Skip to main content
NCBI home page
EPA Author Manuscripts logoLink to EPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Mar 19.
Published in final edited form as: Mar Pollut Bull. 2022 Aug 1;182:113947. doi: 10.1016/j.marpolbul.2022.113947

First evaluation of ballast water management systems on operational ships for minimizing introductions of nonindigenous zooplankton

Sarah A Bailey a,*, Torben Brydges a, Oscar Casas-Monroy a, Jocelyn Kydd a, R Dallas Linley a, Robin M Rozon a, John A Darling b
PMCID: PMC10949186  NIHMSID: NIHMS1841017  PMID: 35926436

Abstract

Ballast water is a leading pathway for the global introduction of aquatic nonindigenous species. Most international ships are expected to install ballast water management systems (BWMS) by 2024 to treat ballast water before release. This study examines if ballast water discharges managed by BWMS are meeting standards for organisms ≥50 μm in minimum dimension (i.e., <10 organisms per m3; typically zooplankton). Representative samples of ballast water were collected from 29 ships (using 14 different BWMS) arriving to Canada during 2017–2018. Fourteen samples (48 %) had zooplankton concentrations clearly exceeding the standard (ranging from 18 to 3822 organisms per m3). Nonetheless, compared to earlier management strategies, BWMS appear to reduce the frequency of high-risk introduction events. BWMS filter mesh size was an important predictor of zooplankton concentration following treatment. Greater rates of compliance may be achieved as ship crews gain experience with operation and maintenance of BWMS.

Keywords: Aquatic invasive species, Biological introductions, Commercial shipping, Treated ballast water, Zooplankton diversity

1. Introduction

International shipping has been recognized as a leading mechanism of unintentional introductions of aquatic nonindigenous species to coastal ecosystems worldwide, resulting in significant ecological and socio-economic impacts (Bailey et al., 2020; Sala and Knowlton, 2006). Ships’ ballast water, used to change draft and trim, regulate stability, or maintain stress loads within acceptable limits, has been found to transport hundreds to tens of thousands of individuals of zooplankton per cubic meter, even following exchange of ballast water on the open ocean (Chan et al., 2014; Cordell et al., 2008; McCollin et al., 2008). This translocation of entire coastal planktonic communities around the world has been equated to ‘ecological roulette’, threatening coastal bays, estuaries and inland waters around the globe (Carlton and Geller, 1993). Such introductions of aquatic invertebrates can negatively impact biodiversity in recipient ecosystems, causing extinctions and significantly altering community structure and ecosystem function, threatening commercial fish stocks and aquaculture (e.g., Groeneveld et al., 2018; Katsanevakis et al., 2016). To reduce the risk of ballast-mediated introductions, the International Convention for the Control and Management of Ships’ Ballast Water and Sediments (hereafter known as the Convention) – which was adopted in 2004 and entered into force on September 8, 2017 – set limits on the concentration of viable organisms that can be discharged in ballast water (Regulation D-2) (International Maritime Organization, 2004). For organisms ≥50 μm in minimum dimension (typically zooplankton), ships shall discharge <10 viable organisms per cubic meter (hereafter the D-2 standard). The Convention will be implemented gradually, with most international ships expected to replace the use of ballast water exchange (BWE) with the use of ballast water management systems (BWMS) to meet Regulation D-2 by 2024.

A variety of BWMS have been developed which typically include two or three steps to treat ballast water. First, many BWMS utilize mechanical processes such as hydrocyclonic separation or filtration (nominal mesh sizes ranging between 20 and 55 μm) to remove larger organisms and particulates. Second, chemical or physical disinfection processes such as electrochlorination, ozonation or ultra-violet (UV) irradiation are applied as main treatments. For BWMS using chemical treatment, a neutralizer may be applied prior to discharging ballast water to the environment to meet maximum allowable discharge concentrations set by the IMO (typically 0.02 mg/L Total Residual Oxidants (TRO)) (Kim et al., 2016). While each BWMS is subjected to type-approval testing across a range of water quality conditions representative of the natural environment (e.g., temperatures, salinities, total suspended solids and plankton concentrations) (International Maritime Organization, 2018; NSF International, 2010), reliable performance of different BWMS across global environmental conditions is not guaranteed. Challenging water quality conditions may be expected at the many global ports located in estuaries and river deltas with high turbidity, known to vary with rainfall events, tidal currents and seasonal plankton blooms (e.g., Seers and Shears, 2015). Further, depending on the BWMS technology, treatment efficacy may vary across taxonomic groups and may be more effective at killing smaller phytoplankton cells than killing larger zooplankton or resistant life stages (Gregg et al., 2009; Lloyd’s Register, 2014).

Historically, zooplankton ballast water samples have typically been collected directly from ballast tanks using plankton nets or pumps and immediately preserved using buffered formalin or ethanol for storage until analysis can be completed (Briski et al., 2013; Gollasch and David, 2010; McCollin et al., 2008; Simard et al., 2011). With the entry-into-force of the Convention and the limits on viable organisms in ballast water discharge set by Regulation D-2, ballast water sampling guidelines now recommend collecting representative samples from the ship’s ballast piping, during discharge, as close as practicable to the point of discharge (ICES/IOC/IMO WGBOSV International Council for Exploration of the Sea/Intergovernmental Oceanographic Commission/International Maritime Organization Working Group on Ballast and Other Ship Vectors, 2017; International Maritime Organization, 2008). Further, as treated ballast water samples may contain a mix of live organisms and detritus, samples should be analyzed within a few hours of collection to determine the number of viable organisms at the time of discharge (International Maritime Organization, 2018; NSF International, 2010).

As the global implementation of the Convention and use of BWMS are in initial stages, the objective of this study was to collect and examine representative samples of ballast water discharge from ships using BWMS under operational conditions (i.e., as part of normal operations rather than during more controlled type approval tests) to determine rates of compliance for organisms ≥50 μm in minimum dimension with the limit set by Regulation D-2 (organisms in the size class ≥10 to <50 μm were also assessed, reported separately in Casas-Monroy and Bailey, 2021). Further, it was explored if zooplankton concentration in treated samples can be related to different management actions (such as BWMS disinfection method or filter mesh size) or ballast water history (e.g., source location or ballast water age). Finally, individuals confirmed as alive in treated ballast water samples were isolated and preserved for detailed taxonomic analysis, to look for patterns in the biodiversity of ballast water taxa following treatment. Diversity assessments were conducted using multiple methods (morphological taxonomy, direct DNA barcoding of specimens, and high-throughput sequence (HTS) metabarcoding of bulk samples) to explore differences in ballast water community profiles obtained through these methods.

2. Methods

2.1. Sample collection

Ships were selected opportunistically, with the requirement that they carried ballast water that had been treated with a BWMS, which could be discharged while our sampling team was on board to collect a representative sample. In 2017, ships arriving to ports across three Canadian regions (Pacific coast, Atlantic coast and Great Lakes) were targeted, while in 2018, sampling occurred exclusively at the port of Vancouver (Pacific coast) since this location receives the majority of shipping traffic in Canada, including ships with BWMS in use. During each test, the ship crew was interviewed to establish detailed information about the BWMS and the ballast water history before sampling.

Treated ballast water was collected during ballast discharge operations via the ship-supplied sample port and sample probe in the ship’s main discharge pipe located in the engine room, or in some cases on deck or in a pump room. A large volume sample collection device (LVCD) was used to collect and measure up to three cubic meters of treated ballast water for analysis of viable organisms in the ≥50 μm size class, collected as an inline continuous sample at targeted flow rates within, or below, the recommended isokinetic range to minimize flow turbulence and injury to organisms (ICES/IOC/IMO WGBOSV International Council for Exploration of the Sea/Intergovernmental Oceanographic Commission/International Maritime Organization Working Group on Ballast and Other Ship Vectors, 2017). Additionally, a small volume sample collection device (SVCD) integrated into the LVCD was used to collect 20–30 L of whole water for analysis of the ≥10–50 μm size class (Moser et al., 2018; results presented in Casas-Monroy and Bailey, 2021) and TRO (see below). The volume of water flowing through the LVCD was determined using a Bürkert S030 paddlewheel body with the SE35 monitor/digital display flow meter (Bürkert Fluid Control Systems, Germany).

The three cubic meters were collected as 1 m3 continuous samples. Each sample was concentrated through a 35 μm (50 μm in diagonal) mesh, 30 cm diameter net (Wildco-Science First, USA) partly submerged in the ballast water filtrate within a large plastic sampling bin (75 L) fitted with two gated (valved) 25 mm discharge lines, located top and bottom to maintain net submergence and direct the filtrate into the bilge of the ship, grey water tank or overboard (as directed by each ships’ Chief Engineer). When the sampling was complete, each net was rinsed down (spraying on the outside) using a spray bottle filled with 10-μm filtered ballast water to wash the sample down into the cod end assembly, and subsequently into a 1 L Nalgene plastic wide-mouth bottle. Environmental variables of the ballast water discharge (temperature, pH, conductivity, and turbidity) were measured using a YSI EXO2 water quality sonde (YSI Incorporated, Yellow Springs, OH) submerged in the bin while sampling. A YSI CHLORINE-900 Colorimeter was used to measure TRO following the vendor-supplied manual prior to leaving the vessel. All sample bottles were kept in a dark, insulated container with ice packs wrapped in bubble wrap (samples kept near or below ambient ballast water temperature) and analyzed within 3–4 h.

2.2. Field counts

Immediately after departing each ship, ballast water samples were transported to a lab setting where each sample was concentrated further using a 35 μm (50 μm in diagonal) mesh sieve for immediate analysis under a microscope. During 2018, samples were also split for analysis by a second method (based on cellular adenosine triphosphate (ATP), results not shown here). Depending on the abundance of live plankton and debris in each concentrated sample, aliquots of up to 2.5 mL were transferred into single channels of a Modified Bogorov counting chamber (Hydro-Bios Apparatebau GmbH, Germany) by Eppendorf pipette for assessment under a Nikon SMz800N Zoom stereoscope under 30–80× magnification (aim of ~25 individuals per channel). The opening of the pipette tip was trimmed to a 5 mm opening to allow for capture and transfer of larger organisms. Prior to loading the channel, a 10 μL aliquot of 50 μm dyed green aqueous fluorescent microspheres (Fluoro-Max; ThermoScientific, Waltham, USA), suspended in double distilled water at 100 mg/mL, was added as a size reference. Live zooplankton ≥50 μm in minimum dimension were enumerated in each aliquot using standard movement/response to stimuli techniques (NSF International, 2010). Multiple aliquots of concentrated ballast water were counted until a time limit of 60–90 min per net sample was reached (to minimize effects/mortality of plankton due to storage and handling). Typically, a volume of at least 1 m3 in total was assessed, though volumes were much smaller when concentrations were high (i.e., <300 mL assessed when concentrations were >500 individuals per m3; Table 1). Counts were converted to ambient concentrations (i.e., number per m3 within ships’ ballast water) based on volumes collected and fraction of sample counted.

Table 1.

Characteristics of the ballast water samples assessed, including year and location of collection, management history - including use of ballast water exchange (BWE) and ballast water management system (BWMS) model, disinfection type and filter mesh size - and total residual oxidant (TRO) level, sample volume, volume counted, time until live analysis completion, numeric count (live individuals per cubic meter) and outcome in comparison to Regulation D-2. Treatment Types: CH = chemical injection; EC = electrochlorination, electrolysis or electrodialysis; OZ = ozonation; UV = ultra violet. Result CTL = close-to-limit, where the confidence intervals of the count span above and below the D-2 standard.

Year Port BWE BWMS model Type Filter (μm) TRO (mg/L) Sample vol. (L) Count vol. (L) Count time (h) Count (ind/m3) Result
2017 Vancouver No Panasia GloEn Patrol P1200 UV 50 <0.02 2989 2989 5.4 2 Pass
2017 Vancouver No JFE BallastAce-500 CH 50 <0.02 3000 3000 5.4 0 Pass
2017 Vancouver Yes NK-O3 BlueBallast System NK-03-030 OZ n/a <0.02 3000 3000 5.1 5 Pass
2017 Sorel No Alfa Laval Pureballast 3.0 UV 20 <0.02 3000 3000 4.0 0 Pass
2017 Sorel No RWO Clean Ballast 1500 EC 55 <0.02 1156 85 4.8 3557 Fail
2017 Hamilton No JFE BallastAce 1500 CH 50 0.11 3000 294 6.5 1052 Faila
2017 Saint John No Techcross Electro-Cleen ECS 600 EC n/a <0.02 3000 1258 7.7 238 Fail
2017 Halifax No Panasia GloEn-Patrol P500 UV 50 <0.02 2589 1073 5.5 362 Fail
2018 Vancouver Yes OptiMarin BWTS 500/600BK UV 40 <0.02 2274 315 5.8 43 Faila
2018 Vancouver Yes Techcross Electro-Cleen ECS-450B EC n/a 2.9 3000 2001 4.0 0 Pass
2018 Vancouver Yes Headway Ocean Guard HMT 1500E EC 50 0.21 700 8 3.9 3753 Fail
2018 Vancouver Yes Miura HK-300 UB UV 50 <0.02 1630 723 5.2 19 Fail
2018 Vancouver No OceanSaver AS MKII EC 40 0.03 900 600 3.0 0 Pass
2018 Vancouver Yes Techcross Electro-Cleen ECS-450A EC n/a 7.66 3000 1304 5.8 0 Pass
2018 Vancouver No Miura HK HK-300 UV 50 0.1 3000 246 6.1 929 Fail
2018 Vancouver Yes Panasia GloEn-Patrol P1500 EX UV 50 <0.02 3000 348 5.5 223 Fail
2018 Vancouver No Miura HK-900 UV 50 <0.02 3000 942 5.8 101 Fail
2018 Vancouver No Alfa Laval Pureballast 2.0 UV 40 <0.02 3000 1712 5.2 0 Pass
2018 Vancouver Yes Alfa Laval Pureballast 3.0 UV 50 0.04 3000 1963 5.9 7 CTL
2018 Vancouver No Alfa Laval Pure Ballast 3.1 EX UV 20 0.04 3000 1994 4.4 0 Pass
2018 Vancouver No Techcross Electro-Cleen ECS-XXX EC n/a 0.1 3000 1268 5.4 2 Pass
2018 Vancouver Yes Techcross Electro-Cleen ECS-450A EC n/a 0.3 3000 1624 3.9 0 Pass
2018 Vancouver Yes Alfa Laval Pureballast 3.0 UV 50 0.05 3000 1325 5.1 22 Fail
2018 Vancouver Yes OptiMarin BWTS 1167/1200BK UV 40 0.05 2712 1803 6.1 32 Faila
2018 Vancouver No NK-O3 BlueBallast System NK-03-075 OZ n/a 0.02 3000 94 5.6 3822 Fail
2018 Vancouver Yes Hyundai HiBallast system HiB-600 EC n/a 0.05 2478 137 6.4 664 Fail
2018 Vancouver No JFE BallastAce-2640 CH 50 0.04 3000 1998 5.4 8 CTL
2018 Vancouver No Techcross Electro-Cleen EX-ECS-1000B EC n/a 0.68 3000 1893 4.9 3 Pass
2018 Vancouver Yes Alfa Laval Pure Ballast 3.1 UV 20 0.02 1734 951 4.8 15 CTL
Open in a new tab
a

Denotes BWMS operational issue noted by sampling personnel.

During field counts, the analyst took notes of types of taxa observed as live (e.g., copepods, nauplii, rotifers, polychaetes, etc.) and recorded the numbers of individuals counted within those broad groups (except for three ships, for which only total numbers were recorded). These counts were used to calculate an average concentration for each of the broad taxonomic groups for each ship (averaging across net samples). For more detailed taxonomy of organisms, when time allowed, live individuals (~10 per distinct type) were isolated by pipette and placed in 20 mL scintillation vials for later analysis by morphology (e.g., mature copepods and rotifers) or DNA barcoding using the mitochondrial cytochrome C oxidase subunit I (COI) locus (e.g., juveniles, such as copepod nauplii).

In addition, high-throughput sequence (HTS) metabarcoding based on 18S small subunit rRNA was conducted on bulk ballast water samples preserved in 95 % ethanol for a subset of ships (n = 17) to supplement live taxonomy results since it was not feasible to isolate all live individuals during field assessments. These HTS samples consisted mainly of uncounted sample portions representing approximately 20 to 90 % of the initial sample volume (380 to 2000 L); for ten of the HTS samples the uncounted sample portion was recombined with the counted sample portion less any live individuals already picked out. These samples likely contain a mix of live and dead individuals and HTS based on 18S captures a very broad spectrum of metazoan diversity, including but not exclusive to zooplankton in the >50 μm size class. Furthermore, given variation in amplification efficiency and biomass, HTS data cannot be translated directly into numerical counts of individual organisms (Piñol et al., 2019). Therefore, species lists resulting from metabarcoding were examined only qualitatively, with the goal of providing another assessment of the overall biodiversity present in treated ballast water rather than providing independent determination of compliance with the discharge standard.

2.3. Statistical analysis

Counts from the three continuous net samples were summed for each ship sampling event and confidence intervals (CI) were calculated assuming a Poisson distribution (as recommended by NSF International, 2010). Locations of last ballast water uptake were mapped using the WGS 1984 Mercator Pacific Ocean projection within ArcGIS Pro 2.7.2 (Environmental Systems Research Institute (ESRI), 2020) and partitioned into ballast water source regions based on the Food and Agriculture Organization of the United Nations (FAO) major fisheries oceanic regions (FAO, 2020). Data points from three non-compliant ship sampling events were excluded from statistical analysis as these ‘outlier’ points could be explained by BWMS maintenance issues (where treatment was not fully applied – see Discussion).

The variation in concentration of zooplankton (cumulative number of live individuals per m3) across ship samples was evaluated against two discrete probability distributions (i.e., negative binomial or Poisson distribution), with the most suitable distribution selected using the Akaike’s information criterion (AIC). Then, a multi-model inference (Calcagno and de Mazancourt, 2010) was run to test all possible model combinations considering potential explanatory or predictor variables for zooplankton concentration using the negative binomial function and model averaging under the GLMulti package using R (R Core Team, 2021). Variables included in the multi-model inference were related to ballast water history (ballast water source region, ballast water salinity, ballast water age (in days)), management actions (BWE + BMWS or BWMS alone, BWMS filtration (presence/absence), BWMS filter mesh size (when present), and BWMS disinfection type (chemical or UV)). In addition, variables were included for sampling methods that could potentially impact zooplankton estimates through artefacts such as holding effects (sample collection flow rate, sampling duration and total volume sampled) (Supplementary data S1). The Akaike Information Criterion corrected for small sample size (AICc) was used to determine the model that best explains the variability of the data across ballast water samples.

The relationships between the predictor variables included in the resulting best fit model and the response variable (cumulative number of live individuals per m3) were subsequently examined using a Generalized Linear Model (GLM). As the negative binomial distribution is more suitable to account for over-dispersion of count data (Lindén and Mäntyniemi, 2011), the GLM was applied using a negative binomial regression model to assess correlations between zooplankton concentration and the selected variables: BWMS disinfection type (chemical or UV), BWMS filter mesh size (when present), sampling duration and total volume sampled. The selected variables were also assessed for interactions generated by the GLM.

Finally, visual assessments were conducted using the R package “pairedData” to examine differences in flow rate, sampling duration, total volume sampled and zooplankton concentration for each net (within ships) followed by Wilcoxon signed rank tests between pairs of data for each factor. All the statistical analyses were performed using R software programming (4.1 version, R Core Team, 2021) and α = 0.05 was used to define statistical significance. P-values were adjusted using the Bonferroni correction method, in which p-values are multiplied by the number of comparisons (Wright, 1992).

2.4. Taxonomic identification

As live individuals must be enumerated within hours of collection to avoid mortality caused by holding effects, taxonomy was limited to coarse assessment (typically no finer than Order) during field counts using low magnification stereoscope microscopy. Mature live individuals isolated from ballast water samples were sent to taxonomic experts for morphological identification to the lowest feasible taxonomic level using microscopy (see Acknowledgements). Juvenile live individuals isolated for taxonomy were transferred into individual wells of 96-well plates and sent for DNA barcoding based on COI gene using methods adapted from (Mychek-Londer et al., 2020).

Briefly, following an ethanol evaporation step, DNA was extracted using a Tecan Freedom EVO 150 Liquid Handling System, a carboxylate magnetic bead-based protocol, three ethanol washes, transferred into TE buffer and stored at −20 °C. The 25 μL PCR mix consisted of 2.5 μL 10× Taq buffer, 2.5 μL MgSO4, 0.5 μL dNTPs, 0.5 μL of each 10 μM COI primer pair (Folmer et al., 1994), 0.1 μL Taq polymerase, 5 μL sample DNA, and ddH2O. Thermal cycling consisted of a 1 min initial cycle at 94 °C, followed by 40 cycles of 94 °C (40 s), 45 °C (40 s) and 72 °C (40 s), and a final extension of 72 °C. Amplification was tested on 2 % agarose gel. PCR products producing solid bands were cleaned using a magnetic bead protocol, re-tested on 2 % agarose gel, with those still containing strong bands cleaned and sent for forward and backward Sanger sequencing. Sequences were BLASTed against the entire NCBI database (National Center for Biotechnology Information, Bethesda MD, USA) with the standard conditions. Identity matches >97 % are reported below.

Residual (uncounted) portions of ballast water samples were processed for HTS metabarcoding of 18S small subunit (SSU) rRNA following methods described in Darling et al. (2018). Briefly, samples were filtered and phenol-chloroform extracted and a fragment amplified using primers SSU_F04 (GCTTGTAAAGATTAAGCC) and SSU_R22 (GCCTGCTGCCTTCCTTGGA) as described by Blaxter et al. (1998). Subsequent cleaning and preparation of amplicons for dual-indexing PCR and MiSeq sequencing was conducted according to standard protocols (Darling et al., 2018). Raw Illumina sequences were trimmed of primers, merged, and phiX contamination removed. A standard length of 350 bps was selected after evaluation of length vs. expected error rates, including only full length sequences with ≤1 expected errors. This subset was dereplicated and unique sequences were used to generate OTUs after clustering at 97 % identity. Merged reads were mapped to OTU sequences to produce a final OTU table, which was blasted against a local copy of the NCBI nucleotide database for taxonomic assignment using the Ribosomal Database Project (RDP) classifier (v2.2) as implemented in QIIME (v.1.9.1). Taxa were included in final tables only if species level assignment received an RDP classifier score (an estimate of confidence in assignment accuracy) of ≥0.97. In order to further facilitate comparisons between approaches, taxa not considered by morphological analysis (e.g., fungi, algae, protists, and other phytoplankton) were removed from the final HTS dataset.

The geographic distribution of each species identified (by any method) was explored to determine its population status, relative to the ballast water receiving port, as nonindigenous, indigenous or cryptogenic through review of data in the World Register of Marine Species (WoRMs - https://www.marinespecies.org/), Marine Planktonic Copepods (https://copepodes.obs-banyuls.fr/en/), AquaNIS (http://www.corpi.ku.lt/databases/index.php/aquanis), the Invasive Species Compendium (https://www.cabi.org/ISC), the Global Biodiversity Information Facility (GBIF - https://www.gbif.org/), and additional sources as required.

3. Results

3.1. Sample population

Treated ballast water samples were collected from 29 ships arriving to Canadian ports during May–Sept 2017 and July–Oct 2018, with the majority sampled in Vancouver, BC (see Table 1). Thirteen ships had used ballast water exchange in combination with the BWMS (BWE + BWMS), loading oceanic ballast water in FAO regions 61, 67, and 77 prior to arrival in Canada (Table 1; Supplementary Fig. S2). At the time of sampling, ballast water age ranged from 5 to 32 days (median 11 days). The sixteen remaining ships had used BWMS alone, loading ballast water at coastal or inland ports in Asia, North America and Europe (FAO regions 21, 27, 31, 61, 71, and 77; Supplementary Fig. S2). At the time of sampling, ballast water age ranged from 1 to 82 days (median 16.5 days). Sampled ships were equipped with 14 different models of BWMS, which primarily used electrochlorination or UV as the disinfection method (Table 1). The majority of BWMS also used filtration prior to disinfection, with nominal mesh sizes ranging from 20 to 55 μm (Table 1). While BWMS using an active substance for disinfection also had the ability to apply neutralization prior to discharge, half of all ballast water discharge samples exceeded the maximum allowable TRO level specified during IMO Type Approval. Though exceedances were usually small, in one case ballast water was discharged having full strength chlorination (8 mg/L; Table 1).

3.2. Field counts

Considering confidence intervals of field counts, twelve samples had concentrations of live zooplankton below the D-2 standard, including seven samples where no live individuals were observed (Fig. 1; Table 1). Fourteen samples had concentrations of live organisms clearly exceeding the standard, while three samples had narrow CIs spanning the D-2 standard (called ‘close-to-limit’) (Fig. 1; Table 1). The samples with clear exceedances were highly skewed, with confidence intervals ranging from 11 to 5090 (median 292) live zooplankton per m3. Nine samples (31 %) had concentrations at least 5× the D-2 standard according to the lower confidence intervals.

Fig. 1.

Fig. 1.

Open in a new tab

Estimates of zooplankton concentration (individuals ≥ 50 μm in minimum dimension) recorded during live field counts. Panel A) shows the frequency of estimates (arithmetic mean) on a logarithmic scale, with the first two bars being lower than the limit in Regulation D-2. Panel B) shows the cumulative estimate of the number of live individuals per m3 with lower and upper confidence intervals (error bars) for each ship sampled.

3.3. Statistical results

Based on AIC values, the negative binomial distribution was the most suitable distribution to represent the concentration of zooplankton (live individuals per m3) across samples for all ships (estimates were size = 0.1, mu = 616.4 with AIC = 256.8 for negative binomial compared to lambda = 616.7 and AIC = 47,204.2 for Poisson). The multi-model inference analysis produced up to 5600 models considering different combinations of possible explanatory variables. Four variables (BWMS disinfection type, BWMS filter mesh size, sampling duration and total volume sampled) were retained in the final model based on the AICc (=169.3), with BWMS filter mesh size and sampling duration showing a relative importance >0.6 across samples (Fig. 2).

Fig. 2.

Fig. 2.

Open in a new tab

Model-averaged importance of main factors (i.e., sampling duration - SpD, mesh, disinfection and total zooplankton volume - TZV) affecting zooplankton concentration.

The GLM showed that zooplankton concentration was positively related to BWMS filter mesh size (Estimate = 1. 2; Std. Error = 0. 3; z-value 4.3; p < 0.001) and inversely related to sampling duration (Estimate = 43.6; Std. Error = 12; z-value = 3.7; p < 0.001). Additionally, the GLM showed that the disinfection type used influenced the response variable (Estimate = −2710; Std. Error = 1200; z-value = −2.2; p < 0.01) and the total volume sampled (Estimate = −0.3; Std. Error = −0.05; z-value = −4.1; p < 0.001) was inversely related to the response variable, both interacting with sampling duration. Factors such as ballast water age, BWMS filtration (presence/absence) or BWMS management action (BWE + BWMS or BWMS alone) showed no effect on the response variable.

The Wilcoxon test for non-parametric data showed that estimates of zooplankton concentration from Net 1 were significantly higher than from Net 2 (v = 9819; p-value < 0.0001) and Net 3 (v = 9414; p-value < 0.0001). The analysis did not show any significant difference in the concentration of zooplankton between Net 2 and Net 3 (v = 4617; p-value = 0.23). Similarly, the flow rate for Net 1 was significantly higher than for Net 2 (v = 6565.5; p-value < 0.0001) and Net 3 (v = 4815; p-value < 0.0001). No difference was found in flow rates for Net 2 and Net 3. Likewise, no differences were found for sampling duration nor total volume sampled between nets (Fig. 3).

Fig. 3.

Fig. 3.

Open in a new tab

Estimates of zooplankton concentration (individuals ≥ 50 μm in minimum dimension) recorded during field counts in relation to A) sampling duration time (hours); B) mesh size of filters (μm), when present; and C) total volume sampled (m3); D) BWMS disinfection type (chemical or UV) as boxplots showing median (bold line), third quartile, and outliers (dots).

3.4. Taxonomic information

Broad taxonomy was recorded for 19 of the 29 samples during field counts, as no live individuals were observed in seven samples, and only total counts (no taxonomic info) were recorded for three samples. At least 13 unique taxa were recorded, with Arthropoda (primarily Copepoda) being most commonly observed followed by Annelida and Mollusca (occurring in 95 %, 32 % and 32 % of the 19 samples assessed, respectively; Table 2; Supplementary Table S3). Median concentrations of each taxonomic group, when present, were highly variable ranging from <1 to 61 individuals per m3. Nematoda, Mollusca and Ostracoda had highest median concentrations, while Copepoda, Rotifera and Annelida had highest maximum concentrations (Table 2). Approximately 2 % of individuals enumerated in the 19 samples assessed were not assigned to any broad taxonomic group (excluding the >800 individuals enumerated in the three samples with total counts only).

Table 2.

The rate of occurrence, median and maximum concentration of live zooplankton (individuals per m3) in treated ballast water samples, broadly identified to Phylum (or lower taxonomic level, where feasible) during field counts (n = 19). Median concentration is calculated only considering samples where taxon was present.

Taxonomic group Occurrence (%) Median concentration Maximum concentration
Arthropoda 95 26 3007
 Copepoda 95 26 3007
 Ostracoda 6 46 89
 Decapoda 5 2 2
 Amphipoda 5 1 1
 Diplostraca 5 1 1
 Cirripedia 5 1 1
Annelida 32 18 245
Mollusca 32 47 121
Ciliphora 16 13 115
Cnidaria 16 3 16
Rotifera 11 13 2663
Nematoda 11 61 121
Tunicata 5 2 2
Unidentified 37 2 242
Total 100 29 3753
Open in a new tab

Live individuals were isolated during field counts from 16 samples (all from ships arriving to the Pacific Coast) for detailed taxonomy. Only 48 of 360 (13 %) individuals isolated for morphological taxonomy could be determined to the species level while 31 of 191 (16 %) individuals isolated for DNA barcoding received confident species-level designations. In total, at least 38 unique taxa (24 to species level) were identified (Table 3; Supplementary Table S3). At least seven species appear to be nonindigenous to British Columbia, including two considered already established (Asian Date Mussel, Arcuatula senhousia and Pacific Oyster, Magallana gigas) and five copepods with unknown population status (Calanus sinicus, Hemicyclops ctenidis, H. japonicas, Ditrichocorycaeus erythraeus, Euterpina acutifrons). Additional taxa may be nonindigenous but were not identified to the species level (e.g., Pseudodiaptomus sp.) or there is a paucity of records to determine native range (e.g., Cyclops kikuchii). While most of these taxa were sampled from ballast water that did not meet the D-2 standard, D. erythraeus was found in ballast water samples that met the standard (having 2.6 individuals per m3).

Table 3.

List of taxa collected as live individuals in treated ballast samples and identified using traditional morphological taxonomy (T) or DNA barcoding (D), with occurrence rate in the 16 samples assessed, status in coastal waters of British Columbia (IN - indigenous, NIS - nonindigenous, EST - NIS with established populations or UK - unknown) and typical habitat (M - marine, B - brackish and/ or F - fresh). Population status and typical habitat were ascertained mainly using the World Register of Marine Species (WoRMS), AquaNIS, and the Invasive Species Compendium (CABI).

Taxonomic group Species No. samples Status Habitat ID method
Crustacea
 Calanoida Acartia longiremis 1 IN M T
Acartia spp. 3 – – T
Calanus pacificus 1 IN M D
Calanus sinicus 1 NIS M T
Pseudocalanus minutus 1 IN M T
Pseudodiaptomus sp. 1 – – T
Microcalanus pusillus 1 IN M T
Microcalanus sp. 1 T
Metridia sp. 1 – – T
Paracalanus parvus 1 IN B/M D
Paracalanus spp. 3 T
Calanoida indet. 4 – – T
 Cyclopoida Cyclops kikuchii 1 UK F D
Hemicyclops ctenidis 1 NIS M D
Hemicyclops 2 NIS M T
japonicus Oithona setigera 2 IN M T
Oithona spp. 2 – – T
Oithonidae indet. 2 – – T
Ditrichocorycaeus affinis 1 IN M T
Ditrichocorycaeus anglicus 1 IN B/M T
Ditrichocorycaeus erythraeus 1 NIS M T
Lubbockia sp. 1 T
Clausidiidae indet. 1 – – T
Corycaeidae indet. 3 – – T
Oncaeidae indet. 2 – – T
Cyclopoida indet. 3 – – T
 Harpacticoida Euterpina acutifrons 1 NIS B/M T
Microsetella norvegica 1 IN B/M T
Ectinosomatidae indet. 1 T
Ameiridae indet. 1 T
Miraciidae indet. 1 – – T
Paramesochridae indet. 2 – – T
Tisbidae indet. 2 T
Harpacticoida indet. 5 – – T
Copepoda indet. 2 – – T
 Polyarthra Longipedia sp. 1 – – T
 Amphipoda Themisto pacifica 1 IN M T
Themisto sp. 1 – – T
Hyperiidae indet. 1 – – T
 Decapoda Crangonidae indet. 1 – – T
 Balanomorpha Balanus crenatus 6 IN T,D
Balanomorpha indet. 3 T
Cnidaria
 Actiniaria Actiniaria indet. 1 – – T
Mollusca Gastropoda indet. 2 – – T
Bivalvia Mytilus trossulus 1 IN M D
Arcuatula senhousia 1 EST M D
Magallana gigas 1 EST M D
Mytilidae indet. 1 – – T
Bivalvia indet. 1 – – T
Nematoda Nematoda indet. 2 – – T
Annelida
 Polychaeta Micronephthys cornuta 4 IN M D
Micronereis nanaimoensis 1 IN M D
Grubeopolynoe tuta 1 IN M D
Harmothoe extenuata 1 IN M D
Sabellariidae indet. 1 – – T
Polychaeta indet. 1 – – T
Rotifera
 Ploima Gastropodidae indet. 1 – – T
 Flosculariaceae Pompholyx sp. 1 – – T
Open in a new tab

Eighteen bulk ballast water samples were analyzed by HTS metabarcoding, including 13 that had live individuals removed for detailed taxonomy as described above, four that had high concentration of live individuals and were only sent for HTS, and three that were sent for HTS as ‘negative controls’ where live individuals were not observed during field counts (or the very few live individuals had been removed from the samples and sent for detailed taxonomy). Although over 900 OTUs were captured overall from bulk water samples by HTS metabarcoding, after removing taxonomic groups not considered in morphological analysis (e.g., plants, fungi, dinoflagellates, etc.) and filtering for only high-confidence taxonomic assignments, 55 species were identified, including species-level identifications for taxa that were not identified to species level by morphology or direct DNA barcoding (e.g., Tunicata, Diplostraca and Cnidaria; Supplementary Table S3). While at least 30 of the species identified are considered nonindigenous to the recipient environment, only one of these (A. senhousia) was confirmed as being live in collected samples (live individuals isolated and identified by DNA barcoding). Further, for the 13 samples which had a portion of live individuals isolated for detailed taxonomy with the remainder sent for HTS metabarcoding, there was very little overlap in the species identified (a total of 3 species-level matches between the methods; 1 with morphology and 2 with DNA barcoding). Finally, there were 22 species-level identifications reported from the ‘negative controls’ by HTS metabarcoding.

4. Discussion

This first assessment of new ballast water management practices provides essential data needed to understand potential changes in the probability of introduction of aquatic nonindigenous species globally. In this study, half (48 %) of managed ballast water samples showed clear exceedances of the D-2 standard and half (52 %) were below or ‘close to’ the discharge limit (having lower confidence interval below and upper confidence above the limit). Although three samples assessed were found to have relatively high concentrations (around 3000 individuals per m3) in comparison to the D-2 standard, the median and maximum concentrations reported (15 and 3822 individuals per m3, respectively) are much lower than reported from ships sampled in Vancouver during 2006–2008 following ballast water exchange (median and maximum values: 1368 and 63,562 individuals per m3, respectively) (DiBacco et al., 2012). Although differences in sample size, sample collection and analysis methods between these two studies limit direct comparison, these results indicate that the frequency of high risk (high propagule pressure) introductions may be dramatically reduced using BWMS even though the targeted discharge concentrations were often exceeded.

Both BWMS filter mesh size and sampling duration were found to be important factors affecting zooplankton concentration, with the former having the greatest influence. Based on the GLM, zooplankton concentration increases as BWMS mesh size increases from 20 μm to 55 μm (with sampling duration held constant), however, this may be an artefact of sample size since only seven samples were collected following use of BWMS with filter mesh size lower than 50 μm. In contrast, zooplankton concentration decreased with longer sampling duration (with filter mesh size held constant at 50 μm), though the magnitude of change was small. Sampling duration (and total volume sampled) may influence zooplankton estimates through holding effects, where mortality may increase as organisms are held in small containers due to predation, deoxygenation or similar containment stresses. Samples managed by chemical treatment had significantly higher zooplankton concentration than samples managed by UV treatment. This might be explained by the absence of filtration, with 60 % of chemical systems having no filter step, although presence of BWMS filtration was not retained in the best-fit model as a predictive variable. It has been reported that efficacy of UV systems is dependent on optimal water quality conditions (e.g., low turbidity) while chemical systems are considered more broadly applicable irrespective of water quality (Tsolaki and Diamadopoulos, 2010), thus it could be important for future studies of BWMS efficacy to account for differences in uptake water quality (e.g., concentration of live organisms, total suspended solids, UV transparency, etc.). With the relatively small sample size, it is not possible to definitively determine the causative factors leading to high zooplankton concentration in this study.

There are multiple reasons why a ballast sample may exceed the discharge standard, some of which may not be a direct indication of the BWMS efficacy. For example, the BWMS could be installed, operated or maintained incorrectly, particularly during this early stage of implementation of the Convention. During this study, operational issues were documented for three (10 %) tests. In one case, the crew were waiting for a service call from the BWMS manufacturer as two of the UV lamps were not reaching the required intensity (limiting the UV dose). Similarly, in the second case, the BWMS self-monitoring system was giving an alarm related to low UV lamp intensity. The samples from these ships were marginally above the discharge standard (32–43 individuals per m3). In the third case, after learning the results of this testing (1052 individuals per m3), a root cause analysis conducted by the shipowner discovered that the BWMS operating system was due for a software update – without the update, an insufficient level of chlorine was being applied during treatment. Such issues could be avoided in the future with more frequent or formal maintenance schedules (and possibly more comprehensive training for ship crew). As uptake concentrations of zooplankton were unknown and it was not possible to determine causes for all exceedances observed, one should not equate the proportion of samples meeting the D-2 standard in this study as any measure of BWMS efficacy. It will be important for future studies to collect paired measurements during ballast water uptake and discharge to directly assess BWMS efficacy.

Ballast water testing outcomes may also be influenced by contamination of treated ballast water with untreated seawater (due to leaky valves or entrapment in ballast piping) or mortality during sample collection and handling. While such issues were not noted during this study, there were two tests during which the sampling team suspected the sample probes were absent or installed facing downstream rather than facing into the ballast discharge. In both cases, the sampling result was an exceedance of the discharge standard; these might be considered definitive failures given that such installations are likely to bias the outcome towards an underestimate (Richard et al., 2008). During another test, the valve for the sample port could not be fully opened, as is recommended for representative sampling to minimize turbulence and mortality. As this sample also exceeded the standard, any valve effects (additional mortality) did not change the outcome. It has been noted elsewhere that sample ports are rarely installed following recommendations to achieve representative sampling and that additional research is needed to understand how such deviations might impact measurement outcomes (Drake et al., 2021).

Although Regulation D-2 sets limits on the total number of viable organisms in ballast discharge, it is important to consider the diversity of taxa (i.e., colonization pressure) and potential for selection of tolerant taxa to comprehensively assess the risk of treated ballast water (Briski et al., 2018). The 30 live individuals observed in samples meeting the D-2 standard included cyclopoid, calanoid and harpacticoid copepods, amphipods (such as Themisto pacifica), barnacles (such as Balanus crenatus), rotifers, ostracods, and unidentified protist-like organisms. The samples with higher concentrations showed greater diversity, noting that 85 % of the 551 individuals isolated for detailed taxonomy could not be identified to the species level. At least 63 % of these individuals were juvenile life stages, showing the great need to advance reference libraries supporting molecular identification methods, particularly for harpacticoid copepods. Due to changes in ballast water sampling and analysis methods for assessment of treated ballast water, it is difficult to determine if the use of BWMS has resulted in any directional change of the types of taxa being discharged. The most recent study of zooplankton in ballast water being discharged at Canadian Pacific ports (prior to use of BWMS) reported 176 taxa, including taxonomic groups not observed during this study (i.e., Cumacea, Mysida, Euphausiacea, and Isopoda (DiBacco et al., 2012); however, as these taxa were infrequently observed, their absence in this study may be the result of small sample size (i.e., fewer ships sampled).

The high proportion of juvenile taxa reported here suggests that the use of BWMS may select for immature life stages although it is not possible to quantitatively compare the proportion of juveniles in samples since the earlier study (DiBacco et al., 2012) used a larger mesh size during sample collection (125 μm vs. 35 μm used in this study). Size selection could also result from the use of BWMS filtration, which may more effectively remove larger, mature individuals. Alternatively, the large proportion of juveniles could be an indicator of eggs hatching within ballast tanks. The efficacy of BWMS for removal of inactive life stages, such as diapausing eggs, from ballast water is unknown. One third of samples assessed in this study contained relatively high levels of fine sediment, indicating that sediment (and associated egg banks) could accumulate in ballast tanks even with the use of BWMS. Although immature life stages may pose lower risk for population establishment than (reproductive) adults, the accumulation of diapausing stages able to withstand harsh environments in ballast tank sediments is known to be a mechanism for transport and introduction of aquatic invertebrates (Bailey et al., 2003; Branstrator et al., 2015; Kipp et al., 2010).

Diversity assessments using HTS metabarcoding revealed poor overlap with detailed assessments. Samples provided for HTS analysis were not split in a representative way, limiting direct comparison of results. Nevertheless, the striking discrepancy between HTS and morphological assessments likely results from multiple factors, including the increased sensitivity and broad taxonomic coverage of HTS, limitations of reference databases for taxonomic assignment, as well as the probability that HTS is capturing components of overall biodiversity that are inaccessible to morphological investigation, including early life stages, organismal fragments, remnants of dead organisms, or even extra-organismal DNA. Thus, while the diversity revealed by HTS provides yet another picture of the communities present in ballast tanks, it is important to recognize that these methods are currently not suitable for assessments in regulatory contexts (Darling and Frederick, 2018).

The difficulty of determining viability status of taxa identified via HTS is widely recognized (Pochon et al., 2017). This is most clearly reflected in samples identified here as ‘negative controls’, for which no individuals were observed as live during field counts (with 80 % of the sample having been assessed) but 10 species were identified using HTS. Similar results were obtained for the two samples where all live individuals were removed and the sample remainder sent for HTS. While it is possible live individuals were missed in these samples during field counts, it is more likely that HTS is capturing biodiversity components such as dead organisms, organismal fragments, and/or extra-organismal DNA. Amplifiable DNA may remain in aquatic environments well after death of the target organism, depending on physical, chemical, and biological characteristics of that environment (Stewart, 2019). Future examination of alternative molecular targets, in particular RNA (which degrades more rapidly than DNA), may better reflect the viable component of ballast water diversity (Cristescu, 2019; Wood et al., 2020).

It is also known that HTS primers amplifying the 18S locus capture a broader subset of biodiversity, including fungi and phytoplankton, that are unlikely to be identified by morphological approaches (Clarke et al., 2017). In this study, over 900 OTUs were originally identified by HTS metabarcoding. However, the majority of these OTUs either received taxonomic assignments with relatively low confidence (in our analysis, those with an RDP classifier score < 0.97) or were representatives of taxonomic groups not included in morphological analysis. Such taxa may be of general interest for future explorations of ballast water diversity, particularly those examining the 10 to 50 μm size class. Surprisingly, many taxa identified by morphology were not captured by HTS. While errors in morphological identification are possible, it is more likely that the discrepancies are due to failures of HTS to accurately identify species due to limitations of reference databases or errors in bioinformatic workflows. These issues are widely recognized and future improvements to databases and assignment algorithms are expected to increase the general utility of HTS for diversity assessments (Steyaert et al., 2020). Similarly, several taxa identified by direct COI barcoding did not appear in HTS taxon lists, reflecting the fact that different loci often capture very different components of sample diversity (Clarke et al., 2017; Piñol et al., 2019), suggesting that HTS analysis targeting multiple loci may provide more comprehensive assessments of total diversity.

This evaluation of live organisms ≥50 μm in minimum dimension (typically zooplankton) in treated ballast water samples provides a first insight into the potential risk reduction that may be achieved using BWMS. While a large proportion of samples were found to exceed the international performance standard under Regulation D-2, the findings are not surprising given approximately 21 % of treated ballast water samples were found to exceed the D-2 standard during commissioning testing of BWMS at the time of initial installation (SGS Global Marine Services, 2020). Promisingly, even though the performance standard was often exceeded, the frequency of high risk (high propagule pressure) introductions appears to be dramatically reduced with the use of BWMS. As this research was conducted during the very early days of BWMS use and entry-into-force of the Convention, one can be hopeful that even greater rates of compliance will be achieved as shipowners and crews move beyond the steep learning curve associated with selection, operation and maintenance of these new and complex technologies.

Supplementary Material

Supplement1
Supplement2
Supplement3

Acknowledgements

Thanks to A. Burt and S. Campbell for participating in field work, C. Remillard for assistance with mapping, and A. Rushwan for assistance gaining access to ships in Vancouver. We thank the ship crews that participated in this research as well as their respective owners and agents. Morphological taxonomy was conducted by Biologica Environmental Services Ltd., Victoria, BC, while COI DNA extraction and sequencing were conducted by the Great Lakes Institute for Environmental Research (University of Windsor, Windsor, ON) and Genome Quebec, respectively. High throughput sequencing was conducted by US EPA; while this research was funded in part by that Agency, the views expressed in this manuscript are those of the authors and do not necessarily reflect the views or policies of the EPA. J. Cordell and C. DiBacco provided information on population status of different taxa along the Pacific coast. This project benefited from discussions held during meetings of the joint Working Group on Ballast and Other Shipping Vectors (WGBOSV) of the International Council for the Exploration of the Sea (ICES), Intergovernmental Oceanographic Commission of UNESCO (IOC) and International Maritime Organization (IMO), as well as those of the US Naval Research Lab and the IMO’s Pollution, Prevention and Response Sub-Committee. Financial support was provided by Transport Canada, Fisheries and Oceans Canada and the US EPA.

Footnotes

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Sarah A. Bailey: Conceptualization, Funding acquisition, Supervision, Methodology, Investigation, Writing - original draft. Torben Brydges: Methodology, Investigation, Writing – review & editing. Oscar Casas-Monroy: Formal analysis, Visualization, Methodology, Investigation, Writing – review & editing. Jocelyn Kydd: Methodology, Investigation, Writing – review & editing. R. Dallas Linley: Methodology, Investigation, Writing – review & editing. Robin M. Rozon: Methodology, Investigation, Writing – review & editing. John A. Darling: Resources, Methodology, Investigation, Writing – review & editing.

Supplementary data to this article can be found online at https://doi.org/10.1016/j.marpolbul.2022.113947.

Data availability

Data are attached as a Supplementary File

References

  1. Bailey Sarah A., et al. , 2003. Viability of invertebrate diapausing eggs collected from residual ballast sediment. Limnol. Oceanogr. 48 (4), 1701–1710. 10.4319/lo.2003.48.4.1701. [DOI] [Google Scholar]
  2. Bailey Sarah A., et al. , 2020. Trends in the detection of aquatic non-indigenous species across global marine, estuarine and freshwater ecosystems: a 50-year perspective. Divers. Distrib. 26 (12), 1780–1797. 10.1111/ddi.13167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Blaxter Mark L., et al. , 1998. A molecular evolutionary framework for the phylum Nematoda. Nature 392 (March), 71–75. 10.1038/32160. [DOI] [PubMed] [Google Scholar]
  4. Branstrator Donn K., Westphal Kelly L., King Breana K., 2015. Analysis of invertebrate resting eggs and other biota in ballast tank sediment of domestic Great Lakes cargo ships. J. Great Lakes Res. 41 (1), 200–207. 10.1016/j.jglr.2014.11.017. [DOI] [Google Scholar]
  5. Briski Elizabeta, et al. , 2013. Taxon- and vector-specific variation in species richness and abundance during the transport stage of biological invasions. Limnol. Oceanogr. 58 (4), 1361–1372. 10.4319/lo.2013.58.4.1361. [DOI] [Google Scholar]
  6. Briski Elizabeta, et al. , 2018. Beyond propagule pressure: importance of selection during the transport stage of biological invasions. Front. Ecol. Environ. 16 (6), 345–353. 10.1002/fee.1820. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Calcagno Vincent, de Mazancourt Claire, 2010. Glmulti: an R package for easy automated model selection with (generalized) linear models. J. Stat. Softw. 34 (12), 29. 10.18637/jss.v034.i12. [DOI] [Google Scholar]
  8. Carlton James T., Geller Jonathan B., 1993. Ecological roulette: the global transport of nonindigenous marine organisms. Science 261 (5117), 78–82. http://science.sciencemag.org/content/261/5117/78.abstract. [DOI] [PubMed] [Google Scholar]
  9. Casas-Monroy Oscar, Bailey Sarah A., 2021. Do ballast water management systems reduce phytoplankton introductions to Canadian waters? Front. Mar. Sci. 8, 691723 10.3389/fmars.2021.691723. [DOI] [Google Scholar]
  10. Chan Farrah T., Briski Elizabeta, Bailey Sarah A., MacIsaac Hugh J., 2014. Richness - abundance relationships for zooplankton in ballast water: temperate versus Arctic comparisons. ICES J. Mar. Sci. 71 (7), 1876–1884. 10.1093/icesjms/fsu020. [DOI] [Google Scholar]
  11. Clarke Laurence J., Beard Jason M., Swadling Kerrie M., Deagle Bruce E., 2017. Effect of marker choice and thermal cycling protocol on zooplankton DNA metabarcoding studies. Ecol. Evol. 7 (3), 873–883. 10.1002/ece3.2667. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cordell Jeffery R., et al. , 2008. Factors influencing densities of non-indigenous species in the ballast water of ships arriving at ports in Puget Sound, Washington, United States. Aquat. Conserv. Mar. Freshwat. Ecosyst. 19 (3), 322–334. 10.1002/aqc.986. [DOI] [Google Scholar]
  13. Cristescu Melania E., 2019. Can environmental RNA revolutionize biodiversity science? Trends Ecol. Evol. 34 (8), 694–697. 10.1016/j.tree.2019.05.003. [DOI] [PubMed] [Google Scholar]
  14. Darling John A., Frederick Raymond M., 2018. Nucleic acids-based tools for ballast water surveillance, monitoring, and research. J. Sea Res. 133, 43–52. 10.1016/j.seares.2017.02.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Darling John A., et al. , 2018. Ballast water exchange and invasion risk posed by intracoastal vessel traffic: an evaluation using high throughput sequencing. Environ. Sci. Technol. 52 (17), 9926–9936. 10.1021/acs.est.8b02108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. DiBacco Claudio, Humphrey Donald B., Nasmith Leslie E., Levings Colin D., 2012. Ballast water transport of non-indigenous zooplankton to Canadian ports. ICES J. Mar. Sci. 69 (3), 483–491. 10.1093/icesjms/fsr133. [DOI] [Google Scholar]
  17. Drake Lisa A., et al. , 2021. Design and installation of ballast water sample ports: current status and implications for assessing compliance with discharge standards. Mar. Pollut. Bull. 167, 112280 10.1016/j.marpolbul.2021.112280. [DOI] [PubMed] [Google Scholar]
  18. Environmental Systems Research Institute (ESRI), 2020. ArcGIS Pro 2.7.2 [Computer Software]. Environmental Systems Research Institute (ESRI), Redlands, CA. [Google Scholar]
  19. FAO, 2020. Major Fishing Areas. Pacific, Northwest (Major Fishing Area 61), Pacific, Northeast (Major Fishing Area 67), Pacific Western Central (Major Fishing Area 71), Pacific Eastern Central (Major Fishing Area 77), CWP data collection. In: FAO Fisheries And Aquaculture Department [fao.org]. FAO, Rome. [Google Scholar]
  20. Folmer O, et al. , 1994. DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Mol. Mar. Biol. Biotechnol. 3 (5), 294–299. [PubMed] [Google Scholar]
  21. Gollasch S, David M, 2010. Testing sample representativeness of a ballast water discharge and developing methods for indicative analysis. European Maritime Safety Agency, Lisbon, Portugal. [Google Scholar]
  22. Gregg Matthew, Rigby Geoff, Hallegraeff Gustaaf M., 2009. Review of two decades of progress in the development of management options for reducing or eradicating phytoplankton, zooplankton and bacteria in ship’s ballast water. Aquat. Invasions 4 (3), 521–565. 10.3391/ai.2009.4.3.14. [DOI] [Google Scholar]
  23. Groeneveld Rolf A., et al. , 2018. Economic impacts of marine ecological change: review and recent contributions of the VECTORS project on European marine waters. Estuar. Coast. Shelf Sci. 201, 152–163. 10.1016/j.ecss.2016.04.002. [DOI] [Google Scholar]
  24. ICES/IOC/IMO WGBOSV [International Council for Exploration of the Sea/Intergovernmental Oceanographic Commission/International Maritime Organization Working Group on Ballast and Other Ship Vectors], 2017. Standard Operating Procedures Collection of Treated Ballast Water Samples Using an Inline Sample Port. ICES, Copenhagen, Denmark. https://www.ices.dk/community/Documents/Expert%20Groups/WGBOSV/SOP_inline%20ballast%20water%20sampling.pdf. [Google Scholar]
  25. International Maritime Organization, 2004. International Convention for the Control and Management of Ships’ Ballast Water And Sediments, 2004. International Maritime Organization, London, U.K.. BWM/CONF/36, 16 February 2004. [Google Scholar]
  26. International Maritime Organization, 2008. Guidelines for Ballast Water Sampling (G2).. International Maritime Organization, London, U.K. MEPC.173(58), 10 October 2008. [Google Scholar]
  27. International Maritime Organization, 2018. Code for Approval of Ballast Water Management Systems (BWMS Code). International Maritime Organization, London, U.K. MEPC.300(72), 13 April 2018. [Google Scholar]
  28. Katsanevakis Stelios, Tempera Fernando, Teixeira Heliana, 2016. Mapping the impact of alien species on marine ecosystems: the Mediterranean Sea case study. Divers. Distrib. 22 (6), 694–707. 10.1111/ddi.12429. [DOI] [Google Scholar]
  29. Kim Eun-Chan, Jeong-Hwan Oh., Lee Seung-Guk, 2016. Consideration on the maximum allowable dosage of active substances produced by ballast water management system using electrolysis. Int. J. e-Navig. Marit. Econ. 4, 88–96. 10.1016/j.enavi.2016.06.008. [DOI] [Google Scholar]
  30. Kipp R, Bailey, Sarah A., MacIsaac HJ., Ricciardi A., 2010. Transoceanic ships as vectors for nonindigenous freshwater bryozoans. Divers. Distrib. 16 (1), 77–83. 10.1111/j.1472-4642.2009.00629.x. [DOI] [Google Scholar]
  31. Lindén Andreas, Mäntyniemi Samu, 2011. Using the negative binomial distribution to model overdispersion in ecological count data. Ecology 92 (7), 1414–1421. 10.1890/10-1831.1. [DOI] [PubMed] [Google Scholar]
  32. Lloyd’s Register, 2014. National Ballast Water Management Requirements. Lloyd’s Register, London, U.K. [Google Scholar]
  33. McCollin Tracy, Shanks Aileen M., Dunn John, 2008. Changes in zooplankton abundance and diversity after ballast water exchange in regional seas. Mar. Pollut. Bull. 56 (5), 834–844. 10.1016/j.marpolbul.2008.02.004. [DOI] [PubMed] [Google Scholar]
  34. Moser CS, First MR, Wier TP, Riley SC, Robbins-Wamsley SH, Molina V, Grant JF, Drake LA, 2018. Design and validation of a ballast water compliance sampling device for shipboard use. Manage. Biol. Invas. 9 (4), 497–504. 10.3391/mbi.2018.9.4.12. [DOI] [Google Scholar]
  35. Mychek-Londer Justin G., Chaganti Subba Rao, Heath Daniel D., 2020. Metabarcoding of native and invasive species in stomach contents of Great Lakes fishes. PLoS ONE 15 (8 August), 1–22. 10.1371/journal.pone.0236077. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. NSF International, 2010. In: Generic Protocol for the Verification of Ballast Water Treatment Technology, 600(146). U.S. Environmental Protection Agency, Washington, D.C., EPA/600/R-10/146, 2010., p. 41 https://cfpub.epa.gov/si/si_public_record_report.cfm?Lab=NRMRL&dirEntryId=230926. [Google Scholar]
  37. Piñol Josep, Senar Miquel A., Symondson William O.C., 2019. The choice of universal primers and the characteristics of the species mixture determine when DNA metabarcoding can be quantitative. Mol. Ecol. 28 (2), 407–419. 10.1111/mec.14776. [DOI] [PubMed] [Google Scholar]
  38. Pochon Xavier, et al. , 2017. Wanted dead or alive? Using metabarcoding of environmental DNA and RNA to distinguish living assemblages for biosecurity applications. PLoS ONE 12 (11), 1–19. 10.1371/journal.pone.0187636. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. R Core Team, 2021. R: A Language And Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria: https://www.R-project.org/. [Google Scholar]
  40. Richard RV, Grant JF, Lemieux EJ, 2008. Analysis of Ballast Water Sampling Port Designs Using Computational Fluid Dynamics. U.S. Coast Guard Research and Development Center, Groton, CT. https://apps.dtic.mil/sti/citations/ADA479103. [Google Scholar]
  41. Sala Enric, Knowlton Nancy, 2006. Global marine biodiversity trends. Annu. Rev. Environ. Resour. 31, 93–122. 10.1146/annurev.energy.31.020105.100235. [DOI] [Google Scholar]
  42. Seers Blake M., Shears Nick T., 2015. Spatio-temporal patterns in coastal turbidity - long-term trends and drivers of variation across an estuarine-open coast gradient. Estuar. Coast. Shelf Sci. 154, 137–151. 10.1016/j.ecss.2014.12.018. [DOI] [Google Scholar]
  43. SGS Global Marine Services, 2020. Commissioning Testing of Ballast Water Management Systems. Geneva. [Google Scholar]
  44. Simard Nathalie, Plourde Stphane, Gilbert Michel, Gollasch Stephan, 2011. Net efficacy of open ocean ballast water exchange on plankton communities. J. Plankton Res. 33 (9), 1378–1395. 10.1093/plankt/fbr038. [DOI] [Google Scholar]
  45. Stewart Kathryn A., 2019. Understanding the effects of biotic and abiotic factors on sources of aquatic environmental DNA. Biodivers. Conserv. 28 (5), 983–1001. 10.1007/s10531-019-01709-8. [DOI] [Google Scholar]
  46. Steyaert Margaux, et al. , 2020. Advances in metabarcoding techniques bring us closer to reliable monitoring of the marine benthos. J. Appl. Ecol. 57 (11), 2234–2245. 10.1111/1365-2664.13729. [DOI] [Google Scholar]
  47. Tsolaki Efi, Diamadopoulos Evan, 2010. Technologies for ballast water treatment: a review. J. Chem. Technol. Biotechnol. 85 (1), 19–32. 10.1002/jctb.2276. [DOI] [Google Scholar]
  48. Wood Susanna A., et al. , 2020. Release and degradation of environmental DNA and RNA in a marine system. Sci. Total Environ. 704, 135314 10.1016/j.scitotenv.2019.135314. [DOI] [PubMed] [Google Scholar]
  49. Wright SPaul, 1992. Adjusted P-values for simultaneous inference. Biometrics 48, 1005–1013. 10.2307/2532694. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement1
NIHMS1841017-supplement-Supplement1.pdf (3.7MB, pdf)
Supplement2
NIHMS1841017-supplement-Supplement2.xlsx (41.1KB, xlsx)
Supplement3
NIHMS1841017-supplement-Supplement3.xlsx (33.8KB, xlsx)

Data Availability Statement

Data are attached as a Supplementary File

RESOURCES