N = 85, p b 0.001) but not the Atlantic. When tanks are classified separately according to ocean basin, exchanged and unexchanged tanks occupy predominately separate positions in multivariate space. In Fig. 6A, tanks exchanged in the Pacific Ocean fall within the red polygon and tanks exchanged in the Atlantic Ocean within the blue ellipse, except for four ballast tanks containing low Ba and P coastal water from Honolulu and Haramachi (Japan). Fig. 6B shows the multivariate distribution of Ba and Mn in ballast water. Ba and Mn were weakly correlated in tanks exchanged in the Atlantic (r = 0.37, N = 42, p b 0.05), but not the

Fig. 5 – Correspondence between dissolved Ba, Mn and P concentrations in ballast tanks and in the ambient ocean during BWE (µgL− 1, mean ± SE). Treatments are initial ballasting ( ), exchanged by FT method (▲); exchanged by ER method (▼); partially (1/3) exchanged ( ); and partially (2/ 3) exchanged (◪). Colors indicate cruise (LF [pink], AS [purple], K1 [red], K2 [orange], BN [blue], Fos [green]).

●

water from Pacific ports overlapped with ranges in tanks that were exchanged in the open ocean. P concentrations in unexchanged ballast tanks varied widely from ~4–10 µgL− 1 in tanks ballasted at the port of Fos Sur Mer on the Mediterranean Sea to N100 µgL− 1 in tanks ballasted in San Francisco Bay. In tanks exchanged in the Atlantic Ocean, P concentrations occupied a tight range between 3 and 6 µgL− 1 (cruises BN and Fos), whereas in tanks exchanged in the Pacific Ocean P varied widely (~6–60 µgL− 1) and entirely overlapped with ranges in coastal ballast water from both oceans. Mn concentrations in unexchanged ballast tanks varied between ~1 and 70 µgL− 1 (Atlantic ports) or between ~2 and 80 µgL− 1 (Pacific ports).

Fig. 6 – Concentrations (µgL− 1, mean ± SE) of dissolved Ba, Mn and P in ballast tanks during nine research cruises. Treatments are unexchanged ( R. ), exchanged by FT method (Δ); exchanged by ER method (∇); partially (1/3rd) exchanged ( ); and partially (2/3rd) exchanged (◪). Colors indicate locations where tanks were ballasted in the Pacific Ocean (Hawaii [pink], Korea [purple], Japan [orange], USA West Coast [red], open N. Pacific [yellow], Gulf of Alaska [black]) or Atlantic Ocean (Rotterdam [navy], Mid Atlantic [light blue], Mediterranean Sea [green]). Polygons enclose means in ballast tanks exchanged at least 200 nautical miles from shore in the Pacific (red) or Atlantic (blue) Oceans.

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Pacific. In Fig. 6B, tanks exchanged in the Pacific Ocean fall within the red polygon and tanks exchanged in the Atlantic Ocean within the blue ellipse, except for five cases of elevated (N1–3 µgL− 1) Mn; these were in two tanks on cruise AS (~1.5 µgL− 1), one tank on cruise LA (~4.3 µgL− 1) two tanks on cruise SF (~6 µgL− 1).

4.

Discussion

4.1.

Tracer utility

Many metals are sensitive tracers of terrestrial and anthropogenic influences (for example, many transition elements including iron and heavy metals), however, most are unsuited to measurement in ballast tanks due to the unacceptably high risk of contamination from ship structural materials. Dissolved concentrations of Ba, P and Mn were examined as potential tracers of BWE in this study because (1) it was considered likely that accurate measurements could be made in ballast tanks due to relatively low contamination risk; (2) concentrations of these elements (together with Mo, U and V) can be determined simultaneously and rapidly at low (oceanic) levels without pre-concentration using ICP-MS (Field et al., 1999, 2007); (3) the elements were considered likely to behave conservatively in ballast tanks; and (4) the multivariate distribution of these elements in coastal waters is likely to differ from their distribution in the open oceans. Dissolved (b0.45 µm) concentrations were considered the most appropriate measure because particulate concentrations in the open oceans are low (Dehairs et al., 1980; Jeandel et al., 1987; Sunda and Huntsman, 1990; Yoshimura et al., 2007) and total concentrations in ballast tanks could vary strongly as a function of sediment settling and resuspension. Cycling of trace elements between particulate and dissolved phases affects dissolved concentrations, but does not preclude the use of dissolved measurements as long as the scale of changes due to cycling is small relative to the difference in concentrations between oceanic and coastal ballast waters.

4.1.1.

Mo, U and V

Mo and U exhibit conservative concentrations throughout their oceanic profiles (Palmer and Edmond, 1993; Sohrin et al., 1998). In this study, Mo and U offered comparable resolution for tracing ballast water sources to salinity measured in-situ. Vanadium occasionally offered additional resolution, possibly due to its slight surface depletion in the open oceans (Collier, 1984). Although salinity can be measured more easily and precisely than Mo, U and V, particularly under controlled conditions in the laboratory, in the absence of salinity data all three elements could assist in discriminating coastal from oceanic ballast water.

4.1.2.

Ba

Ba was a stable and faithful tracer of the mid-ocean exchange of ballast waters in this study, as indicated by stable Ba concentrations in 14 (of 17) ballast tanks containing port water held over time (Tcv below 3.5%, Table 4), together with an overall strong linear correlation between the concentrations of Ba in ballast tanks and concentrations in the ocean where the

19

ballast water was sourced. Much lower stability was observed on cruise BN (Tcv ~20%) however, this correlated with high variability in Mo, U and V (Tcv N10%), suggesting an external explanation for this variability, such as sampling/analytical error or water-mass mixing in ballast tanks during cruise BN. The much higher Rcv values for all three elements on cruise BN (Rcv ~14%) relative to other cruises (Rcv b3%) indicates lower measurement precision and suggests that the sampling/ analytical explanation is most likely. Overall, these results suggest that Ba cycling in ballast tanks is unimportant on the timescales of this study. Ba exhibited close to 1:1 correspondence between concentrations measured in ballast tanks and shipside samples during ballasting, despite often low sample replication. We attribute the anomalously poor correspondence between Ba (and other tracers) in the port of Rotterdam compared to ballast tanks at the beginning of cruise BN to the collection of too few port samples to accurately characterize the water masses that entered the ballast tanks, exacerbated by the fact that tanks were particularly large (~104 m3) and loaded over several days. Our assessment of tracer fidelity for Ba and other tracers would have benefited from more frequent shipside sampling during ballasting — particularly in ports and in the North Pacific ocean where tracer concentrations were most spatially and temporally variable. Barium was usually a more sensitive tracer of BWE than salinity and the conservative elements particularly at higher salinities, but almost always a less sensitive tracer than Mn and P (Fig. 2). Ba concentrations in shipside samples collected N100 miles offshore in this study averaged 4.4–6.6 µgL− 1 in the North Pacific and 5.8–6.7 µgL− 1 in the North Atlantic, corresponding well with previous measurements in the NE Pacific, NE and NW Atlantic and western Mediterranean Sea (Bernat et al., 1972; Chan et al., 1976, 1977; Nozaki et al., 2001) (Table 5). In unexchanged ballast tanks with salinity N30.0, average Ba concentrations were ~5.6–10 µgL− 1 in tanks ballasted in the North Pacific and 9–14 µgL− 1 tanks ballasted in the North Atlantic. These small differences between measured ranges of Ba in ballast tanks sourced from high-salinity coasts versus the open ocean illustrate the limited sensitivity of this tracer. While improved data quality in future studies should enhance the resolution of Ba measurements in ballast water, Ba may be of greatest utility when used in combination with other ballast water tracers.

4.1.3.

P

Phosphorus concentrations were relatively dynamic in ballast tanks held over time, with Tcv ranging from 11–32% (Table 4). These high values of Tcv resulted from fluctuating P in some tanks together with lower measurement precision for P compared to most other elements (Rcv ranging from 7–26%). In oligotrophic oceanic waters near Hawaii and in the Atlantic, low P concentrations greatly increased measurement uncertainty. Even so, P remained a much more sensitive tracer of BWE than salinity or the conservative elements at higher salinities (Fig. 2). While some P variability was attributable to measurements, the fluctuating P concentrations observed in several ballast tanks is suggestive of nutrient recycling. Dissolved P in seawater is present primarily as the orthophosphate anion

20

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Table 5 – Concentration (mean ± σ) of total dissolved barium, manganese and phosphorus (µgL− 1) in shipside samples during this study [TS: number of samples], compared to previous observations in these regions (references are indicated in parentheses in units of µgL− 1) Ba North Pacific Ocean Northwestern (40–60°N, 130–180°E)

6.6 ± 1.6 [TS: 10]

Mn

1.1 ± 1.8 [TS: 10]

37 ± 21 [TS: 10] 30–60 (Garcia et al., 2006) ‡ 9–13 (Fang, 2004) 7–11 (Suzumura and Ingall, 2004)

3.0 ± 0.9 [TS: 18] 0.04–0.12 (Martin et al., 1985) 0.2 ± 0.1 [TS: 6] 0.03–0.07 (Martin et al., 1989)

6.3 ± 2.6 [TS: 18] 6–8 (Wu et al., 2000)

(China Sea, South/ 5.5 (Nozaki East) (Off Japan) et al., 2001)

Northeastern (Offshore California–Hawaii: 15–44°N, 125–180°W) (Gulf of Alaska: 44–60°N, 125–165°W)

North Atlantic Ocean Northwestern (25–40°N, 45–77°W)

Sargasso Sea

Northeastern (30–51°N, 5.5–45°W)

(English Channel, near-shore)

Mediterranean Sea (Western, near-shore)

4.4 ± 0.8 [TS: 18] 4.8 (Chan et al., 1976) 6.8 ± 0.6 [TS: 6]

P, DOP† or phosphate‡

32 ± 8 [TS: 6] 15–30 (Martin et al., 1989)‡ 31–40 (Ridal, 1992)

6.7 ± 0.45 0.5 ± 0.3 [TS: 8] [TS: 11] 6 (Chan ~0.05–0.16 et al., 1977, (Shiller, 1997) stn 29) 0.12 (Bruland and Franks, 1983)

3.0 ± 1.4 [TS: 11] 3.5–7.5 (Sohm and Capone, 2006) 2.3 ± 1.3 (Wu et al., 2000)

5.8 ± 1.0 [TS: 27] 5.3 (Chan et al., 1977, stn 115)

0.12 ± 0.08 [TS: 25] ~0.05–0.22 (Shiller, 1997)

7.6 ± 2.9 [TS: 27]

0.36 ± 0.17 [TS: 27] 0.13–1.3 (Statham et al., 1999) 1.2 ± 1.3 [TS: 19]

3.2 ± 0.7 [TS: 24] 2–4 (Aminot and Kerouel, 2004) † 1–9 (Garcia et al., 2006)‡ 7.7 ± 1.6 [TS: 27] 12–22 (Butler et al., 1979)

8 ± 1.1 [TS: 21] 10 (Bernat et al., 1972)

2.8 ± 1.8 [TS: 19] 2.2 (Aminot and Kerouel, 2004)† 2–6 (Garcia et al., 2006)‡

Samples were from the open ocean N 200 nautical miles offshore except in the English Channel and Mediterranean Sea.

HPO2− and dissolved organic phosphorus (DOP) (Karl and 4 Björkman, 2002). Both the organic and inorganic components of the total dissolved P pool are available to phytoplankton and other microorganisms for growth (Cotner and Biddanda, 2002; Bjorkman and Karl, 2003) and subsequently recycled back to phosphate by digestion, decay and hydrolysis. P uptake by autotrophs is dependent upon solar radiation but heterotrophic bacterial uptake apparently is not (Roberts and Howarth, 2006); the dark ballast tank environment might thus be expected to favor heterotrophic microbial uptake of P. Laboratory investigations of the microbial degradation of organic detritus (Cooper, 1935; Ogura, 1975; Newell et al., 1981; Suzumura and Ingall, 2004) indicate that a relatively labile proportion can be mineralized within days, suggesting that P fluctuations in ballast tanks probably reflected in part the relative rates of P sequestration versus liberation by microbes. Total dissolved P in exchanged ballast tanks ranged widely on Pacific cruises (6–52 µgL− 1) compared to the Atlantic cruises (3–7 µgL− 1), reflecting the much greater surface ocean variability of P along the North Pacific cruise tracks compared to the North Atlantic cruise tracks. Our oceanic P measurements in shipside samples (Table 5) correspond well with published data from the NE and NW Atlantic oceans (Butler et al., 1979; Wu et al., 2000; Sohm and Capone, 2006) and NE Pacific (Ridal, 1992; Wu et al., 2000), and in the NW Pacific and Mediterranean Sea overlap with phosphate ranges according to the World Ocean Atlas (Garcia et al., 2006). P concentrations in ballast tanks were strongly correlated with concentrations in shipside samples during ballasting, however, overall, P concentrations were typically slightly greater in ballast tanks than in the open ocean where BWE took place. This is partly a reflection of differences in endmember concentrations coupled with BWE efficiencies near or below theoretical (95%) levels and partly a reflection of the fact that fewer samples were collected from the ocean during BWE than from any ballast tank. Fig. 5 probably underestimates source variability, especially for tanks exchanged in the highly variable subpolar North Pacific (cruises LF and AS). For most ballast tanks exhibiting deviations from 1:1 correlation, BWE efficiency was unknown, making it impossible to draw definitive conclusions about the origin of additional (~1–10 µgL− 1) P. However, there is some suggestion that ballast tanks may themselves be a minor source of P, possibly via the resuspension of sediments or organic detritus during BWE. Particle resuspension in ballast tanks during ballasting or in inclement weather could affect the speciation of particle-reactive elements like P and Mn (VanCappellen and Wang, 1996; Pan et al., 2002).

4.1.4.

Mn

Mn exhibits high potential for tracing terrestrial water masses due to large coastal versus oceanic concentration differences (Bruland and Franks, 1983; Shiller, 1997), and was almost always the most sensitive tracer of BWE in this study (Fig. 2). Despite this, the performance of dissolved Mn as a ballast water tracer was reduced in this study by several factors. These included (1) several instances where dissolved Mn concentrations in ballast tank decreased significantly over time (low stability); (2) poor correspondence between Mn

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

concentrations in ballast tanks compared to concentrations measured in shipside samples during exchange (low fidelity); and (3) higher sampling and/or analytical uncertainty compared to U, V, Mo and Ba, which impacted the performance of Mn according to all criteria assessed in this study. An essential nutrient for phytoplankton growth, Mn is accumulated in phytoplankton cells under low-light conditions (Bruland et al., 1991), released during biological decay (Collier and Edmond, 1984) and oxidized by bacteria in the dark (Sunda et al., 1983). In six ballast tanks on cruise AS, dissolved Mn decreased by 5–10% day− 1, which is comparable to particulate formation rates measured by Sunda and Huntsman (1988) in dark-stored deep ocean samples of ~ 4.8% day− 1 and exceeds typical rates of particulate formation in lightexposed samples of ~1–2% day− 1 (Sunda and Huntsman, 1988, 1990; Moffett, 1997). In natural environments, Mn-oxidizing bacteria that are suppressed by light catalyze the slow chemical oxidation of Mn(II) species to insoluble Mn oxides at night time, whereas by day photo-mediated reactions dissolve the oxides and regenerate Mn(II) (Hem, 1963; Stumm and Morgan, 1981; Sunda and Huntsman, 1988, 1990). In ballast tanks, the absence of light may favor the bacteriallymediated transformation of dissolved Mn into insoluble oxides which settle from the water column or are excluded by 0.45 µm filters. A number of factors could have contributed to the observed inter-cruise (and inter-tank) variability in dissolved Mn stability during this study. Mn-oxidizing activity of temperate coastal waters can vary strongly by season, an effect sometimes attributed to seasonal changes in temperature and primary productivity (Sunda and Huntsman, 1987; Moffett, 1997). In this study, stability did not correlate with water temperature, however, the two cruises that exhibited decreasing dissolved Mn over time also had ~ 3× greater concentrations of fluorescent dissolved organic matter than cruises with stable Mn (Murphy et al., 2006), suggesting a possible link with primary productivity. Furthermore, significant increases in organic matter fluorescence have been observed in ballast tanks held over time (Murphy et al., 2004b), with fluorescence characteristics consistent with recent, microbially-produced humic material (Murphy et al., 2008). Plankton composition and survival in ballast tanks is known to be highly variable (e.g. Verling et al., 2005), however, the only published timeseries of bacterial populations in ballast tanks (Drake et al., 2002) did not include community composition, and it remains for future research to determine whether ballast tank environments favor the survival of particular microorganism species. Another possible source of between-cruise variability in Mn behavior is suggested by the observation that rates of particulate formation in dark-stored samples are depressed by pre-exposure to sunlight (Sunda and Huntsman, 1988). This effect may be due to the light-mediated production of hydrogen peroxide, an agent recognized to be capable of inducing the reductive dissolution of Mn(IV) oxides (Szymczak and Waite, 1989). If photo-generated reactive oxygen species such as hydrogen peroxide and superoxide are present in ballast tanks, transformations might be expected to be ongoing until redox-active agents such as these are consumed (Waite et al., 1988; Rose and Waite, 2006).

21

In ballast tanks where dissolved Mn decreases over time, its resolution as a BWE tracer is a decreasing function of voyage duration. It may be possible in future studies to circumvent this problem by measuring total (dissolved + particulate) Mn which, due to much higher particulate Mn in coastal versus oceanic environments, would also greatly magnify differences between coastal and oceanic concentrations relative to dissolved measurements (Bruland and Franks, 1983; Moffett, 1997; Wells et al., 2000). A trade-off to this approach is likely to be the much greater dependence of measurements on the quantity and behavior of entrained sediments, which can build up to significant levels in ballast tanks (Bailey et al., 2003; Duggan et al., 2005). Sediments retained in exchanged ballast tanks could greatly elevate total suspended Mn and P beyond oceanic levels and increase the variability of whole-tank concentration estimates as a result of resuspension, scavenging, and settling (VanCappellen and Wang, 1996; Pan et al., 2002). Decreasing concentrations in ballast tanks held over time reduces sensitivity but does not eliminate dissolved Mn as a ballast water tracer as long as the resulting changes in concentration are small relative to the difference between oceanic and coastal ballast waters. This was the case for the ballast tanks examined in this study except on cruise BN. On other cruises, Mn concentrations were always significantly lower following BWE, even where Mn decreased by 50–90%. Thus Mn concentrations were typically below 1 µgL− 1 in tanks exchanged in the Pacific ocean and below 3 µgL− 1 in tanks exchanged in the Atlantic. There were only five cases where Mn remained significantly elevated above these levels following BWE, and all are potentially explained by differences in end-member concentrations together with b100% BWE efficiency: (1) Four tanks on cruises AS and SF with high pre-BWE Mn concentrations (30–60 µgL− 1) and post-exchange concentrations of 1.5–6 µgL− 1, since 95% BWE efficiency accounts for Mn concentrations elevated by ~1.5–3 µgL− 1 above oceanic levels, and (2) One tank on cruise LA, where low BWE efficiency (75% according to independent determination; see below) accounts for Mn concentrations of ~ 3.3 µgL− 1 following BWE of tanks with initial concentrations of 12–16 µgL− 1. On cruise LA where Mn decreased by ~ 75%, BWE resulted in a salinity and Mo reduction of b10%, and reductions in Ba and P of ~ 30% and ~ 40% respectively, demonstrating the much greater sensitivity of Mn to coastal influences. The fidelity results for Mn may therefore reflect relatively large coastal versus oceanic concentration differentials and variable BWE efficiency together with lower measurement precision for this element. Dissolved Mn in open ocean shipside samples in this study ranged from 0.2–3 µgL− 1 in the North Pacific and 0.1–0.5 µgL− 1 in the North Atlantic (Table 5). Mn in shipside samples from the NE Atlantic and English Channel are comparable with previous reports (Bruland and Franks, 1983; Shiller, 1997; Statham et al., 1998); however Mn in NE Pacific samples are elevated 2–10 times relative to previous measurements in the same region (Martin et al., 1985; Martin et al., 1989). Some of our oceanic Mn data may therefore be overestimated; especially since Mn concentrations in oceanic shipside samples often approached detection limits of ~0.2 µgL− 1. Recent modifications to analytical protocols have reduced Mn detection limits in ballast water to below 0.01 µgL− 1 (Field et al.,

22

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Fig. 7 – Mean annual phosphate distribution in the surface global oceans (µgL− 1). Data are from the World Ocean Atlas (Garcia et al., 2006) mapped in Ocean Data View (Schlitzer, 2007).

2007). A larger database of surface ocean Mn distribution is now required particularly for the North Pacific, in order to better define the expected range Mn concentrations in exchanged ballast tanks and reduce the influence of outliers.

4.2.

Implementation

The optimal implementation of tracer techniques for verifying mid-ocean ballast water exchange requires at minimum a twostage procedure. In the first stage, the ballast water should exceed a salinity criterion, since low salinity in ballast tanks is an unambiguous indicator of coastal sources. Only after exceeding the salinity criterion would confirmation be needed that additional tracers lie within the expected ranges for exchanged ballast tanks. From an enforcement standpoint, a high-salinity/ocean-referenced approach is critical for three reasons. First, salinity screening eliminates the unnecessary implementation of more costly and time-consuming techniques. Second, verification should ultimately be possible in the absence of either information regarding, or comparative samples from, the original coastal source port. Instead, tracer concentrations should be compared with ranges for the surface open oceans, where variability is far less than at the coasts and where the position of BWE is known from declarations on the ship's compulsory ballast water reporting form (USCG, 2004). Third, in delineating the boundaries for acceptable multivariate tracer concentrations in exchanged ballast tanks, the opportunity exists to incorporate statistical uncertainties in existing knowledge together with management tolerances toward false determinations of different types. While ICP-MS enables sensitive and accurate multi-element measurements in the laboratory, it is ultimately desirable to assess the BWE status of ships in real-time in order for it to be possible to intercept high-risk ballast water before it is discharged. Colorimetric and chemiluminescence methods are widely used in field measurements of phosphate and Mn (e.g. Chiswell et al., 1990; Okamura et al., 1998; Yaqoob et al., 2004), but are non-trivial and unsuitable for non-scientists. Automated in-

situ instruments increase the practicality of field measurements, while reducing the opportunity for sample contamination. Prototype in-situ instruments have been developed for quantifying Mn and P using spectroscopy (Klinkhammer, 1994; Adornato et al., 2007) and voltammetry (Tercier-Waeber et al., 1998), but many technical and practical hurdles would need to be overcome before their deployment would be feasible in a regulatory setting.

4.3.

Improved predictability

The discrimination of ballast water sources under this framework will rely upon knowledge of the spatial and temporal distribution of target tracers in the open oceans. Developing global scale tracers for oceanic environments requires extensive investigation and sampling directed at existing gaps in spatial and temporal coverage. There is a need for rapid, accurate and precise analytical techniques with appropriate QA/QC management. Overall measurement precision for elements examined in this study was generally high (Rcv b10%), with the exception of a subset of samples primarily from Cruise BN (Rcv 9–30%, Table 4). Recent improvements to our methods for ICP-MS measurement of BWE tracers have resulted in three fold improvements in detection limits, accuracy and precision (Field et al., 2007). Even so, tracer concentrations in shipside samples from this study generally compared well to previous studies that used orthodox sampling methods (Table 5) suggesting that it is possible to collect uncontaminated samples of these elements in the oceans via the water-cooling pipes of commercial ships. Several existing databases that map the global distribution of P are freely-available. For example, the 2005 World Ocean Atlas and World Ocean Database available via NOAA-NODC include mean dissolved inorganic phosphorus (phosphate) data for the global ocean with 1° grid resolution (Garcia et al., 2006), and the Global Open Ocean DOP database (N N 139,000 measurements) records dissolved organic phosphorus (Karl and Björkman, 2002). Shiller (1997) mapped surface Mn distributions in the North Atlantic. We are aware of no extensive databases of

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

surface distributions of Mn in the Pacific or Ba in either ocean south of the Arctic circle. Many ships that travel to the US west coast from Asia follow a Great Circle route transecting the subpolar (western subarctic and Alaskan) gyres (Miller et al., 2004), where surface tracer distributions are influenced by Ekman upwelling of nutrient-rich deep water (Tomczak and Godfrey, 2003; Garcia et al., 2006). The high-P oceanic samples from this study also contained higher Ba relative to the oligotrophic North Pacific south of 40°N. It is clear from the global distribution of phosphate (Fig. 7) that P and possibly other BWE tracers will be more variable in tanks exchanged in the North Pacific than in the North Atlantic, suggesting that different levels of complexity would be associated with BWE verification in either ocean.

4.4.

Exchange efficiencies

A significant complication to verifying BWE by natural tracer methods is that in practice, lower than theoretical exchange efficiencies are frequently demonstrated by vessels that have implemented ballast water exchange to theoretical 95% effectiveness, particularly when ships use the flow-through exchange method (Murphy et al., 2004a; Ruiz et al., 2005; Ruiz and Smith, 2005). This is apparently due to a combination of factors including engineering constraints (Hall and Wilson, 2006; Wilson et al., 2006), complicated physics, and human error, since appropriate pumping times are difficult to calculate if pump capacities are not accurately known (T. Snell, pers. comm.). Regulators may thus be forced to allow for b95% exchange efficiency until technological advances enable better flushing of ballast tanks, or until guidelines on the best-practice implementation of BWE under current technology are available. If regulators have to accommodate lower exchange efficiencies and hence a greater contribution of residual port water to the overall chemical signature of ballast tanks, the overall sensitivity of the method for distinguishing between high-salinity coastal and oceanic ballast waters will be reduced. Currently, the most reliable way to determine BWE efficiency for a given ballast tank is through experimentation and direct empirical measures. Deploying artificial tracers to track the movement of water in tanks allows control over detection sensitivity relative to natural tracers. In the earliest cruises of this study, fluorescent dye was added to tanks prior to BWE allowing independent calculation of BWE efficiency (75% efficient by FT exchange on cruise LA, 93% efficient by FT exchange and 98% efficient by ER exchange on cruise SF) (Murphy et al., 2004a). Artificial dye tracers were discontinued on subsequent cruises due to interference with another potential ballast water tracer (chromophoric dissolved organic matter; Murphy et al., 2006) and concerns about the possible contamination of low-level trace element measurements. Fluorescent 1 µm plastic beads have been deployed in other ballast water studies (Ruiz unpublished, N = 13 cruises), but have tended to stick to surfaces and settle from the water column over time.

5.

Conclusion

Results from sampling the ballast tanks of ocean-going cargo ships indicate that Ba, Mn and P are sensitive and useful

23

tracers of high-salinity coastal ballast water sources. Criteria for effective ballast water tracers include relatively stable concentrations over time, concentrations which accurately reflect the entrained source waters and a predictable range of concentrations in exchanged ballast tanks. In this study, dissolved Ba and P performed well according to these criteria and were more sensitive tracers of source than salinity and the conservative elements Mo, U and V. Dissolved Mn exhibited lower stability and fidelity, however, Mn remained the most sensitive tracer of coastal influences except in a small number of tanks where initial concentrations were low and decreased by an order of magnitude during a 7-day cruise. We conclude that several trace elements aid the discrimination of coastal from oceanic ballast water sources and show promise for verification, for which applications may be most limited by the opportunity for real-time determinations. Our results also suggest that commercial vessels may offer an exceptional and unexploited opportunity for reliable observations (measures) of some trace elements, given the number and global reach of operating ships, trade routes, and frequency of transits. Efforts currently underway to gather data from additional ships, ports and oceanic regions across greater temporal and spatial scales will assist in characterizing the full range of signals expected in exchanged ballast tanks and will further test the utility of trace elements for discriminating ballast water sources.

Acknowledgement We thank SERC staff and volunteers for sampling assistance and the Rutgers Inorganic Analytical Laboratory for analytical services. L. Kalnejais and anonymous reviewers provided invaluable comments on earlier versions of this manuscript. NYK Bulkship (USA) LTD., Gateway Maritime Corp./Sincere Industrial Corp., Matson Navigation Company, Bergesen DY ASA., Sea River Maritime, the Alaska Tanker Company, BP Amoco PLC and Krupp Seeschiffahrt GmbH generously provided experimental platforms. This research was funded by the US Coast Guard's Research and Development Center and the Columbia River Aquatic Nuisance Species Initiative (CRANSI).

REFERENCES Adornato LR, Kaltenbacher EA, Greenhow DR, Byrne RH. High-resolution in situ analysis of nitrate and phosphate in the oligotrophic ocean. Environ Sci Technol 2007;41:4045–52. Aminot A, Kerouel R. Dissolved organic carbon, nitrogen and phosphorus in the N–E Atlantic and the N–W Mediterranean with particular reference to non-refractory fractions and degradation. Deep-Sea Res, Part I, Oceanogr Res Pap 2004;51:1975–99. Bailey SA, Duggan IC, van Overdijk CDA, Jenkins PT, MacIsaac HJ. Viability of invertebrate diapausing eggs collected from residual ballast sediment. Limnol Oceanogr 2003;48:1701–10. Baturin GN. Phosphorus cycle in the ocean. Lithol Miner Resour 2003;38:101–19. Benitez-Nelson CR. The biogeochemical cycling of phosphorus in marine systems. Earth-Sci Rev 2000;51:109–35.

24

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Bernat M, Church T, Allegre CJ. Barium and strontium concentrations in Pacific and Mediterranean sea water profiles by direct isotope dilution mass spectrometry. Earth Planet Sci Lett 1972;16:75–80. Bjorkman KM, Karl DM. Bioavailability of dissolved organic phosphorus in the euphotic zone at station ALOHA, North Pacific Subtropical Gyre. Limnol Oceanogr 2003;48:1049–57. Broecker WS, Peng TH. Tracers in the Sea. NY: Lamont-Doherty Geological Observatory; 1982. Bruland KW, Franks RP. Mn, Ni, Cu, Zn and Cd in the western North Atlantic. In: Wong CS, Boyle EA, Bruland KW, Burton JD, Goldberg ED, editors. Trace Metals in Seawater. New York: Plenum Press; 1983. p. 395–414. Bruland KW, Donat JR, Hutchins DA. Interactive influences of bioactive trace metals on biological production in oceanic waters. Limnol Oceanogr 1991;36:1555–77. Butler EI, Knox S, Liddicoat MI. Relationship between Inorganic and Organic Nutrients in Sea-Water. J Mar Biol Assoc UK 1979;59:239–50. Carlton JT. Transoceanic and interoceanic dispersal of coastal marine organisms: the biology of ballast water. Oceanogr Mar Biol Ann Rev 1985;23:313–71. Chan LH, Edmond JM, Stallard RF, Broecker WS, Chung YC, Weiss RF, et al. Radium and barium at GEOSECS stations in the Atlantic and Pacific. Earth Planet Sci Lett 1976;32:258–67. Chan LH, Drummond D, Edmond JM, Grant B. On the barium data from the Atlantic Geosecs expedition. Deep-Sea Res 1977;24:613–49. Chiswell B, Rauchle G, Pascoe M. Spectrophotometric methods for the determination of manganese. Talanta 1990;37:237–59. Coffey M, Dehairs F, Collette O, Luther G, Church T, Jickells T. The behaviour of dissolved barium in estuaries. Estuar Coast Shelf Sci 1997;45:113–21. Collier RW. Particulate and dissolved vanadium in the North Pacific Ocean. Nature 1984;309:441–4. Collier R, Edmond J. The trace-element geochemistry of marine biogenic particulate matter. Prog Oceanogr 1984;13:113–99. Cooper LHN. The rate of liberation of phosphate in sea water by the breakdown of plankton organisms. J Mar Biol Assoc UK 1935;20:197–202. Cotner JB, Biddanda BA. Small players, large role: microbial influence on biogeochemical processes in pelagic aquatic ecosystems. Ecosystems 2002;5:105–21. Dehairs F, Chesselet R, Jedwab J. Discrete suspended particles of barite and the barium cycle in the open ocean. Earth Planet Sci Lett 1980;49:528–50. Delanghe D, Bard E, Hamelin B. New TIMS constraints on the uranium-238 and uranium-234 in seawaters from the main ocean basins and the Mediterranean Sea. Mar Chem 2002;80:79–93. Drake LA, Ruiz GM, Galil BS, Mullady TL, Friedman DO, Dobbs FC. Microbial ecology of ballast water during a transoceanic voyage and the effects of open-ocean exchange. Mar Ecol, Prog Ser 2002;233:13–20. Duggan IC, van Overdijk CDA, Bailey SA, Jenkins PT, Limen H, MacIsaac HJ. Invertebrates associated with residual ballast water and sediments of cargo-carrying ships entering the Great Lakes. Can J Fish Aquat Sci 2005;62:2463–74. Fang TH. Phosphorus speciation and budget of the East China Sea. Cont Shelf Res 2004;24:1285–99. Field MP, Cullen JT, Sherrell RM. Direct determination of 10 trace metals in 50 mL samples of coastal seawater using desolvating micronebulization sector field ICP-MS. J Anal At Spectrom 1999;14:1425–31. Field MP, LaVigne M, Murphy KR, Ruiz GM, Sherrell RM. Direct determination of P, V, Mn, As, Mo, Ba and U in seawater by SF-ICP-MS. J Anal At Spectrom 2007;22:1145–51. Froelich PN, Bender ML, Luedtke NA, Heath GR, DeVries T. The marine phosphorus cycle. Am J Sci 1982;282:474–511.

Garcia HE, Locarnini RA, Boyer TP, Antonov JI. World Ocean Atlas 2005, Volume 4: Nutrients (phosphate, nitrate, silicate). In: Levitus S, editor. NOAA Atlas NESDIS 64. Washington, D.C.: U.S. Government Printing Office; 2006. p. 396. Gollasch S, Lenz J, Dammer M, Andres HG. Survival of tropical ballast water organisms during a cruise from the Indian Ocean to the North Sea. J Plankton Res 2000;22:923–37. Hanor JS, Chan LH. Non-conservative behavior of barium during mixing of Mississippi River and Gulf of Mexico waters. Earth Planet Sci Lett 1977;37:242–50. Hatje V. Particulate trace metal and major element distributions over consecutive tidal cycles in Port Jackson Estuary, Australia. Environ Geol 2003;44:231–9. Hem JD. Chemical equilibria and rates of manganese oxidation. U.S. Geol Surv Water-supply Pap 1963;1667A:1–64. Hunt CD, Tanis DC, Stevens TG, Frederick RM, Everett RA. Verifying Ballast-Water Treatment Performance. Environ Sci Technol 2005;39:321A–8A A-pages. IMO. (International Maritime Organisation). International Convention for the Control and Management of Ships Ballast Water and Sediments. BWM/CONF/36, 16 February 2004. London: International Maritime Organisation; 2004. Jeandel C, Caisso M, Minster JF. Vanadium behaviour in the global ocean and in the Mediterranean sea. Mar Chem 1987;21:51–74. Karl D, Björkman K. Dynamics of DOP. In: Hansell DA, Carlson CA, editors. Biogeochemistry of marine dissolved organic matter. Academic Press; 2002. p. 249–366. Klinkhammer GP. Fiber optic spectrometers for in-situ measurements in the oceans — the ZAPS probe. Mar Chem 1994;47:13–20. Laslett RE, Balls PW. The behaviour of dissolved Mn, Ni and Zn in the Forth an industrialised, partially mixed estuary. Mar Chem 1995;48:311–28. Martin JH, Knauer GA. Manganese cycling in Northeast Pacific waters. Earth Planet Sci Lett 1980;51:266–74. Martin JH, Knauer GA, Broenkow WW. VERTEX: the lateral transport of manganese in the northeast Pacific. Deep-Sea Res 1985;32:1405–27. Martin JH, Gordon RM, Fitzwater S, Broenkow WW. Vertex: phytoplankton/iron studies in the Gulf of Alaska. Deep-Sea Rese, Part A, Oceanogr Res Pap 1989;36:649–80. Miller AW, Ruiz GM, Lion K. Status and trends of ballast water management in the United States. Second biennial report of the national ballast information clearinghouse (January 2002 to December 2003). Submitted to the United States Coast Guard. Smithsonian Environmental Research Center, Edgewater, Maryland, 2004, pp. 1–32. Moffett JW. The importance of microbial Mn oxidation in the upper ocean: a comparison of the Sargasso Sea and equatorial Pacific. Deep-Sea Res, Part I, Oceanogr Res Pap 1997;44:1277–91. Moore RM, Burton JD, Williams PJL, Young ML. The behavior of dissolved organic material, iron and manganese during estuarine mixing. Geochim Cosmochim Acta 1979;43:919–26. Muller FLL, Tranter M, Balls PW. Distribution and transport of chemical constituents in the Clyde Estuary. Estuar Coast Shelf Sci 1994;39:105–26. Murphy KR. Naturally occurring chemical tracers in seawater and their application to verifying mid-ocean ballast water exchange. PhD Thesis, Sydney: School of Civil and Environmental Engineering. University of New South Wales; 2007. [188 pages]. Murphy K, Ruiz G, Sytsma M. Standardized sampling protocol for verifying mid-ocean ballast water exchange. US Coast Guard Research and Development Center, Groton, CT; 2003. Murphy K, Boehme J, Coble P, Cullen J, Field P, Moore W, et al. Verification of mid-ocean ballast water exchange using naturally occurring coastal tracers. Mar Pollut Bull 2004a;48:711–30. Murphy K, Ruiz G, Coble P, Boehme J, Field P, Cullen J, et al. Mid-ocean ballast water exchange: approach & methods for verification. US

SC IE N CE OF TH E TOTA L E N V I RO N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Coast Guard Research and Development Center Report No. CG-D-02-04, Groton, CT; 2004b. Murphy KR, Ruiz GM, Dunsmuir WTM, Waite TD. Optimized parameters for fluorescence-based verification of ballast water exchange by ships. Environ Sci Technol 2006;40:2357–62. Murphy KR, Stedmon CA, Waite TD, Ruiz GM. Distinguishing between terrestrial and autochthonous organic matter sources in marine environments using fluorescence spectroscopy. Mar Chem 2008;108:40–58. Nakatsuka S, Okamura K, Norisuye K, Sohrin Y. Simultaneous determination of suspended particulate trace metals (Co, Ni, Cu, Zn, Cd and Pb) in seawater with small volume filtration assisted by microwave digestion and flow injection inductively coupled plasma mass spectrometer. Anal Chim Acta 2007;594:52–60. Newell RC, Lucas MI, Linley EAS. Rate of degradation and efficiency of conversion of phytoplankton debris by marine microorganisms. Mar Ecol, Prog Ser 1981;6:123–36. Nozaki Y, Yamamoto Y, Manaka T, Amakawa H, Snidvongs A. Dissolved barium and radium isotopes in the Chao Phraya River estuarine mixing zone in Thailand. Cont Shelf Res 2001;21:1435–48. Ogura N. Further studies on decomposition of dissolved organic-matter in coastal seawater. Mar Biol 1975;31:101–11. Okamura K, Gamo T, Obata H, Nakayama E, Karatani H, Nozaki Y. Selective and sensitive determination of trace manganese in sea water by flow through technique using luminol hydrogen peroxide chemiluminescence detection. Anal Chim Acta 1998;377:125–31. Palmer MR, Edmond JM. Uranium in river water. Geochim Cosmochim Acta 1993;57:4947–55. Pan G, Krom MD, Herut B. Adsorption–desorption of phosphate on airborne dust and riverborne particulates in east Mediterranean seawater. Environ Sci Technol 2002;36:3519–24. Ridal JJ. Dissolved organic phosphorus concentrations in the northeast sub-arctic pacific-ocean. Limnol Oceanogr 1992;37:1067–75. Roberts BJ, Howarth RW. Nutrient and light availability regulate the relative contribution of autotrophs and heterotrophs to respiration in freshwater pelagic ecosystems. Limnol Oceanogr 2006;51:288–98. Rose AL, Waite TD. Role of superoxide in the photochemical reduction of iron in seawater. Geochim Cosmochim Acta 2006;70:3869–82. Ruiz GM, Carlton JT. Invasive species: vectors and management strategies. Washington, D.C.: Island Press; 2003. p. 600. Ruiz GM, Smith G. Biological study of container ships arriving to the Port of Oakland. Final report to the Port of Oakland, March 22 2005; 2005. Ruiz GM, Carlton JT, Grosholz ED, Hines AH. Global invasions of marine and estuarine habitats by non-indigenous species: mechanisms, extent, and consequences. Am Zool 1997;37:621–32. Ruiz GM, Murphy KR, Verling E, Smith G, Chaves S, Hines AH. Ballast water exchange: efficacy of treating ships' ballast water to reduce marine species transfer and invasion success. Final report to the US Fish & Wildlife Service, American Petroleum Institute and Prince William Sound Regional Citizens' Advisory Council. Smithsonian Environmental Research Center, Edgewater; 2005. p. 17. Schlitzer R. Ocean Data View. http://odv.awi.de2007. Shammon TM, Hartnoll G. The winter and summer partitioning of dissolved nitrogen and phosphorus. Observations across the Irish Sea during 1997 and 1998. Hydrobiologia 2002;475:173–84. Shaw TJ, Moore WS, Kloepfer J, Sochaski MA. The flux of barium to the coastal waters of the southeastern USA: the importance of submarine groundwater discharge. Geochim Cosmochim Acta 1998;62:3047–54.

25

Shiller AM. Manganese in surface waters of the Atlantic ocean. Geophys Res Lett 1997;24:1495–8. Sohm JA, Capone DG. Phosphorus dynamics of the tropical and subtropical North Atlantic: Trichodesmium spp. versus bulk plankton. Mar Ecol, Prog Ser 2006;317:21–8. Sohrin Y, Isshiki K, Kuwamoto T, Nakayama E. Tungsten in North Pacific waters. Mar Chem 1987;22:95–103. Sohrin Y, Fujishima Y, Ueda K, Akiyama S, Mori K, Hasegawa H, et al. Dissolved niobium and tantalum in the North Pacific. Geophys Res Lett 1998;25:999–1002. Statham PJ, Yeats PA, Landing WM. Manganese in the eastern Atlantic Ocean: processes influencing deep and surface water distributions. Mar Chem 1998;61:55–68. Statham P, Leclercq S, Hart V, Batté M, Auger Y, Wartel M, et al. Dissolved and particulate trace metal fluxes through the central English Channel, and the influence of coastal gyres. Cont Shelf Res 1999;19:2019–40. Stumm W, Morgan JJ. Aquatic Chemistry: An introduction emphasizing chemical equilibria in natural waters. 2nd ed. New York: Wiley-Interscience Publication; 1981. Sunda WG, Huntsman SA. Microbial oxidation of manganese in a North-Carolina estuary. Limnol Oceanogr 1987;32:552–64. Sunda WG, Huntsman SA. Effect of sunlight on redox cycles of manganese in the southwestern Sargasso Sea. Deep-Sea Res, Part A, Oceanogr Res Pap 1988;35:1297–317. Sunda WG, Huntsman SA. Diel cycles in microbial manganese oxidation and manganese redox speciation in coastal waters of the Bahama-islands. Limnol Oceanogr 1990;35:325–38. Sunda WG, Huntsman SA, Harvey GR. Photoreduction of manganese oxides in seawater and its geochemical and biological implications. Nature 1983;301:234–6. Suzumura M, Ingall ED. Distribution and dynamics of various forms of phosphorus in seawater: insights from field observations in the Pacific Ocean and a laboratory experiment. Deep-Sea Res, Part I, Oceanogr Res Pap 2004;51:1113–30. Szymczak R, Waite TD. Photochemical activity in waters of the Great Barrier Reef. Estuar, Coast Shelf Sci 1989;33:605–22. Taylor JR, Falkner KK, Schauer U, Meredith M. Quantitative considerations of dissolved barium as a tracer in the Arctic Ocean. J Geophys Res-Oceans 2003;108(C12):3374. Tercier-Waeber ML, Belmont-Hebert C, Buffle J. Real-time continuous Mn(II) monitoring in lakes using a novel voltammetric in situ profiling system. Environ Sci Technol 1998;32:1515–21. Tomczak M, Godfrey JS. Regional Oceanography: an Introduction. Delhi: Daya Publishing House; 2003. USCG. (United States Coast Guard). Mandatory Ballast Water Management Program for U.S. Waters: Final Rule, 33 CFR 151, Subpart D. 69. 144, 2004, pp. 44952–44961. VanCappellen P, Wang YF. Cycling of iron and manganese in surface sediments: a general theory for the coupled transport and reaction of carbon, oxygen, nitrogen, sulfur, iron, and manganese. Am J Sci 1996;296:197–243. Verling E, Ruiz GM, Smith LD, Galil B, Miller AW, Murphy KR. Supply-side invasion ecology: characterizing propagule pressure in coastal ecosystems. Proc Royal Soc B 2005;272:1249–56. Waite TD, Sawyer DS, Zafiriou OC. Reactive oceanic transients. Appl Geochem 1988;3:9–19. Wells ML, Smith GJ, Bruland KW. The distribution of colloidal and particulate bioactive metals in Narragansett Bay, RI. Mar Chem 2000;71:143–63. Wilson W, Chang P, Verosto S, Atsavapranee P, Reid DF, Jenkins PT. Computational and experimental analysis of ballast water exchange. Nav Eng 2006;118:25–36. Wonham MJ, Carlton JT. Trends in marine biological invasions at local and regional scales: the Northeast Pacific Ocean as a model system. Biological Invasions 2005;7:369–92.

26

SC IE N CE OF TH E TOTA L E N V IR O N ME N T 3 93 ( 20 0 8 ) 1 1–2 6

Wonham MJ, Walton WC, Ruiz GM, Frese AM, Galil BS. Going to the source: role of the invasion pathway in determining potential invaders. Mar Ecol, Prog Ser 2001;215:1–12. Wu JF, Sunda W, Boyle EA, Karl DM. Phosphate depletion in the western North Atlantic Ocean. Science 2000;289:759–62. Yaqoob M, Nabi A, Worsfold PJ. Determination of nanomolar concentrations of phosphate in freshwaters using flow injection

with luminol chemiluminescence detection. Anal Chim Acta 2004;510:213–8. Yoshimura T, Nishioka J, Saito H, Takeda S, Tsuda A, Wells ML. Distributions of particulate and dissolved organic and inorganic phosphorus in North Pacific surface waters. Mar Chem 2007;103:112–21.