Science of the Total Environment 472 (2014) 1036–1043

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Performance of a biological deoxygenation process for ships' ballast water treatment under very cold water conditions Yves de Lafontaine ⁎,1, Simon-Pierre Despatie 1 Water Science and Technology, St. Lawrence Centre, Environment Canada, 105 McGill St., Montreal, QC H2Y 2E7, Canada

a r t i c l e

i n f o

Article history: Received 6 September 2013 Received in revised form 20 November 2013 Accepted 24 November 2013 Available online 15 December 2013 Keywords: Water treatment Deoxygenation Toxicity Bioreactor Ballast water Cold temperature

a b s t r a c t Water deoxygenation is listed among the promising on-board treatment technologies to treat ships' ballast waters to reduce the risk of species transfer. We assessed the performance of a yeast-based bioreactive deoxygenation process in very cold water (b2 °C) and determined the potential toxicity of the residual treated waters. Experiments using two treatment levels (0.5% and 1% v/v) were conducted in large-volume (4.5 m3) tanks over 19 days at mean temperature of 1.5 °C. Time to hypoxia varied between 10.3 and 16 days, being slightly higher than the predicted time of 9.8 days from previous empirical relationships. Water deoxygenation was achieved when yeast density exceeded 5 × 105 viable cells mL−1 and variation in time to hypoxia was mainly explained by difference in yeast growth. There was no oxycline and no significant difference in yeast density over the 2-m deep water column. Results from six bioassays indicated weak toxic response of treated waters at the 1.0% level, but no potential toxic response at the 0.5% treatment level. Results confirmed that the potential application of a yeast-based deoxygenation process for treating ships' ballast waters extended over the range of water temperature typically encountered during most shipping operational conditions. Time to reach full deoxygenation may however be limiting for universal application of this treatment which should be preferably used for ships making longer voyages in cold environments. There was no evidence that biological deoxygenation at low temperature did increase toxicity risk of treated waters to impede their disposal at the time of discharge. Crown Copyright © 2013 Published by Elsevier B.V. All rights reserved.

1. Introduction The discharge of ships' ballast waters is listed as one of the principal vectors for the global spread of aquatic invasive species which has often resulted in major socio-economic and environmental impacts on aquatic ecosystems of many countries (Carlton and Geller, 1993; Gonçalves and Gagnon, 2012; Lodge et al., 2006). In order to reduce the risk of species introduction and transfer, many countries require that transoceanic ships either conduct ballast water exchange (BWE) at sea before entering coastal and inland waters or use water treatment technologies to eliminate living organisms in their ballast tanks (Gollasch et al., 2007; Gonçalves and Gagnon, 2012). The use of treatment technologies will strongly depend on the demonstration of their effectiveness in treating ballast water under a variety of environmental conditions as typically encountered by ships during their voyage (Hunt et al., 2005; Tsolaki and Diamadopoulos, 2010). Shipboard technologies must be effective in treating both fresh- and saltwater at temperatures ranging from near 0 to 30 °C. They must also be cost-efficient and environmentally safe so that the treated ballast waters can be discharged without causing secondary pollution or having negative impacts to the receiving waters and they should meet local environmental regulations (Gregg and ⁎ Corresponding author at: Tel.: +1 514 496 5025; fax: +1 514 496 7398. E-mail address: [email protected] (Y. de Lafontaine). 1 Tel.: +1 514 496 5025; fax: +1 514 496 7398.

Hallegraeff, 2007). For example, some chemical treatments may be effective in killing organisms under various environmental conditions but the toxicity of residual treated water at discharge will be a function of chemical degradation rates which may vary with water properties (de Lafontaine et al., 2008, 2009; Gonçalves and Gagnon, 2012; Landrum et al., 2003; Sano et al., 2005). Ballast water treatment systems should therefore be tested using laboratory experiments and on-board ship trials to assess their effectiveness and their potential environmental impact before making recommendations on their use to replace BWE. Water deoxygenation was listed among the promising alternative solutions for ballast water management (Gonçalves and Gagnon, 2012; Matheickal and Raaymakers, 2004; Mountfort et al., 1999; Tamburri et al., 2002; Tsolaki and Diamadopoulos, 2010). Prolonged hypoxic conditions (dissolved oxygen b 1 mg L−1) can be detrimental to many aquatic organisms and lead to massive mortality under confined situations (Mountfort et al., 1999; Tamburri et al., 2004; Tamburri and Ruiz, 2005; Tamburri et al., 2002). Moreover, anoxic water could help in reducing corrosion in steel ballast tanks (Tamburri et al., 2004). A biological deoxygenation process using a yeast-based mixture was shown to be effective for generating and maintaining anoxic conditions (dissolved oxygen b 0.3 mg L−1) in both fresh- and saltwater for up to 2 weeks in 200-liter containers under laboratory conditions (de Lafontaine et al., 2013). The treatment uses a mixture of red yeast (Sporidiobolaceae) largely dominated by Rhodotorula glutinis that does not ferment or produce alcohol and that can grow

0048-9697/$ – see front matter. Crown Copyright © 2013 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2013.11.116

Y. de Lafontaine, S.-P. Despatie / Science of the Total Environment 472 (2014) 1036–1043

well at temperatures of b15 °C (Hagler and Ahearn, 1987). Unlike other processes based on nutrients addition to stimulate bacterial growth that failed to fully deoxygenate water (McCollin et al., 2007; Mountfort et al., 1999), this yeast-based bioreactive process can rapidly lead to anoxia between 8 and 175 h after inoculation. Time to full deoxygenation was inversely related to water temperature between 4.5 and 25 °C, presumably due to the effect of temperature on yeast growth and activity (de Lafontaine et al., 2013). Although the process did significantly alter several water quality parameters, ecotoxicity assessment tests indicated that the discharge of treated water would pose no toxic risk to the recipient freshwaters (de Lafontaine et al., 2013). The performance of this bioreactive deoxygenation process has not been tested at water temperatures of b4 °C. When exposed to very low or near-freezing temperature, many yeast species exhibit a cold temperature stress associated to important physiological changes leading to severe growth reduction and to the synthesis of trehalose, a cryoprotectant for cells that increases tolerance to low temperatures and freezing (Inouye and Phadtare, 2004; Kandror et al., 2004). In addition, a very slow growth rate of yeasts in cold water could result in increased sinking of the cells which would in turn reduce the capacity to achieve full deoxygenation over the entire water column. It was thus important to test the performance of the bioreactive deoxygenation process at temperatures of b4 °C in order to assess its potential use as a treatment for ships carrying very cold ballast waters or sailing in very cold environments. This is particularly relevant in the context of the expected increase in shipping activities in Arctic waters as a result of global climatic changes (Chan et al., 2013; Ware et al., 2013). The objective of the present study was to assess the effectiveness of a yeast-based bioreactive deoxygenation process to deoxygenate water under very cold conditions (b 2 °C) and to determine the potential toxicity of the residual treated waters. In laboratory conditions, we specifically tested the following hypotheses that a biological deoxygenation process will: 1) generate and maintain anoxic conditions at very cold water temperatures, 2) produce homogeneous anoxic conditions over the entire depth of large volume tanks at very cold water temperature, and 3) yield treated waters having no residual toxicity and no potential toxicological risk when discharged into aquatic environments. Given that the detrimental effects of prolonged water deoxygenation on a variety of aquatic organisms are well documented (Tamburri and Ruiz, 2005; Tamburri et al., 2002), our objective was not to test whether the biological deoxygenation treatment was effective in eliminating organisms to meet proposed standards for ballast water discharge standards (Gollasch et al., 2007) but rather that the process was capable of inducing water deoxygenation in very cold water conditions. 2. Materials and methods 2.1. The cold water experimental design The cold water experiments were conducted in three large-volume (4.5 m3) cylindrical polyethylene tanks (1.83 m diameter and 1.80 m height), in the fish culture room at the Aquarium du Quebec on the St. Lawrence River waterfront at Quebec City. The tanks design and the experimental set-up were similar to those used in previous experiments to assess other chemical treatments in cold water conditions (de Lafontaine and Despatie, 2006; de Lafontaine et al., 2008). Briefly, the side of the tanks was covered with insulation and a 5-cm-thick styrofoam sheet was placed at the water surface to minimize heat exchange and light penetration. There was no lid on top of each tank allowing contact between air and water at the periphery of the styrofoam sheet. Previous experiments in smaller tanks (200 L) without lids or sheets at water surface demonstrated that air–water contact through water mixing did not affect water the deoxygenation process (de Lafontaine et al., 2013). Feedpipes running from each of the three feedthrough holes located at 10 cm from the top, in the middle and 10 cm from the bottom of each tank permitted water sampling from

1037

the center of the tank. The tanks were filled with unfiltered freshwater from the St. Lawrence River supplied by the pumping station facility of the Aquarium du Quebec. Complete filling of the tanks took approximately 40 min at a mean flow rate of 120 L min−1. The treatment process developed under the trade name BallaClean® for which a patent application is under review consists of a Rhodotorulayeast mixture at a density of ~108 cells mL−1 (solution A). The process is normally activated by mixing equal of quantities of the yeast mixture and a carbon-rich solution (solution B) into water. Initially prepared for another large-tank experiment conducted at warmer temperature (de Lafontaine et al., 2013), the stock (undiluted) solution supplied for this cold-water experiment was 10 months old (M. de Blois, Redal Technologie, treatment supplier, pers. comm.) and had a mean pH of 2.9. Two tanks were inoculated with the stock solution to get treatment concentrations of 0.5% and 1.0% v/v respectively and one tank was filled with untreated water to serve as the experimental control. Inoculation consisted of pouring in three equal quantities of the stock solution while filling the tanks in order to ensure complete mixing. Water was not mechanically mixed during the experiment. Because of logistic constraints, tanks were emptied once water deoxygenation was achieved in a first tank and part of the content of each tank was transferred (via a feedthrough hole located at mid-depth) into 80-liter plastic containers until the end of the experiment on day 19. The monitoring of dissolved oxygen and other water characteristics (temperature, conductivity, pH) revealed no significant changes in water conditions before and after water transfer between tanks. Room ambient temperature fluctuated diurnally between −1.4 and 5.6 °C during experimentation. Because of the limited budget available and the particular logistics associated to the experimental set-up in this study, this cold water experiment could not be replicated over time. Although the two treated tanks had different treatment levels and could not be considered as exact replicates, a series of 15 replicated experiments conducted at warmer temperature showed that there was no significant difference in water deoxygenation performance at treatment concentration of N0.5% v/v (de Lafontaine et al., 2013). These previous experiments always resulted in complete water deoxygenation under all tested conditions indicating a high replicability of the process (de Lafontaine et al., 2013). 2.2. Water analyses Dissolved oxygen (DO) concentration and water temperature were measured twice a day with an in-situ sensor (Solinst LTDO model 3301 levelogger) near the surface and bottom of the tanks. Between measurements, D.O. concentrations were continuously recorded at 15-min intervals with the sensor immersed near the bottom of the 1% treated tank. The detection limit of the sensor was 0.35–0.4 mg L− 1 depending on ambient water temperature. Water temperature was also recorded at 15-min intervals by StowAway Tidbit underwater data loggers (precision: ± 0.4 °C) attached at the end of the middepth feedthrough pipe inside each tank. Water conductivity was also measured with an in-situ sensor probe twice a day near the surface and near the bottom of each tank. Near surface and near bottom water samples were collected twice daily for pH measurements and daily for analysis of nitrates (NO2–NO3 N 0.04 mg-N L−1), ammonia −1 (NH+ ), orthophosphates (PO4 N 0.01 mg-P L−1), total 4 N 0.01 mg-N L phosphorus (TP N 0.01 mg-P L−1), dissolved organic carbon (DOC N 0.25 mg-C L−1), particulate organic carbon (POC N 0.05 mg-CL−1) and particulate organic nitrogen (PON N 0.08 mg-N L− 1), following the standard protocols of Environment Canada — Water Science and Technology Directorate. Samples were also drawn and preserved for the determination of total sulfides (e.g. H2S, HS−) by spectrophotometry based on the methylene blue method, a procedure applicable to sulfide concentrations between 0.1 and 20 mg-S L−1 with a detection limit better than 0.01 mg-S L−1 (American Water Works Association (AWWA), 2005). Total suspended matter (TSS N 1 mg L− 1) was

Y. de Lafontaine, S.-P. Despatie / Science of the Total Environment 472 (2014) 1036–1043

Samples for ecotoxicological assays were taken at 138 h (5.7 days) and 449 h (18.7 days) after the beginning of the experiment. Following the approach used in previous assessment studies (de Lafontaine et al., 2008, 2013), the potent toxicity of treated waters on various components of the food web was assessed by a battery of six bioassays from bacteria to fish, including the Microtox (Vibrio fischeri) test, the algal (Pseudokirchneriella capricornatum) test, the cladoceran (Daphnia magna) test, the microcrustacean (Thamnocephalus platyurus) test, the Coelenterate (Hydra attenuata) test and the trout (Oncorhynchus mykiss) test. The Microtox and algal tests are two sublethal tests measuring the inhibitory effect of toxic compounds on cell physiology and growth. The four others are lethal tests measuring the proportion of dead organisms resulting from toxic exposure over different time periods. Although their experimental protocols vary slightly, all assays are based on a common procedure that consists of exposing, under controlled laboratory conditions, living organisms to a dilution series of water samples and to note and quantify the observed effects. Treated water samples were aerated during testing in order to eliminate the possible confounding factor of anoxia on test response. Data were analyzed to compute various endpoint values (LC50 — 50% lethal concentration; IC50 — 50% inhibitory concentration; or NOEC — no observed effect concentration) depending on the assay. The results are expressed in toxic units (TU), which correspond to the inverse of the dilution factor required to reduce potent toxicity at the selected endpoint concentration. All the toxicity tests were conducted according to standard procedures and protocols currently used by Environment Canada [see de Lafontaine et al. (2008) for complete references].

2.4. Statistical treatment Experimental data were tested for homogeneity of variances and logarithmic transformation of data was applied when appropriate. Vertical difference in parameters between surface and bottom samples in tanks was tested by paired-T test. Differences in parameters between treatment levels (control, 0.5% and 1.0%) were tested by ANOVA. Correlation and regression analyses of parameter values against time were used to test for significant temporal trends in parameters.

3.1. Deoxygenation process Water temperature in treated tanks varied between 0.29 and 2.33 °C over time averaging 1.52 °C (s.d. ± 0.46) for the entire experiment duration (Fig. 1a). Water temperature was, on average, 0.15 °C lower in the 0.5% treated tank relative to the 1.0% treated tank (paired T-test, p b 0.001). Water temperature in the control tank was 0.35 °C lower than in the 0.5% treated tank. This slight difference between tanks was mainly explained by the position of each tank relative to the room heating unit. Differences in temperature at near surface and near bottom of the tanks were not statistically significant, indicating no thermal stratification of water in the tanks (paired T-test, p N 0.05). The D.O. concentrations over time in treated tanks were characterized by an initial phase of minimal change followed by a phase of rapid oxygen depletion leading to hypoxia (Fig. 1b), as previously reported for this yeast-based deoxygenation process (de Lafontaine et al., 2013). Time to reach hypoxia (b 0.5 mg L− 1) was 247.2 h (10.3 days) after inoculation at the 0.5% treatment level and 384 h (16 days) in the 1.0% treated tank. Once anoxic conditions were established in each tank, D.O. levels remained below detection limit (b0.4 mg L−1) until the experiment ended on day 19. Absolute differences in D.O. concentrations measured at the surface and bottom of the treated tanks were less than 0.1 mg L−1 and not statistically different from 0 (paired T-test, p N 0.05) revealing no oxycline in treated waters. Water in the control tank remained fully oxygenated throughout the experiment (mean D.O. = 13.17 mg L−1, range = 12.81– 13.35 mg L−1) and there was no significant difference in D.O. measurements between surface and bottom samples (paired T-test, p N 0.05). The density of viable yeast cells did not differ significantly between near surface and bottom samples (paired T-test, p N 0.05), indicating a vertically well-mixed population of yeasts in each treated tank. Yeasts densities replicates were thus averaged for each sample. At the time of inoculation, the density of viable yeast cells averaged 3.14 × 104 (±0.71 × 104) and 7.00 × 104 (± 0.21 × 104) cells mL−1 in the 0.5%

a) 4

Temperature (oC)

2.3. Ecotoxicological assays

3. Results

3 2 1 0

b) -1

measured on day 6 and at the end (day 19) of the experiment. The biomass of living phytoplankton in the water was estimated daily by fluorometric determination of chlorophyll a and phaeopigment concentrations in 400 mL water samples (Parsons et al., 1984). The ratio of chlorophyll a to phaeopigments was calculated as an index of phytoplankton vitality. Ratio values of N2 are usually interpreted as evidence of healthy and active phytoplankton cells, while values approaching 1 indicate increased amounts of phaeopigments relative to chlorophyll, which is typical of decaying phytoplankton (Pena et al., 1991). The density of viable yeast cells at near surface and bottom of each treated tank was measured once a day using a violet methylene staining procedure (Smart et al., 1999). Cell counts of four replicates taken from every sample were done with a hemacytometer using light microscopy at 400× magnification and the mean density of cells was calculated. The biomass of living microorganisms (including yeasts as well as other bacteria, phytoplankton and microzooplankton present in raw waters when filling the tanks) was estimated by ATP levels in particulate matter (Parsons et al., 1984). Water samples drawn for ATP determination were prefiltered on a series of two on-line filters using sterilized 53 μm mesh Nitex netting and 0.7 μm GF/F filters, respectively. Filters were analyzed to estimate living biomass for the 0.7–53 μm size class. ATP levels expressed in mol L− 1 were determined with a Turner Designs model TD20/20 luminometer using the luceferin–luciferase assay with a detection limit of 10−16 mol mL−1.

Dissolved oxygen (mg L )

1038

14 13 12 11 10 9 8 7 6 5 4 3 2 1 0

Control 0.5 % 1%

0

1

2

3

4

5

6

7

8 9 10 11 12 13 14 15 16 17 18 19

Time since inoculation (days) Fig. 1. a) Continuous record of water temperature at the bottom of treatment tanks. b) Concentrations of dissolved oxygen (mg L−1) over time in the control and the two treatment tanks during the water deoxygenation experiment.

Y. de Lafontaine, S.-P. Despatie / Science of the Total Environment 472 (2014) 1036–1043

and 1.0% treated tanks respectively (Fig. 2a). Viable cells represented between 5 and 10% of the total density of yeast cells inoculated (Fig. 2b). Viable cell density in the 0.5% treated tank remained quite stable during the first week of the experiment and began to increase on day 8 to reach 9.75 × 105 cells mL−1 at the end of the experiment. The percent of viable cells also increased up to 25%, suggesting the development of an active and healthier yeast population over time. Viable cell density in the 1.0% treated tank did not exhibit any significant increase during the first 13 days after inoculation. During the third week, the percent of viable cells increased from 4% to 33% yielding a density of 7.90 × 105 viable cells mL−1 at the end of the experiment, a value close to that observed in the 0.5% treated tank. The temporal variation in the viable cell density was inverse that of dissolved oxygen in each tank, where the increase in yeast cell density and vitality corresponded to the decline in D.O. levels (Fig. 1b). This indicated that the growth of yeasts was the main causative agent for water deoxygenation. The ATP levels in untreated water remained fairly stable varying between 6 × 10−11 and 1 × 10−10 mol L−1 during the experiment, indicating little biological production over time. In contrast, ATP levels in the 0.5% treated water exhibited a 100-fold increase from ~6.0 × 10−11 to ~ 5.0 × 10−9 mol L− 1 over the first 9 days and remained stable afterwards (Fig. 2c). Levels in the 1.0% treated tank declined slightly from 4.8 × 10−11 to 1.5 × 10−11 mol L−1 during the first week before slowly increasing to a maximum of 3.8 × 10−8 mol L−1 at the end of the experiment. This corresponds to a nearly 1000-fold increase in living biomass over the course of the experiment (Fig. 2c). ATP measurements at near surface and bottom of tanks did not differ

1039

significantly (paired T-test, p N 0.05), indicating no vertical gradient of living biomass in the tanks. The difference in the time course of ATP levels between the two treated tanks did match that noted for the deoxygenation curves (Fig. 1b), where maximum ATP levels and anoxic conditions happened in the 0.5% treated water well before that measured in the 1.0% treatment. The regression coefficient (adjusted r2) of the relationship between ATP levels and viable cell density (both on logarithmic scale) was 0.70 and 0.74 for the 0.5% and 1.0% treated tanks respectively, but dropped to 0.47 when pooling data from both treated tanks. This suggests that ATP measurements may be indicative of the growth dynamics of the viable yeast population in individual tanks but cannot be a strong predictor of yeast cell density in all cases. Using the predictive relationship between water temperature and time (in hours) to reach hypoxia previously developed by de Lafontaine et al. (2013) and using the mean temperature of 1.5 °C observed during our experiment, we estimated the predicted time to hypoxia to be 235 h (9.8 days). This was very close to the value of 247 h (10.3 days) observed in the 0.5% treated tank (Fig. 3), indicating that the performance of the bioreactive process can be predicted fairly accurately at temperatures ranging from 1 to 25 °C. The time to hypoxia at the 1% treatment concentration was however 6 days longer than that predicted. This was unexpected since previous laboratory experiments did not reveal such large differences in time to hypoxia when varying treatment concentrations between 0.33% and 1.0% at similar temperature conditions (de Lafontaine et al., 2013). The lower performance of the 1% treated tank was apparently due to the slower growth of yeasts relative to that observed in the 0.5% treated tank. It was previously suggested that the minimum yeasts density required to ensure successful biological deoxygenation at temperature of N4 °C should be between 4 × 105 and 8 × 105 viable cells mL−1 (de Lafontaine et al., 2013). The density of yeasts in the 1% treated tank during the first 13 days of our experiment (Fig. 2) remained well below this threshold, while that in the 0.5% treated waters reached it on day 9. This suggests that water deoxygenation would essentially depend upon the establishment of a critical density of viable yeasts. Our results further suggested that this threshold density would be independent of water temperature per se and would indeed generate hypoxic conditions in a similar way at any water temperature ranging from 1 to 25 °C. The much lower living biomass (as measured by ATP levels) in the 1% treatment water over the first week of the experiment (Fig. 2c) suggests that some factor contributed to depress yeast growth and production in that particular tank, resulting in delayed deoxygenation.

264 Y = 7.35 + 308.9 exp -(0.199 Temp) 2 r = 0.88

240

Time to hypoxia (hours)

216 192 168 144 120 96 72 48 24 0 0

5

10

15

20

25

30

Temperature (oC) Fig. 2. a) Density of viable yeast cells (cells mL−1), b) percentage of viable cells, and c) total ATP concentrations (mol L−1) in treated and untreated water tanks as a function of time since treatment inoculation (days).

Fig. 3. Time to hypoxia in the 0.5% treatment concentration (black triangle) at a mean temperature of 1.5 °C in comparison to the predicted curve for the deoxygenation relationship as a function of water temperature. Open circles correspond to experimental results from previous lab experiments (de Lafontaine et al., 2013).

1040

Y. de Lafontaine, S.-P. Despatie / Science of the Total Environment 472 (2014) 1036–1043

3.2. Water quality The bioreactive process of the yeast treatment did alter water quality parameters to various degrees over the course of the experiment. Except for NO2–NO3 concentrations (0.50–0.55 mg-N L−1) that did not differ significantly between untreated and treated tanks, treated waters were characterized by significantly higher concentrations of PO4, total P, NH4, DOC, POC, PON and TSS than in the control tank at the beginning of the experiment (Table 1). The increase in concentrations was proportional to the treatment concentration in each tank. As a result of yeasts addition and growth dynamics over time, concentrations of POC and PON remained stable during the first week but increased by 6.5 and 5 times in the 0.5% and 1% treatment tank respectively when reaching their maximum at the end of the experiment. The C/N ratio of treated water varied between 5.27 and 7.71 over time but did not differ significantly between treatment concentrations (ANOVA F-test, p N 0.05). Because of the increase in yeast biomass over time, total suspended solids (TSS) increased by 4 to 6 times between day 6 and day 19 reaching concentrations of 12.6 and 31 mg L− 1 in the 0.5 and 1% treated tanks respectively (Table 1). Values in the control tank dropped from 2.5 to 1.2 mg L−1 over the same time period possibly due to some particles settling in untreated waters. NO2–NO3 concentrations (0.50– 0.55 mg-N L−1) declined below detection limits (b 0.04 mg-N L−1) in fully deoxygenated water but remained stable in untreated water. Levels of PO4 and total P also declined by more than 50% over time and PO4 even became exhausted under sustained anoxic conditions in the 0.5% treated tank. DOC that was initially 20 and 40 times more concentrated in treated water relative to that in the control tank, dropped by 20–25% in each treated tank by the end of the experiment. No significant trend in DOC was noted in the control tank. Elevated concentrations of NH4 measured in treated water (0.7–0.9 mg-N L−1) immediately after inoculation continued to increase during the deoxygenation process to reach a maximum of 10.38 and 2.03 mg-N L−1 under full anoxic conditions in the 0.5% and 1% treated tank, respectively. The

Table 1 Descriptive statistics of water properties in each experimental tank. Mean values with standard deviation are in parentheses and the range of values measured in each tank is given for each parameter. The number of measurements (n) for each parameter in each tank is also indicated. Parameter

Control tank

0.5% treated tank

1.0% treated tank

Conductivity (μS cm−1) (n = 31) pH (n = 23) NH4 (mg-N L−1) (n = 6) PO4 (mg-P L−1) (n = 6) TP (mg-P L−1) (n = 6) DOC (mg-C L−1) (n = 6) PON (mg-N L−1) (n = 11) POC (mg-C L−1) (n = 11) C/N ratio (n = 11) NO2–NO3 (mg-N L−1) (n = 6) H2S (mg-S L−1) (n = 11) TSS (mg L−1) @ day 6 @ day 19 Chlorophyll a (μg L−1)

301.7 (±0.8) 300.3–303.9 7.60 (±0.09) 7.37–7.73 0.032 (±0.004) 0.025–0.035 0.021 (±0.002) 0.019–0.024 0.032 (±0.002) 0.029–0.035 5.3 (±0.6) 4.5–6.3 b0.08

305.3 (±3.4) 302.4–314.1 6.02 (±0.21) 5.72–6.61 2.671 (±3.792) 0.788–10.355 0.182 (±0.086) 0.010–0.229 0.237 (±0.029) 0.155–0.259 96.2 (±9.8) 76.8–104.6 0.28 (±0.25) 0.15–1.00 1.79 (±1.64) 1.03–6.60 6.57 (±0.81) 5.27–7.12 b0.04–0.52

318.1 (±1.3) 314.6–320.7 4.68 (±0.13) 4.36–5.08 1.473 (±0.459) 0.988–2.027 0.401 (±0.136) 0.126–0.472 0.464 (±0.081) 0.222–0.502 195.8 (±15.0) 165.3–203.4 0.39 (±0.34) 0.26–1.40 2.69 (±2.69) 1.68–10.80 6.65 (±0.55) 6.35–7.71 b0.04–0.55

0.034 (±0.016) 0.016–0.067 3.4 12.6 0.261 (±0.064) 0.409–0.126 3.01 1.76–5.11

0.076 (±0.019) 0.046–0.106 5.2 31.1 0.165 (±0.074) 0.382–0.062 1.10 0.62–2.17

Chlorophyll a/phaeopigment ratio

0.39 (±0.15) 0.04–0.61 – 0.51 (±0.01) 0.50–0.53 0.022 (±0.016) 0.009–0.051 2.5 1.2 0.288 (±0.069) 0.426–0.097 3.13 1.39–7.29

difference was apparently related to the difference in yeast growth rate in each treated tank. Concentrations of hydrogen sulfide (H2S) in treated water were always less than 0.11 mg L−1, indicating very little production during the experiment. These changes in water properties translated into slightly higher conductivity and lower pH of treated water. Water conductivity was significantly (ANOVA test, p b 0.05) lower in the untreated tank (301.7 μS cm− 1, s.d. ± 0.8, n = 31) than in the two treated tanks (305.3 ± 3.4 μS cm−1 in the 0.5% treated tank; 318.1 ± 1.3 μS cm−1 in the 1.0% treated tank) (Table 1). While conductivity did not fluctuate much in both the control and the 1% treated tanks, values in the 0.5% treated tank gradually increased from 302.5 to 314.1 μS cm−1 during the second half of the experiment. The pH of treated water was 6.02 (s.d. ± 0.21) and 4.68 (s.d. ± 0.13) for the 0.5 and 1.0% treatment concentration respectively and was significantly lower than for untreated water (mean = 7.60, s.d. ± 0.09) (Table 1). The pH dropped immediately after treatment inoculation and showed no significant trend over time. No significant (paired T-test, p N 0.05) difference in water conductivity and pH between surface and bottom samples was noted. Because of the very depauperated phytoplankton population in the river during the mid-winter month, the phytoplankton biomass, as determined by chlorophyll a concentration, was very low (b0.45 μg L−1) in each tank throughout the experiment but was significantly lower (ANOVA F-test, p b 0.05) in the 1% treated tank (Table 1). Chlorophyll concentrations dropped by 50% reaching 0.2 μg L−1 on day 7 and day 9 in the 0.5% and control tanks respectively, suggesting minimal phytoplankton growth due to the lack of incident light at water surface inside the tanks. The decline in living phytoplankton was much more rapid and more important in the 1% treated tank where levels dropped below 0.2 μg L−1 on day 3 and reached 0.08 μg L−1 on day 9. Chlorophyll a concentrations varied between 0.062 and 0.125 μg L−1 in all tanks at the end of the experiment. The chlorophyll a/phaeopigment ratio also declined over time in each tank and was significantly lower (ANOVA F-test, p b 0.05) in the 1% treated water. Ratio values in the control and 0.5% treated tanks remained above 2.0 during the first 9 days of the experiment while values in the 1% treated tank declined gradually from 2.0 to 0.6 during the treatment. The mean value of 1.1 was strongly indicative of an unhealthy and decaying phytoplankton population in this treated tank, as opposed to mean values greater than 3.0 for the two other tanks (Table 1). The large and rapid decline in living phytoplankton biomass within the 1% treated tank did happen well before the water became anoxic which suggested that water deoxygenation was not the causal factor for this unhealthy algal population. Presumably, the poor water quality due to the very low pH might have caused severe acidification stress to phytoplankton in the 1% treated tank, as it has been demonstrated to occur in natural lake systems (Leavitt et al., 1999). 3.3. Ecotoxicological assays Results from six ecotoxicological bioassays indicated that treated waters exhibited no potential toxic response at the 0.5% treatment level, while a weak but significant toxic response was measured in all bioassays performed on the 1.0% treated water samples (Table 2). The toxic response varied between 1 and 4 toxic units (depending on the tests), indicating that a 1:4 dilution factor would be at least needed to eliminate the potential toxic effects of treated waters at the time of discharge. The fact that the positive response was noted for the 1.0% treatment only and, more importantly, was also noted during the first week when waters were still highly oxygenated, strongly suggested that the toxic response might have resulted from poor water quality of treated water rather than effects associated with water deoxygenation. Presumably, the very low pH values of the 1% treated waters would be considered problematic for many aquatic organisms and could have resulted in toxic responses for the various bioassays. Although elevated NH4 concentrations might sometimes represent some toxic stress for

Y. de Lafontaine, S.-P. Despatie / Science of the Total Environment 472 (2014) 1036–1043

1041

Table 2 Results of the six bioassays for potential toxicity of the treated water at 0.5% and 1.0% v/v treatment concentrations expressed in toxic units (TU). Samples were taken from the bottom of tanks. (Temp. = water temperature, D.O. = dissolved oxygen, TSS = total suspended solids). Control

Toxicological tests (expressed in toxic units TU)

Elapsed time (h)

Temp. (°C)

pH

D.O. (mg L−1)

TSS (mg L−1)

Microtox (IC50)

Algae (IC50)

Hydra (LC50)

Daphnia (LC50)

Microcrustacean (LC50)

Trout test (LC50)

3 138 449

1.23 0.89 1.81

7.7 7.7 7.7

12.83 13.09 13.29

– 2.5a 1.2a

– b2.0 b2.0

– b1.0 b1.0

– b1.0 b1.0

– b1.0 b1.0

– b1.0 b1.0

– b1.0 b1.0

0.5%

Toxicological tests (expressed in toxic units TU)

Elapsed time (h)

Temp. (°C)

pH

D.O. (mg L−1)

TSS (mg L−1)

Microtox (IC50)

Algae (IC50)

Hydra (LC50)

Daphnia (LC50)

Microcrustacean (LC50)

Trout test (LC50)

3 138 449

1.36 1.28 1.82

5.9 5.9 6.5

12.94 12.85 0.62

– 3.4 12.6

– b2.0 b2.0

– 1.8 b1.0

– b1.0 b1.0

– b1.0 b1.0

– b1.0 b1.0

– b1.0 b1.0

Elapsed time (h)

Temp. (°C)

pH

D.O. (mg L−1)

TSS (mg L−1)

Microtox (IC50)

Algae (IC50)

Hydra (LC50)

Daphnia (LC50)

Microcrustacean (LC50)

Trout test (LC50)

3 138 449

1.43 1.43 1.92

5.1 4.7 4.7

12.83 12.42 0.56

– 5.2 31.1

– 4.0 4.3

– 1.0–4.0 3.8

– 1.0–4.0 2.8

– 4.8 1.6

– 1.5 1.4

– 1.4 b1.0

1.0%

a

Toxicological tests (expressed in toxic units TU)

Trace amounts.

aquatic systems (Chambers et al., 1997), levels measured in our experiment would not be considered acutely toxic under low pH conditions (Environment Canada/Health Canada, 2001). Moreover, the observation that the much higher NH4 levels measured in the 0.5% treated waters (relative to that in the 1% treated tank) did not result in toxic response would support the hypothesis that ammonia was probably not the main factor causing the potential toxic response of the 1% treated water. 4. Discussion and conclusion Our experimental results provided the first evidence that biological deoxygenation of water using a Rhodotorula-based bioreactor can be carried out successfully at very cold temperatures (1–2 °C). The treatment process was effective in generating and maintaining anoxic conditions over several days. Although the two treated tanks had different treatment levels and could not be considered as exact replicates, complete water deoxygenation was achieved in both tanks as previously observed in 15 replicated experiments conducted at warmer temperature (de Lafontaine et al., 2013). Results of this present study and those from previous trials (de Lafontaine et al., 2013) confirmed that complete biological deoxygenation of water by this yeast-based bioreactor is possible from 1 °C up to 25 °C. Although Rhodotorula yeasts are ubiquitous in the aquatic environment (Nagahama, 2006), in-situ growth conditions for these species have not been documented. Laboratory experiments at temperatures between 20 and 37 °C showed that growth of R. glutinis was optimal at 28 °C (Martinez et al., 2006), while growth parameters of Rhodotorula glacialis cultivated at temperatures between 1 and 20 °C were maximal at 1 °C (Margesin, 2009). The testing and validation of the relationship between temperature and time to hypoxia (Fig. 3) showed that growth and production of the Rhodotorula mixture used in our experiment was indeed slowed but not stopped at very cold temperatures. Our results further demonstrate that, even under such very slow growth conditions, the bioreactor will lead to anoxic water conditions when yeast density eventually reaches a threshold value of ~5 × 105 viable yeast cells mL−1, as previously estimated for warmer water conditions (de Lafontaine et al., 2013). The difference in the time to hypoxia between the two treated tanks (Fig. 1) was obviously not related to the treatment concentration or the initial density of viable yeast cells (Fig. 2), as one might have expected assuming that the higher density of cells in the 1% treated tank would favor faster deoxygenation. It was previously shown that variation in

treatment concentration by a factor of 3 did not have a significant impact on time to hypoxia in trials conducted at warmer temperatures (de Lafontaine et al., 2013). The poorer performance of the 1% treatment concentration in this cold water experiment was presumably related to some factor affecting the quality of the 1% treated water and the yeasts growth during the first week of treatment. The very low pH (averaging 4.68) of the 1% treated waters immediately after inoculation and throughout the experiment was unexpected and unpredicted given that the pH of similarly treated water was around 6.0 in previous laboratory experiments (de Lafontaine et al., 2013). Apparently, this resulted from the dilution of the acidic inoculum (pH = 2.9). A decrease of 1–1.5 pH units was also reported in biologically deoxygenated ballast waters resulting from bacterial growth stimulated by nutrient addition (McCollin et al., 2007). The acidic component causing such low pH values of the inoculum and treated waters was not determined but could have resulted from the utilization of ammonium ion as a nitrogen source by Rhodotorula yeasts causing the culture medium to become acidic (Cho et al., 2001). Other laboratory studies have shown that the growth efficiency and production of yeast cultures, including some Rhodotorula species, remained high in acidic conditions and did not vary much over pH values ranging between 3.0 and 7.0 (Cho et al., 2001; Martinez et al., 2006; Spotholz et al., 1956). The pH in the culture medium for these controlled experiments was usually adjusted using HCl or NaOH, whereas the pH of water in our experimental tanks was not controlled and would merely depend upon the dilution of the stock mixture. Another cause of the very low pH would be contamination of the yeast cultures by lactic or acetic bacteria (Drysdale and Fleet, 1989; Sousa et al., 2012). These biological contaminants can lead to the production of acetic acid or lactic acid over time. Yeasts tolerance to acetic acid exposure varies between species and can elicit a stress response in some yeasts, inhibiting yeast growth and eventually causing cell death (Pinto et al., 1989; Sousa et al., 2012). A third but less plausible explanation would be related to tank filling operations that resulted in some “tank effect” causing water quality differences between tanks. Although the exact cause of the low pH of the 1% treated water in our experiment remained unexplained, it appeared that this acidic condition was sufficiently toxic to significantly reduce yeasts growth and abundance and the overall living biomass as indicated by lower ATP levels during the first week of the treatment (Fig. 2), even in absence of any deoxygenation. Low pH also affected living phytoplankton as chlorophyll a concentrations in the 1% treated water were significantly

1042

Y. de Lafontaine, S.-P. Despatie / Science of the Total Environment 472 (2014) 1036–1043

lower than in the two other tanks (Table 1). This is in agreement with other evidence about the negative impact of severe and prolonged acidification (pH b 5.0) on phytoplankton and other primary producers (Leavitt et al., 1999; Vinebrooke, 1996). The extent to which the age or the quality of the yeast mixture may affect the performance of the Rhodotorula bioreactive process remains to be explored in greater details. Given that the time to reach hypoxia would essentially depend on the time it takes to get a density of around 5 × 105 viable cells mL–1, the physiological conditions and the initial density of viable cells in the inoculum would affect cells growth rate and thus the time to reach full deoxygenation. Presumably, an inoculum with a more viable yeast population could have resulted in a faster deoxygenation rate. Previous study results indicated that quality of the different yeast batches had a significant impact on treatment effectiveness and time to reach hypoxia (de Lafontaine et al., 2013). Results from this cold water experiment further support the recommendation that the production of yeast batches and yeast mixtures be standardized and checked to minimize variability in treatment effectiveness. Despite the long time needed to achieve complete deoxygenation of water and the absence of water mixing or agitation during the course of our experiment, we found no evidence of vertical stratification in oxygen levels, yeast density, temperature, pH or any other measured parameters of water quality. This was similar to that observed in one experiment conducted in a 2-m deep tank at warmer temperature (de Lafontaine et al., 2013). This confirmed that biological deoxygenation in cold water occurred simultaneously over the entire volume of water being treated, suggesting that container size would not be a limiting factor to the bioreactive process. The possible sinking of yeast cells resulting from slower growth in cold water was apparently not important enough to generate vertical heterogeneity in anoxic conditions. Given that most ballast water treatment methods usually add small quantities of active chemicals or products relative to large volumes of ballast water, it is important to verify for nonhomogeneous treatment application that may sometimes result in partial mixing and water stratification inside the ballast tanks with some part remaining “untreated” (de Lafontaine et al., 2009; Herwig et al., 2006; Roed et al., 2005; Wilson et al., 2006). The achievement of complete deoxygenation (irrespective of treatment method) over the entire water column will be crucial to ensure treatment efficacy against organisms present in ballast tanks, because it was shown that mobile organisms can rapidly avoid hypoxic areas by migrating to zones of higher oxygen concentrations as a refuge (Montagna and Ritter, 2006). Changes in all measured parameters of water quality due to treatment application at very cold water conditions (Table 1) were very similar and within that previously measured in warmer experimental conditions (de Lafontaine et al., 2013). Such changes, characterized by increased concentrations of ammonium and DOC (mostly sugars) and severe depletion of nitrates and phosphates of treated water, were also noted in a shipboard deoxygenation experiment by nutrient addition to stimulate bacterial growth (McCollin et al., 2007). The potential environmental impact of these chemically altered waters was previously and fully discussed by de Lafontaine et al. (2013) who concluded that none of the observed changes in water quality of deoxygenated water would result in acute toxic risk at the time of discharge. Water samples from the 0.5% treated tank, which was effective in yielding complete water deoxygenation, did not indicate any residual toxicity, which supports previous bioassay results with the same deoxygenation process (de Lafontaine et al., 2013). In contrast, samples from the 1.0% treated water were found to be slightly toxic to a wide variety of organisms, from bacteria to fish, irrespective of the levels of dissolved oxygen (Table 1). As indicated above, this positive toxic response noted before water deoxygenation began was probably more related to the low pH (4.68) of the treated water rather than the deoxygenated condition. As found with other chemical treatments that may alter water characteristics, this would imply that water quality of the treated water should be monitored and rendered safe to minimize possible environmental

impact at the time of discharge (de Lafontaine et al., 2008, 2009; La Carbona et al., 2010; Sano and Landrum, 2005; Smit et al., 2008; Tsolaki and Diamadopoulos, 2010). As discussed by de Lafontaine et al. (2013), the activity of yeast population after being discharged in the environment would rapidly cease in absence of high organic carbon levels, and would cause little environmental impact to the receiving waters. This study demonstrated that the potential application of a yeastbased deoxygenation process for treating ships' ballast water could be extended over the range of water temperatures typically encountered during most shipping operational conditions (Drillet et al., 2013). Time to reach full deoxygenation at water temperature of b4 °C may however be limiting for universal application of this treatment. For example, ships that travel between Europe and North America and along the eastern seaboard of North America within one week will sail in b4 °C waters during at least 5 months a year (November–April). In these conditions, voyage duration (b7 days) would be shorter than the estimated time to reach full anoxia in ballast water (Fig. 3), making the deoxygenation treatment probably ineffective in eliminating living organisms before the end of the trip. This treatment method should preferably be used for ships making longer voyages in cold water environments thus ensuring a long enough exposure time of organisms to anoxic conditions in the ballast tanks. Time to hypoxia relative to voyage duration was also found to be problematic for another deoxygenation method using nutrient enrichment to stimulate bacterial growth, during a shipboard experiment conducted in warm waters (McCollin et al., 2007). As correctly pointed out by Drillet et al. (2013), despite the fact that often temperature has not been taken into account during test cycles for ballast water treatment technologies, several studies indicated that temperature may affect the efficacy of chemical treatments and will influence the decay rates of the chemicals and their byproducts. All ballast water treatment systems have some limitations and none could actually be considered as a fully efficient solution under all operational conditions and able to fully meet all performance standard criteria (Regulation D2) set by IMO (2004). While the performance of the yeast-based deoxygenation process tested here was found to be strongly dependent on incubating temperature, treatment duration can be predicted using the developed empirical relationships for best management practices. Contrary to other chemical treatments resulting in longer toxic effects at low temperatures (de Lafontaine et al., 2008, 2009), we found no evidence that biological deoxygenation at low temperatures increased the toxic risk of treated water that may impede their disposal at the time of discharge. Acknowledgments We sincerely thank Éloïse Veilleux, Jasmin Perrier and Yan Chambers for their technical assistance and great support during this experiment and to Michel Lagacé, former scientific director at the Aquarium du Québec, for his generous help with lab facilities and logistics. We are also grateful to Michel de Blois from Redal Technologie for providing the stock yeast solution and to Chris Wiley for ensuring the financial support of Department of Fisheries and Oceans. All bioassays using experimental animals during this study were conducted following approved protocols in accordance to guidelines provided by the GLLFAS/ NWRI animal care committee of Environment Canada. References American Water Works Association (AWWA). Standard methods for the examination of water and wastewater. Washington, DC: American Public Health Association, Water Environment Federation, and American Water Works Association; 2005 (1–1368 pp.). Carlton JT, Geller JB. Ecological roulette: the global transport of nonindigenous marine organisms. Science 1993;261:78–82. Chambers PA, Allard M, Walker SL, Marsalek J, Lawrence J, Servos M, Busnarda J, Munger KS, Adare K, Jefferson C, Kent RA, Wong MP. Impacts of municipal wastewater effluents on Canadian waters: a review. Water Qual Res J Can 1997;32:659–713.

Y. de Lafontaine, S.-P. Despatie / Science of the Total Environment 472 (2014) 1036–1043 Chan F, Bailey S, Wiley C, MacIsaac H. Relative risk assessment for ballast-mediated invasions at Canadian Arctic ports. Biol Invasion 2013;15:295–308. Cho DH, Chae HJ, Kim EY. Synthesis and characterization of a novel extracellular polysaccharide by Rhodotorula glutinis. Appl Biochem Biotechnol 2001;95:183–94. de Lafontaine Y, Despatie S-P. Evaluation of the effectiveness and potential toxicological impact of the PERACLEAN Ocean ballast water treatment technology. AEP-TN07-001. WSTD Technical Report SeriesBurlington, ON: Environment Canada Water Science and Technology Directorate; 2006. p. 1–48. de Lafontaine Y, Despatie S-P, Wiley CJ. Effectiveness and potential toxicological impact of PERACLEAN Ocean Ballast water treatment technology. Ecotoxicol Environ Saf 2008;71:355–69. de Lafontaine Y, Despatie S-P, Veilleux E, Wiley CJ. Onboard ship evaluation of the effectiveness and the potential environmental effects of PERACLEAN Ocean for ballast water treatment in very cold conditions. Environ Toxicol 2009;24:49–65. de Lafontaine Y, Chambers Y, Despatie S-P, Gagnon C, Blaise C. Effectiveness and potential environmental impact of a yeast-based deoxygenation process for treating ships' ballast waters. Water Qual Res J Can 2013;48:55–75. Drillet G, Schmoker C, Trottet A, Mahjoub MS, Duchemin M, Andersen M. Effects of temperature on type approval testing of ballast water treatment systems. Integr Environ Assess Manag 2013;9:192–5. Drysdale GS, Fleet GH. The effect of acetic acid bacteria upon the growth and metabolism of yeast during the fermentation of grape juice. J Appl Bacteriol 1989;67:471–81. Environment Canada/Health Canada. Canadian Environmental Protection Act, 1999 — ammonia in the aquatic environment. Priority substances list assessment report. Ottawa: Environment Canada & Health Canada; 20011–96. Gollasch S, David M, Voigt M, Dragsund E, Hewitt C, Fukuyo Y. Critical review of the IMO international convention on the management of ship's ballast water and sediments. Harmful Algae 2007;6:585–600. Gonçalves AA, Gagnon GA. Recent technologies for ballast water treatment. Ozone Sci Eng 2012;34:174–95. Gregg MD, Hallegraeff GM. Efficacy of three commercially available ballast water biocides against vegetative microalgae, dinoflagellate cysts and bacteria. Harmful Algae 2007;6:567–84. Hagler AN, Ahearn DG. Ecology of aquatic yeasts. In: Rose AH, Harrison JS, editors. The yeasts. Biology of yeastsLondon: Academic Press; 1987. p. 181–205. Herwig RP, Cordell JR, Perrins JC, Dinnel PA, Gensemer RW, Stubblefield WA, Ruiz GM, Kopp JA, House ML, Cooper WJ. Ozone treatment of ballast water on the oil tanker S/T Tonsina: chemistry, biology and toxicity. Mar Ecol Prog Ser 2006;324:37–55. Hunt CD, Tanis DC, Stevens TG, Frederick RM, Everett RA. Verifying ballast-water treatment performance. Environ Sci Technol 2005:321A–328A. Inouye M, Phadtare S. Cold shock response and adaptation at near-freezing temperature in microorganisms. Sci SKTE 2004:pe36. International Maritime Organisation (IMO). The international convention for the control and management of ships' ballast water and sediments. London, UK: United Nations IMO; 2004. Kandror O, Bretschneider N, Kreydin E, Cavalieri D, Goldberg AL. Yeast adapt to near-freezing temperatures by STRE/Msn2,4-dependent induction of trehalose synthesis and certain molecular chaperones. Mol Cell 2004;13:771–81. La Carbona S, Viitasalo-Frösen S, Masson D, Sassi J, Pineau S, Lehtiniemi M, Corroler D. Efficacy and environmental acceptability of two ballast water treatment chemicals and an alkylamine based-biocide. Sci Total Environ 2010;409:247–55. Landrum PF, Sano LL, Mapili MA, Garcia E, Krueger AM, Moll RA. Degradation of chemical biocides with application to ballast water treatment. NOAA Technical Memorandum GLERL, 123. Ann Arbor, Michigan: NOAA — Great Lakes Environmental Research Laboratory; 2003. p. 1–37. Leavitt PR, Findlay DL, Hall RI, Smol JP. Algal responses to dissolved organic carbon loss and pH decline during whole-lake acidification: evidence from paleolimnology. Limnol Oceanogr 1999;44:757–73. Lodge DM, Williams S, MacIsaac HJ, Hayes KR, Leung B, Reichard S, Mack RN, Moyle PB, Smith M, Andow DA, Carlton JT, McMichael A. Biological invasions: recommendations for U.S. policy and management. Ecol Appl 2006;16:2035–54.

1043

Margesin R. Effect of temperature on growth parameters of psychrophilic bacteria and yeasts. Extremophiles 2009;13:257–62. Martinez C, Gertosio C, Labbe A, Perez R, Ganga MA. Production of Rhodotorula glutinis: a yeast that secretes á-L-arabinofuranosidase. Electron J Biotechnol 2006;9:407–13. Matheickal J, Raaymakers S. 2nd International Ballast Water Treatment R&D Symposium, IMO London, 21–23 July 2003: proceedings. GloBallast Monograph Series. London: IMO; 2004. McCollin T, Quilez-Badia G, Josefsen KD, Gill ME, Mesbahi E, Frid CLJ. Ship board testing of a deoxygenation ballast water treatment. Mar Pollut Bull 2007;54:1170–8. Montagna PA, Ritter C. Direct and indirect effects of hypoxia on benthos in Corpus Christi Bay, Texas, U.S.A. J Exp Mar Biol Ecol 2006;330:119–31. Mountfort DO, Hay C, Dodgshun T, Buchanan S, Gibbs W. Oxygen deprivation as a treatment for ships' ballast water — laboratory studies and evaluation. J Mar Environ Eng 1999;5:175–92. Nagahama T. Yeast biodiversity in freshwater, marine and deep-sea environments. In: Rosa CA, Peter G, editors. Biodiversity and ecophysiology of yeasts. Berlin Heidelberg: Springer-Verlag; 2006. p. 241–62. Parsons TR, Maita Y, Lalli CM. A manual of chemical and biological methods for seawater analysis. Oxford: Pergamon Press; 1984 (1–173 pp.). Pena MA, Lewis MR, Harrison WG. Particulate organic matter and chlorophyll in the surface layer of the equatorial Pacific Ocean along 135° W. Mar Ecol Prog Ser 1991;72:179–88. Pinto I, Cardoso H, Leao C, van Uden N. High enthalpy and low enthalpy death in Saccharomyces cerevisiae induced by acetic acid. Biotechnol Bioeng 1989;33:1350–2. Roed PC, Stocks DT, O'Reilly M. Heat treatment of ballast tank residues, control of invasive species. BMT Fleet Technology Limited, Internal Report; 20051–9. Sano LL, Landrum PF. Evaluation of different biocides for potential use in treating overseas unballasted vessels entering the Great Lakes. Aquat Invaders 2005;16:1–11. Sano LL, Bartell SM, Landrum PF. Decay model for biocide treatment of unballasted vessels: application for the Laurentian Great Lakes. Mar Poll Bull 2005;50: 1050–60. Smart KA, Chambers KM, Lambert I, Jenkins C, Smart CA. Use of methylene violet staining procedures to determine yeast viability and vitality. J Am Soc Brew Chem 1999;57: 18–23. Smit MGD, Ebbens E, Jak RG, Huijbregts MAJ. Time and concentration dependency in the potentially affected fraction of species: the case of hydrogen peroxide treatment of ballast water. Environ Toxicol Water Qual 2008;27:746–53. Sousa MJ, Ludovico P, Rodrigues F, Leao C, Corte-Real M. Stress and cell death in yeast induced by acetic acid. In: Bubulya P, editor. Cell metabolism — cell homeostasis and stress responseINTECH Open Science, Rijeka, Croatia; 2012. p. 73–100. Spotholz CH, Litchfield JH, Ordal ZJ. The effect of pH, temperature, and composition of the medium on growth characteristics of Rhodotorula gracilis. Appl Microbiol 1956;4: 285–8. Tamburri MN, Ruiz GM. Evaluations of a ballast water treatment to stop invasive species and tank corrosion. SNAME; 20051–9. Tamburri MN, Wasson K, Matsuda M. Ballast deoxygenation can prevent aquatic introductions while reducing ship corrosion. Biol Conserv 2002;103:331–41. Tamburri MN, Little BJ, Ruiz GM, Lee JS, McNulty PD. Evaluations of Venturi Oxygen Stripping as a ballast water treatment to prevent aquatic invasions and ship corrosion. In: Matheickal J, Raaymakers S, editors. GloBallast Monograph Series, No.15. London: IMO; 2004. p. 34–47. Tsolaki E, Diamadopoulos E. Technologies for ballast water treatment: a review. J Chem Technol Biotechnol 2010;85:19–32. Vinebrooke RD. Abiotic and biotic regulation of periphyton in recovering acidified lakes. J N Am Benthol Soc 1996;15:318–31. Ware C, Berge J, Sundet JH, Kirkpatrick JB, Coutts ADM, Jelmert A, Olsen SM, Floerl O, Wisz MS, Alsos IG. Climate change, non-indigenous species and shipping: assessing the risk of species introduction to a high-Arctic Archipelago. Div Distrib 2013:1–10. Wilson W, Chang P, Verosto S, Atsavapranee P, Reid DF, Jenkins PT. Computational and experimental analysis of ballast water exchange. Nav Eng J 2006;3:25–36.