MARINE ECOLOGY PROGRESS SERIESMar Ecol Prog Ser

Vol. 331: 35–47, 2007 Published February 16

INTRODUCTION

The introduction of non-indigenous aquatic nui-sance species (ANS) to coastal areas around the worldhas had profound negative impacts on aquatic ecosys-tems (Carlton et al. 1995, National Research Council1996, Ruiz et al. 2000,) and is considered to be one ofthe most important environmental issues facing themaritime community (USCG 2001). Non-indigenousspecies (NIS) introductions have been cited as the sec-ond greatest threat to biodiversity behind habitat loss(Vitousek et al. 1997) and are recognized by conserva-tion biologists (www.gao.gov/atext/d031089r.txt) asthe second most serious threat to endangered speciesafter habitat destruction. The introduction of ANS has

altered important ecological processes and causedserious economic damage worldwide.

A primary vector for the introduction of marine non-native species is through discharge of ballast waterfrom ships. Ballast water is used to increase the sta-bility, maneuverability, and safety of ocean-goingships. The ballast tanks of a vessel are typically filledwith water from the port of origin and emptied of thatwater at the destination port, potentially leading to thetransport of unwanted, non-native species from oneport to the other.

Regulatory actions have been implemented world-wide to prevent and control the introduction of NISvia ballast water discharge. Originally, many of theseactions were voluntary, including the US Coast

© Inter-Research 2007 · www.int-res.com*Email: huntc@battelle.org

Optical signatures of seawater and potential use forverification of mid-ocean ballast water exchange

Carlton D. Hunt1,*, Deborah Tanis1, Elizabeth Bruce1, Michael Taylor2

1Battelle, 397 Washington Street, Duxbury, Massachusetts 02332, USA2Cawthron Institute, Private Bag 2, Nelson, New Zealand

ABSTRACT: Mandatory requirements for exchange of ships’ ballast water in the open ocean requirean effective means of verifying that open ocean exchange has occurred. This study demonstrates thatUV fluorescence of chromophoric organic matter in seawater has great potential to verify open oceanballast water exchange. Ballast water samples, obtained for excitation-emission spectral analysis dur-ing 2 flow-through ballast water exchange events in the trans-Pacific voyage of a chemical carrier,establish a clear difference between the UV fluorescence characteristics of ballast water before andafter mid-ocean exchange. The exchange of water in mid-ocean shifted the fluorescence signaturesof ballast water from those typical of the coastal water at the donor ports to those of the exchangedmid-ocean water. Estimates based on the fluorescence data suggest that at least 90% of the coastalwater was exchanged. The shifts in fluorescence intensity and structure were concomitant with a>99% reduction in the concentration of tracer dye added to the ballast tanks during ballasting in theports of origin. In contrast, changes in phytoplankton and zooplankton species composition anddiversity were not as great, which limit their potential for verification of mid-ocean exchange. Theconsistently low fluorescence intensity and lack of fluorescence structure of mid-Pacific ocean waterrelative to the coastal waters, and the stability of the fluorescence in the ballast tanks before and afterexchange, suggest that fluorescence techniques can form the basis of verifying the presence of openocean water in ballast tanks.

KEY WORDS: Ballast water exchange · Aquatic nuisance species · Non-indigenous species ·Coloured dissolved organic matter · CDOM · UV fluorescence

Resale or republication not permitted without written consent of the publisher

Mar Ecol Prog Ser 331: 35–47, 2007

Guard’s (USCG) guidelines that all vessels entering USwaters after operating outside of the exclusive eco-nomic zone (EEZ) conduct ballast water management(BWM) including voluntary ballast water exchange.However, the US Secretary of Transportation reportedin 2002 that approximately 74% of all regulatedvessels failed to comply with voluntary BWM andconcluded that compliance with the voluntary guide-lines was insufficient to allow for an accurate assess-ment of the voluntary BWM guidelines (USCG 2001).

Adherence to the voluntary guidelines was not foundto be effective for vessels entering US ports so theUSCG published a rule in 2004 that makes BWMmandatory for all vessels that are equipped with bal-last water tanks and are entering US waters from out-side the EEZ (USCG 2004). The mandatory require-ments included ‘Exchange ballast water beyond theEEZ, in an area more than 200 nautical miles from anyshore’, ‘retention of ballast water on board’, or thatvessels ‘Use an ‘environmentally sound’ US CoastGuard-approved alternative ballast water manage-ment method before the vessel enters the US EEZ’(USCG 2004). Moreover, the recently adopted ‘Inter-national Convention for the Control and Manage-ment of Ships Ballast Water & Sediments’ (Interna-tional Maritime Organization 2004) calls for alternativeballast water management practices and requiresexchange at sea. Thus, until such time that alternativeballast water management methods are approved andbecome widespread, exchange of ballast water atsea will be the primary means of reducing the risk ofinvasive species introduction.

The concept behind ballast water exchange isreplacement of all or most of the original ballast waterfrom a port of origin with water from mid-ocean, inorder to minimize the risk of trans-oceanic transferof unwanted species (Hay & Tanis 1998). There are 2main methods of making mid-ocean exchanges: rebal-lasting and flow-through dilution. Reballasting in-volves completely emptying then refilling the ballasttank (for container vessels) or ballast hold (for bulkcarriers). The flow-through dilution method requirespumping ocean water into the ballast tank or ship’shold and allowing water to overflow. Rigby & Halle-graeff (1994) calculated that if a water volume equal to3 times the ballast tank volume is pumped through thetank, 95% of the original coastal water in a ballast tankshould be replaced.

Exchange operations which aim to exchange at least95% of the original ballast water also aim to kill orremove at least 95% of the organisms in that water.However, dilution of the original water is not the sameas the removal of the organisms as there are indica-tions of plankton retention in ballast tanks duringexchanges. For example, Rigby & Hallegraeff (1994)

found that while a completed exchange carried outunder static conditions (i.e. the vessel was anchored inSingapore Harbour) resulted in the retention of 5% ofthe original water, 25% of the original phytoplanktonmaterial remained. Similarly, Harvey et al. (1999)reported that mid-ocean exchange greatly reduces thediversity and abundance of planktonic organisms inthe ballast water of ships entering the Great Lakes.The presence of coastal species in incoming vesselsthat had reported complete exchanges suggests thatthe exchanges were incomplete and suggest the biotamay be retained in the tanks either by entrainment orsedimentation, or by avoiding the outlets, in the case oflarger species. Taylor & Bruce (1999) found moderateto high biological exchange in a series of ballast waterexchange studies on a vessel conducting exchange atsea in the Pacific Ocean, even when dye tracer studiesdemonstrated >99% exchange of coastal water withocean water. Even so, until more effective BWM meth-ods are available, exchange at sea is the best widelyavailable method of reducing the risk of invasivespecies introductions via ballast water.

Given this, managers who regulate ballast water arefaced with verification of ships’ compliance with mid-ocean exchange regulations. One verification methodpracticed widely is measurement of salinity. Howeversalinity suffers from lack of specificity and is variablewithin a small range across the world’s oceans, thus itsutility for verifying exchange is limited except for portsof origin with low salinity water (Taylor & Bruce 1999).Verification methods that either supplement or replacethe salinity measurements are required to accuratelydiscriminate open-ocean water from coastal water.

In anticipation of the regulatory needs for robust,specific verification methods that are easily deployedin the receiving ports, we explored several potentialverification approaches. One of the potential ap-proaches identified was the structure of dissolvedorganic matter (DOM), which occurs at different con-centrations and is also compositionally different be-tween the open ocean and coastal waters (Coble et al.2004). DOM in natural waters has been referred to asfulvic acid, gelbstoff, yellow substance, and humicsubstances (Kirk 1983). A major source of gelbstoff incoastal regions is river runoff, which includes terres-trial humic and fulvic acids (Coble & Brophy 1994).Other sources that contribute to DOM in the coastalenvironment include wastewater discharges, marshes,and wetlands. DOM concentration varies significantlydepending upon the original source of the matter andits components depend on the specific water body inquestion. Typically, DOM concentrations decreasefrom freshwater to coastal to marine environments,and tend to be highest in coastal regions and harbours,owing to the proximity of terrestrial sources (Coble et

36

Hunt et al. : Optical signatures in ocean ballast water

al. 2004). DOM decreases in concentration further off-shore, where the sources of DOM are greatly reduced.Marine organic matter is primarily derived from thedecomposition of marine organisms such as phyto- andzooplankton: when these organisms die, decompo-sition processes produce a complex of colouredorganic matter.

One of the distinctive properties of coloured DOM isfluorescence in the ultraviolet and blue regions of thelight spectrum. Two distinct types of fluorescence sig-nal attributable to DOM have been identified in sea-water: a humic-like fluorescence and a protein- oramino acid-like fluorescence (Coble 1996). It is possi-ble to use these fluorescence signatures to identify dif-ferent types of DOM using excitation-emission matrix(EEM) spectra. EEM spectra show characteristic exci-tation/emission maxima (Exmax/Emmax) for differentwater types. Average wavelengths for these maximafor different water types have been shown to be: riversExmax/Emmax = 340/48 nm; coastal water Exmax/Emmax =342/442 nm; marine shallow transitional Exmax/Emmax =310/423 nm; marine shallow entropic Exmax/Emmax =299/389 nm; and marine deep water Exmax/Emmax =340/438 nm (Coble 1996). These average maximademonstrate that fluorescence characteristics of watertypes differ and suggest that the specificity of thespectra was potentially useful for verifying mid-oceanballast water exchange.

The objective of this study was to assess whetherfluorescence measurements of seawater were usefulfor identifying different water types in ballast tanksduring exchange at sea on a commercial vessel and toassess whether the method could be used to verifymid-ocean ballast water exchange. To ensure that theefficiency of the exchange was known, a tracer dye(Rhodamine WT) was used as a comparative measureof dilution efficiency while the ballast water wasexchanged. The biological effectiveness of theexchange was also examined (Taylor & Bruce 1999).

MATERIALS AND METHODS

Experimental design. Ballast water exchange trialswere carried out during a voyage of the MT ‘IverStream’ from Kawasaki Harbour, Japan to SingaporeHarbour (first transect: 23 February to 8 March 1999)and from Singapore Harbour to New Plymouth, NewZealand (second transect: 8 to 24 March 1999). The MT‘Iver Stream’ is a methanol carrier that plies a regulartrans-Pacific route between New Zealand, Japan andother parts of Asia, and arrives in New Zealand fullyballasted. It has a dead-weight (i.e. weight of cargothat can be carried) of 35 273 short tons (Taylor & Bruce1999), equivalent to 32 000 metric tonnes.

The ship’s track-line and the location of ballast waterexchanges and mid-ocean sampling during the voyageare shown in Fig. 1. At each of the donor ports(Kawasaki Harbour and Singapore Harbour) and dur-ing the voyage, ballast water and ocean water werecollected for measuring the dilution efficiency (dyeand salinity), optical characteristics of coastal, ocean,exchanged water, and changes in plankton communi-ties (Taylor & Bruce 1999).

The MT ‘Iver Stream’ has 2 pairs of port and star-board ballast tanks (capacity: ~1434 m3 each; depth:15 m; width: 6.5 m; length: 14.5 m). Although the bal-last tanks were not compartmentalized they containedinternal platforms and framing structures and a ladderthat allowed access to all levels. All mid-oceanexchanges were carried out using the flow-throughdilution method (maximum pump rate: 170 m3 h–1).The vessel traveled approximately 900 to 1200 km dur-ing the time it took to exchange 3 times the volume of1 ballast tank, i.e. over 24 to 40 h, depending on thepumps that were available. All ballast water sampleswere collected via Butterworth hatches located on theseaward side of the top of the tanks.

At the beginning of the first transect (23 February1999) all 4 ballast tanks were completely emptied andthen cleaned of the remaining sediments prior to fillingin Kawasaki Harbour. The objective of cleaning thetanks was to minimize the level of interference of thefluorescence intensity measurements used for the opti-cal characteristic studies. During filling, the seawaterdepth was approximately 10 m and the ballast waterintake was at approximately 7 m below the harbour’ssurface. The Rhodamine WT dye was added as thetanks filled to achieve a nominal 0.1 ppm concentra-tion. The final dye concentration in the tanks wasdetermined from samples collected from multipledepths using standard fluorescence methods.

On the first transect, the first exchange was carriedout in the East China Sea (approximate depth: 6400 m)from 26 to 28 February 1999 using the port ballast tank(Tank 6P). The second exchange was carried out in theSouth China Sea (approximate depth: 3700 m) from 1to 2 March 1999, using the starboard ballast tank(Tank 6S). The second port ballast tank (Tank 7P) wasused as a non-exchanged control for both exchanges.The second starboard ballast tank (Tank 7S) wasintended as a second control; however, this tank wasnot completely filled in Kawasaki Harbour and wassubsequently ‘contaminated’ by coastal water upliftedoff the coast of Japan several hours after departure.Consequently, this tank was only used for assessingthe stability of the optical signature. For bothexchanges on the first transect, ocean water waspumped in through 4 to 6 large deck hoses into theman-hole at the top of the tank and out via another

37

Mar Ecol Prog Ser 331: 35–47, 200738

Fig. 1. Voyage of the MT ‘Iver Stream’ from Japan to New Zealand, 22 February to 24 March 1999, showing locations of mid-ocean ballast water exchanges and mid-ocean sampling points

Hunt et al. : Optical signatures in ocean ballast water

pump at the bottom of the tank, near the aft end. Threetank volumes were exchanged.

At the beginning of the second transect (8 March1999) the ballast tanks were partly filled with Singa-pore Harbour water. The filling was completed as thevessel travelled out of the harbour (between 01° 20’ N103° 50’ E and 02° 26’ N 106° 56’ E). The first exchangewas made in the west Pacific off the east coast of WestPapua (approximate depth: 1400 m) from 13 to 14March 1999, using one port ballast tank (Tank 6P). Thesecond exchange was made off the north coast ofPapua New Guinea (depth: 900 to 2000 m) from 15 to17 March 1999, using one starboard tank (Tank 6S).The first exchange was complete. However, the secondexchange was equal to only 2 times the tank volume,owing to vessel time constraints. For both exchangeson the second transect, the water was pumped in at thebottom, near the aft end of the tanks, and allowed tooverflow out of the hatches described above.

The sampling program on the first transect consistedof samples collected from Kawasaki Harbour, themid-ocean during exchanges, and ballast water samplescollected from both of the exchange and control tanks.The ballast tanks were initially sampled several hoursafter uplifting ballast in Kawasaki Harbour, severalhours before and after exchanges, and at the end of thetransect in Singapore Harbour. Kawasaki Harbour waterremained in the control tanks for the entire transect, andso traversed a wide latitudinal range from temperateregions to the tropics (Fig. 1). It was therefore subjectedto considerable variation in temperature.

Fluorescence analyses of samples collected duringthe second transect was limited. During the second ex-change on the second transect, the vessel sailed in closeproximity (10 to 15 km) to the mouth of the Sepik River,hence the results of this exchange were of special inter-est with respect to changes in salinity and the composi-tion of plankton communities within the tank. No con-trol tank was available for this exchange, however, dueto cargo loading constraints. Data not presented in thispaper can be found in Taylor & Bruce (1999).

Sample collection. Unfiltered coastal and ballastwater samples (500 ml) were collected for fluorescenceanalysis using a 5 l van Dorn sampler. Mid-ocean sur-face water samples were collected using a stainlesssteel bottle, which was cast into the sea from the ves-sel’s foredeck. Separate samples for fluorescence anddye analysis were taken from each sampler. Duplicatesamples were collected from mid-water strata of theports and also ballast water tanks from the van Dornbottle to assess variation in the analytical method. Atthe end of each transect the samples for fluorescenceanalysis were shipped to Battelle. All water sampleswere stored in the dark at 4°C in amber polyethylenebottles until analyzed, typically within 30 to 45 d of col-

lection. Water samples were not filtered, as we believethe unfiltered water more accurately reflects whatmight be encountered in a compliance verification pro-gram. This would typically be conducted under tightsample collection timelines, would not have the luxuryof filtering samples in the field, and could, in concept,rely on optical probes to perform the fluorescencemeasurement.

Fluorescence measurements. High-resolution EEMspectra were obtained with a SPEX Fluorolog 2 Spec-trofluorometer. Measurements were performed at theUniversity of Massachusetts, Boston. Samples werelightly shaken and the sampling bottle inverted severaltimes to ensure a homogenous mixture; they were thenpoured into a clean 1 cm quartz cuvette for analysis.Excitation wavelengths ranged from 220 to 400 nm at awavelength increment of 5 nm and a bandwidth of4 nm. The emission wavelengths ranged from 250 to550 nm at an increment of 2 nm and a bandwidth of4 nm. The instrument integration time was set at 0.5 s.The time required to complete all 37 emission scans ona sample was approximately 1.5 h. All fluorescencemeasurements were performed at 20°C. The region ofinterest is shown in Fig. 2.

For the selected excitation wavelength range (220 to400 nm) the individual excitation and emission scanswere concatenated in MATLAB into an EEM matrix.Excitation and emission instrument correction factorswere applied to the data to correct for the spectral vari-ations in the lamp output, optics, gratings and detectorresponse. These correction factors were supplied bythe manufacturer. A quinine sulfate standard wasmeasured periodically for a 335 nm excitation wave-

39

Fig. 2. Different regions of an excitation-emission matrixcontour plot for a typical water sample. Scale: fluorescence

intensity in counts s–1

Mar Ecol Prog Ser 331: 35–47, 2007

length and the maximum fluorescence intensity ofquinine sulfate was recorded. A normalization processwas not applied to the data since variation in the fluo-rescence intensity of the quinine sulfate peak was rel-atively low (≤17% across all samples analyzed). EEMswere also recorded for MilliQ water in case waterRaman scattering corrections were required.

RESULTS

Dilution efficiency

The concentration of Rhodamine WT added to theballast tanks in the harbours was reduced by morethan of 95% at all 3 of the ballast tank depth stratasampled for the mid-ocean exchanges on the voyage(see Fig. 3a and Taylor & Bruce 1999). Time series sam-pling during the first exchange on the second transect(filling from the bottom), showed stratification of thedye in the tank after 75% of the tank volume had beenpumped through the tank (Rhodamine WT dilution:47% at 0.5 m, 68% at 7.0 m, and 87% at 14.0 m). Strat-ification of the dye was less apparent after 1.5 times thetank volume had been pumped through. At this stagethe dye dilution was more than 95% at all 3 depthstrata (Fig. 3b). The dye data from the exchanges using3 times the tank volume resulted in exchange of morethan 99% of the coastal water.

Optical characteristics of the samples

A significant amount of information is containedwithin each EEM spectra. These data are most readily

presented as contour plots of the fluorescence intensityof paired excitation versus emission wavelengths(Fig. 4). The coastal water samples from this studytypically show regions of higher fluorescence intensitycompared to the mid-ocean sample. These were espe-cially evident in the lower regions of the spectra forexcitation wavelengths ranging from 220 to 340 nm.The regions of higher fluorescence indicate the pres-ence of different compounds based on the fluorescingproperties of the water. For example, protein-like fluo-rescence is attributed to aromatic amino acids, eitherfree or as protein constituents in the water (Coble1996). Moreover, there are 2 types of protein-like fluo-

40

0.00

0.05

0.10

0.15

0.20 A BC

once

ntra

tion

Rho

dam

ine

WT

(pp

m) Depth

Afterexchange

Beforeexchange

0.5 m7.0 m14.0 m

Beforeexchange

0.75 1.5 2.25 Afterexchangetank vol.

Fig. 3. Rhodamine WT concentrations at 0.5, 7.0 and 14.0 m depth strata in ballast tanks. (A) Before and after mid-oceanexchange of the starboard tank on the first transect (1 to 2 March 1999); (B) before, during and after mid-ocean exchange of the

port tank on the second transect (13 to 14 March 1999)

Fig. 4. Excitation-emission matrix contour plot for a typicalcoastal water sample, showing protein-like and humic-likefluorescence peaks. Scale: fluorescence intensity in counts s–1

Hunt et al. : Optical signatures in ocean ballast water

rescence: tyrosine-like, which occurs at 300 to 305 nmfrom excitation wavelengths of 220 to 225 nm and 275to 280 nm; and tryptophan-like, which occurs at 340 to350 nm, from excitation wavelengths of 220 to 225 nmand 275 to 280 nm. Fig. 4 shows a tryptophan-like peaktypically observed in the samples from this study.Other peaks in the EEM spectra can be attributed tohumic-like compounds. In general, humic-like emis-sion fluorescence occurs at 420 to 460 nm, from excita-tion wavelengths of 230 to 260 nm and 320 to 350 nm(Coble 1996).

Coastal water optical spectra

A water column profile of EEM spectra for coastalwater from Kawasaki Harbour was developed tocharacterize water at the port of origin and to deter-mine if changes occurred after ballasting (Fig. 5A–C).At the surface (0.5 m), the fluorescence was greaterthan 15 × 105 counts s–1 at almost all wavelengths.Fluorescence intensity decreased with depth, reflect-ing an increase in salinity from 30.6 psu at 0.5 m to32.9 psu at 10 m.

The fluorescence of the water samplescollected from Singapore Harbour alsodecreased as a function of depth; how-ever, fluorescence intensity was gener-ally lower at all depths compared to theKawasaki Harbour samples. The regionof highest fluorescence intensity was theprotein-like peak with a maximum inten-sity of approximately 9 × 105 counts s–1 inthe surface water sample. This intensitycontrasts with the Kawasaki Harbourwater that had a maximum fluorescencegreater than 70 × 105 counts s–1 in thesame region of the spectrum.

Mid-ocean water optical spectra

The EEMs of 3 mid-ocean surfacewater samples (Fig. 5D–F) taken duringthe exchange of the starboard tankballast water uploaded in KawasakiHarbour show a distinct difference in theoptical characteristics of the mid-oceanwater and the Kawasaki Harbour water.The mid-ocean EEMs clearly showhigher salinity and lower fluorescenceintensity than the coastal water samples.While little structure in the mid-oceanwater fluorescence was present therewas a peak in fluorescence around theprotein-peak (excitation: 270 nm; emis-sion: 340 nm, which was very small rela-tive to the protein-like signature in thecoastal water samples.

The 22 mid-ocean surface water sam-ple EEMs from the ship’s entire tracklinewere generally similar, as represented inFig. 6. The optical signature of mid-ocean samples consistently had low fluo-rescence intensity (typically less than 2 ×105 counts s–1) and little structure. Thesamples typically had a small protein-like fluorescence; however, none of the

41

Fig. 5. Excitation-emission matrices for samples collected on the first transect ofthe voyage. (A–C) Coastal water samples collected from Kawasaki Harbour;(D–F) mid-ocean surface samples collected during the second exchange.

Scale: fluorescence intensity in counts s–1

Mar Ecol Prog Ser 331: 35–47, 2007

samples showed clear fluorescence peaks in either ofthe humic-like regions examined.

Emissions scans at single excitation wavelengths areuseful for directly comparing differences in fluorescenceintensity between water samples (Fig. 7). Emission scansof Kawasaki Harbour coastal water from 10 m and a sur-face sample from off the west coast of Japan (31° 27’ N,135° 40’ E) had clear differences in fluorescence inten-sity. At the 250 and 330 nm excitation wavelengths, thehumic-like peaks can be clearly seen in the coastal water emission scan, with peak emissions at 430 and420 nm, respectively. The humic-like peaks tend to bebroad, ranging from approximately 400 to 460 nm. Theprotein-like peak can be clearly seen in the coastal water sample at an excitation wavelength of 275 nm. Thepeak is centered at 342 nm and does not fluoresce overthe broad range of wavelengths observed in the humic-like fluorescence. The Rayleigh and Raman scatterpeaks are also clearly seen in the mid-ocean sample;however, there is no humic-like fluorescence and only aslight protein-like peak at an excitation of 275 nm.

Ballast water optical signatures before and after mid-ocean exchange

The optical stability of the source water fluorescencewas documented by comparing the EEMs of the coastal

water to water in the ballast tanksshortly after ballasting and also be-fore the mid ocean exchange and inthe unexchanged control tank overthe duration of the survey. The EEMsof samples taken at 3 depths (0.5, 7and 14 m) in the ballast tanks severalhours after ballasting in KawasakiHarbour (24 February 1999) showedthat the fluorescence signature inthe tank was structurally similarbut slightly more intense than theKawasaki Harbour water at 10 mdepth. Ballast tank samples taken4 days later showed the fluorescenceintensity of the protein like peak hadincreased slightly at all depths (from~1.9 × 106 to ~2.2 × 106 counts s–1),which may have been due to thegrowth of bacteria in the tank or otherbiological activity such as decom-position of phytoplankton and zoo-plankton. Similarly, the fluorescencesignature of the control tank (notexchanged) did not change signifi-cantly over the duration of the tran-sect (Taylor & Bruce 1999).

The EEMs of ballast tank samples from Tank 6S on28 February 1999 just before exchange was startedwere generally uniform and similar in fluorescencestructure and intensity to the water from KawasakiHarbour (compare Fig. 8A–C to Fig. 5C). The opticalsignature of ballast water several hours after comple-tion of the exchange on 3 March 1999 were low andhad very little structure(Fig. 8D–F). Note the ex-change increased the salinity from ~32.8 to 34.2 psu.Moreover, the optical signature after exchange wassimilar to that of the mid-ocean surface samples col-lected during the exchange (Taylor & Bruce 1999).The compounds containing fluorophores which con-tribute to the humic-like peaks were not present inthe EEMs of the samples collected after the exchange.The EEMs of the exchanged ballast tank 4 d after thecompletion of the exchange remained low and similarto those at the end of the exchange, although therewas a slight increase in the fluorescence intensityaround the protein-peak, which suggests that somebacterial growth may have occurred in the tank overthis period.

Table 1 summarizes the fluorescence intensity of theprotein-like and 430 nm humic-like fluorescencepeaks for coastal and mid-ocean water samples, aswell as ballast water samples collected from theexchanged starboard tank, i.e. Tank 6S. The peakvalues were determined by selecting the maximum

42

400

350

300

250

x105

6

4

2

0

x105

6

4

2

0

x105

6

4

2

0

x105

6

4

2

0

400

350

300

250

400

350

300

250

400

350

300

250

Exc

itatio

n w

avel

engt

h (n

m)

Emission wavelength (nm)

Boo02s Boo04s

Boo21s Boo36s

300 400 500 300 400 500

300 400 500 300 400 500

Fig. 6. Excitation-emission matrices for mid-ocean water samples collectedthroughout the voyage. Scale: fluorescence intensity in counts s–1, scaled to maxi-mum fluorescence of 7 × 105 counts s–1. The sample locations can be found in

Fig. 1 as follows: Boo02s = 2; Boo04s = 5; Boo21s = 7; Boo36s = 10

Hunt et al. : Optical signatures in ocean ballast water

fluorescence intensity in the region surrounding thepeak that included a portion of the spectrum approxi-mately 10 nm wide. The absolute magnitude of thefluorescence intensity of the ballast water samplesollected before the exchange was 5 to 10 times greaterthan the fluorescence intensity of the ballast watersamples collected after the exchange.

The Raman peak value was also determined forthese samples to serve as a reference point. The ratioof Raman value to the fluorescence peaks is presentedin order to illustrate the differences between samplescollected before and after the exchange. The humicand protein ratios ranged from 0.09 to 0.14 and from0.28 to 0.62, respectively, for all of the mid-ocean watersamples. This was consistent with the values for theballast water samples collected after the exchange,which had values ranging from 0.10 to 0.14 for thehumic ratio and from 0.31 to 0.33 for the protein ratio.In contrast, the ballast water samples collected beforethe exchange had values ranging from 0.58 to 0.75 forthe humic ratio and from 1.60 to 2.18 for the proteinratio

The changes in optical characteristics of the samplescollected from the second transect (Singapore to NewPlymouth, NZ) were similar to those observed for thefirst transect, even though the fluorescence intensitiesof the ballast water samples from Singapore Harbourwere 2 to 3 times lower in magnitude than the fluores-cence intensities of the samples from Kawasaki Har-bour. Even so, the mid-ocean exchange on this transectreduced the fluorescence intensity and structure of thecoastal water to that of the mid-ocean water. Thus,even though there were large differences in the fluo-rescence intensity of the 2 source waters, the fluores-cence data clearly demonstrated the exchange ofcoastal waters for oceanic water.

DISCUSSION

The EEM spectra derived from the coastal watersamples collected on the voyage showed a decreasingDOM fluorescence gradient with depth of the watercolumn, with the highest fluorescence intensity in thelower salinity surface water. The coastal water samplesalso exhibited fluorescence peaks in 3 distinct regions:a protein-like (tryptophan-like) peak and 2 humic-likepeaks. In contrast, the EEM spectra derived fromthe 22 mid-ocean samples collected throughout thevoyage had fluorescence signatures that were low inintensity and consistently lacked fluorescence struc-ture. Together these data confirm reports in the litera-ture (e.g. Coble 1996, Coble et al. 2004) that DOMfluorescence is useful for distinguishing betweenwater types.

43

250 300 350 400 450 500 550

A) Emission scans at excitation wavelength250 nm (Bcw03s and Boo02s)

Flu

ore

scen

ce in

tens

ity x

10

6

Coastal waterMid-ocean water

Coastal waterMid-ocean water

Coastal waterMid-ocean water

Humic-likefluorescencepeak

Humic-likefluorescencepeak

Humic-likefluorescencepeak

Ramanscatter

Ramanscatter

2nd orderRayleighscatter

2nd orderRayleighscatter

B) Emission scans at excitation wavelength275 nm (Bcw03s and Boo02s)

Rayleighscatter

Rayleighscatter

Tryptophan-likeproteinfluorescence peak

C) Emission scans at excitation wavelength330 nm (Bcw03s and Boo02s)

Emission wavelength (nm)

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Fig. 7. Emission scans at (A) 250, (B) 275 and (C) 330 nm forcoastal water samples collected at 10 m depth from KawasakiHarbour, and for a mid-ocean water sample collected off

the west coast of Japan

Mar Ecol Prog Ser 331: 35–47, 2007

The sensitivity of florescence measurement for char-acterizing ocean waters was demonstrated in severalways in this data set. For example, there were clearchanges in fluorescence intensity with depth in ports.In addition, the fluorescence signature of ballast watersamples collected several hours after ballasting wasgenerally consistent with the signature of coastal watersamples collected at approximately the same depth atwhich the tanks were filled. Moreover, small changes

in the fluorescence spectra of indepen-dently collected, depth-stratified bal-last water samples varied slightly influorescence intensity. This variationwas not always reflected in the corre-sponding salinity measurements andimplies that ballast water stratificationnot detected on the basis of physicalvariables such as temperature andsalinity may be detected by character-izing the optical signature. The sen-sitivity of fluorescence was furtherdemonstrated by an apparent increasein fluorescence intensity in the proteinregion in ballast water samples col-lected after uptake. This increase influorescence intensity is probably dueto decomposition of organisms or bac-terial growth within the tanks.

In contrast to the subtle changesconsidered above, marked shifts in theoptical signature of the ballast waterwere evident after mid-ocean ex-change, which coincided with a >99%reduction in the added dye concen-tration by the mid-ocean exchanges(Taylor & Bruce 1999). The reductionin dye concentration is significant inthat it confirmed that the ballast waterwas fully exchanged in mid-ocean.

Contrary to the change in dye con-centrations and fluorescence signa-ture, the salinity and biological datafrom the voyage were not found to beeffective in estimating the efficiency ofexchange (Taylor & Bruce 1999). Sal-inity was ineffective due to the smalldifferences between coastal and oceanwater, imprecision in measurement,and the confounding influence of rem-nant river plumes in the offshorewaters exchanged (Taylor & Bruce1999). Moreover, the exchange effi-ciency estimated from the biologicaldata did not match well with the dyeresults, primarily due to the abundance

and diversity of oceanic species that were added to theballast tank during the exchange. Even so, the plank-ton data showed a marked decrease in phytoplanktonand zooplankton abundance and a shift in the speciescomposition from that originally uplifted from the har-bours (Taylor & Bruce 1999). The biological data alsoshow that species richness in the exchanged waterdecreased approximately 55% compared with <37%in the control tanks. Thus, it was difficult to correctly

44

Fig. 8. Excitation-emission matrices for samples collected from 3 depths in star-board ballast Tank 6S before exchange and several hours after exchange on first

transect of voyage. Scale: fluorescence intensity in counts s–1

Hunt et al. : Optical signatures in ocean ballast water

assign the attrition in planktonic organisms to theexpulsion of organisms by exchange, versus naturalmortality. Taylor & Bruce (1999) also conclude that theusefulness of coastal taxa as indicators of open oceanexchange is limited due to the cosmopolitan natureof many species.

One of the study objectives was to ensure that mid-ocean ballast water exchange could be quantifiedusing dye data. The dye study achieved this objective.It is less clear from the data whether the differences inthe optical signatures and intensity before and afterexchange could quantitatively support estimates ofexchange efficiency. Although the fluorescence studywas not designed to develop the highly precise fluores-cence data necessary to quantify exchange efficiency,we explored its potential for estimating exchangeefficiency. Estimating exchange efficiency based onflorescence data can be problematic due to the manyflorescence and excitation wavelengths that could beused to derive efficiency estimates and the ability toprecisely quantify fluorescence intensity of the natu-rally occurring DOM. Moreover, estimates of exchangeefficiency using natural fluorescence assumes themeasured fluorescence is derived from the same orvery similar compounds, which while true at a generallevel, is not explicitly accurate due to the complexnature of the organic matter that contributes to thefluorescence signature and the level of analytical

precision achieved by the study. Recognizing theselimitations, the large difference between the coastaland open ocean fluorescence intensity was used toestimate exchange efficiency for comparison with thedye data. The data in Fig. 7 suggests that the peakfluorescence intensity at the 250 and 275 nm excitationlines was reduced by approximately 90% by theexchange. Exchange efficiencies calculated from theaverage peak protein and humic fluorescence inten-sities before and after exchange (Table 1) suggest anexchange efficiency of 88 and 90%, respectively. Eventhough the relative precision derived from these sam-ples is moderate (~25%), the exchange efficiency esti-mated from the fluorescence data is reasonably similarto that derived from the highly precise dye data.

These findings support our hypothesis that naturalfluorescence can be used to verify exchange. However,further studies are necessary in order to ensure therobustness and defensibility of the method as a verifi-cation method. Among these are identification of ap-propriate instrumentation, sample turn around times,and verification protocols. Moreover, the ability toidentify a small set of wavelengths, rather than use theentire EEM spectra for verification, suggests that opti-cal verification could be conducted without the use ofsophisticated bench scale laboratory instruments.

One of the concerns for our study design was the sta-bility of the optical signature in the ballast tanks and

45

Sample Sample codea Protein Humic Raman Protein/ Humic/in figures (Fig) peak peak peak Raman Raman

Kawasaki Harbour coastal water CW 01 Bcw01s (5) 7.92 × 106 4.84 × 106 2.95 × 106 2.68 1.64CW 02 Bcw02s (5) 4.92 × 106 1.51 × 106 1.59 × 106 3.09 0.95CW 03 Bcw03s (5) 1.90 × 106 7.82 × 105 1.03 × 106 1.86 0.76

Mid-ocean water first transect MO 01 3.23 × 105 1.03 × 105 9.69 × 105 0.33 0.11MO 02 2.04 × 105 6.33 × 104 7.24 × 105 0.28 0.09MO 03 3.67 × 105 7.30 × 104 7.59 × 105 0.48 0.10MO 04 Boo04s (6) 3.71 × 105 1.32 × 105 9.57 × 105 0.39 0.14MO 05 1.75 × 105 6.25 × 104 4.83 × 105 0.36 0.13MO 06 Boo06s (5) 3.11 × 105 6.57 × 104 4.99 × 105 0.62 0.13MO 07 Boo07s (5) 2.02 × 105 6.15 × 104 7.14 × 105 0.28 0.09MO 08 Boo08s (5) 2.61 × 105 8.57 × 104 7.29 × 105 0.36 0.12MO 09 2.32 × 105 6.82 × 104 6.82 × 105 0.34 0.10MO 10 Boo10s (6) 2.28 × 105 9.13 × 104 6.83 × 105 0.33 0.13

Before exchange ballast water BW 01 2.16 × 106 7.78 × 105 1.35 × 106 1.60 0.58(Tank 6S) BW 02 2.25 × 106 7.16 × 105 1.03 × 106 2.18 0.69

BW 03 1.18 × 106 5.41 × 105 9.39 × 105 1.26 0.58BW 10 Bb1w10s (8) 2.58 × 106 9.71 × 105 1.45 × 106 1.78 0.67BW 11 Bb1w11s (8) 2.89 × 106 9.82 × 105 1.38 × 106 2.09 0.71BW 12 Bb1w12s (8) 2.76 × 106 9.74 × 105 1.30 × 106 2.12 0.75

After exchange ballast water BW 23 Bb1w23s (8) 2.24 × 105 8.24 × 104 7.17 × 105 0.31 0.12(Tank 6S) BW 24 Bb1w24s (8) 3.43 × 105 1.40 × 105 1.03 × 106 0.33 0.14

BW 25 Bb1w25s (8) 2.15 × 105 6.79 × 104 6.81 × 105 0.32 0.10

Table 1. Summary of fluorescence intensities (counts s–1) for the protein-like and humic-like fluorescence peaks and correspond-ing ratios to the Raman scattering peak, for various samples collected during the first transect of the voyage. CW: coastal water;

BW: ballast water; MO: mid-ocean water. aSample codes Boo02s, Boo21s and Boo36s (Fig. 6) not in Table 1

Mar Ecol Prog Ser 331: 35–47, 2007

the possibility of changes during shipment. Given theneed for a highly controlled laboratory environmentfor the EEM analysis, time series studies to test stabil-ity could not be conducted on board the vessel. How-ever, the fluorescence signatures of independent watersamples collected from multiple depths in the unex-changed and exchanged ballast tanks over the dura-tion of the transects (e.g. 12 d on the first transect) pro-vide evidence for fluorescence stability. Likewise, thefluorescence of exchanged ballast water sampled frommultiple depths over time was consistent and onlyshowed small changes in the protein like area of thespectra. These data further suggest that the opticalsignature of the water loaded into the ballast tanks isrelatively stable over periods typical of ocean voyages,an important factor when considering this method as ameans of verifying mid-ocean exchange. The changesin these spectra also suggest this spectral region maynot be the most appropriate area to use for verification.

It is also possible that post-collection changes in thefluorescence characteristics of the samples occurred dur-ing storage on board the vessel and subsequent ship-ping. The concern is that biological activity after thesamples are taken could change the fluorescence signa-ture, thereby making it difficult to distinguish betweenthe original signature and any subsequent changes. Thisconcept was considered during the design of this studyand points to one of the important issues surroundingany verification method, minimizing changes and thetime interval between collection and analysis. The factthat EEMs of independent samples taken over severaldays from multiple depths did not change is encouragingand points to the need for verification methods thatminimize the introduction of potential sampling and han-dling artifacts. One way to control for these concerns is totake advantage of in situ fluorescent measurements.Such measurements would also provide practical conve-nience by enabling rapid shipboard measurements andreduce the costs of the analysis.

CONCLUSIONS

A voyage on a trans-Pacific chemical carrier MT ‘IverStream’ confirmed that optical characteristics of sea-water could be used to verify the compliance of shipswith mandatory international ballast water exchangerequirements. The UV fluorescence of organic matterin seawater, expressed through high resolution EEMspectra and selected areas of the fluorescence spectra,readily discriminates between coastal and mid-oceanwater. The lack of optical signatures in mid-Pacificwater is significant for the hypothesis that optical char-acteristics of water can be used to verify ballast waterexchange. We would expect this to hold generally true

of other oceans. This, coupled with the relatively largeand optically distinct fluorescence structure of coastalwaters, supports the hypothesis that fluorescencebased methods have great potential to verify mid-ocean ballast water exchange.

Unfortunately, the nature of this study did not allowus to address questions on the best way to conduct ver-ifications of open ocean exchange. It is clear, however,that any verification procedure must be completedquickly (i.e. <1 h), accurately, preferably onboard thevessel to minimize shipping delays, and at reasonablecosts. It is also clear that collecting samples for analysisusing highly specialized laboratory instruments andmethods will not be appropriate for exchange verifica-tion. The data from this study also suggested that fullEEMs are not necessary for verification; 1 or 2 wave-lengths may suffice, as considered by Murphy et al.2004. Moreover, the data from this study suggest thatshort delays in performing high resolution analyses,while undesirable, will not affect the optical signatureof the water if the sample is stored properly.

In concept, the requirements for rapid low cost veri-fication could be met by field instruments that aretransported and deployed by 1 or 2 individuals. Thereare currently several commercially available fieldinstruments based on florescence techniques whichmay be useful for optical based verification of mid-ocean exchange. However, in a regulatory environ-ment, the methodology must have low potential fordata ambiguity, be accurate/precise, and have a lowchance of sample contamination or sample degrada-tion. Thus, instrumentation specifically developed orused for verification on board vessels must undergorigorous testing to ensure that the data is appropriateand defensible for expected regulatory requirements.The availability of portable field instrumentation willmake the optical verification of ballast water exchangerelatively straight forward, rapid, and cost effective.

Acknowledgements. Staff from both Cawthron and Battellecontributed to this research, in particular: Lincoln MacKenzie,Tim Dodgshun, Cameron Hay, Wendy Gibbs, Henry Kasparand Rod Asher of Cawthron, and Karen Foster and ScottMacomber of Battelle. Bob Chen and Steve Rudnick of theUniversity of Massachusetts, Boston, provided invaluableassistance in analysis and discussion of the EEM data. Sincerethanks to Vroon BV, Ship managers, The Netherlands, forpermission to travel aboard the MT ‘Iver Stream’. Thanks alsoto Pieter van Leeuwen of Waterfront Shipping Co. Ltd, Van-couver, Stephen Bull of Hooker Pacific, New Plymouth andK. Awane of ALL Barwill Agencies, Japan, for their invaluableassistance prior to and during the voyage. This researchwould not have been possible without the generous assis-tance of the captain, officers, and crew of the MT ‘IverStream’. This research was funded by the New Zealand Min-istry of Fisheries. We also thank the peer reviewers, whosecomments greatly improved the presentation and content ofthe paper.

46

Hunt et al. : Optical signatures in ocean ballast water

LITERATURE CITED

Carlton JT, Reid DM, Leeuwen H (1995) Shipping study – Therole of shipping in the introduction of nonindigenousaquatic organisms to the coastal waters of the UnitedStates (other than the Great Lakes) and an analysis of con-trol options. Final report from the Maritime Studies Pro-gram, Williams College, Mystic, CT, to the US CoastGuard, CG-D-11–95. Government Accession NumberAD-A294809

Coble P (1996) Characterization of marine and terrestrialDOM in seawater using excitation-emission matrixspectroscopy. Mar Chem 51:325–346

Coble P, Brophy M (1994) Investigation of the geochemistryof dissolved organic matter in coastal waters usingoptical properties. SPIE Vol 2258, Ocean Optics XII,p 377–389

Coble P, Hu C, Gould RW, Chang G, Wood AM (2004)Colored dissolved organic matter in the coastal ocean.Oceanography 17:50–59

Harvey M, Gilbert M, Gauthier D, Reid D (1999). A prelimi-nary assessment of risks for the ballast water-mediatedintroduction of nonindigenous marine organisms in theEstuary and Gulf of St. Lawrence. Can Tech Rep FishAquat Sci 2268

Hay C, Tanis D (1998) Mid ocean ballast water exchange: pro-cedures, effectiveness and verification. A report preparedfor the New Zealand Ministry of Fisheries. CawthronReport No. 468. Cawthron Institute, Nelson

International Maritime Organization (2004) International con-vention for the control and management of ships ballastwater & sediments, 2004. International Maritime Or-

ganization, London. Available at: www.imo.org/home.asp?topic_id=583&doc_id=2689

Kirk J (1983) Light and photosynthesis in aquatic ecosystems.Cambridge University Press, New York

Murphy K, Boehme J, Coble P, Cullen J and 5 others (2004)Verification of mid-ocean ballast water exchange usingnaturally occurring coastal tracers. Mar Poll Bull 48:711–730

National Research Council (1996) Stemming the tide-controlling introductions of nonindigenous species byship’s ballast water. National Academy Press, Washington,DC

Rigby G, Hallegraeff G (1994) The transfer and control ofharmful marine organisms in shipping ballast water:behavior of marine plankton and ballast water exchangetrials on the MV ‘Iron Whyalla’. J Mar Environ Eng 1:91–110

Ruiz GM, Fofonoff PW, Carlton JT, Wonham MJ, Hines AH(2000) Invasion of coastal marine communities in NorthAmerica: apparent patterns, processes, and biases. AnnuRev Eco Syst 31:481–531

Taylor M, Bruce E (1999) Mid ocean ballast water exchange:shipboard trials of methods for verifying efficiency.Cawthron Report No. 524, Cawthron Institute, Nelson

USCG (United States Coast Guard) (2001) Report to Congresson the voluntary national guidelines for ballast watermanagement. US Coast Guard, Washington, DC

USCG (2004) Mandatory ballast water management programfor US waters: final rule. Federal Register, Vol. 69, No. 144.USCG-2003–14273

Vitousek PM, Mooney HA, Lubchenco J, Melillo JM (1997) Hu-man domination of Earth’s ecosystems. Science 277:494–499

47

Editorial responsibility: Victor de Jonge (Contributing Editor),Haren, The Netherlands

Submitted: March 31, 2005; Accepted: June 26, 2006Proofs received from author(s): January 29, 2007