- Split View
-
Views
-
Cite
Cite
Maria Célia Villac, Irena Kaczmarska, James M. Ehrman, The diversity of diatom assemblages in ships' ballast sediments: colonization and propagule pressure on Canadian ports, Journal of Plankton Research, Volume 35, Issue 6, November/December 2013, Pages 1267–1282, https://doi.org/10.1093/plankt/fbt090
- Share Icon Share
Abstract
This research focused on the specific diversity of diatoms carried in ballast sediments of ships reaching Canadian ports on the Pacific, Atlantic and the Great Lakes during 2007 and 2009. The inventory of 180 taxa included Thalassiosira, Chaetoceros spores, Cyclotella, Actinocyclus, Aulacoseira, Melosira and Pseudo-nitzschia as the most species-rich genera found in 142 samples of the matter settled at the bottom of ship tanks. We also tested and showed evidence that diatom species composition identified in any given tank was the product of: (i) species intrinsic survival strategies, (ii) the most recent ballast water exchange (BWE), and (iii) the cumulative end result of past ballast operations. Multivariate analysis reduced our data set to three assemblages. Propagule pressure of these potentially colonizing assemblages may be enhanced due to repeated introduction attempts by ships belonging to different ballast management categories (transoceanic, intra-coastal with our without BWE) but that actually carried the same assemblage type. A non-linear relationship between colonization and propagule pressure confirms that the early stages of invasion (ballast uptake, survival in the tank) structure the later stages. Coastal floras in tanks that underwent offshore BWE (particularly persisting freshwater species) corroborate concern with the low efficiency of this management option.
INTRODUCTION
Ballast is any material used to help stabilize a vessel at sea, a practice as ancient as shipping itself, although human traffic across the world's oceans intensified only in the late 1400's (Carlton, 2009). Sand, gravel and stone were initially used as solid ballast but, as of the 1880s, modern steel ships changed over to tanks filled with fresh, brackish or marine water, depending on the location of the source port (Carlton 1985, 2010). As a ship takes on ballast water from the environment to compensate for unloaded cargo, organisms and other suspended particles are also picked up and, during the voyage, some of the living and non-living matter settles to the bottom to become the so-called sediment biota of the ballast tank. Even when ballast water is exchanged, the material accumulated on the bottom of the tank is not completely removed (e.g. Rigby and Hallegraeff, 1994; Bailey et al., 2007). The end result is that the composition and abundance of the sedimented biota is a composite of the assemblages of the various bioregions from where ballast water was taken up over months/years of vessel operation, depending on the time elapsed since the ship was last dry-docked and tanks cleaned (Drake et al., 2007).
Among a wide range of taxa present in ballast sediments (e.g. Galil and Hülsmann, 1997; Bailey et al., 2007), the striking occurrence and germinating potential of microalgae have been demonstrated, including toxic species of diatoms and dinoflagellates (e.g. Hallegraeff and Bolch, 1992; Hamer et al., 2001; Casas-Monroy et al., 2011; Villac and Kaczmarska, 2011a). Nevertheless, research and management options have largely focused on ballast water as a vector of aquatic invasion (Gollasch et al., 2007). A search by title in Web of Science revealed 377 publications with the term ‘ballast water’ as opposed to only 26 with ‘ballast sediment’. Management procedures specific for sediments are required only when ballast tanks are cleaned or repaired, when reception facilities should be available for disposal of sediments (IMO, 2004). Indeed, an operational obstacle in working with ship ballast sediments is access to the material built up on the bottom of the tanks. To overcome this difficulty, Villac and Kaczmarska (Villac and Kaczmarska, 2011a) had the support of the Canadian Aquatic Invasive Species Network (CAISN) to retrieve ballast sediment samples and to quantify diatoms in ships arriving to Canadian ports on the West and East coasts during 2007–2009. Their inspection of 130 samples by light microscopy (LM) revealed diatom cell concentrations from undetectable to 1011 cells/tank, in vitro growth rates of total assemblages of 1.8–4.4 doublings per week, and diversity as high as 40 taxa/tank. Only a modest inventory of the most frequently found taxa was included in that study.
Diatoms are ubiquitous especially in lakes and coastal areas, and possibly contain tens to hundreds of thousands of species (Mann, 1999). Some species are apparently cosmopolitan and the literature contains many examples of narrow endemics (e.g. Williams and Reid, 2006; Vanormelingen et al., 2008). That diatoms, and other micro-organisms, have biogeographical patterns is a central premise for those interested in human-mediated dispersal. Some examples of so-called cosmopolitan morphospecies in the sea have proved to be complexes of semi-cryptic taxa, such as Skeletonema (Kooistra et al., 2008) and Paralia (MacGillivary and Kaczmarska, 2012). Even Pseudo-nitzschia pungens, suspected as having a global gene pool (Medlin, 2007), is actually split among three discrete species lineages and gene flow between distant populations of the most widely distributed one (ITS from clade I) is limited and follows a strong pattern of distance-related isolation (Casteleyn et al., 2010). At regional to global scales, diatom diversity and species composition in lakes are not driven by local environmental selection alone, so that dispersal plays a key role in structuring such communities (Verleyen et al., 2009; Bennett et al., 2010; but see Cermeño and Falkowski, 2009). The advent of ships holding ballast waters, concurrent with a dramatic rise in reported cases of aquatic bioinvasions (Carlton, 2010), may further complicate our still scanty understanding of present-day biogeography of diatoms. Likewise, and of concern regarding ecosystem health, non-indigenous diatoms are known to have ‘replaced’ native ones and to have affected biodiversity in their new environments, especially in those cases in which the newcomer has caused bloom events (e.g. Spaulding et al., 2010; Villac and Kaczmarska, 2011a).
We hypothesize that, at any given time, the diatom assemblage found in the sediments of a ballast tank may be interpreted as the product of the following selective processes: (i) differential intrinsic survival strategies of species present in the ballast tank sediments, identified as biotic pressures; and the effects of (ii) the most recent ballast water exchange (BWE) and of (iii) the cumulative end result of past ballast operations, both recognized as abiotic pressures. The first one is based on the ability of an individual diatom to survive the adverse conditions of ballast tanks by forming resting spores or resting cells (reviewed in McQuoid and Hobson, 1996), as well as having other physiological competences related to energy storage and heterotrophic uptake of organic substrates, especially under light limiting conditions (e.g. Armbrust et al., 2004; Tuchman et al., 2006). The stressors found in a ballast tank (darkness, anoxia, fluctuating salinity, temperature and nutrient availability) would eliminate the less tolerant species/genotypes, leaving open niches for the hardiest species and most adaptable individual genotypes. The second process relates to BWE en-route as a management strategy to minimize the transport of biota between bioregions (IMO, 2004). The efficiency in reducing coastal plankton abundance in the ballast waters by substituting them for scarce populations from the open ocean depends on ship/tank characteristics, the BWE procedure used, the site of ballast uptake and exchange and the type of microbes present (e.g. Rigby and Hallegraeff, 1994; Dickman and Zhang, 1999; McCollin et al., 2007; Simard et al., 2011). Although it is not known how much of the tank bottom sediment is re-suspended and expelled during deballast, the assumption is that the number of cells in the sediment after BWE is somewhat reduced, as well as in the overlying water column. Finally, the third type of selective process depends on magnitude and frequency of ballast operations, in addition to higher (or lower) consistency in ship routes that could draw taxa from the same (or different) regional seed stock(s). The argument underlying these abiotic selective processes is that, either by concentrating or diluting total cell numbers, different species may be introduced into the tank during ballast uptake, facilitating and/or hampering their survival until reaching the destination port.
Here, we expanded our earlier work on propagule pressure (Villac and Kaczmarska, 2011a), the number of individuals introduced, to consider colonization pressure, the number of species introduced or released to a single location (sensuLockwood et al., 2009). Bearing in mind that it is generally not possible to identify exotic microeukaryotes as readily as metazoans or macroalgae (see above regarding microbial species richness and biogeography), our assumption was that ‘tank assemblages’ may be comprised of genotypes foreign to Canadian waters, when not of foreign species. We tested the evidence in support of the three species-driven hypotheses stated above using refined specific diversity. Within the constraints imposed by the species assembled by chance in the ‘tank-communities’ and the use of morphology as the taxonomic tool, our study is based on a large data set. Our goal was to single out diatom species best positioned to reach Canadian coasts in appreciable abundances/frequencies and thus provide insights pertinent to ballast management options in the region.
METHOD
Samples were taken during 2007, 2008 and 2009 from ships that came to Canadian ports on the West coast (WC), the East coast (EC) and the Great Lakes (GL). Methodological details are found in Villac and Kaczmarska (Villac and Kaczmarska, 2011a), thus only summarized here. Sediment was manually collected from a tank after water deballast and immediately preserved with Lugol's solution, except for four live samples from the EC that were stored refrigerated in the dark and processed 3–4 days after collection.
A total of 142 samples were inspected with LM, 61 of them found to contain diatom cells regarded as being alive at the time of sampling (25 WC, 28 EC, 8 GL). These cells showed intact protoplasm and distinct chloroplasts in the case of the preserved samples or had a positive reaction to the vital stain fluorescein diacetate (FDA) in the case of the live samples (results in Villac and Kaczmarska, 2011a). Retrieval of diatoms from the sediment (10–100 µm fraction) was achieved by density gradient centrifugation (Bolch, 1997, adjusted for diatoms by Villac and Kaczmarska, 2011a). For the preserved samples, cell counts with initial identification (400× final magnification) were done by the Utermöhl technique (Hasle, 1978). A minimum of 300 cells were counted to attain a 95% probability of finding a species that comprised 1% of the diatom assemblage (Shaw, 1964). Alternatively, at least half of the Utermöhl chamber was inspected for samples that did not reach this threshold. For the live samples, cell counts were made in Sedgewick–Rafter chambers (200× final magnification, a minimum of 400 cell counts per sample), but the use of the FDA fluorescent stain required an expedited analysis that precluded detailed species identification. LM counts were done with two (preserved) or four (live) replicates, i.e. two to four subsamples were independently processed by gradient centrifugation and average cell concentrations calculated.
All preserved samples within 1–3% detection level (9 WC, 10 EC, 3 GL; most with > 20 taxa) and two samples at 6% detection level (WC-66 and EC-03; both with ≥ 16 taxa) were selected for taxonomic refinement using scanning electron microscopy (SEM, JEOL JSM-5600 operating at 10 kV and 8 mm working distance). SEM preparations were made as described in Kaczmarska et al. (Kaczmarska et al., 2005) with material analyzed and set aside from the Utermöhl chambers and, although not strictly quantitative, they followed a standardized procedure. For each sample, a volume was processed for SEM that would place ca. 1000–10 000 cells on the filter membrane used to collect valves following acid cleaning. This range provided abundant valves for examination, but not so many as to result in overlap of valves. Half of the membrane was retained for archival storage, while the other half was inspected by systematic horizontal transects at 400× (SEM initial magnification = LM final magnification). Fine taxonomy (SEM) was also performed with material from the four live samples after 7 days of incubation in f/2 culture media (Villac and Kaczmarska, 2011a).
The inventory of species was based on 61 samples, but the four live samples were not included in any of the statistical analysis because they were processed using a different LM procedure. Cell concentration from LM is reported as cells/g (wet weight = w.w.) and in terms of cells/tank as extrapolated to the amount of sediment estimated in any respective tank (see Villac and Kaczmarska, 2011a). Results from SEM were categorized as presence/absence data. For multivariate analysis, LM data were transformed to log (x + 1) because data distribution was skewed and had a large number of zero occurrences. A binary matrix was used when LM and SEM information were merged.
Hierarchical clustering of samples was based on the Bray–Curtis index and group average (PRIMER v.5). The routine SIMPER discriminated species accountable for the similarity of samples within a cluster, considering up to 80% cumulative contribution. Given that ‘tank-communities’ are a composite of populations from different sources, gradient analysis was used as an exploratory technique. Indirect gradient analysis was chosen because samples/species are displayed along axes of variation considered to be theoretical variables constructed without reference to actual environmental measurements, but from which underlying gradients can be inferred (ter Braak and Prentice, 1988). Two ordination techniques were tested: principal component analysis (PCA) and correspondence analysis (CA) (CANOCO; ter Braak and Smilauer, 2002).
There were three categories of ships: transoceanic with ballast exchanged (TOE), intra-coastal with ballast exchanged (ICE), and intra-coastal with unexchanged ballast (ICU). Diatoms were classified according to their salinity affinities based on the literature. Three categories were usually adopted, freshwater, brackish and marine, and we assumed that this was based on a generally accepted approximation of such water salinities with fresh being less than 0.5, brackish between 0.5 and 30 and marine between 30 and 35 (Anonymous, 1958). For some species, the classification as solely marine or freshwater is a matter of debate, so that these were considered as marine-brackish or freshwater-brackish. Vegetative cells were classified according to their habit as planktonic or as having a relationship with bottom communities (epiphytic, epilithic), including those that are often found in the water column (tychoplanktonic). Taxa found solely as spores were regarded as such, with the understanding that these forms are most often, but not necessarily, associated with sediment assemblages (Garrison, 1981; Pitcher, 1990).
Results
Species composition
LM counts alone contributed 145 taxa, which is an underestimation of the actual species richness since several cell categories were combined at genus or family level or even into general groupings such as ‘unidentified centrics’ of a given diameter range. The two most diverse and frequent genera, Thalassiosira and Chaetoceros spores, were hardly ever observed in chain formation, a key feature for species identification. SEM analysis contributed an additional 162 taxa, although several of them were informally typified (e.g. Thalassiosira sp.1 … n) or maintained in species compounds (e.g. Aulacoseira granulata sensu lato, Stephanodiscus niagarae complex, Paralia longispina-like). This was necessary when diagnostic characters were lacking on the specimens recovered, and/or in recognition of the importance of analyzing a population rather than just one or a few specimens, and/or in the absence of molecular data considered as key discriminating attributes.
Data analysis (further below) considered all 307 taxa found, but the species list available as Supplementary data, Appendix S1 excluded those informally typified taxa so that this selected inventory includes 180 taxa. The most speciose genera were: Thalassiosira (34 + 14 unnamed taxa), Chaetoceros (present as spores, 15 + 15 unidentified taxa), Cyclotella (11 + 6 unidentified taxa), Actinocyclus (5 + 6 unidentified taxa), Aulacoseira (6 + 1 unidentified taxa), Melosira (5 taxa) and Pseudo-nitzschia (5 taxa). The following genera/families were among those for which diversity is still greatly underestimated: Coscinodiscus, Gyrosigma, Navicula, Nitzschia, Podosira, Stephanodiscus, Cym-bellaceae and Fragilariaceae.
Preserved samples from WC had a total of 202 taxa (n = 25), those from the EC a shorter listing of 148 taxa (n = 24) and those from GL only 55 taxa (n = 8). Although colonization pressure seems to be higher on the Pacific coast, it is noteworthy that the inventory for the EC increased to 202 taxa with the contribution of organisms detected in the four live samples that were allowed 7 days of growth in culture media. Some of the species that had not been detected in the preserved EC samples were rather distinct and unlikely to have been overlooked: Diploneis weissflogii, Leptocylindrus danicus, Psammodictyon aff. constrictum, Skeletonema tropicum, Stephanopyxis nipponica and Thalassionema nitzschioides.
Ships sampled from the marine ports had 28% of the taxa exclusively detected either on the WC or the EC, and 26% were shared taxa; only 4% were exclusively found in ships sampled in the GL ports, and 10% were common to all sampling regions (Fig. 1). About 60% of the taxa were identified to a level that allowed for their classification in terms of their habitat (Supplementary data, Appendix S1). The majority was of marine (45%) and brackish (41%) species, followed by freshwater ones (14%). Planktonic taxa accounted for 57% (34% of which were of Thalassiosira). About 34% were regarded as having a link with bottom communities (epiphytic, epilithic, tychoplanktonic), and 9% were planktonic taxa found as spores (Chaetoceros spp., T. nordenskioeldii, Detonula confervacea, Ditylum brightwellii, Stephanopyxis turris). Spores were detected in 50% of samples (LM data), in ships that came to both marine coasts and one GL ship, and their relative contribution to total cell concentrations varied from 0.1 to 67% (average 11 ± 15%).
Descriptive statistics
On a per sample basis, colonization pressure increased with propagule pressure with the best fit curve a polynomial, cubic correlation with r2 = 0.4 (Fig. 2, significant at P ≤ 0.05). This trend was observed in terms of cells/g and of cells/tank, although data points were more spread out for the latter. The highest colonization pressure was found among WC samples, followed by EC samples, although the difference between samples of intermediate richness was not clear-cut (Fig. 2). The increase in cell concentration with increasing amounts of total sediment in the tanks was significant (P ≤ 0.05) both with a linear or a polynomial, cubic correlation (r2 = 0.6 and r2 = 0.7, respectively). The concentration of a given taxa in different samples followed the same trend, that is, its individual specific cell concentration increased with increments in total cell concentration, as illustrated for the two most frequent taxa, Paralia spp. and Actinoptychus senarius (Fig. 3). No pattern was detected between colonization pressure and total amount of sediment in the tanks. Regardless of the amount of sediment, only a few taxa (often ≤ 3–5) were responsible for the majority of cell concentration in any given sample (Supplementary data, Appendix S2).
Cluster analysis
In 61 of 142 sediment samples, diatoms were present in quantities sufficient for their recovery. Of the preserved samples, cluster analysis with the complete LM data set (57 samples, 138 taxa) produced a dendrogram with several nested groups indicating a trend to set GL samples apart from those collected in ships arriving at marine ports, and some distinction between WC and EC samples (data not shown). The analysis including only those samples more comparable in terms of counting error (24 samples, 129 taxa) refined this trend (Fig. 4a). Most WC and EC samples each formed a cluster (A and B, respectively) and EC-07 was a singleton nested outside this A–B grouping. The remainder of the samples (clusters C and D) were nested outside grouping A–B, and included GL samples and a few EC and WC samples. Although actually ranked into two nested clusters, C + D was operationally considered as a sole unit because, as the refinement of this analysis will show, the positions of some samples within this grouping were interchangeable.
With the taxonomic refinement brought into the cluster classification (SEM results added, 258 taxa), the distinction between WC and EC was confirmed (clusters A and B, respectively), although samples WC-48 and EC-07 became singletons nested outside cluster B (Fig. 4b). Within the C + D grouping, samples were re-arranged so as to sort EC-83 and two GL samples into cluster D and EC-45 as a singleton (Fig. 4b). In the following step, those taxa with only one occurrence were considered fortuitous and excluded from the analysis (24 samples, 139 taxa). This outcome (Fig. 4c) corroborated the previous results very closely and increased the similarity within some clusters. The distributions of samples within clusters A and B were exactly the same as in the previous treatment (compare Fig. 4b and c). Re-arrangement within the C + D grouping brought the former singletons EC-07 and WC-48 into cluster C, set cluster D as an outgroup and EC-45 remained as a singleton though between C and D (Fig. 4c). Expressed in terms of ship category, this later classification revealed the following pattern: within the cluster of samples from marine ports only, the WC cluster included only TOE and ICU ships, whereas the EC cluster included one subset of all ICU ships (B1) and another subset of one ICE and two TOE ships (B2). Except for sample WC-51, taken from an ICU ship, cluster C was comprised of samples from TOE ships that crossed the Atlantic and ICE ships from both marine coasts.
A total of 59 taxa were responsible for the observed clusters, considering ca. 80% cumulative contribution within each main cluster (Table I). Paralia spp. was a major factor in the makeup of each of the three main clusters, but it was especially prominent in subset B1. Actinoptychus senarius, although present in all three regions, played a role only in subunit B2 and cluster D. Among the 33 taxa that generated cluster A, major contributors were marine planktonic species (e.g. Asteromphalus hyalinus, Neodenticula seminae, Pseudo-nitzschia multiseries, Thalassiosira spp.), including some Chaetoceros in spore form (Table I).
Assemblage type . | Taxa . | Habitat . | Cluster A . | Subset B1 . | Subset B2 . | Cluster C . | Cluster D . | |
---|---|---|---|---|---|---|---|---|
I | Thalassiosira nordenskioeldii | MB | PLK | 7.2 | ||||
I | Neodenticula seminae | M | PLK | 4.7 | ||||
I | Thalassiosira pacifica | MB | PLK | 4.7 | ||||
I | Shionodiscus oestrupii var. venrickae | MB | PLK | 2.7 | ||||
I | spore Chaetoceros cf. affinis | M | SPR | 2.6 | ||||
I | spore Chaetoceros cf. radicans | M | SPR | 2.6 | ||||
I | Asteromphalus hyalinus | M | PLK | 1.5 | ||||
I | Pseudo-nitzschia multiseries | M | PLK | 1.5 | ||||
I | Thalassiosira eccentrica | MB | PLK | 1.4 | ||||
II | Paralia spp. | MB | EPI | 7.2 | 10.3 | 2.0 | 7.6 | 5.2 |
II | Skeletonema spp. | MB | PLK | 2.6 | 6.6 | 2.0 | 1.4 | |
II | Odontella aurita | M | EPI | 0.5 | 6.6 | 6.8 | ||
II | Spore Thalassiosira nordenskioeldii | MB | PLK | 1.4 | 6.6 | 6.8 | ||
II | Asterionellopsis glacialis | M | PLK | 6.6 | ||||
II | Cyclotella litoralis | MB | PLK | 3.6 | 2.3 | |||
II | Ditylum brightwellii | M | PLK | 3.5 | ||||
II | Spore Chaetoceros diadema | M | SPR | 1.4 | 3.5 | |||
II | Spore Chaetoceros didymus | M | SPR | 4.8 | 3.5 | |||
II | Spore Chaetoceros debilis | M | SPR | 4.5 | 3.1 | 6.8 | ||
II | Actinocyclus octonarius var. tenellus | B | EPI | 3.1 | ||||
II | Hyalodiscus scoticus | MB | EPI | 3.1 | ||||
II | Melosira nummuloides | MB | EPI | 3.1 | ||||
III | Actinoptychus senarius | MB | EPI | 3.6 | 6.8 | 1.4 | 16.1 | |
III | Cymatosira belgica | M | EPI | 6.8 | 7.8 | 16.1 | ||
III | Cyclotella meneghiniana | FB | PLK | 6.8 | 7.8 | 16.1 | ||
III | Melosira varians | FB | EPI | 6.8 | 7.6 | 16.1 | ||
III | Stephanodiscus hantzschii complex | F | PLK | 6.8 | 3.6 | 4.9 | ||
III | Raphoneis amphiceros | MB | EPI | 2.0 | 8.0 | |||
III | Cyclotella scaldensis | FB | PLK | 0.6 | 13.1 | |||
III | Aulacoseira ambigua | F | PLK | 2.3 | 6.0 | |||
III | Surirella brebissonii sensu lato | B | EPI | 2.0 | ||||
III | Thalassiosira licea | MB | PLK | 4.9 | ||||
III | Cyclotella striata | FB | PLK | 4.0 | ||||
III | Aulacoseira granulata sensu lato | F | PLK | 1.3 |
Assemblage type . | Taxa . | Habitat . | Cluster A . | Subset B1 . | Subset B2 . | Cluster C . | Cluster D . | |
---|---|---|---|---|---|---|---|---|
I | Thalassiosira nordenskioeldii | MB | PLK | 7.2 | ||||
I | Neodenticula seminae | M | PLK | 4.7 | ||||
I | Thalassiosira pacifica | MB | PLK | 4.7 | ||||
I | Shionodiscus oestrupii var. venrickae | MB | PLK | 2.7 | ||||
I | spore Chaetoceros cf. affinis | M | SPR | 2.6 | ||||
I | spore Chaetoceros cf. radicans | M | SPR | 2.6 | ||||
I | Asteromphalus hyalinus | M | PLK | 1.5 | ||||
I | Pseudo-nitzschia multiseries | M | PLK | 1.5 | ||||
I | Thalassiosira eccentrica | MB | PLK | 1.4 | ||||
II | Paralia spp. | MB | EPI | 7.2 | 10.3 | 2.0 | 7.6 | 5.2 |
II | Skeletonema spp. | MB | PLK | 2.6 | 6.6 | 2.0 | 1.4 | |
II | Odontella aurita | M | EPI | 0.5 | 6.6 | 6.8 | ||
II | Spore Thalassiosira nordenskioeldii | MB | PLK | 1.4 | 6.6 | 6.8 | ||
II | Asterionellopsis glacialis | M | PLK | 6.6 | ||||
II | Cyclotella litoralis | MB | PLK | 3.6 | 2.3 | |||
II | Ditylum brightwellii | M | PLK | 3.5 | ||||
II | Spore Chaetoceros diadema | M | SPR | 1.4 | 3.5 | |||
II | Spore Chaetoceros didymus | M | SPR | 4.8 | 3.5 | |||
II | Spore Chaetoceros debilis | M | SPR | 4.5 | 3.1 | 6.8 | ||
II | Actinocyclus octonarius var. tenellus | B | EPI | 3.1 | ||||
II | Hyalodiscus scoticus | MB | EPI | 3.1 | ||||
II | Melosira nummuloides | MB | EPI | 3.1 | ||||
III | Actinoptychus senarius | MB | EPI | 3.6 | 6.8 | 1.4 | 16.1 | |
III | Cymatosira belgica | M | EPI | 6.8 | 7.8 | 16.1 | ||
III | Cyclotella meneghiniana | FB | PLK | 6.8 | 7.8 | 16.1 | ||
III | Melosira varians | FB | EPI | 6.8 | 7.6 | 16.1 | ||
III | Stephanodiscus hantzschii complex | F | PLK | 6.8 | 3.6 | 4.9 | ||
III | Raphoneis amphiceros | MB | EPI | 2.0 | 8.0 | |||
III | Cyclotella scaldensis | FB | PLK | 0.6 | 13.1 | |||
III | Aulacoseira ambigua | F | PLK | 2.3 | 6.0 | |||
III | Surirella brebissonii sensu lato | B | EPI | 2.0 | ||||
III | Thalassiosira licea | MB | PLK | 4.9 | ||||
III | Cyclotella striata | FB | PLK | 4.0 | ||||
III | Aulacoseira granulata sensu lato | F | PLK | 1.3 |
Numbers in the cluster columns correspond to the contributions (%) of each taxon within the respective cluster as calculated by the SIMPER routine considering up to ca. 80% cumulative contribution. Species with contributions below 1% are not shown (most of them from cluster A). Assemblage types as defined in the text. Information for habitat is based on the literature and references are listed with the complete inventory as Supplementary data, Appendix S1. HABITAT: M, marine; B, brackish; F, freshwater; EPI, epiphytic, epilithic, and/or tychoplanktonic; PLK, planktonic; SPR, spore, often associated with sediment assemblages though not necessarily (Garrison, 1981; Pitcher, 1990).
Assemblage type . | Taxa . | Habitat . | Cluster A . | Subset B1 . | Subset B2 . | Cluster C . | Cluster D . | |
---|---|---|---|---|---|---|---|---|
I | Thalassiosira nordenskioeldii | MB | PLK | 7.2 | ||||
I | Neodenticula seminae | M | PLK | 4.7 | ||||
I | Thalassiosira pacifica | MB | PLK | 4.7 | ||||
I | Shionodiscus oestrupii var. venrickae | MB | PLK | 2.7 | ||||
I | spore Chaetoceros cf. affinis | M | SPR | 2.6 | ||||
I | spore Chaetoceros cf. radicans | M | SPR | 2.6 | ||||
I | Asteromphalus hyalinus | M | PLK | 1.5 | ||||
I | Pseudo-nitzschia multiseries | M | PLK | 1.5 | ||||
I | Thalassiosira eccentrica | MB | PLK | 1.4 | ||||
II | Paralia spp. | MB | EPI | 7.2 | 10.3 | 2.0 | 7.6 | 5.2 |
II | Skeletonema spp. | MB | PLK | 2.6 | 6.6 | 2.0 | 1.4 | |
II | Odontella aurita | M | EPI | 0.5 | 6.6 | 6.8 | ||
II | Spore Thalassiosira nordenskioeldii | MB | PLK | 1.4 | 6.6 | 6.8 | ||
II | Asterionellopsis glacialis | M | PLK | 6.6 | ||||
II | Cyclotella litoralis | MB | PLK | 3.6 | 2.3 | |||
II | Ditylum brightwellii | M | PLK | 3.5 | ||||
II | Spore Chaetoceros diadema | M | SPR | 1.4 | 3.5 | |||
II | Spore Chaetoceros didymus | M | SPR | 4.8 | 3.5 | |||
II | Spore Chaetoceros debilis | M | SPR | 4.5 | 3.1 | 6.8 | ||
II | Actinocyclus octonarius var. tenellus | B | EPI | 3.1 | ||||
II | Hyalodiscus scoticus | MB | EPI | 3.1 | ||||
II | Melosira nummuloides | MB | EPI | 3.1 | ||||
III | Actinoptychus senarius | MB | EPI | 3.6 | 6.8 | 1.4 | 16.1 | |
III | Cymatosira belgica | M | EPI | 6.8 | 7.8 | 16.1 | ||
III | Cyclotella meneghiniana | FB | PLK | 6.8 | 7.8 | 16.1 | ||
III | Melosira varians | FB | EPI | 6.8 | 7.6 | 16.1 | ||
III | Stephanodiscus hantzschii complex | F | PLK | 6.8 | 3.6 | 4.9 | ||
III | Raphoneis amphiceros | MB | EPI | 2.0 | 8.0 | |||
III | Cyclotella scaldensis | FB | PLK | 0.6 | 13.1 | |||
III | Aulacoseira ambigua | F | PLK | 2.3 | 6.0 | |||
III | Surirella brebissonii sensu lato | B | EPI | 2.0 | ||||
III | Thalassiosira licea | MB | PLK | 4.9 | ||||
III | Cyclotella striata | FB | PLK | 4.0 | ||||
III | Aulacoseira granulata sensu lato | F | PLK | 1.3 |
Assemblage type . | Taxa . | Habitat . | Cluster A . | Subset B1 . | Subset B2 . | Cluster C . | Cluster D . | |
---|---|---|---|---|---|---|---|---|
I | Thalassiosira nordenskioeldii | MB | PLK | 7.2 | ||||
I | Neodenticula seminae | M | PLK | 4.7 | ||||
I | Thalassiosira pacifica | MB | PLK | 4.7 | ||||
I | Shionodiscus oestrupii var. venrickae | MB | PLK | 2.7 | ||||
I | spore Chaetoceros cf. affinis | M | SPR | 2.6 | ||||
I | spore Chaetoceros cf. radicans | M | SPR | 2.6 | ||||
I | Asteromphalus hyalinus | M | PLK | 1.5 | ||||
I | Pseudo-nitzschia multiseries | M | PLK | 1.5 | ||||
I | Thalassiosira eccentrica | MB | PLK | 1.4 | ||||
II | Paralia spp. | MB | EPI | 7.2 | 10.3 | 2.0 | 7.6 | 5.2 |
II | Skeletonema spp. | MB | PLK | 2.6 | 6.6 | 2.0 | 1.4 | |
II | Odontella aurita | M | EPI | 0.5 | 6.6 | 6.8 | ||
II | Spore Thalassiosira nordenskioeldii | MB | PLK | 1.4 | 6.6 | 6.8 | ||
II | Asterionellopsis glacialis | M | PLK | 6.6 | ||||
II | Cyclotella litoralis | MB | PLK | 3.6 | 2.3 | |||
II | Ditylum brightwellii | M | PLK | 3.5 | ||||
II | Spore Chaetoceros diadema | M | SPR | 1.4 | 3.5 | |||
II | Spore Chaetoceros didymus | M | SPR | 4.8 | 3.5 | |||
II | Spore Chaetoceros debilis | M | SPR | 4.5 | 3.1 | 6.8 | ||
II | Actinocyclus octonarius var. tenellus | B | EPI | 3.1 | ||||
II | Hyalodiscus scoticus | MB | EPI | 3.1 | ||||
II | Melosira nummuloides | MB | EPI | 3.1 | ||||
III | Actinoptychus senarius | MB | EPI | 3.6 | 6.8 | 1.4 | 16.1 | |
III | Cymatosira belgica | M | EPI | 6.8 | 7.8 | 16.1 | ||
III | Cyclotella meneghiniana | FB | PLK | 6.8 | 7.8 | 16.1 | ||
III | Melosira varians | FB | EPI | 6.8 | 7.6 | 16.1 | ||
III | Stephanodiscus hantzschii complex | F | PLK | 6.8 | 3.6 | 4.9 | ||
III | Raphoneis amphiceros | MB | EPI | 2.0 | 8.0 | |||
III | Cyclotella scaldensis | FB | PLK | 0.6 | 13.1 | |||
III | Aulacoseira ambigua | F | PLK | 2.3 | 6.0 | |||
III | Surirella brebissonii sensu lato | B | EPI | 2.0 | ||||
III | Thalassiosira licea | MB | PLK | 4.9 | ||||
III | Cyclotella striata | FB | PLK | 4.0 | ||||
III | Aulacoseira granulata sensu lato | F | PLK | 1.3 |
Numbers in the cluster columns correspond to the contributions (%) of each taxon within the respective cluster as calculated by the SIMPER routine considering up to ca. 80% cumulative contribution. Species with contributions below 1% are not shown (most of them from cluster A). Assemblage types as defined in the text. Information for habitat is based on the literature and references are listed with the complete inventory as Supplementary data, Appendix S1. HABITAT: M, marine; B, brackish; F, freshwater; EPI, epiphytic, epilithic, and/or tychoplanktonic; PLK, planktonic; SPR, spore, often associated with sediment assemblages though not necessarily (Garrison, 1981; Pitcher, 1990).
Group B was formed of 27 taxa, 7 of which common to both subsets of clusters, although only Odontella aurita and spores of Thalassiosira nordenskioeldii in comparable, high percentages (Table I). Subset B1 (EC-ICU ships) was formed by the exclusive (within group B) contributions of Asterionellopsis glacialis, Ditylum brightwellii, spores of Chaetoceros diadema and Chaetoceros didymus, Actinocyclus octonarius var. tenellus, Hyalodiscus scoticus and Melosira nummuloides. Subset B2 (EC-ICE/TOE ships) was formed by the exclusive and high contributions of Cymatosira belgica, Cyclotella meneghiniana, Stephanodiscus hantzschii complex, Melosira varians, Aulacoseira ambigua and Surirella brebissonii sensu lato. Paralia spp. and Skeletonema spp. played a relatively greater role in subset B1, whereas A. senarius and spores of Chaetoceros debilis were more important in forming subset B2.
Within the C + D grouping, cluster C was formed by 18 taxa and cluster D by 11 taxa. Cyclotella scaldensis, Rhaphoneis amphiceros, C. belgica, C. meneghiniana, M. varians and Paralia spp. accounted for 52% of cluster C (Table I). Likewise, A. senarius, C. belgica, C. meneghiniana and M. varians alone accounted for 64% of cluster D that also relied on Aulacoseira ambigua, Paralia spp., S. hantzschii complex and Thalassiosira licea (Table I). The singleton EC-45 was indeed an odd sample (Fig. 3): it had the lowest number of species (14 taxa) of all samples and one of the highest cell counts (6.7 × 1010 cells/tank). It was the only sample with high abundances of the category of unidentified Pseudostaurosira-Staurosira spp. and with a composition resembling that found in cluster C (R. amphiceros) or cluster D (A. ambigua, T. licea).
Gradient analysis
Both PCA (linear model) and CA (unimodal model) yielded comparable results in terms of the distribution of samples/taxa on an ordination biplot, thus only the outcome for the first is presented (Fig. 5a and b). Likewise, similar results were obtained with a PCA done with the same binary matrix used in the most restrictive cluster analysis (24 samples, 139 taxa—Fig. 4c) and with an even more constrained data set (24 samples, 115 taxa) in which most taxa that potentially included more than one species were excluded. This additional constraint was used to facilitate ecological interpretation on species-specific bases. Although ordination analysis was performed solely for exploratory purposes, for the record, the cumulative percentage eigenvalues for the first four axes were 15, 25, 33 and 40%.
The gradient observed along axis 1 was revealed by the display of samples corresponding to cluster A on the extreme portion of the positive side, in opposition to samples from grouping C + D (especially some GL samples) and part of cluster B (the ICE and TOE ships) on the negative side (Fig. 5a). Along axis 2, all samples from group B plotted on the positive side and all samples from cluster A and grouping C + D on the negative side (except for WC-38 and WC-48, though these are close to the intercept). On the positive side, the extreme of this second gradient was created by those samples from EC-ICU ships, those with positions the farthest away from the intercept.
The taxa distributed along axes 1 and 2 corresponded very closely to those already described for these groups in the cluster analysis. The majority of the WC samples (cluster A) were characterized by a closely linked assemblage of marine planktonic species (Fig. 5b, Table I), displaced on the lower-right quadrant. The assemblage discriminating the EC-ICU ships (subset B1) had taxa more loosely linked on the extreme positive side of axis 2 (top-right quadrant). It was composed of a mixture of planktonic (A. glacialis, D. brightwellii, Skeletonema spp., spores of various genera) and sediment dwellers (A. octonarius, H. scoticus, M. nummuloides, O. aurita, Paralia and Podosira), classified as marine or as marine-brackish taxa (Fig. 5b, Table I). The biplot coordinates of D. brightwellii and of the spores of Chaetoceros and Stephanopyxis, all marine and planktonic species, may represent a transition between clusters A and B1. The strongest signal of brackish to freshwater species, especially of sand dwellers, was detected on the negative side of axis 1. The only taxa with marine affinity on this portion of the biplot were A. senarius, C. belgica, R. amphiceros and T. licea. On the lower-left quadrant, grouping C + D corresponded to several planktonic taxa and two sand dwellers (C. belgica and M. varians). On the upper-left quadrant, sand dwellers (A. senarius, S. brebissonii sensu lato and R. amphiceros) stood out as the major contributors to ICE/TOE ships from the EC (subset B2).
Integrating the approaches
The initial perception that WC samples had higher colonization pressure (Fig. 2) was confirmed regardless of the level of constraint applied to the data set (Fig. 6). Similarly, there were matching outcomes among cluster analysis (Fig. 4a–c) and between cluster and ordination techniques (Figs 4c and 5, respectively). Some of the species singled out by the SIMPER routine were not the same ones singled out by the PCA (and vice-versa), but they represented analogous ecological habitats. The observation that almost two-thirds of the species were exclusive to ships reaching either the WC or EC ports (Fig. 1) translated into assemblages corresponding to clusters A and B, respectively (Fig. 4c).
The contrast among three assemblage types (compare Figs 4–6) related to ballast management strategies. Assemblage type I, composed by WC-TOE and WC-ICU samples, had high species richness and marine planktonic taxa were tightly linked. In contrast, assemblage type III had median to low species richness and a strong contribution of sand dwellers and taxa from brackish and freshwaters found in EC-ICE, WC-ICE, EC-TOE and GL-TOE ships (the latter came from overseas ports, same as EC-TOE). Assemblage type II, detected in all EC-ICU ships, created a gradient against the other two assemblage types. It was singled out for having its own suite of planktonic and sand dwellers from marine and brackish waters. The most frequently found and relatively abundant taxa, Paralia (Fig. 3), had its strongest signal within assemblage type II.
Three samples (EC-07, EC-45 and WC-48) stood out as having unstable positions throughout the cluster analysis (Fig. 4). EC-7 was border-line between the EC-exchange subunit of cluster B and cluster C; consequently, its coordinates in the PCA biplot had an intermediate position within group C (Fig. 5a). EC-45 was an odd sample with the second highest cell concentration (Fig. 3) and one of the lowest species richness (Fig. 6). Its unstable position related to that of EC-83, likewise, the sample with the highest cell concentration and low species richness. The instability of WC-48 (WC-TOE) and final positioning within cluster C seems to be an artifact of the sequence of constraints applied to data selection. This sample, prior to the exclusion of taxa, had one of the highest species richness (Fig. 6) and, if its original species composition and cell abundances were taken into account, it might be more rightly placed within cluster A (Fig. 4a).
Discussion
A high diversity of diatoms was found in the matter settled out of the ballast waters taken up by ships. Among 307 taxa recognized in a conservative estimate, there were marine, brackish and freshwater species, some in vegetative and/or resting stages. This represents considerable potential colonization pressure of genotypes or species, albeit within the limitations inherent for microbial ecology as discussed below. Such a vast array of taxa could be reduced to three types of diatom assemblages reaching Canadian waters in the sediments of ballast tanks. All levels of data treatment pointed to this configuration, starting from the merging of LM and SEM data, followed by the exclusion of fortuitously occurring taxa.
Increasing evidence implicates on-going propagule pressure as enhancing the probability of establishment of non-indigenous species and in aiding an invasion to spread by continuous introductions, thus expanding genetic variation of the pioneering populations (Simberloff, 2009). As applied to the present study, propagule pressure may be further enhanced due to the potential for repeated introduction attempts by ships belonging to different ballast management categories but that actually exert the same type of colonization pressure, a synergistic effect. Propagule pressure and colonization pressure showed a non-linear relationship (see Briski et al. 2012 for comparison), corroborating the stochastic nature of introductions through ballast waters and recognizing that, in such cases, the very early stages of the invasion process (ballast uptake, survivorship in the tank) strongly structure later stages (Lockwood et al., 2009).
Hypotheses for human-mediated invasion by diatoms are not trivial (Smayda, 2007). It ideally requires data on vector(s) of introduction, baseline assessments of local floras, understanding the biogeography of target organisms as well as the appraisal of cryptic diversity by molecular methods and paleontological records if applicable. Often one or more of these parameters are missing for micro-organisms. Our list of potential (still confined in the tank) new records for Canadian waters included 70 species to the Pacific coast, 60 species to the Atlantic coast and 12 to the GL (Supplementary data, Appendix S1). If any of these species were to be found in the natural environment, we have provided data on the potential transport vector but the hypothesis for its recent introduction will require further scrutiny.
The ecology of the tank assemblages
None of the assemblages revealed the contribution of oceanic-type flora, such as Asterolampra marylandica, Gossleriella tropica, Mastogloia rostrata, Pseudosolenia calcar-avis and other large-sized diatoms (Villac and Kaczmarska, 2011b, and references therein). The lack of oceanic species is puzzling (Supplementary data, Appendix S1). Either (i) oceanic species all die out rapidly or, (ii) the so-called offshore BWE are made too close to shelf waters, or (iii) coastal species persist and outnumber the oceanic ones. Although it is widely documented that diatom populations in ballast waters, either coastal or oceanic, undergo overall decline during passage (e.g. Klein et al., 2009b; Lang and Kaczmarska, 2012), differential effects between coastal and oceanic populations have not been elucidated. All three possibilities may mutually occur.
The overall high contribution of marine (45%) and planktonic (57%) taxa was related to assemblage type I. This was the closest to an ‘oceanic’ flora we recovered in our data set. These taxa were found in samples whose last ports of call were in Asia (China and Japan) and carried out BWE in the middle of the Pacific Ocean, and by WC-ICU samples. To understand TOE and ICU ships clustering together, it is important to recall that the ships were classified as ICU because their most recent port of call was within proximal coastal range. Before that, the ship likely had visited several different ports, potentially across the ocean. At least for two ICU ships (WC-54 and WC-66), ship log records showed that Japan, South Korea, India and China were among frequent ports of call (though with a few stops in European ports). Since recent BWE was not performed, the assemblages of these ICU tanks had similarities with those that also came from the NW Pacific: a diverse marine planktonic flora typical of neritic regions lacking or with modest freshwater contribution from land.
The same line of reasoning can be followed to interpret the split of EC-ICU ships to create assemblage type II, a mixture of marine and brackish water planktonic species with some bottom dwellers. As depicted by the cluster analysis, this group of samples had some resemblance to three EC samples that performed BWE (cluster B taken as a whole), but the PCA analysis clarified a gradient of greater affinity of the latter with the assemblage type III. The taxa found in the EC-ICU ships are not uncommon in shallow coastal areas around the world, individually or as an assemblage alike. The components of assemblage type II could very well represent the flora of a particular estuarine bay anywhere from subtropical to temperate latitudes. More likely, it is a composite of the communities drawn from such systems due to the cumulative effect of ballast operations. It is noteworthy that these same EC-ICU sediment samples had high cell concentrations and species diversity of dinoflagellate cysts (Casas-Monroy et al., 2011).
In total contrast, assemblage type III was composed of loosely bound taxa with one main commonality: TOE and ICE ships crossing the Atlantic and ICE ships navigating along the Pacific coast. It had brackish water, truly freshwater diatoms and occasional marine components. This is an improbable mixture that does not mirror any known condition from the natural environment and does not reflect the expected outcome for ships that underwent BWE offshore. The one WC-TOE sample (WC-48) placed in this assemblage was shown to represent an artifact of the reduction of the data matrix and actually fits best within assemblage type I. The reason why sample WC-51, an ICU ship, landed in this assemblage is unknown and may lie in the ballast history of this particular tank. So, if assemblage type III has a conspicuous brackish and freshwater signal of a large number of bottom dwelling taxa detected in TOE and ICE ships, the query must be how and/or where these ships are carrying out BWE.
Hypothesis 1: differential intrinsic survival strategies
The quantitative dominance of few species in every sample analyzed (Supplementary data, Appendix S2), despite total cell abundance and species richness, indicates that less tolerant species/genotypes were eliminated or decreased in number. Estuarine, benthic and spore forming diatoms are prevalent in coastal floras that were also most common and abundantly represented in our samples, likely compounding propagule pressure on colonization pressure by these species. Indeed, brackish water species are presumably adjusted to a wide range of environmental changes and bottom dwellers, given their original habitat, are anticipated to thrive in light-limiting conditions typical of ballast tanks. The first step in the chain of selective events that lead to the successful transport and introduction of non-indigenous species in ballast tanks is the access of the organism to the tank during ballast uptake (Carlton, 1985). The high incidence of estuarine (41%) and of benthic species (34%) represents an a priori biotic pressure, whereas the occurrence of spores (9%) may be a bottle-neck filter that operates both on a priori and a posteriori stages, when cells are already in the tank.
The formation of spores is a key survival strategy because it may be triggered as soon as a cell meets adverse environments and, given favorable conditions, most diatoms re-start asexual reproduction within 24 h (McQuoid and Hobson, 1996). For example, within a week, total assemblages of EC live samples reached cell concentrations equivalent to 1.8–4.4 doublings of the original inoculation (Villac and Kaczmarska, 2011a). In this case, Melosira nummuloides and Chaetoceros debilis were initially detected only as spores. Although there is no specific mention of the presence of spores or resting cells, diatom growth was observed from ballast sediments that had been stored for 6 months (Hallegraeff and Bolch, 1992) and from natural sediments stored for 4 years (Lewis et al., 1999). It is likely that the spores detected in the sediments of our samples settled as such (not as vegetative forms) from the overlying ballast water. Similar to our findings, Chaetoceros spores were detected in ca. 50% of the ballast water samples taken from ships reaching Canadian ports during the same time-frame as ours (but not the same tanks) and the seven most relevant species in the sediments (Table I) are among the 19 most common taxa found in the water (Klein et al., 2009a).
Hypothesis 2: the effect of the most recent BWE
The expected exchange of coastal floras for offshore communities in the overlying ballast water was far from having a similar effect for the organisms in the sediment. Incomplete sediment removal during BWE too close to the shelf break in the wide-shelf coasts may contribute to persistence of the most tolerant species in ballast sediments, including diatoms known as freshwater. This persistence may potentially expand the ecological range of colonizers, even to inland aquatic systems, although how many of the cells in the sediments could be re-suspended and reach the environment during subsequent ballast operations is unknown. A conservative figure for ships arriving during a 2-week period at the Port of Saint John on the EC of Canada estimated that, if only 1% of viable diatoms in the tank sediments were deballasted, the environment might have harbored 108 to 109 newcomers given 7 days of growth (Villac and Kaczmarska, 2011a).
Canadian regulations (Transport Canada, 2006) require that all transoceanic vessels perform BWE 200 nautical miles away from the coast and with local depth of at least 2000 m. For ships that are not able to comply, BWE is allowed in designated coastal sectors or in areas only 50 nautical miles from land. Intraregional traffic coming from south of Cape Blanco (Oregon, USA) or south of Cape Cod (Massachusetts, USA) fall into this second category and correspond to our ICE ships. BWE of ICE ships was perhaps performed too close to the edge of the shelf-break where diatoms were discharged, but the local flora with coastal components was also taken up. Even when transoceanic vessels complied with the most rigorous recommendation, BWE of our ships coming from Europe to North America was performed fairly close to shelf waters, unlike ships crossing the Pacific (Supplementary data, Appendix S4). Diatom populations in the North Atlantic are laterally exchanged by small-scale eddies so that coast-to-offshore transition is less abrupt than observed in the North Pacific, a wider ocean with narrower continental shelves (Villac and Kaczmarska, 2011b).
Although we do not have salinity data, it is likely that low salinity residues may be detected in tanks that underwent offshore BWE even when performed to its full, expected level of 95% exchange (e.g. Olenin et al., 2000; McCollin et al., 2007; Klein et al., 2009b). The conspicuous contribution of freshwater diatom flora in assemblage type III may result from a chain of events starting from the source port(s) systematically located in a freshwater system followed by the accumulation of cells in the residue left from a number BWE operations. It should be noted that the number of last ports of call inland, a source of freshwater seed-stocks, was greater in Europe and the North American EC than in Asia and the North American WC (Supplementary data, Appendix S5).
Hypothesis 3: the cumulative end result of past ballast operations
Interpretation of patterns would require disclosure of ballast operations from ship logs, information considered as classified by most shipping companies. Data on the previous 10 ports of call were available only for some ships reaching the WC, but even then, without detailed records for the particular tank of interest. Despite this shortcoming, three lines of evidence support the hypothesis that the specific composition of potential colonizers in ballast sediments reflects, at least in part, the cumulative end result of successive ballast uptakes combined with incomplete sediment renewal during en-route BWE.
First, cell concentration increased with increasing amount of sediment in a tank and number of taxa per sample correlated with cell concentration (Fig. 2). This trend was not exclusive to diatoms. Sediments from the same tanks inspected for diatoms also had higher abundances and species richness of invertebrates with increasing amount of sediment (Briski et al., 2011). Second, indirect evidence showed that cell concentration in the sediments may be orders of magnitude higher than in the overlying water column of the ship tank. As mentioned above, the composition of Chaetoceros spores in the sediments mirrored that found in ballast waters of related (though not the same) ships that came to Canadian ports (comparison with Klein et al., 2009a). However, concentrations reached maximum values of 4 × 102 cells/L in the water column as opposed to the average value of 6 × 106 cells/L (± 3 × 107 cells/L) in the sediments. These taxa, in a resistant resting stage, seem to have been continuously sedimented and accumulated on the bottom of ballast tanks. The third evidence was that truly marine and freshwater diatoms were part of the same assemblages, whether the tank had undergone recent BWE or not. Moreover, freshwater taxa contributed 14% of our inventory, a fairly high relative contribution considering that ballast is taken from either inland or ocean-bound ports, but BWE is always carried out in marine waters.
Insights for management options
A study of global hot spots of bioinvasion linked to ballast water transport indicated that reducing the per-ship-visit chance of causing invasion is more effective in reducing the rate of biotic homogenization than eliminating key ports that are the epicenters for global spread (Drake and Lodge, 2004). Risk assessment plans in the ports of entry range from all ships being subject to management to an informed, selective approach that takes into account different levels of risk posed by a given ship route (IMO, 2004; Gollasch et al. 2007). The synergy between propagule and colonization pressure, that is, all per-ship-visits reduced to three assemblage types, reinforces the assessment that all ships coming to Canadian ports are of concern.
In about half of the samples (57%), diatoms were below the detection level of our methods. Although ‘undetectable’ does not mean absent and reduced risk does not mean ‘no risk’, it is encouraging that only 36% of the WC ships and 60% of the EC ships contained detectable diatoms in the sediment of ballast tanks. Concerning the smaller GL data set, diatoms were found in eight out of the nine samples examined, a much higher proportion than in WC and EC ship tanks.
Based on propagule pressure alone, this data set indicated that the Pacific and the Atlantic coasts would require different management practices (Villac and Kaczmarska, 2011a). From that perspective, all ship categories that reach the WC should receive the same level of concern. For the EC, the ICU ships exerted a bioinvasion pattern based on more frequent events though with consistently lower diatom cell concentrations, whereas ICE and TOE ships showed a pattern of less frequent events though of higher magnitude and less predictable. With the taxonomic refinement of the potential colonizers brought to the analysis, the close association among WC samples was confirmed for TOE and ICU ships (assemblage type I). Likewise, EC-ICU samples also stood out on their own (assemblage type II). The novelty was the closer association of WC-ICE samples with EC-ICE and EC-TOE (assemblage type III), to which the GL-TOE samples that were not part of the previous study were also added.
According to Briski et al. (Briski et al., 2010), enforcement of BWE in the GL in 2006 significantly reduced the overall amount of sediment in ballast tanks that served those local ports in comparison with pre-regulation period, indicating that tank sediment (mostly transoceanic ships) were re-suspended and deballasted offshore. In contrast, our ICE ships held larger amounts of sediments than TOE and ICU ships (Supplementary data, Appendix S3). Less stringent regulations for ships in intra-coastal routes are required from the standpoint of shipping logistics and bioinvasions attributed to regional ship traffic are well discussed (e.g. Carlton, 1985; Ruiz et al., 2000). At the same time that we recognize that the international regulatory regime for shipping is complex, management options must be re-assessed, enforced and/or new alternatives developed as we learn more about the biota being translocated. Currently, provisions for sediment management apply only when cleaning or repair of ballast tanks occur (Transport Canada, 2006). Thus, besides fostering more frequent cleaning of ballast tanks, the risk posed by the organisms accumulated in the sediments can only be addressed indirectly through management alternatives available for the ballast water to be used concurrently with (or in place of) BWE.
CONCLUSION
This research focused on the diatom assemblages found in the seldom investigated sediments of ballast tanks that may reach Canadian ports on the Pacific, Atlantic, and the GL. It was based on a large data set, high-resolution species identification and the use of multivariate analysis without a priori assumptions about the type of ship or voyage. A step-wise approach strengthened our views regarding the selective pressures that may drive species composition in a given tank and patterns of transport of species/assemblages and implications for ballast management. The relationship between abundance (propagule pressure) and diversity (colonization pressure) of diatoms in the sediments of ballast tanks follows a key concept in invasion ecology, that of ‘the more you introduce the more you get’ (Lockwood et al., 2009). Reducing an inventory of 307 taxa to only three assemblage types indicated that repeated introduction attempts by ships of different ballast management categories may enhance the probability of successful establishment of non-native diatoms in Canadian waters.
FUNDING
This work was supported by CAISN (Canadian Aquatic Invasive Species Network) and associated sponsors, and by a NSERC (Natural Sciences and Engineering Research Council of Canada) Discovery Grant to I.K.
ACKNOWLEDGEMENTS
We thank all individuals that were part of the CAISN sampling teams. Participation of the shipping companies was greatly appreciated, as well as the suggestions of two anonymous reviewers.
References
Author notes
Corresponding editor: John Dolan